دانلود رایگان مقاله انگلیسی یک شبکه بیزی برای حوادث حمل و نقل مواد خطرناک با کمک ارزیابی زمان - الزویر 2018

Abstract

To improve the efficiency of emergency rescue in transportation accidents of hazardous materials(HAZMAT), a Bayesian network(BN) was developed in this paper to estimate the accident handling time. Also, based on this BN, the difficulty of handling every types of accidents can be quantified. According to theoretical analysis and literature review, 7 nodes (season, time, type of road, type of HAZMAT, the former accident, the secondary accident and handling time) are used to set up the BN. The value of mutual information was calculated to refine the BN. A database of 902 transportation accident of HAZMAT cases was built up for Bayesian parameter learning. Based on the parameter learning of BN, the results were summarized as follow: (1) The BN could be used to estimate the probabilities of handling time in different periods which include ‘0 to 2 hours’, ‘2 to 4 hours’, ‘4 hours and more’. (2) The difficulty of each type of accident can be ordered as follow: rollover> rear-end> internal fault≈ impact> falling> tire fault> vehicle body fire. Leakage>combustion explosion.

5. Conclusion

Based on the statistics of all 902 accidents occurred during 2013-2016 and analyzing the data with the method of Bayesian network, we can get such conclusions:

1) Comparing with fault tree and bow tie model etc, Bayesian network can better illustrate the casual relationships of all nodes in detail.

2) With the help of Bayesian network, an estimate of the handling time of transportation accidents of HAZMAT can be obtained. We can use this model to predict the probability of totally handling the accident within 0-2h, 2-4h and more than 4 h with the preliminary information from alarm calls.

3) The posterior probability of adjusting node D via Bayesian network is 1, it can help us ration the difficulty of handling different kinds of former accidents and secondary accidents. The result is as followed: rollover> rear-end> internal fault ≈ impact> falling> tire fault> vehicle body fire. Leakage>combustion explosion.