The traffic information forecasting functional study device, the traffic information forecasting device, the traffic information fluctuation doctrine acquisition device and

摘要

PROBLEM TO BE SOLVED: To acquire a rule to predict future traffic information from data which are the closest to the time-sequential data of a traffic information vector value with a traveling time and the number of vehicles related with a plurality of blocks as components for prediction, and to acquire a variation law among these many variables as a rule.;SOLUTION: Data are read from a traffic information data storing device 11, and a cyclic fluctuation pattern included in the data is removed by a cyclic component removing device 12, and prediction functions are learnt by a method to optimize an information quantity standard by a prediction function estimating device 13. The parameter to designate the learning prediction function is read for the data excluding the fluctuating pattern, and the data are inputted to the prediction function designated thereby so that a prediction value can be calculated, and a prediction value added with the cyclic fluctuating pattern is outputted by a cyclic component adding device 18. The parameter of the learnt AR model is read by a prediction function interpreting device 15, and inter-variable correlation relations are extracted, and those correction relations are displayed on a map by a traffic information variation law display device 16.;COPYRIGHT: (C)2006,JPO&NCIPI