论文部分内容阅读
本文在分析了摩擦学系统特点以及在BP神经网络的结构的基础上 ,采用BP神经网络对摩擦学系统进行了建模和预测 ,通过对结果的分析 ,给出了这种神经网络在不改变隐层节点数和训练精度的条件下 ,来提高训练速度和拟合精度的一种新的方法 ,并指出了要保证BP网络能够近似任意的连续非线性函数 ,其训练数据应具备的条件。它对BP神经网络在处理大数值的应用中 ,具有较大的指导意义。
Based on the characteristics of tribological system and the structure of BP neural network, this paper used BP neural network to model and predict tribology system. Based on the analysis of the results, Hidden layer nodes and training accuracy to improve the training speed and the fitting accuracy of a new method, and to ensure that the BP network can be approximated to any continuous nonlinear function, the training data should have the conditions. It is of great guiding significance to the application of BP neural network in dealing with large numerical values.