一种面向能源工程数据评估的改进随机森林算法设计  

Design of an improved random forest algorithm for energy engineering data evaluation

作　　者：马林 MA Lin(Economic and Technology Research Institute,State Grid Gansu Electric Power Company,Lanzhou 730050,China)

机构地区：[1]国网甘肃省电力公司经济技术研究院,甘肃兰州730050

出　　处：《电子设计工程》2024年第18期57-61,共5页Electronic Design Engineering

基　　金：国网甘肃省电力公司技改检修项目(2022010640)。

摘　　要：传统电力工程数据稽核与评估方法的准确率偏低且效率较差,不适用于当前日益复杂的信息处理与分析工作。针对此,文中基于改进的随机森林算法提出了一种面向电力工程的异常数据检测算法。对于随机森林算法易受高维数据影响而导致信息特征提取能力不足的问题,该算法利用堆栈稀疏自编码器对高维数据进行降维,以提升数据检测的准确率。同时使用麻雀搜索算法对数据特征提取模型的参数加以优化,进一步提升了算法的性能和效率。在以电力工程造价数据为样本展开的实验测试中,所提算法的AUC与F1值领先于SSAE-RF算法2.73%及0.011,且异常数据识别率可达80%,运行时间也在对比算法中为最短,表明其具有较好的性能和计算效率。The traditional methods of electric power engineering data audit and evaluation have low accuracy and low efficiency,and are not suitable for the increasingly complex electric power engineering information processing and analysis.Based on the improved random forest algorithm,this paper proposes an anomaly data detection algorithm for power engineering.The algorithm is based on the defect that the random forest algorithm is easy to be affected by high-dimensional data,resulting in poor information feature extraction ability.It uses the stack sparse self-coder to reduce the dimension of high-dimensional data,improving the accuracy of data detection.At the same time,the sparrow search algorithm is used to optimize the parameters of the data feature extraction model,which further improves the performance and efficiency of the algorithm.In experimental tests conducted using power engineering cost data as samples,the AUC and F1 values of the proposed algorithm were 2.73%and 0.011 higher than those of the SSAE-RF algorithm,and the abnormal data recognition rate could reach 80%.The running time was also the shortest among the compared algorithms,indicating its good performance and computational efficiency.

关 键 词：工程造价 随机森林 堆栈稀疏自编码器 麻雀搜索算法 异常数据检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]