Study On Short-term Prediction Of Parking Slots Of Intelligent Car Park

With the soaring car ownership,parking difficulties and traffic congestion is a common problem.Part of the problem is due to invalid traffic caused by an invalid parade when the driver is looking for a parking space.The parking guidance system can provide drivers with guidance information to help the driver make a judgment as soon as possible to find parking spaces,greatly reducing parade time,reducing ineffective traffic and improving urban traffic conditions.Short-term effective parking spaces prediction is the difficulty of parking-induced systems,for this problem,this paper studies the prediction technology,compares the relevant forecasting model,studies the methods and algorithms of BP neural network,and puts forward the effective parking spaces prediction method of BP neural network based on genetic algorithm optimization.The main work includes:(1)For the prediction of short-term effective parking spaces,the GM(1,1)model and the BP neural network prediction model are established,and the specific parameter setting in the model establishment process is described in detail.In the Matlab,we combined with the example of the simulation experiment,according to the error criteria,we compare and analyze the experimental results.(2)Shortcomings of BP neural network is easy to fall into the minimum,slow convergence and prone to shock and so on.The BP neural network is optimized by using the genetic algorithm with global optimization,to found the its optimal weights and thresholds.The results of global optimization and local search are also ensured to make up the traditional BP neural network.Value of the defect.The process of parameter setting of optimization model is described in detail.(3)Combining the parking data of a parking lot in Chengdu,we establish BP neural network prediction model and GABP prediction model respectively in Matlab.The experimental results show that the accuracy of GABP prediction model is higher than that of BP neural network model,and the convergence rate is stronger than that of BP neural network model.