基于径向基神经网络的废旧铝分离技术研究  

Research on Sorting Technology of Scrap Aluminum Based onRadial Basis Function Neural Network

作　　者：周晏锋 陈蔚芳[1] 潘立剑 韦子祥 徐鹏行 ZHOU Yanfeng;CHEN Weifang;PAN Lijian;WEI Zixiang;XU Pengxing(College of Mechanical and Electrical Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China)

机构地区：[1]南京航空航天大学机电学院,江苏南京210016

出　　处：《机械与电子》2020年第12期19-23,共5页Machinery & Electronics

基　　金：江苏省重点研发项目(BE2018721);南京航空航天大学研究生创新基地(实验室)开放基金(kfjj20190514)。

摘　　要：建立了一套废旧铝喷气式分离系统,使用不同气压吹不同体积的废旧铝进行实验,测量其被吹离距离,获得230组训练集、80组测试集数据,从而建立径向基网络模型,拟合优度可达93%,均方根误差低至2.99。结果表明该套系统构造简单、识别效果准确,可以根据废旧铝具体情况预测出吹离距离,选取合适大小的气流进行自适应分离操作,实现对不同成分的废旧铝进行精细分离的目的。A set of jet sorting system for scrap aluminum is established.Different air pressures are used to blow different volumes of scrap aluminum for experiments,after measuring the distance it was blown away,230 sets of training set and 80 sets of test set data are obtained to establish a radial basis function network model.The goodness of fit can reach 93%and the root mean square error can reach 2.99.The results show that the system is simple in structure and accurate in recognition.It can predict the blowoff distance according to the specific conditions of scrap aluminum,select a suitable airflow for adaptive sorting operation,and achieve the purpose of fine sorting of scrap aluminum with different compositions.

关 键 词：废旧铝 径向基神经网络 精细分离 自适应 

分 类 号：TP273.5[自动化与计算机技术—检测技术与自动化装置]