Abstract: An intelligent data batch processing control model based on software Agent is introduced，which is mainly used in designing and manufacturing collaborative platform.The behavior character of request processing is analyzed by using Agent’s study ability and recorded in the belief library of Agent.With environment sensor，when a request arrives，Agent gets data about system load，bandwidth of client，and so on.Then combining the old request characteristics and using reasoning processing，Agent works out the optimal data volume to be processed of this request，which will be controlled by control flow that has been changed by execute plan.Integrating designing and manufacturing collaborative platform with this model may ensure the user’s response time，shorten waiting time and smooth the peak value of system load.Performance of this model is also be validated by experiments.

Key words: software Agent, response time, data batch processing, performance optimization

摘要： 介绍了一种用于设计制造协同平台的基于软件Agent的智能数据批量处理控制模型。模型利用Agent的学习能力分析数据处理的行为特征并记录到信念库中，当有请求到达时，Agent利用外部环境传感器获取系统负载、用户带宽等数据，并结合信念库中的请求特征，运用推理过程计算出当前处理的最佳数据量，通过调用执行计划改变请求的控制流来控制本次请求需处理的数据量。在协同平台中集成本模型可以保障用户的请求响应时间，缩短等待时间，平滑系统负载的峰值。通过实验验证了模型的性能。

关键词: 软件Agent, 响应时间, 数据批量处理, 性能优化