基于PSO-SVDD的齿轮箱故障诊断

Gearbox Fault Diagnosis Based on PSO-SVDD Algorithm

针对某热源厂罗茨风机齿轮箱故障诊断中诊断技术难度大,准确率不高的情况。现提出将支持向量数据描述(Support Vector Data Description,SVDD)和粒子群优化算法(Particle Swarm optimization,PSO)相结合来进行齿轮箱设备的故障诊断。分别利用数据采集器采集到的罗茨风机齿轮箱正常状态数据以及特定故障数据构建经PSO算法优化后的SVDD最小超球体模型,用建立好的模型对测试数据进行故障诊断,使诊断结果更加准确。工业应用结果表明:该方法可以有效地处理罗茨风机故障诊断难度大、准确率低的问题,能较好识别已知故障并做出报警处理。

In view of the situation that the diagnosis technology is difficult and the accuracy is not high in the fault diagnosis of the Roots fan gearbox in a heat source factory.Now it is proposed to combine Support Vector Data Description(SVDD)and Particle Swarm optimization(PSO)to diagnose faults of gearbox equipment.The normal state data and specific fault data of the Roots fan gearbox collected by the data collector are used to construct the SVDD minimum super sphere model optimized by the PSO algorithm,and the established model is used to diagnose the test data to make the diagnosis results more accurate.The industrial application results show that this method can effectively deal with the problems of difficulty in diagnosis of roots blower and low accuracy,and can better identify known faults and make alarms.