基于改进BP网络的磨料水射流切割质量研究

The model analysis of abrasive water jet cutting quality by the improving BP neural network

目前建立的磨料水射流切割质量数学模型,对厚板材切割时,其加工精度和表面质量较难控制。基于BP神经网络学习理论,通过对网络权值和阀值进行四种方法的改进,目的是改善网络误差反向学习性能,建立低误差收敛精度、快训练速度的最佳切割质量模型。在获取大量样本数据的基础上,对最佳改进BP网络模型进行训练、预测,结果表明该模型能够快速、准确、可靠地预测磨料水射流切割质量。

The rapid development of computer technology, some mathematical model of abrasive water jet which was established has still not compared to neural network ,the precision machining and su（ace quality is difficult to control. By the BP neural network system,the four contrastive improvements methods are adopted to amend the weights and bias of BP network ,at last,to determine the best model,in which the lower error convergence precision and faster speed training is selected. Based on a large number of sample data,the best improved BP network model is trained,and forecast the cutting quality, the results show the rapid,accurate and reliable forecasting cutting quality,and good to meet production needs of factory.