An Adaptive Conjugate Gradient Neural Network–Wavelet Model for Traffic Incident Detection

Artificial neural networks are known to be effective in solving problems involving pattern recognition and classification. The traffic incident-detection problem can be viewed as the recognition of incident patterns from incident-free patterns. A neural network classifier must be trained first using incident and incident-free traffic data. The dimensionality of the training input is high, and the embedded incident characteristics are not readily detectable. This paper presents a computational model for automatic traffic incident detection using discrete wavelet transform (DWT), linear discriminant analysis (LDA), and neural networks. DWT and LDA are used for feature extraction, denoising, and effective preprocessing of data before an adaptive neural network model is used for traffic incident detection. Simulated and actual traffic data are used to test the model. For incidents with a duration of more than 5 minutes, the model yields a detection rate of nearly 100% and a false-alarm rate of about 1% for 2- or 3-lane freeways.