关键词: 自主式交通系统, 密度峰值聚类, 功能架构, 道路自动驾驶场景, 多属性文本

Abstract:

Autonomous transportation system (ATS) is a new generation of transportation system proposed in response to the new development trend of active intelligent transportation. To scientifically and reasonably construct the functional architecture of the ATS, an optimized Density Peak Clustering (DPC) algorithm for multiattribute text is proposed in this paper. Combined with the basic characteristics of the functional architecture of a traffic system, the algorithm converts multiattribute text into spatial dimension coordinates through improved term frequency-inverse document frequency algorithm and text vector space model. Gaussian function and decision value were used to optimize the DPC algorithm for clustering, and the clustering result was evaluated using a contour coefficient. To test the rationality of the algorithm, this paper uses the functional datasets of road-carrier operation service domain, traffic infrastructure management service domain, and traffic-safety management service domain in ATS to perform an analysis as an example and draws functional architecture diagrams according to the clustering results. The architecture diagram comprises four layers of autonomous perception, autonomous learning, autonomous decision, and autonomous response, thus forming a scientific analysis method for functional architecture in ATS application scenarios. The results of the example show that the proposed algorithm is robust and the average value of the contour coefficient of the example is 0.84. Compared with the original algorithm, the problem of difficulty in defining the clustering center in the process of clustering is solved. Compared with other architecture designs in the intelligent transportation system, the functional architecture is more hierarchical and logical. This optimization algorithm can promote the construction of the functional architecture of the new generation of transportation system and the development of the theoretical system of the ATS.

Key words: autonomous traffic system, density peak clustering, functional architecture, road self-driving scenarios, multiattribute text