Fault Diagnosis Of Rolling Bearing And Epicyclic Gearbox Based On Data And Multi-source Information Fusion Technology

With the development of science and technology in China,the design of large machinery and systems is becoming more precise and complex.In order to reduce the loss on mechanical equipment failure,implement effective mechanical equipment fault diagnosis technology and improve the reliability,this paper focuses on the key parts of mechanical equipment(rolling bearing and epicyclic gearbox),and mainly studies the signal analysis processing technology,data-driven fault diagnosis technology and multisource information fusion technology,etc.The main research work of the thesis mainly consists of the following parts.Firstly,this thesis focuses on the fault signal analysis and processing of rolling bearing and epicyclic gearbox.In this thesis,the failure mechanism of rolling bearing and epicyclic gearbox is studied.Because the original fault signal has some characteristics of redundancy and uncertainty,so the time-frequency domain analysis is used to extract the features of the original signal.The original signal is analyzed from various angles through different morphologies.The rough set theory is adopted for feature screening of multiple fault features obtained from time-frequency analysis to retain fault features that have a greater impact on diagnosis results,so as to simplify the fault diagnosis system model and improve the efficiency and accuracy of fault diagnosis.At the same time,this thesis carries out damage fault simulation of rolling bearing and epicyclic gearbox to different degrees,and collects system signals by simulating the operation state under typical working conditions,so as to provide data support for the following fault diagnosis method research.Then an information fusion method is adopted in this thesis based on the optimal DS evidence theory.Combining the cloud model with the characteristics of uncertainty,randomness and fuzziness,the cloud model and the D-S evidence theory are combined to reduce the high degree of conflict between evidences and achieve effective information fusion diagnosis.The effectiveness and feasibility of the method are verified by the test data of rolling bearing and epicyclic gearbox.Finally,this thesis focuses on the improvement analysis of support vector machines,and proposes a dual optimization SVM fault diagnosis structure model.A dual optimization structure model is used.First,because of the randomness of the cloud model,the cloud model is used to optimize the genetic algorithm(CM-GA),so as to obtain a faster search process and more effective optimization results.A structural model of support vector machine optimization based on cloud genetic algorithm is proposed.This model uses the combination of preoptimization and optimization to greatly reduce the training time,so the performance of the algorithm can be improved effectively.Finally,the rolling bearing data from Case Western Reserve University is studied theoretically,and then the fault data of rolling bearing and epicyclic gearbox collected by the actual platform is used for experimental analysis.The traditional method is compared with the fault diagnosis method based on cloud genetic algorithm.It shows that the method proposed in this paper can achieve higher accuracy in a shorter time.