基于多目标蝗虫优化算法的全国棉花产量预测

National cotton yield forecasting analysis based on multi-objective locust optimisation algorithm

随着全球气候变化和农业发展的影响,棉花产量的预测分析对于农业规划和资源配置至关重要。为了对全国棉花产量进行更精确的预测,本文提出了一种多目标蝗虫优化(MOGOA)组合预测方法。首先运用ARIMA时间序列模型、最小二乘支持向量机LSSVM模型、循环神经网络RNN模型3种单项模型对2009-2023年全国棉花产量数据进行预测。然后,通过多目标蝗虫迭代优化过程,得到了一组最优解,并将单项模型预测结果与组合预测方法预测结果相对比。通过实例验证,运用多目标蝗虫优化的组合预测方法预测结果误差更小、拟合程度更高,证明了该模型在实际应用中具有良好的价值,更好地反映出棉花产量的实际变化情况。最后使用该方法对2024—2026年的全国棉花产量进行预测,为棉花产业发展提供参考。

With the impact of global climate change and agricultural development,forecasting and analyzing cotton yields is crucial for agricultural planning and resource allocation.In order to provide a more accurate prediction of national cotton yield,a multi-objective locust optimal combination forecasting is proposed.Three single models,ARIMA time series model,Least Squares Support Vector Machine LSSVM model,and Recurrent Neural Network RNN model,are firstly applied to forecast the national cotton production data from 2009 to 2023.Then,a set of optimal solutions were obtained through the multi-objective locust iterative optimization process,and the single model prediction results were compared with the prediction results of the combined prediction method.It is verified through examples that the combined prediction method using multi-objective locust optimisation predicts results with smaller error and higher fitting degree,which proves that the model has good value in practical application and better reflects the actual changes of cotton production.Finally,using this method to forecast the national cotton production in 2024-2026 can provide a reference for the development of the cotton industry.