基于改进随机森林的电力变压器试验和监测数据气体浓度预测  

Gas concentration prediction of power transformer test and monitoringdata based on improved random forest

作　　者：赵超 张迅 王冕 范强 黄军凯 陈沛龙 ZHAO Chao;ZHANG Xun;WANG Mian;FAN Qiang;HUANG Junkai;CHEN Peilong(Electric Power Research Institute of Guizhou Power Grid Co.,Ltd.,Guiyang 550002,China)

机构地区：[1]贵州电网有限责任公司电力科学研究院,贵阳550002

出　　处：《电测与仪表》2024年第11期205-210,共6页Electrical Measurement & Instrumentation

基　　金：中国南方电网有限责任公司科技项目(066600KK 58200019)。

摘　　要：针对现有电力变压器油中溶解气体浓度预测方法预测精度低的问题,提出了一种结合经验模态分解和改进粒子群优化的随机森林算法来预测变压器油中气体浓度。利用改进粒子群算法优化的随机森林模型对经验模态分解后的各分量进行预测,并将各分量的预测结果叠加为最终预测结果。通过算例对该模型的预测结果进行分析,验证了该方法的准确性。结果表明,相比于常规预测模型,该模型预测结果更为接近气体浓度实际值,能够有效提高模型的预测精度,为其他领域的预测提供了一定的参考。Aiming at the problem of low prediction accuracy of existing prediction methods of dissolved gas concentration in power transformer oil,a random forest algorithm combining empirical mode decomposition and improved particle swarm optimization is proposed to predict the gas concentration in transformer oil.The random forest model optimized by improved particle swarm optimization is used to predict the components after empirical mode decomposition,and the prediction results of each component are superimposed into the final prediction results.The prediction results of the model are analyzed through an example,and the accuracy of the proposed method is verified.The results show that,compared with the conventional prediction model,the prediction result of the model is closer to the actual value of gas concentration,and can effectively improve the prediction accuracy of the model,which provides a certain reference for the prediction of other fields.

关 键 词：电力变压器 经验模态分解 粒子群算法 随机森林 气体浓度 

分 类 号：TM411[电气工程—电器]