一种新的基于深度置信网络的电能质量扰动分类方法

A novel classificiation method for power quality disturbance based on deep belief network

针对在噪声干扰下多重扰动识别正确率不高的问题,提出一种新的基于深度置信网络的电能质量扰动分类方法。首先,对电能质量扰动信号进行平稳小波多尺度变换,再利用软阈值函数处理估计小波系数重构原始信号,从而实现对电能质量扰动信号的去噪。再利用软阈值函数处理估计小波系数重构原始信号,从而实现对电能质量扰动信号的去噪。然后进一步提出利用深度置信网络对重构后的单一扰动信号和多重扰动信号进行分类识别。最后算例显示,即使在20 dB噪声干扰下,其分类正确率高达到93%以上。结果表明该方法对7种单一扰动和13种多重扰动信号的识别正确率均较高,验证该方法具有较强的抗噪声干扰能力。

Aiming at the problem that the recognition accuracy of multiple disturbances is not high under noise interference,a new classification method of power quality disturbances based on deep belief network is proposed.Firstly,the stationary wavelet multi‑scale transformation is performed on the power quality disturbance signal,and then the soft threshold function is used to process the estimated wavelet coefficients to reconstruct the original signal,thereby realizing the denoising of the power quality disturbance signal.Moreover,it is further proposed to use the deep belief network to classify and identify the reconstructed single disturbance signal and multiple disturbance signals.The calculation example shows that even under the interference of 20 dB noise,the classification accuracy rate is as high as 93%.The results show that the recognition accuracy of the method is high for 7 kinds of single disturbance and 13 kinds of multiple disturbance signals,which verifies that the method has strong anti‑noise interference ability.