Abstract: To solve the problem of high uncertainty of concurrent service access and the problem of unbalanced resource service allocation on-demand for entity industry in distributed environment,a novel multi-task scheduling optimization algorithm based on mulit-communtiy collaborative searching was proposed to implement the resource service scheduling.Based on analyzing the characteristic of science and technology resource service,a multi-objective optimization scheduling model was established by considering the uncertainty of access to concurrent services and the imbalance of resource allocation;then the solution procedure of multi-community collaborative search algorithm was given,and the speed and position of particles were coded by binary system.The mapping of particle swarm optimization algorithm to discrete space was completed by reconstructing particle expression.At the same time,the interactive evolution mechanism between different particle communities was established to enhance the diversity of the population.Furthermore,the algorithm adaptability to the search environment and the solution accuracy were improved.An example of the service task scheduling process of automotive engine fault diagnosis and maintenance resources was used to show the validity of the proposed algorithm.It provided useful means for solving the complex scheduling problem.

Key words: science and technology resource, multiservice tasks, heuristic task scheduling, multi-community collaborative search

摘要： 为了解决分布式环境下科技资源服务过程并发服务访问不确定性高,按需服务实体产业分配不均衡的问题,提出一种基于多群落协作搜索的启发式任务调度策略。在分析科技资源服务调度过程及特点的基础上,搭建了考虑分布式科技资源并发服务访问不确定性和资源分配不均衡性的多服务任务优化调度模型;给出该调度模型的多群落双向驱动进化算法,并采用二进制对粒子的速度和位置进行编码,通过重构粒子表达式完成粒子群算法到离散空间的映射,同时建立不同粒子群落之间的交互进化机制以增强种群的多样性,进而提高算法对搜索环境的适应能力和求解精度。以汽车发动机故障诊断维修资源服务任务调度过程为例,验证了所提方法的有效性,为复杂调度问题的求解提供了有效手段。

关键词: 科技资源, 多服务任务, 启发式任务调度, 多群落协同交互搜索