基于FWA-DBN的航空发电机偏心故障诊断

针对具有多并联支路绕组结构的航空发电机在偏心故障下的输出三相电压、电流故障特征差异小, 造成故障不易识别的问题, 提出一种基于烟花算法(fireworks algorithm, FWA)优化深度置信网络(deep belief network, DBN)的故障诊断方法。首先根据有限元法搭建航空发电机模型, 通过仿真获取不同静态、动态偏心故障输出数据; 然后运用FWA训练优化与极限学习机(extreme learning machine, ELM)相结合的DBN网络, 得到最佳DBN-ELM模型结构; 最后由ELM分类器进行故障诊断分类。诊断结果表明, 相较于传统的故障诊断方法, 应用所提方法进行航空发电机偏心故障诊断, 可以获得更高的准确率, 平均准确率达到99.203%。

Aviation generator eccentricity fault diagnosis based on FWA-DBN

A fault diagnosis method based on deep belief network (DBN) optimized by the fireworks algorithm (FWA) is proposed for the problem of small differences in the characteristics of three-phase voltage and current under eccentricity faults in aviation generators with multi-parallel branch winding structures, which are not easy to classify. Firstly, a finite element model of the aviation generator is built and different static and dynamic eccentricity fault data are obtained by simulation. Then the DBN network combining with extreme learning machine (ELM) is optimized by FWA and the optimal DBN-ELM network structure can be obtained. Finally, the fault diagnosis is performed by the ELM classifier. The diagnosis results show that the application of the proposed method for aviation generator eccentricity diagnosis can obtain a higher accuracy with an average accuracy of 99.203% compared with the traditional fault diagnosis methods.