The Research On The Prediction Of Traffic Flow In Key Places Based On Time Series Analysis

With the rapid development of modern society,more and more people flock to the cities to seek better development opportunities and raise the level of accident treatment in key places.The pre-judgment analysis of the flow of people is of great importance to the healthy development of the national economy and the sustained and steady social order significance.In this dissertation,the thesis make a systematic and in-depth study on the human traffic forecast in key urban areas,improve the theoretical system of human traffic forecast,and provide a scientific and reliable model for the scheduling of relevant departments.As the IMSI simulation data are selected to simulate the human-flow data for predictive analysis,the thesis first introduces the acquisition system and proposes a new research method to improve the transmission reliability of the acquisition system to ensure the accuracy and validity of the collected data.Then,the fuzzy time series model is taken as the focus of the study,and the improvement of the domain division is made.In the improvement,the methods of taking the values of left and right endpoints of the big classifications in the domain and the relative errors of the adjacent data in the same category get a new definition.The improved model and the traditional fuzzy time series model are respectively used to predict the collected data.The results of the experiment are evaluated and analyzed.The results show that the root mean square error of the predicted value of the improved model is 132,compared with the traditional multi-scale The RMSE 176 of the ratio-partitioning fuzzy model prediction has been significantly improved,that is,the improved fuzzy time series prediction model effectively improves the accuracy of the prediction.Finally,aiming at the characteristics of large fluctuation of collected data,it proposes a combined forecasting model,puts forward the time series and carries out effective processing based on ARIMA model's smoothing process,and then improves the algorithm based on fuzzy time series model Predict the value.By experimenting on the same set of data,the average relative error ? between the predicted value and the true value obtained from the improved fuzzy time series model is 2.68%,while the average relative error ? of the combined forecasting model is only 1.42%.The result shows that the combined forecast The model further improves the prediction accuracy.