动态下电磁阀控对靶喷雾流量特性及控制方法  

Flow characteristics and control method for solenoid valve controlled target spraying under dynamic conditions

作　　者：代祥[1] 肖静[1] 徐幼林[1] 宋海潮[1,2] DAI Xiang;XIAO Jing;XU You-lin;SONG Hai-chao(College of Mechanical and Electronic Engineering, Nanjing Forestry University, Nanjing 210037, China;College of Mechanical Engineering, Nanjing Institute of Industry Technology, Nanjing 210023, China)

机构地区：[1]南京林业大学机械电子工程学院,江苏南京210037 [2]南京工业职业技术学院机械工程学院,江苏南京210023

出　　处：《江苏农业学报》2019年第2期476-483,共8页Jiangsu Journal of Agricultural Sciences

基　　金：江苏省基础研究计划青年基金项目(BK20170930);国家林业局"948"项目(2015-4-56);江苏高校"青蓝工程"项目;"333"人才培养工程项目

摘　　要：本研究对电磁阀控对靶喷雾系统动态条件(喷头开启数、开启时间、流量大小适时变化)下各参数[系统压力(P)、电磁阀频率(f)、占空比(D)]控制下的喷雾流量(Q)特性进行试验研究,并利用BP神经网络及遗传算法(GA)优化的BP神经网络对试验数据进行拟合和测试。试验结果表明:P每增大0.05 MPa,Q的最大调节范围增加约200 ml/min;较大的P会轻微减小流量控制线性区间(I),而较高的f则显著减小线性区间(I),f=20 Hz,P为0.10～0.35 MPa时对应的I约为0.3～0.6;Q与各控制参数间均存在非线性关系,利用BP神经网络进行喷头精准流量控制误差较小,平均误差仅0.20,经GA优化的BP神经网络具有更高的精度,误差低至0.15。综合考虑系统动态条件下电磁阀各参数可以实现流量的精准调节,依靠BP神经网络,尤其是GA优化的BP神经网络,可实现电磁阀控对靶喷雾流量的精准控制。The real output flow rate ( Q ) controlled by system pressures ( P ), solenoid valve control frequencies ( f ) and duty cycles ( D ) in the solenoid valve controlled target spraying system was studied under dynamic working conditions. Furthermore, the test results were verified by using BP neural network and genetic algorithm (GA) optimized BP neural network. The results showed that the Q rose 200 ml/min when the P increased 0.05 MPa. Larger P would slightly reduce the linear interval ( I ) of flow control, while higher f would significantly reduce the I . When the P was 0.10-0.35 MPa and f was 20 Hz, the I ranged from 0.3 to 0.6. There was nonlinear relationship between Q and each control parameter. The use of BP neural network (with the error of 0.20), especially the GA optimized BP neural network (with the error of 0.15), was feasible in the solenoid valve controlled target spraying system. Considering the parameters of the solenoid valve under the dynamic condition of the system, the flow rate can be precisely adjusted. The BP neural network, especially the BP neural network optimized by GA, can realize the precise control of solenoid valve controlled target spraying.

关 键 词：电磁阀控对靶喷雾 动态条件 流量特性 BP神经网络 遗传算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] S220.3[自动化与计算机技术—计算机科学与技术]