论文部分内容阅读
针对ＢＰ网络用于高速旋转机械的故障诊断时学习收敛速度慢和易出现局部最小点的不足，采用自适应学习率和绝对误差等距离逼近方法，对其参数进行了研究。结果表明，这两种方法有效地提高了ＢＰ网络的收敛速度，并给出了自适应学习率、慢性因子、隐层单元数的合取值范围和推荐值。
Aiming at the problems of slow learning speed and prone to local minimum point in the fault diagnosis of high speed rotating machinery, BP network is studied by using distance learning method such as adaptive learning rate and absolute error. The results show that these two methods can effectively improve the convergence speed of BP network and give the range of combined values and recommended values of adaptive learning rate, chronic factor, hidden layer unit number.