基于改进决策树算法的智慧电厂设备故障数据特征融合研究  

Research on feature fusion of intelligent power plant equipment fault data based on improved decision tree algorithm

作　　者：杨卫国 YANG Weiguo(Guangdong Huizhou Lng Power Co.,Ltd.,Huizhou Guangdong 510660,China)

机构地区：[1]广东惠州天然气发电有限公司,广东惠州516082

出　　处：《自动化与仪器仪表》2024年第8期90-94,共5页Automation & Instrumentation

基　　金：2022年广东能源集团科技项目(HZL-PK-21002)。

摘　　要：若电厂设备故障数据的采集和提取过程受到环境噪声的影响,易降低数据特征融合的准确率。针对这一问题,提出基于改进决策树算法的智慧电厂设备故障数据特征融合技术。首先,使用多源传感器数据采集电力设备运行过程中的异常数据,并实时收集设备之间的故障信息,建立设备故障数据集;然后,利用半软阈值函数技术,对设备数据集实施去噪处理,基于改进决策树算法对去噪后的设备数据进行分类处理,以分类后的电气设备故障数据为基础对电气设备故障数据的特征进行提取;最后,构建复合判断模型,实现对智慧电厂设备故障的数据特征的融合处理。实验结果表明,通过本研究设计的智慧电厂设备故障数据特征融合技术,能够显著提高对设备故障数据特征的融合准确率。If the collection and extraction process of power plant equipment fault data is affected by environmental noise,it is easy to reduce the accuracy of data feature fusion.To address this issue,a smart power plant equipment fault data feature fusion technology based on improved decision tree algorithm is proposed.Firstly,use multi-source sensor data to collect abnormal data during the operation of power equipment,and collect real-time fault information between devices to establish a device fault dataset;Then,using semi soft threshold function technology,the device dataset is denoised,and the denoised device data is classified using an improved decision tree algorithm.Based on the classified electrical equipment fault data,the features of the electrical equipment fault data are extracted;Finally,a composite judgment model is constructed to achieve fusion processing of data features for equipment faults in smart power plants.The experimental results indicate that the intelligent power plant equipment fault data feature fusion technology designed in this study can significantly improve the accuracy of equipment fault data feature fusion.

关 键 词：改进决策树算法 智慧电厂设备 故障数据特征融合 半软阈值函数 去噪处理 多源传感器 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]