JP7255358B2 - Anomaly detection device, anomaly detection method and program - Google Patents

Description

The present disclosure relates to an anomaly detection device, an anomaly detection method, and a program for detecting whether disturbance input to a controlled object is normal or abnormal.

Generally, in the field of controlling devices, particularly in the field of controlling industrial equipment, for example, by performing feedback control, the operation amount of a controller input to the device is adjusted according to the control amount of the device. As a result, control is performed to bring the control amount of the device closer to the target.

In the control as described above, a technique for detecting an abnormality by monitoring the waveform of the control system has been proposed. For example, in Patent Document 1, regarding the control amount, an integral value of the difference between an actually measured value and an estimated value calculated by calculation is calculated, and an abnormality is detected by comparing it with a predetermined threshold value. Techniques for predicting are described.

Patent No. 4520819

However, the above technique is used on the premise that no disturbance occurs, and if the integrated value of the error between the actual measurement value and the estimated value is larger than a predetermined threshold value, it will be regarded as abnormal. Therefore, when the waveform measured by the control system using this technology fluctuates, for example, when some disturbance is generated, even if the fluctuation of the waveform is normal, it may be determined to be abnormal. . Therefore, it cannot be applied as a means for detecting anomalies in devices where disturbances occur.

Accordingly, an object of the present disclosure is to solve the above problems and provide an anomaly detection device, an anomaly detection method, and a program capable of easily estimating the presence or absence of an anomalous disturbance.

In order to solve the above problems, an abnormality detection device according to an embodiment of the present disclosure includes:
A parameter of a model of the controlled object is estimated based on the manipulated variable output by the controlled variable adjuster and the controlled variable output by the controlled object, and the manipulated variable, the controlled variable, and the disturbance generator are inputted. An anomaly detection device for estimating a parameter of a model of a disturbance control amount caused by the disturbance based on a disturbance trigger signal that generates a disturbance, and detecting an anomaly of the disturbance based on the parameter,
a storage unit that stores the model of the disturbance control amount and parameters of the disturbance in the model of the disturbance control amount;
a disturbance controlled variable waveform estimator for estimating an estimated disturbance controlled variable from the controlled variable and an estimated device controlled variable calculated by subtracting the disturbance controlled variable from the controlled variable;
a parameter estimating unit that calculates the estimated disturbance control amount in a predetermined measurement period based on the disturbance trigger signal and estimates a parameter value of the disturbance during measurement;
a diagnostic unit that evaluates an abnormality of the disturbance based on the parameter value of the disturbance during the measurement.

According to the above configuration, it is possible to quantitatively estimate the parameter value of the disturbance during measurement. Therefore, it is possible to evaluate whether or not there is an abnormality in the disturbance based on the parameter value without directly evaluating the disturbance.

In the anomaly detection device according to one embodiment, the disturbance model may be a transfer function determined based on the measured disturbance trigger signal and the calculated estimated disturbance control amount, and the disturbance parameter is the transfer function. It may be the coefficients of the numerator and denominator polynomials. According to the above configuration, the disturbance model can be appropriately estimated as a transfer function.

Further, in the anomaly detection device according to one embodiment, the disturbance model may be a transfer function determined based on the measured disturbance trigger signal and the calculated estimated disturbance control amount, and the disturbance parameter is a transfer function. Roots of the numerator and denominator polynomials of the function, or inverses of roots may be used. According to the above configuration, the model can be properly estimated as a transfer function by another method.

In addition, in the disturbance controlled variable waveform estimator of the anomaly detection device according to one embodiment, a first estimating device controlled variable calculated from the controlled object model and the manipulated variable is subtracted from the controlled variable. A first estimated disturbance control amount may be estimated, and the parameter estimation unit estimates a parameter value of the disturbance at the time of measurement from the first estimated disturbance control amount and the disturbance trigger signal. good too. According to the above configuration, it is possible to appropriately estimate the parameter value of the disturbance during measurement.

Further, in the disturbance controlled variable waveform estimator of the anomaly detection device according to one embodiment, a second estimating device controlled variable, which is a component of the controlled variable from which fluctuation due to disturbance is removed, is subtracted from the controlled variable. may estimate a second estimated disturbance control amount, and the parameter estimator may estimate a parameter value of the disturbance at the time of measurement from the second estimated disturbance control amount and the disturbance trigger signal. good. According to the above configuration, it is possible to appropriately estimate the parameter value of the disturbance during measurement using another method.

Further, in the storage unit of the abnormality detection device according to one embodiment, the disturbance parameter value when the disturbance is normal may be stored. A determination unit may be provided that sets a threshold range and determines whether the disturbance is abnormal depending on whether or not the parameter value of the disturbance during the measurement is within the first threshold range. According to the above configuration, it is possible to appropriately determine whether or not there is an abnormality in the disturbance during measurement in consideration of the parameter value of the disturbance when the disturbance is normal.

Further, in the anomaly detection device according to one embodiment, the number of disturbance parameters may be n (n is a positive integer), and the storage unit stores a disturbance parameter value when the disturbance is normal. and m types (m is a positive integer) of disturbance parameter values when the disturbance is abnormal, and in the diagnosis unit, n-dimensional parameters corresponding to the number of parameters of n disturbance models In the space, a normal point corresponding to the parameter value of the disturbance in the normal state, and m abnormal mode points respectively corresponding to m types of disturbance parameter values in the abnormal state and associated with the cause of the abnormality Along with positioning, a measurement point corresponding to the parameter value of the disturbance during the measurement is positioned, and a measurement vector directed from the normal point to the measurement point and each abnormal mode point vector directed from the normal point to each abnormal mode point are set. and a processing unit for evaluating a contribution rate of the measurement vector to each of the abnormal mode point vectors, and the degree of abnormality of the disturbance may be diagnosed for each cause based on the contribution rate.

According to the above configuration, each abnormal mode point corresponds to the parameter value of the disturbance when the disturbance is abnormal, and is associated with the cause of the abnormality. The contribution rate represents the degree of abnormality for each cause. Therefore, by evaluating the contribution rate of the measurement vector to each abnormal mode point vector by the processing section, the degree of disturbance abnormality based on the contribution rate is diagnosed for each cause by the diagnosis section.

Furthermore, in the processing unit of the abnormality detection device according to one embodiment, the contribution rate may be evaluated by calculating a cosine of an angle between the measurement vector and each of the abnormal mode point vectors. According to the above configuration, the contribution rate of the measurement vector to each abnormal mode point vector is obtained by projecting the measurement vector onto each abnormal mode point vector.

Further, in the processing unit of the abnormality detection device according to one embodiment, the contribution rate may be evaluated by decomposing the measurement vector by each abnormality mode point vector. According to the above configuration, the contribution rate of the measurement vector to each abnormal mode point vector is obtained by decomposing the measurement vector into each abnormal mode point vector.

In addition, in the processing unit of the anomaly detection device according to one embodiment, in the n-dimensional parameter space, the outline of the allowable range of the degree of anomaly is set as a threshold line of the parameter at which the contribution rate is constant, The diagnosis unit may include a judgment unit that judges abnormality according to whether or not the measurement point exists within the allowable range. According to the above configuration, the determination unit determines the abnormality of the disturbance based on whether the contribution rate evaluated by the processing unit is on or over the threshold line. Therefore, not only is the occurrence of an abnormality detected, but also its cause is estimated.

Then, in the abnormality detection device according to one embodiment, the processing unit sets the length of each abnormal mode point vector to Vref, and the length of the vector of the representative point included in the normal range from the normal point to Vrep , where k is a predetermined coefficient, the threshold Vth, which is the outline of the allowable range of the degree of abnormality, is calculated by the following formula,
Vth=(Vref−Vrep)×k+Vrep
The diagnosis unit may include a determination unit that determines abnormality of the disturbance according to whether or not the measurement point exists within the allowable range. According to the above configuration, the threshold value can be set to the parameter value before reaching each abnormal mode point, and it is estimated that there is a high possibility that an abnormality will occur and what the cause of the abnormality is.

Further, in order to solve the above problems, an abnormality detection device according to an embodiment of the present disclosure is configured such that an operation amount output by a control amount adjuster and a disturbance trigger signal input to a disturbance generation device to generate a disturbance are Based on, an abnormality detection device for detecting an abnormality of the disturbance,
a storage unit that stores a set value that is a manipulated variable in a state where the controlled variable output by the controlled object is stable;
a manipulated variable variation calculation unit that calculates a manipulated variable variation during disturbance measurement based on the manipulated variable, the set value, and the disturbance trigger signal;
a diagnostic unit that determines an abnormality in the manipulated variable variation during the measurement,
The storage unit stores the operation amount fluctuation when the disturbance is normal,
The diagnosis unit sets a second threshold range from the operation amount fluctuation in the normal state, and detects an abnormality of the disturbance according to whether or not the operation amount fluctuation at the time of measurement is within the second threshold range. judge.

According to the above configuration, without creating a disturbance model, the presence or absence of an abnormality in the disturbance is determined based on the set value of the manipulated variable and the manipulated variable fluctuation during measurement of the disturbance with respect to the manipulated variable fluctuation when the disturbance is normal. can do.

Furthermore, in the anomaly detection device according to one embodiment, the operation amount fluctuation may be an integral value of a change amount of the operation amount with respect to the set value. According to the above configuration, it is possible to appropriately determine whether there is an abnormality based on the integrated value of the amount of change in the manipulated variable with respect to the set value.

Further, in the abnormality detection device according to one embodiment, the operation amount fluctuation may be a change amount of the operation amount with respect to the set value. According to the above configuration, it is possible to appropriately determine whether or not there is an abnormality by another method based on the amount of change in the manipulated variable with respect to the set value.

In the abnormality detection device according to one embodiment, the control amount adjuster may be a temperature adjuster.

In order to solve the above problems, an anomaly detection method according to an embodiment of the present disclosure includes:
A parameter of a model of the controlled object is estimated based on the manipulated variable output by the controlled variable adjuster and the controlled variable output by the controlled object, and the manipulated variable, the controlled variable, and the disturbance generator are inputted. An abnormality detection method for estimating a parameter of a model of a disturbance control amount caused by the disturbance based on a disturbance trigger signal that generates a disturbance, and detecting an abnormality of the disturbance based on the parameter,
a step of storing the model of the disturbance control amount and parameters of the disturbance in the model of the disturbance control amount in a storage unit;
a step of estimating an estimated disturbance controlled variable by a disturbance controlled variable waveform estimator from the controlled variable and an estimated device controlled variable calculated by subtracting the disturbance controlled variable from the controlled variable;
a step of estimating a parameter value of the disturbance during measurement by a parameter estimating unit based on the estimated disturbance control amount calculated in a predetermined measurement period based on the disturbance trigger signal;
and evaluating an abnormality of the disturbance by a diagnosis unit based on a parameter value of the disturbance during the measurement.

According to the above configuration, it is possible to quantitatively estimate the parameter value of the disturbance during measurement. Therefore, it is possible to evaluate whether or not there is an abnormality in the disturbance based on the parameter value without directly evaluating the disturbance.

In order to solve the above problems, one embodiment of the present disclosure is implemented as a program for causing a computer to execute the above abnormality detection method. The anomaly detection method is implemented by causing a computer to execute the program of the present disclosure.

Advantageous Effects of Invention According to the present disclosure, it is possible to provide an abnormality detection device, an abnormality detection method, and a program capable of easily estimating the presence or absence of an abnormality in disturbances that occur intermittently in a controlled object.

FIG. 1 is a block diagram showing a schematic configuration of a control system for an abnormality detection device according to Embodiment 1 of the present disclosure. FIG. 2 is a diagram showing an example of changes in control amount over time. FIG. 3 is a diagram showing an example of changes in the manipulated variable over time. FIG. 4A is a diagram showing an example of changes in the control amount over time, FIG. 4B is a diagram showing an example of changes in the manipulated variable over time, and FIG. FIG. 4D shows an example of a change in the first estimated disturbance control amount over time, and FIG. 4E is a change in disturbance trigger signal over time. It is a figure which shows an example. FIG. 5 is a diagram showing an example of disturbance parameter values. FIG. 6 is a flow chart showing an operation example of the control system according to the first embodiment. FIG. 7 is a block diagram showing a schematic configuration of a control system according to Embodiment 2. FIG. 8(A) is a block diagram showing a feedback control system in Embodiment 2, FIG. 8(B) is a block diagram constituting part of FIG. 8(A), and FIG. 8(C) is a block diagram of FIG. 2 is a block diagram of a part thereof; FIG. FIG. 9A is a diagram showing an example of changes in the control amount over time, FIG. 9B is a diagram showing an example of changes in the second estimated device control amount over time, and FIG. 8A to 8C are diagrams showing an example of changes in the second estimated disturbance control amount with respect to progress, respectively. FIG. 10 is a block diagram showing a schematic configuration of a control system for an abnormality detection device according to Embodiment 3 of the present disclosure. FIG. 11 is a diagram for explaining vectorization processing according to the third embodiment. FIG. 12A is a diagram showing an example of numerical values of each parameter value, and FIG. 12B is a diagram showing an example of numerical values of each vector. FIG. 13 is a graph showing the parameter values of FIG. 12(A) as two-dimensional parameters. FIG. 14 is a diagram showing the results of calculating the moisture content abnormality contribution rate and the position deviation contribution rate of the measurement vector. FIG. 15 is a diagram for explaining vectorization processing in the fourth embodiment. FIG. 16 is a diagram showing results of calculation of the moisture content abnormality contribution rate and the position deviation contribution rate of the measurement vector based on each vector shown in FIG. 12(B). 17 is a block diagram showing a schematic configuration of a control system according to Embodiment 5. FIG. FIG. 18 is a diagram showing an example of an allowable range of moisture content anomaly and positional deviation when calculating a contribution rate by projecting a measurement vector onto an anomaly moisture content vector or the like. FIG. 19 is a diagram showing an example of an allowable range of moisture content anomalies and positional deviations when a measurement vector is represented by combining components of a positional deviation vector and moisture content anomaly vectors to calculate a contribution rate. FIG. 20 is a flow chart showing an operation example of the control system according to the fifth embodiment. FIG. 21 is a diagram showing an example of plotting parameter values of three types of causes of anomalies in a two-dimensional parameter space in the sixth embodiment. 22 is a diagram showing an example in which parameter values of three types of abnormal mode points are plotted in a three-dimensional parameter space in Embodiment 6. FIG. 23 is a diagram showing another example in which parameter values of three types of abnormal mode points are plotted in the three-dimensional parameter space in Embodiment 6. FIG. FIG. 24 is a diagram for explaining a threshold determination method according to the seventh embodiment. FIG. 25 is a block diagram showing a schematic configuration of a control system for an abnormality detection device according to Embodiment 8 of the present disclosure. FIG. 26A is a diagram showing an example of changes in the manipulated variable over time, FIG. 26B is a diagram showing an example of changes in the first estimated device controlled variable over time, and FIG. It is a figure which shows an example of a trigger signal. FIG. 27 is a diagram showing an example of the calculation result of the integral value of the manipulated variable change amount. FIG. 28 is a flow chart showing an operation example of the control system according to the eighth embodiment.

An embodiment of a control system relating to an abnormality detection device of the present disclosure will be described with reference to the accompanying drawings.

(Embodiment 1)
First, Embodiment 1 of a control system relating to an abnormality detection device of the present disclosure will be described with reference to FIGS. 1 to 6. FIG.

<Schematic configuration of the control system>
FIG. 1 is a block diagram showing a schematic configuration of a control system 100 relating to an abnormality detection device 2 according to Embodiment 1. As shown in FIG. As shown in FIG. 1, the control system 100 includes a control amount adjuster 1, a controlled object 3, an abnormality detection device 2, and a disturbance generation device 4.

The controlled object 3 is a device or the like that can be controlled by feedback control. As an example of the controlled object 3, when a temperature controller is used as the controlled variable controller 1, a heater capable of heating, an object to be heated by the heater, and the temperature of the object to be heated can be measured. An apparatus including a temperature sensor capable of controlling temperature is conceivable, such as a semiconductor manufacturing apparatus.

A control system including the controlled variable adjuster 1, the controlled object 3, and the disturbance generator 4 is a control system that feedback-controls the controlled object. Feedback control is a control that brings the controlled variable y of the controlled object 3 closer to the target value r by inputting the manipulated variable u determined based on the controlled variable y of the controlled object 3 to the controlled object 3 .

If there is a disturbance regardless of whether it is normal or abnormal, the controlled variable y can be represented by the sum of the controlled variable that brings the controlled variable of the controlled object closer to the target value and the controlled variable caused by the influence of the disturbance. Here, assuming that the disturbance control amount yd is caused by the influence of the disturbance generated by the disturbance generator 4, the control amount y can be expressed as y=yr+yd. Let this yr be the device control amount. The device controlled variable yr is the same as the controlled variable y when there is no disturbance. If the disturbance-free state continues, the controlled variable y becomes equal to or extremely close to the target value r by feedback control. If a disturbance occurs in this state, a difference between the controlled variable y and the target value r occurs. Therefore, feedback control is performed to control the value of the device controlled variable yr so that the controlled variable y approaches the target value r. .

The controlled variable adjuster 1 determines the manipulated variable u to be input to the controlled target 3 based on the controlled variable y of the controlled target 3, the target value r of the controlled variable, and the control parameters. The control parameter is a parameter used for feedback control, and in this embodiment, for example, a control parameter in PID (Proportional-Integral-Derivative Controller) control is used. Control parameters include proportional gain, integral gain, derivative gain, integral time, derivative time, and the like. Control parameters are appropriately determined based on characteristic information such as process gain, dead time, and time constant that indicate the characteristics of the controlled object 3 . An example of the controlled variable adjuster 1 is a temperature regulator, and when a semiconductor manufacturing apparatus is used as the controlled object 3, the target value r and the controlled variable y are the temperature, and the manipulated variable u is the usage amount of the heater. rate can be considered as an example.

The disturbance generation device 4 is a device that generates a disturbance that changes the control amount of the controlled object 3 in response to the input of a disturbance trigger signal, for example. As an example of the disturbance generator 4, when a temperature adjuster is used as the controlled variable adjuster 1 and a semiconductor manufacturing apparatus is used as the controlled object 3, a device for loading a semiconductor wafer into the semiconductor manufacturing apparatus can be considered. In this case, for example, a disturbance means that a semiconductor wafer is put into a semiconductor manufacturing apparatus. By loading a semiconductor wafer into a semiconductor manufacturing apparatus, heat absorption or heat dissipation occurs in the semiconductor manufacturing apparatus. Also, the amount of heat required to maintain the temperature may fluctuate. Therefore, the amount or rate of use of the heater of the temperature controller and the temperature of the semiconductor manufacturing equipment (for example, the ambient temperature in the semiconductor manufacturing equipment or the temperature of the part whose temperature is measured by the temperature sensor), that is, the controlled variable The manipulated variable u of the regulator 1 and the controlled variable y of the controlled object 3 fluctuate.

The abnormality detection device 2 includes a storage section 20 , a disturbance waveform estimation section 21 , a parameter estimation section 22 , a diagnosis section 23 and a display section 24 .

The storage unit 20 is configured by a storage device such as a magnetic hard disk device or a semiconductor storage device. The storage unit 20 includes a model storage area 201 , an operation amount storage area 202 , a control amount storage area 203 , a disturbance trigger signal storage area 204 , a normal parameter value storage area 205 and a threshold storage area 206 . The model storage area 201 stores a model of the controlled object 3, q parameters (q is a positive integer) in the model of the controlled object 3, a disturbance model, and n parameters (n is a positive integer) in the model of the disturbance. ) parameters are stored. The model of the controlled object 3, the model of the disturbance, and the parameters of each model will be described later. The manipulated variable storage area 202 stores the measured manipulated variable of the controlled variable regulator. The controlled variable storage area 203 stores the measured controlled variable y of the controlled object 3 . The disturbance trigger signal d input to the disturbance generator 4 is stored in the disturbance trigger signal storage area 204 . The normal parameter value storage area 205 stores the parameter values of the model when the disturbance is normal (hereinafter referred to as normal parameter values).

The disturbance waveform estimator 21 is composed of a CPU and the like. The disturbance waveform estimation unit 21 stores the model of the controlled object 3 stored in the model storage area 201, the operation amount u at the predetermined time stored in the operation amount storage area 202, and the predetermined time stored in the control amount storage area 203. The estimated apparatus controlled variable is calculated based on the controlled variable y of . Then, an estimated disturbance controlled variable is calculated based on the controlled variable y and the calculated estimated device controlled variable. The measurement of the manipulated variable u of the controlled variable adjuster 1 at a predetermined time and the measurement of the controlled variable y of the controlled object 3 at a predetermined time may be performed by the disturbance waveform estimation section 21 or by a control section (not shown). may

The parameter estimator 22 is composed of a CPU and the like. The parameter estimator 22 estimates a disturbance parameter value at a predetermined time in the disturbance model based on the calculated estimated disturbance control amount and the disturbance trigger signal d at a predetermined time stored in the disturbance trigger signal storage area 204. do. The measurement of the disturbance trigger signal d at a predetermined time may be performed by the parameter estimator 22 or may be performed by a controller (not shown).

The diagnosis unit 23 is composed of a CPU and the like. The diagnosis unit 23 detects an abnormality based on the disturbance parameter value estimated with respect to the disturbance during normal operation and the disturbance parameter value estimated separately. The details of the determination of abnormality related to disturbance will be described later.

The display unit 24 is configured by a liquid crystal display or the like, and displays information regarding the disturbance abnormality determined by the diagnosis unit 23 .

<Model to be controlled>
Next, a model of the controlled object 3 will be described with reference to FIGS. 2 and 3. FIG. FIG. 2 is a diagram showing an example of changes in the control amount y over time. FIG. 3 is a diagram showing an example of changes in the manipulated variable u over time.

The time axis in FIG. 2 and the time axis in FIG. 3 correspond to each other. As shown in FIGS. 2 and 3, the manipulated variable u is large in the period from 0 to t1 when the controlled variable y is low, but as the controlled variable y increases from t1 to t2, that is, the controlled variable y reaches the target. It can be seen that the manipulated variable u becomes smaller as it approaches the value r.

In this way, when a certain target value r is input, the controlled variable adjuster 1 outputs the manipulated variable u for manipulating the controlled object 3 so as to change the controlled variable y of the controlled object 3 to the target value r. It is a device that The controlled object 3 changes and outputs the apparatus controlled variable yr based on the manipulated variable u input from the controlled variable adjuster 1 . Further, the controlled object 3 is configured to feed back to the controlled variable adjuster 1 a controlled variable y, which is the sum of the device controlled variable yr and the disturbance controlled variable yd generated by the disturbance generator 4 . Therefore, the controlled variable adjuster 1 can appropriately change the output manipulated variable u in order to change the controlled variable y to the target value r based on the change in the controlled variable y.

As an example, consider a case where the control amount adjuster 1 is used as a temperature adjuster and a semiconductor manufacturing apparatus is used as the controlled object 3 . A semiconductor manufacturing apparatus includes a heater capable of heating, an object to be heated by the heater, and a temperature sensor capable of measuring the temperature of the object to be heated. It is a device in which an element is generated (for example, a semiconductor wafer is loaded). In this case, the target value r is the target temperature of the object to be heated, and the manipulated variable u is the usage amount or usage rate of the heater. Also, the controlled variable y is the temperature of the object to be heated measured by the temperature sensor. When a disturbance occurs, that is, when a semiconductor wafer is loaded, the semiconductor wafer absorbs heat from the object to be heated or releases heat to the object to be heated. As a result, the controlled variable y, which is the temperature, rises or falls.

Here, the Laplace transform of the manipulated variable u is defined as the manipulated variable U(s), and the same applies to the controlled variable Y(s). The manipulated variable is U(s), the controlled variable is Y(s), the difference between the target value and the controlled variable is E(s), the transfer function of the controlled variable adjuster is C(s), and the transmission of the controlled object Assuming that the function is G(s), the relationship between the manipulated variable U(s), the controlled variable Y(s), and the deviation E(s) can be expressed as the following (Formula 1) and (Formula 2): .

The transfer function C(s) takes as input the deviation E(s), which is the difference between the target value and the feedback control amount Y(s), and the output is the manipulated variable U(s). is taken into consideration, the deviation E(s) is determined to be zero (the target value and the controlled variable match). The transfer function G(s) changes Y(s) according to the operation amount U(s) when the operation amount U(s) is the input and the control amount Y(s) is the output. That is, the manipulated variable U(s) is calculated by the transfer function C(s) so that the deviation E(s) becomes zero, and the manipulated variable U(s) is input to the transfer function G(s). The control amount Y(s) is designed to approach the target value. Here, since the transfer function G(s) has the manipulated variable U(s) as an input and the controlled variable Y(s) as an output, it can be considered as a model of the controlled object 3. .

Such a transfer function, that is, a model of the controlled object 3 can be represented by a mathematical formula as shown in (Equation 3) below. In Equation 3 below, K _a and T _b are transfer function parameters, K _a is the system gain, and T _b is the time constant. Also, a is an integer greater than or equal to 0, b is a positive integer, and s is the Laplace operator.

(Equation 3) can also be transformed into (Equation 4). Here, d _a and c _b , which are the roots of the denominator polynomial, the numerator polynomial, or the reciprocals of the roots, are parameters of the transfer function, respectively, and can be used instead of K _a and T _b described above.

In this embodiment, for the sake of simplification, an example will be described in which the transfer function shown in the following (Equation 5) when a=0 and b=1 in (Equation 3) is adopted as a model of the controlled object 3.

<Acquisition and storage of parameter values to be controlled>
Acquisition and storage of parameter values of the model of the controlled object 3 in the anomaly detection device 2 of the present embodiment will now be described. First, the manipulated variable u of the controlled variable regulator 1 and the controlled variable y of the controlled target 3 are measured when the controlled target 3 is in a normal state and is not affected by disturbance. Then, based on the manipulated variable u and the controlled variable y measured for a predetermined period, the controlled object parameter values (K ₀ , T ₁ ) in the model of the controlled object 3 in (Equation 5) described above are obtained. Here, the predetermined period is assumed to be a period from time 0 to time t2 shown in FIGS. 2 and 3, for example. The model of the controlled object 3 and the acquired controlled object parameter values are stored in the model storage area 201 of the storage unit 20 .

In this embodiment, two parameters, K ₀ and T ₁ , are used as an example, but q parameters may be acquired and stored in the model storage area 201 depending on the model to be adopted. For example, in the case of the model of (Formula 3) described above, q=a+b+1.

<Disturbance model>
Next, a model of the controlled object 3 will be described with reference to FIGS. 4(A) to 4(E). FIGS. 4(A) to 4(E) are graphs showing the control amount and the like in this embodiment.

In this embodiment, the first estimated device controlled variable _yra is calculated from the manipulated variable u and the controlled variable y by removing the influence of the disturbance controlled variable _yd . make an estimate.

As described above, the controlled variable y can be represented by the sum of the device controlled variable yr and the disturbance controlled variable yd, which is the controlled variable due to the influence of the disturbance. Here, the apparatus controlled variable yr can be estimated from the above model of the controlled object and the manipulated variable u. In this embodiment, the device controlled variable yr is an output when the manipulated variable u is input to the model of the controlled object 3 . This estimated value is set as the first estimated device controlled variable _yra . The first estimated device controlled variable yr _a means the controlled variable of the device itself of the controlled object 3, and is the controlled variable considered to be due to the device of the controlled object 3 among the controlled variables y. The disturbance controlled variable is estimated by subtracting the first estimated apparatus controlled variable _yra from the controlled variable y. Assuming that the estimated value is the first estimated disturbance control amount yd _a , yd _a =y−yr _a .

FIG. 4A shows an example of changes in the control amount y over time. FIG. 4B shows an example of changes in the manipulated variable u over time. FIG. 4C shows an example of changes in the first estimated device controlled variable _yr a over time. FIG. 4D shows an example of changes in the first estimated disturbance control amount _yd a over time. FIG. 4(E) shows an example of changes in the disturbance trigger signal d over time.

The time axes in FIGS. 4A to 4E correspond to each other. A period from 0 to t11 is a period from when the device is started until a disturbance occurs. As shown in FIGS. 4A and 4C, the first estimated apparatus controlled variable _yra is the same as the controlled variable y during the period.

At t11, the disturbance trigger signal d is input to the disturbance generator 4, and the disturbance control amount is output. As a result, the controlled variable y in FIG. 4A and the first estimated disturbance controlled variable yd _a in FIG. 4D fluctuate to the negative side. Also, since the controlled variable y is lower than the target value r, the manipulated variable u is increased by feedback control.

Since the manipulated variable u has increased, the first estimated apparatus controlled variable yr _a increases at t12. As a result, the controlled variable y increases.

At t13, the controlled variable y coincides with or very close to the target value r, and then stabilizes. Therefore, although the manipulated variable u also maintains a constant value, the manipulated variable increases compared to immediately before t11.

At t14, the input of the disturbance trigger signal d to the disturbance generator 4 stops, and similarly the output of the disturbance control amount stops. As a result, the controlled variable y and the first estimated disturbance controlled variable _yda fluctuate toward the positive side. Also, since the controlled variable y has become higher than the target value r, the manipulated variable u is decreased by feedback control.

Since the manipulated variable u has decreased, the first estimated apparatus controlled variable yr _a decreases at t15. As a result, the controlled variable y decreases.

At t16, the controlled variable y matches or approaches the target value r, and then stabilizes. Therefore, the manipulated variable u also maintains a constant value. The value of the manipulated variable u at this time is equal to or close to the value immediately before t11.

As shown from t16 onward in FIGS. 4(A) to 4(E), a similar change occurs each time a disturbance trigger signal is input or not.

A transfer function related to disturbance, that is, a model of disturbance can be represented by a formula as shown in (Equation 3) or (Equation 4) above. In this embodiment, for the sake of simplification, similar to the model of the controlled object, an example in which the transfer function shown in (Equation 5) when a=0 and b=1 in (Equation 3) is adopted as a disturbance model. explain.

<Acquisition and storage of disturbance parameter values>
Acquisition and storage of the parameter values of the disturbance model in the anomaly detection device 2 of the present embodiment will now be described. The parameter values can be obtained in a similar manner to the model of the controlled object 3 and its parameters as described above.

First, the first estimated apparatus controlled variable yra is calculated from the model of the controlled object 3 obtained as described above and the manipulated variable _u . Next, when the disturbance is normal, the first estimated device controlled variable _yra is subtracted from the controlled variable y to estimate the first estimated disturbance controlled variable yd _a . It also acquires a disturbance trigger signal. Then, based on the first estimated disturbance control amount yd _a and the disturbance trigger signal d estimated for a predetermined period, the disturbance parameter values (K ₀ , T ₁ ) respectively. Here, the predetermined period is assumed to be, for example, the period during which the disturbance trigger signal d is input, that is, the period from time t11 to time t14 shown in FIGS. 4(A) to 4(E). Alternatively, it may be a period from time t11 to any time between time t13 and time t14. The disturbance model is stored in the model storage area 201 of the storage unit 20 , and the acquired normal state disturbance parameter values are stored in the normal parameter value storage area 205 of the storage unit 20 .

In this embodiment, two parameters, K ₀ and T ₁ , are used as an example. good too. For example, in the case of the model of (Equation 3) described above, n=a+b+1.

<Abnormal judgment>
In the present embodiment, the abnormality is determined by comparing the disturbance parameter value in the normal state and the disturbance parameter value estimated when the disturbance occurs. A disturbance parameter value estimated when a disturbance occurs is hereinafter referred to as a disturbance parameter value at the time of measurement. The abnormality determination processing in this embodiment will be described below.

FIG. 5 is a diagram illustrating an example of normality and abnormality of disturbance parameter values. FIG. 5 shows a two-dimensional parameter space in which the vertical axis is the system gain _K0 , which is one of the disturbance parameters, and the horizontal axis is the time constant _T1, which is another parameter. In a mode in which the control amount adjuster 1 is used as a temperature adjuster and the controlled object 3 is a semiconductor manufacturing apparatus, the unit of the system gain _K0 is the deviation of the target temperature [° C.] from the target temperature [° C.] and the control amount ( Temperature [°C]) is expressed as a percentage [%] of the difference, [°C/%]. The unit of the time constant _T1 is [seconds (s)].

In FIG. 5, a point Aa is a point obtained by plotting the normal disturbance parameter values stored in the normal parameter value storage area 205 of the storage unit 20 in a two-dimensional parameter space. Points Ba and Bb are points obtained by plotting in a two-dimensional parameter space the disturbance parameter values at the time of predetermined measurement estimated based on the estimated disturbance control amount and the disturbance trigger signal d estimated during the measurement period.

By setting a threshold for the point Aa, it is possible to determine whether it is normal or abnormal. Let this threshold be the first threshold. For example, as shown in FIG. 5, predetermined thresholds may be set as upper and lower limits for system gain _K0 and time constant _T1 . In this case, the range enclosed by the solid line is the range of the first threshold. If it is out of the range of the threshold value, like the points Ba and Bb, it can be determined that there is an abnormality. Of course, instead of setting the upper limit value and the lower limit value, the threshold value may be set by another method such as applying a function corresponding to the threshold value to the normal disturbance parameter value.

In this embodiment, there are two parameters, K ₀ and T ₁ , so the space is a two-dimensional space.

When the disturbance parameter value at the time of measurement exceeds a threshold set based on the disturbance parameter value at the time of normality, the display unit 24 displays so that the user can determine the occurrence of an abnormality. In addition, it may be possible to display how much difference there is with respect to the disturbance parameter in the normal state or with respect to its threshold value. By looking at the display unit 24, the user can determine whether an abnormality has occurred or whether there is a high possibility that an abnormality will occur. In addition, it is not always necessary to display it, and when an abnormality occurs, the user may be notified by another means such as sounding an alarm, or the apparatus may be automatically stopped.

<Operation example>
Next, an operation example of the control system 100 regarding the abnormality detection device 2 in this embodiment will be described with reference to the flowchart of FIG. FIG. 6 is a flow chart showing an operation example of the control system 100 regarding the abnormality detection device in this embodiment.

First, when the control system 100 of the present embodiment is powered on and started, for example, the diagnostic unit 23 measures the controlled variable y of the controlled object 3 in a predetermined period ( FIG. 6 : S10). Next, the controlled object parameter values of the model of the controlled object 3 are obtained based on the manipulated variable u of the controlled quantity adjuster 1 and the controlled quantity y of the controlled object 3 ( FIG. 6 : S11).

After the manipulated variable u and the controlled variable y are stabilized, the disturbance generator 4 generates disturbance ( FIG. 6 : S12). Then, the disturbance waveform _estimating unit 21 calculates the first estimating device controlled variable yra from the model of the controlled object 3 and the manipulated variable u, and calculates the first estimating device controlled variable _yra and the controlled variable y. , the first estimated disturbance control amount yd _a is further estimated (FIG. 6: S13).

Next, the parameter estimator 22 estimates parameter values of the disturbance model from the first estimated disturbance control amount yd _a and the disturbance trigger signal d ( FIG. 6 : S14). If the normal disturbance parameter value is not stored, the estimated disturbance parameter value may be learned as the normal disturbance parameter value and stored in the normal parameter value storage area 205 of the storage unit 20 (FIG. 6). : S15, S151). After storing, the process returns to S12. Alternatively, the parameter value of the disturbance model may be estimated by measuring the control amount or the like in the normal disturbance in advance, and stored in the normal parameter value storage area 205 of the storage unit 20 .

When the normal disturbance parameter value is stored, the determination unit 231 of the diagnosis unit 23 compares the estimated disturbance parameter value as the disturbance parameter value at the time of measurement with the normal disturbance parameter value (FIG. 6: S16). The determination unit 231 determines that the disturbance is abnormal when the disturbance parameter value at the time of measurement exceeds a predetermined threshold set from the disturbance parameter value at the time of normal, and when it is within the predetermined threshold, the disturbance is It is determined that there is no abnormality, that is, that it is normal (FIG. 6: S17). If the disturbance is normal, the process returns to S12. If it is determined that there is an abnormality, for example, a message indicating that the disturbance is abnormal is displayed on the display unit 24, and the process ends (FIG. 6: S18).

In the present embodiment, the process returns to S12 when it is not determined to be abnormal, and the method of generating the next disturbance is described. Then, the process may be returned to S12 to generate the next disturbance. In addition, although the determination unit 231 determines abnormality/normality based on a predetermined threshold value, the diagnosis unit 23 evaluates the abnormality based on the disturbance parameter value at the time of measurement without the determination unit 231 determining, and the display unit It may be operated so that the user can make a decision based on the display at 24 .

As described above, according to the present embodiment, it is possible to determine whether intermittent disturbance is abnormal or not. Therefore, for example, in the case where the occurrence of disturbance means that some part is thrown into the controlled object 3 and the controlled object 3 processes the part, it is possible to detect the presence or absence of an abnormality without inspecting the part. .

(Embodiment 2)
Next, Embodiment 2 of the present disclosure will be described with reference to FIGS. 7 to 9. FIG. FIG. 7 is a block diagram showing a schematic configuration of a control system 100 relating to the abnormality detection device 2 according to the second embodiment. As shown in FIG. 7, the abnormality detection device 2 differs from the first embodiment described above with reference to FIG. 1 in that the storage unit 20 includes a target value storage area 207. The target value storage area 207 of the storage unit 20 stores the target value r.

FIG. 8A is a diagram showing a block diagram of feedback control according to this embodiment. FIG. 8(B) is a block diagram showing feedback control excluding disturbance in FIG. 8(A). FIG. 8(C) is a block diagram showing feedback control for disturbance in FIG. 8(A). When examining the transfer function, the case where there is no disturbance and the case where the target value r is zero are separately examined. 8(C) block diagram sum.

FIG. 9 shows the relationship between the control amount and the like regarding the block diagram as described above. FIG. 9(A) shows an example of changes in the control amount y over time, and shows part of the period from 0 to t13 and the period from t13 to t14 in FIG. 4(A). FIG. 9B shows an example of changes in the second estimated device controlled variable yr _b over time. FIG. 9C shows an example of changes in the second estimated disturbance control amount yd _b over time. Also, FIGS. 8A and 9A, 8B and 9B, and 8C and 9C correspond to each other.

As described in the first embodiment, the controlled variable y can be represented by the sum of the device controlled variable yr and the disturbance controlled variable yd, which is the controlled variable due to the influence of the disturbance. Here, the controlled variable y when there is no disturbance, that is, the device controlled variable yr basically does not fluctuate when the controlled variable y coincides with or is extremely close to the target value r and stabilizes in that state. Therefore, it is estimated that the device controlled variable yr changes as shown in FIG. 9B after the device is started. This estimated value is set as the second estimated device controlled variable yr _b . Further, although details will be described later, a value obtained by subtracting the second estimated device controlled variable yr _b from the controlled variable y is taken as a second estimated disturbance controlled variable yd _b .

As described in the first embodiment, the transfer function of the controlled variable adjuster 1 is C(s), the transfer function of the controlled object 3 is G(s), and the transfer function of the disturbance generator 4 is P(s). Then, the relationship between the target value R(s), the controlled variable Y(s), and the disturbance trigger signal D(s) can be expressed as in (Equation 6) below. FIG. 8A is a block diagram schematically showing (Equation 6). Note that, for example, the target value R(s) is the Laplace transform of the target value r, and the same applies to the controlled variable Y(s) and the like.

Further, (Equation 6) is the relationship between the second estimated device control amount YR _B (s) and the second estimated disturbance control amount YD _B (s) represented by (Equation 7) and (Equation 8) below. It can be considered as a summation. FIG. 8B is a block diagram schematically showing (Equation 7). FIG. 8(C) is a block diagram schematically showing (Formula 8).

The transfer function P(s) outputs a predetermined control amount when the disturbance trigger signal D(s) is input to the disturbance generator 4 .

The term related to the second estimated apparatus controlled variable YR _B (s) is a component of the controlled variable Y(s) from which fluctuations in the controlled variable due to the influence of disturbance, including feedback control, are removed. In other words, the second estimated apparatus controlled variable yr _b is a controlled variable obtained by feedback control for bringing the controlled variable y closer to the target value r when there is no disturbance.

The term related to the second estimated disturbance controlled variable YD _B (s) is the variable component of the controlled variable Y(s) due to the influence of the disturbance, including the feedback control. In other words, the second estimated disturbance control amount yd _b is a control amount obtained by feedback control for bringing the control amount fluctuated by the disturbance closer to zero when the target value is zero.

Further, as described above, the second estimated apparatus controlled variable yr _b basically does not fluctuate once the controlled variable y coincides with or is extremely close to the target value r and stabilizes in that state. Therefore, the second estimated apparatus controlled variable yr _b may be estimated based on the measured controlled variable at startup and the stable controlled variable after a predetermined time has elapsed since startup.

As is clear from (Equation 6) to (Equation 8), the second estimated disturbance controlled variable yd _b is the difference between the measured controlled variable y and the second estimated device controlled variable yr _b , that is, yd _b = y-yr _b . Thus, the second estimated disturbance control amount yd _b is estimated.

In this embodiment, instead of the first estimated device controlled variable yr _a and the first estimated disturbance controlled variable yd _a in the first embodiment, a second estimated device controlled variable yr _b and a second estimated disturbance controlled variable yd _b is used to estimate the parameter values of the disturbance model. Then, it is possible to detect disturbance abnormality by the method described in the first embodiment.

(Embodiment 3)
Next, Embodiment 3 of the present disclosure will be described with reference to FIGS. 10 to 14. FIG. FIG. 10 is a block diagram showing a schematic configuration of a control system 100 relating to the abnormality detection device 2 according to the third embodiment. As shown in FIG. 10, the storage unit 20 of the abnormality detection device 2 in this embodiment further includes an abnormality parameter value storage area 208 in addition to the first embodiment. In the abnormal parameter value storage area 208, there are m types (m is a positive integer) of disturbance parameter values (hereinafter referred to as disturbance parameter values during abnormal conditions) of the disturbance model in a state in which some abnormality has occurred in the disturbance. ) is stored. Although this embodiment describes a difference from the first embodiment, it is naturally possible to apply it to the second embodiment.

Diagnosis unit 23 further includes processing unit 232 . The processing unit 232 performs vectorization processing of each disturbance parameter value. The disturbance parameter values to be vectorized include the disturbance parameter value of the disturbance model in the normal state stored in the normal parameter value storage area 205, and the disturbance parameter value in the abnormal state stored in the abnormal parameter value storage area 208. contains m types of disturbance parameter values of the disturbance model in . Each disturbance parameter value also includes the disturbance parameter value at the time of measurement estimated by the parameter estimation unit 22 . Details of the vectorization processing in the processing unit 232 will be described later.

The diagnosis unit 23 determines the degree of abnormality in the disturbance based on the result of vectorization processing by the processing unit 232 . The details of determination of the degree of abnormality in disturbance will be described later.

The display unit 24 is configured by a liquid crystal display or the like, and displays information regarding the disturbance abnormality determined by the diagnosis unit 23, for example, the degree of abnormality.

In Embodiments 1 and 2, a disturbance parameter value of a normal disturbance is estimated and acquired, and the normal disturbance parameter value is compared with the separately estimated disturbance parameter value during measurement related to another disturbance. Therefore, an abnormality was determined. However, in the present embodiment, a disturbance parameter value is acquired in advance and stored in the abnormal parameter value storage area 208 of the storage unit 20 when some abnormality occurs in the disturbance. In a mode in which the controlled variable adjuster 1 is used as a temperature adjuster, the controlled object 3 is the semiconductor manufacturing apparatus, and the disturbance is caused by the input of a wafer (work), the temperature of the semiconductor manufacturing apparatus reaches the target value within a predetermined period after the wafer is input. , it may be determined that an abnormality has occurred.

Various causes are conceivable as the cause of an abnormality, but in this embodiment, as an example, an abnormality is caused by an abnormality in the amount of moisture in a wafer that is loaded as a disturbance, and the placement of the wafer that is loaded. When an abnormality occurs due to deviation of the position from the appropriate value, the disturbance parameter value at the time of the abnormality is acquired and stored in the abnormality parameter value storage area 208 .

If the correlation between the cause of the abnormality and the parameter value at the time of occurrence of the abnormality is known in advance, the parameter value estimated within a predetermined period from the occurrence of the disturbance during each use and the parameter value at the time of the abnormality occurrence can be calculated. , it is possible to estimate the cause of the abnormality. The predetermined period after the occurrence of the disturbance is, for example, the period from when the disturbance is applied until when it is removed. The period is assumed to be the period from time t11 to time t14 shown in FIGS. , is the measurement period. Also, the measurement period may be a period from the input of the disturbance until after the time set by the user has elapsed, for example, from time t11 to any time between time t13 and time t14.

In this embodiment, there are two types of causes of anomalies: abnormal water content in the wafer and misalignment of the wafer. should be stored.

<Vectorization processing>
In this embodiment, the cause of the abnormality is estimated by vectorizing the disturbance parameter value at the time of abnormality occurrence and the disturbance parameter value estimated at each time of use. The vectorization processing in this embodiment will be described below.

FIG. 11 is a diagram for explaining vectorization processing in this embodiment. As in FIG. 5, FIG. 11 shows a two-dimensional parameter space in which the vertical axis is the system gain _K0 , which is one of the disturbance parameters, and the horizontal axis is the time constant _T1, which is another parameter. In a mode in which the control amount adjuster 1 is used as a temperature adjuster and the controlled object 3 is a semiconductor manufacturing apparatus, the unit of the system gain _K0 is the deviation of the target temperature [° C.] from the target temperature [° C.] and the control amount ( Temperature [°C]) is expressed as a percentage [%] of the difference, [°C/%]. The unit of the time constant _T1 is [seconds (s)].

In FIG. 11, point A0 is a point obtained by plotting normal disturbance parameter values stored in normal parameter value storage area 205 of storage unit 20 in a two-dimensional parameter space. do.

In FIG. 11, a point A1 is a point obtained by plotting in a two-dimensional parameter space the disturbance parameter values when an abnormality occurs due to an abnormal water content of the wafer stored in the abnormal parameter value storage area 208 of the storage unit 20. , in the following description, this point is defined as the abnormal mode point of the moisture content abnormality.

In FIG. 11, a point A2 is a point plotted in a two-dimensional parameter space of the disturbance parameter values when an abnormality occurs due to the positional deviation of the wafer stored in the abnormal parameter value storage area 208 of the storage unit 20. In the following description, this point is assumed to be the abnormal mode point of misalignment.

Point B in FIG. 11 is a point where the disturbance parameter value estimated based on the estimated disturbance control amount and the disturbance trigger signal d estimated during the measurement period is plotted in a two-dimensional parameter space. is the measurement point.

An abnormality mode point A1 for moisture content abnormality and an abnormality mode point A2 for positional deviation represent how the disturbance parameter value changes from the normal point A0 when an abnormality occurs due to each cause. Similarly, measurement point B represents the amount of change in the disturbance parameter value from normal point A0. Therefore, the abnormal mode point A1 of the moisture content abnormality, the abnormal mode point A2 of the misalignment, and the disturbance parameter values of the measurement point B can be considered as vectors with the normal point A0 as the starting point.

In the following description, the vector of the abnormal mode point A1 of the moisture content abnormality starting from the normal point A0 is defined as the moisture content abnormality vector A1, and the vector of the abnormal mode point A2 of the positional deviation starting from the normal point A0 is the positional deviation vector. A2. In addition, the vector of the measurement point B starting from the normal point A0 is defined as the measurement vector B.

In this embodiment, in order to determine to what extent the measurement vector B is related to the moisture content abnormality vector A1 and the position deviation vector A2, respectively, the moisture content abnormality vector A1 and the position deviation vector of the measurement vector B Calculate the contribution rate to A2.

In calculating the contribution rate, the measurement vector B is projected onto the abnormal moisture amount vector A1 and the positional deviation vector A2, and the ratio of the projected vector to the abnormal moisture amount vector A1 and the positional deviation vector A2 is calculated.

A method of calculating the contribution rate of the measurement vector B to the moisture content abnormality vector A1 and the positional deviation vector A2 will be described below based on mathematical expressions.

First, when the disturbance parameter value of the normal point A0 is ( _T1A0 , _K0A0 ) and the disturbance parameter value of the abnormal mode point A1 of the moisture content abnormality is ( _T1A1 , _K0A1 ), the moisture content abnormality vector A1 is as follows. is represented as

Next, assuming that the disturbance parameter value of the abnormal mode point A2 of the positional deviation is ( _T1A2 , _K0A2 ), the positional deviation vector A2 is expressed as follows.

Further, when the disturbance parameter value of the measurement point B is (T _1B , K _0B ), the measurement vector B is expressed as follows.

In the following description, the disturbance parameter values (T _1A0 , K _0A0 ) of normal point A0 are assumed to be (0, 0) for simplification of description.

Next, the projection vector B A1 of the moisture content abnormality vector _A1 of the measurement vector B is expressed as follows.

Here, w1 is the contribution rate of the measurement vector B to the water content abnormality vector A1, and the contribution rate w1 is expressed as follows.

The magnitude of the moisture content anomaly vector A1 is expressed as follows.

The magnitude of the projection vector B _A1 can be expressed by the following equation from the cosine of the angle α between the projection vector B _A1 and the measurement vector B and the magnitude of the measurement vector B.

Moreover, the inner product of the moisture content abnormality vector A1 and the measurement vector B can be represented by the following formula.

Therefore, the contribution rate w1 can be obtained as follows from the above equations (15) and (16) and the above (13).

Similarly, the projection vector B A2 of the displacement vector _{A2 of} the measurement vector B is expressed as follows.

Here, w2 is the contribution rate of the measurement vector B to the positional deviation vector A2, and the contribution rate w2 is expressed as follows.

The magnitude of the positional deviation vector A2 is expressed as follows.

The magnitude of the projection vector B _A2 can be expressed by the following equation from the cosine of the angle β formed between the projection vector B _A2 and the measurement vector B and the magnitude of the measurement vector B.

Also, the inner product of the positional deviation vector A2 and the measurement vector B can be expressed by the following equation.

Therefore, the contribution rate w2 can be calculated as follows from the above equations (21) and (22) and the above (equation 19).

As is clear from the above equations (17) and (23), if the disturbance parameter value of the measurement point B is known, the disturbance at the abnormal mode point A1 of the moisture content abnormality and the abnormal mode point A2 of the positional deviation Based on the parameter values, the contribution rate of the measurement vector B to the moisture content abnormality vector A1 and the positional deviation vector A2 can be calculated.

In this embodiment, the formula (17) and the formula (23) are stored in the storage unit 20 as programs. The processing unit 232 calculates the above-described contribution rate based on the estimated disturbance control amount estimated by the disturbance generator 4, the normal disturbance parameter value, and the abnormal disturbance parameter value, according to the program.

The display unit 24 displays the contribution rate calculated as described above. By looking at the contribution rate displayed on the display unit 24, the user can determine whether an abnormality has occurred or whether there is a high possibility that an abnormality will occur. In addition, the user can estimate what causes the occurrence of the abnormality or the possibility of the occurrence of the abnormality.

<Numerical example>
Next, one numerical example of this embodiment will be described with reference to FIGS. 12 to 14. FIG. FIG. 12A is a diagram showing numerical examples of each disturbance parameter value (system gain K ₀ , time constant T ₁ ). No. in FIG. 1 to No. 3 indicates the disturbance parameter values of the normal point, the abnormal mode point of the moisture content abnormality, and the abnormal mode point of the positional deviation. No. 4 to No. At 8, the control amount adjuster 1 as the temperature adjuster and the controlled object 3 as the semiconductor manufacturing apparatus are activated five times by turning on the power, and measurement points 1 to 5 are measured at predetermined timings during each measurement period. The disturbance parameter value for point 5 is shown.

FIG. 12B is a diagram showing a numerical example of each vector. No. in FIG. 2 and no. 3 indicate the values of the moisture content anomaly vector and the positional deviation vector, respectively. No. 4 to No. 8 indicates the values of the measurement vectors 1 to 5 corresponding to the measurement points 1 to 5, respectively.

FIG. 13 is a graph showing the parameter values of FIG. 12(A) as two-dimensional parameters. FIG. 14 is a diagram showing results of calculation of the abnormal moisture amount contribution rate and the position deviation contribution rate of the measurement vectors 1 to 5 from the equations (17) and (23) described above.

As shown in FIG. 14, for measurement vector 1, the moisture content abnormality contribution rate is 40%, but the position deviation contribution rate is 110%, and the contribution rate to the position deviation vector exceeds 100%. Therefore, it is possible to presume that the cause of the occurrence of the abnormality is the positional deviation.

Further, as shown in FIG. 14, for measurement vector 2, the moisture content abnormality contribution rate is 44% and the position deviation contribution rate is 89%, which indicates that the moisture content abnormality and the position deviation are within the allowable range. However, the positional deviation contribution ratio is approaching 100%, and it is also possible to determine that countermeasures against the positional deviation are necessary.

Furthermore, as shown in FIG. 14, for the measurement vector 5, the moisture content abnormality contribution rate is 48%, and the position deviation contribution rate is 85%. However, the water content abnormality contribution rate is approaching 50%, and it is also possible to determine that it is necessary to deal with the water content abnormality.

As described above, according to the present embodiment, it is possible to present not only whether or not some abnormality has occurred in the disturbance that has occurred, but also the possibility of a plurality of causes for the occurrence of the abnormality. In addition, since the characteristics of the generated disturbance can be parameterized, it is possible to clarify not only the aging of the device, but also the difference in characteristics of devices having the same configuration.

(Embodiment 4)
Next, Embodiment 4 of the present disclosure will be described with reference to FIGS. 15 and 16. FIG. In the third embodiment, the measurement vector B is projected onto the moisture content abnormality vector A1 or the like, and the ratio of the projected vector to the moisture content abnormality vector A1 or the like is calculated, whereby the moisture content abnormality vector A1 or the like of the measurement vector B is calculated. We calculated the contribution rate to However, in the present embodiment, the measurement vector B is expressed by combining the components of the positional deviation vector A2 and the component of the moisture content abnormality vector A1, and each component is calculated as a contribution rate, unlike the third embodiment. there is

A method of calculating the contribution rate of the measurement vector B to the moisture content abnormality vector A1 and the positional deviation vector A2 in this embodiment will be described below based on mathematical expressions. FIG. 15 is a diagram for explaining vectorization processing in the fourth embodiment.

Also in this embodiment, the disturbance parameter values of the normal point A0 are ( _T1A0 , _K0A0 ), and the disturbance parameter values of the abnormal mode point A1 of the moisture content abnormality are ( _T1A1 , _K0A1 ). Also, the disturbance parameter value of the abnormal mode point A2 of positional deviation is (T _1A2 , K _0A2 ). The moisture content abnormality vector A1 is expressed as shown in (Equation 9) of the third embodiment, and the positional deviation vector A2 is expressed as shown in (Equation 10) of the third embodiment. Also, the disturbance parameter value of the measurement point B is set to (T _1B , K _0B ) as in the third embodiment, and the measurement vector B is expressed as shown in (Equation 11) of the third embodiment.

In the following description, the disturbance parameter values (T _1A0 , K _0A0 ) of normal point A0 are assumed to be (0, 0) for simplification of description.

In this embodiment, first, as shown in FIG. 15, the measurement vector B is decomposed into a component vector B _A2 in the positional deviation vector A2 and a component vector B _A1 in the moisture content abnormality vector A1. A component vector B _A2 in the displacement vector A2 is expressed as follows.

w2 is the contribution rate of the measurement vector B to the positional deviation vector A2, and the contribution rate w2 is expressed as follows.

On the other hand, the component vector B _A1 in the moisture content abnormality vector A1 of the measurement vector B is expressed as follows.

w1 is the contribution rate of the measurement vector B to the water content abnormality vector A1, and the contribution rate w1 is expressed as follows.

The measurement vector B is represented as follows by a component vector B _A2 in the positional deviation vector A2 and a component vector B _A1 in the moisture content abnormality vector A1.

The above equation (28) can be expressed as follows.

Therefore, from the above (Equation 29), the contribution ratios (w1, w2) can be expressed as follows.

Also, the right side of the above (Equation 30) can be expressed as follows.

Therefore, the contribution ratios (w1, w2) can be expressed as follows.

In this way, if the parameter value of the measurement point B is known, vectorization processing is performed based on the parameter values of the abnormal mode points A1 and A2 of the moisture content abnormality and the positional deviation, and the moisture content of the measurement vector B is calculated based on each vector. Contributions to the quantity anomaly vector A1 and the positional deviation vector A2 can be calculated.

In this embodiment, the formula (32) is stored in the storage unit 20 as a program. The processing unit 232 calculates the above-described contribution rate based on the control amount measured for the controlled object 3, the disturbance parameter value in the normal state, and the disturbance parameter value in the abnormal state according to the program.

The display unit 24 displays the contribution rate calculated as described above. By looking at the contribution rate displayed on the display unit 24, the user can determine whether an abnormality has occurred or whether there is a high possibility that an abnormality will occur. In addition, the user can estimate what causes the occurrence of the abnormality or the possibility of the occurrence of the abnormality.

<Numerical example>
Next, one numerical example of this embodiment will be described with reference to FIG. Numerical examples of each parameter value (system gain K ₀ , time constant T ₁ ) and each vector are the same as those in FIGS. 12A and 12B in the third embodiment.

FIG. 16 shows the results of calculating the abnormal water content contribution rate and the position deviation contribution rate of measurement vectors 1 to 5 from the above equation (Equation 32) based on the vectors shown in FIG. 12(B). It is a diagram.

As shown in FIG. 16, for measurement vector 1, the moisture content abnormality contribution rate is 18%, but the position deviation contribution rate is 110%, and the contribution rate to the position deviation vector exceeds 100%. Therefore, it is possible to presume that the cause of the occurrence of the abnormality is the positional deviation.

Further, as shown in FIG. 16, for the measurement vector 4, the moisture content abnormality contribution rate is 26% and the position deviation contribution rate is 92%, which indicates that the moisture content abnormality and the position deviation are within the allowable range. However, the positional deviation contribution ratio is approaching 100%, and it is also possible to determine that maintenance for the positional deviation is necessary.

As described above, according to the present embodiment, it is possible to present not only whether or not some abnormality has occurred in the disturbance that has occurred, but also the possibility of a plurality of causes for the occurrence of the abnormality. In addition, since the characteristics of the generated disturbance can be parameterized, it is possible to clarify not only the secular change of the device but also the difference in characteristics of devices having the same configuration.

(Embodiment 5)
Next, Embodiment 5 of the present disclosure will be described with reference to FIGS. 17 to 20. FIG. FIG. 17 is a block diagram showing a schematic configuration of a control system 100 relating to the abnormality detection device 2 according to the fifth embodiment. As shown in FIG. 17, the anomaly detection device 2 differs from the embodiment described above in that the diagnosis unit 23 is provided with a judgment unit 231 . Also, the storage unit 20 differs from the above-described embodiment in that it includes a threshold value storage area 206 .

In Embodiments 3 and 4 described above, the processing unit 232 of the diagnostic unit 23 calculates the contribution rate of the measurement vector B to the moisture content abnormality vector A1 and the positional deviation vector A2, and outputs the contribution rate to the display unit. Aspects have been described. On the other hand, this embodiment differs from the above-described embodiments in that a threshold value is set in advance for the contribution rate output from the processing unit 232, and it is determined that an abnormality has occurred when the threshold value is exceeded.

In the threshold storage area 206 of the storage unit 20, a preset threshold for moisture content abnormality and a preset threshold for positional deviation are stored. The threshold may be stored as a data string of disturbance parameter values or as a linear function.

The determination unit 231 determines whether or not an abnormality has occurred by comparing the contribution rate calculated by the processing unit 232 and the threshold stored in the threshold storage area 206 . The display unit 24 displays the determination result.

FIG. 18 shows the moisture content abnormality when calculating the contribution rate of the measurement vector B to the moisture content abnormality vector A1 or the like by projecting the measurement vector B described in Embodiment 3 onto the moisture content abnormality vector A1 or the like. FIG. 10 is a diagram showing an example of an allowable range of positional deviation;

FIG. 18 shows a threshold line Th1 at which the contribution rate of the measurement vector B to the moisture content abnormality vector A1 is 100%, and a threshold line Th2 at which the contribution rate of the measurement vector B to the positional deviation vector A2 is 100%. . Also shown are a threshold line Th3 at which the contribution rate of the measurement vector B to the abnormal moisture vector A1 is -100%, and a threshold line Th4 at which the contribution rate of the measurement vector B to the positional deviation vector A2 is -100%.

In the example shown in FIG. 18, the range from the threshold line Th1 to the threshold line Th3 is the allowable range for the abnormal moisture content, and the range from the threshold line Th2 to the threshold line Th4 is the allowable range for the positional deviation. Therefore, in the present embodiment, the determination unit 231 determines that there is no abnormality when the measurement point B falls within the area surrounded by the threshold lines Th1, Th2, Th3, and Th4 (not including the threshold lines). Further, the determination unit 231 determines that there is an abnormality when the measurement point B falls within the threshold lines Th1, Th2, Th3, and Th4 and the regions outside the respective threshold lines.

The threshold lines Th1, Th2, Th3, and Th4 are set as lines where the contribution rate of the measurement vector B to the abnormal moisture vector A1 or the positional deviation vector A2 is 100% or -100%. Therefore, even if the measurement vector B is on these threshold lines, or even if it exceeds these threshold lines, not only the contribution rate to either the water content abnormality vector A1 or the position deviation vector A2 , the contribution to the other becomes clear. As a result, the present embodiment can appropriately estimate the cause of the abnormality, compared to the case of independently determining each of the predetermined values of the system gain _K0 and the time constant _T1 , which are disturbance parameter values, as thresholds. can be done.

FIG. 19 shows the measurement vector B described in the fourth embodiment, expressed by combining the component of the positional deviation vector A2 and the component of the moisture content abnormality vector A1, and the moisture content abnormality when each component is calculated as the contribution rate. and FIG. 11 is a diagram showing an example of an allowable range of positional deviation.

FIG. 19 shows a threshold line Th5 at which the contribution rate of the measurement vector B to the moisture content abnormality vector A1 is 100%, and a threshold line Th6 at which the contribution rate of the measurement vector B to the positional deviation vector A2 is 100%. . Also shown are a threshold line Th7 at which the contribution rate of the measurement vector B to the moisture content abnormality vector A1 is -100%, and a threshold line Th8 at which the contribution rate of the measurement vector B to the positional deviation vector A2 is -100%.

In the example shown in FIG. 19, the range from the threshold line Th5 to the threshold line Th7 is the allowable range for the abnormal moisture content, and the range from the threshold line Th6 to the threshold line Th8 is the allowable range for the positional deviation. Therefore, in the present embodiment, the determination unit 231 determines that there is no abnormality when the measurement point B falls within the area surrounded by the threshold lines Th5, Th6, Th7, and Th8 (not including the threshold lines). Further, the determination unit 231 determines that there is an abnormality when the measurement point B falls within the threshold lines Th5, Th6, Th7, and Th8 and the regions outside the respective threshold lines.

The threshold lines Th5, Th6, Th7, and Th8 are set as lines where the contribution rate of the measurement vector B to the abnormal moisture vector A1 or the positional deviation vector A2 is 100% or -100%. Therefore, even if the measurement vector B is on these threshold lines, or even if it exceeds these threshold lines, not only the contribution rate to either the water content abnormality vector A1 or the position deviation vector A2 , the contribution to the other becomes clear. As a result, the present embodiment can appropriately estimate the cause of the abnormality, compared to the case of independently determining each of the predetermined values of the system gain _K0 and the time constant _T1 , which are disturbance parameter values, as thresholds. can be done.

As illustrated in FIGS. 18 and 19, different allowable ranges can be set as the allowable ranges for the moisture content anomaly and the positional deviation depending on the calculation method of the contribution rate. Depending on the purpose of use, the user can use the contribution rate calculation method described in Embodiment 3 (hereinafter referred to as the first method) or the contribution rate calculation method described in Embodiment 4 (hereinafter referred to as the second method). ), or both.

The first and second schemes each have advantages and disadvantages. The advantage of the first scheme is that there is no limit to the number of factors that can cause anomalies. Regardless of the number of parameters, that is, the number of dimensions n in the n-dimensional space, multiple elements can be set as the factors that cause anomalies. A drawback of the first method is that it is unknown what is being detected when anomalies occur simultaneously due to non-orthogonal parameter changes.

On the other hand, the advantage of the second method is that even if anomalies occur simultaneously due to non-orthogonal parameter changes, the occurrence of anomalies can be detected. A drawback of the second method is that the number of elements that cause anomalies can only be set to types that are less than or equal to the number of dimensions of the parameter, that is, the number of dimensions n in an n-dimensional space. For example, as in the above-described embodiment, if the parameters are two-dimensional K ₀ and T ₁ , only up to two causes of abnormality can be detected.

Therefore, it is preferable for the user to appropriately select the contribution rate calculation method according to the purpose of use and the situation.

<Operation example>
Next, an operation example of the control system 100 regarding the abnormality detection device 2 in this embodiment will be described with reference to the flowchart of FIG. FIG. 20 is a flowchart showing an operation example of the control system 100 regarding the abnormality detection device 2 in this embodiment.

As described above, in the abnormality detection device 2 of the present embodiment, the disturbance parameter values of the abnormality mode point A1 of the moisture content abnormality and the abnormality mode point A2 of the positional deviation are stored in the abnormality parameter value storage area 208 of the storage unit 20 in advance. Let me remember. Also, the disturbance parameter value of the normal point A0 is stored in the normal parameter value storage area 205 of the storage unit 20 .

First, when the control system 100 of the present embodiment is powered on, for example, the processing unit 232 of the diagnosis unit 23 measures the controlled variable y of the controlled object 3 during a predetermined period (FIG. 20: S50). Next, the processing unit 232 acquires the controlled object parameter values of the model of the controlled object 3 based on the manipulated variable u of the controlled quantity adjuster 1 and the controlled quantity y of the controlled object 3 ( FIG. 20 : S51). .

After the manipulated variable u and the controlled variable y are stabilized, the disturbance generator 4 generates disturbance ( FIG. 20 : S52). Then, the disturbance waveform _estimating unit 21 calculates the first estimating device controlled variable yra from the model of the controlled object 3 and the manipulated variable u, and calculates the first estimating device controlled variable _yra and the controlled variable y. , the first estimated disturbance control amount yd _a is further estimated (FIG. 20: S53).

Next, the parameter estimator 22 estimates parameter values of the disturbance model from the first estimated disturbance control amount yd _a and the disturbance trigger signal d ( FIG. 20 : S54). If the disturbance parameter value of the normal point A0 is not stored, the estimated disturbance parameter value may be learned as the disturbance parameter value of the normal point A0 and stored in the normal parameter value storage area 205 of the storage unit 20 ( Figure 20: S55, S551). In this case, the vectorization process described above is performed, and the equations (17) and (23) are stored as programs in the storage unit 20 (FIG. 20: S552). After storing, the process returns to S52. Further, by measuring the control amount and the like in the normal disturbance in advance, the disturbance parameter value of the normal point A0 is stored in the normal parameter value storage area 205 of the storage unit 20, and the vectorization process described above is performed in advance, ( Equation 17) and Equation 23 may be stored in the storage unit 20 as a program.

When the disturbance parameter value of the normal point A0 is stored, the processing unit 232 uses the estimated disturbance parameter value as the disturbance parameter value of the measurement point B based on the equations (17) and (23). Then, each contribution rate is calculated (FIG. 20: S56). If the disturbance parameter value at the time of measurement exceeds a predetermined threshold value set from the disturbance parameter value at normal time, it is determined that the disturbance is abnormal, and if it is within the predetermined threshold value, the disturbance is not abnormal. It is judged to be normal (FIG. 20: S57). If the disturbance is determined to be abnormal, the display unit 24 displays the calculated contribution rate (FIG. 20: S58). If it is determined that the disturbance is not abnormal, the process returns to S52 to execute the process. As described above, in the present embodiment, every time the control system 100 is powered on, the operation shown in FIG. 20 is performed, the contribution rate is calculated, and if the disturbance is abnormal, the display done.

In the present embodiment, the process returns to S12 when it is not determined to be abnormal, and the method of generating the next disturbance is described. Then, the process may be returned to S12 to generate the next disturbance. Moreover, although the method of displaying when it is determined that there is an abnormality is described, the contribution rate may be displayed on the display unit 24 each time the contribution rate is calculated regardless of whether there is an abnormality.

(Embodiment 6)
Next, Embodiment 6 of the present disclosure will be described with reference to FIGS. 21 and 22. FIG. FIG. 21 is a diagram showing an example of plotting disturbance parameter values of three types of causes of anomalies in a two-dimensional parameter space in this embodiment.

In this embodiment, the contribution rate is calculated by the first method when the number of parameters is two, that is, when the number of abnormal mode points is three in a two-dimensional space.

In FIG. 21, the normal point A0, the abnormal mode points A1 and A2, and the disturbance parameter values at the measurement point B are the same as in the above embodiment. In FIG. 21, an abnormality mode point A3 is plotted when an abnormality is caused by the third factor other than the moisture content abnormality and the positional deviation.

Even in such a case, the projection vector B A3 of the abnormal mode vector _A3 of the measurement vector B can be obtained by the method described in the first embodiment, and the contribution ratio of the measurement vector B to the abnormal mode vector A3 can be obtained. As described above, the use of the first method makes it possible to calculate the contribution rate even if the number of abnormal mode points is greater than the number of parameters.

FIG. 22 is a diagram showing an example in which disturbance parameter values of three types of abnormal mode points are plotted in a three-dimensional parameter space in this embodiment. In the example shown in FIG. 22, the contribution rate is calculated by the first method when the number of parameters is three, that is, when the number of abnormal mode points is three in the three-dimensional space. As a model with three parameters, for example, a model represented by the following (Equation 33) can be considered.

As shown in FIG. 22, even if the number of model parameters increases, the contribution rate can be calculated by the method described in the third embodiment. Further, although details are not shown, in the case of this embodiment, the number of abnormal mode points does not exceed the number of parameters of the model, so the contribution rate can be calculated by the second method as well.

FIG. 23 is a diagram showing another example in which disturbance parameter values of three types of abnormal mode points are plotted in the three-dimensional parameter space in this embodiment. In the example shown in FIG. 23, the contribution rate is calculated by the second method when the number of parameters is three, that is, when the number of abnormal mode points is three in the three-dimensional space. Also in this case, the model represented by the above-mentioned (Formula 33) can be considered as a model with three parameters. Of course, these above-mentioned numbers of parameters, number of abnormal modes and models are exemplary and not limiting.

(Embodiment 7)
Next, Embodiment 7 according to the present disclosure will be described with reference to FIG. FIG. 24 is a diagram for explaining the method of determining the threshold in this embodiment. In the above-described embodiment, a mode has been described in which the disturbance parameter values of the abnormal mode points of the moisture content abnormality and the positional deviation are set as the threshold values of the allowable ranges of the moisture content abnormality and the positional deviation, respectively. However, this embodiment differs from the above-described embodiment in that the threshold value of the allowable range is set using the following formula (Equation 34).

In the above (Formula 34), Vref is the length of the vector of the abnormal mode point with reference to the normal point as shown in FIG. 24, and Vrep is the length of the vector of the normal point as shown in FIG. is the length of the vector of the representative points included in the normal range. For the representative point, the disturbance parameter value is measured multiple times in a normal state, and a value that is 3σ based on the standard deviation σ of the disturbance parameter value of the measured normal point may be used, or the measured normal point may be used. Also, the disturbance parameter value measured last time may be set as the normal point, or the moving average of the disturbance parameter values from the current time to P times before (P is a positive integer) may be used. Coefficient k is a preset coefficient.

24, the threshold value Vth is obtained by multiplying the difference between the length Vref of the vector of the abnormal mode point and the length Vrep of the vector of the representative point multiplied by the coefficient k, is obtained by adding the length Vrep of the vector of .

In this way, by determining the threshold value of the allowable range of the moisture content abnormality, etc., which causes the abnormality, it is possible to appropriately estimate the cause of the moisture content abnormality, etc., which is likely to cause the abnormality before the abnormality occurs. can be done.

(Embodiment 8)
Next, Embodiment 8 of the present disclosure will be described with reference to FIGS. 25 to 28. FIG. FIG. 25 is a block diagram showing a schematic configuration of a control system 100 relating to the abnormality detection device 2 according to the eighth embodiment.

In the above-described embodiment, an aspect of detecting a disturbance abnormality by estimating the disturbance control amount from the operation amount u and the control amount y has been described. The present embodiment is different from the above-described embodiments in that it detects an abnormality in disturbance using the manipulated variable u without the need to measure the controlled variable y.

When the controlled variable y of the controlled object 3 is stabilized at a state that matches or is extremely close to the target value r, the manipulated variable u of the controlled variable adjuster 1 also maintains a constant value and a stable state (that is, a settled state). This manipulated variable is set as a settling manipulated variable u _s . If a disturbance is applied while the manipulated variable is set, a difference occurs between the controlled variable y and the target value r, so that the manipulated variable u varies from the settling manipulated variable _us . Since the amount of change in the manipulated variable u is determined according to the amount of change in the controlled variable y, there is a correlation between these amounts of change. Therefore, by obtaining the amount of change in the manipulated variable u in a normal disturbance in advance, monitoring the manipulated variable u, and calculating the amount of change from the settling manipulated variable _us , the amount of disturbance can be compared with the amount of change in the normal state. Anomalies can be detected.

As shown in FIG. 25, in the control system 100 according to the eighth embodiment, the abnormality detection device 2 does not include the disturbance waveform estimator 21 and the parameter estimator 22, and instead includes a manipulated variable variation calculator 25. However, it differs from the first embodiment.

The storage unit 20 includes an operation amount storage area 202 , a settling operation amount storage area 209 , a disturbance trigger signal storage area 204 , a normal operation amount fluctuation storage area 210 and a threshold storage area 206 . The settling manipulated variable u _s described above is stored in the settling manipulated variable storage area 209 . The settling operation amount u _s is, for example, the operation amount It may be the manipulated variable measured when u does not fluctuate for a certain period of time. Alternatively, the controlled variable y may be measured and used as the manipulated variable measured when the manipulated variable u and the controlled variable y do not fluctuate for a certain period of time after the control system 100 is started and before the disturbance occurs. Alternatively, the operation amount immediately before the disturbance is generated may be used. In the normal operation amount fluctuation storage area 210, an integrated value of the operation amount change amount during a normal disturbance, which will be described later, is stored. The threshold storage area 206 stores a second threshold for determining that the integrated value of the amount of change in the manipulated variable is abnormal. The threshold may be set by the user, or may be set as a difference or a ratio with respect to the integrated value of the amount of change in the manipulated variable during normal operation.

The manipulated variable variation calculator 25 is configured by a CPU and the like. The manipulated variable fluctuation calculator 25 in the present embodiment calculates an integral value of the amount of change in the manipulated variable with respect to the settling manipulated variable _us based on the manipulated variable u, the settling manipulated variable _us , and the disturbance trigger signal. The period for calculation may be the period during which the disturbance trigger signal is input, or may be a certain period of time after the disturbance trigger signal is input, which is set by the user.

The diagnosis unit 23 determines whether or not an abnormality has occurred by comparing the integrated value of the amount of change in the manipulated variable at the time of measurement with a threshold value. The display unit 24 displays the determination result.

FIGS. 26A to 26C are graphs showing manipulated variables and the like in this embodiment. FIG. 26(A) is a graph showing part of FIG. 4(B) and represents an example of the manipulated variable u. FIG. 26(B) is a graph showing part of FIG. 4(C) and represents an example of the first estimated device controlled variable _{yr a} in the first embodiment. FIG. 26(C) is a graph showing part of FIG. 4(E) and represents an example of the disturbance trigger signal d.

The time axes in FIGS. 26A to 26C correspond to each other. A period from 0 to t21 is a period from when the device is started until a disturbance occurs. At t21, when the disturbance trigger signal d is input to the disturbance generator 4, the manipulated variable u increases as described in the first embodiment. After a predetermined period of time has elapsed, for example, at t22, the controlled variable y matches or very close to the target value r, so that the manipulated variable u is maintained at a constant value.

Each shaded area in FIGS. 26A and 26B is an area fluctuated due to the input of the disturbance. As mentioned above, there is a correlation between these regions. Further, as is clear from FIG. 26A, the integrated value of the manipulated variable change amount is an area that can be represented by the settling manipulated variable u _s and the manipulated variable change amount. For this region, for example, a period from when the disturbance trigger signal d represented by t21 is input to when the manipulated variable u represented by t23 is stable at a constant value may be used. Also, although not shown, a period from when the disturbance trigger signal is input until it stops may be used. Let _yra as be the first estimated apparatus controlled variable corresponding to the settling manipulated variable u _s . In this case, _as is clear from FIG. It is the integrated value of the amount of change in quantity.

<Abnormal judgment>
In this embodiment, abnormality is determined by comparing the integrated value of the manipulated variable change amount when the disturbance is normal and the integrated value of the manipulated variable change amount calculated when the disturbance occurs. Hereinafter, the integrated value of the manipulated variable variation calculated when the disturbance occurs is referred to as the integrated value of the manipulated variable variation during measurement. The abnormality determination processing in this embodiment will be described below.

FIG. 27 is a graph showing an example of the calculation result of the integrated value of the manipulated variable change amount during measurement in this embodiment. The horizontal axis represents each disturbance, and the vertical axis represents the integrated value of the manipulated variable change amount for each disturbance. The two dashed lines are the upper and lower threshold limits, respectively. The determination unit 231 of the diagnosis unit 23 may determine that there is an abnormality when the integrated value of the amount of change in the manipulated variable at the time of measurement does not fall within these ranges. The threshold can be set, for example, by obtaining an integral value of the amount of change in the manipulated variable in a normal disturbance and based on the integrated value of the amount of change in the manipulated variable in the normal state.

<Operation example>
Next, an operation example of the control system 100 regarding the abnormality detection device 2 in this embodiment will be described with reference to the flowchart of FIG. FIG. 28 is a flow chart showing an operation example of the control system 100 regarding the abnormality detection device 2 in this embodiment.

First, when the control system 100 of the present embodiment is powered on, the manipulated variable u is controlled so that the controlled variable y approaches the target value r ( FIG. 28 : S80). After the manipulated variable is set, the set manipulated variable u _s is stored in the manipulated variable storage area 202 of the storage unit 20 ( FIG. 28 : S81).

Next, the disturbance generator 4 generates disturbance (FIG. 28: S82). Then, the integrated value of the amount of change in the manipulated variable u of the controlled variable adjuster 1 with respect to the settling manipulated variable u _s is calculated (FIG. 28: S83).

If the integrated value of the amount of change in the operation amount under normal conditions is not stored, the integrated value of the amount of change in the operation amount due to the disturbance is stored as the integral value of the amount of change in the operation amount under normal conditions. It may be stored in the area 210 (FIG. 28: S84, S841). After storing, the process returns to S82. Further, by measuring the manipulated variable or the like in normal disturbance in advance, the integrated value of the normal manipulated variable change amount may be calculated and stored in the normal manipulated variable fluctuation storage area 210 .

When the integrated value of the amount of change in the manipulated variable in the normal state is stored, the determination unit 231 of the diagnosis unit 23 determines that the calculated integrated value of the amount of change in the manipulated variable is the integrated value of the amount of change in the manipulated value during measurement. It is compared with the integrated value of the amount of change in the manipulated variable at time (Fig. 28: S85). The determination unit 231 determines that the disturbance is abnormal when the predetermined threshold is exceeded, and determines that the disturbance is not abnormal, that is, is normal when the predetermined threshold is reached ( FIG. 28 : S86). If the disturbance is normal, the process returns to S82. If it is determined that there is an abnormality, for example, a message indicating that the disturbance is abnormal is displayed on the display unit 24, and the process ends (FIG. 28: S87).

In the present embodiment, when it is determined that there is no abnormality, the process returns to S82, and the method of generating the next disturbance is described. Then, the process may return to S82 and operate to generate the next disturbance. In addition, although the determination unit 231 determines abnormality/normality based on a predetermined threshold value, the diagnosis unit 23 evaluates abnormality based on the integrated value of the amount of change in the operation amount during measurement without the determination unit 231 making a determination. Alternatively, the user may make a decision based on the display on the display unit 24 .

In this embodiment, an abnormality is detected using the integrated value of the amount of change in the manipulated variable, but other methods may be used. For example, an abnormality may be detected using the amount of change in the manipulated variable u. _uc in FIG. 26A represents the amount of change in the manipulated variable due to the occurrence of disturbance. _yrc in FIG. 26(B) represents the amount of change in the first estimated device controlled variable y due to the occurrence of disturbance. There is a correlation between these values. Further, as is clear from FIG. 26A, the amount of change in the manipulated variable is the difference between the settling manipulated variable u _s and the value at which the manipulated variable stabilizes after the occurrence of the disturbance. As is clear from FIG. 26(B), the amount of change in the first estimator controlled variable y is the first settling estimator controlled variable _yrs and the first estimator controlled variable y stabilized after the occurrence of the disturbance. value.

As with the embodiments described above. By setting a threshold such as by acquiring the amount of change in the manipulated variable in a normal disturbance, it is possible to determine whether there is an abnormality in the disturbance from the measured amount of change in the manipulated variable.

As described above, according to the present embodiment, it is possible to determine whether intermittent disturbance is abnormal or not without creating a model of the controlled object 3 or the disturbance.

The present invention is not limited to the embodiments described above, and various improvements and design changes are possible without departing from the gist of the present invention.

As described above, according to the technology according to the present disclosure, it is possible to easily estimate the presence or absence of an abnormality in disturbance that occurs intermittently. be able to cope.

1 Control amount adjuster 2 Abnormality detection device 3 Controlled object 4 Disturbance generation device 20 Storage unit 21 Disturbance waveform estimation unit 22 Parameter estimation unit 23 Diagnosis unit 24 Display unit 100 Control system 201 Model storage area 202 Manipulated amount storage area 203 Control amount storage Area 204 Disturbance trigger signal storage area 205 Normal parameter value storage area 206 Threshold storage area 231 Determination unit

Claims (18)

A parameter of a model of the controlled object is estimated based on the manipulated variable output by the controlled variable adjuster and the controlled variable output by the controlled object, and the manipulated variable, the controlled variable, and the disturbance generator are inputted. An anomaly detection device for estimating a parameter of a model of a disturbance control amount caused by the disturbance based on a disturbance trigger signal that generates a disturbance, and detecting an anomaly of the disturbance based on the parameter,
a storage unit that stores the model of the disturbance control amount and parameters of the disturbance in the model of the disturbance control amount;
a disturbance controlled variable waveform estimator for estimating an estimated disturbance controlled variable from the controlled variable and an estimated device controlled variable calculated by subtracting the disturbance controlled variable from the controlled variable;
a parameter estimating unit that calculates the estimated disturbance control amount in a predetermined measurement period based on the disturbance trigger signal and estimates a parameter value of the disturbance during measurement;
a diagnostic unit that evaluates the abnormality of the disturbance based on the parameter value of the disturbance during the measurement;
Anomaly detection device. The disturbance model is a transfer function determined based on the measured disturbance trigger signal and the calculated estimated disturbance control amount, and the disturbance parameters are coefficients of a numerator polynomial and a denominator polynomial of the transfer function. do,
The abnormality detection device according to claim 1. The disturbance model is a transfer function determined based on the measured disturbance trigger signal and the calculated estimated disturbance control amount, and the disturbance parameter is the root of the numerator polynomial and denominator polynomial of the transfer function, or the inverse of the root characterized by using
The abnormality detection device according to claim 1. The disturbance controlled variable waveform estimation unit estimates a first estimated disturbance controlled variable by subtracting a first estimated device controlled variable calculated from the controlled object model and the manipulated variable from the controlled variable. death,
The parameter estimator estimates a parameter value of the disturbance during the measurement from the first estimated disturbance control amount and the disturbance trigger signal.
The abnormality detection device according to any one of claims 1 to 3. The disturbance controlled variable waveform estimating unit estimates a second estimated disturbance controlled variable by subtracting a second estimated apparatus controlled variable, which is a component of the controlled variable from which fluctuation due to disturbance is removed, from the controlled variable. death,
The parameter estimator estimates a parameter value of the disturbance during the measurement from the second estimated disturbance control amount and the disturbance trigger signal.
The abnormality detection device according to any one of claims 1 to 3. The storage unit further stores a parameter value of the disturbance when the disturbance is normal,
The diagnosis unit further sets a first threshold range from the disturbance parameter value in the normal state, and determines whether the disturbance parameter value in the measurement is within the first threshold range. 6. The abnormality detection device according to claim 4, further comprising a determination unit for determining an abnormality of. The number of parameters of the disturbance is n (n is a positive integer),
The storage unit further stores a disturbance parameter value when the disturbance is normal and m kinds of disturbance parameter values (m is a positive integer) when the disturbance is abnormal,
The diagnosis unit further includes, in an n-dimensional parameter space corresponding to the number of parameters of n disturbance models, a normal point corresponding to the parameter value of the disturbance in the normal state and m types of disturbance parameter values in the abnormal state. respectively, and position m abnormal mode points associated with the cause of the abnormality, position a measurement point corresponding to the parameter value of the disturbance during the measurement, and measure from the normal point to the measurement point A processing unit that sets a vector and each abnormal mode point vector directed from the normal point to each abnormal mode point, and evaluates a contribution rate of the measurement vector to each abnormal mode point vector, wherein the contribution rate is Based on, diagnose the degree of abnormality of the disturbance for each cause,
The abnormality detection device according to claim 4 or 5. The processing unit evaluates the contribution rate by calculating the cosine of the angle formed by the measurement vector and each abnormal mode point vector.
The abnormality detection device according to claim 7. The processing unit evaluates the contribution rate by decomposing the measurement vector with each abnormal mode point vector,
The abnormality detection device according to claim 7. The processing unit sets the outline of the allowable range of the degree of abnormality in the n-dimensional parameter space as a threshold line of the parameter at which the contribution rate is constant,
The diagnosis unit includes a determination unit that determines an abnormality of the disturbance according to whether or not the measurement point exists within the allowable range.
The abnormality detection device according to any one of claims 7 to 9. The processing unit sets the length of each abnormal mode point vector to Vref, the length of the vector of the representative point included in the normal range from the normal point to Vrep, and the predetermined coefficient to k. Calculate the threshold Vth, which is the outline of the allowable range of degree, by the following formula,
Vth=(Vref−Vrep)×k+Vrep
The diagnosis unit includes a determination unit that determines an abnormality of the disturbance according to whether or not the measurement point exists within the allowable range.
The abnormality detection device according to claim 10. The abnormality detection device according to any one of claims 1 to 11, wherein the control amount adjuster is a temperature adjuster. An abnormality detection device for detecting an abnormality in the disturbance based on a manipulated variable output by a control amount adjuster and a disturbance trigger signal input to a disturbance generation device to generate the disturbance,
a storage unit that stores a set value that is a manipulated variable in a state where the controlled variable output by the controlled object is stable;
a manipulated variable variation calculation unit that calculates a manipulated variable variation during disturbance measurement based on the manipulated variable, the set value, and the disturbance trigger signal;
a diagnostic unit that determines an abnormality in the manipulated variable variation during the measurement,
The storage unit stores the operation amount fluctuation when the disturbance is normal,
The diagnosis unit sets a second threshold range from the operation amount fluctuation in the normal state, and detects an abnormality of the disturbance according to whether or not the operation amount fluctuation at the time of measurement is within the second threshold range. judge,
Anomaly detection device. 14. The abnormality detection device according to claim 13, wherein said manipulated variable fluctuation is an integral value of a change amount of said manipulated variable with respect to said set value. 14. The abnormality detection device according to claim 13, wherein said manipulated variable variation is a variation of the manipulated variable with respect to said set value. The abnormality detection device according to any one of claims 13 to 15, wherein said control amount adjuster is a temperature adjuster. A parameter of a model of the controlled object is estimated based on the manipulated variable output by the controlled variable adjuster and the controlled variable output by the controlled object, and the manipulated variable, the controlled variable, and the disturbance generator are inputted. An abnormality detection method for estimating a parameter of a model of a disturbance control amount caused by the disturbance based on a disturbance trigger signal that generates a disturbance, and detecting an abnormality of the disturbance based on the parameter,
a step of storing the model of the disturbance control amount and parameters of the disturbance in the model of the disturbance control amount in a storage unit;
a step of estimating an estimated disturbance controlled variable by a disturbance controlled variable waveform estimator from the controlled variable and an estimated device controlled variable calculated by subtracting the disturbance controlled variable from the controlled variable;
a step of estimating a parameter value of the disturbance during measurement by a parameter estimating unit based on the estimated disturbance control amount calculated in a predetermined measurement period based on the disturbance trigger signal;
A step of evaluating an abnormality of the disturbance by a diagnosis unit based on a parameter value of the disturbance during the measurement,
Anomaly detection method. A program for causing a computer to execute the anomaly detection method according to claim 17.

Images