摘要 |
<P>PROBLEM TO BE SOLVED: To reduce an error in an estimation model to allow precise location. <P>SOLUTION: A current waveform in an accident is extracted from the current waveforms observed by observation instruments 3, 4 arranged in an electric power line 2. An FFT part 5 converts the extracted current waveform from a data of a time domain into a data of a frequency domain. The converted data is classified into a plurality of clusters 7 by deterministic annealing (DA). A fuzzy inference neural network (FINN) 8 is learnt the cluster 7 by the cluster 7, using data included in the each cluster 7 as a learning data, to constitute the estimation model. The cluster 7 to which the input data belongs is found to estimate an accident point, using the FINN of the found cluster 7. <P>COPYRIGHT: (C)2006,JPO&NCIPI |