动量-自适应学习BP算法在加工领域的探索

Research on Momentum-Adaptive Learning BP Algorithm in the Field of Machining

为解决实际加工中试验次数多生产成本高、选取加工参数困难等问题,采用动量-自适应学习BP算法构建BP神经网络预测模型。根据实际情况将典型BP算法改进,得到收敛速度快的动量-自适应学习BP算法模型;用电解加工试验数据对模型结构进行训练,最终建立动量-自适应学习BP神经网络加工预测模型。采用该模型对不同加工参数组合下加工的不锈钢微孔孔径大小进行预测。结果表明,该模型的预测误差低于5%,具有很强的预测能力。

In order to solve the problems of many experiments in actual processing,high production costs,difficulties in selecting processing parameters,and so on,a BP neural network prediction model is constructed using momentum-adaptive learning BP algorithm. According to the actual situation,the typical BP algorithm is improved,and the momentum-adaptive learning BP algorithm model with fast convergence rate is obtained. The model structure is trained with the electrochemical machining experiments data. Finally,the momentum-adaptive learning BP neural network processing prediction model is established.This model is used to predict the pore size of stainless steel micro-holes processed under different processing parameters. The results showthat the prediction error of the model is less than 5%,and it has a high prediction ability.