论文部分内容阅读
建立了钢丝绳断丝定量识别的BP神经网络模型,重点从网络输入特征值的分析与优化、网络训练集与测试集的合理选择、网络训练目标的确定3个方面讨论了优化神经网络参数与性能的方法。经实际网络的训练及测试,证明了合理参数的选择改进了网络性能,提高了钢丝绳断丝定量识别的精度,具有实际工程意义。
The BP neural network model for quantitative identification of broken wire rope is established. The key points are the analysis and optimization of eigenvalue of network input, the reasonable selection of network training set and test set, and the determination of network training goal. The parameters and performance of neural network Methods. The actual network training and testing proves that the selection of reasonable parameters improves the network performance and improves the accuracy of quantitative identification of wire rope broken wire, which has practical engineering significance.