摘要
为了及时监测近海养殖环境的变化,帮助养殖者及时了解养殖环境,研究提出一种基于萤火虫算法(Firefly Algorithm,FA)结合反向传播(Back propagation,BP)神经网络的水质溶解氧浓度预测方法,并提出了卷积神经网络(Convolutional Neural Networks,CNN)结合长短时记忆网络(Long-term and short-term memory network,LSTM)的水质溶解氧浓度控制技术,最后对水质溶解氧浓度预测控制模型进行了测试分析。结果显示,FA-BP模型具有更高的预测精度,预测误差能够较好地控制在0.05以内;CNN-LSTM-FA-BP寻找全局最优的迭代次数最少,说明该模型的寻优能力强。在理想实验环境和增加干扰的环境中,本研究提出模型的系统调节时间更少,具有更小的超调量,能够更好地对溶解氧设定值进行跟踪,同时为近海养殖水质溶解氧浓度预测控制工作提供一定的研究借鉴。
To timely monitor the changes in the offshore aquaculture environment and help farmers understand the aquaculture environment in a timely manner,a prediction method of dissolved oxygen concentration in water quality based on Firefly algorithm(FA) and back propagation(BP) neural network was proposed,The Convolutional neural network(CNN) combined with Long term and short term memory network(LSTM) is proposed to control the dissolved oxygen concentration in water quality.Finally,the prediction control model of dissolved oxygen concentration in water quality is tested and analyzed.The results show that the FA-BP model has higher prediction accuracy,and the prediction error can be well controlled within 0.05;The CNN-LSTM-FA-BP has the least number of iterations to find the global optimal,indicating that the model has strong optimization ability.In an ideal experimental environment and an environment with increased interference,this study proposes a model with less system adjustment time and smaller overshoot,which can better track the set value of dissolved oxygen and provide a certain research reference for the prediction and control of dissolved oxygen concentration in offshore aquaculture water quality.
作者
刘玉洁
唐升
姜源
郎波
王江
LIU Yujie;TANG Sheng;JIANG Yuan;LANG Bo;WANG Jiang(Zhuhai City Polytechnic,Zhu Hai.Guang Dong 519090,China;Zhuhai Zhishan Industrial Development Co.,Ltd.,Zhu Hai.Guang Dong 519090,China)
出处
《自动化与仪器仪表》
2023年第12期56-61,共6页
Automation & Instrumentation
基金
2022年中国高校产学研创新基金-新一代信息技术创新项目课题《复杂海鲈养殖环境下基于深度学习的水质溶解氧预测技术研究》(2021ITA02021)
2022年广东省普通高校重点科研项目《自适应智能增氧控制系统的关键技术研究》(2022ZDZX4105)
2020年广东省普通高校重点领域服务乡村振兴重点领域专项项目(自然科学)《基于人工智能技术的智慧水产养殖系统》(2020ZDX1077)。
关键词
萤火虫算法
BP神经网络
溶解氧浓度
预测控制
卷积神经网络
长短时记忆网络
firefly algorithm
BP neural network
dissolved oxygen concentration
predictive control
convolutional neural net-work
long and short term memory network
