Studying the Method of Adaptive Prediction of Forest Fire Evolution on the Basis of Recurrent Neural Networks

被引：7

作者：

Kozik, V. I. ^{[1
]}

Nezhevenko, E. S. ^{[1
]}

Feoktistov, A. S. ^{[1
]}

机构：

[1] Russian Acad Sci, Inst Automat & Electrometry, Siberian Branch, Pr Akad Koptyuga 1, Novosibirsk 630090, Russia

来源：

OPTOELECTRONICS INSTRUMENTATION AND DATA PROCESSING | 2014年 / 50卷 / 04期

关键词：

computer simulation; forest fire; recurrent neural network; data assimilation; learning; Kalman filter;

D O I：

10.3103/S8756699014040116

中图分类号：

O4 [物理学];

学科分类号：

0702 ;

摘要：

A software system is presented for implementation of a fire model on the basis of a recurrent neural network, which ensures real-time simulation of fire evolution. The quality of traditional learning and learning based on the Kalman filter in experiments performed with the neural network is compared. It is demonstrated that the fire overcomes obstacles in the form of regions consisting of incombustible

引用

页码：395 / 401

页数：7