基于信誉构建信任机制是解决虚拟计算环境［1］中行为可信问题的重要途径。现有信誉模型对不良行为的动态适应能力和对信誉信息的有效聚合能力不足，缺乏鼓励节点积极参与诚实合作的有效机制。本文提出一种基于惩罚激励机制的信誉管理模型PERep，根据自主元素行为特征和信誉状态区分交易行为的合作与偏离，并对偏离行为进行惩罚；给出了PERep的分布式实现。实验表明， PERep能准确地区分诚实交易和恶意行为，有效提高自主元素诚实交易的积极性并减少恶意行为的危害。

Reputationbased trust mechanisms provide an important way to solve the behavior trustworthiness problems in selforganized iVCE environments.The existing reputation models are deficient in the ability of dynamically adapting to the strategically altering behavior and the efficiently aggregating reputation information. Efficient mechanisms to stimulate honest and active collaboration are also needed. In this paper, a penaltyincentivemechanismbased reputation management model PERep is proposed.According to the behavior characteristics and reputation status, collaboration or departure can be distinguished and departure will be punished. Its distributed realization is presented. Simulation results show that PERep has the ability of exactly distinguishing honest trade and malicious behavior,prominently improves the autonomic elements’ enthusiasm to be honest, and greatly deduces the harm of malicious behavior.