遗传神经网络在动态称重中的应用  

Application of Genetic Neural Network in Dynamic Weighing

作　　者：沈小倩[1] 蔡晋辉[1] 姚燕[1] 蒋庆[1] 

机构地区：[1]中国计量学院计量测试工程学院,杭州310018

出　　处：《传感技术学报》2010年第9期1359-1363,共5页Chinese Journal of Sensors and Actuators

基　　金：浙江省科技计划项目资助(2008C21101)"离散制造过程信息智能测控系统的研发与应用"

摘　　要：为提高称重传感器的稳态响应,加快称重速度,提出一种基于人工神经网络预测动态称重终值的方法,即在BP网络的基础上用遗传算法进行优化,通过动态在线建模预测系统稳态输出AD值,再通过加权移动平均法得出物重。试验结果表明,该方法能够改善系统动态性能,提高测量精度,解决称重速度与精度难以同时提高的矛盾。In order to improve the dynamical respond of the weighing sensor and to meet the demand of rapid weighing,a new method based on neural network is introduced. The output AD values of steady state is predicted by the on-line modeling optimized by the genetic algorithm before the weighing sensor has settled to the steady state,then the sample weight is calculated according to weighted moving average. The experimental results proved that the neural network based method can be used to effectively reduce the weighing time and to increase the accuracy simultaneously.

关 键 词：动态称重 终值预测 神经网络 遗传算法 

分 类 号：TH273.5[机械工程—机械制造及自动化]