Algorithm for using artificial intelligence in predicting fire danger in the Semey forest in Kazakhstan

Absalyam Kuanysh, Khuralay Moldamurat, Cengiz Hajizadeh, Kunnur Dosimbayeva, Assel Atyzova

2025

Closed access Metadata (CSL-JSON)
DOI 10.1109/sist61657.2025.11139172
Data confidence 85 high
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

This paper presents an algorithm for utilizing artificial intelligence (AI) to predict forest fire hazards in the Semey region of Kazakhstan. The study analyzes the primary factors influencing wildfire occurrence, including key meteorological parameters such as temperature, humidity, wind speed, and atmospheric pressure, as well as ecological characteristics specific to the region. A mathematical model has been developed to estimate the probability of fire occurrence, and an implementation algorithm is proposed in the MATLAB software environment. Machine learning techniques and neural networks were applied to forecast fire hazard levels.The scientific novelty of this research is highlighted by the following contributions:–A customized Multilayer Perceptron (MLP) neural network model was developed, specifically optimized for the climatic and ecological conditions of the Semey region. This localization significantly improved prediction accuracy compared to general-purpose models.–The model incorporates additional non-standard parameters, such as atmospheric pressure and vegetation moisture, which are typically not included in conventional indices like FWI and FFMC.–A comparative analysis with classical fire hazard assessment methods demonstrated that the proposed AI-based model achieved up to 10–15% higher accuracy on certain days.–A hybrid approach integrating neural network prediction with fuzzy logic (via an ANFIS framework) was implemented to enable a comprehensive and adaptive fire risk assessment under conditions of uncertainty.–For the first time in this context, real sensor data was utilized within a virtual reality (VR) simulation environment to visualize and analyze hazardous fire development scenarios in real time.–An automatic alerting mechanism was designed to notify emergency response services when critical risk thresholds are reached, enhancing proactive intervention capabilities.The results demonstrate the high effectiveness of the proposed approach in identifying high-risk zones and support its potential for improving rapid response mechanisms to wildfire threats.