Research On Decoupling Control Of Hot Strip Mill Plate Shape And Plate Thickness Based On Improved Fish Swarm Algorithm

The development of steel rolling technology often requires adequate theoretical research and rich practical experience to provide technical support.After years of development,the automation level of hot strip rolling has reached a high level,and plate shape and plate thickness are two more important measures used to evaluate the level of hot strip rolling and strip steel product quality.AFC-AGC comprehensive system is a coupled,nonlinear,multivariable complex control system,but the traditional control strategies can no longer meet the requirements of the current stage of automation,nor can they better solve the corresponding control problems,so it is necessary to propose a new control theory and method to achieve a qualitative leap in control effectiveness and performance.This thesis is based on the research background of 1700 hot strip mill site of Anshan Iron and Steel,and takes the integrated control system of plate shape and plate thickness as the research object,establishes the corresponding static and dynamic mathematical models,analyzes and studies the connection between the coupling variables in the system,and presents the uncoupling control strategies by combining the modified fish swarm method and DRNN neural network.The research details of this article are as follows:(1)Based on the data obtained at the hot strip mill site,the mathematical model of the system is established by analyzing the connection between the variables in the integrated system of plate shape and plate thickness,and the input-output relationship of the system.The advantages and disadvantages of the more common decoupling control methods so far are then introduced in relation to rolling-related theories.For the shortcomings,the fish swarm algorithm and DRNN neural network are introduced for optimization based on the original decoupling control.(2)The improvement of the traditional fish swarming method is to obtain a higher convergence accuracy and faster search speed.The improvement measures are as follows: changing the fixed step size and field of view into adaptive step size and field of view;adding a new default behavior,i.e."pending" behavior,to the clustering behavior of the fish;and introducing a population variation strategy in order to go beyond the local extremum search to the global search.Then the performance of the improved fish swarming algorithm is compared with the standard fish swarming algorithm by using the test function in MATLAB to verify the better effectiveness and rationality of the improved algorithm.(3)The plate shape and plate thickness comprehensive system is compared with different decoupling control tactics for decoupling control capabilities and antiinterference capability of recoupling.That is,traditional PID decoupling,DRNN-PID decoupling,DRNN neural network decoupling based on fish swarm algorithm optimization and DRNN neural network decoupling based on improved fish swarm algorithm optimization,these four decoupling methods are compared in simulation experiments.According to the simulation results,the DRNN neural network based on the improved fish swarm method has higher decoupling and stability performance,faster response time and better anti-interference ability than the other methods.