Mechanical System Preventive Maintenance Research Based On Fault Diagnosis And Reliability Analysis

As an emerging discipline, reliability and maintainability is being increasingly recognized and developed. Especially in the field of mechanical engineering, since the device(system) tends to be increasingly large with complicated structures, it requires people to have a deep understanding of the system’s operating rules, as well as system reliability and maintenance activities. With the more complete understanding of failure mechanisms and its development law of system(component), system reliability and maintainability analysis also tends to be more practical in engineering system applications. On this basis, a variety of maintenance decision models for different failure mechanisms and different systems are proposed. These maintenance decision models provide an important theoretical basis for system reliability and maintainability engineering practice.Preventive maintenance, as an important way of maintenance, has been rapidly developed in recent decades, and is widely used in engineering practice. This thesis focus on the continuing degradation of system and system with component dependent failure. Some of the issues existing in the system are studied, such as system suffering degradation and shock of competitive failure modes, imperfect maintenance mode under continuous system deterioration and system with component dependent failure. Then, the reliability analysis and maintenance modeling are conducted in this thesis. The main contents can be summarized as follows:(1)Proposed fault information processing method in fault diagnosis.An important step before condition based maintenance is the implementation of fault diagnosis. This thesis proposes two methods, namely, spectral kurtosis and extraction sparse representation, for vibration information analysis. The blind equalization algorithm is used to remove the effects of the propagation channel in spectral kurtosis method. In sparse representation, the Morlet wavelet is optimized using the particle swarm optimization method, and the adaptive morphological analysis is utilized to process optimal wavelet filtered signal. The study shows that these two methods have their own advantages for fault information processing, and can effectively identify the characteristics of component failure.(2)Established a maintenance model considering system degradation and shock as two competing failure modes.It is very popular in mechanical equipment that system suffers two competing failure modes simultaneously, namely, degradation and shock. Through the analysis of internal degradation and external shock on system reliability, a condition based maintenance model is proposed. This model combines the results of fault diagnosis, and improves the rationality of maintenance decision to some extent.( 3) Established a sequential preventive maintenance model under the circumstances of imperfect maintenance.Compared with the perfect maintenance, the situation of imperfect maintenance is much closer to the practical application. Thus, it becomes a hot topic in maintenance research. At the same time, the sequential maintenance is also closer to the actual situation than the fixed period maintenance in the study of preventive maintenance policy. Combining these two cases, an imperfect maintenance policy is presented based on random failure rate adjustment and reduction of running age. The proposed maintenance policy is used to present a sequential preventive maintenance model. The results show that the proposed sequential preventive maintenance policy is more reasonable in maintenance cycle arrangement, and more efficient in maintenance cost.(4)Established a preventive maintenance model with component dependent failure in system.Component dependent failure is a very common phenomenon in a system. In this thesis, the Copula functions are used to analyze failure dependency in a two-component system and a multi-state k-out-of-n system, respectively. It reveals the dependency between the components, and analyzes its impact on the system reliability. Finally, a preventive maintenance model considering component dependent failure is proposed in this thesis. The results show that compared with the situation of component independent failure, the preventive maintenance model with component dependent failure is closer to the actual situation of maintenance. This model provides a new way for maintenance theoretical study and practical application.