Research · Application
HUANG Shenglong1, YUAN Hongjun1, HU Lingyun2
Cotton yield prediction research is of great significance to improve agricultural production efficiency, protect farmers’ economic interests as well as promote the textile industry and national grain and cotton security. In this paper, a cotton yield prediction model based on Whale Optimization Algorithm and Bidirectional Long Short-Term Memory Network is proposed, aiming to optimize the key parameters of the BiLSTM model in order to improve the accuracy and stability of the prediction. By simulating the whale’s foraging behaviour, the WOA algorithm is able to search the parameter space efficiently, and to optimize the number of hidden-layer units of the BiLSTM and the learning rate of the superparameters, thus improving the prediction performance. In this paper, using the national cotton yield data from 1980 to 2024 for empirical analysis, the results show that the WOA-BiLSTM model has a smaller prediction error and higher prediction accuracy than the traditional subjective parameter selection BiLSTM model. The results not only verify the advantages of the WOA algorithm in BiLSTM parameter optimisation, but also show that the model can accurately reflect the changing law of cotton production, which provides a strong support for decision- making in cotton production and related industries. Through the application of the model, it can more effectively deal with the uncertainty in cotton production, and help the government, farmers and related enterprises to make more scientific and reasonable decisions in cotton yield prediction, policy making and market regulation.