周志朋 谢冬梅



摘 要:针对风电齿轮箱的复合故障诊断问题,本文提出了基于改进变分模态的风电齿轮箱复合故障的特征提取方法,并采用改进鲸鱼算法优化的最小二乘支持向量机(DEWOA-LSSVM)故障诊断模型进行故障诊断。通过变分模态方法将收集的信号分解为K个模态分量,通过加权排序熵对其进行量化处理,得到特征向量,将特征向量输入到经过改进鲸鱼算法优化后的最小二乘支持向量机模型完成故障诊断,并通过试验验证该方法的有效性。
关键词:故障诊断;变分模态分解;加权排序熵;鲸鱼算法;最小二乘支持向量机
中图分类号:TH165.3文献标识码:A文章编号:1003-5168(2020)17-0072-04
Research on Compound Fault Diagnosis Method of Wind Power Gearbox
ZHOU Zhipeng1 XIE Dongmei2
(1. Graduate Department of Shenyang Institute of Engineering,Shenyang Liaoning 110136;2. School of Electric Power, Shenyang Institute of Engineering,Shenyang Liaoning 110136)
Abstract: Aiming at the problem of compound fault diagnosis of wind power gearboxes, this paper proposed a feature extraction method for compound faults of wind power gearboxes based on improved variational modes, and used DEWOA-LSSVM fault diagnosis model for fault diagnosis. The variational modal method was used to decompose the collected signal into K modal components, the weighted ranking entropy was used to quantize it to obtain the feature vector, the feature vector was input to the least square support vector machine model optimized by the improved whale algorithm to complete the fault diagnosis, and the effectiveness of the method was verified through experiments.
Keywords: fault diagnosis;variational modal decomposition;weighted ranking entropy;whale algorithm;least squares support vector machine
齒轮箱是风力发电机组的核心部件,如果齿轮箱发生故障,容易造成内部结构的损害,人们需要停机进行维修,而且故障维修时间较长,工作量较大。这对风力发电系统工作的稳定和安全性来说也是一种很大的威胁,严重时会造成事故和财产损失。
风电机组齿轮箱在实际应用过程中存在多种故障状态,在实际工作中也可能会存在多种工况,而在实际工况中对风电齿轮箱的故障进行检测时,风电齿轮箱噪声和外界环境噪声会影响齿轮箱的正常运行[1]。所以,风电齿轮箱的故障应该是非线性的,这些噪声都会对风电齿轮箱的故障诊断产生误判,因此人们需要对故障信号进行处理。目前,风电齿轮箱故障信号的处理主要集中在时频域分析,主要有小波包分解、傅里叶变化、经验模态分解等时频分析算法[2]。……