复杂环境下多模态交通警示信息推荐算法仿真

Simulation of Multi-Mode Traffic Warning Information Recommendation Algorithm in Complex Environment

单一环境数据会导致推荐结果不够准确,为缓解交通拥堵问题,便于驾驶员实时掌握交通警示信息,提出一种多模态交通警示信息个性化推荐算法。利用毫米波雷达传感器和摄像头传感器组成多模态交通数据采集架构。架构包括采集、推理、服务与查询四大模块;构建数据融合模型,设计融合函数,提高数据采集的精准性;将交通网络当作一个有向加权图,分别从时间和空间两方面分析交通状态的传播过程;利用粗糙集规则提取的方式填充警示信息评分表,通过矢量夹角的余弦值计算信息相似度,根据计算结果预测待推荐信息评分,将得分较高的警示信息推荐给有需求的用户,实现交通警示信息个性化推荐。实验结果表明,所提方法能够采集更加准确的交通环境信息,推荐的结果能够满足用户需求,辅助用户避开故障路线,节约行驶时间。

Single environmental data may lead to inaccurate recommendation results.In order to alleviate traffic congestion,a multi-modal personalized recommendation algorithm for traffic warning information was presented.First of all,millimeter-wave radar sensor and camera sensor were used to construct a multi-mode traffic data acquisition system.This architecture included four modules:acquisition,reasoning,service and query.Then,a data fusion model was built.Moreover,a fusion function was designed to improve the accuracy of data collection.The traffic network was regarded as a directed weighted graph,and the propagation process of the traffic state was analyzed from aspects of time and space.Furthermore,the warning information scoring table was filled by using the method of rough set rule extraction.Meanwhile,the information similarity was calculated through the cosine value of the vector angle.According to the result,the score of the information to be recommended was predicted,and then the warning information with high score was recommended to the user who needed it.Finally,the personalized recommendation of traffic warning information was achieved.Experimental results show that the proposed method can collect more accurate traffic information.And the recommended results can meet user needs.In addition,the method helps users avoid fault routes,and save driving time.