Thread Content
Last time we talked about PID tuning, two common methods are generally used, or rather, the methods I personally use. Trial-and-error method and open-loop step method. First, let’s talk about the trial and error method. By trial and error, it’s not meant that you simply pick a value at random and then see if it works well or whether it is stable or not. For a proper and effective PID parameter, it is first necessary for it to be stable; secondly, it must be able to respond promptly when the set value changes. By \"promptly,\" it does not mean that if you set an integration time of 5 minutes for the flow rate, it will take half an hour to reach the set value. Although it is stable, it does not meet the second requirement, which is the ability to resist interference. The entire engineering operation is a dynamic process, and as such there are many unforeseeable factors: fluctuations in device load, problems with valves and instrumentation, and adverse weather conditions, among others. If the control loop starts to oscillate or diverge as soon as a problem arises, then such a loop cannot be considered to have undergone proper parameter tuning. The trial-and-error method involves, based on the current PID parameters, quickly making judgments regarding these parameters by considering current or historical trends, thereby enabling simple and effective system tuning. It should be understood that PID parameters are not unique; there is no optimal value, only better ones. When using the trial and error method, one first considers the changes in the set value within the current or historical trends. For example, if the current PID parameters are a gain of 1 and an integral value of 1 unit, at the start of such a change in the set value, the valve will produce an output of, say, 2%. Once stability is achieved, the actual opening degree of the valve will be 5%. That means the ratio is too low; it should be 5%/2% = 2.5. The same is true in reverse. This allows for a simple determination of the proportion value. For example, if the current or historical trend shows volatility, the simplest approach is to halve the proportion and double the integration time. This is simply the trial-and-error method. Fast and effective.