一氧化碳泄漏处理全过程演练专家系统的构建

Construction of an Expert System for the whole Process Rehearsal of Carbon Monoxide Leakage Treatment

为有效应对一氧化碳泄漏事故发生时存在的不确定性与快速演变问题,构建了一套一氧化碳泄漏处理全过程演练专家系统。以淮南某燃气公司的露天一氧化碳储罐泄漏为例,基于动态贝叶斯网络和“情景-应对”分析法,建立事故关键状态节点,并合理分配节点转换概率来模拟一氧化碳泄漏事故的全过程发展态势;采用专家打分机制和改进的基于指标相关性的指标权重确定方法(criteria importance though intercrieria correlation,CRITIC)计算情景状态概率,从而推演出情景发展趋势。研究结果表明:基于动态贝叶斯网络的全过程演练专家系统能够较为准确地还原真实事故演变过程,可为事故处理单位提供技术支持。

In order to effectively deal with the uncertainty and rapid evolution of carbon monoxide leakage accidents,an expert system for the whole process of carbon monoxide leakage handling is constructed.Taking the leakage of an open carbon monoxide storage tank of a gas company in Huainan as an example,based on the dynamic Bayesian network and the“scenario-coping”analysis method,we constructed the key state nodes of the accident and reasonably assigned the transition probabilities of the nodes to simulate the development of the whole process of carbon monoxide leakage accident.The expert scoring mechanism and the improved criteria importance though intercrieria correlation(CRITIC)method are used to calculate the probability of scenario states and deduce the development trend of scenarios.The results show that the whole process rehearsal expert system based on dynamic Bayesian network can accurately restore the real accident evolution process and provide technical support for the accident handling units.