Power terminal safety monitoring method based on improved LSTM neural network

A power terminal safety monitoring method based on an improved LSTM neural network belongs to the technical field of power system safety monitoring, on the basis of an LSTM, the basic structure of the LSTM is not changed, two gate operations are added, through interaction of a hidden layer and input information, the context modeling capability of a model is enhanced, the overall performance of the LSTM is improved, and the LSTM is more suitable for a bypass signal detection method. Therefore, the safety detection precision of the power terminal is improved. By adopting the method to screen the features, the calculation amount can be greatly reduced while the monitoring precision is ensured as far as possible. 基于改进LSTM神经网络的电力终端安全监测方法，属于电力系统安全监测技术领域，在LSTM基础上，不改变其基本结构，增加两个门运算，通过隐含层和输入信息的交互，增强模型上下文建模能力，改进LSTM整体性能，使其更适合旁路信号检测方法，从而提高对电力终端的安全检测精度；采用本发明方法筛选特征可以在尽可能保证监测精度的同时，大大降低计算量。