对机械工业中常用的评价滚动轴承故障的冲击脉冲法，提出了一种基于神经网络的故障评价系数自修正策略。这一修正系数模块已用于工业现场滚动轴承状态在线监测系统中，经长期运行的实践证明，自修正系数模块的引入极大地提高了滚动轴承运行状态判断的准确性。 For the common impulse pulse method of evaluating the fault of rolling bearing in the machinery industry, a self-correction strategy of fault evaluation coefficient based on neural network is proposed. The correction coefficient module has been used in the industrial on-line rolling bearing state on-line monitoring system. The long-running practice proves that the introduction of the self-correction coefficient module greatly improves the accuracy of judging the running state of the rolling bearing.