The Study Of The Nnpid Control And The Pipelines Dynamic Characteristic Of The Bending Roll

Flatness is one of the most important factors to evaluating the quality of a strip, it has many influence factors, and its relevant laws are complex. Hydraulic Bending Roll shape control system is one of the shape control means, its control quality on the plate-shape is good or bad control, that has a direct impact on the plate-shape. In this paper, the pipelines that connect servo valve and hydraulic cylinder in the hydraulic bending roll system is the research object, the study aim is to analyze the effect of pipeline parameters on the hydraulic bending roll control performance of the system. In addition, because hydraulic bending roll system is a nonlinear, multi-disturbance servo system, the traditional PID control method can not meet the requirements of precision of its control, in this paper, I design an intelligent PID control in order to improve the system dynamic performance.Firstly, I reference to the achievements of predecessors about the dynamic characteristics of the pipeline, using the Distributed parameter model of the pipelines to conduct research and analysis of pipelines. Build the comprehensive mathematical models of the various components including the pipeline of bending roll force control system, laid the foundation to simulate use of the MATLAB in next step. By the simulation, I obtain some results about some pipeline parameters effects on the stability of the system factors.Secondly, the bending roll system contains many non-linear, time-varying, etc. unstable factors, I Design a neural network PID controller based on BP algorithm. Using the neural network to adjust the three control parameters of the PID of the system to achieve the best control effect. Through the use of the MATLAB simulation, contrast the neural network PID control performance with the traditional PID, get the result that the neural network PID can come to better improve the system overshoot and response time.Finally, through the experiments that change the pipe diameter, pipe length about a pipeline of 300 mm four roller reversible cold rolling mill roll bending systems (positively curved part) in our school mill laboratory, As well as using the C language to program the neural network PID control and insert the program in the mill control program to carry out control experiments, the experiment results are in good line with the simulation results. This shows that analysis of the pipeline parameters affecting on the system stability factors is roughly in line with reality, which lays a foundation to choose a suitable pipe in practice, The designed neural network PID can control the bending roll system better.