| For a manufacturing company, mainteinance is an important way to decline asset failure risk and improve asset reliability. The ultra goal for modern manufacturing companies is to realize “zero fault”, “zero risk”, “zero accident”, “zero pollution” and “selfwareness” during the whole production process under the framework of “Made in China, 2025” and “Industry 4.0”. Thus requires moden companies have to transfer maintenance mode from conventional “breakdown maintenance” and “preventive maintenance” to more intelligent and more personalized maintenance mode which considers relative impacts for maintenance comprehensively.Since many elements can impact maintenance decision making. Currently, there is no general maintenance policy that is suitable for all production systems. When making maintenance decision, two nonignorable impacts should be considered, one is asset degradation information and remaining useful life, the other is production scheduling on these assets. There has a lot of research done in domestic and oversea, and many research achievements have published. While, the research is deficient to some extent. With the foundation of Key Project supported by National Science Foundation of China(51035008) and the Fundamental Research Funds for the State Key Laboratory of Mechanical Transmission, Chongqing University(SKLMT-ZZKT-2012 MS 02), this paper mainly researches on and solves the problems about maintenance decision making based on remaining useful life prognosis information and production scheduling for key transmission components. The main work of this thesis includes the following aspects.(1) According to the character of data amount, the conceptions of complete data of degradation process and incomplete data of degradation process are proposed. For a key transmission component, the concept of adaptive feature window and other concepts are introducted. Then, an artificial neural network-based model and linear-based model and exponential-based model are built to predict degradation and remaining useful life. Thus reliazes remaining useful life prediction and degradation prediction for an individual under no failure or suspension histories condition. According to the predicted degradation feature, a new failure rate model is proposed, then optimal condition based maintenance policy can be got by minimizing maintenance cost rate.(2) Considering the interactive impact on time for maintenance activities and production, and the impacts of maintenance for production system reliability improvement and maintenance cost, joint optimization models of maintenance and production scheduling are built. The effect, time, and cost of different maintenance modes and the failure risk of production system associately are considered. Joint optimization models are built considering the characters of different production systems with the objectives to maximize the total profit or minimize the total cost. A view is proposed that is asset cannot be maintained infinitly under the imperfect preventive maintenance mode. An joint optimization model of group maintenance and production scheduling is proposed, maintenance interval and production scheduling are optimized simultaneously. The maintenance policy is determined from the view of economy.(3) Based on the model analysis above, a prototype system is developed based on remaining useful life prediction information and production scheduling for manufacturing system. This developed system can help maintainers maintain systems and production scheduling makers make scheduling shemes. The architecture of the system is built, the basic process and function modules are analyzed. This system is developed by MATLAB GUI. The system is an intelligent “self-prognosis” and “self-decision” system which integrates remaining useful life prediction, condition based maintenance, maintenance planning and production scheduling. |
