JP5071057B2 - PID control support device - Google Patents

Description

  The present invention relates to a PID control support device, and particularly suitable for a parameter setting method of a PID adjuster used for temperature control of an injection molding machine, a reflow furnace, a flat panel display manufacturing device, a semiconductor manufacturing device, and the like. It is.

There is a method using PID control in the temperature control of a cylinder of a conventional injection molding machine. In this PID control, a PID control parameter is determined while performing auto-tuning using a method such as a step response method or a limit cycle method, and the PID control parameter is determined by repeatedly performing auto-tuning for system fluctuations. Adaptive control is performed while changing to a system corresponding to system fluctuations.
Further, for example, Non-Patent Document 1 discloses a gradient temperature control method using a plurality of heaters and sensors in order to realize non-interference of a control target in which thermal interference exists.

FIG. 12 is a block diagram showing a schematic configuration of a conventional temperature control system.
In FIG. 12, a temperature regulator 100 includes PID regulators P _{0 to} P _n−1 that perform proportional differential integral control, a pre-compensator Gc that realizes transfer characteristics of the controlled object 101, and measured temperature values of the controlled object 101. ch ₀ ~ch _n-1 an average temperature measurement value k ₀ and gradient temperature measurements k ₁ to k mode converter Gm for converting the _n-1, the mean temperature setpoint s ₀ and gradient temperature setpoint s ₁ ~s _{n -1} subtractor M ₀ ~M _n-1 to calculate an average temperature measurement value k ₀ and gradient temperature measurements k ₁ ~k _n-1 difference between each are provided.

When the temperature measurement values ch _{0 to} ch _{n-1 of} the controlled object 101 are input to the mode converter Gm as the control amount PV, the average temperature measurement value k ₀ and the gradient temperature measurement values k _{1 to} k _n-1 are obtained. Converted and input to the subtracters M _{0 to} M _n−1 , respectively. Further, the average temperature set value s ₀ and the gradient temperature set values s _{1 to} s _n-1 are input to the subtracters M _{0 to} M _n-1 as the target value SV, respectively. Then, in the subtracter M ₀ ~M _n-1, the average temperature set point s ₀ and gradient temperature setpoint s ₁ ~s _n-1, the average temperature measurement value k ₀ and gradient temperature measurements k ₁ to k _n- after being calculated ₁ deviation between each, are input to the PID controller P ₀ to P _n-1, PID controller P ₀ to P output from the _n-1 via the predistorter Gc operation amount It is input to the control object 101 as MV.
OMRON TECHNICS vol. 44 No, 2 2004 pp112-117

However, in the conventional PID control, there is a problem in that control that optimizes the entire system cannot be performed because control in consideration of interference occurring between a plurality of control targets cannot be performed.
Further, in the method disclosed in Non-Patent Document 1, since the transfer characteristic of the control object 101 is identified from the response waveform when the stepped operation amount MV is applied to each heater, the transfer characteristic of the control object 101 is identified. There is a problem that a great amount of man-hours are required to do this.
Moreover, it is necessary to adjust the parameters of the PID regulators P _{0 to} P _n-1 for the average temperature measurement value k ₀ and the gradient temperature measurement values k _{1 to} k _n-1 , and trials are performed to optimize the parameters. It was necessary to repeat the mistakes.
Accordingly, an object of the present invention is to provide a PID control support device capable of realizing control in consideration of interference occurring between a plurality of control objects while reducing the labor required for adjusting parameters of a PID adjuster. It is.

In order to solve the above-described problem, according to the PID control support device according to claim 1, the operation amount time series data input to a plurality of control targets and the control target output according to the operation amount. that control the amount of time on the basis of the sequence data, and the transfer function matrix parameter identification means for identifying the parameter of the transfer function matrix for the current advance the controlled object structure is set tables, identified by the transfer function matrix parameter identification means PID parameter calculation means for calculating the parameters of each PID adjuster that controls the controlled object based on the parameters of the transfer function matrix, and parameters of the transfer function matrix identified by the transfer function matrix parameter identification means Based on the model predictive control parameter for calculating a parameter for controlling the set value of the PID adjuster in model predictive control And the model predictive control parameter calculation unit regards the control object and the PID adjuster as an enlarged control object, and expresses the enlarged control object including interference between the control objects. The transfer function matrix is used as an internal model, the off-diagonal term of the transfer function matrix is an interference term between the plurality of control objects, and the model predictive control parameter calculation means uses the transfer function matrix used as the internal model. of the diagonal terms representing based on the primary delay element and a dead time element, that you expressed based on unpaired square section primary delay element and a dead time element and a differential element of the transfer function matrix used as the internal model Features.

Further, according to the PID control support device according to claim 2, the model predictive control parameter calculation means uses a time constant and a coincidence point of a reference trajectory that are parameters of the model predictive control as a transfer function matrix. It is given as a function of gain, time constant and dead time .

Further, according to the PID control support device of claim 3, the model predictive control parameter calculation means uses a time constant and a coincidence point of a reference trajectory that are parameters of the model predictive control as a transfer function matrix. The time constant is given by a linear function .
According to the PID control support apparatus of the fourth aspect, the structure of the control target is set based on a first-order lag element and a dead time element .

Further, according to the PID control support device according to claim 5, the transfer function matrix parameter identification means uses time series data of the manipulated variable and data during closed loop control as the time series data of the controlled variable. It is characterized by.
Further, according to the PID control assistance device according to claim 6, wherein the transfer function matrix parameter identification means comprises a time-series data of the operation amount, as time-series data of the control amount, and Mochiiruko data at startup It is characterized by.

Further, according to the PID control support apparatus of the seventh aspect, the PID parameter calculation means gives the parameter of the PID adjuster as a function of the parameter to be controlled .
According to the PID control support device of claim 8, the PID parameter calculation means ignores the off-diagonal term of the transfer function matrix whose parameter is identified by the transfer function matrix parameter identification means, and PID controller parameters are calculated .

As described above, according to the present invention, it is possible to represent a control object using a transfer function matrix, and it is possible to incorporate interference between control objects into the transfer function matrix, and to control the parameter of the control object. A parameter of the PID regulator can be given as a function, and it is possible to realize control in consideration of interference occurring between a plurality of controlled objects while reducing the effort for adjusting the parameter of the PID regulator.
Further, by using the transfer function matrix expressing the control object and the PID regulator as an enlarged control object as an internal model of the model predictive control, the set value of the PID regulator is controlled by the model predictive control, Even when the parameters of the PID adjuster are adjusted without considering the interference between them, it is possible to realize the control that takes into account the interference that occurs between a plurality of control objects, and realize the control that optimizes the entire system be able to.

Hereinafter, a PID control support apparatus according to an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing a schematic configuration of a PID control system according to an embodiment of the present invention, and FIG. 2 is a flowchart showing a parameter setting method in the PID control system of FIG.
In FIG. 1, a PID adjuster 12 that performs PID control of the control target 13 is connected to the control target 13, and the set value SV _PID of the PID adjuster 12 is controlled by model predictive control at the preceding stage of the PID adjuster 12. The model prediction control means 11 to be connected is connected.
Examples of the control target 13 include an injection molding machine, a reflow furnace, a flat panel display manufacturing apparatus, and a semiconductor manufacturing apparatus.

Here, the model prediction control means 11 obtains a predicted value of the future control amount PV using the internal model 11a corresponding to the control target 13, and the predicted value approaches the target value SV _MPC (or reference trajectory) in the prediction section. Thus, it is possible to determine the setting value SV _PID of the control input section, and to input only the setting corresponding to the current time among the obtained setting values SV _PID .
Further, the PID control system is provided with model predictive control parameter calculation means 14, PID parameter calculation means 15, and transfer function matrix parameter identification means 16.

  Here, the transfer function matrix parameter identifying means 16 is based on the time series data of the manipulated variable MV input to the controlled object 13 and the time series data of the controlled variable PV output from the controlled object 13 according to the manipulated variable MV. Thus, it is possible to identify the parameters of the transfer function matrix that expresses the control target 13 having the structure set in advance including the interference between the control targets 13. The structure of the control target 13 can be set based on the first-order lag element and the dead time element. Further, the transfer function matrix parameter identification means 16 may use the data during the closed loop control as the time series data of the manipulated variable MV and the time series data of the controlled variable PV, or use the data at the time of activation. May be.

  The PID parameter calculation means 15 can calculate the parameters of the PID adjuster 12 that controls the control target 13 based on the parameters of the transfer function matrix identified by the transfer function matrix parameter identification means 16. The PID parameter calculation means 15 can give the parameters of the PID adjuster 12 as a function of the parameters of the control target 13. Further, the PID parameter calculation means 15 can calculate the parameters of the PID adjuster 12 ignoring the off-diagonal terms of the transfer function matrix whose parameters are identified by the transfer function matrix parameter identification means 16.

Based on the parameters of the transfer function matrix identified by the transfer function matrix parameter identifying means 16, the model predictive control parameter calculating means 14 calculates parameters for controlling the set value SV _PID of the PID regulator 12 by model predictive control. can do. Note that the model predictive control parameter calculation means 14 regards the control target 13 and the PID adjuster 12 as an enlarged control object 18 and expresses the enlarged control object 18 including interference between the control objects. Can be used as the internal model 11a. The model predictive control parameter calculation means 14 expresses the diagonal terms of the transfer function matrix used as the internal model 11a based on the first order lag element and the dead time element, and the off-diagonal terms of the transfer function matrix used as the internal model 11a. Can be expressed based on a first-order lag element, a dead time element, and a differential element. Further, the model predictive control parameter calculating means 14 gives the time constant and coincidence point of the reference trajectory, which are parameters of the model predictive control, as a function of the gain, time constant and dead time of the transfer function matrix used as the internal model 11a. For example, it can be given as a linear function with a time constant of a transfer function matrix used as the internal model 11a.

The model prediction control means 11 can be realized on the panel computer 1, and the model prediction control parameter calculation means 14, the PID parameter calculation means 15, and the transfer function matrix parameter identification means 16 can be realized on the personal computer 2. it can.
When the manipulated variable MV output from the PID adjuster 12 and the controlled variable PV output from the controlled object 13 are input to the transfer function matrix parameter identifying unit 16, the transfer function matrix parameter identifying unit 16 selects the manipulated variable MV. Based on the time series data and the time series data of the control amount PV, the parameters of the transfer function matrix expressing the control target 13 are identified and output to the PID parameter calculation means 15 (step S11).

  When the PID parameter calculation means 15 receives the parameter of the transfer function matrix representing the control target 13 from the transfer function matrix parameter identification means 16, the PID adjustment for controlling the control target 13 based on the parameter of the transfer function matrix. The parameter of the device 12 is calculated, set in the PID adjuster 12, and output to the model predictive control parameter calculating means 14 (step S12).

When the model predictive control parameter calculating unit 14 receives the parameters of the PID adjuster 12 from the PID parameter calculating unit 15, the model predictive control parameter calculating unit 14 sets the internal model 11a based on the parameters of the PID adjuster 12, and sets the parameters of the model predictive control. Calculate and set in the model prediction control means 11 (step S13).
When the model predictive control parameter calculation means 14 sets the internal model 11a and parameters of the model predictive control means 11, the model predictive control means 11 predicts the future control amount PV ^* using the internal model 11a. , the predicted value is determined set value SV _PID control input area so as to approach the target value SV _MPC in the prediction interval, only the setting value SV _PID corresponding to the current time of the obtained set value SV _PID subtractor 17 to output.

FIG. 3 is a diagram conceptually showing the operation of the model prediction control means of FIG.
In FIG. 3, the case where the target value SV _MPC is changed at the current time k is considered. In this case, the future set value SV ^* _PID is obtained at the coincidence point Pset so that the difference between the reference trajectory RT set in advance and the future control amount PV ^* obtained from the internal model 11a is minimized. At this time, it is assumed that the future set value SV ^* _PID operates only in a section called a control horizon Hu.

Next, among the calculated future set values SV ^* _PID , the set value SV ^* _PID (t + 1) at time t + 1 is set as the current set value SV _PID and is output to the subtractor 17. By sequentially repeating this operation, it is possible to always perform optimal control while predicting the future.
Then, in the subtractor 17, a deviation between the set value SV _PID of the PID adjuster 12 and the control amount PV is calculated and output to the PID adjuster 12. Then, the PID adjuster 12 calculates the operation amount MV so that the deviation between the set value SV _PID of the PID adjuster 12 and the control amount PV becomes zero, and outputs it to the control target 13.

  As a result, the control target 13 can be expressed by a transfer function matrix, interference between the control targets 13 can be taken into the transfer function matrix, and the PID adjuster 12 can be used as a function of the parameters of the control target 13. Thus, it is possible to realize control in consideration of interference occurring between the plurality of control objects 13 while reducing the labor for adjusting the parameters of the PID adjuster 12.

The set value SV _PID of the PID adjuster 12 is used in the model predictive control while the transfer function matrix expressing the control target 13 and the PID adjuster 12 as the enlarged control target 18 is used as the internal model 11a of the model predictive control. By controlling, even when the parameters of the PID adjuster 12 are adjusted without considering the interference between the controlled objects 13, it is possible to realize the control considering the interference occurring between the plurality of controlled objects 13. Control that optimizes the entire system can be realized.

FIG. 4 is a block diagram showing an example of a schematic configuration when an injection molding machine is used as a control target of the PID control system of FIG.
In FIG. 4, the injection molding machine 20 is provided with a cylinder 21 for injecting resin, and the cylinder 21 is divided into _four zones Z ₁ , Z ₂ , Z ₃ , and Z ₄ . Then, the cylinder 21, the resin hopper supplied into the cylinder 21 22, the motor 24 for rotating the screw 23 and the screw 23 for stirring the resin in the cylinder 21 is provided, also, the zone Z ₁ cylinder 21 , Z ₂ , Z ₃ and Z ₄ are provided with heaters H _{1 to} H ₄ , respectively, and temperature measuring means J 1 to measure the temperature of the cylinder 21 for each of the zones Z ₁ , Z ₂ , Z ₃ and Z _4. J4 is provided. For example, thermocouples can be used as the temperature measuring means J1 to J4.

The injection molding machine 20 is connected with PID regulators P1 to P4 that perform temperature control of the zones Z ₁ , Z ₂ , Z ₃ , and Z ₄ of the cylinder 21 by PID (proportional integral derivative) control, respectively. Yes.
In addition, as a method of performing the temperature control of the zones Z ₁ , Z ₂ , Z ₃ , and Z ₄ of the cylinder 21, other control methods such as proportional control, integral control, proportional integral control are used in addition to PID control. It may be.
Then, the cylinder 21, at each heater H1~H4 heated per zone _{Z 1, Z 2, Z 3} , Z _4, each zone _{Z 1, Z 2, Z 3} , the temperature of Z ₄ is a temperature measuring means J1 ~ Measured at J4. The temperatures measured by the temperature measuring means J1 to J4 are sent to the PID adjusters P1 to P4 as control amounts PV1, PV2, PV3, and PV4, respectively.

When the PID adjusters P1 to P4 receive the control amounts PV1, PV2, PV3, and PV4 from the temperature measuring means J1 to J4, respectively, the control amounts PV1, PV2, PV3, and PV4 are set values SV1, SV2, SV3, and SV4. The operation amounts MV1, MV2, MV3, and MV4 are calculated so as to coincide with each other. Then, the PID adjusters P1 to P4 calculate the operation amounts MV1, MV2, MV3, and MV4 so that the control amounts PV1, PV2, PV3, and PV4 respectively match the set values SV1, SV2, SV3, and SV4. By controlling the outputs of the heaters H1 to H4 based on the amounts MV1, MV2, MV3, and MV4, the heaters H1 to H4 are controlled to be heated for each of the zones Z ₁ , Z ₂ , Z ₃ , and Z ₄ . Note that PID control by the PID adjusters P1 to P4 can be performed independently for each of the zones Z ₁ , Z ₂ , Z ₃ , and Z ₄ .

1 receives the control amounts PV1, PV2, PV3, and PV4 from the heaters H1 to H4, respectively, the model predictive control means 11 of FIG. While predicting PV2, PV3, and PV4, the optimal control problem is solved at a certain time, and the set values SV1 and SV2 at that time are set so that the future control amounts PV1, PV2, PV3, and PV4 are as close as possible to the target value SV _MPC. , SV3, SV4 can be determined and output to the subtractor 17.

For example, the model prediction control unit 11 can calculate the set values SV1, SV2, SV3, and SV4 so that the output in the prediction horizon matches the reference trajectory.
Here, the plant output at the current time k is represented by y (k), and the set value trajectory at an arbitrary time t that the plant output y (k) should follow is represented by s (t). If the error between the plant output y (k + i) after i steps and the set value trajectory is s (t + i) is ε (k + i), the reference trajectory r (k + i | k) can be expressed by the following equation.
r (k + i | k) = s (k + i) −ε (k + i)
= S (k + i) -e- ^{i [} epsilon] ^{Ts / Tref} (k)
Ts is a sampling period, and Tref is a time constant of the reference trajectory. The notation r (k + i | k) means that the reference trajectory depends on the condition at time k.

When the section to predict the plant output y (k) is called a prediction horizon H _P, the input trajectory u between the prediction horizon _{H P (k + i | k} ) (i = 0,1, .., H P -1 ) is assumed to be applied, the plant output y (k) is the prediction horizon H reference trajectory between _P r (k + i | can select k) | k) to as close as possible to the input trajectory u (k + i . When the future input trajectory u (k + i | k) is selected, only the first element of the trajectory can be applied to the plant as an input signal.

Here, when the control target 13 in FIG. 1 is the injection molding machine 20 in FIG. 3, the transfer function matrix parameter identification unit 16 in FIG. 1 operates the operation amounts MV1, MV2, MV3, and MV4 respectively input to the heaters H1 to H4. Based on the time series data and the time series data of the control amounts PV1, PV2, PV3, and PV4 output from the temperature measuring means J1 to J4 according to the operation amounts MV1, MV2, MV3, and MV4, respectively. The parameters of the transfer function matrix that express the set injection molding machine 20 including the interference between the zones Z ₁ , Z ₂ , Z ₃ , Z ₄ are identified.
Here, assuming that the transfer function for each zone Z ₁ , Z ₂ , Z ₃ , Z ₄ of the injection molding machine 20 is expressed based on the primary delay element and the dead time element, the transfer function of the injection molding machine 20 is assumed. The matrix can be expressed by the following equation (1).

However, G ₁₁ ~G ₄₄ gain, T ₁₁ through T ₄₄ is a time constant, L ₁₁ ~L ₄₄ a dead time. The diagonal term of the transfer function matrix in equation (1) indicates a non-interference term between the zones Z ₁ , Z ₂ , Z ₃ , and Z _4, and the non-diagonal term in the transfer function matrix of equation (1) is each zone _{Z 1, Z 2, Z 3} , showing the interference term between Z _4. For example, the transfer function G ₁₁ e ^−sL11 / (1 + sT ₁₁ ) in the first row and the first column of the transfer function matrix of the equation (1) ^indicates the temperature rise in the zone Z ₁ when the heater H1 in the zone Z ₁ is driven. The transfer function G ₂₁ e ^−sL21 / (1 + sT ₂₁ ) in the second row and first column of the transfer function matrix of equation (1) indicates the temperature rise in the zone Z ₂ when the heater H1 in the zone Z ₁ is driven. The transfer function G ₃₁ e ^−sL31 / (1 + sT ₃₁ ) in the third row and first column of the transfer function matrix of the equation (1) indicates the temperature rise in the zone Z ₃ when the heater H1 in the zone Z ₁ is driven. The transfer function G ₄₁ e ^−sL41 / (1 + sT ₄₁ ) in the 4th row and the 1st column of the transfer function matrix of the equation (1) indicates the temperature rise of the zone Z ₄ when the heater H1 of the zone Z ₁ is driven. Show.

The gains G _{11 to} G ₄₄ , the time constants T _{11 to} T ₄₄ , and the dead times L _{11 to} L _{44 that} are transfer function parameters of the transfer function matrix of the expression (1) are expressed by the following expressions (2) to (5 ) can be obtained by minimizing the objective function J ₁ through J ₄ of formula.
minJ ₁ = ∫ (PV1 ^* −PV1) ² dt (2)
minJ ₂ = ∫ (PV2 ^* −PV2) ² dt (3)
minJ ₃ = ∫ (PV3 ^* −PV3) ² dt (4)
minJ ₄ = ∫ (PV4 ^* −PV4) ² dt (5)
However, PV1 ^{* to} PV4 ^* are estimated control amounts.

Equations (2) to (5) are nonlinear optimization problems. For example, parameters that minimize the objective functions J _{1 to} J ₄ can be obtained by using the Newton method. The time series data of the manipulated variables MV1, MV2, MV3, and MV4 and the time series data of the controlled variables PV1, PV2, PV3, and PV4 may be the data during the closed loop control, or at the time of startup. Data may be used.

5 to 8 are block diagrams showing configuration examples of the transfer function matrix parameter identification means of FIG.
In FIG. 5, a transfer function of G ^* ₁₁ exp (−sL ^* ₁₁ ) / (1 + sT ^* ₁₁ ) is set in the block B11, and G ^* ₁₂ exp (−sL ^* ₁₂ ) / (1 + sT) is set in the block B12. ^* ₁₂₎ is the transfer function is set as the block B13 _{^{is, G * 13 exp (-sL *}} 13) / (1 + sT * 13) that the transfer function is set, the block B14 is, G ^* ₁₄ exp (-sL It is assumed that a transfer function of ^* ₁₄ ) / (1 + sT ^* ₁₄ ) is set.

Then, from the manipulated variables MV1, MV2, MV3, MV4 and the controlled variable PV1, gains G _{11 to} G ₁₄ , time constants T _{11 to} T ₁₄ , and dead time, which are parameters of the transfer function of the transfer function matrix of equation (1). When identifying L _{11 to} L ₁₄ , the time series data of the manipulated variables MV1, MV2, MV3, and MV4 are input to the blocks B11 to B14, respectively, and the outputs from the blocks B11 to B14 are added by the adder Q1. Thus, the estimated control amount PV1 ^* is calculated.
Then, in the subtracter R1, calculates the deviation between the estimated control quantity PV1 ^* and actually measured control quantity PV1, the gain G ₁₁ ~G ₁₄ by minimizing the objective function J ₁ (2) below, The time constants T _{11 to} T ₁₄ and dead time L _{11 to} L ₁₄ can be obtained.

In FIG. 6, a transfer function G ^* ₂₁ exp (−sL ^* ₂₁ ) / (1 + sT ^* ₂₁ ) is set in the block B21, and G ^* ₂₂ exp (−sL ^* ₂₂ ) / A transfer function of (1 + sT ^* ₂₂ ) is set, a transfer function of G ^* ₂₃ exp (−sL ^* ₂₃ ) / (1 + sT ^* ₂₃ ) is set in the block B23, and G ^* ₂₄ exp ( It is assumed that a transfer function of -sL ^* ₂₄ ) / (1 + sT ^* ₂₄ ) is set.

Then, from the manipulated variables MV1, MV2, MV3, MV4 and the controlled variable PV2, gains G _{21 to} G ₂₄ , time constants T _{21 to} T ₂₄ which are parameters of the transfer function of the transfer function matrix of equation (2), dead time If the identification of L ₂₁ ~L _24, the operation amount MV1, MV2, MV3, respectively input series data in the block B21~B24 when MV4, adding the outputs from each block B21~B24 by an adder Q2 Thus, the estimated control amount PV2 ^* is calculated.
Then, in the subtracter R2, calculates the deviation between the estimated control quantity PV2 ^* and actually measured control quantity PV2, the gain G ₂₁ ~G ₂₄ by minimizing the objective function J ₂ (3) below, The time constants T _{21 to} T ₂₄ and dead time L _{21 to} L ₂₄ can be obtained.

In FIG. 7, a transfer function of G ^* ₃₁ exp (−sL ^* ₃₁ ) / (1 + sT ^* ₃₁ ) is set in the block B31, and G ^* ₃₂ exp (−sL ^* ₃₂ ) / A transfer function of (1 + sT ^* ₃₂ ) is set, a transfer function of G ^* ₃₃ exp (-sL ^* ₃₃ ) / (1 + sT ^* ₃₃ ) is set in the block B33, and G ^* ₃₄ exp ( It is assumed that a transfer function of -sL ^* ₃₄ ) / (1 + sT ^* ₃₄ ) is set.

Then, from the manipulated variables MV1, MV2, MV3, MV4 and the controlled variable PV3, gains G _{31 to} G ₃₄ , time constants T _{31 to} T ₃₄ which are parameters of the transfer function of the transfer function matrix of equation (1), dead time When identifying L _{31 to} L ₃₄ , the time series data of the manipulated variables MV1, MV2, MV3, and MV4 are input to the blocks B31 to B34, respectively, and the outputs from the blocks B31 to B34 are added by the adder Q3. Thus, the estimated control amount PV3 ^* is calculated.
Then, in the subtracter R3, calculates the deviation between the estimated control quantity PV3 ^* and actually measured control quantity PV3, the gain G ₃₁ ~G ₃₄ by minimizing the objective function J ₃ of (4), The time constants T _{31 to} T ₃₄ and dead time L _{31 to} L ₃₄ can be obtained.

Further, in FIG. 8, in block B41, the transfer function of ^{_{G * 41 exp (-sL * 41}} ) / (1 + sT * 41) is set, the block B42 ^{_{is, G * 42 exp (-sL *}} 42) / A transfer function of (1 + sT ^* ₄₂ ) is set, a transfer function of G ^* ₄₃ exp (−sL ^* ₄₃ ) / (1 + sT ^* ₄₃ ) is set in the block B43, and G ^* ₄₄ exp ( It is assumed that a transfer function of -sL ^* ₄₄ ) / (1 + sT ^* ₄₄ ) is set.

Then, the operation amount MV1, MV2, MV3, from MV4 and the controlled variable PV4 Prefecture, (1) a transfer gain G ₄₁ ~G ₄₄ is a parameter of the transfer function of the function matrix, the time constant T ₄₁ through T _44, dead time If the identification of L ₄₁ ~L _44, the operation amount MV1, MV2, MV3, respectively input series data in the block B41~B44 when MV4, adding the outputs from each block B41~B44 by an adder Q4 Thus, the estimated control amount PV4 ^* is calculated.

Then, in the subtracter R4, calculates the deviation between the estimated control quantity PV4 ^* and actually measured control quantity PV4, the gain G ₄₁ ~G ₄₄ by minimizing the objective function J ₄ of equation (5), The time constants T _{41 to} T ₄₄ and dead time L _{41 to} L ₄₄ can be obtained.
Then, gains G _{11 to} G ₄₄ , time constants T _{11 to} T ₄₄ , and dead times L _{11 to} L _{44 that} are parameters of the transfer function of the transfer function matrix of equation (1) are the transfer function matrix parameter identification means 16 of FIG. Is calculated, the PID parameter calculation means 15 calculates the parameters of the PID adjusters P1 to P4 from the diagonal terms of the transfer function matrix of equation (1).

Here, the transfer function C (s) of the PID adjusters P1 to P4 can be expressed by the following equation (6).
C (s) = K _P (1 + 1 / (T _I s) + T _D s) (6)
Here, K _P is a proportional constant, T _I is an integral constant, and T _D is a differential constant.
Then, the proportional constant K _P , the integral constant T _I , and the differential constant T _{D that} are parameters of the PID adjusters _{P1 to} P4 can be given as functions of transfer function parameters of the transfer function matrix of the equation (1).
For example, when the adjustment rules of chien, Hrones and Rewick are used, the parameters of the PID controller P1 can be given as shown in Table 1.

  Further, when the adjustment rule of Ziegler and Nichols is used, the parameters of the PID adjuster P1 can be given as shown in Table 2.

The parameters of the PID adjusters P2 to P4 can be set in the same manner using the adjustment rules of chien, Hrones and Rewick, or the adjustment rules of Ziegler and Nichols.
Next, when the parameters of the PID adjusters P1 to P4 are set by the PID parameter calculating means 15, the model predictive control parameter calculating means 14 sets the internal model 11a based on the parameters of the PID adjuster 12, and Parameters for model predictive control are calculated and set in the model predictive control means 11.

Here, when the transfer function of the control target 13 in FIG. 1 is P (s) and the transfer function of the PID regulator 12 is C (s), the closed loop from the set value SV _PID of the PID regulator 12 to the control amount PV is shown. The transfer function can be expressed by the following equation (7).
PV / SV _PID = C (s) P (s) / (1 + C (s) P (s)) (7)
When the equation (7) is approximated by a first order lag system of the time constant T _exd , the following equation (8) is established.
C (s) P (s) / (1 + C (s) P (s)) = 1 / (1 + sT _exd ) (8)
However, it is assumed that the PID controller 12 has an integral element, and the gain of the equation (8) is set to 1.

Further, when the dead time is ignored, the transfer function P (s) of the control target 13 is G ₁₁ / (1 + sT ₁₁ ), and the formula (8) is arranged, the following formula (9) is obtained.
C (s) = − T ₁₁ / (G ₁₁ T _exd ) (1 + 1 / (sT ₁₁ )) (9)
Therefore, when the parameters of the PID regulator 12 are set as shown in the following equations (10) and (11), the transfer function matrix when the control object 13 and the PID regulator 12 are regarded as the enlarged control object 18 is shown. Can be approximated as 1 / (1 + sT _exd ).
K _P = −T ₁₁ / (G ₁₁ T _exd ) (10)
T _I = T ₁₁ (11)

On the other hand, the non-diagonal terms of the transfer function matrix when the control object 13 and the PID regulator 12 are regarded as the enlarged control object 18 are considered in consideration of the influence of the PID regulator 12 and sG _exnd e ^-sLexnd / (1 + sT _exnd ). Where G _exnd is the gain of the off-diagonal term of the transfer function matrix of the controlled object 18, T _exnd is the time constant of the off-diagonal term of the transfer function matrix of the enlarged control object 18, and L _exnd is enlarged. It is the dead time of the off-diagonal term of the transfer function matrix of the controlled object 18.
The diagonal term of the transfer function matrix of the expanded control object 18 is 1 / (1 + sT _exd ), and the non-diagonal term of the transfer function matrix of the expanded control object 18 is sG _exnd e ^−sLexnd / (1 + sT _exnd ), The transfer function matrix when the injection molding machine 20 and the PID adjusters P1 to P4 in FIG. 4 are regarded as the enlarged control object 18 can be expressed by the following equation (12).

Where G _{exnd12 to} G _exnd43 are gains, T _{exd11 to} T _exd44 and T _{exnd12 to} T _exnd43 are time constants, and L _{exnd12 to} L _{exnd43 are dead} times.
FIG. 9 is a diagram for explaining a configuration method of the internal model of the model prediction control means of FIG.
In FIG. 9, assuming that a step input is given as the set value SV1, the PID adjuster P1 outputs the operation amount MV1 so that the control amount PV1 of the zone Z1 of the injection molding machine 20 approaches the set value SV1. The diagonal term of the transfer function matrix when the injection molding machine 20 and the PID controllers P1 to P4 are regarded as the expanded control object 18 is expressed by a first-order lag element 1 / (1 + sT), where T is a time constant. can do.

On the other hand, if a step input is given as the set value SV1, the temperature rise in the zone Z1 is transmitted to the zone Z2, and the temperature in the zone Z2 tends to rise even when the set value SV2 is constant. Here, since the PID controller P2 outputs the operation amount MV2 so that the control amount PV2 of the zone Z2 of the injection molding machine 20 approaches the set value SV2, if the temperature of the zone Z2 increases, the temperature of the zone Z2 It is controlled by the PID regulator P2 so that the rise is suppressed, and the temperature of the zone Z2 tries to return to the original value. For this reason, the non-diagonal term of the transfer function matrix when the injection molding machine 20 and the PID regulators P1 to P4 are regarded as the enlarged control object 18 is the first-order lag differential element s / ( 1 + sT).
Note that the parameters of the transfer function matrix of the controlled object 18 that has been expanded can be determined from the parameters of the transfer function matrix of the control object 13 that are identified by the transfer function matrix parameter identifying means 16.

FIG. 10 is a diagram showing a parameter setting method for the internal model of the model prediction control means of FIG.
In FIG. 10, when the step-like set value SV _PID is input to the enlarged control object 18 of FIG. 1, the time taken until the control amount PV becomes 63% can be set as the time constant T _exd . can be a T _exd = T _11/20.
Then, the model predictive control parameter calculating means 14 indicates the time constant Tref and the coincidence point Pset of the reference trajectory RT, which are parameters of the model predictive control, as shown in the following formulas (13) and (14), the internal model 11a. As a function of the gain, time constant, and dead time of the transfer function matrix used as
Tref = f ₁ (T _exd , G _exd , L _exd , T _exnd , G _exnd , L _exnd ), (13)
Pset = f ₂ (T _exd , G _exd , L _exd , T _exnd , G _exnd , L _exnd ) (14)

Further, the time constant Tref of the reference trajectory RT and the coincidence point Pset, which are parameters of the model predictive control, are expressed by the time constant of the transfer function matrix used as the internal model 11a as shown in the following equations (15) and (16). It can be given as a linear function.
Tref = α ₁ T _exd + β ₁ T _exnd (15)
Pset = α ₂ T _exd + β ₂ T _exnd (16)
However, α ₁ , α ₂ , β ₁ , β ₂ are constants. Further, T _exnd can be determined as an appropriate value by experimentation or the like.
Alternatively, β ₁ = β ₂ = 0 may be set, and the time constant Tref and the coincidence point Pset of the reference trajectory RT may be obtained by the following equations (17) and (18).
Tref = α ₁ T _exd (17)
Pset = α ₂ T _exd (18)

When the set amounts SV1, SV2, SV3, SV4, operation amounts MV1, MV2, MV3, MV4 and control amounts PV1, PV2, PV3, PV4 at the time of starting the injection molding machine 20 in FIG. 4 are set as shown in FIG. The transfer function matrix parameter identification unit 16, the PID parameter calculation unit 15, and the model predictive control parameter calculation unit 14 were calculated.
As a result, the value of Table 3 is obtained as the calculation result of the transfer function matrix parameter identification means 16, the value of Table 4 is obtained as the calculation result of the PID parameter calculation means 15, and the calculation result of the model predictive control parameter calculation means 14 is obtained. The values in Table 5 were obtained.
However, the values in Table 4 are assumed to be no disturbance of the adjustment rules of chien, Hornes and Resick in Table 1 and no overshoot. The values in Table 5 were set to α ₁ = 0.3, α ₂ = 0.4, β ₁ = 0.0, β ₂ = 0.0 in the equations (15) and (16).

It is a block diagram which shows schematic structure of the PID control system which concerns on one Embodiment of this invention. It is a flowchart which shows the parameter setting method in the PID control system of FIG. It is a figure which shows notionally the operation | movement of the model prediction control means of FIG. It is a block diagram which shows an example of schematic structure at the time of using an injection molding machine as a control object of the PID control system of FIG. It is a block diagram which shows the structural example of the transfer function matrix parameter identification means of FIG. It is a block diagram which shows the structural example of the transfer function matrix parameter identification means of FIG. It is a block diagram which shows the structural example of the transfer function matrix parameter identification means of FIG. It is a block diagram which shows the structural example of the transfer function matrix parameter identification means of FIG. It is a figure explaining the structure method of the internal model of the model prediction control means of FIG. It is a figure which shows the parameter setting method of the internal model of the model prediction control means of FIG. It is a figure which shows the waveform of the setting amount at the time of starting of the injection molding machine of FIG. 4, the operation amount, and the control amount. It is a block diagram which shows schematic structure of the conventional temperature control system.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Panel computer 2 Personal computer 11 Model prediction control means 11a Internal model 12, P1-P4 PID regulator 13 Control object 14 Model prediction control parameter calculation means 15 PID parameter calculation means 16 Transfer function matrix parameter identification means 17 Subtractor 18 Enlarged controlled object 20 injection molding machine 21 cylinder 22 Ho Tsu Pas <br/> 23 screw 24 motor H1~H4 heater J1~J4 temperature measuring means

Claims (8)

Based on time-series data of operation amounts input to a plurality of control targets and time-series data of control amounts output from the control targets according to the operation amounts, the control targets whose structures are set in advance are expressed. Transfer function matrix parameter identifying means for identifying parameters of the transfer function matrix to be performed;
PID parameter calculation means for calculating parameters of each PID adjuster for controlling the control object based on the parameters of the transfer function matrix identified by the transfer function matrix parameter identification means;
Model predictive control parameter calculating means for calculating a parameter for controlling the set value of the PID adjuster in model predictive control based on the parameter of the transfer function matrix identified by the transfer function matrix parameter identifying means. ,
The model predictive control parameter calculation means regards the control object and the PID adjuster as an enlarged control object, and represents a transfer function matrix expressing the enlarged control object including interference between the control objects as an internal model. And the off-diagonal term of the transfer function matrix is an interference term between the plurality of controlled objects,
The model predictive control parameter calculation means expresses a diagonal term of a transfer function matrix used as the internal model based on a first-order lag element and a time delay element, and a non-diagonal term of the transfer function matrix used as the internal model A PID control support device, characterized by being expressed based on a delay element, a dead time element, and a differential element. The model predictive control parameter calculation means provides a time constant and a coincidence point of a reference trajectory that are parameters of the model predictive control as a function of a gain, a time constant, and a dead time of a transfer function matrix used as the internal model. The PID control support device according to claim 1 . The model predictive control parameter calculation means provides a time constant and a coincidence point of a reference trajectory that are parameters of the model predictive control as a linear function of a time constant of a transfer function matrix used as the internal model. The PID control support apparatus according to 1 or 2 . The PID control support device according to any one of claims 1 to 3 , wherein the structure to be controlled is set based on a first-order lag element and a dead time element. The transfer function matrix parameter identification means comprises a time-series data of the operation amount, as time-series data of the control amount, to any one of claims 1, which comprises using the data in closed loop control 4 PID control support device. The transfer function matrix parameter identification means comprises a time-series data of the operation amount, as time-series data of the control quantity, according to any one of claims 1 to 4, which comprises using the data at the time of starting PID control support device. The PID parameter calculation means, PID control assisting apparatus according to any one of 6 claim 1, wherein the providing a parameter of the PID controller as a function of the control target parameter. The PID parameter calculating unit calculates a parameter of the PID regulator ignoring a non-diagonal term of a transfer function matrix whose parameter is identified by the transfer function matrix parameter identifying unit. The PID control support apparatus according to any one of 1 to 7 .

Images