Multi-and Many-objective Scheduling Optimization For The Green Collaborative Operation Of Production And Logistics

In recent years,the continuous development of modern logistics has promoted the close integration of logistics activities and industrial production,and the demand for collaborative operation between production and logistics is more and more frequent.On the other hand,with the global energy shortage and increasingly prominent environmental problems,green production has become a new trend of enterprise development.Scheduling is an important link in the implementation of industrial production system.In the context of production and logistics collaboration and green production,based on cost reduction and efficiency improvement,scheduling plan needs to take into account the objectives of energy conservation and emission reduction.Different from traditional production scheduling problems,the production and logistics collaborative operation scenario with the consideration of energy efficiency is more complex and needs to schedule more elements.It is a multiand many-objective NP-hard problem with complex collaborative operation constraints.Intelligent optimization methods,such as evolutionary algorithm and swarm intelligence algorithm,can solve this kind of complex scheduling optimization problems,but they need to overcome the problems such as inefficient evaluation and prone to fall into local optimum for multi-and many-objective problems(MOPs/MaOPs).Therefore,it is necessary to design efficient multi-and manyobjective scheduling optimization methods to meet the challenges brought by the collaborative scheduling of production and logistics.Combined with the scheduling problems on green collaborative operation of production and logistics,this thesis carries out a study on multi-and many-objective scheduling optimization methods.This thesis focuses on two kinds of scheduling problems with different characteristics of production and logistics cooperation:(1)cooperation between multiple production equipment;and(2)cooperation between production equipment and transportation equipment.Several multi-and many-objective optimization methods are desigened in line with the characteristics of the scheduling problems.The main contents and innovative achievements of this thesis are summarized as follows.(1)The key issue on solution evaluation and selection in MOP/MaOP is deeply studied.Combined with the characteristics of fuzzy correlation entropy(FCE),a novel fitness evalution mechnism(FEM)is proposed.In order to realize the effective connection between fuzzy information entropy and MOP/MaOP,reference points(include ideal point and nadir point)are constructed via serial,parallel and dynamic methods,respectively.Comparison points are constructed with the objective function values of solutions in population.The ideal point and comparison points are then mapped into fuzzy sets via a semi trapezoidal membership function.A judgment criterion based on fuzzy correlation entropy coefficient is established to evaluate the population solutions in MOP/MaOP.Based on the proposed FEM,an optimization framework for complex multi-and many-objective scheduling problem is proposed.It provides a basis for the subsequent multi-and many-objective green cooperative scheduling of production and logistics.(2)A multi-objective scheduling problem of green parallel machine with multiple production equipment cooperation in container port is discussed.Based on the classical parallel machine scheduling theory,constraints such as movement and safety distance of HQCs are considered.A three-objective optimization model including container completion time,average operation time of HQCs and energy consumption of HQCs is constructed.An enhanced multi-objective evolutionary algorithm(EMOEA)is designed.EMOEA uses the FCE-based FEM to evaluate solutions,and utilizes the opposition-based learning(OBL)strategy to improve population diversity.With the CPLEX,the effectiveness of the scheduling model is verified.The influence of HQCs on the three shceduling objectives is discussed.Simulation experiments verify the impact of the new strategies on the performance of EMOEA,and the significant advantages of EMOEA on solving the studied problem.(3)A multi-and many-objective scheduling problem on green job-shop with the collaboration of multiple automatic guided vehicles(AGVs)is studied.Considering the production and logistics collaborative operation characteristics including manufacturing processing,setup activities and AGV transportation,a many-objective model minimizing the completion time,total tardiness,total idle time and total energy is constructed.A many-objective differential evolutionary algorithm(MaDE)is proposed to solve the model.A four-layer encoding mechanism is studied to respresent the problem’s solutions.Population initialization based on heuristic rules and average entropy is used,and the population is evaluated with FCE-based FEM.A coevolution strategy with multiple subpopulations and an interaction strategy based on chaotic migration are adopted to achieve an effective balance between convergence and diversity.An adaptive local intensification strategy is adopted to improve its local search ability.The effectiveness of the scheduling model is verified.The influences of setup and AGV transportation on scheduling objectives are analyzed.The effects of new strategies on MaDE’s performance are discussed.Comparative experiments show that MaDE has superior optimization performance.(4)Engineering case studies of multi-and many-objective scheduling with different green collaborative characteristics of production and logistics are carried out.With the research on multi-and many-objective scheduling theory of green collaborative operation of production and logistics,combined with the container port production that can reflect the parallel collaboration characteristics of multiple production equipment,and the workshop production of robot intelligent manufacturing that can reflect the collaboration of production equipment and transportation equipment,the multi-and many-objective scheduling cases are refined.The availability of scheduling models and algorithms are tested,and their application effects in practical engineering scheduling problems are discussed.