基于改进DenseNet模型的电压暂降故障识别方法

Voltage sag fault recognition method based on improved DenseNet model

为了解决电力系统中常见的电压暂降故障的识别问题,文章提出了一种基于改进DenseNet模型的识别方法。首先,将电压信号从时域空间转换到频率空间,并对瞬时电压进行空间坐标变换,将三维静止坐标系转换到二维静止坐标系。其次,将二维坐标矢量轨迹图、电压瞬时波形图和频率空间图像作为改进DenseNet模型的输入,并对这些图像进行二值化处理。最后,在过渡层中引入注意力机制,帮助DenseNet模型更好地集中注意力于关键的特征,在全局平均池化层增加自适应池化功能,以适应处理任意尺寸的图像。实验结果表明该方法能够有效地识别电压暂降故障。

To address the recognition problem of voltage sag faults in power systems,this paper proposes a recognition method based on an improved DenseNet model.Firstly,the voltage signals are transformed from the time domain to the frequency domain,and the instantaneous voltage is subjected to spatial coordinate transformation,converting the three-dimensional stationary coordinate system to a two-dimensional stationary coordinate system.Then,the two-dimensional coordinate vector trajectory plot,voltage instantaneous waveform plot,and frequency domain image are used as inputs to the improved DenseNet model,and these images are subjected to binary processing.Finally,an attention mechanism is introduced in the transition layer to help the DenseNet model focus more on key features,add adaptive pooling functionality to the global average pooling layer,enabling it to handle images of different sizes.Experimental results demonstrate that this method can effectively identify voltage sag faults.