基于RBF网络的盐密光纤在线监测系统的研究

ESDD Fiber Online Monitoring System Based on RBF Network

针对传统等值附盐密度(盐密)测量方法的局限性,结合现有的光纤技术,开发了一套绝缘子盐密在线监测系统。通过对光纤通路中光功率衰减与光传感器表面附着盐分、环境温度、相对湿度等复杂关系的研究,建立了以光通量衰减、相对湿度和尘埃比率作为输入,盐密作为输出的RBF神经网络模型,该模型较好地解决了具有严重非线性的复杂系统的建模和控制问题,采用正交最小二乘(OLS)算法对模型进行训练,模型输出准确度较高。同时应用该建模结果开发了适于现场使用的盐密在线监测系统,数据监测中心的工作站根据神经元网络模型计算得到盐密值,并最终生成盐密的参考曲线图。

Aiming at the limitations of traditional equivalent salt deposit density （ESDD） measurement method, combined with existing optical fiber technology, a set of insulator ESDD online monitoring system is developed. By studying the influence of the salt attached to the surface of light sensor, environment temperature and relative humidity on the flux attenuation in fiber channel, taking the flux attenuation, relative humidity and ash density as inputs as well as taking the ESDD as output, this paper establishes a RBF neural network model which can solve the modeling and control problem preferably in complex system with serious nonlinear. This model showed high accuracy after trained by orthogonal least square （OLS） learning algorithm .Then the ESDD online monitoring system is developed for fieldwork based on the modeling results, the ESDD can be calculated by a work station in data monitoring system with neural network model, and finally the reference curve for the ESDD is formed.