Research On Online Condition Monitoring Based Prognostics Of Key Parts And Components Of Mechnicanic Equipment

With the development of the modern industry, the relationship between mechanical equipment and human becomes more and more close. Sophisticated and complicated machineries bring our life much more convenience, but at the same time, this situation is a great challenge for prognostics and health management of the equipment. Equipment failure will affect the normal work of the whole mechanical system directly, even cause malignant events.The study of residual life is one of the most important part of prognostics and health management in the key parts and components of the mechanical equipment. Based on online condition monitoring, life study of the key parts and components of the mechanical equipment, the main contents of this thesis as follows:(1) The failure theories of the mechanical parts and components are provided firstly. The relationship between output parameters and damage, the characteristics and growth of fault are discussed. Rolling bearing and gear, which are two of the key parts of a mechanical equipment, are studied as an example both in theoretical and practical. And the fault signals of rolling bearing and gear are studied respectively.(2) Based on wavelet de-noising, feature extraction and multi-feature fusion, a method to track the degradation index of the key parts and components was proposed. The wavelet de-noising method used to keep the noises from condition monitoring signal. Then, multi-feature extracted from time domain, frequency domain and time-frequency domain. And a principal component analysis method was used to multi-feature fusion. An example of condition monitoring based data sets of rolling bearing was provided.(3) The degradation trend prediction and life prediction were studied in this part. For trend prediction, a model based on cumulative sum, particle swarm algorithm and least squares support vector regression was proposed. Cumulative sum used to control the errors, particle swarm algorithm used to optimize the parameters of least squares support vector regression. An example of condition monitoring based data sets of rolling bearing was provided. For life prediction, based on weight application to failure times theory, a multi-model prediction was proposed. Condition monitoring data sets of four rolling bearings was discussed for life prediction.(4) Based on MATLAB, a graphical user interface of trend prediction and life prediction was developed. And a practical example was discussed.