一种实时电能质量扰动识别分类方法

针对电能质量扰动数据大、识别算法繁琐,难以实现在线实时识别等问题,提出了基于深度卷积神经网络AlexNet的电能质量扰动识别数法,首先将各类电能质量扰动转化为图片格式,然后输入到AlexNet算法,通过学习、调整电能质量扰动信号的特征参数,迭代收敛,最后将实时的电能质量扰动通过训练好的AlexNet,直接实现扰动识别分类。实时仿真结果表明,所提出的方法能精确识别包括3种复合扰动在内的17种电能质量扰动问题,只需要对电能质量扰动信号进行学习,即可以直接对电能质量扰动信号进行识别与分类,识别算法简单且处理的时间短,达到了实时性的目的。

Real-time power quality disturbance recognition method

In view of such problems as the large amount of data of disturbance signals in power system, the numerous links of current identification algorithms, and the lack of real-time power quality disturbance identification, AlexNet classification algorithm is used in deep convolution neural network for power quality disturbance identification. Firstly, the power quality disturbances are converted into picture format and input into AlexNet algorithm. By learning and adjusting the characteristic parameters of the power quality disturbance signal, iterative convergence is realized. Finally, the real-time power quality disturbances are recognized and classified directly by AlexNet. Real-time simulation shows that this method can accurately identify 17 kinds of power quality disturbances including three kinds of compound disturbances. The method can recognize and classify power quality disturbance signals directly after learning the power quality disturbance signals. The recognition algorithm is simple and the processing time is short, which achieves the purpose of real-time.