JPH0798268A - Dynamic system diagnostic and tire pneumatic-pressure diagnostic device and device for detecting fluctuation of car body weight using the diagnostic device - Google Patents

Abstract

PURPOSE:To locate a fault simply and accurately by providing a means for specifying the corresponding place of a dynamic system based on the component associated with the separated internal disturbance. CONSTITUTION:The internal state of a dynamic system 10 is changed by a control input 14 and external disturbance 15, and a control output 16 is changed. The output 16 is fed back to a control device 12, and the system 10 controlled. A disturbance estimating means 32 estimates integrated disturbance vector (w) and internal-state- quantity vector (x) and outputs the vectors into a correlation operating means 34. The means 34 operates the correlation of the vector (w) and the vector (x). The component associated with the internal disturbance is separated from each element of the vector (w) and outputted into a diagnostic means 36. The means 36 determines the place of the diagnosis of the system 10 based on the input component. Each element of the component associated with the internal disturbance corresponds to each fault plate generated in the system 10. Therefore, the means 36 can detects the occurrence of the fault and the fault place in the system 10 based on each element of the component associated with the separated internal disturbance.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a dynamic system diagnosing device, a tire pressure diagnosing device using the same, and a vehicle body weight fluctuation detecting device. To a diagnostic device for diagnosing a failure of a dynamic system, detecting an abnormality in tire air pressure, detecting a change in vehicle body weight, a tire air pressure diagnostic device using the same, and a vehicle body weight variation detecting device.

[0002]

2. Description of the Related Art Conventionally, a device for diagnosing a fault in a dynamic system is well known. This diagnostic device determines the presence / absence of a failure and the failure location by using the residual between the response of the dynamic system estimated from the normal model of the dynamic system and the response actually measured by a sensor or the like.

As a representative of this, the generalized likelihood comparison test method (ASWillsky & HLJones, “AGeneralized Likelih
ood Ratio Approach to Detection and Estimation of J
umps in Linear Systems, "IEEE Trans. AC-21, No.1)
Is mentioned.

FIG. 7 shows an apparatus embodying this likelihood comparison test method. The failure diagnosis device 20 targets the dynamic system 10 controlled based on the control input 14 from the controller 12. Where u, d
Are control input vectors and external disturbance vectors input to the dynamic system 10. y is the control output vector of the dynamic system. x is an internal state quantity vector of the dynamic system 10 measured using a sensor or the like.

The failure diagnosis device 20 includes an observer 22 for a normal model and a plurality of observers 24 for a failure model.
-1, 24-2, ..., Likelihood ratio test units 26-1, 26-2, ... Corresponding to each failure model observer,
The failure determination unit 28 is included.

The normal model observer 22 has a control input vector u and a control output vector y of the dynamic system 10.
From this, the state of the dynamic system 10 is estimated and output based on the normal model. A residual 25 between the estimated output signal 23 and the state quantity x of the dynamic system 10 actually measured by a sensor or the like is input to the likelihood ratio test units 26-1, 26-2.

Failure model observers 24-1, 24-2
... estimates and calculates the state of the dynamic system 10 based on different failure models. Then, the estimated output 27- of each of these failure model observers 24-1, 24-2 ...
1, 27-2 ..., and the actually measured dynamic system 10
The residuals 29-1, 29-2, ... Of the state x are input to the corresponding likelihood ratio test units 26-1, 26-2.

The likelihood ratio test units 26-1, 26-2, ...
From the residual signal 25 from the normal model observer 22 and the residual signal 29 from the fault model observer 24, the probability (likelihood) that the corresponding fault model matches the current dynamic system 10 is calculated. The calculation result is output to the failure determination unit 28.

In this way, the likelihood ratio test units 26-1 and 26-2
6-2 ... Calculates the probability that the failure model matches the current dynamic system 10 for each assumed failure model. Then, the failure determination unit 28 determines the failure model with the highest likelihood from the input signal, and determines the failure occurrence and failure location of the dynamic system 10 by using this.

[0010]

However, the failure diagnosis device 20 has the following problems.

First, in this conventional device, the observer 22,
The model corresponding to the failure is obtained from the residuals 25 and 29 between the states 23 and 27 estimated by 24 and the actual measured value x. However, this residual largely depends on the design of the observer, and if an attempt is made to accelerate the state detection speed (fault detection speed) of the observer, the residual itself becomes small, and there is a problem that the fault detection sensitivity decreases.

In particular, in a noisy system or the like, there is a problem that only a large and suddenly changing failure can be detected.

Further, in this conventional apparatus, the calculation of the likelihood for detecting a failure is complicated. Moreover, this calculation must be done according to the individual failure model. For this reason, the amount of calculation becomes enormous, and there is a problem that diagnosis cannot be performed in real time.

Further, this conventional device has a problem that if even one internal state quantity x of the dynamic system 10 to be diagnosed cannot be measured, the internal failure to be diagnosed cannot be detected and specified. That is, the internal state quantity x of the dynamic system 10 is detected as a state quantity vector including a plurality of elements. Therefore, if even one of the elements of the state quantity vector x cannot be measured, the internal failure to be diagnosed cannot be detected or specified.

Furthermore, this conventional device does not consider the separation of the external disturbance d entering from the outside and the internal disturbance generated by the internal failure. Therefore, there is a problem that the measurement accuracy is easily affected by external disturbance.

The present invention has been made in view of such conventional problems, and an object thereof is to make a diagnosis while specifying a diagnosis portion of a dynamic system, and particularly to diagnose a plurality of portions. An object of the present invention is to provide a diagnostic device for a dynamic system that can be accurately performed with a small amount of calculation without using many observers corresponding to locations.

Another object of the present invention is to provide a tire air pressure diagnostic device capable of easily and accurately diagnosing tire air pressure of a wheel in a dynamic system including a suspension and wheels. .

Another object of the present invention is to provide a vehicle body weight variation detecting apparatus for detecting a vehicle body weight variation of a dynamic system composed of a suspension and wheels.

[0019]

[Means and Actions for Solving the Problems]

(1) Invention relating to dynamic system diagnostic device A dynamic system diagnostic device (claim 1) of the present invention is a dynamic system diagnostic device for detecting a failure of a dynamic system, the dynamic system diagnostic device comprising: Based on the internal state quantity vector, a disturbance estimating means for estimating a total disturbance vector as a sum of an external disturbance vector and an internal disturbance vector of the dynamic system, the estimated total disturbance vector, and the internal state quantity vector, Correlation calculating means for calculating the cross-correlation of the internal disturbance and separating the component related to the internal disturbance from the total disturbance vector, and the corresponding portion of the dynamic system is identified from the separated component related to the internal disturbance and its diagnosis And a diagnostic means for performing the diagnosis of the dynamic system.

Here, the correlation calculating means calculates a cross-correlation between an element of the comprehensive disturbance vector and an element of the internal state quantity vector having no correlation with the external disturbance, and the element of the comprehensive disturbance vector. Can be formed to separate the components associated with the internal disturbance from.

Further, the correlation calculation means performs the cross-correlation calculation such that the temporal sum of squares of errors between the total disturbance vector and the product of the internal disturbance vector and the internal state quantity vector is minimized. So that the calculation of the direction vector of the total disturbance using the internal state vector as a base vector is performed to separate the components related to the internal disturbance from the elements of the total disturbance vector. And

Further, the correlation calculating means calculates a cross-correlation between a plurality of elements of the estimated total disturbance vector and an element having no correlation with the external disturbance of the internal state quantity vector to calculate the total disturbance. The diagnostic means is formed to separate each element of the component related to the internal disturbance from the plurality of elements of the vector, and the diagnostic means identifies the diagnostic point of the dynamic system from each element of the separated component related to the internal disturbance. However, it is formed so as to make the diagnosis.

Further, the diagnosing means includes a memory unit in which a failure detection reference value corresponding to each element of the component related to the internal disturbance is stored in advance, and each separated element of the component related to the internal disturbance. It is preferable that a corresponding fault detection reference value is compared and a fault identifying unit that identifies a fault location of the dynamic system is included.

Further, it is preferable that the diagnostic device is formed to include a sensor means for measuring all or a part of each element of the internal state vector of the dynamic system.

Further, the disturbance estimating means can be formed so as to estimate and calculate a part or all of the elements of the internal state vector of the dynamic system.

FIG. 1 shows a dynamic system diagnostic apparatus according to the present invention. Here, the dynamic system 10 to be diagnosed by the diagnostic device 30 is controlled based on the control input 14 from the controller 12. In this dynamic system 10, the internal state changes based on the control input 14 and the external disturbance 15 input from the outside, and the control output 16 changes accordingly. The controller 12 controls the control output 16
Is used as a feedback signal to control the dynamic system 10.

Here, the dynamic system 10 has n internal state quantities (ie, the degree of the system is n). u represents the control input vector 14 to the system 10 of m elements. y represents a control output vector 16 composed of p elements output from the system 10. Also, d is the same degree as the dynamic system 10, n
The external disturbance vector 15 having a number of elements is shown.

Further, the diagnostic device 30 of the present invention includes a disturbance estimating means 32, a correlation calculating means 34, and a diagnosing means 36, and is configured to detect a failure of the dynamic system 10 as an internal disturbance.

The disturbance estimating means 32, based on an internal state quantity vector of the dynamic system 10 (a vector composed of elements representing the internal state quantity of the dynamic system 10), external disturbance of the dynamic system 10 The total disturbance vector w, which is the sum of the vector d and the internal disturbance vector, is estimated and output to the correlation calculation means 34.

In FIG. 1, the control output vector y of the dynamic system 10 is input to the disturbance estimating means 32. Then, the disturbance estimating means 32 estimates and calculates the internal state quantity vector x of the dynamic system 10 from the control output vector y, and outputs it to the correlation calculating means 34. In such an estimation calculation, the control output vector y is added to the internal state quantity vector quantity x
It is performed when the information that can calculate each element of is included. The estimation calculation of the internal state vector x described above is performed simultaneously with the estimation of the total disturbance vector w. Specifically, a new state quantity composed of the total disturbance vector w and the state quantity x shown in Equation 8 is converted into a conventional linear control theory (for example, Furuta and Sano: “Basic System Theory”, Corona Publishing Co., Ltd.).
1978, pp127-137).

If the amount of information contained in the control output vector y is insufficient to estimate the internal state quantity vector, a sensor for detecting the internal state quantity is provided inside the dynamic system 10 as necessary. The sensor output may be input to the disturbance estimation means 32.

When the information of all internal state quantity vectors x can be directly obtained from the control output vector y of the dynamic system 10 or the internal state quantity sensors provided in the dynamic system 10 as necessary, The internal state quantity x may be directly output to the correlation calculation means 34.

The correlation calculating means 34 calculates the cross-correlation between each element of the estimated total disturbance vector w and the element of the internal state quantity vector x to calculate the total disturbance vector w.
The components related to internal disturbance are separated from each element of. The components related to the separated internal disturbance are output to the diagnostic means 36.

The diagnostic means 36 is configured to identify the diagnostic location of the dynamic system 10 from the separated internal disturbance-related components and determine its condition.

Here, the correlation calculating means 34 is
It is preferable to calculate a cross-correlation with respect to a plurality of elements of the total disturbance vector w so as to separate the component related to the internal disturbance from the plurality of elements of the total disturbance vector w.

Each element of the component related to the internal disturbance separated at this time corresponds to each failure point occurring in the dynamic system 10. Therefore, the diagnostic means 36
From each element of the component related to the separated internal disturbance, the occurrence of the fault in the dynamic system 10 and the location of the fault can be identified.

In this case, the diagnosing means 36 has a memory unit 40 in which a reference value for failure detection corresponding to each element of the component related to the internal disturbance is stored in advance, and the separated component related to the internal disturbance. It is preferable that the element is compared with a corresponding reference value for failure detection, and a failure specifying unit 38 that specifies a failure point of the dynamic system 10 is included.

The present invention is constructed as described above, and its operation will be described below.

First, the disturbance estimating means 32 will be described.

When a failure occurs in the dynamic system 10 to be diagnosed, the internal state quantity of the diagnostic object shows a response different from that in the normal state. From a different point of view, this response can be considered to be the response at the time of normal operation with some kind of disturbance corresponding to the failure. This disturbance is subject to diagnosis 1
It is generated not internally from 0 but from inside. The disturbance estimating means 32 estimates this internal disturbance caused by the failure as a total disturbance w as the sum with the external disturbance.

The disturbance estimation principle of the disturbance estimation means 32 will be described below.

First, it is assumed that the dynamic system 10 is expressed by the following equation of state.

[0043]

[Equation 1] Here, x (t) is the dynamic system 10 to be diagnosed.
Internal state vector of u, u (t) is a control input vector,
y (t) represents a control output vector, and d (t) represents an external disturbance vector. Further, the matrices A, B, and C are constant matrices (parameters of the system constituting the diagnosis target) determined by the structure of the diagnosis target.

Therefore, the equation 1 is expressed by the following equation.

[0045]

[Equation 2] The diagnosis target 10 at the time of failure can be expressed equivalently using the fluctuations of the matrices A and B (parameter fluctuations). That is, if the matrix A fluctuates by ΔA (t) and the matrix B fluctuates by ΔB (t) depending on the failure, the diagnostic target after the failure, that is, the dynamic system 10 after the failure is written as follows. To be done.

[0046]

[Equation 3] However, the Dw (t) is expressed by the following equation.

[0047]

[Equation 4] Here, the matrix D represents in which path of the diagnosis target 10 the disturbance occurs due to the failure, and is set corresponding to the failure assumed to be the intrusion path of the external disturbance.

The above equations (3) and (4) can be expressed by a general determinant as follows.

[0049]

[Equation 5]

[0050]

[Equation 6] In this way, the response of the state vector of the diagnosis target at the time of failure can be represented by the sum of the response at normal time {Ax (t) + Bu (t)} and the disturbance {Dw (t)}. The disturbance estimating means 32 is configured to estimate the disturbance Dw (t).

The disturbance estimation at this time is performed as follows.

First, in the first step, an expanded system of the diagnosis object 10 including the disturbance w (t) in the state is constructed. For that purpose, the following assumption is made for w (t), and w (t) is added as a state to be diagnosed.

[0053]

[Equation 7] As a result, the expanded system of the system including w (t) in the state is expressed by the following equation.

[0054]

[Equation 8] Then, in the second step, the state of [Equation 8] [xTw
T] T is estimated and calculated using the conventional linear control theory.

The assumption of the equation 7 means that the disturbance 100, which originally changes continuously as shown in FIG. 2, is approximated in a stepwise manner as indicated by 110 in the figure. The smaller the width of this staircase, the better the accuracy of approximation. This width corresponds to the disturbance estimation time of the disturbance estimation means 32,
In practice, the estimated time can be made very short compared to the speed at which the disturbance changes, so this approximation is sufficiently practical.

As described above, if the expanded system is observable, the disturbance estimating means 32 estimates the unmeasurable state and the disturbance corresponding to the failure even when all the states to be diagnosed cannot be measured. You can

Next, the correlation calculating means 34 will be described.

As described above, the total disturbance vector w (t) estimated by the disturbance estimating means 32 penetrates into the diagnosis target from the outside regardless of whether the disturbance is the internal disturbance caused by the failure of the diagnosis target 10 or whether the disturbance is normal or not. The external disturbance d that comes in is mixed.

Although the external disturbance d is an irregular signal, it has a characteristic that its value becomes zero when the average over a fixed time is taken. In the present invention, focusing on the characteristics of the external disturbance d, for separating the components (parameter fluctuation components ΔA, ΔB) related to the internal disturbance caused by the failure from the estimated total disturbance w (t). Calculate. A typical method for such calculation is the least squares method.

First, the following equation is defined from the equation (4).

[0061]

[112] Then, using the method of least squares, from N data,

[0062]

[Equation 113]

Minimize

[Outside 59] Ask for.

This is

[Outside 60] The partial differential equation is obtained with zero and is expressed by the following equation.

[0065]

[Equation 114] Furthermore, if this is converted into a recurrence formula,

[0066]

[Equation 115] Then, it becomes possible to successively estimate the parameter variations ΔA and ΔB.

Next, a specific example of such cross-correlation calculation will be described by taking the case where there is no correlation with each element of the vector ζ as an example. In this case, the total disturbance w (t),
By obtaining the correlation with the internal state quantity x (t) having no correlation with the external disturbance, the component related to the internal disturbance can be separated from the total disturbance w (t).

For example, the estimated value i of the disturbance vector w (t) is i.
Taking the second element, this is given by

[0069]

[Equation 9] As is clear from this equation 9, the i-th element of the estimated disturbance is the linear combination of the quantity Δaij representing the failure of the diagnosis object 10 and the respective elements x1, x2 ... Of the internal state quantity vector x (t). It consists of

Therefore, in order to extract the failure amount from the i-th element of the estimated disturbance, the i-th element of the estimated disturbance and the internal state quantity of the diagnosis object 10 are cross-correlated. At this time, as the internal state quantity of the diagnostic target used for the cross-correlation calculation, a value directly measured by a sensor or the like provided inside the dynamic system 10 may be used, and as described above, the disturbance estimation means 32 may be used. The estimated value may be used.

First, let us consider a case where the cross-correlation function between the i-th element of the estimated disturbance and the j-th element xj of x shown in the above equation 9 is obtained. At this time, the cross-correlation function is defined by the following equation.

[0072]

[Equation 10] Here, as described above, it is assumed that there is no correlation between each element of the external disturbance vector d (t) and each element of the internal state quantity vector x due to the action of the controller 12.

The cross-correlation function and auto-correlation function are expressed by the following equations.

[0074]

[Equation 11] Each value in this equation 11 is expressed by the following equation.

[0075]

[Equation 12] The calculation using the equation 9 assumes that the internal state quantity xj is directly measured using a sensor or the like. On the other hand, when xj is not directly measured, the cross-correlation function may be obtained based on the following equation using the estimated value from the disturbance estimation means 32.

[0076]

[Equation 13]

[0077]

[Equation 14] Since the disturbance estimating means 32 can estimate the internal state quantity xj of the diagnosis target without error regardless of the presence or absence of a failure or external disturbance, the internal state can be directly estimated even if the correlations of the equations 13 and 14 are taken. It is possible to obtain a result that is almost the same as when measured.

Next, the diagnostic means 36 will be described.

The diagnosing means 36 is a correlation function Ci which is a component related to the internal disturbance calculated by the correlation calculating means 34.
The occurrence of a failure is detected and the location of the failure is specified by j. That is, the parameter variation Δaji can be detected by dividing the correlation Cij by the state autocorrelation vxj and normalizing it. When organized,

[0080]

[Equation 116] Becomes The above equation is equivalent to the case where there is no correlation between the elements of the vector ζ in the equation 114 when the average value is zero.

For example, the dynamic system 10 to be diagnosed
There are components I and II in the components of, and the parameter expressing the components I is at the 1st row and 1st column of the A matrix of the dynamic system expressed by the equation 1 of the state equation,
It is assumed that the parameters expressing the constituent element II are at the 1st row and 1st column and the 1st row and 2nd column of the A matrix. At this time, if the correlation function C12, that is, Δa12 has a value, it is immediately determined that the component II has a failure. Also, there is no value in C12,
If C11, that is, Δa11 has a value, it is determined that the component I is in failure.

As described above, according to the present invention, it is possible to surely detect a failure that occurs in each constituent element of the dynamic system 10 and to accurately specify the failure location.

By the way, in the prior art which obtains the quantity corresponding to the failure from the residual between the state estimated by the observer and the directly measured state, this residual does not have a simple relationship with the failure (the above-mentioned equation 9). It's not such a simple relationship)
The failure cannot be specified by a simple calculation such as correlation calculation.

The present invention can take the following forms depending on the design of the disturbance estimating means 32.

In the first mode, all the internal state quantities x (t) of the diagnostic object 10 can be measured by using a sensor or the like, and the disturbance estimating means 32 estimates only the total disturbance vector w (t). This is the case when it is formed. In this case, since the order of the disturbance estimating means 32 is only the order of the disturbance, it has the advantages of the simplest configuration and the highest fault detection accuracy.

The second mode is a case where a part of the internal state quantity x (t) of the diagnostic object 10 cannot be measured or can be estimated by the disturbance estimating means without the measurement. In these cases, the disturbance estimation means 32 is configured to estimate and calculate the total disturbance vector w (t) and the internal state quantity that cannot be measured or is not measured.

In this case, since it is not necessary to measure a part of the internal state quantity of the diagnosis object 10, the number of sensors can be reduced. Since the disturbance estimating means 32 also estimates the internal state quantity that cannot be measured together with the disturbance quantity, the correlation can be used for this estimated value to measure the failure almost as if all internal states were measured.

In the third mode, when a part of the internal state quantity x (t) of the diagnosis object 10 cannot be measured, the disturbance estimation means 32.
Is a case in which the mutual disturbance vector w (t) and all internal state quantities x (t) including unmeasurable internal state quantities are formed to be estimated and calculated.

Also in this case, similarly to the second mode, the number of sensors can be reduced, and the failure can be specified almost in the same manner as the case where all the internal state quantities are measured.

There is also an advantage that the design of the disturbance estimating means 32 is slightly simpler than the second countermeasure.

The calculation in the case of adopting the first to third modes will be described in detail below. First, as described in the first aspect, it is a case where all the internal state quantities of the diagnosis target 10 are measured.

In this case, when the average of the disturbance vector of the above equation 9 is taken, this arithmetic expression is expressed by the following equation.

[0093]

[Equation 52] Therefore, when all the state quantities of the diagnosis target can be measured, the value of the cross-correlation function between the estimated disturbance vector and the state quantity xj is given by the following equation.

[0094]

[Equation 53] Here, from the assumption that there is no correlation between the state quantities to be diagnosed, each term in the above equation 53 is expressed by the following equation.

[0095]

[Equation 54] Further, since it is assumed that there is no correlation between the state xj to be diagnosed and the external disturbance, the relationship shown by the following equation holds.

[0096]

[Equation 55] As a result, the value of the correlation function is finally expressed by the following equation.

[0097]

[Equation 56]

[0098]

[Equation 57] When the cross-correlation is calculated using the estimated value from the disturbance estimating means 32 as in the second and third modes, the calculation formula is represented by the following formula.

[0099]

[Equation 58] Here, since it is assumed that there is no correlation between the state quantities to be diagnosed, each term in the above equation 58 is expressed by the following equation.

[0100]

[Equation 59] Therefore, the value of the cross correlation is expressed by the following equation.

[0101]

[Equation 60]

[0102]

[Equation 61] By using the correlation function thus obtained, the occurrence of a failure and the location of the failure can be similarly identified.

(Effect of the Invention) As described above, according to the present invention, the disturbance detecting means for estimating the disturbance caused by the failure as one state of the diagnosis object is used as the failure detecting means. There is no trade-off between the failure detection speed and the failure detection sensitivity that is seen in the above, and the failure detection speed and the failure detection sensitivity are significantly improved compared to the conventional technology.

Further, since the relationship between the estimated disturbance and the failure point can be expressed by a simple mathematical expression, a simple calculation of obtaining the correlation with the internal state quantity of the diagnosis object
It is possible to easily separate the disturbance from the outside and the internal disturbance caused by the failure to identify the failure location.

At this time, as the internal state quantity used for the correlation calculation, not only the one directly measured by the sensor but also the internal state quantity estimated by the disturbance estimating means at the same time as the disturbance can be used, if necessary. As a result, detailed failure detection can be performed even if not all internal state quantities to be diagnosed have been measured.

As described above, according to the present invention, it is not necessary to use a large number of observers depending on the failure as in the prior art, and the diagnosis can be made without measuring all the states of the diagnosis object as necessary. The target failure location can be specified with high sensitivity.

Furthermore, the calculation result used to identify the failure is
Since it corresponds to the failure scale of the diagnosis target, it can be used as information for parameter identification after failure and control system redesign.

(2) Invention Concerning Tire Pressure Diagnosis Device [Explanation of the Invention] Next, a tire pressure diagnosis device of the present invention (claim 5) constructed by using the principle of the above-mentioned dynamic system diagnosis device. The details will be described.

(Structure and Operation) The present invention relates to a tire air pressure diagnostic device for diagnosing a tire air pressure condition of a dynamic system composed of a suspension and wheels, which is based on the internal state quantity vector of the dynamic system. Disturbance estimating means for estimating a total disturbance vector as a sum of an internal disturbance vector generated by a change in tire air pressure in the system and an external disturbance vector input to the dynamic system from the road surface, and the estimated said Correlation calculation means for calculating a cross-correlation between the total external disturbance and the internal state quantity vector, and separating a component related to the internal external disturbance from the element of the total external disturbance vector, and a component related to the separated internal external disturbance And diagnostic means for determining a tire pressure state of the dynamic system.

Here, the correlation calculating means calculates the cross-correlation between the element of the total disturbance vector and the element of the internal state quantity vector having no correlation with the external disturbance, and the element of the total disturbance vector. Can be formed to separate components associated with internal disturbances from.

Further, the correlation calculating means determines, as the cross-correlation calculation, the temporal sum of squares of the errors between the total disturbance vector and the product of the internal disturbance vector and the internal state quantity vector. As described above, the calculation of the direction vector of the total disturbance using the internal state quantity vector as a base vector is performed, and the component related to the internal disturbance can be separated from the element of the total disturbance vector.

Further, the correlation calculating means calculates a cross-correlation between a plurality of elements of the estimated total disturbance vector and an element having no correlation with the external disturbance of the internal state quantity vector to calculate the total disturbance. The diagnostic means is formed to separate each element of the component related to the internal disturbance from the plurality of elements of the vector, and the diagnostic means identifies the diagnostic point of the dynamic system from each element of the separated component related to the internal disturbance. And can be configured to make that diagnosis.

The tire pressure diagnostic device of the present invention will be described in more detail based on the dynamic system diagnostic device of FIG. 1 described above.

The dynamic system 10 to be diagnosed in the present invention is a system including a suspension and wheels. When the tire air pressure changes, each state quantity of the system shows a different response from the normal state.
From a different point of view, it can be considered that this response is a response in the normal state with an internal disturbance corresponding to a change in air pressure. Therefore, if this disturbance, that is, the total disturbance vector is estimated by using the disturbance estimating means 32, it is possible to detect the fluctuation of the tire air pressure.

Now, it is assumed that the dynamic system 10 including the suspension and the wheels is expressed by the following equation of state.

[0116]

[Equation 20] Here, x is an internal state quantity vector of the system 10. u is a control input, and corresponds to the operation amount when the suspension is an active suspension. Further, y is a control output vector (internal state quantity vector) directly detected and output from the sensor of the system 10. d represents a road surface disturbance received from the road surface. The matrices A, B, and C are constant matrices determined by the physical parameters of the system 10.

Hereinafter, the suspensions constituting the dynamic system 10 are divided into those in which the input u is present in the system 10 as in the active suspension and those in which the input u is not present in the system 10 as in the conventional suspension. The present invention will be described below. When system input is present In this case, fluctuations in tire air pressure, etc.
It is replaced with zero physical parameter variation. This variation can be expressed using the variation of matrix A. That is, assuming that the matrix A fluctuates by ΔA due to fluctuations in air pressure or the like, the system after fluctuation is expressed as follows.

[0118]

[Equation 21] Here, the Dw is represented by the following equation.

[0119]

[Equation 22] From this number 22, ΔA can be changed by the change of the tire air pressure.
It can be seen that a new disturbance of x occurs. Also, the D
Is a matrix composed of 1 and 0 elements, and is set according to the intrusion route of the road surface disturbance and the source of the disturbance generated by the fluctuation of the parameters of the system 10.

As described above, the response of the state vector at the time of fluctuation can be represented by the sum of the response at the normal time and the disturbance Dw (t). The disturbance estimating means 32 will be configured to estimate this disturbance w.

The disturbance estimating means 32 is constructed by being divided into the following two steps.

In the first step, an expanded system of the system including the disturbance w (t) in the state is constructed. Therefore, w
The following assumption is made in (t), and w (t) is added as the state of the diagnosis target 10.

[0123]

[Equation 23] As a result, the expanded system of the system 10 including the disturbance w in the state is expressed by the following equation.

[0124]

[Equation 24]

[0125]

[Equation 25] Then, in the second step, the state [x
The estimation means 32 for estimating TwT] T is configured by using the conventional linear control theory. The disturbance w is estimated by the estimating means 32 designed in this way.

It has already been explained that the assumption of the above equation 23 means that the continuously changing disturbance w (t) is approximated in a stepwise manner as shown in FIG.

Disturbance estimating means 3 configured in this way
2 estimates the state that cannot be measured even when all the state quantity vectors x of the system 10 cannot be measured,
It is possible to estimate a disturbance corresponding to a change in air pressure or the like.

As can be seen from the above equation 22, the estimating means 3
The disturbance estimated by 2 is the external disturbance d received from the road surface,
It is the sum of the internal disturbance ΔAx generated inside the system 10 due to fluctuations in tire air pressure and the like. Expression 22 is expressed as a vector, and for example, taking the first element of this vector, this is expressed by the following equation.

[0129]

[Equation 26] Hereinafter, a specific description will be given based on the equation 26.

In the above equation 26, it is assumed that Δa11 is an element due to a change in tire air pressure, and the other elements are caused by a change other than the tire air pressure. In this case, it can be seen that the internal disturbance generated by the fluctuation of the tire air pressure depends on the state quantity x1. That is,
In this case, the system 10 may be affected by variations in tire air pressure.
The state quantity that influences is x1.

Therefore, the disturbance [D
In order to remove the external disturbance d1 such as the road surface disturbance from w] 1 and detect only the component related to the internal disturbance due to the tire air pressure fluctuation, the estimated disturbance [Dw] 1 and state quantity x1
Calculate the cross-correlation with. This calculation is performed by the correlation calculation means 34.
Will be held in. The cross-correlation calculated at this time is C ([D
w] 1, x1), this results in the values given by:

[0132]

[Equation 27] Here, it is described above that the terms of external disturbances such as road surface disturbances and other internal disturbances due to changes other than air pressure have no correlation with the state quantity x1. Therefore, only the internal disturbance corresponding to the tire air pressure fluctuation can be extracted from the estimated disturbance w by performing the correlation calculation of the equation 27.

The correlation function calculated in this way has a value corresponding to the frequency component of the internal disturbance generated only by the fluctuation of the spring constant of the tire from among the various frequency components included in the estimated disturbance w. . Therefore, the variation of the spring constant can be detected from the calculated value of this correlation function.

The cross-correlation function expressed by the equation (27) can be expressed by the product of the term Δa11 representing the fluctuation of the spring constant and the autocorrelation function of the state quantity expressed by the following equation. Here, the diagnostic means 3
6 can quantitatively detect the variation amount Δa11 of the spring constant by dividing the cross-correlation function by the autocorrelation function of the state quantity.

[0135]

[Equation 28] Then, the diagnosing means 36 determines the variation amount Δ of the obtained spring constant.
When a11 reaches the variation amount of the spring constant corresponding to the tire air pressure that should be determined to be abnormal, the air pressure is determined to be abnormal. Case where Input is not Present in System 10 Next, the case where the suspension is not the active suspension and there is no input in the system 10 as in the conventional suspension will be described.

In this case, the dynamic system 10 including the suspension and the wheels is expressed by the following state equation.

[0137]

[Equation 29] Here, x is a state quantity vector of the system 10, y is an output vector directly measured from a sensor of the system, and d is a road surface disturbance received from the road surface.
The matrices A, B, and C are constant matrices determined by the physical parameters of the system 10. The difference from the active suspension is that there is no input u to the system 10 seen in the equation (20).

If the matrix A fluctuates by ΔA due to fluctuations in air pressure or the like, the system after fluctuation is represented by the following equation.

[0139]

[Equation 30] Here, the Dw is represented by the following equation.

[0140]

[Equation 31] From the equations 30 and 31, the extended system of the system 10 including the disturbance w is expressed by the following equation.

[0141]

[Equation 32]

[0142]

[Expression 33] The subsequent configuration and operation including the configuration of the disturbance observer 32 are exactly the same as those of the active suspension except the input u.

Thus, tire pressure abnormalities can be detected even when there is no input to the system 10 for a conventional suspension.

Comparison with Prior Art Conventionally, there are roughly two types of means for detecting abnormality in tire air pressure.

The first means is to mount a pressure sensor on the tire (rotation) side and a wireless type signal transmission device for transmitting a signal from the pressure sensor to the vehicle body side, and This is for judging the abnormality of the tire air pressure.

However, in this prior art, special processing for embedding the pressure sensor in the tire was required, and measures such as air leakage from the sensor mounting portion were also required.
In addition, the pressure sensor and signal transmission device mounted on the rotating tires are exposed to the severe environment where they are directly exposed to vibration, shock, centrifugal force, temperature change, cold water, snow, etc. It was difficult to maintain sex.

The second means is to detect an abnormal tire pressure without directly mounting the sensor on the tire.

Examples of such a method include a method of measuring the distance between the axles of the four wheels and the ground to judge that the air pressure of the tire in the short distance portion is abnormal, or detecting the number of rotations of the four wheels to rotate the wheels. There are known a method of considering the air pressure of a tire having a high speed as an abnormality, a method of detecting an air pressure abnormality of a tire by using acceleration signals above and below the axle, and the like.

For example, in the proposal of Japanese Patent Laid-Open No. 63-22707, the abnormality of the air pressure is detected by utilizing the phenomenon that when the tire air pressure changes, the acceleration frequency (acceleration spectrum) received by the vehicle body also changes. is doing.

FIG. 17 shows a block diagram of this conventional device. In this conventional device, the acceleration converter 1 detects the vertical acceleration of a vehicle wheel to be detected, and the detected acceleration is passed through an amplifier 2 to each filter 3
It outputs to a and 3b. Each filter 3a, 3b
The level converters 4a and 4b are formed as bandpass filters having different pass frequency bands, and the level converters 4a and 4b include the filters 3a and 3b.
The respective effective values of the output signals of b are converted into signals of DC voltage levels V1 and V2 and output to the divider 5. Divider 5
Divides the input voltages V1 and V2, and the value V =
V2 / V1 is output to the comparator 6.

Here, the vertical acceleration level of the wheel obtained from the acceleration converter 1 has a large value in accordance with a certain frequency band, and that frequency band is irrelevant to the road condition and the traveling speed, and It fluctuates depending on the air pressure and shifts to the low frequency side especially when the air pressure becomes low.

Therefore, the filter 3a is formed so as to pass only the signal in the frequency band when the air pressure is normal, and the filter 3b is formed so as to pass only the signal in the frequency band of the air pressure which should be determined to be abnormal. . Thus, when the tire pressure is normal, the voltage V1 corresponding to the effective value of the output of the filter 3a is large, and the value of the voltage V2 corresponding to the effective value of the filter 3b is small. Therefore, the output V obtained by the divider 5 has a small value.

On the other hand, when the air pressure decreases and the frequency band of the acceleration level approaches the frequency band set by the filter 3b, the value of V2 becomes small and the value of the voltage V1 becomes large. Therefore, the output V obtained by the divider 5 has a large value.

Therefore, the voltage V output from the divider 5
Is compared with the reference value set by the setting device 7, and when V is larger than the reference value, a signal is output from the comparator 6 and the alarm device 8
To operate.

As described above, this prior art has the advantage that the air pressure can be determined only by the signal from one wheel, but when the air pressure fluctuation is small,
There was a problem that the frequency band of the acceleration level hardly changed from the normal value and the air pressure detection accuracy was low.

FIG. 18 shows the spectrum of the vertical acceleration of the wheel when the tire pressure is normal, and the frequency in the area of arrow A where the acceleration level is maximum is around 10 hertz.

On the other hand, FIG. 19 shows the vertical acceleration spectrum of the wheel when the air pressure decreases and the spring constant of the tire decreases by 20%. The frequency at which the acceleration level becomes maximum is around 10 hertz indicated by arrow B. Becomes

As is apparent from FIGS. 18 and 19,
When the fluctuation of the air pressure is small, the frequency band of the acceleration level hardly moves, so it is impossible to detect this, and there is a problem that this abnormality cannot be detected until a large fluctuation of the air pressure occurs.

On the other hand, in the tire pressure diagnostic device of the present invention, unlike the prior art, it is not necessary to mount the pressure sensor and the signal transmission device directly on the tire side, and the reliability of the pressure sensor and the signal transmission device is improved. There are no problems such as durability.

In addition to this, in the present invention, as the air pressure detecting means, the disturbance estimating means 32 for estimating the disturbance caused by the fluctuation of the air pressure as one state of the diagnosis object is used.
Moreover, the relationship between the estimated disturbance and the air pressure can be expressed by a simple mathematical expression. Therefore, the value of the air pressure can be easily detected and the abnormality can be determined by a simple calculation that correlates with the state of the diagnosis target.

In particular, according to the present invention, compared to the prior art using the acceleration converter, the value can be reliably detected even when the change in air pressure is small.

(Effects of the Invention) As described above, according to the present invention, as the air pressure detecting means, the disturbance estimating means for estimating the disturbance caused by the fluctuation of the air pressure as one state of the diagnosis object is used and further estimated. With the configuration in which the relationship between the disturbance and the air pressure is obtained by a simple correlation calculation, there is an effect that it is possible to obtain a tire pressure diagnostic device that can accurately obtain the variation in the tire air pressure with a simple configuration. is there.

Further, when the device of the present invention is applied to a pneumatic pressure diagnosis for a tire mounted on an active suspension, the sensor used for controlling the suspension, such as an acceleration detector, is used as it is, and the pneumatic pressure of the tire is changed. Since the diagnosis can be performed, there is also an effect that it is not necessary to provide a new acceleration sensor for detecting the air pressure abnormality.

When diagnosing a tire mounted on a conventional suspension that is not active, the tire air pressure is diagnosed only by providing an acceleration sensor that detects vertical accelerations above and below the spring of the suspension. be able to.

In addition to this, according to the present invention, the reference value of the tire air pressure fluctuation amount which should be judged to be abnormal is provided, and the tire air pressure abnormality judgment is made by comparing this with the fluctuation amount obtained by the disturbance estimating means. You can also do

Further, the diagnosis means includes an autocorrelation calculation unit for calculating the autocorrelation of an element having no correlation with the external disturbance of the internal state quantity vector, and the tire correlation is calculated based on the cross-correlation value and the autocorrelation value. If it is formed so as to diagnose the state of the air pressure, it is possible to accurately detect the fluctuation of the tire air pressure as the amount of fluctuation of the spring constant as it is, so it can be used as a pneumatic pressure monitor that constantly displays the abnormality of the pneumatic pressure to the driver. Further, if this information is used for the control law of the active suspension or for the suspension that can change the attenuator constant, it is possible to realize an appropriate ride comfort corresponding to the fluctuation of the air pressure.

(3) Invention relating to vehicle body weight variation detecting apparatus Next, a vehicle body weight variation detecting apparatus (claim 11) according to the present invention, which is constructed by using the principle of the above-mentioned dynamic system diagnostic apparatus, will be described. .

The present invention relates to a vehicle body weight variation detecting device for diagnosing a variation in vehicle body weight of a dynamic system composed of a suspension and wheels, in which the vehicle body is moved in the dynamic system based on the internal state quantity vector of the dynamic system. Disturbance estimating means for estimating the total disturbance vector as the sum of the internal disturbance vector generated by the fluctuation of the weight and the external disturbance vector input to the dynamic system from the road surface, and the estimated total disturbance vector, Correlation calculation means for calculating a cross-correlation with the internal state quantity vector and separating a component related to the internal disturbance from the total disturbance vector, and a vehicle body of the dynamic system from the separated components related to the internal disturbance. And a detection unit that detects a change in weight.

Here, the correlation calculating means calculates the cross-correlation between the element of the total disturbance vector and the element of the internal state quantity vector having no correlation with the external disturbance, and the element of the total disturbance vector. Can be formed to separate components associated with internal disturbances from.

Further, the correlation calculating means determines, as the cross-correlation calculation, the temporal sum of the squares of the errors between the total disturbance vector and the product of the internal disturbance vector and the internal state quantity vector. As described above, the calculation of the direction vector of the total disturbance using the internal state quantity vector as a base vector is performed, and the component related to the internal disturbance can be separated from the element of the total disturbance vector.

In this way, the apparatus of the embodiment can diagnose the fluctuation of the vehicle body weight of the dynamic system by using the method of the dynamic system diagnostic apparatus described above.

(4) Another Invention Concerning Dynamic System Diagnosis Apparatus Next, an invention of a dynamic system diagnosis apparatus that is effective when there is a correlation between an internal state quantity and an external disturbance (claim 9). )
Will be explained.

The present invention is a dynamic system diagnostic apparatus for detecting a fault in a dynamic system, wherein the external disturbance vector and the internal disturbance vector of the dynamic system are summed based on the internal state quantity vector of the dynamic system. Disturbance estimating means for estimating a total disturbance vector as, a correction value storage means for storing the correlation between the internal state quantity vector in a predetermined reference state and the external disturbance vector as a correction value, and the estimated total disturbance vector And a correlation calculation means for calculating a cross-correlation with the internal state quantity vector, and the cross-correlation calculated by the correlation calculation means based on the internal state quantity vector and the correction value to correct the total correlation. Correlation compensation means that separates components related to internal disturbance from the disturbance vector without being affected by external disturbance, and the separated internal disturbance From the components communicating, wherein the diagnostic means for performing the specified diagnosis, the corresponding part of the dynamic system, and performs a diagnosis of a dynamic system.

Here, the correlation compensating means corrects the cross-correlation calculated by the correlation calculating means based on the internal state quantity vector and the correction value.
A component related to the internal disturbance can be formed so as to be separated from the elements of the total disturbance vector without being affected by the external disturbance.

Here, as the reference state, it may be assumed that the dynamic system is normally operating under a predetermined condition. For example, when diagnosing the tire air pressure as described below, a state in which the tire air pressure has a certain value under normal conditions may be set as the reference state.

With the above configuration, when the correlation calculation means calculates the cross-correlation between the element of the total disturbance vector and the element of the internal state vector, the correlation correction means is set in advance in the correction storage means. By reading out the correction value that is present and correcting the cross-correlation, it is possible to separate the component related to the internal disturbance from the element of the total disturbance vector without being affected by the disturbance.

In this way, the dynamic system can be accurately diagnosed without being affected by external disturbance.

Further, by using such a method, the tire pressure of the dynamic system can be diagnosed and the fluctuation of the vehicle body weight can be detected.

Further, another invention (Claim 10) according to the present application is a tire air pressure diagnostic apparatus for diagnosing a tire air pressure state of a dynamic system including a suspension and wheels, wherein the internal state of the dynamic system is Disturbance estimating means for estimating a total disturbance vector as a sum of an internal disturbance vector generated by a change in tire air pressure in the dynamic system and an external disturbance vector input from the road surface to the dynamic system based on the quantity vector A correction value storage means for storing the correlation between the internal state quantity vector in a predetermined reference state and the external disturbance vector as a correction value, the estimated total disturbance vector, and the cross-correlation between the internal state quantity vector. Correlation calculating means for calculating, and mutual calculation for calculating by the correlation calculating means based on the internal state quantity vector and the correction value. Correlation compensation means for separating the component related to the internal disturbance from the total disturbance vector without being affected by the external disturbance by correcting the relationship, and the dynamic system based on the separated component related to the internal disturbance. And a diagnostic means for determining the tire air pressure state.

With the above configuration, the tire pressure condition of the dynamic system can be diagnosed without being affected by external disturbance.

Further, another invention according to the present application (Claim 13) is a vehicle body weight variation detecting device for diagnosing a variation in vehicle body weight of a dynamic system including a suspension and wheels, wherein Disturbance estimating means for estimating a total disturbance vector as a sum of an internal disturbance vector generated by a change in vehicle body weight in the dynamic system and an external disturbance vector input from the road surface to the dynamic system based on the state quantity vector A correction value storage means for storing the correlation between the internal state quantity vector in a predetermined reference state and the external disturbance vector as a correction value, the estimated total disturbance vector, and the cross-correlation between the internal state quantity vector. Correlation calculating means for calculating, and correcting the cross-correlation calculated by the correlation calculating section based on the internal state quantity vector and the correction value. Thereby, the correlation compensation means for separating the component related to the internal disturbance from the total disturbance vector without being affected by the external disturbance, and the component related to the separated internal disturbance, And a detecting means for detecting a change.

Here, as the reference state, it may be assumed that the vehicle weight has a certain value.

With the above-mentioned structure, it is possible to detect the fluctuation of the vehicle body weight without being affected by the external disturbance.

[0184]

【Example】

(First Embodiment) Next, an embodiment in which the present invention (claim 1) is applied to a failure detection of an active suspension control system of an automobile will be described as a first embodiment.

FIG. 3 shows a specific example of the dynamic system 10 which is the object of diagnosis in this embodiment. The dynamic system 10 of the embodiment represents a vibration model of an automobile single wheel suspension. In the figure, the wheel 41 is represented by the unsprung mass portion represented by the parameter m1 and the spring portion of the wheel (tire) represented by the spring constant k1. Further, 42 represents a vehicle body part having a sprung mass constant m2, and 46
Represents a gas spring having a spring constant k2, 48 represents an attenuator having a damper constant Dm, and 56 represents a road surface displacement represented by a variable x0. Further, 52 is the unsprung displacement represented by the variable x1, 50 is the unsprung range represented by the variable x2, 5
Reference numeral 4 represents a relative displacement (x1-x2) represented by a variable y, and 44 represents a control force generator that generates an active control force f required for control from an operation amount u output from a controller that controls the suspension.

From the figure, the state equation of Equation 1 is expressed as follows.

[0187]

[Equation 15] Here, T represents the response time of the control force generator 44, that is, the time delay between the manipulated variable u and the active control force f.

A and b are expressed by the following equations.

A = k1 / m1 + k2 / m1 + k2 / m2 b = 1 / m1 + 1 / m2 In this embodiment, the road surface displacement 56 is the external disturbance x0.

In this embodiment, as the failure, the tire pressure is abnormal, the gas spring 46 is abnormally pressured, and the attenuator 48 is abnormal.
Assume a failure. Then, these failures are treated as fluctuations of the parameters k1, k2, and Dm, respectively. Then, the disturbance observer 32 is configured based on the vibration model 10.

That is, in the equation 15, the parameters k1, k2 and Dm are in the second and fourth elements on the right side. Therefore, the invasion routes of internal disturbances caused by assumed failures are also set at these two locations. The setting is performed based on the matrix D represented by the above-mentioned mathematical expression 3. in this case,
It may be set as in the following equation.

[0192]

[Equation 16] Using this D, the expansion system shown in the above-mentioned equation 8 is created, and FIG.
The disturbance observer 32 of FIG.

In this embodiment, the disturbance observer 32 is divided into the following three modes.

The first mode is the suspension model 10
In this case, the sprung displacement x2, the sprung speed, the relative displacement y, the relative speed, and the active control force f can be measured. Disturbance observer 32 used in such a case
Does not need to estimate the internal state quantity. The sprung speed and the relative speed are expressed by the following equations.

[0195]

[Equation 17] The second mode is a case where only the sprung displacement x2 and the relative displacement y can be measured, and the disturbance observer 32 estimates other unmeasured internal state quantities (here, the sprung velocity and the relative velocity).

The third mode is a case where only the sprung displacement x2 and the relative displacement y can be measured, and the disturbance observer 32 estimates all internal state quantities including these.

In the following, the disturbance observer 32 used for each of the above modes and the diagnostic device 3 constructed using the disturbance observer 32.
0 will be described in detail.

First Mode FIG. 4 shows a diagnostic device 3 used as the first mode .
0 block diagram is shown. In the figure, 10 represents the dynamic system (single-wheel suspension) shown in FIG. 3, 12 represents a controller controlling the suspension, and 14 represents a controller 1 for controlling the suspension.
2 represents an operation amount output from 2, and 16 represents all internal state quantities of the suspension measured using a sensor (not shown). Here, since all of the internal state quantities are measured using the sensor, the disturbance observer 32 needs to estimate a part or all of the internal state quantities as in the second and third aspects described later. There is no.

The disturbance observer 32 of the embodiment catches an abnormality in the tire air pressure, an abnormality in the pressure of the gas spring and a failure of the attenuator in the diagnosis object 10 as an internal disturbance of the dynamic system 10 to be diagnosed. Then, the total disturbance vector w as the sum of the external disturbance vector and the internal disturbance vector of the dynamic system 10 is estimated and calculated from the manipulated variable 14 of the controller 12 and all the measured state quantities 16 of the suspension 10. , And outputs to the correlation calculation unit 34.

The correlation calculator 34 calculates the cross-correlation between the estimated value of the total disturbance vector estimated and calculated by the disturbance observer 32 and the sprung displacement, relative displacement and relative velocity which are the internal state quantities of the dynamic system 10. To do. At this time, the internal state quantities such as the sprung displacement, the relative displacement and the relative velocity are factors that have no correlation with the external disturbance. Therefore, by calculating the cross-correlation between these internal state quantities and the estimated value of the total disturbance vector, the influence of road surface displacement that invades as an external disturbance is excluded, and the components related to the internal disturbance generated by the fault are separated. However, the calculation result can be output to the diagnosis unit 36.

The diagnostic section 36 detects the occurrence of a failure in the dynamic system 10 based on the calculation result thus input from the cross-correlation calculation section 34, and at the same time the internal calculation calculated by the correlation calculation section 34 is performed. It is formed so as to identify a specific failure location of the suspension based on the location of the disturbance.

The failure diagnosis device 30 of the embodiment has the above-mentioned structure, and its operation will be described below.

It is assumed that an assumed failure occurs in the vibration model 10 shown in FIG. In this case, the disturbance observer 32 determines, based on the control input f to the suspension 10 and the state of the diagnosis target, the second and fourth elements w2 and w4 of the comprehensive disturbance vector previously set according to the failure.
To estimate. At this time, assuming that the estimated delay of these disturbances is small enough to be ignored, these estimated values are expressed by the following equations.

[0204]

[Equation 18] Here, Δk1 represents the parameter fluctuation caused by the tire air pressure abnormality, Δk2 represents the parameter fluctuation caused by the gas spring pressure abnormality, and ΔDm represents the parameter fluctuation caused by the attenuator failure.

The estimated total disturbance is the above-mentioned parameter variation and suspension state variables (relative displacement y,
It is the sum of the product sum of the relative velocity and the sprung displacement x2) and the road surface displacement x0 added as an external disturbance.

Therefore, the cross-correlation unit 34 and the diagnosis unit 36.
Separates the external disturbance from the internal disturbance and judges the failure from the estimated total disturbance as follows.

First, the estimated value of w2 and the correlation function C21 of the measured value y, and the estimated function of w4 and the correlation function C41 of the measured value y, C41 and w4
The correlation function C42 between the estimated value and the measured value (relative speed) is calculated. Although the road surface displacement is included in the estimated value of w2 as an external disturbance, by taking a correlation with the measured value y,
The influence of road surface displacement that is not correlated with y is removed, and only the failure component is extracted as the value of the correlation function C21, for example.

Next, the failure location is specified from the values of these correlation functions. For example, only the fluctuation of the gas spring appears in the correlation function C41. Therefore, if there is an abnormality in this value, it can be immediately identified as a gas spring failure.

Similarly, if the correlation function C42 is abnormal, it is determined that the attenuator is faulty.

The remaining tire pressure abnormality is the correlation function C21.
It is judged by. C21 has a value due to a failure of the gas spring or an abnormal tire pressure, but since the failure of the gas spring is detected by the determination by the above C41, if there is no failure in the gas spring and there is an abnormality in the value of C21, the tire It can be judged that the air pressure is abnormal. It should be noted that if the gas spring and the tire air pressure abnormality occur at the same time, they cannot be distinguished from each other, but since such a thing can rarely occur in practice, there is no problem.

As described above, in the present embodiment, the relation between the estimated value of the total disturbance and the failure point can be expressed by a simple mathematical expression, and the simple calculation of obtaining the correlation with the state of the diagnosis object facilitates the external It is possible to identify the location of the failure by separating the external disturbance from the internal disturbance generated by the failure.

Second Mode FIG. 5 shows a disturbance observer 32 configured according to the second mode described above, and a diagnostic device 3 configured using the disturbance observer 32.
0 block diagram is shown. The same reference numerals will be given to cases corresponding to the embodiment shown in FIG. 4, and description thereof will be omitted.

In this embodiment, only the sprung displacement x2 and the relative displacement y are detected as the internal state quantities from the dynamic system 10 which constitutes the vibration model of the suspension, and other internal state quantities such as the relative velocity are detected. It represents the case where it is not measured directly.

In such a case, the disturbance observer 32 determines, from the manipulated variable 14 and the measured value 16, the sprung velocity and the relative velocity, which are the unmeasured states of the internal state quantity of the suspension 10 together with the comprehensive disturbance vector. And the active control force are estimated and calculated and output to the correlation calculator 36.

This observer is designed as follows.

First, an extended system is created based on the equations (8), (15) and (16).

[0217]

[Equation 62] Next, in equation 62, the measurable state quantities y and x2 are separated from the measurable state quantities.

[0218]

[Equation 63] For the sake of simplicity, Formula 63 is set as the following formula.

[0219]

[Equation 64] And the state quantity xb that cannot be measured including the total disturbance
Is estimated based on the following equation.

[0220]

[Equation 65] Here, from the equations 64 and 65, the error between the true value xb and its estimated value can be written as the following equation.

[0221]

[Equation 66] Therefore, if the real number matrix G is determined such that the eigenvalues of all the vectors of (A22-GA12) of the equation 66 are all negative, the error represented by the equation 66 converges to zero with time. That is, the estimated value converges to the true value.

The above equation (65) is illustrated in FIG. As shown in the figure, the unmeasureable state quantity xb can be estimated by the control input u and the measurable state quantity xa = [y x2] T.

As described above, the disturbance observer 65 of the embodiment is
Is the total disturbance w based on the control input to the suspension and the measured internal state quantities x2, y of the suspension,
It is configured as a minimum dimensional observer for estimating an unmeasured internal state quantity, for example, a relative velocity. Even if the disturbance observer 32 is configured in this way, the estimated total disturbance is expressed by the above-mentioned equation 18, so that the correlation calculation unit 34 performs the correlation calculation as in the case of the first aspect described above. ,
The diagnosis unit 36 can identify the occurrence of a failure and the location of the failure.

That is, the estimated value of the relative velocity required for the correlation calculation is expressed by the following equation.

[0225]

[Formula 19] Therefore, the relative speed is replaced with the estimated value, and the first
Correlation calculation similar to that of the above embodiment may be performed.

Since the disturbance observer 32 estimates the internal state quantity with almost no error even if a failure occurs, even if the correlation calculation is performed using the estimated value as described above, the correlation is almost the same as in the first mode. You can get the function. After the correlation function is obtained, the failure can be identified in the same manner as in the first aspect, so the description thereof will be omitted here.

As described above, in the present embodiment, the state quantity used for the correlation function is not only that directly measured by a sensor or the like but also the internal state quantity estimated by the disturbance observer 32 at the same time as the total disturbance. be able to.
Therefore, detailed failure detection can be performed without directly measuring all the internal state quantities of the diagnosis target 10.

Third Mode FIG. 6 shows a disturbance observer 32 used according to the above-mentioned third mode, and a diagnostic device 3 constructed using the disturbance observer 32.
0 block diagram is shown.

The difference between this embodiment and the embodiment of the second mode is that the disturbance observer 32 is constructed as a full dimensional observer.

That is, the disturbance observer 32 of the embodiment.
Is configured to estimate all of the total disturbance and the internal state quantity of the diagnosis target 10 by the control input f of the suspension 10 and the internal state quantities x2, y directly measured in the diagnosis target 10.

This observer has the following equations (8), (15) and (1):
From 6, the design is made based on the following equation.

[0232]

[Equation 67] Also at this time, a differential equation relating to the error between the estimated value and the true value similar to the equation 66 is created, and the real number matrix G is determined so that the eigenvalues of all the vectors of this equation are all negative. The larger the absolute value of the eigenvalue, the better, but -300 to-
It is desirable to set it to about 700 rad / s.

In this way, the estimated value shown by the following equation can be obtained from the control input u and the estimated values y and x2.

[0234]

[Equation 68] Since the estimated total disturbance is represented by the above equation 18,
Correlation calculation is performed in the same manner as in the first aspect. For the correlation calculation, the estimated values of the relative displacement and the relative velocity are used, and after the correlation function is obtained, the failure point is specified as in the first aspect.

As described above, in this embodiment, the location of the failure can be specified by a simple calculation. In addition to this, since the disturbance observer 32 used in the embodiment is configured as a full-dimensional observer, there is an advantage that the procedure is simpler than that of the minimum-dimensional observer used in the second mode.

(Second to Sixth Embodiments) Next, embodiments according to the present invention (claim 5) will be described as second to sixth embodiments. Here, the case where the present invention is applied to a dynamic system including an active suspension and wheels is used in the second and third cases.
As an example, a case in which the present invention is applied to a dynamic system including wheels and suspensions including a spring and a damper that have been conventionally used will be described in detail as fourth to sixth examples.

Application to Dynamic System Including Active Suspension (Second and Third Embodiments) (Second Embodiment) FIG. 8 shows a specific example of the dynamic system 10 to be diagnosed in this embodiment. It is shown. The members corresponding to those in the dynamic system shown in FIG.
The description is omitted.

The dynamic system 10 of this embodiment is a system including an active suspension and wheels. In this dynamic system 10, the wheel 41 has a parameter m1.
It is represented by the unsprung mass part represented by and the spring part of the tire represented by the spring constant k1. 42 is the mass on the spring m
2 represents a vehicle body part having 2; 46 represents a gas spring having a spring constant k2; 56 represents road surface displacement represented by variable x0; 52 represents unsprung displacement represented by variable x1;
Represents a sprung displacement represented by a variable x2, and 54 represents a variable y.
Represents the phase displacement (x1-x2). Further, 44 is a control force generator that generates an active control force f required for control from an operation amount u output from the controller 12 that controls the suspension.

Usually, in such an active suspension system, a pressure sensor 60a is provided as an element required for active control of the suspension. The pressure sensor 60a is installed in the control force generator 16 so that the active control force f can be measured. Further, the acceleration sensors 60b, 60 are used to detect abnormal tire pressure.
c is provided, and these acceleration sensors 60b, 6
0c is installed on the upper spring and the lower spring, respectively, so that the acceleration of vertical vibration can be detected.

In the dynamic system 10 as described above, the above-mentioned equation of state expressed by the equation 24 is specifically expressed by the following equation.

[0241]

[Equation 34] FIG. 9 shows a block diagram of the system of this embodiment. The suspension system 10 to be diagnosed here receives the manipulated variable u output from the controller 12 and outputs the vector included in the equation 34.

[0242]

[Outer 1] It has been described above that the system is configured so that the internal state quantity is.

Of the respective elements included in the internal state quantity vector x, the sprung acceleration is directly detected by the acceleration sensor 60b shown in FIG. 8, and the sprung speed is obtained by integrating the sprung acceleration. The relative acceleration is obtained from the difference between the unsprung acceleration detected by the acceleration sensor 60b and the sprung acceleration. The relative velocity is obtained by integrating the relative acceleration. The respective calculation units that perform such calculations are the suspension system 1 of the embodiment.
It shall be always embedded in 0. Therefore, the output of the suspension system 10 is the internal state quantity vector included in the equation 34.

[0244]

[Outside 2] Becomes

The controller 12 of this embodiment receives the internal state quantity vector x thus output as an input, and calculates and outputs the manipulated variable u which is an input signal of the active suspension system 10.

Further, the diagnostic device 30 of the embodiment in which the active suspension system 10 is a diagnostic target includes a disturbance observer 32, a correlation calculator 34, and a diagnostic unit 36.

The disturbance observer 32 receives the output u of the controller 12 and the output x of the active suspension system 10 as input signals, and estimates the fluctuation of the tire spring constant k1 as an internal disturbance generated in the suspension system 10. It is formed to calculate.

The state quantity detection unit 60 extracts, from the output x of the active suspension system 10, the state quantity (in this case, the relative speed) that affects the system 10 by the fluctuation of the tire air pressure.

The correlation calculator 34 is the cross-correlation calculator 7
0, a normalization unit 72, and an autocorrelation calculation unit 74 are included.

Then, the cross-correlation calculating section 70 calculates and outputs the cross-correlation between the total disturbance estimated by the disturbance observer 32 and the relative velocity.

The autocorrelation calculator 74 calculates the autocorrelation function of the relative velocity.

Then, the normalization unit 72 calculates the cross-correlation function calculated by the cross-correlation calculation unit 70 from the auto-correlation calculation unit 7
4 is divided by the autocorrelation function calculated and normalized to extract and output only the variation amount of the tire spring constant.
Output to.

Diagnosis unit 36 includes a memory 76 in which a reference value is stored and an abnormality determination unit 78. Then, the tire spring constant variation amount obtained by the normalizing unit 72 is compared with the reference value of the spring constant variation amount corresponding to the air pressure to be determined to be abnormal stored in the memory 76, and the tire It is formed so as to judge and output an abnormality in the air pressure.

The second embodiment has the above construction, and its operation will be described below.

In the system 10 consisting of the active suspension and the wheels shown in FIG. 8, when the tire air pressure fluctuates and the tire spring constant fluctuates, each state quantity x of the system 10 differs from that under normal air pressure. Will show a response. This response can be regarded as a normal response to which internal disturbance corresponding to a change in air pressure is added.

The disturbance observer 32 receives the output signal x of the suspension system 10 and the manipulated variable u applied to the suspension system 10, and receives the internal disturbance and the external disturbance (external intrusion from the outside of the system such as road surface displacement). Disturbance) and a comprehensive disturbance including and is calculated and output.

Therefore, assuming that the tire air pressure is normal and the spring constant of the tire at that time is k1, the disturbance observer 32 calculates and outputs the external disturbance represented by the following equation.

[0258]

[Equation 35] The external disturbance thus estimated and output as output corresponds to the differential value of the road surface disturbance, and is a random disturbance having no correlation with the state of the system 10.

In such a state, the tire air pressure fluctuates, and its spring constant fluctuates by Δk1 (k1 +
Assuming that Δk1), the disturbance observer 3
2 outputs the total disturbance represented by the following equation.

[0260]

[Equation 36] Therefore, it is necessary to remove external disturbance such as road surface disturbance from the estimated value of the total disturbance estimated and calculated by the disturbance observer 32 and detect only the internal disturbance due to tire air pressure fluctuations. For this purpose, the correlation calculation unit 70 calculates the cross-correlation between the estimated total disturbance and the element of the internal state quantity having no correlation with the external disturbance. The formula 36 includes the relative velocity and the sprung velocity as the corresponding state quantities,
In the present embodiment, the correlation with the relative speed obtained from the suspension system 10 is calculated. This correlation function

[0261]

[Outside 3] Then, this is calculated by the following equation.

[0262]

[Equation 37] By calculating such a cross-correlation, the tire air pressure fluctuation amount Δk can be extracted and the road surface disturbance term can be separated. That is, the output estimated total disturbance and relative velocity are sampled continuously for N points, and the average value shown by the following equation is obtained.

[0263]

[Equation 38] Then, the correlation calculation of Expression 37 is performed using these average values. Thus the correlation function

[0264]

[Outside 4] , The sprung speed term and the road surface displacement rate disappear,
This value is as shown in the following equation.

[0265]

[Formula 39] The correlation function calculated in this way is the product of the term (Δk1 / m1) representing the fluctuation of the spring constant of the tire and the autocorrelation function of the relative velocity of the suspension used in the correlation calculation (shown by Equation 40). Can be expressed. Therefore, if the obtained correlation function is divided by the autocorrelation function of the state quantity (relative velocity), the variation of the spring constant can be quantitatively detected.

[0266]

[Formula 40] That is, as can be seen from the above Equation 39, from the various frequency components of the comprehensive estimated disturbance, the correlation function described above is selected.

[0267]

[Outside 5] Has a value corresponding to the frequency component of the internal disturbance generated only by the fluctuation of the spring constant of the tire. Therefore, this correlation function

[0268]

[Outside 6] From this, the fluctuation amount of the spring constant can be detected.

The above-described relative velocity autocorrelation function is calculated by the autocorrelation calculator 74 and input to the normalizer 72. Then, the normalization unit 72 outputs the correlation function output from the cross-correlation calculation unit 70.

[0270]

[Outside 7] Is an autocorrelation function output from the autocorrelation calculation unit 74

[0271]

[Outside 8] By dividing by, the variation amount of the tire spring constant is detected and output as shown in the following equation.

[0272]

[Formula 41] Here, since the parameter m1 is the mass of the tire and is a known value, it is possible to accurately know the variation Δk1 of the spring constant from the output J of the normalization unit 72.

Then, the abnormality determining section 78 compares the spring constant variation Δk1 thus obtained with the variation reference value that should be determined to be abnormal, and determines the air pressure abnormality.

FIG. 11 is a flow chart of the tire pressure abnormality detection algorithm thus performed.

In this embodiment, along the flow 100, disturbance estimation and data sampling are performed every 5 ms, N = 4.
Repeated 00 times.

Next, in accordance with flow 110, the previous 40
Using the data obtained by sampling 0 times (sampling for the past 2 seconds), the cross-correlation and auto-correlation are calculated, and the normalized value J
To determine whether the tire pressure is normal or abnormal.

In this embodiment, the disturbance observer 32
Since the response of 5 ms is used, the relation that the differential value of the total disturbance W of the dynamic system 10 becomes almost 0 is established.

FIG. 12 shows the correlation between the estimated tire spring constant and the actually measured spring constant according to the abnormality detection algorithm shown in FIG.

At this time, the measurement experiment was carried out by the following method.

First, the fluctuation amount Δk1 / k1 is set with respect to the spring constant k1 when the air pressure is normal, and the air pressure is reduced in advance so as to reach the set value.

Next, the vehicle is run, and the estimated value of the spring constant variation amount 2 seconds after the start of estimation is recorded. Subsequent estimates will be the same as the estimates at that point.

Next, the set value of Δk1 / k1 is changed in 0.1 steps, and the experiment is repeated. The data thus obtained is as shown in FIG. From this experimental result, it is understood that by using the diagnostic device 35 of the present embodiment, the variation amount of the spring constant can be estimated with high accuracy within ± 3%.

In this embodiment, the relative speed is used for the cross-correlation calculation, but the sprung speed may be used for the correlation calculation. In this case, the cross-correlation calculation unit 70 calculates the cross-correlation between the estimated total disturbance and the sprung mass velocity based on the following equation, and the autocorrelation calculation unit 74 calculates the autocorrelation of the sprung mass velocity.

[0284]

[Outside 9] And the normalization unit 72 is configured to detect an abnormality in tire air pressure by dividing the cross-correlation by the autocorrelation.

[0285]

[Equation 42] As described above, according to the second embodiment, it is not necessary to directly attach the pressure sensor, the wireless device, and the like to the tire as in the conventional technique, and therefore, problems such as reliability and durability may not occur. Absent.

Also, as the sensor necessary for implementing the second embodiment, the sensors 60a, 60b, 60c conventionally used for controlling the active suspension can be used as they are, and a new sensor for detecting the abnormal air pressure can be used. There is no need to provide a sensor.

In the second embodiment, the tire air pressure fluctuation amount reference value that should be judged to be abnormal is set in advance, the fluctuation amount obtained by the disturbance observer 12 is compared with the reference value, and The air pressure abnormality was judged. The device of the present invention is not limited to this, and can be used for various applications. For example, since the amount of change in the tire spring constant can be accurately detected as it is, it can be used as an air pressure monitor that constantly indicates the detected air pressure state to the driver. Furthermore, if the information on the detected air pressure is used for the control law of the active suspension, it is possible to realize an appropriate ride comfort corresponding to the fluctuation of the air pressure. Further, if the information on the air pressure is used for controlling the suspension that can change the attenuator constant, it is possible to appropriately respond to the fluctuation of the air pressure and realize a more comfortable riding comfort.

Further, in the second embodiment, the system 10 including the suspension and the wheels is modeled by the state equation shown in the above-mentioned equation 34, and the suspension spring is
Only the acceleration under the spring can be measured by the sensor,
Other state quantities are obtained by integrating these signals.
However, even if the sensor that measures the state of the suspension is not the acceleration sensor but another sensor, if the model is changed according to the state quantity measured by that sensor,
The tire pressure abnormality can be determined by the same method.

For example, when the state quantity measurable by the sensor is not the acceleration signal but the sprung displacement x2 and the relative displacement y, the model corresponding to the equation 34 is set as the following equation.

[0290]

[Equation 43] In this number 43, the state of the suspension

[0291]

[Outside 10] In the above, the sprung speed and the relative speed are obtained from the difference between the sprung displacement x2 and the relative displacement y, respectively.

(Third Embodiment) FIG. 10 shows a diagnostic device 30 when such a suspension model 10 is used as a diagnostic object.
This embodiment is shown as a third embodiment.

When the tire air pressure fluctuates and the tire spring constant fluctuates, the disturbance observer 32 is configured to estimate and calculate the total disturbance represented by the following equation.

[0294]

[Equation 44] Therefore, by dividing the estimated cross-correlation between the total disturbance and the displacement y by the autocorrelation of the relative displacement, it is possible to determine the tire pressure abnormality, as in the above-described embodiment. At this time, even if the cross-correlation between the estimated disturbance and the sprung displacement x2 is divided by the autocorrelation function of the sprung displacement x2, the tire pressure abnormality can be similarly determined.

As described above, in the second embodiment and the third embodiment of the present invention, the state quantity used for the cross-correlation calculation with the estimated total disturbance can be obtained by using the state quantity that affects the system by the fluctuation of the air pressure. Good.

Application to Conventional Inactive Suspension (4th to 6th Embodiments) FIG. 13 shows a suspension system 10 including inactive conventional suspensions and wheels. In this suspension system 10,
The wheel 41 is represented by the unsprung mass portion represented by the parameter m1 and the spring portion of the tire represented by the spring constant k1. Further, 42 represents a vehicle body part having a sprung mass m2,
46 represents a spring having a spring constant k2, 48 represents a damper having a damper constant Dm, 56 represents a road surface displacement represented by a variable x0, 52 represents an unsprung displacement represented by a variable x1, 50 Represents the sprung displacement represented by the variable x2, 5
4 represents the relative displacement (x1-x2) represented by the variable y.
In the dynamic system 10 as well, similarly to the above-described embodiment, the acceleration sensor 60b and the acceleration sensor 60c for detecting the vertical vibration of the wheel are provided on the upper and lower springs.

Such a suspension system 10
Is expressed by a state equation, it is expressed as the following equation.

[0298]

[Equation 45] (Fourth Embodiment) FIG. 14 shows, as a fourth embodiment, an embodiment in which the suspension system 10 shown in FIG.

The current suspension system 10
In the above, although the one in which the damping force of the suspension is changed according to the condition of the road surface or the judgment of the driver is also realized, this can be modeled by assuming that the attenuator constant Dm changes. At this time, Dm shown in FIG.
Represents the representative value of the changing Dm.

The respective elements of the state quantity x included in the above equation 45 are the same as in the above embodiment.
It is calculated and output by a calculation unit provided therein. That is, the sprung acceleration is directly output from the output of the sensor 62b, and the sprung speed is obtained by integrating the value of the sprung acceleration. The relative acceleration is obtained from the difference between the unsprung acceleration and the unsprung acceleration, and the relative velocity is obtained by integrating the relative acceleration and the like. Therefore, suspension system 1
The output of 0 becomes the internal state quantity vector x of the above-mentioned equation 45.

The diagnostic device 30 of the present embodiment, which targets the suspension system 10 having the above-mentioned configuration, for diagnosis is
The details will be described below.

When the tire air pressure is normal and the tire spring constant at this time is k1, the disturbance observer 32 estimates and calculates the total disturbance represented by the following equation.

[0303]

[Equation 46] Here, it is assumed that the damping constant Dm of the suspension changes, and the amount of change from the representative value is represented by ΔDm.

When the tire air pressure fluctuates and the spring constant thereof fluctuates by Δk1 to (k1 + Δk1), the disturbance observer 32 outputs a signal represented by the following equation.

[0305]

[Equation 47] As described above, the total disturbance estimated by the disturbance observer 32 includes the disturbance caused by the fluctuation of the attenuator constant, the internal disturbance caused by the fluctuation of the tire air pressure, and the external disturbance received from the road surface. I understand.

Therefore, the cross-correlation calculation unit 70 calculates the cross-correlation between the relative velocity and the second element of the estimated total disturbance including the internal disturbance caused by the tire air pressure fluctuation. Further, the autocorrelation calculation unit 74 calculates and outputs the autocorrelation of the relative speed by the same method as in the above-mentioned embodiment.

Then, the normalization section 72 divides the output of the cross-correlation calculation section 70 by the output of the auto-correlation calculation section 74,
The amount of fluctuation of the spring constant is detected and output to the abnormality determination unit 78.

Therefore, the abnormality determining unit 78 compares the variation amount of the spring constant input in this way with a predetermined reference value to determine the abnormality of the air pressure.

As described above, in the present embodiment, even in the case of the wheel mounted on the conventional suspension which is not active, the acceleration sensors 60b and 60c for detecting the vertical acceleration of the sprung and unsprung portions of the suspension are simply provided. It is possible to detect an abnormality in tire air pressure.

In the fourth embodiment, the system 10 consisting of the suspension and the wheels is modeled by the equation of state shown by the above equation 43, and
Assuming that only the unsprung-under acceleration can be measured by the sensor, other state quantities are obtained by integrating these signals. However, not limited to this, even if the sensor for measuring the suspension state is not the acceleration sensor but another sensor, if the model is changed in accordance with the state amount measured by the sensor, the tire is measured in a similar manner. An abnormal air pressure can be determined.

(Fifth Embodiment) The fifth embodiment of the present invention is characterized in that the state quantity measurable by the sensor is not the acceleration signal but the sprung displacement x2 and the relative displacement y. In the fifth embodiment, the model corresponding to the equation 43 is expressed by the following equation.

[0312]

[Equation 48] In the equation, the sprung speed and the relative speed in the suspension state x are obtained from the difference between the sprung displacement x2 and the relative displacement y, respectively.

FIG. 15 shows an apparatus of the fifth embodiment in which such a suspension system 10 is the object of diagnosis.

When the tire air pressure fluctuates and the tire spring constant fluctuates, the disturbance observer 32 estimates and calculates the total disturbance represented by the following equation.

[0315]

[Equation 49] Then, using the correlation calculation unit 70 and the autocorrelation calculation unit 74,
The calculation of the cross-correlation between the estimated second element of the total disturbance and the relative displacement y and the calculation of the auto-correlation of the relative displacement y are performed, and the cross-correlation value and the auto-correlation value thus obtained are normalized. By dividing by the conversion section 72, it is possible to determine the tire pressure abnormality as in the previous embodiment. At this time, the second element of the estimated total disturbance and the sprung displacement x
Even if the cross-correlation function with 2 is divided by the autocorrelation function of the sprung displacement, the tire air pressure abnormality can be similarly determined.

In the fifth embodiment, the disturbance included in the second element of the total disturbance estimated by the calculation of the cross correlation.

[0317]

[Outside 11] Although the internal disturbance (which is the internal disturbance caused by the change of the damping ratio constant) is removed in the same manner as the external disturbance, in the present invention, the term of the internal disturbance is deleted from the second element of the estimated total disturbance in advance. Then, the correlation calculation can be performed.

(Sixth Embodiment) FIG. 16 shows a block diagram of a diagnostic apparatus according to a sixth embodiment of the present invention incorporating this method. In the figure, the damping constant compensator 80 is
The following calculations are performed using the estimated total disturbance from the disturbance observer 32.

[0319]

[Equation 50] This formula is

[0320]

[Equation 51] Therefore, it can be seen that the term due to the fluctuation of the attenuator constant has been eliminated. Then, by calculating the cross-correlation between w12 and the relative speed or the sprung speed, it is possible to similarly detect the abnormality of the air pressure.

In the fifth embodiment shown in FIG. 15 as well, by providing the attenuator constant compensator 80 in the same manner,
Before the cross-correlation calculation, the disturbance due to the fluctuation of the attenuator constant is canceled in advance. Therefore, when the present invention is applied to the suspension in which the variation of the attenuator constant is extremely large, there is an advantage that the accuracy of detecting the abnormality in the air pressure is improved.

(Seventh Embodiment) In the first embodiment, it is assumed that the internal state quantity of the dynamic system 10 has no correlation with the external disturbance, but the structure of the dynamic system and the controller Depending on the control law, there may be a correlation between the internal state quantity and the external disturbance.

An example in this case will be described as a seventh example in correspondence with the first to third aspects of the first example. The members corresponding to those in the first embodiment are designated by the same reference numerals, and the description thereof will be omitted.

First Mode In the first mode of the present embodiment, all the internal state quantities x (t) of the diagnosis object 10 can be measured by using a sensor or the like, and the disturbance estimation means 32 can use the total disturbance vector w ( This is the case when it is formed to estimate only t).

The block diagram of the diagnostic device 30 is constructed as shown in FIG.

The diagnostic device 30 of the embodiment includes the correlation compensator 35.
And a correction value memory 37.

That is, the correlation calculator 34 calculates the cross-correlation between the internal state quantity and the estimated value of the total disturbance vector, but there is a slight difference between the internal state quantity and the road surface disturbance entering as the external disturbance. If there is a correlation, the effect of the road surface disturbance that invades as an external disturbance cannot be effectively removed as it is.

Therefore, in the correlation compensator 35, the correlation calculation result, the internal state quantity, and the correction value memory 3 are stored in advance.
The correlation calculation result is corrected by the correction value set to 7, and the influence of road surface disturbance is excluded.

The diagnosis section 36 can correct the corrected correlation calculation result, detect the occurrence of a failure, and specify a specific failure location.

The structure and operation of the correlation compensator 35 will be described below.

From Expression 18, it is the second element of the total disturbance that includes the road surface disturbance x0 that is an external disturbance in the total disturbance, so the correlation between the estimated value and the measured value y of the second disturbance. The function C21 is corrected.

First, the correction value preset in the memory 37 is obtained as follows.

The disturbance is estimated by the disturbance observer 32 while the dynamic system is operating in the reference state (for example, in the state where there is no failure). This estimated value is

[0334]

[Equation 69] Is equal to That is,

[0335]

[Equation 70] Becomes

Therefore, the estimated value of w2 and the measured value y
The correlation with is equivalent to the correlation between the road disturbance x0 and the measured value y. That is, it can be written as the following equation.

[0337]

[Equation 71] Therefore, under the reference condition, this correlation calculation is performed, and this value is divided by the autocorrelation C (y, y) of the measured value y and the normalized value is stored in the memory 37 as the correction value h. This correction value is obtained by the following equation.

[0338]

[Equation 72] The correction value h has a substantially constant value regardless of the magnitude of the road surface disturbance. Road disturbance x0 in other operating conditions
The correlation between the measured value y and the measured value y is
It can be obtained by multiplying by the autocorrelation.
That is, the correlation C between the estimated value of w2 at the time of failure and the measured value y
21 can be approximated by the following equation.

[0339]

[Equation 73] Therefore, at the time of failure diagnosis, the correlation compensator 35 calculates the autocorrelation C (y, y) of the measurement value y. Then, using this value C (y, y) and the correction value h stored in the memory 37 in advance, the output C of the correlation calculator 34 is calculated.
21 is corrected according to the following equation, and the result is again output as C21 to the diagnosis unit 36.

[0340]

[Equation 74] The corrected correlation C21 can be expressed by the following equation.

[0341]

[Equation 75] Similar to the first embodiment, the diagnosis unit 36 detects the occurrence of a failure based on the corrected correlation calculation result and identifies the specific failure location. For example,

[0342]

[Outside 55] Since only the variation of the gas spring appears in the correlation C41 between y and y, if there is an abnormality in this value, it can be immediately identified as a failure of the gas spring. Similarly,

[0343]

[Outside 56] And measured value

[0344]

[Outside 57] If the correlation C42 is abnormal, it is determined that the attenuator is defective. The remaining tire pressure abnormality is determined by the correlation function C21. C21 has a value due to a gas spring failure or tire air pressure abnormality.
Since the failure of the gas spring is detected by the judgment by 41,
Therefore, if there is no failure in the gas spring and the value of C21 is abnormal, it is determined that the tire is abnormal.

As described above, in this embodiment, even when the internal state quantity of the dynamic system 10 has a correlation with the external disturbance, the external disturbance and the internal disturbance caused by the failure are easily separated, and the failure point is detected. Can be specified.

Further, in the present embodiment, the correction value h is shown to be a substantially constant value regardless of the road surface disturbance. The present invention is not limited to this, and when the correction value h changes depending on the road surface disturbance x0, the correlation calculation between the measured value y and the road surface disturbance x0 is performed, and the calculation result is stored in the memory 37 as the correction value h. ,
In the correlation compensator 35, it is possible to make a correction with the correction value h corresponding to the measurement value y.

Second Mode In the second mode of the seventh embodiment, when a part of the internal state quantity x (t) of the diagnostic object 10 cannot be measured or is estimated by the disturbance estimating means 32 without measurement. If you can. In these cases, the disturbance estimator 32 is configured to estimate the total disturbance vector w (t) and the internal state quantity that cannot be measured or cannot be measured.

FIG. 22 is a block diagram of the diagnostic device at this time.
Shown in. The correlation compensating unit 35 and the correction value memory 37 in the figure have the same configurations and operations as those of the embodiment shown in FIG. 21, and therefore their explanations are omitted here.

Third Mode In the third mode of the seventh embodiment, the disturbance estimating means 32 is used when a part of the internal state quantity x (t) of the diagnosis object 10 cannot be measured.
In this case, the total disturbance vector w (t) and all internal state quantities x (t) including the internal state quantities that cannot be measured are estimated and calculated.

FIG. 23 is a block diagram of the diagnostic device at this time.
Shown in. The correlation compensator 35 and the correction value memory 37 in the figure have the same configuration and operation as those of the embodiment shown in FIG.

(Eighth to twelfth embodiments) The second to the twelfth embodiments
In the sixth embodiment, the failure diagnosis device 30 is configured on the assumption that there is no correlation between the road surface disturbance and the internal state quantity. However, depending on the configuration of the suspension 10 and the control algorithm of the controller 12, the road surface disturbance may be There may be a case where there is a correlation with the internal state quantity. Pneumatic diagnostic devices in such a case will be described as eighth to twelfth embodiments.

(Eighth Embodiment) First, the second embodiment shown in FIG.
An example in which there is a correlation between the road surface disturbance and the internal state quantity of the suspension 10 will be described as an eighth example.

FIG. 24 shows a block diagram of this embodiment.

In the figure, the cross-correlation calculating section 70 calculates the cross-correlation between the relative velocity of the suspension, which is the internal state quantity, and the estimated value of the total external disturbance estimated by the disturbance observer 32. At this time, if there is a correlation between the two, it is not possible to eliminate the influence of the road surface disturbance that enters as an external disturbance.

Therefore, in the correlation compensator 35, the correlation calculation result is corrected by the correlation calculation result, the internal state amount and the correction value set in the memory 37 in advance, and the influence of the road surface disturbance is excluded.

In the diagnosis section 36, based on the corrected correlation calculation result, by the same operation as that of the embodiment shown in FIG.
Detect abnormal tire pressure.

The operation of the correlation compensator 35 will be described below.

It has been fully described that the disturbance observer 32 estimates the disturbance expressed by the equation 36 when the tire air pressure fluctuates and the tire spring constant fluctuates. Then, in the same manner as in the above embodiment, the correlation calculation unit 70 calculates the cross-correlation between the measured value of the disturbance and the relative speed, and the result is normalized by the normalization unit
When divided by the autocorrelation function of the relative velocity at 2, the output of the normalization unit 72 is as follows.

[0359]

[Equation 76] here,

[0360]

[Outside 58] Is shown in the following equation.

[0361]

[Equation 77] However,

[0362]

[Equation 78] Is.

As described above, if there is a correlation between the road surface disturbance and the relative speed, the second term on the right side of the equation 76 affects the correlation calculation result.

On the other hand, under normal air pressure conditions, the output J of the normalizing section 72 is given by

[0365]

[Equation 79] Is equal to This value can be expressed as:

[0366]

[Equation 80] Therefore, this output result J is divided by k1 / m1

[0367]

[Equation 81] Is stored in the memory 37 in advance as a correction value h.

The correction value h has a substantially constant value regardless of the magnitude of the road surface disturbance. Therefore, when the air pressure is normal, the output J of the normalizing unit 72 and the calculation result based on the equation 81 are stored in the memory in association with each other under a proper operating condition of the suspension 10 and used as a correction value. Good.

From the expressions 76 and 81, the output of the normalizing section 72 is

[0370]

[Equation 82] Can be expressed as

[0371]

[Equation 83] Can be found at.

Therefore, in the correlation compensator 35, the normalizer 7
The correction value h is read using the J output from the data 2, the calculation is performed using the correction value h according to the arithmetic expression of Expression 83, and the calculation result is output to the diagnosis unit 36 as J again.

Then, the diagnosis unit 36 determines the abnormality of the air pressure by the same operation as that of the embodiment of FIG.

As described above, in this embodiment, even when the internal state quantity of the suspension has a correlation with the road surface disturbance,
It is possible to easily separate the external disturbance and the internal disturbance caused by the abnormality in the air pressure to determine the abnormality in the air pressure.

FIG. 29 shows a flowchart of the tire pressure abnormality detection algorithm thus performed. Along the flow 120, cross-correlation and auto-correlation are calculated, the normalized value J and the correction value h shown in Formula 76 are used to calculate Formula 83, and the calculated value is used as J again. The tire pressure is judged to be normal or abnormal.

FIG. 30A shows a state in which the controller is adjusted so that the relative velocity of the suspension and the road surface disturbance are intentionally generated, and the air pressure is 2.0 Kg / cm 2 (an example of the standard state). The time variation is shown as a result of running the car under and calculating the equation 81. FIG. 30 (b) shows
FIG. 30 (a) shows the result of running a car on a rougher road surface than the time of measurement and performing the same calculation. From these results, it is understood that the calculation result of the equation 81 hardly changes with time, and hardly changes even if the road surface state changes.
Therefore, it is understood that the representative value of this value can be used as the correction value h.

FIG. 31 (a) shows a value obtained by estimating the spring constant of the tire according to the abnormality detection algorithm shown in FIG. 11 under the condition that the relative velocity of the suspension and the road surface disturbance are correlated. , The correlation of the value with the actually measured spring constant is shown. From the figure, it can be seen that when there is a correlation between the relative speed and the road surface disturbance, the fluctuation amount of the spring constant cannot be accurately estimated.

FIG. 31 (b) shows the correlation between the estimated tire spring constant and the actually measured spring constant in accordance with the abnormality detection algorithm shown in FIG. In this algorithm, the correction value shown in FIG.
The average value of the values indicated by 0 (a) is used. From the figure, it is understood that the variation amount of the spring constant can be estimated with high accuracy by performing the correction calculation according to Formula 83.

(Ninth Embodiment) Next, in the third embodiment shown in FIG. 10, an embodiment in which there is a correlation between the road surface disturbance and the relative displacement of the suspension 10 will be described as a ninth embodiment. To do.

FIG. 25 shows a block diagram of this embodiment. In the correlation compensator 35, the correlation calculation result, the relative displacement, and the correction value h preset in the memory 37.
Thus, the correlation calculation result can be corrected and the influence of the road surface disturbance can be excluded. In this case, as the correction value h, a value calculated based on the output of the normalizing unit 72 when the air pressure is normal is used.

(Tenth Embodiment) Next, in the fourth embodiment shown in FIG. 14, the road surface disturbance and the suspension 10
An example in which there is a correlation with the relative speed of No. 10 will be described as a tenth example.

FIG. 26 shows a block diagram of this embodiment. In the figure, the correlation calculator 70 calculates the cross-correlation between the relative velocity of the suspension and the estimated value of the second element of the total disturbance estimated by the disturbance observer 32.

In the correlation compensator 90, the correlation calculation result is corrected by the correlation calculation result, the relative speed, and the correction value h preset in the memory 37, and the influence of road surface disturbance is excluded.

In the diagnosis section 36, based on the corrected correlation calculation result, by the same operation as that of the embodiment shown in FIG.
Detects abnormal air pressure. As the correction value h, a value calculated based on the output of the normalizing unit 72 when the air pressure is normal is used.

(Eleventh Embodiment) Next, in the fifth embodiment shown in FIG. 15, road surface disturbance and suspension 10
An example in which there is a correlation with the relative displacement of is described as an eleventh example.

FIG. 27 shows a block diagram of this embodiment. In the figure, a correlation calculator 70 calculates the cross-correlation between the relative displacement of the suspension and the estimated value of the total external disturbance w2 estimated by the disturbance observer 32.

The correlation compensating section 90 corrects the correlation calculation result by the correlation calculation result, the relative displacement, and the correction value h preset in the memory 37, and eliminates the influence of the road surface disturbance.

In the diagnosis section 36, based on the corrected correlation calculation result, by the same operation as that of the embodiment shown in FIG.
Detects abnormal air pressure. As the correction value h, a value calculated based on the output of the normalizing unit 72 when the air pressure is normal is used.

(Twelfth Embodiment) Next, in the sixth embodiment shown in FIG. 16, road surface disturbance and suspension 10
An embodiment in the case where there is a correlation with the relative speed of is described as a twelfth embodiment.

FIG. 28 shows a block diagram of this embodiment. In the figure, the correlation calculator 70 outputs w12 which is the output of the relative velocity of the suspension and the attenuator constant compensator 80.
Calculate the cross-correlation with.

The correlation compensator 35 corrects the correlation calculation result by the correlation calculation result, the relative speed, and the correction value h preset in the memory 37, and eliminates the influence of the road surface disturbance.

In the diagnosis section 36, based on the corrected correlation calculation result, by the same operation as that of the embodiment shown in FIG.
Detects abnormal air pressure. As the correction value h, a value calculated based on the output of the normalizing unit 72 when the air pressure is normal is used.

As described above, in the present invention, even when the internal state quantity of the suspension has a correlation with the external disturbance, the road surface disturbance and the internal disturbance caused by the air pressure abnormality are easily separated, and the tire pressure abnormality is detected. be able to.

(Thirteenth Embodiment) Next, a detailed description will be given of an embodiment of a vehicle body weight detecting device for detecting a change in vehicle body weight using the principle of the dynamic system diagnostic device of the present invention. In addition,
The members corresponding to those in the above-mentioned embodiment are designated by the same reference numerals and the description thereof will be omitted.

The case where this embodiment is applied to the active suspension and the case where it is applied to the conventional passive suspension will be described. When Applied to Active Suspension FIG. 32 shows a suspension system to be diagnosed in this embodiment. Since the same figure is the same as the diagnosis target shown in FIG. 3, the corresponding members are denoted by the same reference numerals, and the description thereof will be omitted. In addition, in this dynamic system 10, as in the second and third embodiments, the sprung portion and the unsprung portion are
It is assumed that an acceleration sensor 60b and an acceleration sensor 60c that detect vertical vibrations of the wheels, and a pressure sensor 60a for active control of the suspension are provided. The suspension system 10 as described above is represented by the following equation when expressed by a state equation.

[0396]

[Equation 84] Where a = k1 / m1 + k2 / m1 + k2 / m2, b
= 1 / m1 + 1 / m2, T means a delay time between the manipulated variable u and the active control force f.

In FIG. 32, the sprung mass m2 represents the weight of the vehicle body. Therefore, in the present embodiment, the variation of the vehicle body weight is represented by the variation of the parameter m2, and at the same time, the tire 40
The air pressure abnormality, the gas spring 46 abnormality, and the attenuator 48 abnormality are assumed and treated as fluctuations of the parameters k1, k2, and Dm, respectively.

FIG. 33 is a block diagram of the system of this embodiment. The suspension system 10 to be diagnosed receives the operation amount u output from the controller 12 as an input, and a vector included in the equation 84.

[0399]

[Outside 12] Is a system configured such that

Among the elements included in x of the internal state quantity vector, the sprung acceleration is directly detected by the acceleration sensor 60b shown in FIG. 32, and the sprung speed is obtained by integrating the sprung acceleration. The relative acceleration is obtained from the difference between the unsprung acceleration and the unsprung acceleration detected by the acceleration sensor 60c. The relative velocity is obtained by integrating the relative acceleration. The respective calculation units that perform such calculations are the suspension system 10 of the embodiment.
Shall always be incorporated within. Therefore, the output of the suspension 10 is the internal state quantity vector included in the equation 84.

[0401]

[Outside 13] Becomes

The controller 12 of the present embodiment receives the internal state quantity vector x thus output as an input, and calculates and outputs the manipulated variable u which becomes the input of the active suspension system 10.

Further, the diagnostic device 30 of the embodiment which targets the active suspension system 10 includes a disturbance observer 32, a correlation calculator 34, and a variation detector 90.

The disturbance observer 32 receives the output u of the controller 12 and the output x of the active suspension system 10 as inputs, and estimates the fluctuation of the parameters assumed above as an internal disturbance generated in the active suspension system 10. Is formed.

The state quantity detection unit 60 extracts from the output x of the active suspension system 10 the state quantity (fu in this case) which affects the system 10 by the fluctuation of the vehicle body weight m2. The reason for selecting (fu) as the state quantity will be described later in the operation.

The correlation calculator 34 is the cross-correlation calculator 7
0, a normalization unit 72, and an autocorrelation calculation unit 74 are included.

Then, the cross-correlation calculating section 70 calculates and outputs the cross-correlation between the total disturbance estimated by the disturbance observer 32 and (fu).

The auto-correlation calculator 74 calculates the auto-correlation function of the state quantity (fu).

Then, the normalization section 72 calculates the cross-correlation function calculated by the cross-correlation calculation section 70 from the auto-correlation calculation section 7
4 is divided by the autocorrelation function calculated and normalized, and the fluctuation amount detection unit 90 detects fluctuations in the vehicle body weight.

This embodiment is constructed as described above, and its operation will be described below.

The assumed parameter fluctuation is defined as follows.

[0412]

[Equation 85] When the above-described parameter fluctuation occurs, the response of the system 10 that has changed due to the fluctuation can be regarded as a normal response to which internal disturbance corresponding to the parameter fluctuation is added.

The disturbance observer 32 receives the output signal x and the input signal u of the suspension system 10 as input, and estimates and outputs a comprehensive disturbance vector w including the internal disturbance and the external disturbance. The estimated value of the total disturbance at that time is

[0414]

[Equation 86] Becomes

In the above equation, the term represented only by the fluctuation of the vehicle body weight m2 is the disturbance factor.

[0416]

[Outside 14] Δa25f of the third term, Δb2u of the fourth term of

[0417]

[Outside 15] The fourth term is Δa45f and the fifth term is Δb4u. To extract the fluctuation of the vehicle body weight m2 from the total disturbance represented by the above equation, the road surface disturbance that is an external disturbance is used.

[0418]

[Outside 16] Disturbance element that does not exist

[0419]

[Outside 17] Is preferred.

The above formula

[0421]

[Outside 18] Can be rewritten as

[0422]

[Equation 87] From the above equation, the amount Δa25 representing the fluctuation of the vehicle body weight m2 is extracted. Therefore, the correlation calculator 70 estimates the total disturbance.

[0423]

[Outside 19] And (fu) are calculated. This correlation

[0424]

[Equation 88] Then, this is calculated by the following equation.

[0425]

[Equation 89] By calculating such cross-correlation, the term of Δa25 is separated from the other terms, and only the fluctuation of the vehicle body weight m2 is extracted. That is, the estimated value of the total disturbance output

[0426]

[Outside 20] And (fu) are sampled continuously for N points to obtain the average value shown by the following equation.

[0427]

[Equation 90] Then, using the average value of these, the correlation calculation of the equation 89 is performed. When this correlation function is performed, the terms other than the term of Δa25 disappear, and the value becomes as shown in the following equation.

[0428]

[Formula 91] The correlation function calculated in this way can be expressed by the product of the term representing the fluctuation of the vehicle body weight m2 and the autocorrelation function of the suspension state quantity (fu) used for the correlation calculation. Therefore, if the obtained correlation function is divided by the autocorrelation function of the state quantity (fu), the fluctuation of the vehicle body weight m2 can be quantitatively detected.

The autocorrelation function of this state quantity (fu) is
It is calculated by the autocorrelation calculation unit 74 and input to the normalization unit 72. Then, the normalization unit 72
By dividing the correlation function output from the cross-correlation calculation unit 70 by the auto-correlation function output from the auto-correlation calculation unit 74, the fluctuation of the vehicle body weight m2 is detected and output as shown by the following equation.

[0430]

[Equation 92] Here, the parameters m2 and T are known.
Therefore, the fluctuation amount detection unit 90 can accurately calculate the fluctuation amount Δm2 of the vehicle body weight m2 by using the output J of the normalization unit 72 and performing the calculation shown in the following equation.

[0431]

[Equation 93] In this example, the state quantity used for the correlation calculation is (f-
Although u) is selected, this may be selected as f. At that time, the cross-correlation calculation unit 70 calculates the correlation between the estimated value of the first element of the total disturbance and the state f, and outputs the value shown by the following equation.

[0432]

[Equation 94] Since the state f is the operation amount u delayed by the time T, the correlation C (u, f) between the operation amount u and the state f may not be zero. Therefore, in order to obtain the fluctuation of the vehicle body weight m2, first, simultaneously with the autocorrelation calculation of the state f, the cross-correlation C (u, f) between the operation amount u and the state f is obtained. Next, the difference C (f, f) -C (u, f) between this value and the autocorrelation function C (f, f) of the state f is obtained. Then, in the normalization unit 72, the output of the cross-correlation calculation unit 70 is divided by the calculated difference value, so that the variation in the vehicle body weight m2 can be detected. The output J of the normalization unit 72 at that time is as follows.

[0433]

[Formula 95] In addition, in the same manner, the cross-correlation calculation unit 70 causes the disturbance.

[0434]

[Outside 21] Estimated value and relative velocity

[0435]

[Outside 22] Correlation function with

[0436]

[Outside 23] And the relative velocity is calculated in the autocorrelation calculation unit 74.

[0437]

[Outside 24] Autocorrelation function of

[0438]

[Outside 25] And the normalization unit 72 calculates the correlation function

[0439]

[Outside 26] The relative speed

[0440]

[Outside 27] Autocorrelation function of

[0441]

[Outside 28] Dividing by, the output J of the normalization unit 72 is the amount represented by the fluctuation of the gas spring constant k2 and the fluctuation of the vehicle body weight m2.

[0442]

[Equation 96] Is required. However, the variation amount Δm2 of the vehicle body weight m2
Is obtained by the method shown in the above example,
By using this Δm2, the variation amount Δk2 of the gas spring constant k2 can also be obtained.

Further, in the cross-correlation calculation unit 70, the disturbance

[0444]

[Outside 29] Estimated value and relative acceleration

[0445]

[Outside 30] Correlation function with

[0446]

[Outside 31] And the relative acceleration is calculated in the autocorrelation calculation unit 74.

[0447]

[Outside 32] Autocorrelation function of

[0448]

[Outside 33] And the normalization unit 72 calculates the correlation function

[0449]

[Outside 34] Relative acceleration

[0450]

[Outside 35] Autocorrelation function of

[0451]

[Outside 36] When divided by, the attenuator constant D is obtained as the output J of the normalization unit 72.
Amount represented by fluctuations in m and fluctuations in vehicle weight m2

[0452]

[Numerical Expression 97] Is calculated. By using the variation amount Δm2 of the vehicle body weight m2 obtained in the above embodiment, the variation amount Δ of the attenuator constant Dm is obtained.
Dm is required.

Similarly, the estimated value of the total disturbance is calculated by the cross-correlation calculation unit 70.

[0454]

[Outside 37] And relative speed

[0455]

[Outside 38] Correlation function with

[0456]

[Outside 39] And the relative acceleration is calculated in the autocorrelation calculation unit 74.

[0457]

[Outside 40] Autocorrelation function of

[0458]

[Outside 41] And the normalization unit 72 calculates the correlation function

[0459]

[Outside 42] The relative speed

[0460]

[Outside 43] Autocorrelation function of

[0461]

[Outside 44] Dividing by, the variation amount Δk1 of the tire spring constant k1 and the variation amount Δ of the gas spring constant k2 as the output J of the normalization unit 72.
Amount represented by variation amount Δk2 between k2 and vehicle body weight m2

[0462]

[Equation 98] And the gas spring constant k2 has already been obtained by the above method.
The variation amount Δk2 of the tire spring constant k1 and the variation amount Δk2 of the vehicle body weight m2 are obtained.
Can also know.

As described above, according to the present embodiment, the variation amount of the vehicle body weight m2 can be easily detected, and the other parameter variation amounts can be sequentially obtained using the obtained values.

By utilizing the variation amount of the vehicle body weight and the variation amounts of various parameters thus obtained in the control law of the active suspension, the optimum riding comfort corresponding to these variations can be achieved. You can

FIG. 36 shows the result of simulation performed based on the block diagram of FIG. In this simulation, disturbance estimation and data sampling are performed every 5 ms, and cross-correlation and auto-cross-correlation are performed using information obtained in the last 2 seconds. In the figure, the vehicle body weight is changed from 0% to 40% with respect to the reference weight, and the variation rate is calculated from the estimated value of the vehicle body weight variation amount at that time. From this result, it is understood that the variation rate of the vehicle body weight is estimated with an accuracy within ± 10%.

When applied to a passive suspension The suspension system to be diagnosed in this example is
Since it is the same as the diagnosis target in the tire air pressure diagnosis device shown in FIG. 13, the description thereof will be omitted. It has already been stated that the equation of state of the suspension system 10 is as shown in Equation 45.

In FIG. 13, the sprung mass m2 represents the weight of the vehicle body. Therefore, in the present embodiment, the variation of the vehicle body weight is represented by the variation of the parameter m2, and at the same time, the tire 40
The air pressure abnormality, the gas spring 46 abnormality, and the attenuator 48 abnormality are assumed and treated as fluctuations of the parameters k1, k2, and Dm, respectively.

FIG. 34 is a block diagram of the system of this embodiment. Moreover, it has already been described that the output of the suspension system 10 to be diagnosed is the internal state vector x of Equation 45.

Further, the diagnostic device 30 of the embodiment in which the suspension system 10 is a diagnostic object includes a disturbance observer 32, a correlation calculator 34, a state quantity detector 60, and a variation quantity detector 90.

The disturbance observer 32 is formed to receive the output x of the passive suspension system 10 as an input and estimate the fluctuation of the parameters assumed above as an internal disturbance generated in the passive suspension system 10.

The state quantity detection unit 60 extracts, from the output x of the passive suspension system 10, the state quantity (in this case, the relative velocity and the relative acceleration) that affects the system 10 by the fluctuation of the vehicle body weight m2.

The relative calculator 34 is the cross-correlation calculator 7
0, a normalization unit 72, and an autocorrelation calculation unit 74 are included.

Then, the cross-correlation calculator 70 calculates and outputs the cross-correlation between the total disturbance estimated by the disturbance observer 32 and the relative velocity and relative acceleration.

Further, the auto-correlation calculator 74 calculates auto-correlation functions of the relative velocity and the relative acceleration.

Then, the normalization section 72 calculates the cross-correlation function calculated by the cross-correlation calculation section 70 from the autocorrelation calculation section 7
4 is divided by the autocorrelation function calculated and normalized, and the fluctuation amount detection unit 90 detects fluctuations in the vehicle body weight.

This embodiment is constructed as described above, and its operation will be described below.

The assumed parameter fluctuation is defined as follows.

[0478]

[Numerical expression 99] When the above-mentioned parameter fluctuation occurs, the response of the suspension system 10 changed by the fluctuation can be regarded as a normal response to which an internal disturbance corresponding to the parameter fluctuation is added.

The disturbance observer 32 receives the output signal x and the input signal u of the suspension system 10 as inputs, and estimates and outputs a comprehensive disturbance vector w including the internal disturbance. The estimated total disturbance at that time is

[0480]

[Equation 100] Becomes

In the above equation, since there is no term represented only by the variation of the vehicle body weight m2, the variation of the vehicle body weight m2 is detected by using some variation terms.

First, the first element and the second element of the total disturbance estimated in the cross-correlation calculation unit 70.

[0483]

[Outside 45]

[0484]

[Outside 46] And the relative correlation between the relative velocity and the relative acceleration is calculated.

Here,

[0486]

[Outside 47] The cross-correlation function between

[0487]

[Outside 48] age,

[0488]

[Outside 49] And the cross-correlation function of relative acceleration

[0489]

[Outside 50] age,

[0490]

[Outside 51] The cross-correlation function between

[0491]

[Outside 52] age,

[0492]

[Outside 53] And the cross-correlation function of relative acceleration

[0493]

[Outside 54] Then, these values are as follows.

[0494]

[Equation 101]

[0495]

[Equation 102] By calculating such cross-correlation, each variation term of the estimated total disturbance is separated from other terms.

Then, the variation term can be extracted by dividing the correlation function thus calculated by the autocorrelation function of the relative velocity and the relative acceleration. The normalizer 72 divides the correlation function output from the cross-correlation calculator 70 by the autocorrelation function output from the autocorrelation calculator 74 to output four values as shown in the following equation.

[0497]

[Equation 103] The value Jij thus obtained (where i, j = 1,
From 2), the fluctuation amount detection unit 9 detects fluctuations in the vehicle body weight. The assumed parameter fluctuations are Δm2, Δk1, Δ
There are four types, k2 and ΔDm, and the output of the normalization unit 72 is J
Since there are four, 11, J12, J21, and J22, the variation amount can be detected by using a method such as simultaneous equations.

For example, if J22 is added to J12,

[0499]

[Equation 104] Becomes Here, since m1 and Dm are known, the variation amount ΔDm of the attenuator constant can be obtained from the above equation.

Then, ΔDm obtained in this way
And J12, the variation amount Δm2 of the vehicle body weight is obtained.

Further, Δm2 thus obtained
And J11, the fluctuation amount Δk2 of the gas spring constant is obtained,
Since Δm2 and Δk2 have been obtained up to this point, the variation Δk1 of the tire spring constant can be obtained from these values and J21.

As described above, in this embodiment, even if the output of the normalization unit 72 cannot directly detect the amount of change in the vehicle body weight, the output of the normalization unit 72 can be used to detect the amount of change in the weight of the body. It is also possible to detect other parameter variations as well.

The example applied to the passive suspension in this embodiment can be applied to the active suspension as well.

Further, in this embodiment, the diagnostic device 30 is constructed on the assumption that there is no correlation between the road surface disturbance and the internal state quantity. However, depending on the configuration of the suspension 10 and the control algorithm of the controller 12, There may be a case where there is a correlation between the road surface disturbance and the internal state quantity. A vehicle body weight variation detecting device in such a case will be described as a next embodiment.

(Fourteenth Embodiment) FIG. 35 shows an embodiment in which there is a correlation between the road surface disturbance and the internal state quantity of the suspension 10.

In the figure, the correlation calculator 70 calculates the cross-correlation between the relative velocity and relative acceleration of the suspension, which are the internal state quantities, and the estimated value of the total disturbance estimated by the disturbance observer 32. If there is a correlation between them, it is not possible to eliminate the influence of road surface disturbance that enters as external disturbance.

Therefore, in the correlation compensating unit 35, the correlation calculation result is corrected by the correlation calculation result, the internal state amount, and the correction value h preset in the memory 37.
Exclude the effects of road surface disturbances.

The fluctuation amount detecting section 90 detects the fluctuation amount of the vehicle body weight based on the corrected correlation calculation result by the same operation as that of the embodiment shown in FIG.

The operation of the correlation compensator 35 will be described below.

First, as in the above-mentioned embodiment, the assumed parameter fluctuation is defined as follows.

[0511]

[Equation 105] It has been fully described that the disturbance observer 32 estimates the disturbance represented by the expression 100 when the above-described fluctuation occurs.
Then, as in the case of the above embodiment, the correlation calculating unit 70 calculates the correlation between the estimated value of the total disturbance and the relative velocity, and when the result is divided by the normalizing unit 72 by the autocorrelation function of the relative velocity, The output of the conversion section 72 is as follows.

[0512]

[Equation 106] Thus, if there is a correlation between the road surface disturbance and the relative velocity and the relative acceleration, the second term on the right side of J21 and the third term on the right side of J22 in Formula 106 affect the correlation calculation result.

On the other hand, under the normal condition in which the parameter fluctuation does not occur, the output of the normalizing unit 72 is

[0514]

[Equation 107] Is equal to This value can be expressed as:

[0515]

[Equation 108] Therefore, in this output result, J21 and J22 are set to k1 /
Value divided by m1

[0516]

[Equation 109] Is previously recorded in the memory 37 as a correction value h.

That is, the correction values h1 and h2 are
It has a substantially constant value regardless of the magnitude of the road surface disturbance. Therefore, when the parameter value is normal in advance, under the proper operating condition of the suspension 10, J21 and J22 of the output of the normalization unit 72 and the calculation result based on the equation 109 are recorded in the memory 37, and the correction value is recorded. Can be used as

The correlation compensator 35 uses the correction values to make the following corrections.

[0519]

[Equation 110] As a result, the correlation compensator 35 outputs the following values.

[0520]

[Equation 111] Therefore, the fluctuation amount detection unit 90 detects the fluctuation amount of the vehicle body weight based on the above equation. That is, there are four variables to be obtained in the equation 111, Δk1, Δk2, Δm2, and ΔDm, and there are four equations in the equation 111. The fluctuation amounts Δk1, Δk2, Δm2, ΔDm can be obtained.

In the present embodiment, the vehicle body weight at the time of the experiment for recording the correction value h is set as the reference state of the vehicle body weight.

[0522] The actual variation of the vehicle body weight according to the present embodiment,
The correlation with the estimated value is shown in FIG. According to the present embodiment, it is possible to accurately estimate the fluctuation of the vehicle body weight.

(Fifteenth Embodiment) Next, an embodiment corresponding to Claim 3 will be described.

In the above embodiments, it is assumed that there is no correlation between the states of the dynamic system, and the system failure is diagnosed from the total disturbance. However, depending on the system, there may be a correlation between the states. In this case, an error may occur in diagnosis in the method of normalizing the correlation between the total disturbance and the state by the autocorrelation of the state, as shown in the correlation calculator 34.

In such a case, the correlation between states is compensated by using a method such as the least squares method. FIG. 37
Shows its block diagram. In the figure, fluctuations in the tire air pressure and the like are replaced by fluctuations in the physical parameters of the system 10, and the fluctuations occur in the disturbance observer 32.
As described above, the disturbance observer estimates the total disturbance vector represented by the sum of the external disturbance d and the internal disturbance ΔAx. Taking the first element of the total disturbance vector at that time as Equation 2
You can write like 6.

In Expression 26, it is assumed that Δa11 is an element due to a change in tire air pressure, and the other elements are caused by a change other than the tire air pressure. Here, for example, it is assumed that the states x1 and x2 have a correlation. In this case, if the correlation C ([Dw] 1, x1) between [Dw] 1 and x1 is taken to detect Δa11,

[0527]

[Expression 117] Becomes Here, the above equation is applied to the autocorrelation C (x1,
When divided by x1), the following equation is obtained, but the second term on the right side of the above equation does not become zero because the states x1 and x2 are correlated as described above.

[0528]

[Equation 118] Further, in this case, the correlation C between the states x1 and x2 is previously set.
Even if (x1, x2) is obtained, .DELTA.a12 is not known, so it cannot be compensated. That is, unless Δa12 is zero, Δa11 cannot be accurately obtained.

Therefore, the equation 26 is rewritten as the following equation,

[0530]

[Equation 119] It is considered to find Δa that minimizes the squared error index represented by the following equation.

[0531]

[Equation 120] This is obtained by partially differentiating the above equation by Δa and setting it to zero.

[0532]

[Equation 121] This method is characterized in that not only Δa11 but also other fluctuations of the system are simultaneously obtained as a fluctuation vector Δa.

The above equations 119, 120 and 121 are
It corresponds to the numbers 112, 113 and 114 described above.

Further, this method has an advantage that the correlation between states is automatically canceled. For example,

[0535]

[Equation 122] Then, each term on the right-hand side of the equation 121 is expressed as follows.

[0536]

[Equation 123] So, if we calculate the right side of equation 121,

[0537]

[Equation 124] Then, the term including the correlation C (x1, x2) between the states is canceled, and the system fluctuations Δa11 and Δa12 are simultaneously obtained.

With this operation, the least squares calculation unit 134 estimates the system fluctuation based on the right side of the equation 121 and outputs it to the diagnosis unit.

Further, although the inverse matrix has to be calculated in the equation 121, when the matrix has a high degree, the calculation of the inverse matrix requires a large amount of memory and calculation time. However, this can be avoided by using the recurrence formula shown below. That is,

[0540]

[Equation 125] Therefore, the equation 121 can be rewritten as the following equation. This number 121 corresponds to the above-mentioned equation 115.

[0541]

[Equation 126] In the above method, the initial value of the recurrence formula is

[0542]

[Equation 127] Then, the calculation may be started from N = 0. In this method, it is not necessary to calculate the inverse matrix of the matrix, so when the order of the matrix is high, it is possible to reduce the memory and the calculation time as compared with the method of directly calculating equation 121.

Further, instead of the error index number 120,

[0544]

[Equation 128] Using, the following recurrence equation is obtained.

[0545]

[Numerical Expression 129] Here, λ is called a forgetting factor. This method tries to obtain the system fluctuation by giving a smaller weight to old data, and is effective when convergence of the system is faster than that of Eq.

An embodiment applied to the dynamic system shown in FIG. 3 will be described below. FIG. 38 is a block diagram showing this embodiment. In the present embodiment, as in the first embodiment, it is assumed that the tire pressure is abnormal, the gas spring 46 is abnormal in pressure, and the attenuator 48 is abnormal. At that time, the equation of state of the dynamic system is expressed by Expression 15, and the total disturbance estimated by the disturbance observer 32 at the time of failure is expressed by Expression 18.

Therefore, in the least squares calculation unit 134,

[0548]

[Equation 130] Toki,

[0549]

[Equation 131] Or by using the corresponding recurrence formula,
Fluctuation expressed by the following formula

[0550]

[Equation 132] Can be asked. More new

[0551]

[Expression 133] Toki,

[0552]

[Equation 134] Alternatively, by using the recurrence formula corresponding to it, the variation

[0553]

[Equation 135] Can be asked.

From these Δa, Δk1, Δk2, Δ
Dm can be calculated.

In this embodiment, the active suspension in which u is not zero has been described as an example, but it goes without saying that the present invention can be applied to a passive suspension in which u = 0.

[Brief description of drawings]

FIG. 1 is a block diagram showing an example of a diagnostic device for a dynamic system according to the present invention.

FIG. 2 is a curve diagram showing an approximation method of disturbance used in the first embodiment.

FIG. 3 is a schematic explanatory diagram of a suspension system of an automobile which is a diagnosis target of the first embodiment.

FIG. 4 is a block diagram showing a diagnostic device of a first aspect in a first embodiment in which the suspension model shown in FIG. 3 is a diagnostic target.

FIG. 5 is a block diagram showing a diagnostic device of a second aspect in the first embodiment.

FIG. 6 is a block diagram showing a diagnostic device of a third aspect in the first embodiment.

FIG. 7 is an explanatory diagram of a conventional diagnostic device.

FIG. 8 is a schematic explanatory diagram of a dynamic system including an active suspension and wheels.

9 is a block diagram of a diagnostic device according to a second embodiment of the present invention, which targets the dynamic system shown in FIG. 8 as a diagnostic object.

FIG. 10 is a block diagram of a third embodiment of the present invention which is a modification of the diagnostic device shown in FIG.

11 is a flowchart showing the operation of the diagnostic device shown in FIG.

12 is an actual measurement data diagram showing a correlation between a variation amount of a spring constant estimated by using the diagnostic device shown in FIG. 9 and an actually measured spring constant.

FIG. 13 is a schematic explanatory diagram of a dynamic system including a conventional suspension and wheels.

FIG. 14 is a block diagram of a diagnostic device according to a fourth embodiment of the present invention, which targets the dynamic system shown in FIG. 13 for diagnosis.

FIG. 15 is a block diagram of a fifth embodiment of the present invention, which is a modification of the diagnostic device shown in FIG.

16 is a block diagram of a sixth embodiment of the present invention which is a modified example of the diagnostic device shown in FIG.

FIG. 17 is a block diagram of a conventional tire pressure diagnostic device.

FIG. 18 is a data measurement diagram in a normal state measured by using the conventional apparatus shown in FIG.

19 is a data measurement diagram at the time of abnormality measured using the conventional apparatus shown in FIG.

FIG. 20 is a block diagram of a minimum dimension observer according to the present embodiment.

FIG. 21 is a block diagram of a seventh embodiment of the present invention.

FIG. 22 is a block diagram of a modification of the seventh embodiment.

FIG. 23 is a block diagram of another modification of the seventh embodiment.

FIG. 24 is a block diagram of an eighth embodiment of the present invention.

FIG. 25 is a block diagram of a ninth embodiment of the present invention.

FIG. 26 is a block diagram of a tenth embodiment of the present invention.

FIG. 27 is a block diagram of the eleventh embodiment.

FIG. 28 is a block diagram of the twelfth embodiment.

FIG. 29 is a flow chart showing the operation of the device of the eighth embodiment.

FIG. 30 is an explanatory diagram of a temporal change of a correction value obtained when the vehicle is run under a constant condition.

FIG. 31 is an explanatory diagram of a correlation between an estimated value of a tire spring constant and an actually measured spring constant.

FIG. 32 is a schematic explanatory diagram of a vehicle suspension system which is a diagnosis target of the thirteenth embodiment.

FIG. 33 is a block diagram of a thirteenth embodiment.

FIG. 34 is a block diagram of a modification of the thirteenth embodiment.

FIG. 35 is a block diagram of a fourteenth embodiment.

FIG. 36 is an explanatory diagram of the result of an experiment conducted using the device shown in FIG. 33.

FIG. 37 is a block diagram of a fifteenth embodiment.

FIG. 38 is a block diagram of a specific example of the fifteenth embodiment.

[Explanation of symbols]

 10 Dynamic System 30 Diagnostic Device 32 Disturbance Observer 34 Correlation Calculator 36 Diagnostic Unit 38 Failure Specifier 70 Cross Correlation Calculator 72 Normalizer 74 Autocorrelation Calculator 76 Memory 78 Abnormality Judge 80 Attenuator Constant Compensator u Control Input Vector d External disturbance vector y Control output vector x State quantity vector

Claims (14)

[Claims] 1. A diagnostic device for a dynamic system for detecting a failure of the dynamic system, comprising: an external disturbance vector of the dynamic system and a sum of the internal disturbance vector based on the internal state vector of the dynamic system. Disturbance estimating means for estimating a comprehensive disturbance vector, calculating a cross-correlation between the estimated comprehensive disturbance vector and the internal state quantity vector, and separating a component related to the internal disturbance from the comprehensive disturbance vector. A dynamic operation system including: a correlation calculation means; and a diagnosis means for identifying a corresponding portion of the dynamic system from the separated components related to the internal disturbance and performing the diagnosis. System diagnostic device. 2. The correlation calculating means according to claim 1, wherein the correlation calculating means calculates a cross-correlation between an element of the comprehensive disturbance vector and an element of the internal state quantity vector having no correlation with the external disturbance, A diagnostic device for a dynamic system, which is formed to separate a component related to an internal disturbance from an element of a disturbance vector. 3. The correlation calculation means according to claim 1, wherein, as the cross-correlation calculation, the total disturbance vector,
The calculation of the direction vector of the total disturbance using the internal state vector as a base vector is performed so that the temporal sum of squares of the error between the product of the internal disturbance vector and the internal state vector is minimized. The dynamic system diagnostic device is characterized by being formed so as to separate a component related to an internal disturbance from an element of the comprehensive disturbance vector. 4. The correlation calculation means according to claim 1, wherein a plurality of elements of the estimated total disturbance vector are elements having no correlation with an external disturbance of the internal state quantity vector. Is formed to separate each element of the component related to the internal disturbance from the plurality of elements of the total disturbance vector, and the diagnostic means is configured to separate each element of the component related to the separated internal disturbance. From the above, a diagnostic device for a dynamic system is formed so as to identify a diagnostic point of the dynamic system and perform the diagnostic. 5. A tire air pressure diagnostic device for diagnosing a tire air pressure state of a dynamic system composed of a suspension and a wheel, wherein a tire air pressure in the dynamic system is determined based on an internal state quantity vector of the dynamic system. Disturbance estimating means for estimating a total disturbance vector as a sum of an internal disturbance vector generated by fluctuations and an external disturbance vector input to the dynamic system from the road surface, the estimated total disturbance, and the internal disturbance vector Correlation calculating means for calculating a cross-correlation with a state quantity vector, and separating a component related to internal disturbance from an element of the total disturbance vector; and a tire of the dynamic system from the separated component related to internal disturbance. A tire pressure diagnostic device comprising: a diagnostic means for determining a state of pneumatic pressure. 6. The correlation calculation means according to claim 5, wherein the correlation calculation means calculates a cross-correlation between an element of the comprehensive disturbance vector and an element of the internal state quantity vector having no correlation with the external disturbance, A tire pressure diagnostic device formed to separate a component related to internal disturbance from an element of a disturbance vector. 7. The correlation calculation means according to claim 5, wherein, as the cross-correlation calculation, the total disturbance vector,
The calculation of the direction vector of the total disturbance using the internal state vector as a base vector is performed so that the temporal sum of squares of the error between the product of the internal disturbance vector and the internal state vector is minimized. However, the tire pressure diagnostic device is characterized in that it is formed so as to separate the component related to the internal disturbance from the element of the comprehensive disturbance vector. 8. The cross-correlation according to claim 5, wherein the diagnosis means includes an autocorrelation calculation unit that calculates an autocorrelation of an element having no correlation with an external disturbance of the internal state quantity vector. A tire air pressure diagnostic device for diagnosing a tire air pressure state based on a value and an autocorrelation value. 9. A dynamic system diagnostic device for detecting a failure of a dynamic system, comprising: an external disturbance vector of the dynamic system and a sum of the internal disturbance vector based on the internal state vector of the dynamic system. Disturbance estimating means for estimating a total disturbance vector, an internal state quantity vector in a predetermined reference state, a correction value storage means for storing the correlation between the external disturbance vector as a correction value, the estimated total disturbance vector, Correlation calculation means for calculating a cross-correlation with the internal state quantity vector, and the comprehensive disturbance vector by correcting the cross-correlation calculated by the correlation calculation means based on the internal state quantity vector and the correction value. Correlation compensation means for separating components related to internal disturbance without being affected by external disturbance, and From the components, the anda specified diagnosis means for performing the diagnosis, the corresponding part of the dynamic system, diagnostic apparatus of a dynamic system, characterized in that the diagnosis of a dynamic system. 10. A tire air pressure diagnostic device for diagnosing a tire air pressure state of a dynamic system composed of a suspension and wheels, wherein a tire air pressure in the dynamic system is determined based on an internal state quantity vector of the dynamic system. Disturbance estimating means for estimating the total disturbance vector as the sum of the internal disturbance vector generated by the fluctuation and the external disturbance vector input to the dynamic system from the road surface, the internal state quantity vector in a predetermined reference state, and the external disturbance Correction value storage means for storing a correlation with a vector as a correction value; correlation calculation means for calculating a cross-correlation between the estimated total disturbance vector and the internal state quantity vector; the internal state quantity vector; The total disturbance is corrected by correcting the cross-correlation calculated by the correlation calculating means based on the correction value. Correlation compensating means for separating components related to internal disturbance without being affected by external disturbance from the koutor, and diagnostic means for determining the tire pressure state of the dynamic system from the separated components related to internal disturbance, A tire pressure diagnostic device comprising: 11. A vehicle body weight variation detecting apparatus for diagnosing a variation in vehicle body weight of a dynamic system including a suspension and wheels, wherein a vehicle body weight variation in the dynamic system is determined based on an internal state quantity vector of the dynamic system. Disturbance estimating means for estimating a total disturbance vector as the sum of an internal disturbance vector generated by fluctuations and an external disturbance vector input to the dynamic system from the road surface, the estimated total disturbance vector, and the internal disturbance vector Correlation calculation means for calculating the cross-correlation with the state quantity vector and separating the component related to the internal disturbance from the comprehensive disturbance vector; and the component of the body weight of the dynamic system from the separated component related to the internal disturbance. A vehicle body weight variation detection device comprising: a detection unit that detects a variation. 12. The correlation calculating means according to claim 11, wherein the correlation calculating means calculates a cross-correlation between an element of the comprehensive disturbance vector and an element of the internal state quantity vector having no correlation with the external disturbance, A vehicle body weight variation detecting device, which is formed so as to separate a component related to an internal disturbance from an element of a disturbance vector. 13. A vehicle body weight variation detecting apparatus for diagnosing a variation in vehicle body weight of a dynamic system including a suspension and wheels, wherein a vehicle body weight variation in the dynamic system is determined based on an internal state quantity vector of the dynamic system. Disturbance estimating means for estimating the total disturbance vector as the sum of the internal disturbance vector generated by the fluctuation and the external disturbance vector input to the dynamic system from the road surface, the internal state quantity vector in a predetermined reference state, and the external disturbance Correction value storage means for storing a correlation with a vector as a correction value; correlation calculation means for calculating a cross-correlation between the estimated total disturbance vector and the internal state quantity vector; the internal state quantity vector; By correcting the cross-correlation calculated by the correlation calculator based on the correction value, Correlation compensation means for separating components related to internal disturbance without being affected by external disturbance, and detection means for detecting a change in vehicle body weight of the dynamic system from the separated components related to internal disturbance, A vehicle body weight variation detecting device including: 14. The correlation calculation means according to claim 11, wherein the cross-correlation calculation includes the total disturbance vector,
The calculation of the direction vector of the total disturbance using the internal state vector as a base vector is performed so that the temporal sum of squares of the error between the product of the internal disturbance vector and the internal state vector is minimized. However, the vehicle body weight variation detecting device is formed so as to separate the component related to the internal disturbance from the element of the comprehensive disturbance vector.