Possibilities of artificial neural networks in solving inverse problems of electrical prospecting by the method of vertical electrical sounding
| Autor: | R.N. Petrosyan, N.V. Ryzhov |
|---|---|
| Jazyk: | ruština |
| Rok vydání: | 2024 |
| Předmět: | |
| Zdroj: | Вестник Камчатской региональной ассоциации "Учебно-научный центр". Серия: Науки о Земле, Vol 62, Iss 2, Pp 109-119 (2024) |
| Druh dokumentu: | article |
| ISSN: | 1816-5524 1816-5532 |
| DOI: | 10.31431/1816-5524-2024-2-62-109-119 |
| Popis: | An algorithm for solving the inverse problem of electrical prospecting by the method of vertical electrical sounding (VES) using neural networks (NN) is presented. The use of NN is aimed at identifying complex patterns and dependencies that may not be available for traditional methods of quantitative interpretation of electrical prospecting data. The algorithm includes the formation of a training sample; training NN and directly obtaining solutions; combination of solutions found with the help of NN and selection of the optimal solution to the inverse problem of VES. The algorithm is tested on model data and practical materials in order to evaluate its capabilities. Geoelectric sections constructed using NN based on the results of field observations were compared with the results of quantitative interpretation performed in the «ZOND» program. The advantages and disadvantages of the created algorithm for solving the inverse problem of VES, as well as ways for its further development are characterized. |
| Databáze: | Directory of Open Access Journals |
| Externí odkaz: |
|