Intelligent Fault Diagnosis Of Rotating Machinery Based On Cluster Analysis

As an important part of modern mechanical equipment,rotating machinery has crucial significance for material production.With the development of science and technology and the improvement of production demand,machanical equipment are developing towards integration,intelligence and precision.At the same time,it also greatly increases the challenge of equipment and diagnostic maintenance.Intelligent diagnosis technology also rises.In this context,based on cluster analysis,an intelligent fault diagnosis model including signal processing and feature extraction and selection is constructed.The method is checked by the application of laboratory equipment to collect bearing and gear working condition data.Specific studies are as follows:Firstly,the research status of rotating machinery fault diagnosis and cluster analysis is briefly introduced.It also indicates the broad application prospect of clustering analysis in fault diagnosis.Secondly,after briefly introducing the related theories and concepts of clustering analysis technology.Three different clustering methods are chosen to construct and introduce the intelligent fault diagnosis model.Including affine propagation(AP)clustering,improved fuzzy C means clustering and GG clustering.These three clustering algorithms have unique advantages,disadvantages and application range.Such as when the original vibration data contains a lot of noise and unstable signals,GG clustering has better diagnostic effect than the improved fuzzy C mean clustering used in this paper.Thirdly,three intelligent fault diagnosis methods proposed in this paper are introduced in detail.From the time domain,frequency domain and energy domain of the signal,the multi-domain features are extracted and input into the self-weight algorithm to select the sensitive features.In the third chapter,we select the optimal feature and eliminate the redundant feature by affine propagation clustering after selecting the sticky feature using self-weight algorithm.The selected features were input into the pattern recognition method after dimensionality reduction using principal component analysis(PCA)to obtain diagnostic results.At the same time,the benefits of the proposed intelligent fault diagnosis method are shown by comparing with other fault diagnosis methods.Finally,for three distinct intelligent fault diagnosis methods,bearing and gear fault classification data are used to verify its effectiveness and robustness.Bearing data used inExample 1 of Chapter 3 are derived from Case Western Reserve University.The data used in the other five examples are collected from the automated fault comprehensive simulation experimental platform(MFS-MG)and the wind turbine motor drive fault comprehensive test bench(WTDS).For the three experiments collected on the MFS-MG,at the second and the first time,bearing fault classification data are all selected,but at the second time,the data acquisition under various working conditions including rotor imbalance,friction and composite fault,and whether there is loading or not.The third data acquisition in this platform is mainly gear fault classification data.The consequences of fault diagnosis of bearing and gear by three intelligent fault diagnosis methods show that the better diagnosis effect can be achieved with less artificial dependence and prior knowledge.At the same time,three intelligent fault diagnosis methods have altered diagnostic effects for different types of signals.It also provides new ideas and methods for diagnosing different rotating machinery faults.