Abstract: To overcome the separation problem of business units and knowledge resource allocation during the business process,a business-knowledge allocation strategy based on the Grey Wolf Optimizer for Support Vector Regression (GWO-SVR) and improved Simulated Annealing (SA) algorithm was proposed.In this strategy,knowledge resources were encapsulated as knowledge modules based on the analysis of user requirements and the business scenario.On this basis,through the function of the configurator,the initial configuration between the business units and knowledge modules was realized.A set of synthesis evaluation indicator systems was developed based on the analysis of the evaluation parameters,and the CRITIC—fuzzy comprehensive evaluation method was used to obtain the evaluation scale of business-knowledge configuration.Based on the evaluation scale,a dynamic evaluation model based on GWO-SVR for the composition of business-knowledge configuration was constructed and trained.The functional relationship of this trained model could be imported as the objective function during the improved SA optimization process since GWO-SVR was a regression model.Through the optimization iteration,an optimal composition plan corresponding to the optimal value was found to satisfy business needs.Thus,the optimal allocation of knowledge resources was achieved.Using the processing of the reducer box as an example,the effectiveness of the proposed models and algorithms could be verified.

Key words: knowledge-business configuration, knowledge module, support vector regression, grey wolf optimizer, simulated annealing algorithm, knowledge service

摘要： 为解决业务流程下业务单元与知识资源配置分离的问题,提出一种基于灰狼算法优化支持向量回归(GWO-SVR)和改进模拟退火算法(SA)的知识-业务优化配置策略。该策略基于用户需求和业务情景分析,将知识资源封装为知识模块。在此基础上,通过配置器作用实现知识模块与业务单元间的初始配置。然后,依据知识模块评价指标参数分析,构建综合评价指标体系,并运用CRITIC-模糊综合评估法得到知识-业务配置组合评价量表;基于此评价量表,构建和训练基于GWO-SVR的知识-业务配置组合动态评价模型。由于GWO-SVR是回归模型,可将该训练好的模型的函数关系式作为改进SA算法优化的目标函数导入,通过寻优迭代找到最优值对应的最优组合方案,实现满足业务需求的知识资源最优配置。以减速器箱体加工为例进行验证,证明了所用模型和算法的有效性。

关键词: 知识-业务配置, 知识模块, 支持向量回归, 灰狼算法, 模拟退火算法, 知识服务