Research On Intelligent Optimization Of Rolling Force Energy Model For Ultra-heavy Plate

As an important steel product,ultra-heavy plate is widely used in construction engineering,mechanical manufacturing,marine manufacturing and other aspects of national economic construction.In order to solve the common problems caused by insufficient compression ratio with continuous casting slab,such as loose internal structure and coarse grain,variable gauge rolling technology is proposed to produce ultra-heavy plate without changing the existing rolling equipment.Based on the theory and technology of variable gauge rolling,this paper studies the force energy model of variable gauge rolling and its intelligent optimization method.The main contents of this paper are as follows:(1)For the trainsient bitting process,based on the main drive system of one 5000mm heavy plate rolling mill,the main drive system was simplified to a 5-DOF "mass spring" system by the method of centralized mass,and the dynamic model and torsional vibration mathematical model of the main drive systemwere established.The fourth order Runge Kutta method was used to solve the differential equation of torsional vibration and the dynamic response simulation curve were drawn.The influence on the torsional vibration of the main drive system was analyzed,such as whether there is a clearance,the size of clearance,the biting time and other factors.According to the simulation results,the prediction model of bitting peak torque was established.The maximum error of the model prediction is within ±12%,and the average error is within ±10%;(2)Based on the analysis of the process characteristics of thinning rolling and flattening rolling in the variable gauge rolling process,the theory and regression calculation model of variable gauge rolling were studied.Based on the basic theory of geometry and rolling mechanics,the parameters of variable gauge rolling were deduced.Combined with the actual data and the characteristics of variable gauge rolling process,it was found that the torque arm coefficient is mainly related to the geometry coefficient of deformation zone and the reduction rate.The torque arm coefficient model and rolling force energy model which more suitable for the ultra-heavy plate rolling were established.The maximum error of the torque model is within±12.3%and the average error is within ±10%;(3)Combining the BP neural network improved by particle swarm optimization algorithm with the prediction model of bitting peak torque,the influence factors of bitting peak torque were analyzed according to the rolling process characteristics of ultra-heavy plate in the transient biting process,and the neural network structure was established and optimized based on the measured actual data.PSO-BP network algorithm was used to intelligently optimize the bitting torque model,and a high-precision prediction modelwas established.After optimization,the maximum error of prediction of bitting peak torque is reduced from 12%to 9.20%,and the average error is reduced from 9.36%to 5.27%;(4)PSO-BP neural network algorithm was used to intelligently optimize the torque model in steady-state stage,and a high-precision torque model in steady-state stage was established.According to the characteristics of variable gauge rolling process,the influence factors of torque in the steady-state rolling stage were analyzed.Based on the practical data,the network structure were established and optimized.After optimization,the maximum error of the torque model for steady rolling stage is reduced from 12.30%to 8.21%,and the average error is reduced from 8.27%to 4.26%.