【摘要】 为了解决传统的板形识别与控制中的识别精度低,控制速度慢等问题,将粒子群优化(particle swarm optimization,PSO)算法和误差反传递(back propagation,BP)算法混合训练的PSO-BP网络引入到板形的识别与控制中.首先根据板形轧制的历史数据,建立预测板形的神经网络,得到反映板形控制手段对板形特征参数影响的效应矩阵,同时根据理论数据建立对板形进行模式识别的神经网络.这些都是离线进行的,而且对一批板材只需训练一次神经网络,在线轧制过程中只需要根据识别网络的识别结果和效应矩阵,便可以很快的得到需要的控制量.这种方法可以简化板形控制过程,提高控制速度,最后的仿真实验进一步说明了这种方法的有效性.更多还原

【Abstract】 In order to solve the problems of low-precision and slow control of the traditional algorithms in the pattern recognition and control of flatness,the neural network trained by hybrid algorithms of particle swarm optimization(PSO) and back propagation(BP)is introduced.According to the rolling data in history,the PSO-BP network for predicting flatness is trained.As a result,the effective matrix reflecting the effects of adjustable parameters on the eigen-parameters of flatness is obtained.At the same time,the network for recognizing flatness is trained based on theoretical data.The networks are trained only once for a batch of strips.And the corresponding adjustments of parameters can be quickly calculated on line based on the effective matrix.Therefore,this approach can simplify and speed up the control of flatness. Finally,its effectiveness is proved by the given case study.更多还原