基于声纹识别技术的调度下令验证系统研究

Research on Verification System of Dispatch Order Based on Voiceprint Recognition Technology

电网调度下令操作对于电网运行方式和人员安全具有重要意义。当前的下令环节主要依靠调度电话或网络发令形式,缺乏人员验证与防误,存在一定的安全隐患。声纹识别技术作为一种非接触式生物验证方法,有助于提高调度下令环节的安全性。通过对比声纹识别技术中的全变量系统和卷积神经网络模型,得出卷积神经网络模型可以深刻挖掘语音频谱中的声纹信息,识别效果更好;采用基于卷积神经网络的声纹识别技术,构建了调度下令验证系统。通过验证实验,可以验证该系统对于人员的识别度高达95.42%,同时克服了传统方案需要全变量建模的缺陷,降低了系统复杂度,具有抗噪性。该系统可帮助远程识别现场接令人员与调度下令人员的真实身份,提高了调度运行的安全性与效率,具有实际应用价值。

The operation of the power grid dispatching order is of great significance to the power grid operation mode and personnel safety. The current order link mainly relies on dispatch telephone or network order form, lack of personnel verification and error prevention, and there are certain security risks. Voiceprint recognition technology, as a non-contact biological verification method, helps to improve the security of the dispatch order link. The full-variable system and convolutional neural network model in the voiceprint recognition technology are compared, and the convolutional neural network model that can deeply mine the voiceprint information in the speech spectrum, and have better recognition effect is concluded. Using voiceprint recognition technology based on convolutional neural network, a dispatch order verification system is constructed. Through the verification experiment, it can be seen that the system can recognize the personnel as high as 95.42%,and at the same time overcomes the defects of the traditional scheme that requires full-variable modeling, reduces the system complexity, and has noise resistance. The system can help identify the true identities of the on-site ordering personnel remotely and dispatching order personnel, improving the safety and efficiency of dispatching operations, and has practical application value.