Decision-making Optimization Of Predictive Maintenance Oriented To Customer’s Personalized Requirement

The popularization of mass production technology makes most products with excess capacity and serious homogenization,and it has greatly intensified the competition among manufacturing enterprises.In order to win in the fierce market competition,the manufacturing industry is facing deep transformation and change,including from mass production of products to the provision of customized products and services.A huge demand for personalized services has been created.The traditional maintenance decision-making model does not fully consider the personalized requirements of customers,thus it is difficult to meet diversified maintenance services.By providing personalized and refined maintenance services for the customers,it can improve the product’s availability,enhance customer’s satisfaction and increase the revenue,and it has become an important research direction in the field of maintenance decision-making.This thesis carries out research on predictive maintenance decision-making optimization from the perspective of customers’ personalized requirements,including failure prediction based on grey neural network model,selective maintenance based on maintenance time threshold,and maintenance service contract design based on game theory.The major research contents are as follows:(1)Failure prediction based on grey neural network combined model is studied.In order to improve the accuracy of maintenance decision-making,the improved grey model GM(1,1)and BP neural network model are proposed for the customer’s products with a small amount of feature monitoring data.Based on the idea of combination prediction,the grey neural network combination model is constructed to carry out failure prediction.Case study shows that compared with the single model such as grey model GM(1,1)and BP neural network model,the proposed grey neural network combination model has higher accuracy for failure prediction,and it can more accurately predict the maintenance time of the product.(2)Multi-objective selective maintenance optimization is conducted with consideration imperfect maintenance.Considering the relationship between the maintenance cost and maintenance quality,the age reduction coefficient and hazard rate adjustment coefficient are constructed on the basis of maintenance cost.The hybrid failure rate model is used to represent the improvement degree of system reliability.In view of the limited maintenance time and by considering system reliability and maintenance cost,a multi-objective selective maintenance model is established based on nonlinear stochastic programming.Taking series-parallel system as the object,the case study is completed.The result demonstrates that the proposed multi-objective selective maintenance model can better allocate maintenance resources,flexibly choose maintenance programs and meet customer preferences.(3)Based on the non-cooperative game theory,the design and pricing of maintenance service contract is studied,where the punishment and reward mechanism are considered according to the set multi-threshold maintenance time.Multiple maintenance service contracts are designed to take into account the maintenance actions,such as corrective maintenance and preventive maintenance.The customer and service agent adopt complete information non-cooperative game theory to optimize the pricing of maintenance service contract,so as to achieve a win-win situation.Through numerical example analysis,the result shows that by considering punishment and reward mechanism,the maintenance service contract can improve the revenue for customers and service agents.For service agents,with the increase of repair rate and preventive maintenance improvement factor,the revenue of maintenance service contract increases gradually.Moreover,with the increase of the contract time,the revenue presents a trend of first increase,then decrease.The major contributions of this thesis include:(1)it systematically carries out the predictive maintenance decision-making optimization research for customer’s personalized requirements,by considering a small amount of product’s feature data,limited maintenance time and maximizing the product’s expected revenue.(2)From various views of failure prediction,selective maintenance and maintenance service contract,novel prediction methods and optimization models are proposed for predictive maintenance,and it enriches the corresponding theory.