基于膜分离的变压器在线监测系统中气体体积分数的预测算法  

Gas concentration prediction algorithm based on membrane extraction for transformer online monitoring system

作　　者：李洋流[1] 赵学增[1] 郭春志[2] 何鹏[1] 

机构地区：[1]哈尔滨工业大学传感与测控实验室,黑龙江哈尔滨150001 [2]沈阳超高压局,辽宁沈阳110001

出　　处：《电力自动化设备》2010年第12期36-41,共6页Electric Power Automation Equipment

摘　　要：采用高分子膜对溶解于变压器油中的气体进行分离的变压器在线监测系统需要在气体渗透达到平衡后才能给出油中气体的准确体积分数。在对气体渗透过程的影响因素进行分析后,提出了一种在非平衡状态下利用气室中气体体积分数数据预测平衡后气室中气体体积分数的算法,并采用特氟隆膜与电化学气体传感器构建在线监测系统进行了现场实验。在弱化温度对渗透过程的影响的条件下,利用预测算法对非平衡状态下的测量数据计算后,与离线气相色谱对H2的测量值26μL/L的最大相对误差为26.8%,小于国际电工委员会对实验室离线色谱测量结果误差调查的平均值。实验结果证实了该预测算法的可行性,并在充分利用大量实时的在线监测数据的同时,提高了变压器在线监测系统测量结果的准确性与可靠性。The transformer online monitoring system based on macromolecule membrane extraction acquires the concentration of gas dissolved in transformer oil only when the permeation equilibrium state is realized.Based on the analysis of gas permeation process,an algorithm is proposed to predict the gas concentration in permeation equilibrium state with the data of gas concentration in permeation nonequilibrium state.With Teflon membrane and electrochemical gas detection module,an online gas monitoring system is constructed and field tests are carried out.When the effect of fluctuant temperature is depressed,the biggest relative error between the predicted and the measured concentrations is 26.8 %（H2,26 μ L / L）,better than the mean error of lab offline measurements surveyed by IEC,which demonstrates the feasibility of the proposed prediction algorithm.

关 键 词：变压器 在线监测 油气分离 特氟隆膜 非平衡状态 预测算法 

分 类 号：TM406[电气工程—电器]