A Performance evaluation of neural network models in traffic volume forecasting

Public transportation trip flow modeling with generalized regression neural networks

Modelling public transport trips by radial basis function neural networks

Research into advanced traveler information systems (ATIS) for rural roads is limited However, highway agencies expect to implement intelligent transportation systems (ITS) in both urban and rural areas In this paper, genetic algorithms (GAs) are used to design both time delay neural network (TDNN) models as well as locally weighted regression (LWR) models to predict short-term traffic for two rural roads in Alberta, Canada A top-down refinement was used to study the interactions between modeling techniques and underlying data sets for obtaining highly accurate models It is found that LWR models achieve faster accuracy improvement than TDNN models over the refinement process Compared with previous research, the models proposed here show higher accuracy The average errors for the best LWR models obtained through the model-refining process are less than 2% in most cases For refined TDNN models, the average errors are usually less than 6–7% The resulting models indicate a level of high robustness ove

Refining Genetically Designed Models for Improved Traffic Prediction on Rural Roads

The objective of this study is the development of the short-term prediction models to predict average spot speeds of the subject location in the short-term periods of 5, 10 and 15 minutes respectively. In this study, field data were used to see the comparison of the predictability of Regression Analysis, ARIMA, Kalman Filtering and Neural Network models. These field data were collected from image processing detectors at the urban expressway for 17 hours including both peak and non-peak hours. Most of the results were reliable, but the results of models using Kalman Filtering and Neural Networks are. more accurate and realistic than those of the others.

Short-term travel speed prediction models in car navigation systems

This paper applies three state estimation algorithms to an hourly traffic volume system. The observation values of traffic volume by detectors are contaminated by errors depending on type, setting, sensitivity, and so on. From the statistical analysis of hourly traffic volume data, it is assumed that hourly traffic volumes vary depending on the day of the week and time. The hourly traffic volume system is described by a linear time-varying discrete dynamic system. State and observation equations are described in such linear equations that the state variables are additively affected by the state and observation noises respectively. In order to remove errors caused by type, setting, and sensitivity of a detector, the authors apply such state estimation algorithms as the Kalman filter, the fixed-interval smoother, and the MIPA Kalman filter to the hourly traffic volume system. The Kalman filter and the fixed-interval smoother are optimal for the linear dynamic system with Gaussian noises. The MIPA Kalman filter, based on the M-interval polynomial approximation method, is a robust Kalman filter for the linear dynamic system with non-Gaussian noise. Three state estimation algorithms are applied to the hourly traffic volume system using the observation data obtained from 7 a.m. to 7 p.m. in Fukuyama, Japan. Estimates of the Kalman filter are fairly improved compared with the observation values, and estimates of the fixed-interval smoother are more accurate than the Kalman filter estimates. Estimates of the MIPA Kalman filter are the most accurate and robust as compared with other estimates.

H Shimizu

Papers

Applications of state estimation algorithms to hourly traffic volume system