基于平稳小波与BP神经网络的换相失败检测算法

针对高压直流输电系统中换相失败检测问题,提出了一种采用平稳小波分析和BP神经网络的换相失败检测算法.通过平稳小波提取换相失败信号不同尺度的小波能量,作为特征向量输入神经网络中进行训练,并得到能够进行自动化识别的分类模型.在实际采集得到的200组数据集上进行了算法验证,结果表明,文中算法可以有效地区分直流输电系统中的换相失败和正常信号,其平均检测精确度达到95%以上,为进一步系统准确无功补偿提供保障.

Commutation failure detection algorithm based on stationary wavelet and BP neural network

In order to solve the commutation failure detection problem in high voltage direct current transmission(HVDC)system, a commutation failure detection algorithm based on stationary wavelet analysis and BP neural network was proposed. The commutation failure signals and wavelet energy with different scales were extracted as the feature vectors through the stationary wavelet. In addition, the feature vectors were inputted into the neural network for training, and a classification model for automatic recognition was obtained. The proposed algorithm was verified with the actually collected 200 data sets. The results indicate that the proposed algorithm can effectively distinguish the commutation failure and normal signals in the HVDC system, and the average detection accuracy can reach above 95%. It is obvious that the proposed algorithm can provide the guarantee for the further precise reactive power compensation in the system.