针对滚动轴承故障分析的特点 ,构造了基于 p-范数模糊推理神经网络 ,指出它可以对 Sugeno- Takagi模型进行逼近 ,因而更便于学习 ,并克服了单纯前向神经网络训练中容易陷入局部极小及收敛速度较慢的缺点。该神经网络应用于滚动轴承的四类故障诊断 ,与实验结果符合很好 ,取得了良好的故障诊断效果 Aiming at the characteristics of the fault analysis of rolling bearing, the fuzzy neural network based on p-norm is constructed and it is pointed out that it can approximate the Sugeno-Takagi model, which makes it easier to learn and overcomes the shortcoming of simple forward neural network And the slow convergence of the shortcomings. The neural network is applied to four kinds of fault diagnosis of rolling bearing, which is in good agreement with the experimental results and has achieved a good fault diagnosis effect