基于改进鲸鱼算法优化LSTM的粮油温度预测  

Grain and oil temperature prediction based on improved Whale algorithm optimized LSTM

作　　者：史红伟[1] 叶明昊 谢酶 武士奇 SHI Hong-wei;YE Ming-hao;XIE Mei;WU Shi-qi(College of Electronics and Information Engineering,Changchun University of Science&Technology,Changchun 130000,China;Bibiyou(Changchun)Technology Co.,Ltd.,Changchun 130000,China)

机构地区：[1]长春理工大学电子信息工程学院,吉林长春130000 [2]必必优(长春)科技有限公司,吉林长春130000

出　　处：《陕西科技大学学报》2024年第6期208-214,共7页Journal of Shaanxi University of Science & Technology

基　　金：吉林省自然科学基金项目(YDZJ202301ZYTS412)。

摘　　要：针对粮油温度预测问题,提出一种基于改进鲸鱼优化算法(IWOA)优化长短时记忆神经网络(LSTM)的粮油温度预测模型.针对传统WOA算法收敛速度慢和容易陷入局部最优等问题,提出了应用Logistic混沌映射、Levy飞行策略等方法来提升WOA算法的种群丰富度和搜索能力的方法.采用真实粮油温度值与多种模型预测值对比,IWOA-LSTM的MAE、RMSE比其他模型分别降低了13.64%~68.33%、6.06%~60.39%,R^(2)提高了1.65%~14.27%.结果表明,本文所提模型可以准确预测未来粮油温度变化趋势.This paper proposes a grain and oil temperature prediction model that optimizes the parameters of a Long Short-Term Memory(LSTM)neural network using the Improved Whale Optimization Algorithm(IWOA).To address the slow convergence and susceptibility to local optima issues of the traditional WOA algorithm,methods such as the application of Logistic chaotic mapping and Levy flight strategy were proposed to enhance the population diversity and search capability of the WOA algorithm.Through the comparison of real grain and oil temperature data with multiple models,the IWOA-LSTM model reduced the Mean Absolute Error(MAE)and Root Mean Square Error(RMSE)by 13.64%to 68.33%and 6.06%to 60.39%,respectively,compared to other models.Additionally,the coefficient of determination(R^(2))was improved by 1.65%to 14.27%.The results demonstrate that the proposed model accurately predicts future trends in grain and oil temperature changes.

关 键 词：粮油温度预测 鲸鱼优化算法(WOA) 长短时记忆神经网络(LSTM) 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]