摘要 |
PROBLEM TO BE SOLVED: To prepare more useful support information at the time of using a train by predicting the change of a train congestion state using a train boarding rate by a method of neural network or the like to prepare support information at the time of selecting a train to board. SOLUTION: A boarding rate input means (a) is provided to input a train boarding rate at each station. The inputted train boarding rate is stored in a storage means (b), and the boarding rate result value is computed by a computing means (c) from the train boarding rate. The boarding rate predictive value is obtained respectively on the basis of weight coefficients in a predictive hierarchical neural network (e) and a learning hierarchical neural network (f), and the weight coefficient correction value is computed by a hierarchical neural network learning means (h). A boarding rate table is prepared on the basis of the train boarding rate at each station, and support information for selecting a train to board is prepared using this table and outputted. |