摘要
为了解决目前用于检测高压隔离开关分合闸状态的电子式传感器易受电磁干扰的影响,导致测量数据不准确,且安装到通有高电压的导电臂上存在取电难的问题。为此,根据隔离开关分合闸时其闸刀旋转的特性,提出了一种利用光纤测角度的方案,将闸刀旋转角度反映在直线上的距离进行角度测量,并采用遗传算法(GA)优化BP神经网络模型(GA-BP)对设计的光纤角度传感器存在的非线性问题进行补偿。通过实验仿真得出,通过遗传算法优化过的BP神经网络模型补偿后的光纤角度传感器具有较高的性能,相比于传统BP神经网络而言,其均方误差由0.084563提高到0.011828,拟合度由90.68%提高到了97.08%,线性度由8.26%提高到了0.83%。在此基础上,以SBE27.5kV/2000A隔离开关为研究对象搭建实验平台对光纤角度传感器检测的准确度进行可行性实验测试。实验结果表明:进行非线性补偿后的光纤角度传感器精确度从7.585提高到了0.269,表明了该方案具有较高的精确度和较强的可行性。
In order to solve such problem that the electronic sensor,which is used presently to detect the opening and closing state of high voltage disconnector,is susceptible to electromagnetic interference,results in inaccurate measurement data and,in case of installing on the conductive arm with high voltage,has the difficulty in obtaining the electricity,a kind of scheme using optical fiber to measure the angle is proposed in accordance with the characteristics of the blade rotation in case of opening and closing operation of the disconnector.The angle measurement is made by reflecting the rotation angle of the blade on the distance of the line.The genetic algorithm(GA)is adopted to optimize BP neutral network(GA-BP)so to compensate the non-linear problem of the designed optical fiber angle sensor.It is concluded by experimental simulation that the optical fiber angle sensor of the BP neutral network model,which is optimized through genetic algorithm and after compensation,has higher performance and,compared with the traditional BP neural network,its mean square error is increased from 0.084563 to 0.011828.The degree of fitting is increased from 90.68%to 97.08%and the linearity is increased from 8.26%to 0.83%.On this basis,an experimental platform isset up with SBE27.5 kV/2000 A disconnector as the research object to perform the feasible experimental test on the accuracy which is detected by the optic fiber angle sensor.The experimental results show that the accuracy of the optical fiber angle sensor after non-linear compensation is increased from 7.585 to 0.269,indicating that the scheme has high accuracy and strong feasibility.
作者
刘子英
张靖
付智辉
朱琛磊
LIU Ziying;ZHANG Jing;FU Zhihui;ZHU Chenlei(School of Electrical and Automation Engineering,East China Jiaotong University,Nanchang 330013,China)
出处
《高压电器》
CAS
CSCD
北大核心
2021年第10期9-17,共9页
High Voltage Apparatus
基金
国家自然科学基金(51767006)。
关键词
高压隔离开关
光纤传感器
遗传算法
BP神经网络
非线性补偿
high voltage disconnector
optical fiber sensor
genetic algorithm
BP neural network
nonlinear compensation
