基于支持向量回归的电力变压器状态评估

Power transformer condition evaluation based on support vector regression

为提高电力变压器状态评估的准确性,将变压器健康状态分为5级。鉴于支持向量机对小样本具有良好的拟合能力,而变压器数据具有小样本、贫信息的特点,提出了基于支持向量回归的电力变压器状态评估模型。将变压器的油色谱分析数据和电气实验数据利用半岭模型确定变压器各个参数的分值,评分项目结果作为支持向量机的自变量,通过多层动态自适应优化算法优化了支持向量回归的参数,形成变权重的预测。实例验证了变压器状态评估模型的正确性及可行性,其结果更接近变压器的真实运行状态。

To improve the accuracy of transformer condition evaluation,its health status is divided into five levels.Since SVM（Support Vector Machine） has good fitting ability for small sample set and the transformer data has the features of less samples and poor information,a condition evaluation model based on SVR is proposed for power transformer.The transformer dissolved gas analysis data and electrical test data are used to determine the scores of transformer parameters by the semi-mountain model and the results of scoreevaluation are taken as the independent variables of SVM.The multi-layer dynamic adaptive algorithm is used to optimize the parameters of SVR to predict the changeable weights.Case study verifies the proposed model is efficient and feasible and the conclusion of transformer condition evaluation is closer to its actual condition.