A fusion model for multi-source detect data of section average velocity based on BP network

As section average velocity that is one of the most important traffic flow parameters has a wide range of sources of data, different sources of data vary in standards, advantages and disadvantages. Single-detect equipment can´t meet the needs of multi-purpose and different environments. What´s more, under certain conditions, the detector performance is defective, and it can´t get rich and high-quality section average velocity information. The paper will try to use B-P neural network to do date fusion, to get more realistic traffic flow speed information, to provide a basis for traffic management, control, and induction measures. Taking Beijing as the research background, the expressway section average velocity of multi-source data is adopted to do data fusion in the final section of the study.