Research On Forecasting And Analysis Technology For Spatio-Temporal Data In The Intelligent Transportation System Based On Machine Learning

The intelligent transportation systems are the core of a smart city.It is of great significance for realizing intelligent traffic scheduling and path planning to analysis traffic data with spatio-temporal characteristics and obtain valuable knowledge.Based on machine learning and deep learning methods,we focus on spatio-temporal prediction,pattern discovery technologies for spatio-temporal data such as passenger flow,traffic flow data on intelligent transportation systems.We deeply extract temporal,spatial and other hidden features with spatio-temporal data to achieve improved performance.Our research covers the main traffic scenes such as the urban road system,the light rail system and the bus system.The characteristics of each scene and the relationship between them are fully considered.In particular,the thesis proposes a dynamic graph convolutional neural network for traffic forecasting,which is of great significance for similar problems.Therefore,the thesis has a good application prospect and academic value.Our main contributions include:1)This thesis presents a novel hybrid(DTMGP)model to accurately forecast the volume of passenger flows multi-step ahead with the comprehensive consideration of factors from temporal,OD spatial,frequency and self-similarity perspectives.We first apply discrete wavelet transform(DWT)to decompose the traffic volume series into an appropriation component and several detailed components.Then we propose a more efficient tracking model to forecast the appropriation component and a novel Gaussian Process model to forecast the detailed components.The forecasting performance is evaluated with real-time passenger flow data in Chongqing,China.Simulation results demonstrate that our hybrid model can achieve on average 20%-50% accuracy improvement,especially in rush hours.2)Graph convolutional neural networks(GCNN)have become an increasingly active field of research.It models the spatial dependencies of nodes in a graph with a pre-defined Laplacian matrix based on node distances.However,in many application scenarios,spatial dependencies change over time,and the use of fixed Laplacian matrix cannot capture the change.To track the spatial dependencies among traffic data,we propose a dynamic spatiotemporal GCNN for accurate traffic forecasting.The core of our deep learning framework is the finding of the change of Laplacian matrix with a dynamic Laplacian matrix estimator.To enable timely learning with a low complexity,we creatively incorporate tensor decomposition into the deep learning framework,where real-time traffic data are decomposed into a global component that is stable and depends on long-term temporal-spatial traffic relationship and a local component that captures the traffic fluctuations.This thesis proposes a novel design to estimate the dynamic Laplacian matrix of the graph with above two components based on our theoretical derivation,and introduce our design basis.The forecasting performance is evaluated with two real-time traffic datasets.Experiment results demonstrate that our network can achieve up to 25% accuracy improvement.3)This thesis collects two months of real data sets of a city in China,including passenger flow data in the bus and light rail system,covering the main trip modes.Based on the complex network and machine learning methods,the thesis focuses on the spatio-temporal pattern,the trip displacement distribution,the trip duration distribution,the trip interval distribution,and the community migration of various types of people in the transportation system.To the best of our knowledge,this thesis is the first systematic study of mobility patterns for different groups of people(including ordinary people,the elderly,students,the disabled etc).These findings can provide a sufficient theoretical basis for the relevant departments to further improve urban planning and design.