基于Wavelet leader和优化的等距映射算法的回转支承自适应特征提取

Adaptive feature extraction method for slewing bearing based on Wavelet leader and optimized isometric mapping method

为了解决回转支承振动信号微弱,特征信息不易提取的问题,提出基于Wavelet leader方法和经混合灰狼算法优化的等距映射算法(HGWO-ISOMAP)的多分形自适应特征提取方法.利用Wavelet leader计算多分形特征,挖掘振动数据的几何结构信息,构造高维特征矩阵;通过HGWO优化后的ISOMAP算法对高维特征矩阵进行自适应特征筛选;将筛选后的特征矩阵输入到经遗传算法(GA)优化的最小二乘支持向量机(LSSVM)中进行故障状态识别.为了验证所提方法的优越性,采用课题组自主研发的回转支承综合性能试验台对某型号回转支承进行全寿命实验.结果表明,相比一般时域、时频域、频域特征提取方法,所提方法能提高识别精度,缩短计算时间,为回转支承特征提取提供新的有效途径.

Multi-fractal adaptive feature extraction method based on Wavelet leader method and isometric mapping method optimized by hybrid grey wolf optimization algorithm(HGWO-ISOMAP)was proposed,in order to solve the problem that the vibration signal of slewing bearing is weak and the feature information is difficult to extract.Wavelet leader is utilized to calculate the multi-fractal features,mine the geometric structure information of vibration data,and construct a high-dimensional multi-fractal feature matrix.Adaptive feature selection of high-dimensional feature matrix is carried out through ISOMAP method optimized by HGWO.The selected feature matrix is input into the least squares support vector machine(LSSVM)optimized by genetic algorithm(GA)for fault state identification.A full life experiment of a certain type of slewing bearing was carried out by using self-developed comprehensive performance test platform of slewing bearing,in order to verify the superiority of the proposed method.Results show that compared with general time domain,time-frequency domain and frequency domain feature extraction methods,the proposed method can improve the recognition accuracy and reduce the calculation time,providing a new effective way for feature extraction of slewing bearing.