基于神经网络的出菇房CFD温度均匀性预测模型  

Prediction model of CFD temperature uniformity of mushroom house based on neural network

作　　者：景亮[1] 方倩 Jing Liang;Fang Qian(School of Electrical and Information Engineering , J iangsu University ,Zhenj iang,212013, China)

机构地区：[1]江苏大学电气信息工程学院

出　　处：《中国农机化学报》2019年第6期71-75,共5页Journal of Chinese Agricultural Mechanization

基　　金：江苏省高校优势学科建设工程资助项目(PAPD)(苏政办发[2011]6号);江苏省政策引导类专项资金项目(BN2015095);镇江市科技计划项目(NY2016017)

摘　　要：为预测出菇房内环境性能指标,采用CFD建立菇房模型并通过试验数据验证仿真结果准确性,对比可知温度的平均相对误差为4.9%,引入温度均匀性指标,设计正交试验进行CFD数值模拟,利用模拟数据训练GMDH(group method of data handling,数据处理组合法)型神经网络,最后得出温度均匀性指标的预测模型。分析结果表明,预测值与CFD仿真值相关系数达到0.942 5,平均绝对误差仅为0.042,预测精度较高,为出菇房的进一步优化提供可靠依据。Aiming at predicting the environmental performance index of mushroom house, CFD is used to establish the model and the accuracy of the simulation results is verified by comparing with experimental data. The average relative error of the temperature is 4.9%. The temperature uniformity index is introduced and the orthogonal test is designed for CFD numerical simulation. Using the simulation data to train the GMDH(group method of data handling) type neural network, and finally obtains the prediction model of the temperature uniformity index. The analysis results show that prediction accuracy of GMDH type neural network is high enough. Because it can be seen that the correlation coefficient between the predicted value and CFD simulation value reaches 0.942 5 and the average absolute error is only 0.042 K. It can provide a reliable basis for further optimization of the mushroom house.

关 键 词：CFD模拟 温度均匀性 GMDH型神经网络 预测 

分 类 号：S24[农业科学—农业电气化与自动化]