On PID Parameter Tuning as Well

Thread Content

Actually, before I started writing this, I saw many, many similar posts in the group. When I first began working with PID parameter tuning, I was completely confused; the more posts I read, the more unsure I became about where to start. For example, that famous mnemonic with a ratio of two to one, high at the beginning and low at the end, and so on. These three small parameters, proportional, integral, and derivative control, possess magical powers. It can move wildly like a demon, or remain completely still. There’s actually a lot more nonsense, but I won’t go into it for now. From not knowing where to start before, to handling it with ease now. It was nothing more than going through a particularly painful process. First of all, when tuning PID parameters, you need to be able to read trend charts. Open the trend charts and compare the trends. Some people might say, who can’t do that? How many curves are needed in a trend chart? How long should the curve be? The current value, valve output, and set value are required. As for how long to watch, it depends on whether you are using manual, automatic, or cascade mode at the moment Secondly, it depends on what kind of DCS it is. Different DCS systems use different terms for PID. Some DCS systems do not use a proportional element but rather a proportional band; this is the case with companies like Yokogawa and Hollyland, where the relationship between proportionality and the proportional band is given by proportionality = 100/proportional band. In other systems, it is not referred to as proportionality but as gain; this is the case with DeltaV. ABB and Siemens also use slightly different terms, but the basic calculation principle remains the same. Again, before starting debugging, remove the differentiation first, if it is present. It will cause unnecessary interference, affecting judgment and calculations. As agreed, let’s get back to the main topic. Generally speaking, there are two common methods for tuning PID parameters: the trial and error method and open-loop calculation. I have never used the method for calculating the period when a circuit starts to oscillate, as mentioned in many posts or textbooks. Because in actual production processes, especially in chemical manufacturing, such actions are basically not allowed. Workers or engineers prefer to do it manually rather than have fluctuations in the curve. This also goes against the original purpose of parameter tuning. Write it here for now; I’ll continue next time.