利索-全球专利检索

  1. 登录
  2. 注册
发明公开
CN107733089A 一种基于SVM的变电站刀闸二次回路故障预测方法 无效 - 驳回 转让

一种基于SVM的变电站刀闸二次回路故障预测方法
摘要：
本发明提供了一种基于SVM的变电站刀闸二次回路故障预测方法，包括如下步骤：步骤一：整理数据源；步骤二：对上述数据源进行预处理获得一致的单调性，并归一化使数值范围处在[0,1]；步骤三：通过步骤二得到归一化的数值后，结合历史数据构成的训练样本(xi,yi)，训练样本(xi,yi)通过SVM工具箱中函数train-svm的计算，即可得到偏差b和拉格朗日系数α，得出预测模型：步骤五：用历史数据训练好的预测模型进行刀闸二次回路故障预测，输出为-1认为正常运行，若输出为+1认为存在故障，即可结合停电计划提前进行故障排查。
摘要（英）：

The invention provides a substation disconnector secondary return circuit fault predicting method based on an SVM. The method comprises the following steps of 1, sorting a data source; 2, preprocessing the data source to obtain consistent monotonicity, and conducting normalization to make the range of numerical values in [0,1]; 3, constituting a training sample (xi, yi) through the numerical values normalized in the second step in combination with historical data, wherein the training sample (xi, yi) is calculated through a function train-svm in an SVM tool box to obtain deviation b and a Lagrange coefficient alpha, and a prediction module is obtained; 5, using the prediction module trained by the historical data to conduct disconnector secondary return circuit fault predicting, wherein ifminus 1 is output, it is determined that running is normal, if plus 1 is output, it is determined that faults occur, and fault removal can be conducted in advance in combination with a power-out scheme.

0/0