基于tesseract训练的电力告警信号特征识别方法  

Feature recognition method of power alarm signal based on tesseract training

作　　者：陈东海 王猛 吴昱浩 杨淇 CHEN Donghai;WANG Meng;WU Yuhao;YANG Qi(State Grid Ningbo Electric Power Supply Company,Ningbo 315000,China)

机构地区：[1]国网宁波供电公司,浙江宁波315000

出　　处：《电子设计工程》2023年第17期137-140,145,共5页Electronic Design Engineering

摘　　要：为缩小电力告警信号峭度指标极值差水平,增强电网主机对告警信号特征的识别准确性,提出基于tesseract训练的电力告警信号特征识别方法。在tesseract训练布局环境中,分别识别信号字符的特定行区域与列区域,完成基于tesseract训练的信号字符标记。在此基础上,提取关键电力告警信号特征,通过计算识别阈值方式,完成特征信号参量暂态传输行为处理,实现基于tesseract训练的电力告警信号特征识别方法设计。实验结果表明,与多特征组合识别方法相比,文中方法在tesseract训练引擎作用下,电力告警信号峭度指标极值差水平得到较好控制,提升了电网主机对告警信号特征的准确识别能力。In order to reduce the extreme value difference level of power alarm signal kurtosis index and enhance the recognition accuracy of power grid host for alarm signal characteristics,a power alarm signal feature recognition method based on Tesseract training is proposed.In the testeract training layout environment,the specific row area and column area of signal characters are recognized respectively,and the signal character marking based on testeract training is completed.On this basis,the key power alarm signal features are extracted,and the transient transmission behavior processing of characteristic signal parameters is completed by calculating the recognition threshold,so as to realize the design of power alarm signal feature recognition method based on Tesseract training.The experimental results show that compared with the multi feature combination recognition method,under the action of Tesseract training engine,the power alarm signal kurtosis index limit value difference level is better controlled,and the accurate recognition ability of power grid host to the characteristics of alarm signal is improved.

关 键 词：tesseract训练 电力告警信号 特征识别 行区域 列区域 暂态行为 

分 类 号：TN911[电子电信—通信与信息系统]