Improved single-neuron PID tuning method based on discrete system

Abstract

The invention relates to an improved single-neuron PID tuning method based on a discrete system. When a control system is in the dynamic state, a single-neuron PID controller with the self-adaptive proportionality coefficient lambda is adopted, and the single-neuron proportionality coefficient is completely controlled by the input-output error e(k), so that parameters of the single-neuron PID are completely and automatically tuned by the system, and manual evaluation is not required; the single-neuron PID is poor in steady state effect, so that after the system is in the steady state, the conventional PID with the self-adaptive integral grain Ki is adopted to improve the steady-state performance, and the integral grain parameter Ki is also completely controlled by the input-output error e(k) of a controlled object. According to the method, all the parameters are automatically tuned, the method is high in robustness, and the control effect cannot be degraded along with aging and damage to the controlled object; the parameters can be adjusted automatically during interference, so that the effect caused by the interference on the control system is reduced.

Description

technical field [0001] The invention relates to a PID parameter tuning method of a discrete system, in particular to a PID tuning method based on an improved single neuron of a discrete system. Background technique [0002] It is known that the PID controller is widely used in various industrial control systems, and has good control effect and simple structure. In discrete systems, PID parameter tuning is mainly for proportional gain Kp (Proportional Gain), integral gain Ki (Integral Gain) and Differential Gain k (Derivative Gain). At present, the setting of these three parameters is still artificially based on experience. Although good results can be achieved, for non-professionals and beginners, parameter setting is not an easy task. With the development of intelligent control technology, some methods of self-adaptive tuning of PID parameters have begun to attract people's attention. Among them, the single-neuron PID controller has a relatively simple structure, has ...

AI Technical Summary

Problems solved by technology

[0012] Although the single-neuron PID controller has many advantages, it also has a big defect: the output of the single-neuron PID controller Scale factor with single neuron The choice has a great relationship, the final The choice of is also an empirical choice, not an adaptive controller in a complete sense; due to the steady state, the controlled object input with output error is always present, resulting in a single-neuron output It is not stable, but fluctuates up and down. The size of the fluctuation directly affects the control effect. That is to say, the effect of the single neuron PID controller is not ideal in the steady state.

Images