皮带秤动态称重用神经网络算法设计  

Design of Neural Network Algorithm for Dynamic Weighing of Belt Weigher

作　　者：董祥辰 李冰莹 李永新[1] 王海涛[2] DONG Xiang-chen;LI Bing-ying;LI Yong-xin;WANG Hai-tao(School of Mechanical Engineering,Nanjing University of Science and Technology,Nanjing,Jiangsu 210094,China;Jiangsu Institute of Metrology,Nanjing,Jiangsu 210023,China)

机构地区：[1]南京理工大学机械工程学院,江苏南京210094 [2]江苏省计量科学研究院,江苏南京210023

出　　处：《计量学报》2020年第5期573-577,共5页Acta Metrologica Sinica

摘　　要：针对皮带秤在使用中难以保持标称计量精度的缺点,提出将过程神经网络引入皮带秤动态称重误差的补偿中。将动态称量过程中皮带秤单位长度上的重量、皮带速度、皮带垂度变化作为模型输入,设计了应用于皮带秤动态称重误差研究的单隐层过程神经网络误差反传播学习算法,利用Matlab软件对算法模型进行训练和测试,模型经过149次学习优化达到网络精度要求,测试组误差为1%,较使用网络前的原误差明显降低,验证了算法的可行性和有效性。To improve the weighing precision of the belt weigher,it was proposed to introduce process neural network( PNN) to compensate the dynamic weighing error of belt weigher. The weight per unit length of the belt weigher,speed of belt,and variation in belt sag in dynamic weighing process were used as model input. The single hidden layer PNN error back propagation learning algorithm was designed to apply to the study of the dynamic weighing error of the belt weigher.The algorithm model was trained and tested by MATLAB software and the model achieved network accuracy requirements after 149 learning optimizations. The test group error reaches 1%,which is significantly lower than the original error before using the network,and verifies the feasibility and effectiveness of the algorithm.

关 键 词：计量学 皮带秤 动态称重 误差补偿 学习算法 过程神经网络 

分 类 号：TB932[一般工业技术—计量学]