Abstract: The bi-level parallel computation model is proposed in this paper,which is oriental to the grid computation and based on the message passing.Bi-level parallel model is applied into evolutionary algorithm which adopts the mind-evolution-based machine learning and special partition strategy. Besides several communication models are integrated in order to balance the load and allocate grid resource dynamically. And the performance is testified on the Shanghai university E-institution grid computation platform.

摘要： 提出一种面向网格的基于消息通信方式的二级计算模型以求解问题。将二级模型与思维进化机器学习以及空间分解技术相结合（思维进化与空间分解并行演化计算－PMEBML-SP）,采用多种通信模型实现处理器间负载均衡、支持网格动态资源分配等功能,最后在上海高校网格E网格计算应用平台上实例验证。