突发水污染风险的马尔科夫链贝叶斯网络预测

Risk Prediction of Sudden Water Pollution Under Markov and Bayesian Networks

突发水污染事故诱因众多且不确定,为预测此类不确定性因素进而进行防控,提前做好风险预防具有较大的经济意义。通过研究突发水污染事故风险因素、马尔科夫链和贝叶斯网络方法,根据现场调研获取的指标因素,提出马尔科夫链和贝叶斯网络相结合的突发水污染风险预测新方法,从纵、横两方面预测突发水污染事故发生的概率,弥补仅用马尔科夫链不能预测上层指标的不足,并为突发水污染事故的预防提供参考依据。实例分析结果表明:人为因素、机械设备因素和环境因素诱发黄河流域突发水污染事故的概率分别为52%、29%、12%,人为、机械设备和环境三方面应重点关注的风险因素分别是违规操作、管道破裂和自然灾害。

Influencing factors of sudden water pollution accidents are numerous and uncertain,so it is a greater economic significance to predict the uncertainty of such factors and then be able to prevent and control early.By studying the risk factors of sudden water pollution accident,Markov Chain and Bayesian Network,a new method of sudden water pollution risk prediction was proposed by combining with Markov Chain and Bayesian Network based on indicators adopted on site.This method can predict the risk probability of sudden water pollution accident from vertical and horizontal,and make up for the shortcoming that the Markov Chain can’t predict the upper indicators alone,and pro-vide a reference for the prevention of sudden water pollution accident.The results of example analysis are as follows.1)the probability of human factors,mechanical equipment and environmental factors to induce the Yellow River basin sudden water pollution accidents are 52%,29%and 12%respectively;2)Three aspects of human,mechanical equipment and environment should be focused on risk factors are illegal operation,pipeline rupture and natural disasters respectively.