JP2000148205A - Coefficient deciding device for command arithmetic expression of feedback controller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily change and control a feedback controller by giving a command signal to a control object which operates the physical value in response to the command signal and outputs the observation value showing the operated physical value to the feedback controller in order to turn the observation value into the target value and then deriving an arithmetic expression. SOLUTION: An actuator 1 operates the physical value in response to a command signal and outputs the observation value showing the operated physical value to a controller 3 serving as a feedback controller. The test signal input means 4 and 6 give the pulse signals, i.e., command signals to the actuator 1 and a plant 2. Then a signal recording means 7 reads-in time series the observation value, the pulse signal value and the target value that is given from a host computer. Based on the observation value that is read in time series, a means 8 generates a Markov parameter. A means 9 calculates a minimum square evaluation function from the Markov parameter and obtains an arithmetic expression that minimizes the calculated evaluation function and turns it into a low dimension to turn the observation value into a command signal.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION The present invention relates to a device or equipment (hereinafter referred to as "control object") which operates a physical quantity in response to a command signal and outputs an observed quantity representing the resulting physical quantity to the controller, and sets the observed quantity to a target value. In particular, the present invention relates to an auxiliary device used for setting an arithmetic expression for generating a command signal based on an observed value in the feedback controller.

[0002]

2. Description of the Related Art Feedback control is employed in many technical fields. For example, a roll driving device for rotating a roll such as a rolling mill includes a driving motor, a reduction gear or a speed increasing gear, a roll, a shaft (spindle), and the like. In a control system of this type of device, generally, a speed signal from a speed detector attached to a driving motor,
A feedback controller for controlling the driving state of the motor according to the result of comparison with the target speed is used.

In the design, setting, change, and adjustment of a feedback controller, a time series input signal (command signal) and an output signal (observed value) of a control object are used to determine an operation model of the control object (output characteristics with respect to input). The parameters (coefficients of the mathematical expression) of the mathematical expression representing the mathematical expression are identified (set by obtaining a value that matches the operation characteristic of the controlled object), and the input of the controller that is highly consistent with this operation model is identified. In some cases, an output characteristic (a formula for calculating a command signal corresponding to an observed value) is obtained.

However, when an operation model of a controlled object is identified, if a feedback controller exists in the controlled object,
Since there is a correlation between the input signal and the output signal, it is impossible to perform mechanically accurate identification. Therefore, the identification is manually performed by trial and error, and the parameters of the operation model of the control target cannot be easily determined.

In order to identify an operation model of a controlled object, the order of the controlled object (the number of parameters of the operation expression) is always determined, and a modeling error that ignores components (parameters) higher than that is determined. May occur and may not accurately represent the actual behavior of the controlled object. When there are many parameters to be controlled or when the operation is non-linear, the operation characteristics of the controlled object are simulated by a linear equation, so that it is difficult to design a feedback controller. Adjustments are being made.

In order to improve this, for example, Japanese Patent Laid-Open No. 7-36
Japanese Patent Publication No. 505 proposes a method for identifying an operation model using a neural network. This method is effective when the behavior of the control target is non-linear or when there is a parameter change during the feedback control.

[0007]

However, a lot of trial and error is required until the neural network configuration can be sufficiently determined and learned. If the control target is, for example, a manufacturing or processing process device or equipment, It produces a large amount of trial products and has low productivity before entering high-quality commercial operations. Also, if it is necessary to change or adjust the feedback controller, there is a concern that a considerable number of products will be produced until the production quality converges within the standard after the change and adjustment, and the yield will be significantly reduced. , Changes, adjustments, etc., are difficult to perform appropriately or easily.

[0008] In addition, when a feedback controller is included in the controlled object, it has not been possible to identify mechanical parameters with high productivity.

The present invention relates to a feedback controller,
The first object is to make the adjustment appropriately or easily, and the second object is to reduce the control error of the feedback controller, and to design, set, change or adjust the feedback controller immediately. It is a third object to improve the reliability of the control quality.

[0010]

According to the present invention, there is provided a coefficient determining apparatus for a command operation expression of a feedback controller, wherein a physical quantity is manipulated according to a command signal from the controller, and an observed quantity representing the resulting physical quantity is sent to the controller. The control target (1, 2), which is the output device or equipment,
Test signal input means for giving a pulse signal as a command signal
(4, 6); signal recording means (7) for reading in time series the amount of observation after giving a pulse signal, the pulse signal value, and a target value given from a host computer (host); Means (8) for generating a Markov parameter from the observed quantity of (8); means for generating a Bode diagram from the generated Markov parameters (BDG); calculating a least square evaluation function from the generated Markov parameter, and performing least square evaluation Means (9) for deriving a controller for minimizing a function; and lowering the controller into a lower-order controller of order n for mounting on the controller, Means for deriving an arithmetic expression for converting the observed quantity into a command signal
(11); In addition, in order to facilitate understanding, the symbols of the corresponding elements in the embodiments shown in the drawings and described later are added in the parentheses for reference.

According to this, it is possible to derive an arithmetic expression for the feedback controller for the controller, which is suitable for the closed loop while the controller (3) and the control object (1, 2) form a closed loop. .

After shifting the gain of the feedback controller to a small value (making the gain close to zero), when the target value from the host computer (host) shows the same value as the observed value of the controlled object during operation, the signal application means A pulse is applied to the command signal from the pulse signal generating means (6) through (4). The response of the control target (1, 2) to the command signal to which the pulse is applied is recorded as an observation amount. Because the gain of the feedback controller is small and the control target usually has a delay, the input signal and the output signal can be treated as uncorrelated for a short time (100 ms if the actuator is an electric motor and a drive device). . Then, since the difference between the recorded observation amount and the target value is obtained, the feedback calculation expression of the controller can be determined based on the response-based control method. This approach reduces controller set errors without breaking the closed loop and without the conventional modeling errors affecting the controller. Since the arithmetic expression of the feedback controller is determined based on the amount of observation that appears due to the actual operation of the control target, the reliability of the control quality immediately after design, setting, change, or adjustment is high. By applying the pulse signal to the control target and reading the amount of observation from the supply of the pulse signal until the response of the control target stops, the coefficient of the feedback operation formula of the controller can be determined. The setting or adjustment of the feedback controller is completed only by setting the adjusted coefficient in the controller, so that the controller can be changed or adjusted appropriately or easily.

The controlled object (1, 2) includes a rotating machine (rolling mill) that generates torsional vibration of the shaft. A Bode diagram is created from the generated Markov parameters, and the torsion of the controlled object is controlled. Find the resonance point.

[0014] Other objects and features of the present invention will become apparent from the following description with reference to the drawings.

[0015]

FIG. 1 shows an embodiment of the present invention. An object to be controlled includes an actuator 1 and a plant (equipment) 2 driven by the actuator.
The physical quantity of the plant 2 or an object manufactured, processed, or operated by the plant 2 is detected by a sensor provided to the controller 2 and supplied to the controller 3 as a controller as an observation quantity.
The target value of the amount of observation is given to the controller 3 from the host computer (host). Controller 3
According to the feedback calculation formula set inside,
A command signal for adjusting the amount of observation to the target value is generated and given to the actuator 1.

An adder 4 is interposed between the controller 3 and the actuator 1 in order to carry out the present invention, and one of two inputs of the adder 4 is constantly provided. The command signal generated by the controller 3 is added to the other input. The response of the control target 1 or 2 is input to the other input for the design, setting, change or adjustment of the controller 3 as a feedback controller by the operation of the operator. When exploring,
A pulse signal (step signal or impulse signal) generated by the signal generator 6 is added. The adder 4 gives the added value of the two inputs to the actuator 1. Therefore, only when a signal is generated by the operation of the operator, a signal obtained by adding a pulse signal to the command signal (command value of controller 3 + pulse signal level value) is supplied to actuator 1.

In order to design, set, change or adjust the controller 3 which is a feedback controller,
The operator closes the change-over switch 4 to the contact on the A side when the controller is created and searches for the
, And outputs a step output or an impulse output to the actuator 1 through the switch 4. FIG.
In the embodiment shown in (1), the arithmetic device 5 is a coefficient setting device, that is, an auxiliary device used for setting an arithmetic expression in the controller 3.

FIG. 2 shows an outline of the configuration of the controller 3 which is a feedback controller. Controller 3 is the key,
The basic controller 31 and the sub-controller 32 are serially connected (series-coupled). In the case where the controller 3 is constituted by an operation program, it means that the sub-controller operation is performed (serial) after the basic controller operation. However, the basic controller 3
The order of 1 and the sub-controller 32 may be interchanged.
The basic controller 31 is a PI controller or P
In addition to the ID controller, a P controller, an I controller, a D controller, or a PD controller may be used. P means a proportional operation, I means an integral operation, and D means a differential operation. The sub controller 32 is a P controller.
, A D controller, a phase advance / delay operation controller or a first order delay operation controller.

The basic controller 31 has No. 1 shown in Table 1 below. 1 to No. Any of the six embodiments may be used. N
o. 1 to No. FIG. 2 is a hardware representation of the configuration of the controller in each of the three aspects. ~ 3. And N
o. 4-No. FIG. 4 shows the configuration of the controller in each of the modes 6 in hardware representation in FIG. ~ 6. As shown.

[0020]

[Table 1]

The sub-controller 32 has N as shown in Table 2 below.
o. (A) -No. Any of the embodiments (d) can be used.
5 (a) to 5 (d) show No. (A) -No. (D)
That is, proportional operation (P), differential operation (D), 1
The configuration of the hardware representation of each of the next delay operation and the phase advance / delay operation will be described.

[0022]

[Table 2]

The arithmetic unit 5 has functions "signal generation" 6,
"Signal recording" 7, "Markov parameter generation" 8, "Board diagram generation" BDG, "Design controller to minimize least squares evaluation function" 9, "Singular value reduction by singular value""10" and "approximate to a specific controller" 11 are stored therein, and each function is developed according to an operator input.

"Signal generation" 6 is for providing the actuator 1 with a time-series command signal output pattern (time lapse vs. command signal level) previously input by the operator via the adder 4. In this embodiment, one of two patterns, a step signal output mode and an impulse output mode, is output in accordance with the operator's designation and in accordance with the operator's pulse signal output instruction. . "Signal recording" 7
Is started by the above-mentioned pulse signal output instruction of the operator, and the operator reads the output (observed amount: detected value of the sensor) of the plant 2 according to a preset sampling period and sampling number. Each time the period elapses, the amount of observation is written to the observation data memory at the read time. This pulse signal output and observation data
Data reading is repeated the number of times specified by the operator.

FIG. 6 shows an outline of the processing functions of the arithmetic unit 5, and FIGS. 7 to 9 show details. The operator inputs the sampling period, the number of samples, and the number of times of data sampling. When the amount of observation of the controlled objects 1 and 2 is stable at the target value, the gain of the controller 3 is reduced. Is applied to the input side of the control object via the adder 4.

From the time when the step signal is output from the signal generator 6, the execution of the above-mentioned collection of the observation amount (response data) is started. This step signal output and observation value reading are repeated for the number of times of sampling (step 7 in FIGS. 6 and 7).
1). When reading is performed for the number of times of data collection, the arithmetic unit 5
Calculates the average value of the observed values (for the number of times of collection) having the same elapsed time (time) from the start of the step signal output (step 72 in FIGS. 6 and 7). Then, the differential value of the average value at each time is calculated from the time series distribution of the average value of the observation amount at each time (steps 73 and 81 in FIGS. 6 and 7). These are referred to as Markov parameters H.

When the operator designates the impulse output mode and instructs the execution of response data collection, the arithmetic unit 5 outputs the above-mentioned observation amount from the time when the signal generator 6 outputs the impulse signal. Start execution of (response data) collection. The output of the impulse signal and the reading of the observation value are repeated for the number of times of collection (step 74 in FIGS. 6 and 7). When reading is performed for the number of times of data collection, the arithmetic unit 5 calculates an average value of data at each time (each time of data collection) (step 75 in FIGS. 6 and 7). These average values are used as Markov parameters H (FIGS. 6 and 7).
Step 81).

As described above, when the collection of the Markov parameter H is completed, the operator stops the signal output,
The input to the control target is only the command signal, and the controller 3
Restore the gain of. Thereby, the control targets 1 and 2 and the controller 3 can restart the feedback control before the pulse application.

Next, the arithmetic unit 5 outputs the Markov parameter H
, The Markov parameter from the disturbance w (step signal or impulse) to the observed quantity y (Markov parameter H)
Is calculated (step 8 in FIG. 6, step 8 in FIG. 7).
2).

Next, when it is necessary to confirm the frequency band (region), the operator generates a Bode diagram in the arithmetic unit 5 and determines the required frequency region. When a resonance point to be controlled appears on the Bode diagram, a frequency region is determined in a region outside the resonance point (step BDG in FIGS. 6 and 7). Then, it instructs the arithmetic unit 5 to generate a controller. Then, according to the response control method, the high-dimensional controller C1
(Controller) (steps 9 and 8 in FIG. 6)
And steps 91 and 92 in FIG. 9). The contents of the calculation formula in step 91 shown in FIG.
Steps 91a and 91b of FIG.

Next, the arithmetic unit 5 calculates a singular value of the observable gramian Pk of the high-dimensional controller C1, and displays the calculated singular value on a display (step 10 in FIG. 6, step 101 in FIG. 9). , 102), and prints out when an operator print instruction is issued. The operator arranges the singular values in descending order, evaluates the degree of influence by the sum of the squares, determines the order n for lowering the order of the high-dimensional controller C1, inputs the order n to the arithmetic unit 5, and inputs the lower order n Enter the conversion instruction.

[0032]

[Table 3]

In response, the arithmetic unit 5 reduces the order of the high-dimensional controller C1 to the order n (step 1 in FIG. 6).
0, step 103 in FIG. 9). Control with reduced dimensions
La is called a normative controller. The operator may or here (e.g., before instructing the execution of response data collection) provide a minimum frequency, a maximum frequency, and a score N in a frequency band desired by the operator or specified by another. as well as the input arithmetic expression of the form (eg PID) and the response time constant T _1. And the control in the arithmetic unit 5
When the dimensionality of the model has been reduced, the computer is instructed to determine the arithmetic expression.

In response, the arithmetic unit 5 calculates an approximate frequency (step 11 in FIG. 6 and step 111 in FIG. 10), and calculates the frequency response of the reference controller (steps 11 and 11 in FIG. 6). A coefficient for minimizing the square error of the frequency response is obtained (step 112 in FIG. 10) (step 11 in FIG. 6 and step 113 in FIG. 10).

After confirming the operator, when the operator instructs the mounting, the arithmetic unit 5 calculates the calculated reference control.
La (arithmetic expression) and coefficients are stored in the controller 3 shown in FIG.
Transfer to The controller 3 includes a feedback operation unit (or a feedback operation program) therein.
It is configured (rewritten) to the one specified by the transfer data of the arithmetic unit 5. The data (arithmetic expressions and coefficients) output by the arithmetic unit 5 to the display and / or the printer is operated.
When the data is input to the controller 3, the operator can rewrite the feedback calculation expression of the controller 3.

A method for reducing the dimension of the high-dimensional controller C1 to the order n is a known representative root method (Davison method, for example,
Computrol 1989, No. 23 pp92-96)
Alternatively, this can be omitted and the high-dimensional controller C1 can be used as a reference controller.

FIG. 10 shows the contents of the derivation of the controller C2 in step 11 in FIG. 6 and the case where the PID operation is set in the feedback operation expression (Tables 1, 3 and 4) of the controller 3. The outline was shown. When the controller 3 uses only the PID operation expression (the basic controller 31 has the PID operation and the sub-controller 32 has no operation), the PID operation expression (controller C2) to be set therein is derived (see FIG. FIG. 13 shows the contents of step 11) of FIG. 6 more specifically.

FIG. 14 shows the derivation of a feedback operation expression when the controller 3 is composed of a basic controller 31 for PID operation and a sub-controller 32 for pseudo differential operation. The contents of step 11) will be described.

FIG. 15 shows the derivation of a feedback operation formula when the controller 3 is composed of a basic controller 31 for PID operation and a sub-controller 32 for first-order lag operation (FIG. 6). The contents of step 11) will be described.

FIG. 16 shows the derivation of a feedback operation expression when the controller 3 is composed of a basic controller 31 for PID operation and a sub-controller 32 for phase lead / lag operation (FIG. 6). Shows the contents of step 11).

As described above, the basic controller 31 has a PID
An example in which the calculation is performed is shown. Next, PID
A case other than the operation will be described.

[0042]

[Table 4]

[0043]

[Table 5]

[0044]

In FIG. 17, the control object (1 + 2) is a rolling mill, and the controller 3 sets the roll rotation speed (observed amount) of the rolling mill to the target speed indicated by the host computer (host). An embodiment applied to a rolling system that performs feedback control will be described. In this case, as shown in FIG. 18, the outline of the control object is that two loads (2) of an intermediate inertia and a load inertia (rolling roller) are driven via two shafts as a drive motor which is an actuator. 1 and their specifications are as follows.

[0045]

[Table 6]

While the control target (rolling mill) of the system shown in FIG. 17 is operating (during rolling or idling),
The controller 3 obtains a Markov parameter, derives a feedback operation expression suitable for the dynamic characteristic of the control target, and
The procedure set in is described below. (1) When the rotation speed (observed value) of the drive motor 1 to be controlled is stable at the target speed, the gain of the controller 3 is lowered, and the control device outputs a step signal to the adder 4. As a result, a command value obtained by adding the step signal level value to the command value of the controller 3 was input to the INV board (motor driver) that drives the motor 1, and the rotation speed increased, and then decreased and increased. The rotation speed of the drive motor 1 was read into the arithmetic unit 5 using a pulse generator, and was written into the internal memory of the arithmetic unit. The sampling period Ts is 2 msec, and the number of samples M is 100
The number Nd of step response sampling times is 10, and the arithmetic unit 5 averages the same 10 readings of the same elapsed time (time) from the start of step signal output. The results are shown in FIG. (2) The Markov parameter H was calculated by time-differentiating the step response (average value of the time series distribution) by the arithmetic unit 5. FIG. 20 shows the calculated Markov parameter H. (3) The arithmetic unit 5 calculates the Markov parameter G from the disturbance w to the observation amount y from the Markov parameter H from the observation amount u to the observation amount y. (4) The arithmetic unit 5 creates a Bode diagram based on the Markov parameter G. FIG. 23 shows the obtained Bode diagram.
Is shown by a solid line (closed loop frequency characteristic). Note that FIG.
For reference, the closed-loop frequency characteristic is indicated by a two-dot chain line. (5) In the arithmetic unit 5, controllers KY and KU that minimize the evaluation function J are calculated from the Markov parameters H and G. That is, a controller that minimizes the square evaluation function J is created from the Markov parameter H from the control input u to the observation amount y and the Markov parameter G from the disturbance w to the observation amount y.

[0047]

[Table 7]

(6) Dimension reduction from KY and KU using singular values was performed. The arithmetic unit 5 calculated the singular value of the observable Gramian of the controller and displayed it on the display.
The calculated singular values are shown below.

[0049]

[Table 8]

In this example, it is determined that the first order (n = 1) is sufficient for the operator, and the first order is designated to the arithmetic unit 5 to instruct the lower order. Normative control in which arithmetic unit 5 is reduced in dimension
Lam is the sampling period T specified by the operator.
s: In 0.002Sec, was assumed to be described by the following state _{equation: x k + 1 = 0.4399 ·} x k +0.0265 · U k y = 0.8953 · x k +0.0 · U k i.e., Ac = 0.4399, Bc = 0.0265, Cc = 0.895
3, Dc = 0.0. (7) Code Control - maximum frequency ω _{mi n} = 1 rad / sec as a frequency band of approximation to La Cm, ω _max = 100
rad / sec, N = 100, PI controller [C (s) = K1 · (1 + K2 / s)] as the controller to be approximated
Is input to the arithmetic unit 5 by the operator, and the arithmetic unit 5 calculates the PI controller. The calculation results of the arithmetic unit 5 are K1 = 0.042 and K2 = 0.515. (8) The signal generation from the pulse signal generation means is stopped, and a control closed loop including the control targets 1 and 2 and the controller 3 is formed, and the coefficient K1 of the PI controller 3 is 0.042, K
A step response experiment was performed in which 2 was set to 0.515 and a target speed of 50 rpm was supplied to the controller 3 as a step input (Example). FIG. 21 shows the motor speed and the load speed according to this embodiment. FIG. 22 shows Ziegler- as a comparative example.
K1 = 0.04 designed using Nichols 1/4 design rule
5, the results of a step response experiment (comparative example) in which a PI controller with K2 = 105.8 is used as the controller 3 and a target speed of 50 rpm is input to the controller 3 as a step input.

Example (FIG. 21) and Comparative Example (FIG. 22)
In both cases, the solid line is the drive motor speed and the broken line is the load rotation speed. The rising speed, which indicates the responsiveness of the step response, does not change in both cases, but the overshoot amount is smaller in the embodiment, which is an excellent result.

According to the present invention described above, the setting, change, or adjustment of the feedback arithmetic expression of the controller is performed using the impulse response or the step response that can be easily applied in the field while the controlled object is driven. In this case, it is possible to reduce the number of trials and errors, and to design a controller for any plant including equipment and facilities in which it is difficult to create a physical model that is difficult to design with the conventional method.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a functional outline of an embodiment of the present invention.

FIG. 2 is a block diagram showing an outline of a functional configuration of a controller 3 shown in FIG.

FIG. 3 is a diagram of the basic controller 31 shown in FIG.
FIG. 3 is a block diagram showing a part of a variety of a callback operation function in a hardware configuration.

FIG. 4 is a diagram of the basic controller 31 shown in FIG.
FIG. 10 is a block diagram showing another example of the variety of the callback operation function in a hardware configuration.

5 is a block diagram showing a variety of arithmetic functions of the sub-controller 32 shown in FIG. 2 in a hardware configuration.

FIG. 6 is a flowchart showing an outline of a data processing function for deriving a feedback operation equation of the arithmetic unit 5 shown in FIG. 1;

FIG. 7 is a flowchart showing a part of the arithmetic function shown in FIG. 6;

FIG. 8 is a flowchart showing a part of the arithmetic function shown in FIG. 6;

FIG. 9 is a flowchart showing a part of the arithmetic function shown in FIG. 6;

FIG. 10 is a flowchart showing the contents of the rest of the arithmetic function shown in FIG. 6;

FIG. 11 is a block diagram showing details of a part of the calculation in step 91 shown in FIG. 8;

FIG. 12 is a block diagram showing details of the remainder of the calculation in step 91 shown in FIG. 8;

FIG. 13 is a flowchart showing the contents of step 11 in FIG. 6 when deriving a controller for PID calculation.

FIG. 14 is a flowchart showing the contents of step 11 in FIG. 6 when deriving a controller for PID operation + pseudo differential operation.

FIG. 15 is a flowchart showing the contents of step 11 in FIG. 6 when deriving a controller for PID operation + first order delay operation.

FIG. 16 is a flowchart showing the contents of step 11 in FIG. 6 when deriving a controller for PID calculation + phase lead / lag calculation.

FIG. 17 is a block diagram showing a functional outline of an embodiment of the present invention.

FIG. 18 shows an actuator 1 and a plant 2 shown in FIG.
FIG. 3 is a block diagram showing a mechanism connection relationship.

FIG. 19 shows an INV board (motor driver) shown in FIG.
Is a time chart showing a change in the rotation speed of the actuator (drive motor) 1 when a step signal is given to the motor.

20 is a time chart showing a differential value of the rotation speed shown in FIG.

21 derives a PI operation formula in the embodiment of the present invention shown in FIG. 17 and sets it in the controller 3;
This is a time chart showing changes in the rotation speed of the actuator (drive motor) 1 and the plant load (roll of the rolling mill) when a step signal is given to the motor driver.

22 derives a PI operation expression by a conventional method in the system configuration shown in FIG. 17 and sets it in the controller 3;
This is a time chart showing a change in the rotation speed of the actuator (drive motor) 1 and the plant load (roll of the rolling mill) when a step signal is given to the NV board (motor driver).

FIG. 23 is a Bode diagram showing frequency characteristics of the control target (1, 2) shown in FIG. 17;

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Takashi Hirayama 20-1 Shintomi, Futtsu-shi, Chiba Prefecture Nippon Steel Corporation Technology Development Division (72) Inventor Harutoshi Okai 20-1 Shintomi, Futtsu-shi, Chiba Prefecture F-term in the Technology Development Division of Nippon Steel Corporation (reference) 5H004 GA03 GA04 GA21 GB03 HA08 HB08 JA29 KA32 KB02 KB04 KB06 KB26 KC08 MA38

Claims (6)

[Claims] 1. A test signal for giving a pulse signal as a command signal to a controlled object which is a device or equipment that operates a physical quantity in response to a command signal from a controller and outputs an observed quantity representing the resulting physical quantity to the controller. Input means; signal recording means for reading in a time series the amount of observation after giving a pulse signal; means for generating a Markov parameter from the read time-series observation amount; least-squares from the generated Markov parameter Control that calculates the evaluation function and minimizes the least squares evaluation function
Means for deriving the controller; and the controller is reduced to a lower-order controller of order n for equipping the controller, and the amount of observation of the lower-order controller is converted into a command signal. Means for deriving an arithmetic expression to be performed, the coefficient determining device for a command arithmetic expression of the feedback controller. 2. A low-order controller of the controller, comprising a PID
2. The coefficient determining device for a command operation expression of a feedback controller according to claim 1, further comprising an operation. 3. The test signal input means supplies a step signal to a control object, and the Markov parameter generating means calculates a differential value of the read time-series observation quantity, and uses the Markov parameter from the differential value. Generating the data.
A coefficient determining device for the command operation expression of the feedback controller. 4. The test signal input means supplies an impulse signal to a control target, and the Markov parameter generation means generates a Markov parameter from the read time-series observation quantity. , A coefficient determination device for the command operation expression of the feedback controller. 5. A physical quantity is manipulated according to a comparison deviation between a command signal from the controller and an observed quantity representing a physical quantity of the operation result through a feedback controller, and outputs the observed quantity to the controller. Means for adjusting the gain of a feedback controller of a controlled object which is a device or equipment; means for providing a pulse signal together with a target value to the command signal; the amount of observation after providing the pulse signal, the pulse signal value and a host computer ( A signal recording means for reading a target value given from the host) in time series; a means for generating a Markov parameter from the read time series observation amount; calculating a least square evaluation function from the generated Markov parameter, and minimizing the least square evaluation function Means for deriving a controller to be implemented; and an order n in a predetermined frequency domain for equipping said controller with said controller. And reduced-lower order controller, the low-order controller, means for deriving an arithmetic expression for converting an observation amount to a command signal; comprises a feedback controller command arithmetic expression coefficient determiner. 6. The control object includes a rotating machine that generates axial torsional vibration. A Bode diagram is created from the generated Markov parameters to determine a resonance point of the axial torsional vibration, and a predetermined frequency region is narrowed down. A method for setting the feedback controller according to claim 5.