qjj1020: A commonly used method in PID debugging is the critical proportionality method:
1. After the controlled system is stable, set the integral time of the controller to the maximum and the differential time to zero (equivalent to removing the integral and differential action, and only using the proportional action).
2. Observe the oscillation of the measured value caused by external disturbance or a step change of the controller setting value.
3. From large to small, gradually reduce the proportionality of the controller to see if the oscillation of the measured value is divergent or attenuated? If it is attenuated, the proportionality should continue to decrease; if it is divergent, the proportionality should be enlarged. .
4. Repeat steps 2 and 3 continuously until the measured value oscillates with a constant amplitude and period, that is, it continues to oscillate 4--5 times with equal amplitude. The proportionality indication value at this time is the critical proportionality PB. Then the PID calculation is performed according to the critical oscillation formula.
What I want to ask is, is the so-called constant-amplitude oscillation at the upper and lower (greater or less than the set value) of the set value arbitrarily set to do constant amplitude and two adjacent peaks or troughs with constant period of oscillation? However, the constant amplitude oscillation only below the set value is not the required oscillation. Simply put, this constant amplitude oscillation is the constant amplitude oscillation above and below the set value.
The following is my reply: a "pure" linear system is very difficult to have stable constant amplitude oscillation. Due to various nonlinearities in the actual system, such as the saturation nonlinearity of the PID controller (limiting effect on the output), it is not difficult to observe the constant amplitude oscillation. This kind of oscillation is generally equal amplitude oscillation above and below the set value.
Many books have critical scaling methods to determine the initial parameters of the PID. Some systems do not allow critical oscillation, and the amplitude of the critical oscillation may be larger, or it may become a divergent oscillation with an increasing amplitude, so this method is dangerous to some systems.
The closed-loop can be debugged by the following methods: First, select the PI controller, and try to keep the initial parameters as conservative as possible (the proportional coefficient should be as small as possible, and the integration time should be as large as possible) to prevent large overshoot or system instability during the initial operation. The controller parameters are then adjusted according to the characteristics of the step response.
If the overshoot of the step response is too large, the proportional coefficient of the controller should be reduced and the integral time should be increased.
If the step response has no overshoot, but the controlled variable rises too slowly and the transition time is too long, the parameters should be adjusted in the opposite direction.
If the speed of eliminating errors is slow, the integration time can be appropriately reduced.
Repeatedly adjust the proportional coefficient and the integration time, if the overshoot is still large, you can add a differential part to gradually increase the differential time.