论文部分内容阅读
采用实验与神经网络预测相结合的方法,对基于温度控制的激光相变硬化工艺参数进行了研究。首先,使用基于温度可控的大功率半导体直接输出激光加工系统对45~#钢进行设定温度下的激光相变硬化实验。然后,通过改进的BP神经网络算法构建神经网络模型,并采用所获得的实验样本数据训练该人工神经网络模型。模型中所采用的改进BP神经网络算法是遗传算法和基于新型误差函数的批量训练神经网络算法相结合的混合算法。为验证改进算法的性能,在Lab Windows/CVI软件上应用C编程语言实现了该算法。通过运行程序发现,采用此算法的收敛速度比传统算法提高了约80%,预测输出的指标与实际值之间的偏差小于4%。
Based on the combination of experiments and neural network prediction, the parameters of laser phase-change hardening based on temperature control were studied. First of all, using the direct output laser processing system based on temperature controllable high power semiconductors, the laser phase-change hardening experiment of 45 ~ # steel was carried out at the set temperature. Then, the neural network model is constructed by the improved BP neural network algorithm, and the artificial neural network model is trained by using the experimental sample data obtained. The improved BP neural network algorithm used in the model is a hybrid algorithm which combines genetic algorithm with batch training neural network algorithm based on new error function. In order to verify the performance of the improved algorithm, this algorithm is implemented in Lab Windows / CVI using C programming language. By running the program, it is found that the convergence speed of this algorithm is improved by about 80% compared with the traditional algorithm, and the deviation between the predicted output index and the actual value is less than 4%.