Date of Award
Doctor of Philosophy
Civil, Construction, and Environmental Engineering
In Intelligent Transportation System (ITS), the emergence of big data technologies makes a wide variety of data being collected and accessed possible. With these multi-source traffic data available, an intelligent diagnosis for the transportation system is feasible and necessary. This dissertation explores several data-driven methods for three types of system diagnosis: a) anomaly diagnosis for highway system, b) performance diagnosis for signalized intersection, and c) crash detection and risk diagnosis for real-time traffic conditions.
The first study focuses on the system health and proposes a systematic data mining technique to diagnose highway system anomalies in a batch-processing fashion. Built on the concepts of symbolic dynamics, a spatiotemporal pattern network (STPN) architecture is developed to capture the system characteristics. This novel spatiotemporal graphical modeling approach is shown to be able to extract salient time series features and discover spatial and temporal patterns for a traffic system, which could provide anomaly diagnosis for highway system.
The second study focuses on system efficiency. To improve the efficiency in signalized intersection system, this study addresses some shortcomings in current automated traffic signal performance measures (ATSPM), which lack of data quality control, demand pattern assessment, and intelligent control support. This study proposes a data-driven intelligent traffic signal performance measures (ITSPM) embedding machine learning method and data visualization to diagnose the system performance.
The last study focuses on system safety. This study proposes a deep learning approach to identify crashes from traffic characteristics. Several deep neural network structures and training operation combinations are examined and applied to achieve higher classification performance. By predicting the probability of a crash on highway traffic, this study helps to diagnose the risk in the system based on big traffic data.
Overall, this dissertation studies on different data-driven methods to diagnose the transportation system regarding system health, efficiency and safety, contributes to the solutions of transportation system diagnostic problem with big data in ITS, provides decision support for practitioners.
Huang, Tingting, "Big data driven diagnostics for intelligent transportation systems" (2018). Graduate Theses and Dissertations. 17211.