摘要
考虑到传统方法在电力工程造价数据异常识别中存在识别精度低、质量差的问题,提出了一种基于卷积神经网络的电力工程造价数据异常识别方法。通过搭建密度分布函数,计算工程造价异常数据的波动系数;根据工程造价异常数据的包络特征,计算工程造价异常数据权重;利用工程造价异常数据矩阵,求解工程造价异常数据的聚类中心。实验结果表明,文中方法在识别工程造价异常数据时可以提高工程造价异常数据识别的精度和质量。
Considering the problems of low recognition accuracy and poor quality of traditional methods in abnormal recognition of power engineering cost data,a method of abnormal recognition of power engineering cost data based on convolution neural network is proposed.By constructing the density distribution function,the fluctuation coefficient of abnormal data of project cost is calculated.The weight of abnormal data of project cost according to the envelope characteristics of abnormal data of project cost is calculated as well.By using the abnormal data matrix of project cost,the clustering center of abnormal data of project cost is solved.The experimental results show that the method,in this paper,can improve the accuracy and quality of identifying abnormal data of project cost.
作者
戴小凤
朱卫东
DAI Xiaofeng;ZHU Weidong(Department of Engineering Management,Anhui Audit College,Hefei 230601,China;School of Management,Hefei University of Technology,Hefei 230009,China)
出处
《兰州工业学院学报》
2022年第4期62-66,共5页
Journal of Lanzhou Institute of Technology
基金
2020年度安徽省教育厅高校自然科学研究重点项目(KJ2020A1129)
2019年度安徽省教育厅高校优秀青年人才支持计划重点项目(gxyqZD2019144)
2018年度安徽省审计厅重点科研课题(AHSJ20180202)。
关键词
卷积神经网络
异常数据
权重计算
工程造价
识别方法
convolution neural network
abnormal data
weight calculation
project cost
identification method
data preprocessing
