基于CEEMDAN-IGWO-SVM的轴承故障诊断研究

为了提高支持向量机(SVM)在轴承故障诊断时的准确率和识别效率,提出了一种基于具有自适应白噪声的完整集成经验模态分解方法(CEEMDAN)、改进灰狼优化算法(IGWO)和支持向量机(SVM)相结合的故障诊断方法。首先用CEEMDAN与Shannon熵对振动信号消噪、分解,获得典型故障的敏感信号;其次,将粒子群算法(PSO)惯性权重w与粒子“飞行”速度v引入灰狼优化算法(GWO),得到IGWO,通过IGWO算法优化SVM得到诊断模型的最优参数,增强SVM的学习能力和泛化能力;最后,利用美国西储大学的轴承试验数据验证优化模型的有效性。结果表明,IGWO算法优化SVM的模型可以准确、高效地对轴承进行故障诊断;与GA、PSO、和GWO算法优化的SVM模型相比,该方法的故障诊断准确率和识别效率更高。

Research on Bearing Fault Diagnosis Based on CEEMDAN-IGWO-SVM

In order to improve the accuracy and recognition efficiency of support vector machine(SVM)in bearing fault diagnosis,a complete integrated empirical mode decomposition method(CEEMDAN)with adaptive white noise and improved grey wolf optimization algorithm(IGWO)are proposed.A fault diagnosis method combined with a support vector machine(SVM).Firstly,CEEMDAN and Shannon entropy are used to denoise and decompose the vibration signal to obtain the sensitive signal of typical fault.Secondly,the particle swarm optimization(PSO)inertia weight w and the particle"flight"velocity v are introduced into the grey wolf optimization algorithm(GWO).IGWO optimizes SVM by IGWO algorithm to obtain the optimal parameters of the diagnostic model,and enhances the learning ability and generalization ability of SVM.Finally,the effectiveness of the optimization model is verified by the bearing test data of the Western Reserve University.The results show that the IGWO algorithm optimizes the SVM model to diagnose the bearing accurately and efficiently.Compared with the SVM model optimized by GA,PSO and GWO algorithm,the fault diagnosis accuracy and recognition efficiency of the method are higher.