To provide a method and a device for predicting road-surface temperature or a road-surface state which can make it speedy and accurate to take countermeasures against ice and snow in the window season by accurately predicting the road-surface temperature or road-surface state at a prediction object point several hours later by using meteorological data at the prediction object point in several past hours from the current time by using a hierarchical type neural network capable of higher-order nonlinear analysis.
This device comprises a meteorological sensor part 1 which detects meteorological data needed to predict the road-surface temperature at the prediction object point, a data storage part 2 stores meteorological data outputted from the meteorological sensor part for several past hours from the current time, a road-surface temperature prediction part 3 which predicts the road-surface temperature at the point several hours later by using the hierarchical type neural network having learnt by using respective pieces of tutor data several hours ahead on the basis of the meteorological data of several past hours from the current time outputted from the data storage part, and a display part 4 which displays the prediction result of the road- surface temperature outputted from the road-surface temperature prediction part in specific form.
ASAI RITSUO
YAGI TADAO
YAMADA MUNEO
UEDA KOJI