In order to further control the bending force accuracy of tandem cold rolling,an improved PSO-SVM prediction model was constructed.Particle swarm optimization(PSO)with compression factor is used to optimize the parameters of support vector machine(SVM)more efficiently,and the roll bending force model is obtained by reverse normalization of regression rolling parameters.The verification process of the prediction model is completed according to the actual rolling results on site.The results show that the im-proved PSO-SVM model has the lowest prediction performance index among the above optimization methods,and the improved PSO-SVM model has the best prediction effect.The installation of reliable compensation greatly reduces the workload of the AFC system and promotes a significant increase in the efficiency of the strip shape.The bending force compensation value is almost the same as the bending force,which has ex-cellent prediction performance.
Cold rollingShape controlBending roll forceParticle swarm optimizationSupport vector ma-chine
NETL
NSTL
万方数据
