Finding Robust Pareto-Optimal Solutions Over Time for Dynamic Disassembly Sequence Planning

Disassembly sequence planning plays a crucial role in the reuse and remanufacturing of end-of-life products, which is a combinatorial optimization problem and has been studied by many researchers. However, it is challenging to obtain optimal disassembly sequences due to great uncertainty owing to various unpredictable factors. We note that some of the uncertainties accompanying the products disassembly process are characterized by dynamic changes and can actually be regarded as dynamic disassembly sequence planning problem. Robust Pareto-optimal over time (RPOT) is a good approach to aovid the inconvenience of tracking optimization by finding solutions that remain acceptable over an extended period. Since there are few studies on applying RPOT to combinatorial optimization, the autoregressive prediction model in RPOT is more suitable for continuous search space problems than combinatorial optimization. In this paper, we develop a dynamic disassembly sequence planning problem considering the uncertainty caused by dynamically changing product states. Finding robust Pareto-optimal solutions over time for dynamci disassembly sequence planning to avoid the consumption of frequent switching solutions. To better apply RPOT to combinatorial optimization, online prediction model is proposed to replace the autoregressive prediction model. Experiment is executed in the three scale problems and compared with tracking optimization. The results indicate that online predictors can effectively improve the accuracy of prediction and improve the performance of the algorithm, and RPOT with new predictor is a more practical and time-saving method of addressing dynamic disaseembly sequence planning problem than tracking optimization.