【摘要】 基于人工神经网络所体现出的良好的非线性逼近特性,运用遗传算法优化BP网络的权值和阈值,避免BP网络陷入局部极小点,同时通过灰色关联分析优化网络的结构,使网络具有更好的泛化性和收敛速度。运用改进的BP网络发掘油管接箍加工中各个加工参数对最终加工圆度误差的潜在关系,从而实现对油管接箍加工圆度误差的预测。结果表明改进型BP神经网络具有较快的收敛速度和较好的泛化性能够准确预测油管接箍加工圆度误差。更多还原

【Abstract】 Neural network has good non-linear approach characteristic.The weight value and the threshold value of BP neural network improved by genetic algorithm,so the BP neural network can prevent itself from getting into local minimum point,and the use of grey correlation analysis in optimizing the structure of network can improve generalization performance and increase convergence speed.The improved network can be used in the prediction of the roundness error of tubing coupling by digging the hidden relation between processing parameters and roundness error.The results showed that the improved BP neural network has fast convergence rate,strong generalization capability and,accurately predicting the roundness error of tubing coupling.更多还原