摘要
基于无线传感网络水质监测中水温数据质量差、预测精度低、稳定性差等问题,提出一种遗传算法(GA)优化改进极限学习机(SELM)的工厂化水产养殖水温预测模型(GA-SELM)。首先,在分析水温影响因素的基础上,通过天气指数的计算对无线传感网络中采集的异常数据进行校正;然后通过皮尔森相关分析研究影响因子与水温之间的关系;最后,采用Softplus函数作为ELM的激活函数,利用GA算法获取ELM的最佳初始权值和偏置,实现工厂化水产养殖水温预测。实验结果表明,GA-SELM模型有较好的预测性能,与传统BP神经网络、标准ELM网络模型和GA优化ELM算法相比,GA-SELM的预测指标MAE、MAPE和RMSE分别为0.154 3、0.005 4和0.187 6,性能均优于其他算法,能高效、稳定地实现水温的预测。
Considering the low quality of water quality data,low prediction accuracy and bad stability,an improved extreme learning machine optimized by genetic algorithm(GA-SELM)for predicting the water temperature is proposed in this paper.Firstly,the outlier data is emendated by weather index which is calculated on the basis of analyzing the influence factor of water temperature.Secondly,Pearson Correlation is utilized to explore the relations between influence factors and water temperature.Finally,Genetic algorithm(GA)is applied to optimize the initial weights and biases of extreme learning machine(ELM)with the Softplus function as activation function.The performance of proposed approach is evaluated using experiments.The mean absolute error(MAE),mean absolute percentage error(MAPE)and root mean square error(RMSE)of GA-SELM were 0.154 3,0.005 4 and 0.187 6 respectively.The results demonstrate that the efficiently and steadily of GA-SELM that forecasting better than other algorithms(BPNN,ELM,GA-ELM)in real-world industrial aquaculture water temperature monitoring.
作者
施珮
袁永明
匡亮
张红燕
李光辉
SHI Pei;YUAN Yongming;KUANG Liang;ZHANG Hongyan;LI Guanghui(Key Laboratory of Freshwater Fisheries and Germplasm Resources Utilization,Ministry of Agriculture Freshwater Fisheries Research Center, Chinese Academy of Fishery Sciences,Wuxi Jiangsu 214081,China;School of IoT Engineering,Jiangnan University,Wuxi Jiangsu 214122,China;School of IoT Engineering,Jiangsu Vocational College Of Information Technology,Wuxi Jiangsu 214153,China)
出处
《传感技术学报》
CAS
CSCD
北大核心
2018年第10期1592-1597,1612,共7页
Chinese Journal of Sensors and Actuators
基金
中央级公益性科研院所基本科研业务费项目(2016HY-ZD1404)
现代农业产业技术体系专项项目(CARS-46)
国家自然科学基金项目(61472368,61174023)。
关键词
无线传感网络
水温预测
极限学习机
遗传算法
激活函数
wireless sensor network
water temperature prediction
extreme learning machine(ELM)
genetic algorithm(GA)
activation function
作者简介
施珮(1988-),女,安徽宣城人,博士研究生,中国水产科学研究院淡水渔业研究中心,助理研究员,研究方向为无线传感网络;袁永明(1961-),男,江苏常熟人,本科,中国水产科学研究院淡水渔业研究中心,研究员,研究方向为渔业信息化研究。
