基于RPA的审计抽样软件机器人故障自动识别系统研究  

Research on the Automatic Robot Fault Identification System Based on RPA

作　　者：白云 郭蓬松[2] BAI Yun;GUO Pengsong(XI’AN Traffic Engineering lnstitute,School of humanities and management,Xi’an 710000,China;Air Force Engineering University,Air and Missile Defense College,Xi’an,710000,China;Northwestern Polytechnical University,Xi’an 710129,China)

机构地区：[1]西安交通工程学院人文与管理学院,西安710030 [2]空军工程大学防空反导学院,西安710000 [3]西北大学,西安710000

出　　处：《自动化与仪器仪表》2022年第8期170-173,178,共5页Automation & Instrumentation

基　　金：校级中青年基金课题《财务机器人在企业财务管理中的应用》(21KY-75);陕西省高等教育学会2021年高等教育科学研究项目《产教融合视角下应用型本科院校财务管理专业“课证融通”实践探索》(XGH21294)。

摘　　要：在审计抽样软件机器人运行状态下,受到样本数据之间差量分布影响,机器人抽取策略无法精准映射到每一个样本个体上,导致抽取的样本存在差量偏移,失去了抽取样本的客观性。为了避免这一问题,通常采用故障识别系统对其异常数据进行识别,但是与传统数据故障不同,神经抽样数据的差量精度较高,对应特征系数精度远超传统识别系统的识别阈值,因此,传统的识别系统的识别结果整体输出误差偏大,且识别效果稳定相差。基于审计抽样数据的特征,引入机器人流程自动化(Robotic Process Automation,RPA),根据审计流程数据特点,构建硬件平台;在硬件平台基础上,设计RPA软件流程,通过对RPA审计数据故障点向量机模型的建立、抽样数据故障特征识别与RPA故障识别数据判定输出,完成提出系统的设计,实验数据表明:经过RPA算法优化后的系统,整体故障识别率明显提升,且系统稳定性远远优于现有识别系统的参数标准,能够满足审计抽样软件机器人故障自动识别的准确度要求。In the operation state of the audit sampling software robot,affected by the distribution of the difference between the sample data,the robot extraction strategy cannot be accurately map to each sample individual,resulting in the deviation of the extracted samples and losing the objectivity of sampling samples.In order to avoid this problem,usually fault identification system is used to identify the abnormal data,but unlike the traditional data fault,the difference of high accuracy,the corresponding characteristic coefficient accuracy far exceeds the traditional recognition system recognition threshold,so the traditional recognition system recognition result overall output error is large,and the identification effect is stable.Based on the characteristics of audit sampling data,the RPA algorithm is introduced to construct the audit process data,the RPA software process is designed.The RPA audit data and the RPA model,the overall fault identification rate is improved to meet the accuracy requirements of the existing identification system.

关 键 词：RPA 审计抽样 故障 识别系统 

分 类 号：TP29[自动化与计算机技术—检测技术与自动化装置] TM773[自动化与计算机技术—控制科学与工程]