基于ARIMA-IPOA-CNN-LSTM的太湖水体溶解氧浓度预测模型  

Prediction Model of Dissolved Oxygen Concentration in Taihu Lake Based on ARIMA-IPOA-CNN-LSTM

作　　者：杨焕峥 崔业梅[1,2] 徐玲 薛洪惠[2,4] YANG Huan-zheng;CUI Ye-mei;XU Ling;XUE Hong-hui(Jiangsu Research and Development Center of Application Technology for Wireless Sensing System,Wuxi 214153,China;Wuxi Vocational Institute of Commerce,Wuxi 214153,China;School of Microelectronics and Control Engineering,Changzhou University,Changzhou 213164,China;Key Laboratory of Modern Acoustics,MOE,Nanjing University,Nanjing 210093,China)

机构地区：[1]江苏省无线传感系统应用工程技术研究开发中心,江苏无锡214153 [2]无锡商业职业技术学院,江苏无锡214153 [3]常州大学微电子与控制工程学院,江苏常州213164 [4]南京大学近代声学教育部重点实验室,江苏南京210093

出　　处：《水电能源科学》2024年第10期55-59,共5页Water Resources and Power

基　　金：国家自然科学基金项目(61873111);中央高校基本科研业务费专项资金(020414380195);江苏高校“青蓝工程”项目(RS20QL01)。

摘　　要：为了提高太湖水体中溶解氧浓度(DOC)参数的预测准确性,设计了一种基于ARIMA-IPOA-CNN-LSTM的预测模型。首先,采用差分自回归移动平均模型(ARIMA)捕捉数据的时间序列趋势和季节性特征;其次,引入卷积神经网络(CNN)和长短期记忆网络(LSTM)模型,分别从数据中学习空间和时间特征;再次,提出了一种改进的鹈鹕优化算法(IPOA)来优化模型参数,算法增加了Logistic混沌映射种群初始化、反向差分进化、萤火虫扰动的方法,CEC2005函数的测试结果显著优于传统鹈鹕优化算法;最后,将“剪枝”模型部署于STM32嵌入式设备。试验结果表明,在溶解氧浓度预测方面,该模型具有高的准确性和鲁棒性,为水环境保护提供了一种高效、可靠的解决方案。In order to improve the prediction accuracy of dissolved oxygen concentration(DOC)parameters in Taihu Lake,a prediction model based on ARIMA-IPOA-CNN-LSTM was designed.Firstly,the auto regressive integrated moving average(ARIMA)model was used to capture the time series trends and seasonal characteristics of the data.Secondly,convolutional neural networks(CNN)and long short-term memory networks(LSTM)models were introduced to learn spatial and temporal features from the data,respectively.An improved pelican optimization algorithm(IPOA)was proposed to optimize model parameters.The proposed algorithm added methods such as Logistic chaotic mapping population initialization,reverse differential evolution,and firefly disturbance.The test results of the CEC2005 function were significantly better than those of traditional pelican optimization algorithms.Finally,the"pruning"model was deployed on the STM32 embedded device.Experimental results show that the model has high accuracy and robustness in predicting dissolved oxygen concentration,providing an efficient and reliable solution for water environment protection.

关 键 词：差分自回归移动平均 鹈鹕优化算法 卷积神经网络 水体 溶解氧浓度 

分 类 号：TV211.1[水利工程—水文学及水资源] P338[天文地球—水文科学] TP18[天文地球—地球物理学]