JPH04195302A - Adaptive controller - Google Patents

Abstract

PURPOSE:To improve the accuracy of an arithmetic value by using differential values of an output and history values of an input and to speed up the convergence of an estimated value by providing a control input characteristic arithmetic means, a control input characteristic judging means, and a control input arithmetic means. CONSTITUTION:For the controlled system 100 of a one-input one-output first order delay system whose actual input distribution characteristic is unknown and different from a control input characteristic, the control input characteristic arithmetic means 110 performs arithmetic operation by using many differential values of the output and history values of an input to improve the accuracy of individual arithmetic values, and the estimated value can be converged fast, so that the output of the controlled system 100 can excellently follow up a target track. When the control input arithmetic means 104 is composed of a digital circuit system, the adaptive controller can easily be realized by setting time delay to an integral multiple of a sampling period and similar effect can be obtained.

Description

[Detailed Description of the Invention] Industrial Field of Application The present invention relates to an adaptive control device for controlling objects that are significantly affected by unknown characteristics or disturbances. High precision and intelligence are required, and in cases where the characteristics of the target change unpredictably or where the influence of external disturbances is significant, L is able to estimate these unknown factors in real time from the input/output signals of the target. Adaptive control devices that enable devices and systems to always exhibit desired characteristics by adjusting the control system in real time based on the results are attracting attention.
Paper by Ito Narayana in the Proceedings of the 31st Automatic Control Association Conference, published in October 1988, pages 361-362 Conventional control based on ``time delay control of systems with unknown input characteristics'' Time delay control using a calculation method of input characteristics will be explained.

Hereinafter, an example of a conventional adaptive control device will be described with reference to the drawings. Motor body 50
1 is a motor rotation line 502 is a load support member, 503 is a viscous friction member, and 504 is a movable load.As shown in FIG. Even if the member 502 is fixed, let C be the unknown viscous frictional resistance mainly caused by the viscous friction member 503 fixed outside.
Further, the load support member 502 is a linear motor, and the load 504 can be moved to any position on the load support member 502 even while the motor is rotating, and the moment of inertia can be varied.

For such a controlled object, the input torque and rotational angular velocity of the servo motor body 500 are expressed as u(t) and y, respectively.
(t) Let us consider the case where the speed of the servo motor is controlled.The differential equation governing the motion of the servo motor (mu (t) = J (t)・y (t) + c (t) )・
y (t) (1) The above equation is a differential equation for one person's car output, and all the coefficients are square. Laplace transform the above equation. Here, a=c/JX b=1/J
(3) From equations (2) and (3), the controlled object is a first-order lag system, and 1/J represents the variable input distribution characteristic. Here, the input distribution characteristic is 0, i=n to 1
FIG. 4 shows a block diagram of a conventional adaptive control device configured to control the speed of a servo motor that varies up to and over a range of 100 to 100 m. In FIG. 4, 100 is the servo motor of FIG. 5 which is the controlled object; 102 is an output value detection means; 104 is a control input calculation means; 106 is input application half-immersion; 408 is a control input characteristic judgment means; 410 is a control input characteristic calculation means. The operation of the conventional adaptive control device configured as described above will be explained. First, the rotational angular velocity and rotational angular acceleration of the servomotor, which is the controlled object 100, are detected by the output value detection means 102, which includes a tachometer, an acceleration sensor, and a processing circuit thereof.

Then, the output y output from the output value detection means 102
(t) and the differential value y(t) of the output and the control input calculation means 10
The target trajectory y of the rotational angular velocity freely given by the time trajectory output from the target trajectory output circuit (not shown) in 4.
Using d(t) and the differential value yn(t) of the target trajectory, the error calculation circuit (not shown) in the control input calculation means 104 similarly calculates the relationship of the following formula % formula % (4) and calculates the error. Then, the control input calculation means 104 outputs the differential value of the target trajectory output from the control input calculation means 104.The error and output value detection means 102
The configuration is such that the control input u (t) is obtained by calculating equation (5) using the output output from the
) 10 y −(t)+ k −e (t)) / b−
t (5) In the control input calculating means 104, L is an arbitrary time delay h u (t-L) is the input applied to the controlled object 100 before the clock time y (t-L) is the input signal applied to the controlled object 100 before the clock clock ticks The rotational angular acceleration & detected by the output value detection means 102 from the object 100 has an error feedback coefficient b.
-1 is the control input characteristic used to find the control input u (t). Here, since the current b is unknown, the initial value of b set is set to 1, which is the maximum value of b, so as not to oscillate. It's only LB. t is large (if the control input calculation means 104 calculates equation (5) while it is different, the followability of the target trajectory of the controlled object 100 is poor t~
Also, since b fluctuates, it remains constant. Since tracking performance is poor at t, it is necessary to change b·-1 to the vicinity of the current b.

The actual input distribution characteristic is estimated and set as the control input characteristic b set.

For this purpose, the control input characteristic determining means 408 determines the following equation: u(t) −u(t−L)>w
(6) Here, W may be a small positive value set in advance to prevent division by zero and maintain the accuracy of the calculated value. If equation (6) holds true, the control input characteristic calculation means 410 calculates the following equation. Calculated value of control input characteristics b0-+(t) (
This is also called the calculated value. ) even if you ask for it! (iii) The control input characteristic determining means 408 directly changes the calculated value calculated by equation (7) to the current control input characteristic and determines it as a new control input characteristic b・-5(t)=b
o−+(t) (8) Or after averaging the calculated values or performing LPF processing as shown in the following equation (hereinafter, the calculated value after processing is referred to as the estimated value), the current control input characteristics are This is changed to create a new control input characteristic.

b--1(t)=Σb-s+(t)/n
(9) The input characteristic determining means 40 controls the above determinations.
8 will be done.

For example, there are various possible judgment criteria for the sumo wrestler lO b
If the condition that the variation in ee1 is within 20% (1o) is satisfied, the current control input characteristics are changed to new control input characteristics.Here, if equations (6) and (10) do not hold, then the calculation and Do not change the control input characteristics (°C).The control input characteristics obtained by the control input calculating means 104 using
is converted into an input torque u(t) and applied to the controlled object 100 by an input applying means 106 composed of an actuator and its drive circuit. Then, the output y(t) of the controlled object 100 is determined by the applied input torque u(t).
However, the above configuration has the following problems: 4 [1] Influence of unknown characteristics other than input distribution characteristics Variations in the calculated values of the control input characteristics become large due to the influence of the delay difference, dead time, and the inability to directly detect the differential value of the output. The ability to follow the target trajectory may deteriorate or the operation may become unstable.
Therefore, even if LPF processing or averaging processing is performed on the calculated values to increase stability, the accuracy of each calculated value is low, and it takes time to obtain control input characteristics (estimated values) close to the input distribution characteristics. . Controlling a controlled object using its slow convergence control input characteristics has the problem that the output of the controlled object cannot follow the target trajectory well. Similarly to the controlled object of the system, a high-order delayed system with a one-car output had the problem of not being able to follow the target trajectory well [3] Similarly, when the output of the controlled object was not set to the target trajectory, The problem is that it is not possible to follow the output of the reference model well.In addition, the estimation method for an adaptive control device proposed earlier in Japanese Patent Application No. 2-274.96 is for multiple inputs and multiple outputs. , - Input - Output have similar problems. In view of the above problems, the present invention [1] [2] The actual input distribution characteristic is unknown and the control input characteristic b *st is different from [1] - Input-output first-order delay [2] - For the controlled object of the input-output higher-order delay system,
The object of the present invention is to provide an adaptive control device in which the accuracy of each calculated value of unknown input distribution characteristics is high and the convergence of estimated values is quick, and the output of a controlled object follows a target trajectory well.

[3] Similarly, an adaptive control device is provided in which the output of a controlled object closely follows the output of a reference model. Means for Solving the 4 Problems In order to solve the above problems, the adaptive control device of the present invention has the following features: It has the following configurations. Immediate board [1] y (s)/ u (s) = b/ (s +
For a controlled object with one input and one output that can be expressed by
Let any m of them be Y1, and the remaining m be Y2, and use the inputs and their antecedent values for a total of 2m ((among them, m that correspond to each time in Y1 are U1, and the remaining m are U2 Let the calculated value be-+(t) of the control input characteristic be, l (t)
= (Y IY2) / (U l - U 2) A control input characteristic calculation means calculates, and the control input characteristics calculation means determines whether or not to perform the calculation and calculates the value after processing of the calculated value of the control input characteristic. a control input characteristic determining means for determining whether to change to a new control input characteristic bea%;
dy (t-L)/dt+dy m(t)/dt-
e (t)]/bomtJ: a control input calculation means [2] y (s)/ u (S) = b/ (s +
For a controlled object with one input and one output that can be expressed as a Bind any m items as Y1 and the remaining m as Y2.For a total of 2m inputs and their antecedent values (1), set the m pieces corresponding to each time of Y1 as U1, and the remaining m pieces as U2.Control The calculated value bO-1(t) of the input characteristic is expressed as bom1(t)=(Yl
-Y2)/(01-U2), the control input characteristic calculation means determines whether or not to perform the calculation, and the processed value of the control input characteristic is used as a new control input. control input characteristic determining means for determining whether or not to change the characteristic to the best characteristic;
L)/dtn”+dny *(t)/dtn+ k−+
・dn-1e (t)/dtn 10”・+k @・
e (t)]/bomt [3] y (s)/u (s)
= b / (s + a) For a controlled object with one input and one output, a total of 2m output differential values and their antecedent values are used (any m of them are Y1, and the remaining Let m pieces be Y2, and use a total of 2m inputs and their antecedent values (of which, m pieces corresponding to each time of Y1 are U1,
Assuming the remaining m pieces as U2, the calculated value of the control input characteristics is −+(
t) to b・-1(t)=(YIY2)/(U 1-U
2), the control input characteristic calculation means calculates whether or not to perform the calculation, and converts the processed value of the control input characteristic into a new control input characteristic ba*t. control input characteristic determining means for determining whether to change or not;
(t-L) + [-dy (t-L)/dt -a
,・y (t)+ b−・r (t)k −e (t
)]/bomt The effect of the above-described structure of the present invention (1) The actual input distribution characteristics are unknown. [1] - input-output-order delay system [2] - input-output higher-order delay monitoring By calculating the control input characteristics using a large number of prior history values of the input and output values, the accuracy of the calculated values can be improved.The convergence speed of the estimated values is faster, and the output of the controlled object can better follow the target trajectory. becomes.

Here, it takes time to obtain the calculated value because many antecedent values are used (as in the conventional example, a small number of antecedent values are used to quickly obtain the b oal, and various processes are performed to obtain the estimated value). It takes time, so in the end, the better each calculated value is, the faster the estimated value will converge to the actual value.
(iii) If the control input calculating means 104 uses an estimated value that fluctuates before convergence, the input will oscillate, so the estimated value will converge even more slowly [3] Similarly, the output of the controlled object can be adjusted to the output of the reference model. Embodiment First, the basic concept of the method of calculating control input characteristics of the present invention will be explained below. Even if we use the method of using the differential value of or the antecedent value of the input, we can consider a controlled object similar to the conventional example. Here, S is the Laplace operator, and y is the output of the controlled object.
If a is an unknown characteristic, ib is an unknown input distribution characteristic, and iu is an input, the above equation can be expressed as a differential equation as shown in the following equation b (t)
・u (t)= y (t)+ a (t)・y (t
) (12) Here, a (t)・y (t
) as the disturbance d(t), and the times t-t, t-L,...
, by substituting t-(2n-1)・L, we get b (t)・
u (t)-y (t)+ d (t)b (t-L)
・u (t-L)=V (t-L)+d (t-L
) (13)b (t-(2m-1)・L)・u
(t-(2m-1)・L)=y(t-(2m-1)
)・L)+d (t-(2+n-1)・L)In the above formula, b(t)'=b(t-L)'q,-,=b(t-(2m
-1)・L) (14) d(t)=d(t-L)',
,・=,==d(t-(2m-1)・L) (15)
Then, take out any m of the 2m equations in equation (13) and add them. Here, for simplicity, we select m locations near the current time, and then b (t) i u (
t)+...-+u (t-(m-1)・L))=y(t
)+”・+y(t-(m-1)・L)+m-d(t)
(16) Furthermore, by adding the remaining m equations, we get
b (t)(u (t-m-L)+----+u (t-
(2m-1)・L))=y (t-n+-L)+・+
y (t-(2m+-1)・L) 10m-d (t)(1
7) By subtracting and rearranging equation (17) from equation (16), the following equation is obtained. )−y (
t-+n-L)-・-y (t-(2m-1)・L)/
(u (t)+・−+u (t−(, m−1)・L)−
u (t-m-L)--u (t-(2m-1)4))
(18) As above, the calculated value of b can be found. Theoretically, equation (18) is derived. Let M L be the sampling period, and if we consider that input and output are detected at each sampling period, input u (t) The differential value of the latest output that shows the effect of adding is y (t + t,).a Nodal plate If we consider that the input corresponding to the output y (t) is u (t - L), then the above equation can be calculated by dividing the input only at time t. Shift b ,, l (t)= b (t)=y (t)+・
・・+di(t-(m-1)・L)-y (t-m-L)
−・・= −y (t−(2n−1)−L)/(u
(t-L)+...+u (t-m-L)-u (t
-(m+1')・L>------u (t-2m4)
) (19). As shown in the above formula, it is necessary to match the input and the output where the influence of the input appears. For example, when using eight inputs or the previous history value of the input, the formula (19
) is written as the following formula, 1(t)=y(t)+y(t-3L)-y(t-
4L) -...-y (t-7L)/(u (t-L)
+・-・+u (t-4L)-u (t-5L)--u
(t-8L)) If the differential value of y cannot be detected and is calculated using the backward difference of y, then the following equation is obtained:b--+(t)=y(t)-2y(t- 4L)+y (t-8L)/
L・(u (t-L)+-・+u (t-4L)-u
(t-5L)--...-u (t-8L))
(21) Next - Consider a controlled object expressed by a high-order delay system of input and output as shown in the following equation u (s) s''+an-+・s'-'+-・+
am (22) Similarly, for the controlled object of the Mu equation (22) with 1+(i-n-1~0) as the unknown characteristic,
) can be easily calculated by replacing the -th differential value of (t-(+n-1)・L
)-u (t-m・L)-...-u (t-(2+n-
1))) (23) Here, the conditions for obtaining a good calculated value are as follows.

・Corporate formula calculation formula (20
) is constant and changes smoothly. (25)・Furthermore, the formula (2
(26) The concept of obtaining a better calculated value by performing calculations only when the above conditions are satisfied. Based on this, the purpose of the adaptive control device in the first embodiment of the present invention is to: The object of the present invention is to provide an adaptive control device in which the accuracy of each calculated value is high and the estimated value converges quickly, and the output of a controlled object follows the target trajectory well. 1 is a block diagram of an adaptive control device according to a first embodiment of the present invention.
00 is the servo motor in FIG. 5 which is the controlled object; 102 is the output value detection means; 104 is the control input calculation means; 106 is the input application means; 108 is the control input characteristic judgment means; 2
Reference numeral 10 denotes control input characteristic calculation means.The operation of the adaptive control device according to the first embodiment of the present invention constructed as described above will be explained.

Considering a controlled object 1°O that has the same configuration and operation as the conventional example, y(s)/u(s)=b/(s+a)0, i=
n≦b≦1 (27) The rotational angular velocity and rotational angular acceleration of the servo motor, which is the controlled object 100, are detected by the output value detection means 1゜2 consisting of a tachometer, an acceleration sensor, and a processing circuit thereof. Output y output from the detection means 102
(t) and the differential value y(t) of the output and the control input calculation means 10
The target trajectory y of the rotational angular velocity freely given by the time trajectory output from the target trajectory output circuit (not shown) in 4.
Using n(t) and the differential value Ωd(t) of the target trajectory, the error calculation circuit (not shown) in the control input calculation means 104 similarly calculates the relationship of the following formula % formula % (28) and calculates the error. Then, the control input calculation means 10.4 is a differential basket of the target trajectory output from the control input calculation means 104.The error and output value detection means 10
The configuration is such that equation (29) is calculated using the output from 2 to obtain the control input u (t).

u (t)= u (t-L)+(-V (t-L)+
Sl dt) + k −e (t)) /
b --t (29) In the control input calculation means 104, L is an arbitrary time delay h u (t-
L) is the ambition applied to the controlled object 100 before the time Ω
(t-L) is the rotational angular acceleration detected by the output value detection means 102 from the controlled object 100 before the time, k is the error feedback coefficient, and bas( is the control input used to obtain the control input u (t) The characteristics and operations above are the same as those of conventional adaptive control devices.

In the present invention, in order to use the arithmetic expression (20), the control input characteristic determining means 108 determines the following equation. 4 u (t-L)+...-+u (t-4L)-u (t
−5L)−・−u (t−8L)>w (w is a small positive value) (30) If equation (30) holds true, the control input characteristic calculation means 110 calculates the control input characteristics by calculating the following equation. Also find the value bo=+(t), self+ (1)=9(t)+...+ji(t-3L)-Ω(t-4L)-
...-Ω(t-7L)/(u (t-L)+-+
u (t-4L)-u (t-5L)-・-u (t-
8L)) Then, the control input characteristic determining means 108 determines the condition of the following equation: The variation of the 10 bo1 values is within 20% (32) If the above condition is satisfied, the control input characteristic determining means 108
Further, the 10 calculated values are averaged as shown in the following equation.

b-, ai(t) = Σ b o-+(t)/
10 (33) Then, change the current control input characteristic to a new control input characteristic. If equations (30) and (32) do not hold, do not change the I'L calculation or control input characteristics (℃Here, since the current b is unknown, the initial value of b est is set so that b est does not oscillate. is set to l, which is the maximum value of .

Then, the control input u (t
) is converted into an input torque u(t) and applied to the controlled object 100 by an input applying means 106 composed of an actuator and its drive circuit. Then, the output y(t
) is a target trajectory type adaptive control device.

As described above, according to this embodiment, the control input characteristic calculation means 110 differentiates the output for the controlled object 100 of the one-person car output car next delay system whose actual input distribution characteristic is unknown and different from the control input characteristic bit. By performing calculations using a large number of prior history values of values and inputs, the accuracy of each calculated value improves, it becomes possible to quickly converge the estimated value, and it is possible to make the output of the controlled object 100 follow the target trajectory well. It goes without saying that when the control input calculating means 104 is configured with a digital circuit system, an adaptive control device can be easily realized by setting the time delay L to an integral multiple of the sampling period, and the same effect can be obtained. (℃) It goes without saying that the output value detection means 102 cannot directly detect the differential value of the output, but obtains the same effect by calculating the differential value of the output. Similarly, the second embodiment of the present invention will be described. do.

The purpose of the adaptive control device according to the second embodiment of the present invention is to perform individual calculations on a controlled object of a one-person car output high-order delay system whose actual input distribution characteristics are unknown and whose control input characteristics differ from b and 6t. The purpose of the present invention is to provide an adaptive control device that has good value accuracy, fast convergence of estimated values, and allows the output of a controlled object to follow the target trajectory well.The adaptive control device in the second embodiment of the present invention will be described below. This will be explained with reference to the drawings, and FIG. 2 shows a block diagram of an adaptive control device according to a second embodiment of the present invention.
00 is the controlled object, 202 is the output value detection means, 204 is the control input calculation means, 106 is the input application half-immersion, 208 is the control input characteristic judgment means, and 210 is the control input characteristic calculation means. Means for performing operations similar to those in the first embodiment are given the same numbers as in the first embodiment, and the points different from the first embodiment regarding the adaptive control device of the second embodiment of the present invention configured as 4 or more are as follows. We will mainly explain its operation.

Control object 2 of a single car output high-order delay system that can be expressed by the following equation
Considering 00, y (s)/ u (s) ==b/(s'+an-+・s'-'+...-+as)
(34) Then, the output y (t) output from the output value detection means 202 and the differential value dy (
t)/dt~dny(t)/dtn and the target trajectory ym(t) freely given by the time trajectory output from the target trajectory output circuit (not shown) in the control input calculating means 204.
and the differential value dyn(t)/dtn- of the target trajectory up to the nth floor.
Using dny n(t)/dtn, the error calculation circuit (not shown) in the control input calculation means 204 similarly calculates the relationship of the following formula % formula % (35) and outputs the error e. (t) and error n-
Differential value up to the first floor de (t)/dt-d” e (t
)/dtn-'a Then, the control input calculating means 204 calculates equation (36) to obtain the control input u (t) and outputs it.
(t-L)+[-d″'l (t-L)/dtn+ d
n y 4(t)/dtn+ k n-16'-' e
(t)/dtn-1+...10 km me-e (t)]
]/b--136) In the control input calculation means 204, dny(t-L
)/dtn is the nth-order differential tt of the output y(t) detected by the output value detection means 102 from the controlled object 200 before the time
L k + (>0, i=n-n-1~0) is the error feedback coefficient b-t is the control input characteristic used to obtain the control input u (t) 4 Here, b・-1 is In order to make the change, the control input characteristic determining means 208 first determines the following equation.

u (t-L)+・-・+u (t-4L)-u (t
−5L)−・−u (t−8L)>w (w is a small positive value) (37) If equation (37) holds, the control input characteristic calculation means 210 calculates the control input characteristics by calculating the following equation. Find the value bO-+(t).

boa+(t)=boa(t-L)+・-・+u(t-4L)-u (
t-5L)--u (t-8L)) Then, the control input characteristic determining means 208 determines the condition of the following formula, and the variation of 10 bo・1 is within 20% (39)
If the above conditions are satisfied, control input characteristic determining means 20
8 further averages the ten calculated values as shown in the following equation.

b −−s(t)=Σb−+ (t)/10
(40) Then, change the current control input characteristic to a new control input characteristic. Here, equation (37),
If (39) does not hold, the calculation and control input characteristics are not changed (°C, and the control input u (t) obtained in the same manner as in the first embodiment is applied to the input torque U (1) by the input applying means 106. and is applied to the controlled object 200.

As described above, according to the present embodiment, the control input characteristic calculation means 210 is used for the controlled object 200 of a one-input, one-output, high-order delay system whose actual input distribution characteristic is unknown and which differs from the control input characteristic b *a (. By calculating using many nth-order differential values of the output and prior history values of the input, the accuracy of each calculated value improves, and it becomes possible to quickly converge the estimated value.
This means that the output of 00 can be made to follow the target trajectory well.

Similarly, the third embodiment of the present invention will be described.The purpose of the adaptive control device of the third embodiment of the present invention is to control the control input characteristics because the actual input distribution characteristics are unknown.

For the controlled object of the one-person car output car next delay system that differs from t, the accuracy of each calculated value is high and the convergence of the estimated value is fast.
The present invention provides an adaptive control device in which the output of a controlled object closely follows the output of a reference model. The adaptive control device according to the third embodiment of the present invention will be explained below with reference to the drawings, but FIG. 3 shows a block diagram of the adaptive control device according to the third embodiment of the present invention. 3, 100 is the servo motor of FIG. 5 which is the controlled object, 102 is the output value detection means, 304 is the control input characteristic hand i, 106 is the input application hand, 108 is the stump ratio for each control input characteristic, and 110 is the control input characteristic calculation. Here, the same numbers as in the first embodiment are given to the means for performing the same operations as in the first embodiment, but the adaptation of the third embodiment of the present invention configured as above is The operation of the control device will be explained with a focus on the differences from the first embodiment, but a controlled object 100 having the same configuration and operation as the first embodiment will be described.
Similarly, the output or the differential value of the output is the output value detection means 1
Detected by 02. and control input calculation means 30
The reference model response output circuit (not shown) in 4 is a(bs
is the coefficient of the normative model and r(s) is the input of the normative model, ym(s)/r (s) = bm/ (s + am
) One input defined by (41) −
An error calculation circuit (not shown) in the control input calculation means 304 outputs the output Y@(S) of the output reference model.
is the output ym(t) of the reference model and the output y of the controlled object (
t) difference e (t) = y plow (t) −y (t)
(42) and outputs the error 4 And the control input calculation means 30
4 calculates the control input u(t) by calculating equation (43) and outputs it.
) −am・'t (t) tensu,・r (t)+k
−e (t)) / b,,t (43) where,
In order to change b s, ai, the control input characteristic determining means 108 first determines the following formula: mu (t-L)+-.-+u (t-4L)-u (t
−5L)−・”−u (t−8L)>w (w is a small positive value) (44) If equation (44) holds, the control input characteristic calculating means 110 calculates the following equation to calculate the control input characteristic. Calculating the calculated value b・, ai(t), b−−+(t)=y(t)+…+V(t−3L)bow(t−4L)
--- Sadness (t-7L)/(u (t-L) ten--
-10u (t-4L)-u (t-5L)=---u
(t-8L)) And control input characteristic determining means 108
determines the condition of the following equation.

The variation of 10 b o-1 is within 20% (46
) If the above conditions are satisfied, control input characteristic calculation means 1
08 further averages the 10 calculated values as shown in the following formula: b s, ai(t) = Σb o-+(t)/10
(47) Then, change the current control input characteristic to a new control input characteristic. Here, equation (44)
, If (46) does not hold, the calculation and control input characteristics will not be changed.1,% And if such changes are possible...The control input u (t) obtained by the control input calculation means 304 using t. is converted into an input torque U(1) by the input applying means 106 and applied to the controlled object 100.

It is a reference model type adaptive control device that controls the output y(t) of the controlled object 100 based on the applied input torque u(t).

As described above, according to this embodiment, the actual input distribution characteristics are unknown and the control input characteristics are unknown. One input and one output different from t -
Control input characteristic calculation means for the controlled object 100 of the next lag system! 10, by performing calculations using many output differential values and input history values, the accuracy of each calculated value improves and it is possible to quickly converge the estimated value. Then, the obtained control input characteristics were used. By calculating the control input calculation means 304 to obtain and output the control input u(t), it is possible to make the output of the controlled object 100 closely follow the output of the reference model. It goes without saying that the same effect can be had on the controlled object of a high-order delay system with a one-person car output of 1.X.

Effect of the invention As described above, the present invention has the following effects.

y(s)/u(s)=b/(s+a) For the controlled object of one-person car output, a total of 2m differential values of the output and its antecedent values are used (\any m of them) Let Y1 be Y1, and let Y2 be the remaining m inputs. For a total of 2m inputs and their antecedent values, t, X, and m corresponding to each time of Y1.
The number is U1, the remaining m number is U2, the calculated value b0-+(t) of the control input characteristic is bo, +(t)=(Yl-Y2)/
(Ul-U2), the control input characteristic calculation means determines whether or not to perform the calculation, and the calculated value of the control input characteristic is converted into a new control input characteristic b... control input characteristic determining means for determining whether to change the control input u (t) to u (t);
= u (t-L)+ [-dy (t-L)/dt+
dy *(t)/dt+ k -e (t)ko/b
By providing a control input calculation means that calculates based on −t, the accuracy of the calculated values is improved by calculating the control input characteristics using many output differential values and input previous history values. It gets faster. Since the control input characteristic calculation means calculates the control input using the control input characteristics, it is possible to cause the output of the controlled object to follow the target trajectory well.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an adaptive control device according to a first embodiment of the present invention. FIG. 2 is a block diagram of an adaptive control device according to a second embodiment of the present invention. FIG. 3 is a block diagram of an adaptive control device according to a second embodiment of the present invention. FIG. 4 is a block diagram of the adaptive control device in the embodiment. FIG. 5 is a perspective view of a servo motor taken as an example of the object to be controlled by the adaptive control device. Servo motor 102... Output value detection half-immersion 104, 204, 3o4... Control input calculation means,
106... Input application early fall 108.208... Control input characteristic-specific stamp i 110, 210... Name of control input characteristic calculation means agent Patent attorney Akira Kokaji and two others 9θ・-BooxS motor body 5θ/ - Axis fOZ to the motor @ - Single stomach

Claims (3)

[Claims] (1) s is Laplace operator, y(s) is output, u(s)
As input, a is the unknown characteristic, and b is the unknown input distribution characteristic,
For a controlled object with one input and one output that can be expressed as y(s)/u(s)=b/(s+a), t is time and y_d
(t) is the target trajectory of the output, e(t) is e(t)=y
Error defined by _d(t)-y(t), k(>
0) is the error feedback coefficient, de(t)/dt=-
When k·e(t) is a target error characteristic and b_o_m_t is a control input characteristic, an output value detection means for outputting the output y(t) of the controlled object and the differential value dy(t)/dt of the output;
A total of 2m differential values of the output and its antecedent values are used, any m of them are designated as Y1, and the remaining m are designated as Y2, a total of 2m of the input and its antecedent values are used, and each of Y1 among them is used. The m pieces corresponding to the time of 1 are U1, the remaining m pieces are U2, and the calculated value b_o_m_1(t) of the control input characteristic is b_o_m
A control input characteristic calculating means that calculates by _1(t)=(Y1-Y2)/(U1-U2), and determining whether or not to perform the calculation by the control input characteristic calculating means and determining the calculated value of the control input characteristic. control input characteristic determining means for determining whether or not to change the processed value to the new control input characteristic b_o_m_t;
-L)+[-dy(t-L)/dt+dy_d(t)/
dt+k·e(t)]/b_o_m_t; and input applying means for applying the control input u(t) obtained by the control input calculating means to the controlled object. Features adaptive control device. (2) s is Laplace operator, y(s) is output, u(s)
is input, a_i (i=n-1~0) is an unknown characteristic, and b is an unknown input distribution characteristic, y(s)/u(s)=b/(s
^n+a_n_-_1・s^n^-^1+...+a_
m), where t is the time, y_d(t) is the target trajectory of the output, and e
(t) is the error defined by e(t) = y_d(t) - y(t), k_i (>0, i = n-1~0) is the error feedback coefficient, d^ne(t)/dt ^n=-k
_n_-_1・d^n^-^1e(t)/dt^n^-
^1-...-k_m・e(t) is the target error characteristic, b_
When o_m_t is the control input characteristic, the output y(t) of the controlled object and the differential value dy(t)/dt of the output up to the nth order
Using an output value detection means that outputs ~d^ny(t)/dt^n, a total of 2m nth-order differential values of the output and its antecedent values, any m of them are set as Y1, and the remaining m Let Y2 be the input, and use a total of 2m of the inputs and their previous history values, of which Y1
The m pieces corresponding to each time of are U1, and the remaining m pieces are U2.
and the calculated value b_o_m_1(t) of the control input characteristic is b
_o_m_1(t)=(Y1-Y2)/(U1-U2)
control input characteristic calculation means for calculating, and determining whether or not to perform calculation by the control input characteristic calculation means, and changing the processed value of the calculated value of the control input characteristic to the new control input characteristic b_o_m_t. control input characteristic determining means for determining whether the control input u(t) is
=u(t-L)+[-d^ny(t-L)/dt^n+
d^ny_d(t)/dt^n+k_n_-_1・d^
n^-^1e(t)/dt^n^-^1+...+k_
m・e(t)]/b_o_m_t; and input applying means for applying the control input u(t) obtained by the control input calculating means to the controlled object. Features adaptive control device. (3) s is Laplace operator, y(s) is output, u(s)
As input, a is the unknown characteristic, and b is the unknown input distribution characteristic,
For a controlled object with one input and one output that can be expressed by y(s)/u(s)=b/(s+a), with a_m and b_m as coefficients of the reference model, y_m(s)/r(s)= b
The input of the one-input, one-output normative model defined by _m/(s+a_m) is r(s), the output is y_m(s), t is time, and e(t) is e(t)=y_m(t). −
Error defined by y(t), k(>0) as error feedback coefficient, de(t)/dt=-(a_m+k
)・e(t) is a target error characteristic and b_o_m_t is a control input characteristic, an output value detection means for outputting the output y(t) of the controlled object and the differential value dy(t)/dt of the output;
A total of 2m differential values of the output and its antecedent values are used, any m of them are designated as Y1, and the remaining m are designated as Y2, a total of 2m of the input and its antecedent values are used, and each of Y1 among them is used. The m pieces corresponding to the time of 1 are U1, the remaining m pieces are U2, and the calculated value b_o_m_1(t) of the control input characteristic is b_o_m
A control input characteristic calculating means that calculates by _1(t)=(Y1-Y2)/(U1-U2), and determining whether or not to perform the calculation by the control input characteristic calculating means and determining the calculated value of the control input characteristic. control input characteristic determining means for determining whether or not to change the processed value to the new control input characteristic b_o_m_t;
-L)+[-dy(t-L)/dt-a_m・y(t)
+b_m・r(t)+k・e(t)]/b_o_m_t
1. An adaptive control device comprising: a control input calculation means for calculating the control input by the control input calculation means; and an input application means for applying the control input u(t) obtained by the control input calculation means to the controlled object.