The invention discloses a signal intersection average passing time estimation method based on data fusion of two types of floating cars, comprising the steps as follows: acquiring original GPS data of buses and taxies; extracting running track data of all floating cars entering a target signal intersection area, carrying out steering feature identification, and building a GPS data set; carrying out model feature identification, and building a bus passing time data set and a taxi passing time data set; extracting bus and taxi passing time data in each direction of the intersection; and calculating the average passing time. According to the signal intersection average passing time estimation method provided by the invention, both GPS data of large cars and GPS data of small cars are taken into consideration in analysis of urban road passing time based on GPS data. Thus, the demand of real-time application is met, a high-accuracy signal intersection average passing time estimation method is realized, and the service level at signal interactions is improved. The average passing time of a road section can be reflected more truly.