一种基于图像文字识别的变电站防误操作系统

A Substation Anti-misoperation System Based on Image Character Recognition

随着电网安全生产管理要求的不断提高,要求进一步加强变电站防误操作。对此,首先提出了一套标识牌文字识别算法,通过文字垂直投影、霍夫曼变换边界分割、滤波降噪、 ANN(人工神经网络)文字识别以及匹配实现了变电站标识牌文字识别;接着,设计了一套由视频采集设备、门禁设备、后台服务器、无线局域网等硬件设备,以及文字识别软件系统构成的防误操作系统,并进行了实验验证。文字识别结果表明,所提出的文字识别算法准确度高达99.5%,未识别的少数文字也可以通过标识牌上下文推断加以精准识别;室内试验和现场试验结果验证了所提出的防误操作系统的有效性和工程适用性。

With the continuous improvement of the requirements of power grid safety production management,substation anti-misoperation management is required to be enhanced. For this reason, this paper proposes a set of signboard character recognition algorithm, which, through the vertical projection of text, Hoffman transform boundary segmentation, filtering noise reduction, Ann(artificial neural network) character recognition and matching, achieves the substation signboard character recognition;then, an anti-misoperation system comprising video acquisition equipment, access control equipment, back server, wireless LAN and other hardware equipment as well as the character recognition software system is designed and verified. The results of character recognition show that the accuracy of the proposed algorithm is up to 99.5%, and the unrecognized characters can also be recognized by inferring from the signboard context. According to the indoor and field test results, the effectiveness and engineering applicability of the anti-misoperation system are verified.