JP5013244B2 - Characteristic amount estimation device and mounted object determination device - Google Patents

Description

  The present invention relates to a characteristic amount estimation device and a mounted object determination device, and relates to estimation of a characteristic amount of a weight body mounted on a vehicle that performs posture control of the vehicle body and determination of the weight body.

Vehicles that use inverted pendulum attitude control (hereinafter simply referred to as inverted pendulum vehicles) are attracting attention and are currently being put into practical use.
For example, Patent Document 1 proposes a technique that has two drive wheels arranged on the same axis and detects and drives the posture of the drive wheel by the driver's movement of the center of gravity.
Further, a vehicle that moves while controlling the posture of one conventional circular drive wheel or one spherical drive wheel is proposed in Patent Document 2, and various inverted pendulum vehicles are also pointed out in Patent Document 2. Yes.

JP 2004-276727 A JP 2004-129435 A

In such a vehicle, the vehicle is maintained in a stopped state or traveled while performing posture control based on the amount of weight shift by the driver, the amount of operation from the remote controller or the control device, the travel data inputted in advance. It has become.
Then, the design center of gravity (design vehicle center of gravity) position is used from the known body weight and body center of gravity position in advance and the model weight and model center of gravity position assuming a general weight body (for example, luggage or passengers). Thus, control parameters for the attitude control system are defined.

However, in actuality, vehicles of various weights and body types will get on the vehicle, and luggage of various weights and shapes (hereinafter referred to as weight bodies) will be on board, and the mass, center of gravity height, Since the moments of inertia differ greatly, the characteristic amount of the attitude control system deviates from the design value.
Also, the weight of the boarded body is not always kept constant, and the posture of the passenger may change (move, fall, etc.) and the luggage may increase. This also changes the mass, the height of the center of gravity, and the moment of inertia (hereinafter referred to as the mechanical characteristic amount), and the characteristic amount of the designed attitude control system deviates from the design value.

And in the vehicle using the inverted pendulum posture control assumed in the present embodiment, the smaller the vehicle body becomes, the lighter the vehicle, and the higher the riding part, the more the influence on the control system characteristics due to the change in the vehicle is growing.
For this reason, if the control parameter based on the passenger assumed at the time of design is maintained, stable posture control cannot be performed, and comfortable boarding cannot be provided to the passenger, and the control parameter needs to be corrected.
As a premise for correcting the control parameter, it is necessary to know the actual mechanical characteristic amount of the weight body (vehicle) and the type of the weight body (person, luggage, etc.).

Accordingly, a first object of the present invention is to provide a parameter estimation device that estimates an actual mechanical characteristic amount of a weight body mounted on a vehicle.
In addition, a second object of the present invention is to provide a mounted object determination device that determines a weight body mounted on a vehicle from the estimated mechanical characteristic amount.

(1) The invention described in claim 1 is a characteristic amount estimation device that estimates a mechanical characteristic amount of a weight body mounted on a vehicle that controls the posture of the vehicle body, and places a weight body such as a luggage or a passenger. From the boarding part, the load sensor arranged in the boarding part, the height sensor for measuring the height of the weight body, and the measured values of the load sensor and the height sensor, the mechanical characteristic amount of the weight body The characteristic amount estimation device is provided with an estimation means for estimating the first object to achieve the first object.
( 2 ) The invention described in claim 2 is a characteristic amount estimation device for estimating a mechanical characteristic amount of a weight body mounted on a vehicle that controls the posture of a vehicle body, and places a weight body such as a luggage or a passenger. Directly estimating a mechanical characteristic amount from a riding part, a load sensor disposed in the riding part, a height sensor that measures the height of the weight body, and measured values of the load sensor and the height sensor Estimating means, indirect estimating means for estimating a mechanical characteristic quantity by a disturbance observer, and estimating means for estimating a mechanical characteristic quantity based on the estimated values by the direct estimating means and the indirect estimating means, An apparatus is provided to achieve the first object.
( 3 ) In the invention described in claim 3 , in the characteristic amount estimation apparatus according to claim 2 , an error determination of the estimated value is performed using both estimated values estimated by the direct estimation means and the indirect estimation means. An error determination means is provided, and the estimation means estimates the other estimated value as a mechanical characteristic amount when one estimated value is determined to be an error.
( 4 ) In the invention described in claim 4 , in the characteristic quantity estimation apparatus according to claim 2 or 3 , the estimation means is based on a frequency component of an estimated value by the direct estimation means and the indirect estimation means. The mechanical characteristic quantity is estimated.
( 5 ) In the invention described in claim 5 , in the characteristic amount estimation device according to any one of claims 1 to 4 , the estimation means uses the weight as a mechanical characteristic amount. The mass of the body, the height of the center of gravity of the weight body, and the moment of inertia of the weight body are estimated.
( 6 ) In the invention described in claim 6 , in the characteristic quantity estimation apparatus according to claim 1 or 2 , the acceleration sensor and the tilt angle sensor are provided, and the estimation unit or the direct estimation unit is A mechanical characteristic amount of the weight body is estimated from a measurement value of the load sensor, the height sensor, and detection values of the acceleration sensor and the tilt angle sensor.
(7) In the invention according to claim 7, the characteristic quantity estimating apparatus according to any one of claims of claims 1 to 6, were estimated by the characteristic quantity estimating device, mechanical properties amount The mounted object determining device is provided with a type determining means for determining the type of the weight body mounted on the vehicle using the above-mentioned object, thereby achieving the second object.

In the first aspect of the present invention, measurement or these mass and height of the weight body, the measured value and the disturbance observer of the mass and height of the weight body in the second aspect of the present invention, the mechanical parameters of the weight body Can be estimated.
Thereby, in the present invention, it is possible to estimate the actual mechanical parameter of the controlled object in the posture control system of the vehicle body from the estimated dynamic parameter of the weight body, and to correct the control parameter in the posture control system. Thus, more stable posture control can be performed.

Hereinafter, preferred embodiments of the characteristic amount estimation apparatus and the mounted object determination apparatus of the present invention will be described in detail with reference to FIGS. 1 to 12.
In the following embodiments, the characteristic amount estimation device and the mounted object determination device are vehicles that perform attitude control, and the actual mechanical characteristic amount is estimated and the control parameter is corrected using the estimated mechanical characteristic amount. Thus, an example of a vehicle capable of performing posture control suitable for the state of a weight body mounted on the vehicle will be described.
(1) Outline of Embodiment FIG. 1 shows an outline of a control system characteristic amount correction process for correcting a control system characteristic quantity to an optimum value according to an actual weight body according to this embodiment.
In the present specification, a portion of the vehicle that is tilted for balancing is referred to as a “controlled object”.
“Vehicle” refers to all persons boarding from the outside of the vehicle, such as people, luggage, animals, etc., and means “heavy body”.
A portion excluding “vehicle” from “controlled object” is referred to as “vehicle body”.

In the present embodiment, as shown in FIGS. 1A and 1B, a vehicle mounted on the current vehicle is directly measured by a measuring instrument and / or estimated from a control result (history). Obtain the actual mechanical characteristics of the entire controlled object, including the sensing (sensing).
Then, as shown in FIGS. 1C and 1D, the reaction is used for control by correcting the control system characteristic amount in the attitude control system based on the estimated mechanical characteristic amount as a reaction.

FIG. 1A shows a method for directly measuring the amount of mechanical characteristics of a vehicle using a measuring instrument.
That is, a weight scale (load meter) is placed under the riding part (seat), a sitting height meter is placed on the backrest part, the weight and sitting height of the vehicle are measured, and each mechanical characteristic amount of the vehicle is estimated, Then, the mechanical characteristic amount of the entire controlled object is calculated (the characteristic amount of the vehicle body is known).

FIG. 1B shows a case in which the mechanical characteristic amount is estimated from the posture change of the controlled object and the history of torque input using, for example, a disturbance observer.
That is, when the posture of the vehicle body (the controlled object) changes with respect to the applied torque (torque input), for example, when it occurs late / early, the vehicle is heavier than the assumed value or the center of gravity is higher / lighter or Should be low.
Such an influence is estimated using a disturbance observer, and the mechanical characteristic amount of the entire controlled object is estimated.

Using the actual mechanical characteristic amount of the controlled object estimated as described above, the actual posture control system characteristic amount is corrected.
FIG. 1C illustrates a case where the control parameter is corrected.
A control system characteristic quantity such as a feedback gain in the attitude control system is corrected from the estimated mechanical characteristic quantity.
In this case, there are a method of calculating an optimal control parameter value for the estimated mechanical characteristic amount and a method of correcting the control parameter value so as not to change the control system characteristic amount as much as possible. In the present embodiment, either one of the two may be used, or both may be used. When both are used, for example, the initial predetermined time when the change of the mechanical characteristic amount is detected is calculated by the characteristic invariant parameter calculation, and after the predetermined time has elapsed, the characteristic optimization parameter calculation is performed.

FIG. 1D shows control system characteristic amount adjustment by weights.
The control system characteristic amount is brought close to the reference value by moving a weight disposed on the rear surface of the vehicle body so as to be movable in the vertical direction. For example, when the occupant raises his / her baggage from his feet to his seat or raises his arm, he adjusts the position of the center of gravity of the controlled object by lowering the position of the weight, and reduces the moment of inertia, Move closer to the design value.

(2) Details of Embodiment FIG. 2 illustrates an external configuration of a vehicle according to the present embodiment.
As shown in FIG. 2, the vehicle includes two drive wheels 11a and 11b arranged on the same axis.
Both drive wheels 11a and 11b are each driven by a drive motor 12.

On the upper part of the drive wheels 11a, 11b (hereinafter referred to as the drive wheels 11 when referring to both drive wheels 11a and 11b) and the drive motor 12, a boarding section 13 (on which a heavy load such as luggage or passengers rides) Sheet) is arranged.
The riding section 13 includes a seat surface section 131 on which a driver sits, a backrest section 132, and a headrest 133.

  The riding section 13 is supported by a support member 14 fixed to a drive motor housing 121 in which the drive motor 12 is accommodated.

  A control device 30 is arranged on the left side of the riding section 13. This control device 30 is for giving instructions such as acceleration, deceleration, turning, rotation, stop, braking, etc. of the vehicle by the operation of the driver.

  Although the control device 30 in the present embodiment is fixed to the seat portion 131, it may be configured by a remote control connected by wire or wirelessly. Further, an armrest may be provided and the control device 30 may be arranged on the upper part thereof.

  Further, the control device 30 is disposed in the vehicle of the present embodiment, but in the case of a vehicle that automatically travels according to predetermined travel data, a travel data acquisition unit is disposed instead of the control device 30. The For example, the travel data acquisition unit may be configured by a reading unit that reads travel data from various storage media such as a semiconductor memory, and / or a communication control unit that acquires travel data from outside by wireless communication. Good.

  In FIG. 2, the boarding unit 13 displays a case where a person is on board. However, the boarding part 13 is not necessarily limited to a vehicle driven by a person, and only a baggage is placed to travel by remote control operation from the outside. In the case where the vehicle is stopped or stopped, only the baggage is loaded and the vehicle is driven or stopped according to the driving data, or the vehicle may be driven or stopped in a state where nothing is boarded.

In this embodiment, control such as acceleration / deceleration is performed by an operation signal output by the operation of the control device 30. For example, as shown in Patent Document 1, the driver is inclined forward and tilted forward and backward. By changing the angle, the vehicle may be switchable so as to perform vehicle attitude control and travel control according to the inclination angle.
When performing posture control and traveling control based on a tilt moment by the driver, the posture control according to the present embodiment is not performed.

A load meter 51 (not shown), which will be described later, is disposed on the lower side (back surface side of the seating surface portion 131) of the riding portion 13.
A weight (weight) 134 is disposed on the back surface of the boarding portion (the back side or the inside of the backrest portion). The weight 134 is configured to be movable in the vertical direction by a weight drive actuator 62 described later.

A control unit 16 is disposed between the riding section 13 and the drive wheel 11.
In the present embodiment, the control unit 16 is attached to the lower surface of the seat portion 131 of the riding portion 13, but may be attached to the support member 14.

FIG. 3 shows the configuration of the control unit 16.
A control ECU (electronic control unit) 20 that performs various types of control such as vehicle travel, posture control, and control system characteristic amount correction control in the present embodiment is provided. The control ECU 20 includes a control device 30, travel, and posture. The control sensor 40, the mechanical characteristic amount estimation sensor 50, the actuator 60, and other devices such as a battery are electrically connected.

  The battery supplies power to the drive motor 12, the weight drive actuator 62, the control ECU 20, and the like.

  The control ECU 20 includes a ROM that stores various programs and data such as a travel control program, an attitude control program, and a control system characteristic amount correction processing program, a RAM that is used as a work area, an external storage device, an interface unit, and the like. It consists of a system.

The control ECU 20 includes a vehicle body basic control system 21 that performs running and posture control, and a mechanical characteristic amount estimation control system 22.
The mechanical characteristic amount estimation control system 22 estimates the actual mechanical characteristic amount of the controlled object in the posture control system from the measured value of the mechanical characteristic amount estimation sensor 50, and from the control result (history) By estimating using the observer 23, it functions as an estimation means.
Further, the mechanical characteristic amount estimation control system 22 supplies a control parameter correction value to the vehicle body basic control system 21 in order to correct the control system characteristic amount of the attitude control system based on the estimated actual mechanical characteristic amount. The weight drive actuator 62 is supplied with a command value indicating the amount of movement of the weight 134 (FIG. 1).

The travel and attitude control sensor 40 includes a travel speed meter (wheel tachometer) 41 that detects the speed (wheel rotation angle) of the vehicle, and a vehicle body tilt angle meter (angular speed meter) 42 that detects the vehicle body tilt angle (tilt angular speed). It has.
The detection value by the running / attitude control sensor 40 is supplied to the vehicle body basic control system 21 and the mechanical characteristic amount estimation control system 22.

The mechanical characteristic amount estimation sensor 50 includes a load meter 51 (or load distribution meter) and a sitting height meter (or shape measuring instrument) 52.
FIG. 4 shows the arrangement of the load meter 51 and the sitting height meter 52.
As shown in FIG. 4B, the load meter 51 is disposed on the lower side of the riding part 13, specifically on the lower surface part of the seating surface part 131, and the mass of the rided object is measured to determine the mechanical characteristic amount. It is supplied to the estimation control system 22. The load meter 51 is arranged on the lower side of the riding section 13, so that it is arranged not only on the vehicle placed on the riding section, but also on the load on the backrest 132 and the headrest 133 and other places. It is configured to be able to measure the load of all mounted vehicles.
The weight of the vehicle body (hereinafter referred to as the vehicle body weight) and the position of the center of gravity (hereinafter referred to as the vehicle body center of gravity position) are fixed and determined in advance at the time of design, and thus are not measured by the load meter 51.

As the load meter 51 of the present embodiment, the mass of the occupant is measured by a single component load meter during upright low-speed traveling, but the main body tilt angle sensor and the translational acceleration sensor used for posture control, or 3 A component load cell may be used.
As a result, it is possible to measure the load at the time of inclination and acceleration.
Moreover, you may make it estimate a passenger | crew's magnitude | size by arrange | positioning a several load meter and measuring load distribution.

As shown in FIGS. 4A and 4B, a seat height meter 52 for measuring the seat height (height) of a vehicle such as an occupant is disposed on the backrest portion 132.
The sitting height meter 52 arranges a plurality of fixed optical sensors in the z-axis direction (height direction), discretely measures the occupant's sitting height, and sends the measured values to the mechanical characteristic amount estimation control system 22 (FIG. 3). It comes to supply.
Note that the sitting height meter 52 may be configured to scan a moving (scanning) type optical sensor in the z-axis direction, thereby enabling highly accurate measurement.
In addition, it is possible to identify the size and shape of the occupant and distinguish the vehicle (person, baggage, etc.) by distributing the fixed photo sensor on the plane or scanning the scan photo sensor on the plane. May be.

FIG. 5 shows the weight 134 disposed on the backrest 132.
As shown in FIG. 5A, the weight 134 is disposed on the backrest 132 and is configured to be movable in the vertical direction. The movement in the vertical direction is realized by various other methods by using a ball screw or by moving on the rail using a linear motor.
By moving the weight 134 up and down, it functions as a vehicle body deformation means for changing the position of the center of gravity and the moment of inertia of the vehicle body.
Note that, as shown in FIG. 5B, the center of gravity position and the moment of inertia of the vehicle body may be changed by changing the height of the riding section 13 (the controlled object).
Further, the center of gravity position and the moment of inertia of the vehicle body may be changed using weight distribution changing means for changing the weight distribution of the vehicle body.

  In FIG. 3, the actuator 60 includes a tire rotation actuator 61 that drives the drive wheels 11 according to the command value supplied from the vehicle body basic control system 21, and a weight 134 according to the command value supplied from the mechanical characteristic quantity estimation control system 22. A weight drive actuator 62 that moves in the vertical direction is provided.

Next, the control system characteristic amount correction process in the vehicle configured as above will be described.
FIG. 6 is a flowchart showing the contents of the control system characteristic amount correction process.
In this control system characteristic amount correction processing, an actual mechanical characteristic amount of the controlled object (vehicle body + vehicle) is estimated (estimating means: step 11 to step 17), and then based on the obtained mechanical characteristic amount. Then, the control system characteristic quantity in the attitude control system is corrected (control system characteristic quantity correction means: step 18 to step 23).
Hereinafter, the contents of each step will be described.

The actual mechanical characteristic amount is estimated from Step 11 to Step 13 using the measured values of the load meter 51 and the sitting height meter 52, and from Step 14 to Step 16 using the disturbance observer 23.
Note that both processes are performed in parallel for the estimation of the actual mechanical characteristic amount.

  First, in the mechanical characteristic amount estimation sensor 50, the weight and height of the vehicle are measured by the load meter 51 and the seat height meter 52 arranged in the riding section 13 and supplied to the mechanical characteristic amount estimation control system 22. (Step 11).

The mechanical characteristic estimation control system 22 calculates the mass (m ₁ ), the primary moment (m ₁ l ₁ ), and the moment of inertia (I ₁ + m ₁ l ₁ ² ) of the controlled object from the acquired measurement data. The estimation is performed as follows (step 12).
That is, the mechanical property quantity estimation control system 22 determines what the vehicle is based on, for example, the following threshold value from the values of the mass m _H and the sitting height ζ _H obtained by measurement.
(A) When m _H <0.2 kg and ζ _H <0.01 m, it is determined that the vehicle is “none”.
(B) When m _H > 8 kg, ζ _H > 0.3 m, and m _H / ζ _H > 30 kg / m, the vehicle is determined to be “person”.
(C) In other cases (other than the cases (a) and (b) above), it is determined that the vehicle is “luggage”.
In the discrimination conditions described above, the weight of the person discrimination condition (b) is as small as 8 kg because it is assumed that a child is on board. Further, by adding the weight per unit seat height (m _H / ζ _H ) to the person determination condition, it becomes possible to make the person determination more accurately. In this case, m _H / ζ _H <p (for example, 80 kg / m) is added as an upper limit to the determination condition (and condition) in order not to determine that the person is loaded even when a small and heavy load (for example, an iron ingot) is put on May be.
Note that each determination condition and determination value are examples, and are appropriately changed and determined according to an assumed use condition.

Hereinafter, the mechanical characteristic amount estimation control system 22 determines the height of the center of gravity (height from the seat surface portion 131) h _H and the moment of inertia (around the center of gravity) I _H according to the determined type of the vehicle. presume. In this way, by determining the vehicle and evaluating it with an expression corresponding to the type, it is possible to estimate a more accurate mechanical characteristic amount.

(A) When the vehicle is “None” h _H = 0
I _H = 0
(B) When the vehicle is “person” h _H = (ζ _H / ζ _{H, 0} ) h _{H, 0}
I _H = (m _H / m _{H, 0} ) (ζ _H / ζ _{H, 0} ) ² I _{H, 0}
Here, ζ _{H, 0} , h _{H, 0} , and I _{H, 0} are standard values of the sitting height, the center of gravity, and the moment of inertia (around the center of gravity) of the human body. In this embodiment, ζ _{H, 0} = 0.902 m, h _{H, 0} = 0.264 m, and I _{H, 0} = 5.19 kgm ² are used as standard values.

(C) When the load is “luggage” h _H = ((1−γ) / 2) ζ _H
I _H = ((1-3γ ² ) / 12) m _H ζ _H ²
Here, γ is an eccentricity representing a downward shift in the center of gravity.
For example, γ = 0.4 is used as the degree of eccentricity in the present embodiment, but can be appropriately changed according to the assumed use conditions.

Next, the mechanical property quantity estimation control system 22 is the actual mechanical characteristic quantity of the controlled object from the measured vehicle mass m _H , the estimated vehicle center of gravity height h _H , and the inertia moment I _H. The mass (m ₁ ), first moment (m ₁ l ₁ ), and inertia moment (I ₁ + m ₁ l ₁ ² ) are calculated (step 13).
That is, the dynamic characteristic amount estimation control system 22 determines the mass, the height from the axis of the drive wheel 11 to the center of gravity, and the moment of inertia of the vehicle and the vehicle body (m _H , l _H , I _H ), ( m _c , l _c , I _c ), the mass m _{1 to} be controlled, the height l ₁ from the axis of the drive wheel 11 to the center of gravity, and the moment of inertia I ₁ are expressed by the following (a) to (c) I want.
(A) m ₁ = m _H + m _c
(B) l ₁ = (m _H l _H + m _c l _c ) / m ₁
(C) I ₁ = I _H + m _H (l _H −l ₁ ) ² + I _c + m _c (l _c −l ₁ ) ²

The mass m _H of the loaded article is the measured value in step 11, the moment of inertia I _H is the moment of inertia I _H calculated in step 12.
The height l _H to the center of gravity is defined as l _H = h ₀ + h from the center of gravity height h _H calculated in step 12, where h ₀ is the height from the axis of the drive wheel 11 to the seating surface of the seating surface portion 131 _Obtained by _H.

  On the other hand, the mechanical characteristic quantity estimation control system 22 estimates a disturbance (actual mechanical characteristic quantity) by the disturbance observer 23 from the motion and posture control result (time history) of the vehicle body (steps 14 to 16).

The mechanical characteristic amount estimation control system 22 includes an input u (→) in the posture control (a torque command value for the tire rotation actuator 61 in FIG. 1) and an output y (→) (the running and posture control sensor 40 in FIG. 1). (The detected value (speed, vehicle body inclination angle)) is acquired (step 14).
The dynamic characteristic amount estimation control system 22 is based on the disturbance that the behavior different from the behavior based on the designed mechanical characteristic amount based on the assumed passenger is boarding a vehicle different from the assumption. Assuming that, disturbance is estimated using the acquired input u, output y, and disturbance observer 23 (step 15).
Next, the mechanical property quantity estimation control system 22 determines the mass, first moment, and inertia moment of the controlled object from the obtained data (step 16).

Hereinafter, processing for estimating an actual mechanical characteristic amount using the disturbance observer 23 will be described.
FIG. 7 shows an outline of the flow for estimating the mechanical characteristic amount.
First, a disturbance estimated value d (→ ∧) is calculated using a disturbance observer (step 31), and the tire rotation detected (or calculated based on the detected value) by a vehicle control sensor (running and attitude control sensor 40). A vehicle state quantity ξ (→) consisting of the angular acceleration θ _W (··), the main body inclination angular acceleration θ ₁ (··), and the main body inclination angle θ ₁ is determined (step 32).
Note that in this specification, for convenience of description, the notation differs from the drawings. For example, among symbols in parentheses after characters such as d (→ ∧) and θ (···), “→” represents a vector quantity (matrix) and “∧” is an estimated value. In addition, “·” and “··” represent the first and second derivatives of the character before the parenthesis.

Next, the dynamic characteristic quantity estimation control system 22 estimates the fluctuation matrix Λ by the least square method from the disturbance estimated value d (→ ∧) and the vehicle state quantity ξ (→) (step 33), and the fluctuation matrix Λ From these, fluctuations in the respective mechanical characteristic quantities of the main body (controlled object) are estimated (step 34).
Details of the estimation of the fluctuation amount of the mechanical characteristics will be described below.

FIG. 8 conceptually shows the state of disturbance estimation by the disturbance observer 23.
In FIG. 8, the controller corresponds to the vehicle body basic control system 21, and the actual controlled object corresponds to the tire rotation actuator 61 (input) and the entire vehicle (controlled object).

The disturbance observer 23 includes a control target model and an estimator.
The controlled object model is a theoretical model of an attitude control system that uses a mechanical characteristic amount in design based on an assumed value of a passenger.

The input u (→) from the controller is input to both the actual control object and the control object model.
Then, an output y (→ ∧) as a theoretical value is output from the controlled object model.
On the other hand, from the actual controlled object (controlled object), an output y (→) (actual value such as the inclination angle θ) for the state in which the disturbance d (→) is acting is output.

From the difference between the output y (→ ∧), which is the theoretical value, and the output y (→), which is the actual measurement value, the estimated value d (→ ∧) of the disturbance acting by the estimator is expressed by the following equation: Calculate according to 1.
In the estimation by the estimator (Equation 1), the influence due to the passengers of body weight, body type, etc. differing from the design values is estimated as the cause of the disturbance. That is, the influence of the difference between the fluctuation amount = the nominal value (assumed value) of the mechanical characteristic amount is regarded as a disturbance and is estimated by the disturbance observer 23.
In this estimation, in this embodiment, the minimum dimension observer is used to shorten the calculation time. However, when priority is given to robustness, the same dimension observer may be used.
The estimated speed is determined by the feedback gain L of the disturbance observer 23. Here, if the estimation speed is increased too much, the estimation becomes unstable, so that a certain amount of time is required for the estimation.

Next, the derivation of Equation 1 for calculating the disturbance estimated value d (→ ∧) will be described.
FIG. 9 illustrates a dynamic model of the vehicle attitude control system.
The balancer in FIG. 9 is a weight body for controlling the attitude of the vehicle, and illustrates a case where the balancer moves in a direction perpendicular to the axle and the vehicle central axis. A weight 134 that moves in the vertical direction is included in the main body.
The symbols in FIG. 9 are as follows.
(A) State quantity θ _W : Tire rotation angle [rad]
θ ₁ : tilt angle of main body (vertical axis reference) [rad]
λ ₂ : Balancer position (vehicle center point reference) [m]
(B) Input τ _W : Drive motor torque (two wheels total) [Nm]
S _B : Balancer driving force [N]
(C) Physical constant g: Gravitational acceleration [m / s ² ]
(D) Parameter m _W : Tire mass [kg]
R _W : Tire radius [m]
I _W : Tire inertia moment (around axle) [kgm ² ]
D _W : Viscous damping coefficient for tire rotation [Nms / rad]
m ₁ : Mass of the main body (including passengers) [kg]
l ₁ : Distance from the center of gravity of the main unit (from the axle) [m]
I ₁ : Moment of inertia of body (around center of gravity) [kgm ² ]
D ₁ : viscous damping coefficient [Nms / rad] with respect to rotation of the main body
m ₂ : Mass of the balancer [kg]
l ₂ : Balancer center of gravity distance (from axle) [m]
I ₂ : Balancer's moment of inertia (around the center of gravity) [kgm ² ]
D ₂ : Viscosity damping coefficient for balancer translation [Ns / m]

The dynamic model of FIG. 9 is expressed by a linear second-order differential equation of Equation 2.
Each value in Formula 2 is as shown in Formula 3.
Further, I _{W, a} and I _{12, a} in Expression 3 are as follows.
I _{W, a} = I _W + (m ₁ + m ₂ + m _W ) R _W ²
I _{12, a} = (I ₁ + m ₁ l ₁ ² ) + (I ₂ + m ₂ l ₂ ² )

The disturbance d (→ ∧) due to the passenger parameter variation is expressed by the following Equation 4, where ξ (→) is the vehicle state quantity, Λ is the fluctuation amount matrix, and P _d is the disturbance approach path.
In Formula 4, in the vehicle state quantity ξ (→), θ ₁ , θ ₁ (··), θ _W (··) are the main body (vehicle body) tilt angle, the main body tilt angle acceleration, and the tire rotation angular velocity as described above. It is.
Since the third row element P _d31 = 0 and P _d32 = 0 of the disturbance approach path P _d , the parameter variation of the main body does not directly affect the balancer motion characteristics.

Further, for easy handling by the disturbance observer 23, Formula 2 is expressed by Formula 5 when it is converted into a general state equation.
Each value in Formula 5 is as shown in Formula 6, and I is a unit matrix.

The disturbance observer 23 acquires an input u in the posture control (a torque command value for the tire rotation actuator 61 in FIG. 1) and an output y (the detected value (speed, vehicle body inclination angle) of the traveling and posture control sensor 40). .
Then, the disturbance observer 23 calculates the disturbance estimated value d (→ ∧) by solving Equation 1 with an estimator (FIG. 8) (step 32).
Note that the disturbance observer uses the minimum dimension observer, but can be estimated using the same dimension observer.

Here, in the model expressed by Equation 1, it is assumed that the estimated value d (→ ∧) of the disturbance is d (→ ∧ ·) = 0 (the fluctuation speed of the disturbance is slower than the estimated speed). The component reliability is low.
For this reason, as will be described later in the present embodiment, for the high-frequency component, values calculated from the measurement values of the measuring instrument (mechanical characteristic amount estimation sensor 50) (step 11 to step 13) are used instead of the disturbance observer 23. Thus, the reliability of the high frequency component is ensured.

On the other hand, the disturbance observer 23 acquires the vehicle state quantity ξ (→) from the running / attitude control sensor 40 (step 32).
Then, the disturbance observer 23 estimates the fluctuation amount matrix Λ by the least square method from the time history of the vehicle state quantity ξ (→) and the disturbance estimated value d (→ 二) (step 33).
That is, for the N discrete-time data sequences ξ ^(k) (→) and d ^(k) (→ 変 動), the variation matrix Λ (equation in Equation 4) The same).
Note that the reference time T _ref = NΔt (Δt is the time increment of the discrete data) is set longer than the estimated time of the observer.

In Equation 7, the calculation is simplified by ignoring the correlation of ξ (→), that is, by setting the off-diagonal component of the tensor product ξ ^(k) (→) ξ ^(k) (→) to 0. Also good.

Next, the disturbance observer 23 calculates the mass Δm ₁ , the primary moment Δ (m ₁ l ₁ ), and the moment of inertia Δ () as the three parameter variations of the main body (controlled object) from the values of the components in the estimated variation matrix. I ₁ + m ₁ l ₁ ² ) is estimated as shown in Equation 9 (step 34).

Formula 9
Δm ₁ = (1 / R _w ² ) Λ ₁₁
Δ (m ₁ l ₁ ) = − (1 / g) Λ ₂₃ = (1 / R _w ) Λ ₁₂ = (1 / R _w ) Λ ₂₁
Δ (I ₁ + m ₁ l ₁ ² ) = Λ ₂₂

The fluctuation amount Δ (m ₁ l ₁ ) of the _first moment can be estimated (calculated) from any of the three elements Λ ₂₃ , Λ ₁₂ , and Λ ₂₁ of the fluctuation amount matrix as shown in Equation 9.
Each element Λ ₂₃ corresponds to the state quantity θ ₁ , Λ ₁₂ corresponds to the state quantity θ ₁ (··), and Λ ₂₁ corresponds to the state quantity θ _w (··).
Therefore, in the present embodiment, the fluctuation amount Δ (m ₁ l ₁ ) of the _first moment varies within the reference time among the three state quantities θ ₁ , θ ₁ (··), θ _w (···). A highly accurate value can be calculated by using an expression corresponding to the state quantity having the maximum width.

When the mechanical characteristic amount is estimated from the measurement value of the mechanical characteristic amount estimation sensor 50 and the mechanical characteristic amount is estimated from the disturbance observer 23 by the above processing, the mechanical characteristic amount estimation control system 22 The two data are compared, and the mass, primary moment (center of gravity height), and moment of inertia, which are mechanical characteristics, are determined (FIG. 6, step 17).
In the following description, the mechanical characteristic amount Pk estimated from the measurement value of the mechanical characteristic amount estimation sensor 50 is represented, and the mechanical characteristic amount estimated by the disturbance observer 23 is represented by Pg.

In this embodiment, the mechanical characteristic amounts Pk and Pg are estimated by two methods. However, by using both of them as in the following (1) to (4), more accurate fluctuations in the mechanical characteristic amount are obtained. The estimation of is realized.
FIG. 10 shows an example of proper use of the mechanical characteristic amounts Pk and Pg.
(1) Proper use according to reliability Since the estimation system with high reliability of evaluation is different for the three elements of each mechanical characteristic quantity, as shown in FIG. 10 (a), a weight according to the reliability is given. .
The measuring instrument has a high reliability of mass (m ₁ ), and the disturbance observer has a high reliability of first moment (m ₁ l ₁ ).
Therefore, as illustrated in FIG. 10A, the mechanical characteristic amounts Pk and Pg are weighted at the following ratios and used. In addition, about the ratio of both, it is an illustration and it is also possible to set it as another value.
About mass, Pk is used 90% and Pg is used 10%.
For the first moment, 30% Pk is used and 70% Pg is used.
As for the moment of inertia, Pk is used 50% and Pg is used 50%.

(2) Use properly according to frequency component The mechanical characteristic amount Pk calculated using the measured values (steps 11 to 13) and the mechanical characteristic amount Pg estimated by the disturbance observer 23 (steps 14 to 16) are: There are upper limit frequencies f1 and f2 that can be evaluated.
That is, the mechanical characteristic amount Pk has an upper limit f1 as hardware such as a natural frequency and response performance.
On the other hand, the mechanical characteristic amount Pg by the disturbance observer 23 has an upper limit f2 for stability (robustness) determined from the estimated speed (pole).

  Therefore, in the present embodiment, the weighting based on each frequency component based on the upper limits f1 and f2 is applied to the frequency components of the two mechanical characteristic amounts Pk and Pg, as illustrated in FIG. It has been decided.

(A) For the mechanical characteristic amount Pg by the disturbance observer 23, the weight is reduced as it becomes larger than the upper limit f2 of the frequency.
(B) For the mechanical characteristic amount Pk based on the measured value, the weight is reduced as it becomes larger than the upper limit f1 of the frequency.
(C) For a range in which the disturbance observer 23 is highly reliable (frequency range below the upper limit f2), the weight for the mechanical characteristic amount Pg by the disturbance observer 23 is increased, and the weight for the mechanical characteristic amount Pk by the measured value is Make it gradually smaller.
(D) A weight of 1 or less is used for a frequency of f1 or more using only the mechanical characteristic amount Pk based on the measured value, and the sum of both weights for a frequency of less than f1 is 1.

(3) Use according to the driving state Disturbance observers have large changes in acceleration and change in posture, and cannot be estimated with high accuracy unless there is sufficient observation time.
Therefore, at the start of control (between the start of control and a predetermined time T1) or during gentle running (when the rate of change in acceleration and the rate of change in posture angle is α1% or less), the mechanical characteristic amount based on the measured value 100% of Pk is used, and its value is given as the initial value of the disturbance observer.
For other cases, it is according to (1) or (2) above.

(4) Combined use as fail-safe In this embodiment, one value is used as a fail determination index of the other value for both mechanical characteristic amounts Pk and Pg.
That is, for each frequency component, the difference between the two values is evaluated, and if the difference is large, a detailed study is performed, and the one that is more likely to be an error is regarded as a failure.
When one is determined to be failed, the other characteristic amount is used with a weight of 1.
In addition, about use of Pk and Pg as this fail safe determination parameter | index, it is always used and judged independently of said (1)-(3).

As described above, when the dynamic characteristic quantity variation (Pf) to be adopted for each frequency component is determined (step 17), the mechanical characteristic quantity estimation control system 22 determines the corresponding mechanical characteristic quantity variation Pf to each of the frequency components. Appropriate distribution to the corresponding mechanism (system) (step 18).
Here, each corresponding mechanism (system) is a control parameter characteristic correction (hereinafter referred to as a parameter change system) and a control parameter characteristic adjustment (hereinafter referred to as a parameter change system) as outlined in FIGS. 1 (c) and 1 (d). Each mechanism).

Distribution of the mechanical characteristic amount variation Pf to be dealt with to the parameter changing system and the weight adjusting mechanism is performed by any of the following methods.
(1) Use according to frequency component In other words, in the parameter changing system, since recalculation of the control parameter takes time, a weight adjusting mechanism is used in a relatively high frequency range.
In the parameter changing system, there is an upper limit frequency f4 due to calculation time and stability (robustness), and the weight adjustment mechanism has an upper limit frequency f3 due to hardware limitations, and to prevent coupling with the attitude control system. In addition, it is necessary to avoid the fundamental frequency band of the attitude control system.
Therefore, as shown in FIG. 11, the dynamic characteristic amount variation Pf determined in step 17 is distributed to the parameter changing system and the weight adjusting mechanism by weighting according to the frequency component.
Note that the system according to this embodiment is intended for quasi-stationary fluctuations in mechanical characteristic quantities (fluctuations slower than the characteristic speed of the attitude control system, for example, movements in units of several seconds), as shown in FIG. By cutting out high frequency components, fairly fast fluctuations are ignored.

(2) Substitution for capacity limit On the other hand, the weight adjustment mechanism has a qualitative limit that the mass cannot be adjusted (increase / decrease) and a quantitative limit that the mass of the weight and the movable stroke are finite.
Therefore, a part that exceeds the limit of the weight adjustment mechanism is handled and corrected by the parameter changing system.

(3) Substitution as fail safe For example, in a parameter change system, there is a possibility that the solution may diverge when the control parameter recalculation is broken, for example, when an implicit equation is solved by an iterative calculation method.
In addition, a hardware failure of the weight adjustment mechanism may occur.
Therefore, in the present embodiment, when one side breaks down, it is used to make up for it with the other side.

When the dynamic characteristic amount variation Pf to be distributed is determined, the mechanical characteristic amount estimation control system 22 corrects the control system characteristic amount of the posture control system by the parameter change system and the weight adjustment mechanism (steps 19 to 20 and step 21). -23).
FIG. 12 conceptually shows the correction of the control parameter by the parameter changing system.
As shown in FIG. 12, the control parameters are corrected to be suitable for the conditions of distribution (mass, first moment, moment of inertia) of the dynamic characteristic amount variation Pf.

This controlled object model is expressed by Expression 10, and is expressed by Expression 11 when expressed in the form of a state equation of the system so that it can be easily handled.
In Equation 10, x _s (→), u (→), M _s , C _s , and K _s are as in Equation 3 above.
In Equation 10, unlike the disturbance observer model (Equation 2) for estimating the disturbance, the term P _d d (→) does not exist. Also in Equation 11, the system state equation (Equation 5) In contrast, there is no Dd (→) term.

The dynamic characteristic quantity estimation control system 22 recalculates the optimal control parameter (feedback gain G) for the distributed dynamic characteristic quantity fluctuation Pf according to the above formula 10 or formula 11 (step 19). The control parameters are introduced into the control system (step 20).
Here, there are two control parameter calculation methods for the distributed dynamic characteristic amount fluctuation Pf: characteristic optimization parameter calculation and characteristic invariant parameter calculation.

In the characteristic optimization parameter calculation, the optimum gain G for the condition (distributed mechanical parameter amount Pf) is obtained by recalculating the feedback gain G from scratch by using an optimal regulator, for example.
On the other hand, in the characteristic invariant parameter calculation, a gain G that does not change the control system characteristic at the time of design as much as possible is calculated.

In the characteristic optimization parameter calculation, an optimum gain can be obtained for the conditions, but there is a demerit that the calculation takes time.
On the other hand, the characteristic invariant parameter calculation has an advantage that a reasonable gain can be calculated in a short time, although it is not an optimum value, and a detailed adjustment at the time of designing a control system can be utilized.
In the present embodiment, two control parameter calculation methods (setting systems) are properly used as follows.
That is, first, provisional setting is performed by characteristic invariant parameter calculation, and after the optimal value is calculated by characteristic optimization parameter calculation, the control parameter is finally changed to the optimal value.
In order to prevent the posture control from becoming unstable by changing the gain abruptly, the gain is changed smoothly by linearly changing each parameter.

Here, the characteristic invariant parameter calculation in the control parameter setting will be described.
In the attitude control system based on state feedback, a model (nominal model) before the control parameter is corrected is expressed by Expression 12, and this characteristic equation is expressed by Expression 13.

Then, when the actual characteristic of the control target expressed by Expression 14 is considered with respect to the nominal model before correction, the estimated model is expressed by Expression 15. In this case, the characteristic equation is expressed by Equation 16.
In the control target characteristics of Expression 14, M _s and K _s include distributed mechanical characteristic amount fluctuations (mass, first moment, inertia moment) as shown in Expression 3 described above.

In the mechanical characteristic estimation control system 22, the feedback gain G is corrected so that the pole arrangement of the closed loop system (basic characteristics of the feedback control system) does not change as much as possible.
The condition that the closed loop system characteristic equations (Formula 13 and Formula 16) are the same before and after the change of the control parameter (gain) is expressed by Formula 17.

The mechanical characteristic amount estimation control system 22 obtains gain correction amounts ΔG _K and ΔG _C from the two equations (17).
Here, if the matrix P _u is regular, accurate calculation is possible, and the gain can be changed so as not to change the pole arrangement at all. On the other hand, if the P _u is regular, the gain is not such as to satisfy the above equation completely, with a least square method, approximately obtaining the gain.
In general, Pu is not regular in the attitude control of a unicycle or a motorcycle, and therefore the gain correction amount (ΔG _k , ΔG _c ) is approximately obtained in the mechanical characteristic amount estimation control system 22 of the present embodiment.
In addition, after correcting the feedback gain G with the calculated gain correction amount (ΔG _k , ΔG _c ), it is necessary to confirm and calculate the stability of the closed-loop system. Attitude control is continued with the value before correction.

Next, correction of the control system characteristic amount of the attitude control system by the weight adjustment mechanism will be described.
The mechanical characteristic quantity estimation control system 22 calculates the arrangement of weights 134 (see FIG. 5) that minimizes the difference between the value based on the distributed mechanical characteristic quantity and the nominal value (control assumed value) ( Step 21) The calculated position is set as a weight target position (Step 22).
Then, the mechanical property quantity estimation control system 22 outputs a command value for moving the weight to the target position to the weight drive actuator 62. As a result, the weight drive actuator 62 moves the weight 134 attached to the vehicle body to the target position, and the value of the mechanical characteristic amount approaches the nominal value (step 23).

Specifically, the deviation of the actual mechanical characteristic amount from the nominal value (control assumed value) is expressed as follows: primary moment Δ (m ₁ l ₁ ), inertia (secondary) moment Δ (I ₁ + m ₁ l ₁ ² ) ( In order to cancel this deviation, the weight m _b ^(k) is moved in the vertical direction by Δh _b ^(k) in order to cancel this deviation.
Here, Expression 18 is an expression assuming that N weights that can move up and down independently are arranged, ^(k) indicates the kth weight among N, and l _b ^{( k)} represents the distance from the axle to the weight reference position.

Servo control to the target position determined from Equation 18 is performed for each weight (however, it is moved slower than the characteristic speed of the existing attitude control system).
In this case, if two weights 134 are used, both parameters (primary moment and moment of inertia) can be matched.
Further, by further increasing the weight 134, it is possible to reduce the weight moving distance and energy consumption.
For example, when there is one weight 134, one of the first moment and the moment of inertia is adjusted from Equation 18 (a) or (b). In this case, there is only one direct adjustment, but in general, by adjusting one of the primary moment and the moment of inertia, the other also changes in a good direction accordingly.

When there are two weights 134, both the primary moment and the moment of inertia are adjusted according to the mathematical expressions 18 (a) and (b).
Furthermore, when there are three or more weights 134, in addition to adjusting both parameters, it is possible to perform minimum movement distance control and minimum energy consumption control of the weight 134.

  As shown in FIG. 5B, the control can be performed by changing the height of the sheet in addition to the movement of the weight 134 or in place of the weight 134.

  In the embodiment described above, the attitude control in the front-rear direction in the single-shaft motorcycle has been described as an example. It is also possible to apply the method of estimating the characteristic quantity and correcting the characteristic quantity of the control system.

It is explanatory drawing showing the outline | summary of the control system characteristic quantity correction process which corrects a control system characteristic quantity to the optimal value according to the actual vehicle state. It is an external appearance block diagram of the vehicle in this embodiment. 2 is a configuration diagram of a control unit 16. FIG. It is arrangement | positioning explanatory drawing of a load meter and a sitting height meter. It is explanatory drawing of the weight arrange | positioned at the backrest part. It is a flowchart showing the contents of control system characteristic amount correction processing. It is a schematic explanatory drawing of the flow which estimates a mechanical characteristic quantity. It is explanatory drawing which represented notionally the method of the disturbance estimation by a disturbance observer. It is explanatory drawing showing the dynamic model of the attitude control system of a vehicle. It is explanatory drawing which showed an example of proper use of the mechanical characteristic amount fluctuation | variation Pk and Pg. It is explanatory drawing about distribution of the determined dynamic characteristic amount fluctuation | variation Pf. It is explanatory drawing which represented conceptually about correction of the control parameter by the parameter change system.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 Drive wheel 12 Drive motor 13 Riding part 131 Seat surface part 14 Support member 16 Control unit 20 Control ECU
DESCRIPTION OF SYMBOLS 21 Vehicle body basic control system 22 Mechanical characteristic amount estimation control system 23 Disturbance observer 30 Steering device 40 Traveling and attitude control sensor 41 Traveling speed meter 42 Vehicle body tilt angle meter 50 Mechanical characteristic amount estimation sensor 51 Load meter 52 Seat height meter 60 Actuator 61 Tire rotation actuator 62 Weight drive actuator 134 Weight

Claims (7)

A characteristic amount estimation device for estimating a mechanical characteristic amount of a weight body mounted on a vehicle that performs posture control of a vehicle body,
A boarding section for carrying heavy objects such as luggage and crew,
A load sensor disposed in the riding section;
A height sensor for measuring the height of the weight body;
Estimating means for estimating the mechanical characteristic amount of the weight body from the measurement values of the load sensor and the height sensor;
A characteristic quantity estimation apparatus comprising: A characteristic amount estimation device for estimating a mechanical characteristic amount of a weight body mounted on a vehicle that performs posture control of a vehicle body,
A boarding section for carrying heavy objects such as luggage and crew,
A load sensor disposed in the riding section;
A height sensor for measuring the height of the weight body;
Direct estimation means for estimating a mechanical characteristic amount from measured values of the load sensor and the height sensor;
An indirect estimation means for estimating a mechanical characteristic amount by a disturbance observer;
An estimation means for estimating a mechanical characteristic amount based on the estimated values by the direct estimation means and the indirect estimation means;
A characteristic quantity estimation apparatus comprising: Using both estimated values estimated by the direct estimation means and the indirect estimation means, comprising error determination means for performing error determination of an estimated value;
The estimation means estimates the other estimated value as a mechanical characteristic amount when one estimated value is determined to be an error;
The characteristic quantity estimation apparatus according to claim 2 , wherein: The estimation means estimates a mechanical characteristic amount based on a frequency component of an estimated value by the direct estimation means and the indirect estimation means;
The characteristic quantity estimation apparatus according to claim 2 or claim 3 , wherein The estimation means estimates the mass of the weight body, the height of the center of gravity of the weight body, and the moment of inertia of the weight body as mechanical characteristic amounts.
The characteristic amount estimation apparatus according to any one of claims 1 to 4 , wherein the characteristic amount estimation apparatus is characterized in that: An acceleration sensor;
An inclination angle sensor,
The estimation unit or the direct estimation unit estimates a mechanical characteristic amount of the weight body from the load sensor, the measurement value of the height sensor, and the detection value of the acceleration sensor and the inclination angle sensor.
The characteristic amount estimation apparatus according to claim 1 or 2 , characterized in that A characteristic quantity estimating apparatus according to any one of claims of claims 1 to 6,
A type determination means for determining a type of a weight body mounted on the vehicle using a mechanical characteristic amount estimated by the characteristic amount estimation device;
A mounted object determination apparatus comprising:

Images