文献类型：期刊文献

中文题名：基于BP神经网络的旁热式辐射与对流粮食干燥过程模型

英文题名：Model of Drying Process for Combined Side-heat Infrared Radiation and Convection Grain Dryer Based on BP Neural Network

作者：代爱妮[1,2];周晓光[1];刘相东[3];刘景云[4];张驰[1]

第一作者：代爱妮

通讯作者：Zhou, Xiaoguang

机构：[1]北京邮电大学自动化学院;[2]青岛农业大学理学与信息学院;[3]中国农业大学工学院;[4]北京联合大学自动化学院

第一机构：北京邮电大学自动化学院,北京100876

年份：2017

卷号：48

期号：3

起止页码：351-360

中文期刊名：农业机械学报

外文期刊名：Transactions of the Chinese Society for Agricultural Machinery

收录：CSTPCD;;EI(收录号：20172203711647);Scopus(收录号：2-s2.0-85019732616);北大核心:【北大核心2014】；CSCD:【CSCD2017_2018】；

基金：国家粮食公益性行业科研专项(201413006)

语种：中文

中文关键词：粮食;红外辐射与对流干燥;BP神经网络;预测模型

外文关键词：grain; combined infrared radiation and convection drying; BP neural network; prediction model

摘要：针对旁热式辐射与对流粮食干燥机的干燥特点,建立了一种粮食干燥机干燥过程的BP神经网络预测模型。该模型采用了3层神经网络结构(8-10-1),模型输入为粮食干燥机的8个变量,模型输出为出口粮食水分比或干燥速率。通过编写Matlab建模程序,基于实际干燥实验的样本数据训练与测试网络,实现了红外辐射与对流联合干燥的动力学模型,并给出了相应的模型数学表达式,模型预测的出口水分比与干燥速率的R2分别为0.998 9和0.998 0,均方根误差分别为0.009和0.004 1,预测结果与实际测量数据拟合较好;另外,结合实验干燥条件对模型干燥性能的预测结果进行了分析与总结,并依据同样方法建立了顺逆流粮食干燥过程的出口粮食水分比预测模型,对比了2种干燥方式的干燥性能。仿真预测表明用BP神经网络方法建模简单,具有自适应性、灵活性和自学习性等特点,相比于其他粮食干燥的经验数学模型,能综合考虑多种影响因素,可为红外辐射与对流联合干燥过程提供一种新的建模方法。
The drying mechanism of combined side-heat infrared radiation and convection （IRC） grain dryer is more complicated compared with that of the traditional convection drying. In order to explore the model of uncertain system like the grain drying and application of BP artificial neural network method, a new intelligent prediction model for the combined side-heated IRC dryer used to predict the outlet core moisture content ratio and drying rate is developed based on BP neural network algorithm. The model which has three layer neural network structures （8 - 10 - 1 ） is trained and tested based on the train data set and test data set by programming the model in Matlab. The model inputs are the eight influence variables of grain dryer, and the model output is the outlet grain moisture ratio of the dryer or the drying rate. The corresponding mathematical expressions of moisture ratio and drying rate model are also given, and the determination coefficients （R^2 ） of model prediction are 0. 998 9 and 0. 998 0, and the root mean square errors （RMSE） are 0. 009 and 0. 004 1, respectively. The predicted results are fitted well with the measured data, and the prediction accuracy is high. In addition, combined with the experimental drying conditions, the prediction results of the model are analyzed and summarized. According to the same method, the prediction model of outlet moisture ratio for the counter-current grain drying is also successfully established. By the comparison of predicted performance curves for two types of drying process, it is proved that the combined side-heat IRC drying has faster drying rate and less time to dry to the target moisture value than those of the conventional hot air convection drying. It can be used to predict the drying performance of different drying processes and to realize the comparison of different drying processes. In addition, compared with other grain drying mathematical models, various influence factors of grain drying can be comprehensively considered, which can provide a new modeling method for the complex system like the drying of combined side-heat IRC.