参数优化的IZOA-SVM机械设备故障诊断方法

IZOA-SVM algorithm with parameter optimization for mechanical equipment fault diagnosis

在复杂的工作环境中,机械设备振动信号的复杂性常常会导致机械设备故障诊断的准确性不高,为解决设备运行中因信号复杂性引发的故障诊断难题,提出了一种参数优化的斑马优化算法优化支持向量机(IZOA-SVM)的故障诊断方法。首先,引入了柯西变异和反向学习的改进策略到斑马优化算法(ZOA)中,提出了改进的斑马优化算法(IZOA),旨在改善原有斑马优化算法在迭代后期容易陷入局部极值等问题,从而有效增强了其全局搜索能力;其次,利用IZOA优化支持向量机(SVM)的核参数g和惩罚参数c以寻找SVM最优参数组合[c,g],并构建了IZOA-SVM模型;然后,计算了样本的13个时域特征以构成特征向量,并将特征向量分别输入到IZOA-SVM模型、斑马优化算法优化支持向量机(ZOA-SVM)模型、粒子群算法优化支持向量机(PSO-SVM)模型、遗传算法优化支持向量机(GA-SVM)模型和支持向量机模型,进行了故障分类;最后,通过旋转机械振动及故障模拟试验验证了该方法的有效性。研究结果表明:IZOA-SVM模型在分类准确率方面得到了明显的提高,达到了98.33%;该模型能够精准而稳定地识别故障类型,提高故障识别的准确性,在准确率方面相较于其他对比方法表现出更为显著的优势。因此,该方法在全局搜索和故障分类准确性方面都取得了明显的改进,为复杂环境下的故障诊断提供了可参考的解决方案。

In complex working environment,the complexity of vibration signals of mechanical equipment often leads to low accuracy of fault diagnosis.In order to solve the problem of fault diagnosis caused by signal complexity in equipment operation,a fault diagnosis method of support vector machine optimized by zebra optimization algorithm(IZOA-SVM)with parameter optimization was proposed.Firstly,the improved strategies of Cauchy variation and reverse learning were introduced into the zebra optimization algorithm(ZOA),and the improved zebra optimization algorithm(IZOA)was proposed,it aimed to improve the local extreme value problem of the original zebra optimization algorithm in the late iteration,so as to effectively enhance its global search capability.Secondly,the kernel parameter g and penalty parameter c of support vector machine(SVM)were optimized by IZOA to find the optimal combination of SVM parameters[c,g],and the IZOA-SVM model was constructed.Then,13-time domain features of the sample were calculated to form the feature vector.The eigenvectors were input to IZOA-SVM model,support vector machine optimized by zebra optimization algorithm(ZOA-SVM)model,particle swarm optimization support vector machine(PSO-SVM)model,support vector machine optimized by genetic algorithm(GA-SVM)model and support vector machine model(SVM)for fault classification.Finally,the effectiveness of the method was verified by the vibration and fault simulation tests of rotating machinery.The results show that the fault accuracy of IZOA-SVM model is 98.33%,which is improved in different degrees comparing with other models.Therefore,the method has achieved obvious improvement in the accuracy of global search and fault classification,and provides a reference solution for fault diagnosis in complex working environments.