It is crucial to keep an eye on the water levels in reservoirs in order for them to perform at peak, as they are one of the, if not, the most vital part in water resource management. The water stored is essential in providing water supply, generating hydropower as well as preventing overlasting droughts. Thus, efficient forecasting models are essential in overcoming the issues revolving around hydropower reservoir stations. This paper reviewed the previous research on application of machine learning techniques in forecasting water level in reservoirs. In this review, the discussed machine learning techniques are ANN, ANFIS, BA, COA, SVM, etc., and their main benefits, as well as the literature, are the main focus. Initially, a general study regarding the fundamentals of the respective methods were made. Furthermore, the affecting conditions of water level forecasting, as well as the common issues faced, was also identified, in order to achieve the best results. The advantages and distadvatanges of the algorithms are extracted. In conclusion, hybrid metaheuristic algorithm produced more efficient results. This review paper covered researches conducted from the year 2000 to 2020.

密切关注水库的水位以使其在高峰期发挥作用至关重要，因为它们是水资源管理中最重要的部分之一，如果不是的话。储存的水对于供水、发电以及防止持续干旱至关重要。因此，有效的预测模型对于克服围绕水电站水库站的问题至关重要。本文回顾了以往机器学习技术在水库水位预测中的应用研究。在这篇综述中，讨论的机器学习技术是 ANN、ANFIS、BA、COA、SVM 等，它们的主要好处以及文献是主要关注点。最初，对各自方法的基本原理进行了一般性研究。此外，还确定了水位预测的影响条件以及面临的共同问题，以取得最佳结果。提取了算法的优缺点。总之，混合元启发式算法产生了更有效的结果。这篇综述论文涵盖了从 2000 年到 2020 年进行的研究。