Advanced Watershed Water Level Prediction Method Based on an Enhanced Informer Model
Yuexin Liu, Weiwei Shou, Dongqiang Chen, Xuefeng Wang, Chu Zhang · 2024
With the rapid development of urbanization and industrialization, the problems of water pollution and water scarcity are becoming more and more serious, and climate change and human activities have led to the increasing unpredictability of water levels in river basins. Water level prediction is of great significance in water resources management and disaster prevention and mitigation. In this study, a watershed water level prediction method based on improved Informer is proposed. The parameters of the Informer model are optimized by improving the honey badger optimization algorithm (HBA) to enhance the accuracy of the water level prediction model. Specifically, Chebyshev chaotic mapping initialization, Gaussian variation strategy and lens imaging reverse learning strategy are introduced into the original HBA algorithm to avoid falling into local extremes, which effectively improves the ability of exploring and exploiting in the optimization process of the algorithm parameters. The experimental results show that the model proposed in this paper exhibits higher accuracy and stability in water level prediction, which is of great practical significance and long-term value for water resources management, flood control and disaster warning.