A Dynamic Model Identification Method for Nonlinear Processes

Abstract

The invention relates to a nonlinear process dynamic model identification method, which includes using an experiment module and an identification module, the experiment module is connected with the nonlinear industrial process through DCS or PLC or other control machines, and the experiment module and the identification module are connected to each other. The experiment module generates experiment signals and performs automatic experiments; the identification module uses the existing process experiment data input by the experiment module to automatically identify the nonlinear process dynamic model, checks the quality of the model, and adjusts it according to the quality of the model The signal is input to the experiment module to adjust the current experiment parameters. The invention can carry out identification experiment and model identification on the nonlinear industrial process, and the nonlinear industrial process can be continuous, intermittent or feeding intermittent. The obtained nonlinear process dynamic model can be used in model predictive controllers, conventional PID (proportional, integral and differential) controllers and other advanced process controllers, and can also be used in reasoning models and soft sensors for product quality prediction .

Description

technical field [0001] The invention belongs to the technical field of control, and relates to a model predictive control (MPC) technology, in particular to an identification method of a nonlinear dynamic model in model predictive control (MPC) equipment, which is used to identify Nonlinear dynamic models of production processes in process industries such as pharmaceuticals, metallurgy, food and paper. The device is capable of handling large-scale industrial processes with multiple controlled variables (MV) and multiple controlled variables (CV). The nonlinear model obtained by the present invention can be used in model predictive control (MPC) and other advanced control (APC), and can also be used in reasoning models or soft sensors to predict product quality that cannot be frequently measured due to high cost. Background technique [0002] Model predictive control (MPC: Model Predictive Control) or model predictive control has become a standard advanced control technolog...

AI Technical Summary

Problems solved by technology

However, linear MPC has limitations: when an industrial process works over a large area or operates intermittently, the linear model is usually not accurate enough, so linear MPC may not be able to achieve satisfactory control results

For nonlinear MPC, one of the most challenging tasks is to identify the dynamic model of the nonlinear process. However, there is no method to systematically conduct low-cost identification experiments and establish a reliable nonlinear dynamic process model (Qin and Badgwell, 2000: An overview of nonlinear model predictive control applications, Proceedings Nonlinear Model Predictive Control, edited by F.Allgower and A.Zheng)

Images