Mobile Characteristics Analysis And Traffic Prediction Of Street Crowd Based On Mobile Signaling

With the accelerating process of urbanization in China,urban population continues to grow.According to the data of Beijing Municipal Bureau of statistics,by the end of 2019,the permanent resident population in Beijing will reach 21.536 million,and the urbanization rate of the permanent resident population will reach 86.6%.The high density of population will inevitably produce a large number of crowded places,and the key blocks with multiple composite functions are one of the important ones.And with the vigorous development of tourism,the key blocks often have a short-term surge in the flow of people.This kind of crowd gathering phenomenon often leads to major safety accidents.Due to the large permanent population and complex population composition in key blocks,the population change has obvious temporal and spatial characteristics.Therefore,how to analyze the behavior patterns of the people in key blocks,accurately predict the flow of the people in the blocks,and reduce the safety risks caused by the large number of people gathering are of great significance to the prevention of safety accidents in key blocks,and have become an urgent problem to be solved.The real-time passenger flow of key blocks has the characteristics of randomness,complex personnel composition,and non-linearity,with many influencing factors.The neural network model has strong nonlinear fitting capabilities that traditional time series data prediction models do not have.It provides an efficient and accurate method for crowd flow prediction in densely populated areas.In order to achieve accurate prediction and real-time warning of passenger flow in crowded scenic spots,this paper constructs a classification model of crowd movement behavior patterns based on mobile phone signaling data.By identifying crowd movement behavior patterns,a loop based on multi-source input and attention mechanism is adopted.The neural network crowd prediction method,which is used for crowd pattern recognition and passenger flow prediction in key neighborhoods,solves the problem of crowd behavior pattern classification in a small area and improves the accuracy of crowd flow prediction.And take the Nanluoguxiang block as an example to classify different types of crowd movement behavior patterns and predict crowd flow,and finally verify the effectiveness of the method through multiple sets of comparative experiments.The main research contents are as follows:(1)Design of preprocessing algorithm for mobile phone signaling data.According to the mobile phone signaling data to obtain the characteristics of the movement of people and the law of temporal and spatial changes,the format conversion and data quality enhancement of the mobile phone signaling data are preprocessed to obtain high-quality mobile phone signaling data to facilitate the crowd identification and identification of mobile phone users.(2)Construct a model of discrimination and classification of crowd movement behavior patterns.According to the user's behavior trajectory and residence time characteristics,a crowd movement behavior pattern recognition model based on mobile phone signaling data is constructed.Through the research on the processing algorithm of mobile phone signaling big data,a combination of K-Means and the shortest common subsequence algorithm is proposed.The clustering algorithm is used to calculate the flow of people in different mobile behavior patterns,which improves the efficiency and accuracy of using mobile phone signaling data to count different types of people.(3)Propose a crowd flow prediction model in scenic spots based on crowd behavior model identification.The result of crowd movement feature classification is used to generate multi-source time series data,which is used as the training input data of the passenger flow prediction model.Using the attention mechanism and the long and short-term memory neural network model,the block crowd flow prediction algorithm is designed and the model is optimized.