摘要 |
PROBLEM TO BE SOLVED: To provide a method and an apparatus for estimating a wave propaga tion loss whereby a highly accurate estimation is carried out. SOLUTION: In a learning process, a central value calculation part 3 forms a central value sequence data. A multilayer neural network process part 6 is formed of, e.g. three artificial neural networks, and constituted so that an output value sequence of a plurality of output nodes corresponds to a central value sequence of propagation loss values at a plurality of detection points. An output node count determination process part 4 sets an output node count to correspond equally to a segment distance when a central value is calculated. The learning process is conducted with using a distance from a transmission station and a propagation condition value as input, data and the central value sequence data as teacher's data. At the estimation time, a required propagation condition value and a required distance from the transmission station are input to the artificial neural network after learning, so that an estimation value of the central value of the propagation loss is obtained. |