Research On Maintenance Strategy Optimization Of Complex System

Reasonable maintenance strategies can improve the availability of the manufacturing system and reduce the production cost.In reality,it is difficult to establish and solve the maintenance optimization model of complex system because of the large state space and the correlation between components.Therefore,it is challenging to find a reasonable maintenance strategy.This paper focuses on the research of condition-based maintenance optimization of complex systems.Taking common serial manufacturing systems,large-scale complex manufacturing systems and regional power grid systems considering performance sharing as examples.Optimization models are established according to the structure and characteristics of different systems.Optimization algorithms are designed and optimal maintenance strategies are solved.Firstly,based on the continuous time discrete state hypothesis,the maintenance optimization problem of serial manufacturing system with intermediate buffers is studied.An iterative algorithm is designed to generate the state transition matrix of the system.Three maintenance strategies are proposed and their effects on the objective function are compared.Numerical examples show that the proposed maintenance strategy can increase production revenue by 36% compared with the single threshold maintenance strategy.On the basis of series manufacturing system,the maintenance optimization problem of complex manufacturing system with feedforward and feedback loop is further studied.Because of its large state space,analytic method is no longer applicable to the calculation of objective function.This paper presents a SB&B algorithm based on simulation optimization.Numerical study shows that the multi-threshold maintenance strategy considering the states of other components can obtain higher economic benefits.In addition,sequential simulation can allocate simulation resources to different solutions more reasonably,improving the convergence efficiency of the algorithm and improve the quality of solution.Finally,this paper studies the maintenance optimization problem of the manufacturing system considering performance sharing.Due to the performance transmission between subsystems in this system,the correlation between components is enhanced.In this paper,continuous decision is adopted to optimize the maintenance strategy,and the least square time series difference(LSTD)algorithm is used.In the maintenance strategy optimization problem of small-scale system,the results of LSTD algorithm are compared with the theoretical optimal values obtained from policy iteration,and the effectiveness of the algorithm is verified.In large-scale system,LSTD algorithm,genetic algorithm and multi-agent DQN are compared respectively,and the rationality of the selected basis function and agent is verified.The maintenance strategy obtained by the LSTD is analyzed,which shows the correlation between components.To sum up,this paper studies several common manufacturing system maintenance strategies and corresponding optimization algorithms.The advantages and disadvantages of various optimization algorithms are compared.A multi-threshold maintenance strategy is proposed.The SB&B algorithm based on simulation is improved,the simulation resources are allocated reasonably,and the efficiency of the algorithm and the quality of the solution are improved.The application of approximate reinforcement learning in maintenance optimization is studied,and the selection of basis function and agent is given.