油浸式大型电力变压器表面温度预测模型

为了早期发现油浸式电力变压器长期连续运行时可能存在的温升异常问题,该文提出了一种新的油浸式电力变压器表面温度预测模型,可以预测出变压器表面的正常温度。通过比较预测值与实测值的误差来预知变压器表面温度是否存在异常,进而判断变压器运行状态是否正常。预测模型参数是基于现场数据采用偏最小二乘法估计出来的,克服了传统最小二乘法参数估计结果不稳定的问题。并且从传热学原理、预测误差的均值、方差及其正态分布性等方面,合理地选择用于参数估计的数据长度。用该模型计算分析了一年的现场数据,预测结果表明该模型是正确、有效的。

Prediction of surface temperatures of large oil-immersed power transformers

The temperatures in oil-immersed power transformers can increase abnormally during long-term, continuous operation. This paper presents a model to predict the normal surface temperatures of oil-immersed power transformers. The predicted value can then be compared to the measured value to identify abnormal transformer temperatures and improper operating conditions. The model parameters are estimated from field data using the partial least squares method, which overcomes the instabilities of estimated results using the traditional least squares method. Furthermore, the appropriate amount of data needed to estimate the parameters was confirmed from the principle of heat trarsfer and the statistical characteristic of the mean, variance and normal distribution of the predicted deviations. Analysis of one year of field data verified the accuracy of the predicted temperatures.