JPH0738124B2 - Feedback / feedforward controller - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION Object of the Invention (Field of Industrial Application) The present invention is suitable for application to various process instrumentation systems, and is suitable for a feedback (hereinafter, referred to as FB) control system and a feedforward (hereinafter, referred to as FB). , FF) control system in combination with the feedback feedforward control device, and more particularly, to a feedback feedforward control device that compensates the dynamic characteristics according to the magnitude of the static characteristic compensation component signal.

(Prior Art) This type of FB / FF control device is composed of an FB control system and an FF control system, as shown in FIG. subtracting the current process variable value PV _n of a controlled object 2 from the value SV _n, the resulting deviation e _n is introduced into position type PI controller calculating means 3, the adjustment signal by performing position type PI controller operations where After it is obtained, it is applied to the controlled object 2 via the adding means 4.

On the other hand, the latter FF control system is provided with a feedforward control model 6, and this feedforward control model 6 multiplies the disturbance signal D _n by the FF gain k to obtain the FF control signal and the FF control signal. A lead / lag calculating means 6b for obtaining a disturbance compensating signal in which the time timing of disturbance compensation is adjusted by advancing or delaying the disturbance compensating signal obtained by the lead / lag calculating means 6b. In addition to the means 4, a so-called disturbance-compensated operation signal MV _n is obtained, and the operation signal MV _n is given to the controlled object 2 to perform control.

By the way, as a transfer function for disturbance compensation in the control system, for example, a transfer function G _P (s) between process variables when an operation signal MV _n is applied is expressed as a denominator sequence, G _P (s) = K _P / (1 + T _P1・ s + T _P2・ s ² ＋ …… ＋
T _Pn · s ⁿ ) (1)

Further, when the transfer function G _D (s) between process variables when a disturbance is added is also expressed by a denominator series, G _D (s) = K _D / (1 + T _D1 · s + T _D2 · s ² + …… +
T _Dn · s ⁿ ) (2) In the above equation, K _P is a control parameter and K _D is D
A control parameter, T _Pi is a process time constant, T _Di is a disturbance time constant, and s ⁱ is a Laplace operator.

Therefore, from the transfer functions G _P (s) and G _D (s), the feedforward control model G _F (s) is G _F (s) = {G _D (S) / G _P (S)} = ( K _D / K _P ) ・ {(1 + T _P1・ s + T _P2・ s ² ＋ ……
+ T _Pn・ s ⁿ ) / (1 + T _D1・ s + T _2D・ s ² ＋ …… + T
_Dn · s ⁿ )} (3) = k · f (S) …… (4) k = K _D / K _P …… (5) f (S) = (1 + T _P1 · s + T _P2 · s ₂ + …… ＝ T _Pn・ s
ⁿ ) / (1 + T _D1・ s + T _D2・ s ² ＋ …… + T _Dn・ s ⁿ )
… (6) Therefore, in this control device, the influence of disturbance is suppressed by adding the output of the feedforward control model 6 as a disturbance compensation signal to the PI adjustment calculation output of the position type PI adjustment calculation means 3.

(Problems to be Solved by the Invention) However, when the above control device is applied to an actual process control system, the operating characteristics and the characteristics of disturbance of the controlled object 2 cannot be accurately approximated, and the nonlinearity, and Since the characteristics change from moment to moment, the following points are a problem.

Since the configuration is such that the FF control system output is additively coupled to the FB control system at all times, for example, when the change in the control output becomes small in both the FB control and the FF control, the feedforward control model 6 has static and dynamic errors. If the FF control output is added as a disturbance compensation signal because it is generated or is affected by the nozzle or the like, there is a problem that the output of the FB control system is disturbed and the control accuracy is deteriorated.

In addition, when the calculation output of the position PI adjustment calculation means 3 and the disturbance compensation signal are added and combined as a position signal, the method is a method of directly calculating the entire operation signal for each sampling cycle, and therefore the control system Since it is difficult to provide a so-called dead zone because it is necessary to perform adjustment calculation and use it as an operation signal even in the area where the output is small, when the change in the disturbance compensation signal becomes small, the output of the FF control system is controlled by the FB control. There is a problem that adversely affects the output of the system.

, FF control system, the dynamic characteristic compensation system, even if it is a dynamic characteristic compensation component signal with a slight step change, is generally added to the FB control system as a dynamic characteristic compensation signal in an expanded state. There is a problem that accurate control cannot be performed if the control system output is used incorrectly.

The present invention has been made in view of the above situation, only the FB control by stopping the FF control system depending on the magnitude of the static characteristic compensation component signal, or FB control and FF control by utilizing the FF control system. An object of the present invention is to provide a feedback / feedforward control device which, when combined, realizes improvement of control accuracy and stable control while maximizing the characteristics of FB control and FF control.

[Structure of the Invention] (Means for Solving the Problems) In order to solve the above problems, the present invention provides at least a velocity type I (integral) in order to make the deviation between a process variable value and a target value from a controlled object zero. ) A feedback control system for performing PI (P is proportional) or PID (D is differential) adjustment calculation output converted to a position type signal after performing adjustment calculation as an operation signal to control the object to be controlled, and disturbance compensation In a feedback feedforward control device in combination with a feedforward control system for performing the above, the feedforward control system is configured to multiply a disturbance signal by a feedforward gain to obtain a static characteristic compensation component signal and a dynamic characteristic compensation signal. And a feed-forward control model that is separated into and a predetermined dead band are set in advance, and are output from the feed-forward control model. When the velocity type signal obtained by converting the characteristic compensation component signal is in the dead band, zero is output, and when it is outside the dead band, a predetermined signal is adjusted at least the velocity type I according to the magnitude of the velocity type signal. Static characteristic compensating means to be added to the calculation output, and the dynamic characteristic compensation component signal added to the PI or PID adjustment calculation output when it is determined whether or not the velocity type signal is in the dead zone And a dynamic characteristic compensation canceling means for forcibly setting the speed control signal to zero, and only the feedback control is performed when the velocity type signal is within the dead band.

(Operation) Therefore, according to the present invention, by taking the above-mentioned means, after converting the static characteristic compensation component signal obtained by the feedforward control model into the velocity type signal, the velocity type signal is given a predetermined insensitivity in advance. When the velocity type signal is in the dead band, the static characteristic compensation component signal is released by passing through the dead band setting means in the band, and at the same time, the velocity characteristic signal is in the dead band by the dynamic characteristic compensation releasing means. Only the FB control is executed by discriminating and releasing the dynamic characteristic compensation signal.

On the other hand, when the magnitude of the change in the static characteristic compensation component signal is above a certain level, it is necessary to proactively cancel the influence of the disturbance. Therefore, when the level change of the static characteristic compensation component signal is large, the velocity-type signal is in the dead zone. FF control is performed by determining that it is outside the range and performing static characteristic compensation according to the level of the level, and at the same time making use of dynamic characteristic compensation, by executing control by combining FB control and FF control. This is to carry out highly accurate and stable control while making appropriate use of it.

(Embodiment) Prior to describing an embodiment of the present invention, a speed type PID adjustment calculation method applied to the device of the present invention will be described below. First, the basic expression of the PID algorithm in the general PID adjustment calculation method is as follows: MV = K _P {e + (1 / T _I ) ∫edt + T _D (de / dt) MV ₀ } ...
... (7) However, in the above equation, MV is an operation signal, e is a deviation, K _P is a proportional gain, T _I is an integration time, T _D is a differentiation time, and MV ₀ is an initial value of the operation signal.

By the way, in the digital calculation method using the basic formula of the above formula (7), the sampling period τ is determined in advance, and necessary data is taken in every sampling period τ to perform the calculation. Therefore, the current sampling time is set to nτ (n
And the sampling time immediately before that is (n−
If 1) tau, the deviation of the current sampling time point obtained from the control system e _n, the deviation of the previous sampling time point e
It can be omitted by _n-1 .

On the other hand, there are two types of digital arithmetic methods, one of which is a position type arithmetic method and the other of which is a velocity type arithmetic method. This position type calculation method is a method of directly calculating the entire operation signal MV _n in each sampling cycle, and the velocity type calculation method obtains only the change ΔMV _n of the operation signal from the previous time in each sampling cycle of this time. Later, this change Δ
By adding MV _n to previous operation signal MV _n-1, is a method of the present operation signal.

Therefore, when the position type calculation method and the velocity type calculation method are executed based on the basic expression of the PID algorithm of the expression (7), in the former position type calculation method, In represented, the latter velocity type operation _{method, ΔMV n = K P {(} e n -e n-1) + (τ / T I) e n +
_{(T D / τ) (e} n -2e n-1 + e n-2)} ...... (9
a) MV _n = MV _n-1 + ΔMV _n (9b)

Therefore, these two arithmetic expressions, that is, (8) and (9a)
Comparing equation (9b) and equation (9b), the speed-type PID algorithm of equations (9a) and (9b) eliminates Σ from the integral term and can be easily calculated.
Substitute the operation signal obtained by the manual operation at the present time into MV _{n-1 of the} equation (9b), then switch to automatic control and add the change ΔMV _n to the MV _n-1 from the next sampling time. Control can be continued as it is, so-called balanceless bumpless switching of so-called manual-automatic switching can be easily performed, reset windup can be easily performed by the integral term, and only the change in the operation signal needs to be calculated. It has features such as the ability to modify the DDC (DDC (D
The speed type PID calculation method is often used for irect Digital Control).

Therefore, in the device of the present invention, the speed type calculation method is effective in the PID control in the FB control, and the dead band is set by setting the dead band by using the static characteristic compensation signal as the speed type signal.
It is to improve the conventional problem by appropriately eliminating the FF control output that disturbs the control output.

An embodiment of the present invention will be described below with reference to FIG. In the figure, 11 is the control target from the current target value SV _n
12 of the current process variable value PV _n deviation calculation means for calculating a deviation e _n is subtracted, 13 on the basis of the deviation e _n (9a)
The adjustment operation of the formula is executed, and the obtained adjustment signal ΔMV _n is introduced into the velocity type / position type signal converting means 15 via the adding means 14. The signal converting means 15 executes the operation of the equation (9b), that is, MV _n-1 + ΔMV _n ′ to obtain the operation signal MV _n , and then controls the signal MV _n ′ via the adding means 16 to the control target 12 is applied to, performs control of the deviation _{e n = 0 (SV n =} PV n).
Note that the FB control system is configured by removing the adding means 14 and 16 from the components 11 to 16.

On the other hand, the FF control system receives the disturbance signal D _n and receives the disturbance signal D _n.
A static characteristic compensation component and a dynamic characteristic compensation component are separated by using a feedforward control model 21 including a coefficient means 21a for multiplying _n by the FF gain k and a dynamic characteristic compensation means 21b. Here, it is proved that the feedforward control model 21 can be separated into a static characteristic compensation component and a dynamic characteristic compensation component. Since the feedforward control model is expressed by G _F (S) = k · f (S) from the formulas (3) to (6) described above, from this formula, G _F (S) = k {1+ [f (S) −1]} = k {1 + [(T _P1 −T _D1 ) · s + (T _P2 .T _D2 ).
^{· S 2 + ...... + (T} Pn -T Dn) · s n ] / (1 + T _D1 · s
+ T _D2 · s ² + …… + T _Dn · s ⁿ )} (10), which can be separated into the static characteristic compensation component of the front stage underline and the dynamic characteristic compensation component of the rear stage underline that are not related to time at all. . Then, when the transfer functions of the equations (1) and (2) are linearly approximated, the equation (10) is: G _F (S) = k {1 + [(T _P1 −T _D1 ) · s] / (1+
T _D1 · s} (11) = k {1 + [(T _P1 · s] / (1 + T _D1 · s) −
1]} ... (12) can be obtained.

Then, the static characteristic compensation component signal A _n separated in the feedforward control model 21 is introduced to the static characteristic compensation means. This static characteristic compensating means uses the static characteristic compensation component signal A.
Cut the following small signal level is out of the converted velocity type signal .DELTA.A _n and the position-type / velocity type signal converting means 22 in this signal conversion means 22 for converting the _n to the speed-type signal .DELTA.A _n,
It is constituted by the dead zone setting unit 23 for outputting a signal .DELTA.B _n for canceling proactively a large signal above a certain level, adding the static characteristic compensation signal .DELTA.B _n passing through the dead band setting hand 23 to the summing means 14 It is composed.

Further, in this device, in addition to the static characteristic compensating means,
Dynamic characteristic compensation releasing means is provided. The dynamic characteristic compensation canceling / removing means is composed of a dead zone discriminating means 24 for discriminating whether ΔA _n is in the dead zone or out of the dead zone from the input ΔA _n and the output ΔB _{n of the} dead zone setting means 23, and the dead zone discriminating means 24. When it is determined to be within the dead zone, the introduction of the dynamic characteristic compensation component signal ΔE _n separated by the feedforward control model 21 to the adding means 16 is released, while when it is determined to be outside the dead zone, the dynamic characteristic compensation component signal ΔE _n is released. And a switch means 25 for introducing the above into the adding means 16.

Next, the operation of the apparatus configured as described above will be described.
First, in the FB control system, the deviation _{e n} receiving speed PI controller calculating means 13 is above the deviation calculating means 11 (9a) _{equation, ΔMV n = K p {(} e n -e n-1) + (τ _/ T _I) running _e n} consisting adjusting operation to obtain an adjustment signal .DELTA.MV _n. And
This adjustment signal ΔMV _n is made to be ΔMV _n ′ via the adding means 14 and introduced into the following speed type / position type signal converting means 15, where MV _n = MV _n−1 based on the equation (9b). After calculating + ΔMV _n ′ to obtain the position type operation signal MV _n , the position type operation signal MV _n is applied to the controlled object 12 via the adding means 16 to execute the FB control.

On the other hand, in the disturbance compensation system, when the disturbance signal D _n is introduced into the feedforward control model 21, the disturbance signal D _n is multiplied by the FF gain k to obtain the static characteristic compensation component signal A _n, and dynamic characteristic compensation means 21b from the compensation component signal A _n
Then, the dynamic characteristic compensation component signal ΔE _n is obtained.

The static characteristic compensation component signal A _n of the disturbance compensation component signal obtained as described above is guided to the position / velocity signal conversion means 22, where ΔA _n = A _n −A _n−1 ( 13) The following calculation is performed to obtain the difference, that is, the velocity type signal ΔA _n , which is the variation of the static characteristic compensation component signal. As is apparent from the equation (13), the velocity type signal ΔA _n is (1) when A _n = A _n−1 , that is, when there is no change ...
When ΔA _n = 0 (2) A _n > A _n-1 , that is, when it is increasing ΔA _n > 0 (3) When A _n <= A _n-1 , that is, when it is decreasing ...... ΔA n _<= 0 becomes, .DELTA.A _n is when the a _n unchanged zero, changes to the positive and negative around the zero in proportion to the magnitude of the change when the a _n increases or decreases.

Therefore, the velocity-shaped signal ΔA _n having the above-mentioned properties is passed through the dead zone setting means 23 having a dead zone of a predetermined magnitude in advance, so that the magnitude of the velocity-shaped signal ΔA _n is set to a predetermined value centered on zero. ΔB _n = 0 is output when the size is equal to or smaller than the predetermined value, that is, when the value is within the dead zone, and ΔA _n = ΔB _n is output when the value is _equal to or larger than a predetermined value, that is, when the value is outside the dead zone. That is, if the size of the dead zone [delta], _{(b) | ΔA n | ≦ δ ΔB} n = 0 ( _{ii) | ΔA n |> δ ΔB} n = ΔA n , and the change in .DELTA.A _n of the static characteristic compensation component signal _{A n} Is smaller than the predetermined magnitude δ, static characteristic compensation is not effective. On the other hand, when the predetermined magnitude δ is exceeded, the static characteristic compensation signal ΔB _n is added to the adding means 14 in accordance with the magnitude of the velocity type signal ΔA _n. Will be introduced to.

The adding means 14 uses the speed-type adjustment signals ΔMV _n and ΔB _n of the speed-type PI adjustment calculating means 13 to calculate ΔMV _n ′ = ΔMV _n + ΔB _n (14) Position / speed type signal conversion means 15
Then, after being converted into a position type signal by the calculation of MV _n = MV _n−1 + ΔMV _n ′ (15), it is introduced into the adding means 16.

The dynamic characteristic compensation component signal ΔE _n separated by the feedforward control model 21 is introduced into the adding means 16, and in this case, the dynamic characteristic compensation is performed depending on whether the velocity type signal ΔA _n is within the dead zone or outside the dead zone. I do. The reason is that the dynamic characteristic compensation component signal ΔE _n generally changes greatly when the disturbance changes sharply. Therefore, the magnitude of change in the static characteristic compensation component A _n can be judged more accurately to perform compensation. Because you can.

Therefore, in this device, the input Δ of the dead zone setting means 23
Leads to the A _n and the output .DELTA.B _n the dead-zone determination unit 24, where, (a) when the change .DELTA.A _n of the static characteristic compensation component signal A _n is in the dead zone, the dynamic characteristic compensation as a non-conductive switching means 25 released, (b) when the change .DELTA.A _n of the static characteristic compensation component signal _{a n} is the outer,
The switch means 25 is turned on to perform dynamic characteristic compensation.

During the operation of compensation, _'give consisting operation signal, the operation signal MV _n' MV _n by adding the operation compensation amount signal Delta] E _n to the output MV _n adding means 16 in the signal converting means 15 to the controlled object 12 Apply and control.

Therefore, as described above, according to the configuration of the embodiment, in the FF control system, the FF control signal is separated into the static characteristic compensation component signal and the motion compensation component, and the static characteristic compensation component is taken as a velocity type signal by taking the difference. By doing so, it becomes possible to pass this velocity type signal through the dead zone setting means 23 having a dead zone of a predetermined size, and when the change in the static characteristic compensation component signal is small, FB control due to static error or noise of FF control. Can be avoided. Also, when the change in the static characteristic compensation component signal is small, that is, when the signal is in the dead zone, the FB control is utilized only by forcing the dynamic characteristic compensation component signal to zero, while when it is outside the dead zone, the dynamic characteristic compensation component is used. By combining FB control and FF control by utilizing the signal as it is, FB control
The controlled object 12 can be controlled by appropriately combining both of these controls while taking advantage of the characteristics of the FF control.

This means that FF control is used to suppress the influence of disturbance in advance when the change of the disturbance is large, and when the change of the disturbance is small, the FF control is stopped and only the FB control is performed. It is possible to realize limit control by making the most of the control features.

In the above embodiment, the speed calculation PI is used as the adjustment calculation means 13.
However, a speed type PID may be used.
In addition, the present invention can be modified in various ways without departing from the scope of the invention.

[Effects of the Invention] As described above, according to the present invention, the FF control system is stopped according to the magnitude of the static characteristic compensation component signal, and only FB control is performed.
Alternatively, by utilizing the FF control system to combine the FB control and the FF control, it is possible to improve the control accuracy and realize stable control while maximizing the characteristics of the FB control and the FF control. It can greatly contribute to the flexibility and improvement of controllability.

[Brief description of drawings]

FIG. 1 is a block diagram for explaining one embodiment of the device of the present invention, and FIG. 2 is a block diagram of a conventional device. 11 …… Deviation calculation means, 12 …… Control target, 13 …… Speed type PI
Adjustment calculation means, 14 ... Addition means, 15 ... Position type / speed type signal conversion means, 16 ... Addition means, 21 ... Feed forward control model, 22 ... Position type / speed type signal conversion means,
23 ... dead zone setting means, 24 ... dead zone discrimination means, 25 ...
... switch means.

Claims (1)

[Claims] 1. PI (P is proportional) converted into a position type signal after at least a velocity type I (integral) adjustment calculation is performed to make a deviation between a process variable value and a target value from a controlled object zero. Alternatively, feedback / feedforward control in which a feedback control system for applying a PID (D is a derivative) adjustment calculation output as an operation signal to the controlled object to control it and a feedforward control system for performing disturbance compensation using the disturbance signal are combined In the device, the feedforward control system separates a signal obtained by multiplying a disturbance signal by a feedforward gain into a static characteristic compensation component signal and a dynamic characteristic compensation component signal, and a predetermined insensitivity. The band is set, and the velocity-type signal obtained by converting the static characteristic compensation component signal output from the feedforward control model is Static characteristic compensating means for outputting zero when in the band and adding a predetermined signal to at least the speed type I adjustment calculation output according to the magnitude of the speed type signal when outside the dead band, and the speed type signal. Is in the dead band, and when it is in the dead band, the dynamic property compensation canceling means for forcibly setting the dynamic property compensation component signal added to the PI or PID adjustment calculation output to zero is provided. A feedback / feedforward control device characterized in that