Abstract
The development of an adaptive incident detection algorithm based on Probabilistic Neural Network (PNN) and optimised using a Genetic Algorithm for use in decision making by Traffic Management personnel is described. The training and testing of the GAPNN is described and the results compared with those obtained for the PNN alone. Since it is more accurate, the GAPNN detected more incidents than the PNN in a given time period after transfer to a new site. The better performance of the GAPNN tested before retraining than that of the PNN in the same circumstances is however at the expense of a longer period of training. (A*)