Intelligent Controller Design And Experiment Study Oncold-rolling APC System

In the production of cold-rolled strip, the accuracy of strip thickness is an important quality evaluation, and the performance of the cold rolling APC(Automatic Position Control, APC) system is directly related to the thickness precision of the cold-rolled strip products. Currently the PID controller, which is most widely used in the cold rolling fields requires manual adjustment of parameters, has poor adaptability in response to time-varying system and its nonlinearness and there are many limitations. With the continuous development and improvement of control theory, it has a very important research value to apply the intelligent algorithm into APC system and make corresponding improvement and innovation, which is based on actual production needs.First, on the basis of in-depth presentation structure and operating principles of the cold-rolling APC system, a linear mathematical model and a nonlinear mathematical model of hydraulic cold rolling APC system are built, and a homologous circuit board is also made based on the linear model.Secondly, a novel PID controller based on the DE(Differential Evolution) algorithm is proposed according to the linear model above. In order to make a more comprehensive evaluation, a multi-index weighted fitness function is introduced. Mutational zooming factor based on Individual health value and crossover probability based on prematurity index are added in so as to enhance optimizational speed and avoid partial optima, obtaining IDE-PID(Improved Differential Evolution-PID) controller finally. MATLAB simulation and PLC experiment are carried out at the end of the chapter.Again, with regard to inefficient adjustment of IDE-PID algorithm, another innovative auto-tuning fuzzy-PID controller is proposed. A thorough search algorithm for the fuzzy rules is also put forward to refine the fuzzy rule base. Combined with the IDE algorithm, IDE-fuzzy PID controller is obtained. In the end, MATLAB simulation and PLC experiment are carried out.Finally, the non-linear model acts as the research object and then a network controller based on Elman network identification is developed. Dynamical Elman network may identify nonlinear factor accurately. Furthermore, At the same time, a feedforward network is added in on the basis of Elman network. The advantages of the GPC are learned in order to improve system performance. MATLAB simulation is carried out finally.