【摘要】 针对BP神经网络存在局部极小值和收敛速度慢等问题,提出了一种resilient backpropagation(RPROP)的改进BP网络。RPROP神经网络具有优良的非线性映射能力,可以很好地描述频率特征和诊断结果之间的关系,经改进算法训练的网络适合旋转机械故障诊断。更多还原

【Abstract】 A kind of BP neural network based on resilient backpropagation(RPROP)optimization algorithm is introduced in detail to overcome the disadvantages of standard BP algorithm.RPORP neural network is effective for dealing with non-linear mapping which can satisfactorily describe the non-linear relations between frequency character and diagnosis results.The improved BP neural network is suitable for diagnosis of rotating mechanical fault.更多还原