基于模糊推理的海底管道泄漏爆炸概率分析

Probability analysis for the leakage and explosion of the submarine installed pipeline based on the fuzzy inference

针对海底管道可燃气体泄漏引起的爆炸,进行了风险识别并构建故障树。将权重修正、输出模糊数修正等方法与模糊推理方法相结合,利用Matlab和Simulink建立基于故障树的模糊推理系统,进行爆炸概率的计算。此外,给出了故障树子事件概率的计算方法,包括概率和模糊数。最后,给出算例并与重要度计算结果进行比较,研究表明此方法更加科学,具有灵活性和可拓展性。

The given paper is aimed at making an evaluation of the explosion risk probability likely to be caused by the leakage of combustible gas explosion of the submarine pipeline through a method of fuzzy inference system. For this purpose,we have first of all performed a risk identification and a fault tree construction analysis by way of the accident statistics and FMEA analysis through the exploration of the relationship of all the involved risks. At the same time,we have also built up a Mamdani algorithm and a fuzzy inference model based on the said fault-analysis tree. What is more,by altering the logic gate of the fault tree to the inference engine,we have also managed to convert the triangular membership function into the normal membership function,for the membership function can have a inference engine to be adopted,to which the designed fuzzy rules can be added to the inference engine when the relative importance of basic events should be taken into consideration. The correction function has been adopted to figure out the differences between the input interval and the output one in the inference engine. This paper has chosen Matlab to build each inference engine and Simulink to establish the complete model of the fuzzy inference system for the purpose of calculation. Besides,we have also illustrated the method for calculating the basic event probability of the fault tree in this paper,including the probability value and fuzzy number,with the sub-events being input into the model to gain the explosion probability. And,finally,an example has been given to indicate its uses of the model based on the actual situation of the platform. Thus,comparing the calculated results by the fuzzy inference with the results of the probability of importance,it has made the description of the method adopted in this paper in a more scientific way. Thus,as a matter of fact,the method is qualified to fill the blanks of the traditional fault tree analysis method in solving the problem of the so-called relative importance,which tends to be more flexible and effective. Hence,the fuzzy inference system should enjoy a very nice application prospect by adding itself to the neural network.