Available Parking Space Prediction Based On Phase Space Reconstruction And CS-SVM

In recent years,with the rapid development of China's economy and automobile industry and the continuous improvement of people's living standards,the number of private cars owned by Chinese citizens has increased.The resulting urban traffic problems have become more and more serious,among which the problem of difficult parking has become increasingly prominent.Therefore,it is reasonable to provide the driver with the prediction information of available parking space and parking guidance for the driver in the parking lot.It can effectively reduce the time for drivers to find available parking spaces outside and inside the parking lot,improve the utilization of parking spaces,and alleviate the pressure of urban traffic congestion.In response to the above problems,this thesis conducts the following main research work:Firstly,through the analysis of the current domestic and foreign research status of parking space prediction,it is proposed to use support vector machine(SVM)to predict available parking space.To slove the influence of the parameters in the support vector machine on the prediction accuracy of the model,the cuckoo search(CS)algorithm is used to optimize the parameters.To slove the problems of low accuracy and slow convergence of the traditional CS algorithm,the traditional CS algorithm is improved,and 7 test functions are used to compare and analyze with the traditional CS algorithm.Finally,the SVM parameters are optimized by the improved CS algorithm,and the CS-SVM available parking space prediction model is proposed.Secondly,the characteristics of available parking space changing with time are analyzed.Since the available parking space sequence data is affected by various random factors,it shows irregular changes.In order to mine the information implicit in the available parking space time series,phase space reconstruction is used to study the nonlinear characteristics of available parking space time series,and a CS-SVM available parking space prediction model based on phase space reconstruction is proposed.In order to verify the feasibility and applicability of the model,compared with CS-SVM and wavelet neural network,the prediction performance of the above model is evaluated through evaluation indicators.Finally,in the parking guidance inside the parking lot,the optimal parking space selection model is constructed from the perspective of the driver,and the arrival path of the optimal parking space from the entrance of the parking lot is reasonably planned by Dijkstra algorithm based on hierarchical search of road weight.An underground parking lot is taken as an example for experimental simulation,and compared with the traditional Dijkstra algorithm to verify the search efficiency of this method and the rationality of decision-making route.