Research On Resource Allocation And Task Scheduling For Uncertainty In Cloud Manufacturing Mode

In recent years,operational management under the cloud manufacturing model has become an important management model for manufacturing enterprises and related organizations to improve resource utilization,enhance manufacturing efficiency,and promote collaborative innovation.Resource allocation and task scheduling are the core elements of cloud manufacturing operation management.Developing scientific resource allocation plans and task scheduling strategies can maximize the satisfaction of user order needs and achieve dual social and economic benefits for manufacturing enterprises.However,facing the high complexity and random uncertainty of cloud manufacturing processes,how to accurately allocate user order tasks to various collaborative manufacturing resources in a timely manner and carry out reasonable and effective task scheduling has become one of the research difficulties in cloud manufacturing operation management.Therefore,this article aims to enhance the comprehensive social and economic benefits of manufacturing enterprises,and deeply explore the uncertainty oriented resource allocation and task scheduling problems under the cloud manufacturing model.The research will focus on the model construction and algorithm design of cloud manufacturing resource allocation and task scheduling problems in uncertain environments,in order to provide practical theoretical support and scientific decision-making basis for manufacturing enterprises and related organizations in cloud manufacturing operation management.Firstly,we study the event-driven cloud manufacturing resource allocation problem under uncertain manufacturing capabilities.Based on the Wasserstein metric,conditional support set,conditional mean,and conditional expected dispersion,an event-driven hybrid ambiguity set is constructed to characterize the uncertainty of manufacturing capacity in manufacturing cloud service units.At the same time,taking into account key service quality indicators such as the establishment cost of manufacturing cloud service units,service cost,energy consumption,and task service time,a cloud manufacturing resource allocation adaptive distributed robust optimization model is constructed,and the model is equivalently transformed into a tractable mixed integer programming model.Thoroughly analyze the structural properties of the model and propose relaxation and sensitivity bounds with optimal error guarantees to reduce computational complexity.To effectively solve the transformed equivalent model,a constraint generation algorithm based on CPLEX is designed.This algorithm decomposes the mixed integer programming model into a relaxed main problem and a series of bilinear programming subproblems,and transforms the bilinear programming subproblems into 0-1 mixed integer programming problems with integer solution properties.In addition,the introduction of In-out constraint generation strategy and Pareto constraint generation strategy improves the efficiency of the algorithm.Finally,through a large number of numerical experiments,the feasibility of the proposed model and the effectiveness of the algorithm were verified.Based on the analysis of test samples,the superiority of event-driven hybrid ambiguity sets over event-driven moment information and Wasserstein information ambiguity sets,as well as the advantages of considering the robustness of event-driven distributional ambiguity,were evaluated.The quality of the proposed relaxation and sensitivity bounds was evaluated,and the influence of model parameters on solution structure and performance was analyzed.Secondly,we study the goal oriented cloud manufacturing resource allocation problem under the uncertainty of task service time.Using an event-driven hybrid ambiguity set composed of Wasserstein metric,conditional support set,conditional mean,and conditional expected dispersion to characterize the uncertainty of task service time.With the goal of minimizing the manufacturing capacity shortage of manufacturing cloud service units,and with the constraint that the service cost and total energy consumption cost of manufacturing cloud service units do not exceed the corresponding values of traditional distributionally robust models,based on the Herwitz criterion,a ”Targeted-oriented” distributionally robust optimization model for cloud manufacturing resource allocation facing manufacturing capacity shortage is constructed,which takes into account all situations from the worst to the best.The model is equivalently transformed into a tractable mixed integer programming model.To effectively solve the transformed equivalent model,a Benders branch-and-cut algorithm based on CPLEX is designed,and three types of acceleration strategies are introduced to further improve the efficiency of the algorithm,including In-out Benders cut generation strategy,Benders dual improvement strategy,and dual variable normalization strategy.Finally,through a large number of numerical experiments,the feasibility of the proposed model and the effectiveness of the algorithm were verified.Based on test samples,the advantages of event-driven hybrid ambiguity sets over event driven moment information and Wasserstein information ambiguity sets were analyzed,as well as the advantages of considering targeted-oriented distributional robustness.The key parameters were evaluated and the impact of considering the Herwitz criterion on the model solution structure and performance was also considered.Finally,the event-driven cloud manufacturing task scheduling problem is studied under the uncertainty of task service time.Using an event-driven hybrid ambiguity set composed of Wasserstein metric,conditional support set,conditional mean,and conditional expected dispersion to characterize the uncertainty of task service time,with the service cost,energy consumption,and worst-case expected task service time of manufacturing cloud service units as the objectives,a distributionally robust chance constraint program is constructed,considering that the probability of the worst-case expected task service time of manufacturing cloud service units not exceeding the maximum manufacturing capacity is not less than 1-ε.A distributionally robust chance constraint optimization model for cloud manufacturing task scheduling is constructed,and the model is equivalently transformed into a tractable mixed integer programming model.To effectively solve the transformed equivalent model,a Benders dual branch-and-cut algorithm based on CPLEX is designed,and an In-out Benders dual cut generation strategy,weak Benders dual cut generation strategy,and aggregated sample Benders dual cut generation strategy are introduced to improve the efficiency of the algorithm.Finally,through a large number of numerical experiments,the superiority of the proposed algorithm over the CPLEX commercial solver and traditional Benders branch-and-cut algorithm was verified.Based on test sample analysis,the advantages of event-driven hybrid ambiguity sets over eventdriven moment and Wasserstein information ambiguity sets,as well as the advantages of considering distributional robustness,were considered.The impact of key parameters on the model solution structure and performance was evaluated.