| The manufacturing industry,as the foundation of a country,is undergoing extensive and profound changes.Sustainable development has a top priority for the update of manufacturing industry,due to the increasing prominence of global issues,such as environmental pollution,energy shortage and social conflicts.To promote the construction of sustainable manufacturing,production scheduling plays an important role by allocating and using limited resources for completing tasks.However,traditional scheduling focuses on the economic indicators such as maximum completion time,output and production cycle.For the balance between the positive performance and negative impact of manufacturing enterprises,sustainable manufacturing scheduling further expands the consideration of environmental dimension and social dimension.Therefore,sustainable manufacturing scheduling is of great theoretical and practical importance to promote the sustainable development of manufacturing industry.As an extension of traditional scheduling,sustainable manufacturing scheduling problems present properties like multi-objective,multi-decision variables and multiple constraints,which are usually strong NP-hard optimization problems.In addition,the complexity of the actual production process is a factor that cannot be ignored in constructing a sustainable scheduling solution,and it further increases the difficulty of problem solving.The complex sustainable manufacturing scheduling problem considers both scheduling sustainability and model complexity,which poses a more severe challenge to the design of scheduling optimization method.Therefore,this paper deeply studies the sustainable manufacturing scheduling problems with different complex situations,and proposes targeted collaborative metaheuristic methods.The solution performance of the proposed methods is verified by a series of simulation experiments.The main research work of this paper is as follows:(1)For the sustainable manufacturing scheduling problem of unrelated parallel machines,an operators-collaborative particle swarm optimization algorithm is proposed.Considering the optimization requirements of meeting the deadline constraints and reducing machining energy consumption,a categorical fitness function is designed to effectively evaluate particles.In order to further improve the optimization performance,search operators with different optimization purposes are constructed,and a collaborative framework based on reinforcement learning is designed.In the proposed method,multiple operators can cooperate with each other to complete the efficient updating of particles.Through experiments and comparison,the effectiveness of categorical fitness function is verified.Experimental results also reveal that the operators-collaborative method significantly improves the solution results.Compared with other two algorithms from the parallel machine scheduling literature,the proposed algorithm has the superiority in solving the problem.(2)For the sustainable manufacturing scheduling problem of hybrid flow shop with batch production at last stage,a parameters-collaborative genetic algorithm is proposed.Considering the production planning constraints of the last stage and the differences of multiple stages,a hierarchical structure is designed including hierarchical coding,hierarchical crossover and hierarchical mutation,which effectively ensures the feasibility of the scheduling scheme.In order to improve the adaptability of crossover and mutation probability parameters to the problem,a parameters-collaborative method based on improved judgment points is designed.The probability parameters of the algorithm are configured according to the individual fitness value distribution information in the population and nonlinear function.The experimental results show that the hierarchical structure can significantly improve the performance of the algorithm,and the proposed parameters-collaborative method is more suitable for the problem studied than the other existing parameter configuration methods.Moreover,the proposed algorithm significantly outperforms the other three algorithms in hybrid flow shop scheduling literatures.In addition,the sustainability of the studied scheduling model is verified by comparing with the traditional scheduling model with time-related objective consideration.(3)For the sustainable manufacturing scheduling problem of unrelated parallel machines with fuzzy time parameters,a genetic-chaos collaborative algorithm is proposed.This problem considers two fuzzy parameters including fuzzy processing time and fuzzy due time,and aims to minimize the fuzzy total cost consisting of tardiness time and energy consumption.In view of the difficulty brought by fuzzy parameters,the genetic algorithm with the breadth of search and the chaos search algorithm with the depth of search are designed to solve the model respectively.Then,based on the analysis of the necessity of collaboration,a conditional collaboration method is designed to fully combine the advantages of the two algorithms.Experimental results verify that the collaborative algorithm significantly outperforms genetic algorithm and chaos search algorithm.Moreover,computation results reveal the conditional collaboration approach plays an important role in strengthening the algorithm.In addition,the superior performance of the proposed collaborative algorithm is proven by comparing with other three algorithms in the fuzzy parallel machine scheduling literature.(4)For the three-dimensional sustainable manufacturing scheduling problem of unrelated parallel machines with multiple types of tasks,a sustainable goal-oriented strategies-collaborative multi-objective memetic algorithm is proposed.The problem considers four different types of scheduling tasks to optimize the three objectives from economic,environmental,and social sustainable dimensions.In order to cope with the difficulty of solving a large decision space with many objectives,a multi-objective memetic algorithm with variable neighborhood descent is firstly designed to efficiently obtain the near-optimal non-dominated solutions.Furthermore,the relationship between scheduling tasks and optimization objectives is analyzed,and optimization strategies for different sustainable dimension objectives are designed.These optimization strategies are played out in a step-by-step manner to gradually optimize each of the different objectives.To achieve efficient collaboration between the directed strategies and the stochastic search,the elite set of multi-objective memetic algorithm adopts these optimization strategies.Through experiments and comparison,the effectiveness of the variable neighborhood descent method is verified,and the experimental results also validate the significant effect of the sustainable goal-oriented strategies-collaborative method on enhancing the algorithm.Furthermore,the proposed algorithm can significantly outperform two commonly used multi-objective scheduling algorithms in terms of both diversity and convergence of the non-dominated solutions. |
