JPS6118207B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention is a process for determining the transient step response characteristics of a process by changing the input amount of the process by a predetermined amount in steps, and temporally sampling the output amount of the process that changes accordingly. Concerning the test method. In process control, in order to perform optimal control, it is necessary to know the dynamic characteristics such as the transfer function of the process. In order to determine the dynamic characteristics of such processes, dynamic characteristics tests are performed. By changing the input amount of the process by a predetermined amount in steps and capturing the change in the output amount of the process over time, the step response characteristics of the process can be determined. If we evaluate this step response characteristic diagrammatically, in the case of a localization process, x in Figure 1a is
The characteristic line will be as shown by x(t) in FIG. 1b in the case of a non-localizing process. In the diagrams in Figures 1a and b, the vertical axis represents the process output x, and the horizontal axis represents the time t.
represents. In the diagram of Figure 1a, if we calculate the area S _T of the part surrounded by the characteristic line x(t) and the straight line 1e indicating the final stable value x(∞) of the output quantity x, we can calculate the overall delay in the process. The time T _T can be calculated, and the characteristic line x
The gain K of the process can be determined from the final stable value x (∞) of (t), and from these, the transfer function etc. indicating the dynamic characteristics of the localization process can be determined. In addition, in Figure 1b, a tangent ls is drawn at the point where the differential coefficient becomes almost constant during the rising process of the characteristic line x(t), and the intersection of this tangent ls with the time axis and the differential coefficient of the characteristic line x(t) From these, the delay time, integration time, etc. of the process can be calculated, and from these, the transfer function etc. indicating the dynamic characteristics of the non-orientating process can be determined. In this way, dynamic characteristic values such as process transfer functions can be determined from the step response characteristics.
Dynamic characteristic tests to determine the dynamic characteristics of a process are mainly performed using step response tests. Therefore, in order to obtain the complete step response characteristics of a process, the test input that changes the step input to the process input must be continued until the output quantity settles down to the final stable value, so this test input must be continued. It takes a very long time. However, since the test input that changes in steps for such a test becomes a disturbance to the process, it is necessary to shorten the duration as much as possible to avoid adversely affecting the process. SUMMARY OF THE INVENTION In view of the above, an object of the present invention is to provide a test method that can shorten the duration of test input of a step change applied to a process in a test for determining step response characteristics of a process. In order to achieve such an object, the present invention changes the input amount of the process by a predetermined amount in steps, and temporally samples the output amount of the process that changes accordingly to determine the step response characteristics of the process. In the test method, the collection of the output amount of the process is started from the time when the input amount of the process is changed, and the temporal change in the sampled output amount of the process is detected, and the amount of change increases and remains approximately constant. or exceeds the maximum value and decreases to a certain ratio of the maximum value, the input amount of the process is returned to the original amount, and if the amount of change becomes constant, the sample is collected at least up to this point. Find the dynamic characteristic value of the process based on the output amount of the process that has been changed, and if the amount of change decreases, collect the output amount until the output amount of the process returns to the original amount, and then calculate the dynamic characteristic value based on the output amount of the collected process. This method is characterized by determining the dynamic characteristic values of the process. Next, the present invention will be explained with reference to the drawings. FIG. 2 is a block diagram showing a schematic configuration of a test apparatus for a process based on the method of the present invention. In this figure, 1 is a process control device and P
is the process to be controlled, FC is the operation unit that operates the process P, C is the adjustment unit that gives the operation signal C _S to the operation unit FC, and the adjustment unit C is the target value S _O
and the output control amount P _S of the process P, and generates an operation signal C _S according to the deviation. Reference numeral 2 designates a test device, in which an operating unit M includes a test signal generator TD that provides a step-like test operation signal T _S of a predetermined amplitude that can be varied via a changeover switch SW, and a test signal generator TD that generates a control variable P _S of a process P for a certain period of time. A memory M that reads and stores information at intervals, and a change detector that reads the control amount stored in the memory M and detects changes over time.
Constructed by AD. The change detector AD has a function of generating a detection signal A _S and giving it to the test signal generator T _D when the amount of change in the controlled variable increases and becomes almost constant, or exceeds the maximum value and decreases to a certain ratio of the maximum value. has. When the test signal generator T _D receives the detection signal A _S , it operates to stop the test signal T _S . In the apparatus shown in FIG. 2, the switch SW is normally closed to the adjustment section C side, and automatic control is performed so that the controlled variable P _S of the process P matches the target value S _O. Therefore, when performing a dynamic characteristic test, the switch SW is switched to the test signal generator TD side of the test device 2, and the operation section FC is separated from the adjustment section C. In this way, at time t _p the test signal generator TD
Therefore, when a test signal T _S of a predetermined step-like amplitude as shown in FIG. The controlled variable P _S of the process P changes over time according to its own dynamic characteristics as shown in FIG. 3b. Memory M is test signal T
The control amount P _S is collected at regular time intervals from a given point in time t _p of _S and stored sequentially. The change detector AD compares the controlled variable P _Si at each point in time stored in the memory M with the controlled variable P _Si-o at the previous point or an appropriate time before, and detects the amount of change over time. , to compare it with the maximum value of the amount of change held up to that point, and if the amount of change at that point is greater than the maximum value held, continue to update the maximum value of the amount of change held to the new maximum value. , the change detector AD always holds the maximum value of the amount of change in the control amount P _S up to each sampling point. Therefore, the period in which the amount of change detected by the change detector AD is equal to the maximum value of the amount of change held continues for a predetermined period, and it is determined that the amount of change has become constant or the maximum value of the amount of change Dmax is the value of a predetermined ratio of that value, for example, the maximum value of change
Then, when it is determined that the value has decreased to a value of 0.368·Dmax, the change detector AD generates a detection signal As and supplies it to the test signal generator TD. Now, if the process P has a first-order transfer function of G(S)=K/1+TS, then the temporal change Ps(t) in the controlled variable Ps when the manipulated variable is changed in a stepwise manner is , It is indicated by. Differentiating this equation (1), we get This indicates the amount of change in the control amount Ps. As is clear from equation (2), Ps(t)' is maximum when t=o, and its value Ps'(o) is Ps'(o)=K/T. Taking the ratio of this Ps′(o) and Ps′(t), we get

[Formula], when t=T, Ps'(T)/Ps'(o
)=e ^-1 =0.368. Therefore, if the process P is a first-order delayed process, the time point at which the change detector AD generates the detection signal As is the time point when t=T, that is, the test signal
This is the time when a time corresponding to the time constant T of the process P has elapsed from the given to time point of Ts. Furthermore, considering the case where process P has a quadratic transfer function of G(s)=K/(1+Ts) ² , the response of the controlled variable to a step change in the manipulated variable is a time function as shown in equation (3). becomes. The first derivative of this equation (3) is becomes. This Ps(t), that is, the maximum value of the amount of change in Ps, appears when the equation (4) is further differentiated and the value becomes 0, and when that point is found, it is the time t=T. t=T in equation (4) and the overall time constant of the process from the time when the manipulated variable is changed in a stepwise manner.
The value of t=2T after the time corresponding to 2T has passed is
Ps'(T)=K/Te ^-1 , Ps'(2T)=2K/Te ^-2 . Therefore, Ps'(2T)/Ps'(T)=2e ^-1 ==0.
It becomes 736. In addition, the value of equation (4) at time t = T when the amount of change reaches its maximum value and further at time t = 3T, which corresponds to the time constant 2T of the process, is Ps' (T) = K/Te ^-1 , Ps′(3T)=3K/Te
^-3 . Therefore, Ps'(3T)/Ps'(T)=3e ^-2
=0.406. As is clear from this consideration, G(s)=
For a process with a transfer function of K/(1+Ts) ² , the overall time constant of the process is calculated from the time when the amount of change in the controlled variable in response to a step change in the manipulated variable reaches its maximum value.
When a time corresponding to 2T has elapsed, the amount of change in the control amount decreases to a value that is 0.406 times its maximum value. In addition, in order to obtain the dynamic characteristics of the process using a pulse response, the duration of the test input is required to be a length corresponding to the time constant of the process. 1 mentioned above
In a process with the following transfer function, if the test input is cut off when the amount of change decreases to 0.368 times the maximum value, the duration of the test input will be equal to the time constant T of the process, and in a process with a quadratic transfer function, If the test input is cut off when the amount of change has decreased to 0.406 times the maximum value, the duration of the test input will be the process time constant 2T plus T. In a process having an n-th order transfer function, the larger n becomes, the larger the ratio between the amount of change and the maximum value becomes when the time constant of the process has elapsed from the time when the amount of change reached the maximum value. The step response characteristics in the case of the localization process are as shown in Figure 1a, and the higher the transfer function, the more the characteristic line x(t) loses its initial response characteristics and the response is less likely to appear immediately. However, since the amount of change in the control amount always reaches 0 after a time period at which it becomes maximum, it is possible to set a ratio that necessarily decreases with respect to the maximum value. Therefore,
If the test input is turned off when the amount of change decreases to 0.368 times the maximum value based on a process with a first-order transfer function, it is possible to test a process with an n-order transfer function for which the characteristics of the process are completely unknown. Even in this case, the duration of test input can be made longer than the time constant of the process. In an actual process test, since noise and the like can be detected in the control amount, it is preferable to make the ratio of the amount of change to the maximum value for turning off the test input smaller than 0.368 and to make the duration of the test input somewhat longer. In this way, the change detector AD detects the point at which the temporal change in the process control amount for the step-like test signal exceeds the maximum value and decreases to a predetermined ratio of the maximum value, and at this point the test signal Ts By cutting , a length corresponding to the time constant of the process P can be secured as the duration of the test signal even if the characteristics of the process are not known. Considering a general transfer function with an n-th delay, the transfer function is G(s) = K/(1+Ts) ⁿ , and the response of the controlled variable to a step change in the manipulated variable is (5)
It becomes a time function as shown in the formula. The first derivative of this equation (5) is Also, the second derivative is becomes. Therefore, the maximum value of the amount of change in Ps appears at the time t=(n-1)T from equation (7). The above explanation is for the case where the process has localization, but when the process has no localization, the step response of the process diverges, so the amount of change is maximum at the time of delay corresponding to the time constant, and after that, Indicates a change that does not change and remains almost constant. In order to cope with this, the change detector AD also has a function to detect a change in which the amount of change increases and its value remains approximately constant. Therefore, even if the process has nonorientation, the change detector AD
provides the detection signal As to the test signal generator TD at a time delay corresponding to the time constant of the process. In this way, at time ta in Fig. 3, the control amount Ps
When the amount of change D over time exceeds the maximum value Dmax,
When a decrease to 0.368Dmax is detected, the detection signal As is output from the change detector AD as shown in Figure 3c.
is generated. As a result, the test signal generator TD stops supplying the test signal Ts, so the operation amount Fs given to the process P from the operation unit FC is returned to its original value.
As a result, as shown in FIG. 3b, the control amount Ps of the process P increases slightly after time ta, then decreases and returns to the original amount. If the change detector AD detects that the amount of change in the control amount Ps decreases beyond the maximum value, it means that the process is localizing, so the control amount Ps remains the original amount even after time ta. The sampling is continued until it returns and is read and stored in the memory M. Therefore, in this case, all the temporal changes Ps(t) of the control amount Ps shown in FIG. 3b are recorded in the memory M. In the present invention, since the test signal Ts is cut off before the process control amount Ps reaches the final stable value, the control amount stored in the memory M
Ps does not represent a complete step response. However, if the process is linear, the pulse _response characteristic line Ps (t ), the step response characteristic Ps(t)' line shown by the dotted line in FIG. 3b can be obtained by calculation. That is, as is clear from FIG. 3b, during period T _A , Ps (t)' = Ps (t), and during period T _B , Ps (t)' = Ps (t) + Ps (t - T _A ). It can be found as From the step response characteristic line Ps(t)' obtained in this way, the dynamic characteristic value of the process can be obtained using the method described above. In addition, if the change detector AD detects that the amount of change in the control amount Ps is constant at the maximum value, it means that the process is non-localizing, so in this case, the process has already been performed up to this point. Since it is possible to sufficiently obtain the dynamic characteristic values of the process from the data collected at this point, it is possible to stop collecting the control variables at this point. If the process is ataxic, then Figure 4b
As shown in Fig. 4a, simply cutting off the test signal Ts by the detection signal As of the change detector AD does not return the control amount Ps to its original _state . If a test signal -Ts of opposite polarity and a time width equal to that of the first test signal Ts is given after a period equal to
The process control amount Ps can be restored as shown in . The setting ratio to the maximum value for detecting a decrease in the amount of change in the change detector AD is 0.368.
There is no need to limit it to. As explained above, according to the present invention, in order to obtain the step response characteristics of the process, the duration of the input that changes in a step-like manner given to the process is detected by the amount of change in the control amount of the process, and the time constant of the process is determined. This can be shortened by limiting the amount of time that can be secured. Therefore, it is possible to prevent the process from being adversely affected by the test input. In addition, by making the duration of the input to the process longer than the time corresponding to the time constant of the process, it is possible to change the response control amount of the process to a relatively large value of 60% or more of the final stable value. Therefore, the method of the present invention is useful because it has the advantage that data processing for detecting the amount of change can be easily and accurately performed.

[Brief explanation of the drawing]

FIG. 1 is a process step response characteristic diagram, FIG. 2 is a block diagram of a test apparatus based on the method of the present invention, and FIGS. 3 and 4 are explanatory diagrams of the operation of the present invention. 1... Process control system, 2... Test equipment, C...
...Adjustment unit, FC...Operation unit, P...Process, M
...Memory, AD...Change detector, TD...Test signal generator.

Claims (1)

[Claims] 1. In a process testing method for determining the transient response characteristics of a process by changing the input amount of the process by a predetermined amount in steps and temporally sampling the output amount of the process that changes accordingly, the input amount of the process is Start collecting the output amount of the process from the moment when the output amount of the process is changed, detect the temporal change in the sampled output amount of the process, and determine whether the amount of change becomes approximately constant or exceeds the maximum value by a predetermined ratio of the maximum value. At the point when the input amount of the process has decreased to the original amount, and if the amount of change has become approximately constant, the process operation is changed based on at least the output amount of the process collected up to this point. Find the characteristic value, and if the amount of change decreases beyond the maximum value, collect the output amount until the output amount of the process returns to almost the original amount, and change the process based on the collected output amount of the process. A process testing method characterized by determining dynamic characteristic values.