| In recent years,with the continuous acceleration of the urbanization process,China's tourism industry has developed rapidly.The rapid growth of passenger traffic has brought a series of challenges to the environmental and safety management of various tourist attractions.Accurate forecasting and early warning of passenger flow in the scenic area will help the scientific management of the scenic area and reduce the risks brought by dense crowds.The scientific management of scenic areas reduces the risks brought by dense crowds.Therefore,the establishment of an accurate passenger flow forecasting model and an efficient passenger flow early-warning platform are of great significance for the formulation of tourism development policies,the optimization of the allocation of tourism market resources,and the implementation of tourism enterprise strategies.Aiming at the problems of single source of passenger flow forecasting data in the past,and the dependence of traditional time series models on the stability of data distribution,this paper proposes a method of forecasting passenger flow in the scenic area based on a hybrid deep neural network model that takes into account the multi-source public transport passenger flow around the area.A passenger flow early warning system was set up with the passenger flow forecasting model as the core to realize the practical application of comprehensive passenger flow early warning in scenic areas.The main research contents of this article:(1)Multi-source data cleaning and normalization.In order to perform passenger flow prediction in the core area of the time dimension,this paper first performs data cleaning on multi-source passenger flow data to remove noise data,and then extracts relevant data sets from the original data according to the scope of the study area.And convert it into multi-source passenger flow time series data.(2)Spatio-temporal analysis of multi-source passenger flow in scenic areas.Because the passenger flow of the scenic area is disturbed by many factors and has non-linear and non-stationary characteristics,this paper analyzes the passenger traffic of the multi-source public transportation in the core area of the scenic area and its surroundings from the two dimensions of time and space.The differences in distribution of traffic passenger flows,the law of periodic changes,and correlations.(3)Construct a passenger flow prediction model in the scenic area.In order to realize the short-term passenger flow prediction in the scenic area,this article takes Nanluoguxiang Scenic Area as an example.Based on the space-time characteristics of the environment and multi-source passenger flow,combined with deep neural network related research,a hybrid neural network passenger flow prediction model based on CNN-LSTM is designed.The model uses the processed data to construct a spatiotemporal dataset of the passenger flow in the scenic areas,uses the CNN network model to perform data fusion to extract feature vectors,and then inputs the feature vectors into the LSTM network model in time series to perform passenger flow prediction in the core area of the scenic areas.The experimentalresults show that the method has higher prediction accuracy than the traditional passenger flow prediction method ARIMA model and standard LSTM network prediction method,the passenger flow prediction accuracy is higher and it has certain robustness.(4)Design and implement a passenger flow early warning system.In order to serve the passenger flow prediction model for the actual scenic areas,this paper designs and develops a large passenger flow early warning system based on the B / S architecture in Nanluoguxiang.This system can help managers respond to large-scale passenger flows in a timely manner through data management and multiple visualization technologies.Gathering risks,while providing a basis for traffic conditions for finding the source of key passenger flows and effectively evacuating people,will help improve the overall management level of the scenic areas. |
PDF Full Text Request