基于机器学习算法的变压器寿命损耗预测研究

Research on life loss of transformer based on machine learning algorithms

变压器寿命评估和故障诊断对提高电网安全可靠、经济性运行具有重要的意义。环境温度和负载曲线是影响变压器绝缘老化的重要因素之一,影响变压器的寿命。为此基于GB/T 1094.7—2008推荐的环境温度和变压器负载计算变压器寿命损耗的模型,开发了估算变压器寿命损失的数据驱动静态模型和基于SVM算法建立变压器寿命损耗计算机器学习模型。研究发现:使用RBF核函数的SVM算法模型可有效、准确地解决变压器寿命评估问题。

Transformer life assessment and failure diagnostics are always important problems for power grid companies. Ambient temperature and load profle are the main factors which affect the transformer insulation, and lifetime. Based on the model of transformer life loss calculated by GB/T 1094.7-2008 recommended ambient temperature and transformer load, a data-driven static model for estimating transformer life loss and a learning model for transformer life loss based on SVM algorithm are developed. The research fnds that the SVM algorithm model using RBF kernel function can effectively and accurately solve the problem of transformer life assessment.