基于多特征提取与IPSO＿LSSVM的故障诊断

Multi-feature Extraction and Improved PSO Optimization LSSVM Fault Diagnosis

针对滚动轴承运行过程中故障难以识别的问题,提出一种多特征提取与改进粒子群算法(improved particle swarm optimization,简称IPSO)优化最小二乘支持向量机的诊断方法。首先用小波包变换对振动信号消噪、分解,提取频域特征;然后结合时域特征等参数,用核主元分析法对多维特征空间进行优选和降维,获得典型故障的敏感信号;最后用改进的粒子群优化最小二乘支持向量机的核参数和惩罚因子解决在寻优中陷入边缘局部最优、收敛精度差的问题,提升故障诊断的识别率。实验结果表明:该方法有效提取故障特征,提高了故障识别的准确率和实时性,是一种可靠的轴承故障诊断方法。

Aiming at the problem that the fault is difficult to identify during the running of rolling bearing,a multiple feature extraction and IPSO LSSVM is proposed.Firstly,the WPT is used to denoise and decompose the vibration signal,and the frequency domain features are extracted.Then,with the parameters of time domain and average frequency,the kernel principal component analysis method is used to optimize and reduce the dimensionality of the multidimensional feature space to obtain the sensitive signal of typical faults.Finally,the kernel parameters and penalty factors of the IPSO LSSVM are used to solve the problem of local optimization and poor convergence precision in the optimization,and improve the recognition rate of fault diagnosis.The experimental results show that the proposed method effectively extracts fault characteristics and improves the accuracy and real-time performance of fault identification.It is a reliable method for fault diagnosis of bearing faults.