基于多维数据的WTD-GA-BP海参养殖水温预测
- 分享到:
摘要：
在海参健康生长过程中海水温度是最为关键的因素,针对海参养殖环境复杂,预测水温难度大,精确度较低等问题,提出一种基于多维数据的WTD-GA-BP海参养殖水温预测模型.首先利用相关性分析方法,系统地获取海参养殖气象及水体环境因子与水温的相关关系;其次使用WTD方法对影响海参养殖水温变化的关键影响因子降噪,增强其数据的平滑性;最后,构建GA-BP水温预测模型,以多维数据作为模型输入,以水温作为输出完成水温序列的预测.结果表明,WTD方法有效去除了数据信号噪声,增强了数据平滑性.同时,以多维数据作为输入,充分挖掘了水温数据的变化特征.基于多维数据的WTD-GA-BP水温预测模型具有良好的预测性能,其评价指标MAE、MAPE和MSE分别为0.1468、0.0060和0.0503,能为海参养殖环境的水温预测提供数据参考依据.
The seawater temperature is the most critical factor in the healthy growth of sea cucumbers.To address the problems of complex sea cucumber culture environment,difficulty in predicting water temperature and low accuracy,a WTD-GA-BP sea cucumber culture water temperature prediction model based on multidimensional data is proposed.Firstly,the correlation analysis method was used to systematically obtain the correlation between the meteorological and water environment factors and water temperature of sea cucumber culture.Secondly,the WTD method was used to reduce the noise of the key influencing factors affecting the change of sea cucumber culture water temperature and enhance the smoothness of its data.Finally,the GA-BP water temperature prediction model was constructed,and the prediction of water temperature series was completed with multidimensional data as model input and water temperature as output.The results show that the WTD method effectively removes the data signal noise and enhances the data smoothing.At the same time,the variation characteristics of water temperature data are fully explored by using multidimensional data as the input.The WTD-GA-BP water temperature prediction model based on multidimensional data has good prediction performance,and its evaluation indexes MAE,MAPE and MSE are 0.1468,0.0060 and 0.0503,respectively,which can provide data reference basis for water temperature prediction of sea cucumber culture environment.
作者：
李晓梅 杨健浩 李俐 盖荣丽 汪祖民
Li Xiaomei;Yang Jianhao;Li Li;Gai Rongli;Wang Zuming(School of Information Engineering,Dalian University,Dalian 116000,China)
机构地区：
大连大学信息工程学院
出处：
《河南师范大学学报:自然科学版》 CAS 北大核心 2023年第6期30-38,共9页
Journal of Henan Normal University(Natural Science Edition)
基金：
大连市科技创新基金项目(2020JJ27SN106).
关键词：
海参养殖 水温预测 多维度数据 遗传算法 BP神经网络 小波阈值降噪
sea cucumber farming water temperature prediction multidimensional data genetic algorithm BP neural network wavelet threshold noise reduction
分类号：
TP18 [自动化与计算机技术—控制理论与控制工程]
