An autonomous approach to road temperature prediction

Abstract

To provide ice warnings for road maintenance, a forecasting model has been developed that computes the state of the road several hours in advance. The model, which is called HS4Cast, runs autonomously and does not need human input. The current version of the forecast model features a fuzzy logic engine that determines the best forecast by evaluating certain patterns in the measured parameters. Using this forecast engine, predictions up to 24 hours in advance are possible with reasonable accuracy. If available, additional forecasts from a weather centre may be imported into the model so that the effects of precipitation are taken into account, this further reduces forecast errors. Statistics summarising the model's performance are presented using data from several measurement sites along an Austrian motorway, where HS4Cast has been in use since about 1990.