基于ARMA模型的电力变压器故障预报方法

Fault Prediction Method for Transformers Based on ARMA Model

为了保证在安全运行的前提下提高变压器的运行效率,分析了变压器在异常时的油色谱数据,提出了基于ARMA模型的多维混合回归系统模型,对故障的发展趋势进行预报。通过最小二乘法,结合AIC准则估计出预报模型的最终参数和阶次。仿真结果证明,该方法适用于大型油浸式电力变压器的故障预报。

According to the analysis of the unusual oil ehromatogram data, a fault prediction method for transformers based on ARMA model is proposed. By using of the least squares and AIC criterion, the final parameters and orders of the predictive model are estimated. A multi-dimensional and mixing regressive system model is presented to predict the developing trend of the faults. Computer simulation results show that the proposed approach has high accuracy and is feasible for the fault prediction of large transformers.