VMD-GA-BiLSTM-based monthly precipitation prediction method

• Main Authors: LI BINGJIE, LOU YUQUAN, LIU XIANKANG, YU XIA, ZHANG BOZHEN, CHEN XIAODA, SONG JIE, LIU DINGXIN, DUAN YONG
• Format: Patent
• Language: Chinese; English
• Published: 31.10.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; MEASURING; METEOROLOGY; PHYSICS; TESTING

The invention belongs to the field of deep learning and meteorological prediction research, and particularly relates to a monthly precipitation prediction method based on VMD-GA-BiLSTM. The method comprises the following steps of: 1, preprocessing data: cleaning dirty data, normalizing the data and reconstructing a rainfall data set; 2, decomposing the original rainfall data into a plurality of subsequences by adopting a VMD method; and a third step, establishing a BiLSTM network model, optimizing network parameters by using a genetic algorithm, inputting each subsequence into the model for training and prediction, and comparing and analyzing experimental results. By using the method, the time sequence characteristics of precipitation are autonomously learned, the instability of precipitation data is reduced, and the future monthly precipitation change trend is predicted. 本发明属于深度学习及气象预测研究领域，具体涉及一种基于VMD-GA-BiLSTM的月降水量预测方法。第一步骤：数据的预处理：脏数据清洗、数据归一化和降水数据集的重构；第二步骤：采用VMD方法将原始降水数据分解为若干个子序列；第三步骤，建立BiLSTM网络模型，并使用遗传算法优化