Method for evaluating control performance of PID (Proportion Integration Differentiation) loop based on subspace identification

Abstract

The invention discloses a method for evaluating the control performance of a PID (Proportion Integration Differentiation) loop based on subspace identification. The method comprises the following steps of: effectively segmenting a controller output signal u and a process output signal y according to the loop opening and closing states of a PID loop controller and the upper and lower limit constraints of the controller output signal u; solving a random system model by using a subspace identification algorithm; calculating a moment performance index and a performance index sequence at the reach of steady-state time according to the sampling time of the controller output signal u and the process output signal y, set PID loop steady-state time and desired time; and detecting oscillation in the loop according to a model obtained by using space identification and a self-correlation function of the process output signal y. The random system model is established by using the subspace identification, so that setting of a system order is avoided. By adopting the method, loop related information including 95 percent of steady-state time, 95 percent of dynamic time, overshoot and a peak moment is given by making full use of the model obtained by the identification. The performance index is calculated by using reference time set by a user, and is accordant with the practical situation of the PID loop.

Description

technical field [0001] The invention relates to a PID loop control performance evaluation method based on subspace identification. The utility model is used for performance monitoring of a PID loop widely used in the field, and provides loop oscillation information, and belongs to the technical field of PID control application. Background technique [0002] A high-quality control system requires a good controller. However, 60% of PID loops widely used in industrial sites have performance problems. Even if the controller works well in the early stage of the control system, after a period of work, many controllers will have performance degradation, so effective control performance evaluation technology becomes very important. [0003] At present, the commonly used control performance evaluation benchmarks include methods based on minimum variance control, user-specified indicators, and LQG criteria. The control performance evaluation method based on the LQG criterion not on...

AI Technical Summary

Problems solved by technology

However, the performance evaluation standard based on the minimum variance control is too harsh, which makes the performance evaluation results too conservative; moreover, when establishing the stochastic system model, the model of the stochastic system is obtained by using the forecast error method through nonlinear least squares optimization

However, since the model order of the stochastic system is unknown, the forecast error method can only artificially set an order or use the AIC criterion to estimate the order of the system

This will lead to two problems: (1) The order of the system is unknown. Using a too high model order will lead to overfitting and cannot describe the model reasonably. Using a too low model order will inevitably lead to the dynamic characteristics of the system. Not fully described; (2) Using the AIC criterion to estimate the order of the system greatly increases the computational complexity of identification, because the prediction error algorithm itself is also an optimization process, which requires a relatively complicated calculation process

In addition, the existing performance evaluation algorithms do not take into account the experience and knowledge of engineers, so that their experience cannot have a positive impact on the performance evaluation of the control loop

Images