基于神经网络的多级行星齿轮箱故障诊断专家系统

Fault diagnosis expert system for multilevel planetary gear boxes based on neural networks

多级行星齿轮传动比大、结构复杂,按照传统的事后检修、计划检修难以满足实际生产需要.因此采用基于人工神经网络的故障诊断专家系统来实现多级行星齿轮增/减速器的不解体故障诊断.根据多级行星齿轮的初始条件,得出齿轮箱的各轴端的特征频率,分析了齿轮箱的各种常见故障.将专家系统与神经网络结合,采用产生式规则表示知识的方法,运用基于模型的推理方法构建专家系统的知识库和推理机,通过人工神经网络的样本分析,改进了专家系统的学习和推理功能,并提出了1种能有效解决多级行星齿轮增/减速器各种故障的诊断方法.

Due to large transmission ratio and complicated structure of multilevel planetary gear boxes, such traditional measurements as post- and planned-maintenance could not meet the practical demands. By applying the fault diagnosis expert system via artificial neural networks, the non-destructive fault diagnosis is conducted for the multilevel planetary gear increaser and reducer. According to the initial conditions, the feature frequencies are detected regarding gear shaft ends, whereas the frequent faults are analyzed on gear boxes. By combining the expert system with artificial neural networks, the knowledge is represented on the basis of rules generated. In this manner, the knowledge based, together with a reasoning mechanism, is developed via the model-based reasoning. Through analyzing the samples using the artificial neural networks, the learning and reasoning functionalities are enhanced and proven effective in terms of fault diagnosis on multilevel planetary gear increasers and reducers.