| 摘要 |
PURPOSE:To make it possible to improve the accuracy of evaluation by providing a method wherein a neural net model is made to learn the frequency distribution of physical or electrochemical signals with respect to an object under the known state and the soundness of an unknown object is judged based on the model. CONSTITUTION:A sensor 2 is provided on the wall surface or in the inside of an object to measured 1. The signal transmitted from the object 1 is sensed. The signal is converted into a potential signal. Thereafter, the signal undergoes frequency spectrum conversion with a signal analyzer. With the shape, strength, dispersion and the like of the frequency spectrum as input layers, the cause and the degree of the transmission of the signal from the object 1 are displayed on a display system 5 by using a neural net model. Thus, the evaluation accuracy can be improved. |
