| With the fast increasing of vehicle population, the demand for parking is growing, however the construction speed of vehicle parking lot has lagged behind the growth speed of motor vehicle, which causing a serious shortage of parking. The conflict between parking supply and demand is increasingly serious. When the Large-scale construction of parking lot is restricted by factors such as land and capital, the rational use and planning of parking resources, and intelligent parking guidance are the fundamental method to solve the growing congestion of parking. Promoting the planning and construction of parking lot and facilities, and achieving market-orienting and industrialization of parking industry are the main method to solve the parking problem.The marketing and industrying of parking lot in metropolitan cities can not be implemented without construction and planning of parking facilities, and these are inseparable from the support and guidance of scientific knowledge. The investigation and collection of information are necessary before the parking lot planning, and data mining can extract implicit, credible,’and potentially useful model from a large number of data information. Under the current circumstances, the use of data mining technology can play a sufficient supporting role in the parking decision support.For the actual traffic flow data of parking lot, the thesis uses the statistics and mathematical methods to preprocess the original data, and to research some kind of similarity measurement in the data set of a single parking lot. And finally the parking vehicle traffic has been normalized and analysed by cluster. The main work is as follows:1. Designing the database platform, importing the original data into the platform, and preprocessing the data, such as the judging of the error data and the filling of the missing data.2. Sampling the vehicle traffic and expressing with the time series. Comparing their features between different days of a single parking lot and verifying the similarity by defining similarity measure function.3. In collection with sampled time series of different parking lot in a month, firstlly normalizing the data, then using the clustering method to classify the normalized data set. Combined with actual location of the parking lot and classified results, analysising the classified results and verifying the effectiveness of the clustering method. |
