Research On System Reliability Growth Prediction Theory And Method

The reliability growth theory as well as the reliability prediction theory have been become hot topics receiving much concern in reliability engineering, widely used in statistical theory, information theory, failure physics, system engineering, mechanical engineering, computer science and technology and other subjects at present. Furthermore, it's an urgent research task to guarantee product reliability and scientifically predict product future reliability utilizing current information during the research and development stage. Therefore, a new proposition of system reliability growth prediction theory and method study was proposed based on random process theory, information fusion technology, and small sample statistical theory, through integrating prediction theory into reliability growth technology, and taking the study of major equipment reliability as a breakthrough point. Then, reliability growth prediction technology during the trial operation phase, system reliability synthesis and prediction method, and related prediction technology in the operation period for wind turbine were further discussed combined with the engineering background. The achievements not only open up new methods to study wind turbine reliability, but also become demonstration for other major equipment reliability study.The main achievements of this PhD work were listed as follows:1. Research status and development trends of reliability growth technology and reliability prediction theory at home and abroad were systematically summarized, and research on frontier and hot topics was analyzed and discussed by consulting extensive literature. Taking theory innovation and engineering application as breakthrough points, the proposition of system reliability growth prediction theory and method study was put forward according to the situation of the nation's and the project's requirements. Example verification was carried out by taking wind turbine as the research object.2. The theoretical basis of AMSAA model and the problems of engineering practice were analyzed deeply, aiming at the problem of reliability growth process with instant corrective action. And prediction problem of reliability growth was further studied. In view of small sample problem, reliability growth prediction method was studied under uniform priori information and no priori information, bayes predictor and prediction limits of MTBF were given. Future failure time probability distribution and reliability characteristic quantities were discussed based on the random process theory, and then two reliability growth prediction models based on generalized gamma distribution and conditional distribution were proposed and established, thus, predictor and prediction interval of the future failure time were given. Reliability growth prediction model based on conditional distribution had higher prediction accuracy compared with other prediction methods published at home and abroad. This achievement provided new theoretical support for making reliability growth plans and carrying out reliability prior analysis.3. Concerned with the reliability growth process with hybrid fix modes, the reliability growth potential was discussed, and reliability level prediction after taking corrective actions was studied with failure analysis and fault pattern classification. Based on the assumption that failure rate obeys gamma distribution. First of all, reliability growth prediction model was established in the case of only taking delayed corrective actions. And then, reliability growth process was extended to the cases of taking instant and delayed corrective actions, and reliability growth prediction model was established, the MLE and Fisher approximate prediction interval of reliability characteristic quantities were given. Finally, fix effectiveness factor was modified based on D-S evidence theory. This achievement provided a new way to predict system reliability growth with hybrid fix modes.4. On the basis of theory studies, wind turbine reliability growth assessment and prediction methods were studied systematically. System reliability synthesis and prediction method of the MW level of wind turbine was discussed during the scheme demonstration and design stage. Wind turbine reliability growth bayes statistical inference was analyzed based on non-information prior and uniform prior information in the field debugging and trial operation stage. In the operation stage, dynamic reliability assessment and prediction method based on AMSAA model was studied using field data, further research on the problems of wind turbine reliability growth prediction was studied with instant and delayed corrective actions. And the reliability growth prediction for the key components reserve number of the wind turbine was discussed for maintenance and management. The above studies play an important role in the wind turbine reliability analysis, operation and maintenance, scientific management and decision-making.5. In order to satisfy the project demands, a"wind turbine reliability growth prediction analysis system"software was developed. It solved the problems of chaos management of reliability growth test data and difficult data analysis, which put forward new ideas and tools for reliability growth assessment and prediction.