A benchmark dataset for ensemble framework by using nature inspired algorithms for the early-stage forest fire...

Area of Science:

• Computer Science
• Artificial Intelligence
• Optimization

Background:

• Forest fires pose significant threats, necessitating efficient rescue operations.
• Dynamic optimization problems are crucial for real-time decision-making in emergency response.
• Existing benchmarks may not fully capture the complexities of dynamic forest fire rescue scenarios.

Purpose of the Study:

• To introduce a novel benchmark dataset for evaluating forest fire rescue algorithms.
• To facilitate the development and comparison of nature-inspired algorithms for dynamic optimization problems in rescue scenarios.
• To support research on minimizing key performance indicators like aircraft flight time and burnt forest area.

Main Methods:

• Development of a rescue ensemble comprising a simulator and a rescue algorithm.
• Integration of real-world map data from Google Maps and relevant operational parameters.
• Creation of a benchmark dataset with 10 distinct maps for algorithm evaluation.

Main Results:

• The benchmark dataset supports dynamic simulation of forest fire rescue operations.
• The rescue algorithm aims to simultaneously minimize the longest aircraft group flight time and the increase in burnt forest cost.
• The dataset provides a standardized platform for comparing algorithm performance.

Conclusions:

• The introduced benchmark dataset is valuable for advancing research in forest fire rescue optimization.
• It enables rigorous evaluation and comparison of nature-inspired and other optimization algorithms.
• The dataset contributes to improving the efficiency and effectiveness of early-stage forest fire rescue efforts.