基于PSO-BP-PID神经网络的注塑机料筒温度预测算法研究

Temperature prediction algorithms of injection molding machine barrel based on PSO-BP-PID neural network

提出了一种基于粒子群优化(PSO)算法、BP神经网络及比例积分微分(PID)控制的复合算法的注塑机料筒温度预测模型,即PSO-BP-PID神经网络模型,并进行了仿真研究。结果表明:使用PSO算法确定该模型的输出权重,并且对混合核函数参数进行优化升级;在模型训练过程中,使用更大的容许度处理正误差,保证预测误差始终处于正值,使预测结果科学可靠;将高斯核函数与多项式核函数结合,生成一个新型混合核函数,提高核函数极限学习机性能;PSO-BP-PID神经网络模型的预测效果整体较传统PID模型好,温度总体趋势与实际预测数据相近,具有更好的拟合度。

A temperature prediction model of injection molding machine barrel based on particle swarm optimization(PSO)algorithm,back propagation(BP)neural network and proportional integral differential(PID)control,the PSO-BP-PID neural network model was proposed and simulated.The results show that PSO algorithm is used to determine the output weight of the model and optimize the parameters of the hybrid kernel function.In the process of model training,the positive error is handled with greater tolerance to maintain the positive prediction error,so that the prediction results are scientifically reliable.The Gaussian kernel function is combined with the polynomial kernel function to generate a new hybrid function to improve the performance of kernel extreme learning machine.The prediction made by PSO-BP-PID neural network model is more accurate than that by traditional PID model,whose overall temperature trend is similar to the actual prediction data with a better fit.