中文摘要：

探索性分析（Exploratory Analysis,EA）是研究复杂的不确定性问题的有效方法,然而计算量过大的问题阻碍了EA的广泛应用。解决计算量过大的问题通常从建模的角度出发通过各种优化措施减小探索空间,但这可能造成有效信息的损失。文中从高性能计算的角度出发,针对EA计算的特点提出使用并行仿真克隆技术提高EA计算效率,从而可以扩大探索空间。文章阐述了递增克隆、虚拟逻辑进程、虚拟消息、消息的复制与转发等概念,介绍了递增克隆的原理和实现方法,通过实验对比了仿真克隆与仿真复制的性能,结果显示仿真克隆能显著提高仿真性能,从而说明仿真克隆技术是提高EA计算效率的有效方法。

英文摘要：

Exploratory analysis （EA） is an effective method in the research of complex uncertain problems. But too much computing load prevents the EA method from being widely used. The conventional method for solving this problem is to reduce exploratory space using optimization techniques, but the side effect is the loss of some useful information. This paper proposes a cloning method to improve the EA computation efficiency according to the characteristic of EA computing from the aspect of high performance computing, then describes the concepts of incremental cloning, virtual logical processes, virtual messages, and message replication ＆ forward. The principle and implementation method of incremental cloning are studied in this paper, and the performance is evaluated by comparing with simulation replication. The result shows a remarkable performance improvement, which makes out that simulation cloning is an effective method for improving the EA computing efficiency.