Research On Uncertainty-aware Multi-task Scheduling Optimization In Cloud Manufacturing

With the continuous development and integration of information technology and manufacturing technology,cloud manufacturing as a new manufacturing paradigm has attracted extensive attention.Manufacturing tasks in cloud manufacturing system have the characteristics of large scale and personalization,and it chooses the best manufacturing services for users based on the personalized needs of each other.Scheduling problem is one of the core problems in cloud manufacturing system.Different from the previous simple matching and combination of subtasks and services,scheduling problem consists of two subproblems: service assignment and task sequencing.Task assignment assigns manufacturing services to each subtask and task sequencing sequences subtasks with corresponding services to obtain superior solutions.By solving the two key problems of service assignment and task sequencing,the resource utilization rate can be effectively improved,the manufacturing cost can be reduced,and the sustainable development of the manufacturing system can be promoted.In the cloud manufacturing environment,scheduling multiple heterogeneous tasks to meet user customization needs is a challenging research topic.On the one hand,when the manufacturing environment changes,a previous schedule may become infeasible owing to the inherent uncertainty of the scheduling problem in cloud manufacturing.Consider that the scheduling system and manufacturing resources are connected remotely through the Internet,which makes it difficult to respond to dynamic interference events timely and effectively.It brings great troubles to the cloud manufacturing scheduling problem.On the other hand,in addition to basic attributes such as time,cost and reliability,conflicts of interests of more stakeholders in cloud manufacturing paradigm need to be considered,such as service providers,service consumers and service managers.Therefore,in the cloud manufacturing environment,how to dynamically schedule multiple tasks to balance the conflicts of interests among service providers,service consumers and service managers is an urgent problem to be solved.To counter the problems above,this paper takes personalized manufacturing tasks submitted to the cloud manufacturing system as the research object,aiming to investigate the uncertainty-aware scheduling problem in cloud manufacturing to achieve service optimization allocation.To solve the cloud manufacturing scheduling problem under different uncertain conditions,two optimization models are proposed respectively.At first,for scenarios where uncertain information,including precise values and service interruption information,is available.A multi-task scheduling optimization model based on proactive strategy is established to reduce the impact of uncertainty by considering the uncertainty of manufacturing service interruption.Then,for scenarios where uncertain information is difficult to obtain,a multi-task scheduling optimization model based on fuzzy theory is established to solve the fluctuation problem of attribute values and the difficulty in obtaining service interruption information,and the optimization objectives of multiple stakeholders in the cloud manufacturing system are considered.Finally,based on the basic genetic algorithm,two extended genetic algorithms are proposed to improve the performance for solving the two scheduling optimization models proposed in this paper.The main contributions of this paper can be conducted as follows:1.A multi-task scheduling optimization model based on proactive strategy is proposed.This model assumes that precise values and service interruption information are available,and builds a proactive scheduling method based on cloud service interruption information,aiming to balance the fitness,robustness and stability to obtain a robust and stable proactive schedule,so as to reduce the disturbance of uncertainty factors on the schedules and avoid causing unnecessary losses to users.Then,the multi-stage method,adaptive parameters,and the local search algorithm are integrated to avoid getting into local optimization.The experimental results confirm that the obtained proactive schedule produces great performance when applied to multi-task scheduling problems with service interruptions.Furthermore,the results also show that the proactive schedule obtained by the proposed approach is more robust and stable than other baseline algorithms taken from previous literature for the proposed model.2.A multi-task scheduling optimization model based on fuzzy theory is proposed.The model replaces precise values with triangular fuzzy numbers to show the dynamic of manufacturing process.Besides,it is not limited to the interests of users,but comprehensively considers the interests of users,manufacturers and manufacturing platform.In addition,the preference of quality of service attributes and the importance between tasks are calculated by using multi-criteria fuzzy decision method to replace the weighting method based on the experience of decision makers,aiming to increase the reliability and effectiveness of schedules.Then,the migration operation,local search and restart strategy are integrated to maintain the diversity of the population.The experimental results demonstrate that the proposed extended genetic algorithm solves the proposed model effectively,providing better optimal solutions compared with the baseline algorithms.