基于熵-流特征和樽海鞘群优化支持向量机的故障诊断方法

针对旋转机械设备故障特征提取困难的问题,提出一种熵-流特征和樽海鞘群优化支持向量机(salp swarm optimization support vector machine,SSO-SVM)的故障诊断方法。利用改进多尺度加权排列熵(improved multiscale weighted permutation entropy,IMWPE)提取机械设备不同工况下的故障特征;采用监督等度规映射(S-Isomap)流形学习进行降维处理,获取低维的熵-流特征集;将熵-流特征输入至SSO-SVM多故障分类器进行识别与诊断。行星齿轮箱故障诊断实验分析结果表明:IMWPE+S-Isomap熵-流特征提取方法优于现有的多尺度排列熵(multiscale permutation entropy,MPE)、多尺度加权排列熵(multiscale weighted permutation entropy,MWPE)和IMWPE等熵值特征提取方法以及IMWPE+等度规映射(Isomap)和IMWPE+线性局部切空间排列(linear local tangent space alignment,LLTSA)等熵-流特征提取方法;樽海鞘群算法对支持向量机参数寻优效果优于粒子群、灰狼群、人工蜂群和蝙蝠群等算法;所提故障诊断方法识别精度达到100%,能够有效诊断出行星齿轮箱各工况类型。

Fault diagnosis method based on the entropy-manifold feature and SSO-SVM

Aiming at the difficulty of features extraction of rotating mechanical equipments,a fault diagnosis method based on the entropy-manifold feature and salp swarm optimization support vector machine(SSO-SVM)was proposed.First,the improved multiscale weighted permutation entropy(IMWPE)was utilized to extract fault features of the mechanical equipment under different working conditions.Then,the S-Isomap,as a manifold learning method,was employed to reduce the dimension,and obtain the low-dimensional entropy-manifold feature set.Finally,the entropy-manifold features were input into a SSO-SVM multi-fault classifier for identification and diagnosis.The experimental results of the planetary gearbox fault diagnosis show that the IMWPE+S-Isomap entropy-manifold feature extraction method is superior to the existing entropy-based feature extraction methods of multiscale permutation entropy(MPE),multiscale weighted permutation entropy(MWPE)and IMWPE.It is also more advantageous than the existing entropy-manifold feature extraction methods of IMWPE+isometric mapping(Isomap)and IMWPE+linear local tangent space alignment(LLTSA).The salp swarm algorithm is better than the particle swarm algorithm,gray wolf algorithm,artificial bee colony algorithm and bat algorithm for the optimization of support vector machine parameters.The proposed fault diagnosis method has a diagnostic accuracy of 100%,which can effectively examine the types of working conditions of planetary gearboxes.