Floating Car Data Ming And Its Application In Path Planning

With the rapid development of economy and technology, the possession of cars in large and middle cities is increasing, which leads to the traffic congestion. Obviously, carrying out the public planning will help to ease the traffic congestion. It is a key issue if we can plan the bus route in a reasonable way so that more people can travel in a convenient way. The traditional method of bus route planning mainly relies on manpower survey. Although this method has been proven to be feasible, quite amount of time and material resources were taken. In addition, the traditional method cannot adapt to the rapid development of the city. As a consequence, the dynamic route planning for the bus based on the GPS data of the floating car brought into being.The taxi drivers are the “expert” in transportation planning, acquiring a lot of hidden city traffic information through mining the collected GPS data. According to the collected huge amounts of GPS taxi data, this thesis proposes a night bus route planning method. This thesis proposes a night bus route planning method. First, in data preprocessing including the data of GPS and GIS, a weighted map matching algorithm is proposed which has a low sampling rate to match the data of GPS and GIS. Using the map matching algorithm, we obtain the average time, average frequency, average speed and the "hot" area and the "hot" path. We use these data to make a "hot" road network.Second, on the hot spot, the candidate stations set of the hot spot is determined by the aggregation and separation algorithm of the hot spots. To determine the departure destination(O-D pairs) form a bus route set, a series of rules is applied to simplify the complex bus routes set for the effective bus routes.At last, to improve the shortcomings of the large amount of calculation in the traditional way, this thesis introduces a new idea to generate the bus route based on the Naive Bayes classification search algorithm. In order to overcome the weakness which the Naive Bayes hardly to select a right training set, we analyze the relevance of the number of passengers and the running time of the bus, the correlation of correlation of heuristic search algorithm is proposed. In addition, to improve the efficiency of the algorithm, we improved the algorithm and proposed the Correlation heuristic search algorithm.This simulation data of this thesis is based on the GPS data and map data of Beijing. The simulation experiments are carried out by simulation software and programming software. The proposed algorithm is compared with internal algorithm and the classical algorithm from the passenger number, hours of operation, and the elapsed Station number and other aspects of the comparison. The experimental result shows that the correlation heuristic searching algorithm which takes the relevance among candidate bus stations into consideration produces bus route with the maximum number of passengers in setting time.