摘要
为了进一步提高海水养殖产量预测精度,考虑多因素对海水养殖产量的影响,文章基于变权组合预测模型,充分结合长短期记忆(Long Short-Term Memory,LSTM)神经网络预测模型、GM(1,N)预测模型和偏最小二乘回归预测模型等传统统计预测模型的优点,构建出一种基于灰色多元变权组合预测模型,并对山东省海水养殖总产量和分类产量进行了预测.实证结果显示,基于灰色多元变权组合预测模型对山东省海水养殖产量的预测精度高达99.13%,预测精度显著优于LSTM神经网络等各单项模型,并能综合于 LSTM神经网络、GM(1,N)预测模型和偏最小二乘回归预测模型的优点,弥补各单项模型的不足,提高预测精度和可靠性.根据预测结果,到 2025 年山东省海水养殖产量仍将保持良好发展,海水养殖产量将达到 579.28×105 t,平均增长速度为 3.11%,而鱼类、甲壳类、贝类、藻类以及海参海水养殖产量将分别达到 6.27×105、26.27×105、445.83×105、68.65×105 和 9.5×105 t.
Abstract
To further improve the accuracy of seawater aquaculture yield prediction and consider the impact of multiple factors on seawater aquaculture yield,this study tested a variable weight combination prediction model by fully combining the advantages of the conventional statistical prediction models,such as the Long Short-Term Memory(LSTM)neural network prediction model,the GM(1,N)prediction model,and the partial least squares regression prediction model,and by constructing a gray multiple variable weight combination prediction model.As a result,the total and classified yields of seawater aquaculture in the Shandong province were predicted.The empirical results indicated that the prediction accuracy of the gray multiple variable weight combination prediction model for seawater aquaculture production in the Shandong province is as high as 99.13%,which is remarkably higher than those of various single-item models,such as the LSTM neural network.This model also integrated the advantages of the LSTM neural network,GM(1,N)prediction model,and partial least squares regression prediction model,thereby making up for the shortcomings of the individual single-item models and improving the prediction accuracy and reliability.Based on the predicted results,total production of marine aquaculture products in the Shandong province will maintain good development at least until 2025,reaching 5.7928 million tons at an average growth rate of 3.11%.The production of fish,crustaceans,shellfish,algae,and sea cucumber in marine aquaculture is estimated to reach 62 700 tons,262 700 tons,4.458 3 million tons,686 500 tons,and 95 700 tons,respectively.
国家科技期刊平台
NSTL
万方数据