光纤传感技术结合SVM-ELM的风力机齿轮箱故障诊断分析

Fault Diagnosis and Analysis of Wind Turbine Gearbox Based on Optical Fiber Sensing Technology and SVM-ELM

针对风力机组齿轮箱故障信号采集、特征提取和故障诊断方面所存在的问题,提出了一种光纤布拉格光栅检测技术与故障诊断算法相结合的健康状态评估方法。首先通过对光纤动态监测系统构架进行设计并规划故障健康状态评估的整体流程进而实现实验采集平台的搭建,其次通过对采集信号进行变分模态分解和多尺度排列熵算法的分析并完成特征向量集的构建,最后通过支持向量机、概率神经网络、极限学习机算法进行测试正确率和时间的对比分析。结果表明,该方法能够准确实现故障模式的分类且提高了故障预测概率。

For wind turbine gearbox fault signal acquisition,feature extraction and fault diagnosis of problems,a health status assessment method combining fiber Bragg grating detection technology and fault algorithm was proposed.First of all,the dy-namic monitoring system based on optical fiber architecture design and planning failure state of health assessment of overall pro-cess to realize the construction of the experimental platform,secondly by the acquisition signal variation mode decomposition and multi-scale analysis of permutation entropy algorithm and completed the construction of characteristic vectorset,finally,support vector machine(SVM),probabilistic neural network(PNN),limit the comparison and analysis of machine learning algorithm for testing accuracy and time.Results show that the method can achieve accurate classification of failure mode and improve the prob-ability of failure prediction.