An investigation was made as to how short-term traffic forecasting on motorways and other trunk roads is related to the density of detectors (e.g. inductive loops). Forecasting performances with respect to different detector spaces have been investigated by applying pruning techniques to the input variables used for neural networks. Simulated data and field data in different geographical locations were used in the work to evaluate the reality, reliability and transferability of the methodology. It was concluded that, on the basis of current evidence, a detector spacing of 1km may be optimal. Increasing coverage to a spacing of 500m gives little extra benefit and may actually be counterproductive in certain circumstances. Algorithm developers are tempted to use all available data, when a more streamlined approach often gives better results. The question of incident detection was also briefly considered, as it is closely related and is likely to use the same equipment. (A)
Abstract