Electrical Equipment Monitoring Data Processing Based On Text Recognition Technology

With the continuous improvement of the social economy,the requirements for the work of electrical equipment are getting higher and higher.As an important part of the current energy industry,a large amount of fault text data is generated in the production process.How to effectively deal with a large amount of Chinese text failure data accumulated in the power system is of great significance for the research of electrical equipment monitoring data.At present,most of the power system fault texts are classified using a single model,and the accuracy needs to be further improved.Aiming at this situation,constructing a variety of text classification models and comparing them is beneficial to improve the accuracy of fault classification of power equipment.Power equipment failure texts require manual screening by professionals before classification.The traditional machine learning model needs to feature extraction,space vector representation and model training on the filtered data.Deep learning model for manual labeling of text sequences,and directly submits the filtered text to the classifier for training.Next,according to the accuracy of machine learning classification model and the deep learning classification model are used as the starting point,a variety of classification models are constructed and compared to obtain a model with high classification accuracy.According to the classification of electrical equipment fault texts,five traditional machine learning classification models(SVM?KNN?NB?GBM?LR)were constructed to study the fault data classification accuracy of these models.The research process includes text segmentation and removal of stop words by the Viterbi algorithm of the hidden Markov model for the filtered fault data.Then,the pre-processed data is extracted from the text data by the CHI and MI,and the M-CHI method improved by the CHI and MI.The space vector model is used to obtain the word vector matrix,and then the word vector matrix is respectively placed into the five machine learning training models.The experimental results show that the improved M-CHI feature extraction method improves the feature extraction value of fault data and the support vector machine model improves the accuracy of the classification model based on this method.Aiming at the research of text classification in deep learning,based on the LSTM network model,in order to strengthen the association of fault text context information and improve the information discontinuity leading to classification errors,a two-layer bidirectional LSTM model is constructed.Then use the Skip-Gram architecture of the Word2 vec to train the word vector.Combining the deep attention mechanism with pay attention to the weight of different word vectors in text data,and a DA-BiLSTM model based on deep attention mechanism is constructed and compared with machine learning model.The experimental results show that by performing comprehensive training tests on the performance of the model proves that the classification accuracy of the model is better than the machine learning classification model,which provides a valuable reference for Chinese text classification of power system.