GASEL: Genetic algorithm-supported ensemble learning for fault detection in autonomous underwater vehicles

Autonomous Underwater Vehicles (AUV) have a significant role in ocean research and the marine environment with the aim of performing different scientific and commercial tasks. During an underwater operation, fault detection for AUV is essential for critical operational decisions such as precautions, maintenance, and repair. This study presents an approach, GASEL, which combines Genetic Algorithm (GA) and Ensemble Learning (EL), for fault detection of AUVs. The proposed GASEL approach is effective since the ratio of 97.70% of the feature vector was eliminated by using GA in the feature selection stage. Thus, this paper proposes an efficient way to solve this problem by removing redundant and irrelevant data, which usually reduces computational complexity and provides a better understanding of the data and learning model. The experiments were conducted on a real-world AUV dataset. The results show that the GASEL approach successfully predicted the faulty state of AUV by 99.70% accuracy and fault type detection (severe propeller fault, slight propeller fault, load increase fault, and sensor fault) with 98.96% accuracy. Furthermore, the results also showed that our method outperformed the state-of-the-art methods on the same dataset.