在气体绝缘组合电器(gas insulated switchgear,GIS)实体模型中分别放置了针-板、悬浮金属颗粒和绝缘子表面固定金属颗粒放电模型,用超声波传感器采集到其放电波形。对放电波形提取的特征向量进行局部线性嵌入(local linear embedding,L...在气体绝缘组合电器(gas insulated switchgear,GIS)实体模型中分别放置了针-板、悬浮金属颗粒和绝缘子表面固定金属颗粒放电模型,用超声波传感器采集到其放电波形。对放电波形提取的特征向量进行局部线性嵌入(local linear embedding,LLE)算法降维处理,用降维后的向量作为输入对BP_Adaboost分类器进行训练和测试类型识别。识别结果表明,用这样方法进行GIS绝缘缺陷类型识别可以在减少计算量的同时保持较高的识别率,说明了其在局部放电模式识别应用中的有效性。展开更多
Empirical mode decomposition (EMD) and BP_AdaBoost neural network are used in this paper to model the oil price. Based on the benefits of these two methods, we predict the oil price by using them. To a certain extent,...Empirical mode decomposition (EMD) and BP_AdaBoost neural network are used in this paper to model the oil price. Based on the benefits of these two methods, we predict the oil price by using them. To a certain extent, it effectively improves the accuracy of short-term price forecasting. Forecast results of this model are compared with the results of the ARIMA model, BP neural network and EMD-BP combined model. The experimental result shows that the root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE) and Theil inequality (U) of EMD and BP_AdaBoost model are lower than other models, and the combined model has better prediction accuracy.展开更多
文摘在气体绝缘组合电器(gas insulated switchgear,GIS)实体模型中分别放置了针-板、悬浮金属颗粒和绝缘子表面固定金属颗粒放电模型,用超声波传感器采集到其放电波形。对放电波形提取的特征向量进行局部线性嵌入(local linear embedding,LLE)算法降维处理,用降维后的向量作为输入对BP_Adaboost分类器进行训练和测试类型识别。识别结果表明,用这样方法进行GIS绝缘缺陷类型识别可以在减少计算量的同时保持较高的识别率,说明了其在局部放电模式识别应用中的有效性。
文摘Empirical mode decomposition (EMD) and BP_AdaBoost neural network are used in this paper to model the oil price. Based on the benefits of these two methods, we predict the oil price by using them. To a certain extent, it effectively improves the accuracy of short-term price forecasting. Forecast results of this model are compared with the results of the ARIMA model, BP neural network and EMD-BP combined model. The experimental result shows that the root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE) and Theil inequality (U) of EMD and BP_AdaBoost model are lower than other models, and the combined model has better prediction accuracy.