深度信念网络在电能质量复合扰动识别中的应用

Application of Deep Belief Network in Power Quality Compound Disturbance Identification

针对电能质量复合扰动信号数据量大、识别效率低的问题,提出一种基于压缩感知理论和深度信念网络的电能质量复合扰动识别方法。首先通过压缩感知理论得到原始扰动信号的测量值,并基于正交匹配追踪算法求得稀疏向量;其次构建深度信念网络分类模型,将稀疏向量作为网络的输入,通过对比散度学习算法训练网络,实现扰动信号的分类;最后对14种常见扰动信号进行仿真验证,仿真结果表明,该方法有效地减少了所需处理的扰动数据量,并且对单一扰动和复合扰动都有效且具有很高的识别效率。

For composite power quality disturbance signal of large amount of data and inefficient identification prob-lems,put forward a kind of method based on compressed sensing and deep belief network to achieve composite power quality disturbance identification.Firstly,through compressed sensing theory to gain a measured value and obtain sparse vector based on orthogonal matching pursuit(OMP)algorithm.Secondly,construct a classification model of deep belief network,the sparse vector as input to this network,the classification of disturbance signals are realized by train-ing this network based on contrastive divergence learning algorithm.Finally,The 14 common disturbances are simulat-ed,it is indicated that this method can effectively reduce the amount of disturbance data needed for processing,it is ef-fective to single and compound disturbance and had high recognition efficiency.