摘要
将BP神经网络建模技术和遗传算法(GA)应用于套料钻性能预测。利用BP神经网络建立套料钻性能预测模型,通过比较实际误差梯度值与给定误差梯度值,来确定BP网络是否处于局部极小状态;GA仅在BP网络处于局部极小时进行学习,对BP网络的连接权值进行优化。BP神经网络和GA两者的有效结合可以解决BP算法固有的缺陷,如收敛速度慢、易陷入局部极小等。套料钻加工性能试验结果表明预测结果和实际结果吻合程度较好,验证了GA-BP网络模型在套料钻性能预测中的有效性和准确性。
A new algorithm for the performance prediction of the diamond core-drill was presented. The development of the algorithm is based on BP neural networks and GA algorithm. The presented GA-BP algorithm uses BP neural networks to generate GA-BP model. By computing error gradient and checking it with given error, local minimum position is identified. At this time, if BP neural network is in local minimum position, GA algorithm is used to optimize the BP neural networks＇ weight. The main advantage of the presented algorithm is the fact that it can solve disadvantages of BP neural networks, such as low rate of convergence, easily falling into local minimum point and so on. The effectiveness and accuracy of the algorithm is demonstrated by the performance prediction experiments of the diamond core-drill.
出处
《兵工学报》
EI
CAS
CSCD
北大核心
2006年第3期494-497,共4页
Acta Armamentarii
关键词
计算机科学技术基础学科
机械制造自动化
遗传算法
BP神经网络
套料钻
性能预测
basic disciplines of computer science and technology
mechanical manufacturing ＆ automation
GA
BP neural network
core-drill
performance prediction
