Research On Occupancy Rate Of Street Parking Based On Neural Network

Using neural network to train time series data can more accurately predict the performance of time series in the future.The occupancy rate of street parking space changes with time,which is a set of time series data.By predicting the occupancy rate of street parking spaces in the future,we can reasonably allocate and utilize the resources of street parking spaces,make up for the defects of real-time broadcasting,and improve the management of street parking spaces.In this paper,for the mixed distribution time series with periodic changes,a combined time series prediction model based on K-means clustering algorithm and GRU neural network,K-GRU hybrid model,is proposed.The specific work is as follows:1.Carry out numerical simulation experiments in different scenarios,generate random data with R language,process the data with K-means algorithm,and finally use GRU neural network for training and prediction.The prediction performances of polynomial,Fourier,LSTM,GRU and K-GRU are compared,and the effects of different cluster numbers on the prediction results are analyzed.The results show that the proposed method can effectively reduce the prediction error,and the error is lower than the other four methods.Finally,the elbow rule is used to judge the best cluster number,determine the value of K,and obtain the most appropriate hybrid prediction model.2.According to the comparative experimental results,it is proposed to add attention mechanism to the training process of K-GRU prediction model,so that the prediction model can focus on dealing with important parts and save processing cost.The results show that the addition of attention mechanism can effectively reduce the prediction error.The experimental results show that the prediction error of K-GRU-attention is always the lowest.3.By analyzing the time series data of the real street parking space occupancy rate of merchants Road,Nanshan District,Shenzhen,it is concluded that the series has the characteristics of periodicity and mixed distribution.The K-GRU hybrid prediction model and the K-GRU model based on attention mechanism are applied to the street parking occupancy rate data of merchants road for empirical analysis to verify the feasibility of the method proposed in this paper in the real scene.By using K-GRU hybrid prediction model and K-GRU model based on attention mechanism to predict the occupancy rate of street parking spaces,we can get high-precision prediction results,provide decision-making basis for drivers’ travel,and also provide basis for the planning and allocation of street parking spaces.