Power operation problem classification system and method based on deep learning and application

The invention discloses a power operation problem classification system and method based on deep learning and application. According to the invention, a data acquisition module acquires operation problem data during operation of a worker, and the operation problem data comprises problem description, problem types and attributes. The data preprocessing module is used for dividing the collected data into a training set and a test set, and operation problem data classification of the training set is labeled manually and comprises problem description, problem types and attributes; and the problem description feature expression module is used for inputting the preprocessed problem description data into a pre-trained model BERT to obtain a problem description feature expression. The model construction module comprises a dimension reduction unit and a feature splicing unit. According to the method, the deep neural network model is utilized to classify the operation problems, objectivity, accuracy and high efficiency