基于增量学习优化的故障录波文件通道名称识别方法

智能变电站不同建设时期各类录波厂家配置的双套录波通道名称命名习惯不同，导致故障录波文件相同通道不同设备命名不同。后期采用人工方式修改工作量大、所需时间长，且无法保证结果的正确性。针对此问题，提出一种基于增量学习优化的录波文件通道名称识别方法。首先，从故障录波配置文件中提取通道名称并进行文本预处理。其次，使用基于增量学习优化的Word2vec模型实现通道名称中文词向量的生成与在线学习。然后，采用余弦相似度和逆文本频率相结合的文本相似度匹配算法实现录波文件通道名称识别。最后，通过录波文件中提取的通道名称构成实验数据进行实验。算例结果表明，所提方法有效地提高了录波文件通道名称识别的自适应性和准确性。

Recognition method of fault recorder file channel name based on incremental learning optimization

The naming conventions of dual sets of fault recorders configured by various manufacturers in different construction periods of a smart substation are different. This leads to inconsistent names of different devices in the same channel of fault recorder files. Manual amendment at a later stage involves a heavy workload, takes a long time and cannot ensure the correctness of the results. Thus a method is proposed for identifying the names of recorder file channels based on incremental learning optimization. First, the channel names are extracted from fault recorder configuration files and text preprocessing is implemented. Second, the Word2vec model based on incremental learning optimization is employed to achieve the generation and online learning of Chinese word vectors for the channel names. Third, a text-similarity matching algorithm that combines cosine similarity and inverse document frequency (IDF) is adopted to realize channel name identification. Finally, experiments are carried out with empirical data obtained from the channel names extracted from recorded files. The results show that the proposed method has effectively improved the adaptivity and accuracy of name identification of recorder file channels.

This work is supported by the National Natural Science Foundation of China (No. 51877084).