微型电脑应用

针对电能质量扰动信号种类繁多、分类精度低以及现有分类算法抗噪性低、鲁棒性差的问题，文章提出一种改进卷积神经网络的电能质量扰动信号分类方法。采用S变换提取电能质量扰动信号特征，采用分类回归树算法对提取特征进行选择，采用改进CNN卷积神经网络对电能质量扰动信号进行分类。测试结果表明，在SNR=15 dB和SNR=25 dB的强噪声环境下，3种改进子模型在一定的迭代次数后，对8种不同扰动信号的训练精度都可达99%以上，同时对电能质量扰动信号的平均识别率达96%以上，表现出很强的抗噪性、高鲁棒性和和高分类精度的特点。

Aiming at the problems of disturbance signals, low classification accuracy, low noise resistance and poor robustness, an improved convolution neural network power quality disturbance signal classification method is proposed. S-transform is used to extract the characteristics of power quality disturbance signal. Then the classification regression tree algorithm is used to select the extracted features. The improved CNN convolution neural network is used to classify the power quality disturbance signals. The testing results show that under the strong noise environment with SNR = 15 dB and SNR = 25 dB, after a certain number of iterations, the training accuracy of the three improved sub models for 8 different disturbance signals can reach more than 99%, and the average recognition rate for power quality disturbance signals can reach more than 96%, showing the characteristics of strong noise resistance, high robustness and high classification accuracy.