基于PSO-BP神经网络的纱线质量预测

Combining the Particle Swarm Optimization with BP Neural Network for Yarn Quality Forecasting

针对复杂纺纱过程中成纱断裂强度难以预测的问题,提出一种基于粒子群优化算法(PSO)优化BP神经网络的成纱断裂强度预测方法.该方法采用PSO优化神经网络的权值和阈值,用来提高神经网络的收敛速度和获得全局最优解的能力.以纺纱车间大量现场质量检测数据为对象,进行预测验证,结果表明,PSO-BP神经网络在预测相关性(预测值与实际值的一致性程度)上与传统BP算法相比提高5.0%,与GA-BP算法相比提高4.6%,在预测精度上均要好于BP神经网络与GABP神经网络.

In view of the predication difficulty of yarn breaking strength in the process of complex spinning,aprediction method based on particle swarm optimization（PSO）is put forward to optimize the yarn breaking strength of BP neural network.This method uses the PSO to optimize weights and threshold of neural network,improve convergence speed of neural network and obtain the ability of global optimal solution.A lot of field quality testing data in spinning workshop is taken as the objects to conduct the prediction verification.The results show that the prediction correlation（the degree of consistency between the actual values and estimated values）of PSO-BP neural network has been improved by 5.0% and 4.6%respectively compared with the traditional BP algorithm and GA-BP algorithm.Besides,it has higher prediction precision than BP neural network and GA-BP neural network.