Abstract:

In order to improve the navigational system of route planning with safety as the first factor, highlight the purpose of people-oriented, based on OD（Origin-Destination） route selection, and traffic accident and weather data from Guiyang, which are affecting the line risk factor, are taken into account synthetically based on the actual situation. The Poisson distribution model and Markov model are setting up to analyze the probability of traffic accident points and route probability under sunny, cloudy and rainy days as risk coefficient indexes, the optimal route is selected by using the maximum decision method according to the route safety factor. The results show that the frequency of traffic accidents is different under different weather conditions in Guiyang and the number of traffic accidents at each location obeys Poisson distribution. The first route recommended by the existing navigation software is not the optimal route, while the second route is the route with the greatest safety factor. The calculation of traffic accident probability based on Poisson distribution model and Markov model adds a humanized setting with safety factor as the first reference value to the existing navigation software, which has good social value and generalization performance.

Key words: Poisson distribution, Markov model, traffic accidents, safety factor, maximum decision

摘要：

为完善安全为第一因素的路线规划导航系统，突出“以人为本”的理念，基于OD（Origin-Destination）的路段选择，综合考虑贵阳市交通事故数据和天气数据两大影响线路风险系数的因素。建立泊松分布模型和马尔可夫模型，分析区域内交通事故易发点在晴天、阴天、雨天发生交通事故点概率和路线概率，构建风险系数指标。针对模型求得的路线安全系数采用最大决策法选出最优安全路线。结果表明，贵阳市不同天气下发生的交通事故频率有差异，检验每个地点发生交通事故数服从泊松分布；现有导航软件推荐的第一路线并不是最优路线，第二条路线才是安全系数最大的路线。基于泊松分布模型和马尔可夫模型计算交通事故发生概率，为现有导航软件中增添了安全系数为第一参考值的人性化设置，有较好的社会价值和推广性。

关键词: 泊松分布, 马尔可夫模型, 交通事故, 安全系数, 最大决策