Research On Dynamic Disassembly Sequence Planning Based On Multi-Objective Robust Optimization Over Time

Disassembly sequence planning seeks the optimal disassembly sequence for end-of-life products from the perspectives of improving resource utilization and reducing environmental hazards.Various unpredictable factors in the disassembly environment lead to significant uncertainties in the actual disassembly process,which makes the optimal disassembly sequence change dynamically under the influence of the environment.Considering the dynamic nature of the existence of uncertain disassembly disassembly environment,disassembly sequence planning is actually a dynamic optimization problem.Most of the existing studies focus on finding the optimal disassembly sequence under the current disassembly environment by the idea of tracking optimization.Tracking optimization requires detecting changes in the environment in a timely manner and reacting quickly to find a new optimization solution.However,tracking optimization in an uncertain disassembly environment incurs high switching costs and consumes a large amount of computational resources due to frequent solution finding and deployment changes,which makes tracking optimization difficult to apply in real disassembly.In contrast,time-domain robust optimization can avoid the inconvenience of trace optimization by finding deployment solutions that are applicable in multiple environments.Most of the existing studies on time-domain robust optimization have verified its effectiveness on dynamic single-objective optimization problems with continuous search space,but extending its application to dynamic multi-objective optimization problems with discrete search space is a challenge.To address the above situation,this paper proposes a dynamic disassembly sequence planning model considering product state uncertainty,and designs a multi-objective robust optimization over time algorithm to find disassembly sequences applicable in multiple disassembly environments to avoid the inconvenience of frequent solution finding.The main research contents are as follows.(1)Research on dynamic disassembly sequence planning problem model.The dynamic disassembly sequence planning problem is proposed for the phenomenon of uncertainty in the state of end-of-life products.Based on the improved AND/OR diagram to represent the structure of end-of-life products and the task priority relationship,the disassembly tasks are refined to the operation sequences affected by the state of product sub-parts,and combined with the disassembly sequence planning theory,a mathematical model of dynamic disassembly sequence planning considering the uncertainty of product state is constructed with the objective of minimizing disassembly consumption and time.(2)Research on dynamic disassembly sequence planning algorithm based on preference-free multi-objective robust optimization over time.For the dynamic demolition sequence planning problem with preference-free demand of decision makers,a dynamic demolition planning algorithm based on tracking optimization is proposed to find the optimal Pareto solution set for the current environment under the dynamically changing demolition environment.According to the characteristics of the dynamic disassembly sequence planning problem,the coding and decoding method and the crossover and variation operators are designed.In order to avoid the inconvenience of tracking optimization for frequent solution finding in dynamic disassembly environment,the disassembly sequence planning algorithm based on time-domain robust Pareto optimization is proposed to find the Pareto solution set with optimal average fitness in a fixed time window for the dynamic disassembly sequence planning problem.Based on the time-domain robust Pareto optimization algorithm,an online predictor is proposed to improve the performance of the algorithm by enhancing the accuracy of prediction.The time-domain robust Pareto optimization algorithm is able to find a more robust Pareto solution set for more disassembly environments than the tracking optimization method.(3)Research on dynamic disassembly sequence planning algorithm based on preference multi-objective robust optimization over time.The decision maker is faced with a frequently changing dynamic demolition environment,and the application of selecting a solution from the Pareto solution set becomes a challenge.Based on the existence of explicit preference information of decision makers,two multi-objective time-domain robust optimization algorithm frameworks are proposed for the dynamic demolition sequence planning problem.Based on the case that the decision maker does not have explicit preference information,a multi-objective time-domain robust optimization algorithm based on the average fitness preference is proposed to find the demolition sequence that maximizes the average fitness within a fixed time window for the dynamic demolition sequence planning problem.The algorithm pinpoints the natural preference solution that maximizes the demolition benefit in the dynamic demolition sequence planning problem by expanding the dominated region of the solution and combining the advantage of the knee point.Based on the case where the decision maker has explicit preference information,a multi-objective time-domain robust optimization algorithm based on survival time preference is proposed to find the demolition sequence for the dynamic demolition sequence planning problem that is used for a long time while satisfying the constraints.The proposed algorithm is applied to dynamic demolition sequence planning instances of different sizes and compared with the frontier algorithm to verify the effectiveness and superiority of the algorithm.