Research On Model And Algorithm Of Multi-Robotic Disassembly Line Balancing With Uncertainty Processing Time

Advances in technology and the demanding market conditions recently have shortened the lifecycle of products.In the process that a product was produced,used,finally discarded to get recycled,a large amount of resource would be consumed.It would also cause a lot of pollution to deal with these End-Of-Life(EOL)products and faulty treatment would have a huge impact on the environment.On the other hand,as more rigid environmental legislation is improved,reasonable disposal of industrial waste is required for all industrial manufacturing industries.For EOL products in largescale production industries such as the automobile and electronics,some electronic chips or mechanical components are still useful.For environmental protection,resource saving or economy,it is of great significance to the recycling EOL products.As a crucial step during the recycling,building an efficient automatic disassembly system on the industrial production line and solving the problem of disassembly line balance is an important area of research.This study takes an extended type of disassembly line as the research object,and researches on the disassembly line balance with uncertain processing time.A mixedmodel disassembly line balancing problem is proposed for the greater variety of EOL products.Considering the uncertainty of processing time,a mixed-integer mathematical model is developed.By introducing the concept of multi-objective optimization and evolutionary computation,a multi-objective evolutionary simulated annealing algorithm is proposed.The main research contents are as follows:(1)Research on modeling of multi-objective disassembly line balance problem.To handle the change of disassembly structure caused by uncertain disassembly conditions,the task-oriented task sequence is utilized to integrate the task precedence among different products.Due to the different capacity of disassembly robots,the task processing time are assumed to shown strong differences and uncertainties.Thus,the proposed problem is defined as that a sequence of tasks of mixed-model products are assigned to the robots in fixed number of multi-robotic workstations where the combined task precedence relations are satisfied and the cycle time,peak energy consumption and total energy consumption are minimized.(2)Research on modeling of uncertain processing time.Considering the uncertainty of the disassembly process,the processing time is assumed to be stochastic variables with known distribution.The fuzzy number is also used to represent the uncertain the processing time and the fuzzy multi-objective programming is utilized.The processing time is assumed to be interval with known lower and upper bound,and the interval possibility degree and order relation are used to change the uncertain inequality with intervals to deterministic inequality.(3)Research on multi-objective evolutionary algorithms of multi-robotic disassembly line balance problem.An evolutionary simulated annealing algorithm based on Pareto dominance criterion is proposed to achieve the convergence and diversity of the Pareto optimal set.To keep the balance between the dimension of encoding and complexity of decoding,a multi-dimensional encoding method is applied for the mapping relationship between task sequence,assignment of robots and workstations.An efficient neighborhood search strategy is proposed to fully search the decision space,including the swap operator,division operator and partial sequence reorder operator.A Pareto-domination based acceptance criterion is proposed for population updating and maintenance.Computational experiments are conducted and the results demonstrate the proposed algorithm has better optimization quality and computational efficiency comparing with latest multi-objective algorithms.(4)Application research on model and algorithm of multi-robotic disassembly line balancing problem.A multi-robotic disassembly line system is realized,which can verify the feasibility and practicability of the proposed models and algorithms in this study.