弹性约束CSP及其基于遗传算法的交互式求解Agent

本文在回顾了约束满足问题(CSP)及其演进优化算法的基础上,提出了弹性约束CSP模型(ECSP),并将该模型形式化为六元组。ECSP模型是对已有的PCSP模型的改进。为了寻求ECSP问题的决策满意解,我们还设计了集成多Ageng、GA优化以及分布式并行计算技术的一种交互式多Ageng体系。我们详细阐述了其中的GA求解器算法,包括适应函数的确定、编码方式的选择、算子定义以及初始种群定义等。最后,我们用一个简单的算例证明了方法的有效性。

Elastic CSP and Its GA-Based Interactive Solving Agent

Constraint satisfaction problems (CSP) and the relevant evolutionary algorithms are reviewed first.Then we introduce a new CSP paradigm,elastic CSP model(ECSP),which is the improvement on the existing partial CSP model.We formularize ECSP as 6-tuple .For the purpose of finding the satisfactory solution(s) for decision,an interactive multi-agent architecture is designed,which integrates several technologies including multi-agent,genetic algorithm and distributed parallel computing.Furthermore,we elaborate the GA solver algorithm in the solving agent.How to determine the adaptive function?Which encoding method is chosen?How to define GA operators and initial population?Finally,we give a simple example to demonstrate the validity of our approach.