Intelligent Operation And Maintenance Knowledge Base System Based On Deep Learning

In the iron and steel metallurgy industry,operation and maintenance work is the basis for ensuring the normal operation of various equipment,the knowledge base is an effective way to improve the efficiency of equipment operation and maintenance management.By building an operation and maintenance knowledge base,iron and steel plants can effectively accumulate knowledge and reduce the technical threshold of operation and maintenance.At the same time,the knowledge base can also reduce operation and maintenance costs.Aiming at the problems of single fault identification data source and difficulty in updating rules in the traditional operation and maintenance knowledge base,a method combined with artificial intelligence algorithm is adopted to construct an intelligent operation and maintenance knowledge base system based on deep learning.The main research contents and innovations of this paper are as follows:(1)An intelligent operation and maintenance knowledge base system architecture based on deep learning is proposed.Starting from the equipment operation and maintenance requirements of iron and steel plants,this paper analyzes the equipment operation and maintenance process based on the traditional knowledge base,and summarizes the problems that the traditional operation and maintenance knowledge base has a single source of fault identification data and the difficulty of updating the rules of the knowledge base.By integrating the inspection image classification model and the short text similarity model on the basis of the traditional operation and maintenance knowledge base system,combined with the representation of the knowledge graph,the design of the new operation and maintenance knowledge base system is completed.(2)Aiming at the problem of identifying fault phenomena in the operation and maintenance knowledge base,a patrol image classification model DRSF-Caps Net(Deep Residual Shrinkage Fusion Caps Net)is proposed.The feature extraction structure of capsule network is improved by using deep residual shrinkage network combined with feature fusion strategy.The model robustness problem caused by image noise is solved by the deep residual shrinkage network structure.At the same time,the feature fusion strategy effectively copes with the multi-scale problem of images.Experiments show that the improved model has high classification accuracy and good robustness.(3)Aiming at the semantic matching between work order fields and knowledge base rule fields,a short text similarity model TSN-MAF(Text Similarity Network based on MultiAttention Fusion)is proposed.Through the improved multi-attention mechanism fusion module,the model can screen out important features more comprehensively,thereby solving the problem of model information overload.Experiments show that the short text similarity model has a high accuracy in the semantic matching task between the work order field and the knowledge base rule field,and can better assist the knowledge base rule update.In this paper,combined with inspection images and operation and maintenance work order data,the deep learning algorithm is integrated into the operation and maintenance knowledge base,which improves the knowledge base’s ability to mine operation and maintenance data.By studying the inspection image classification model,more equipment failure information that cannot be found only by the structured data collected by the Internet of Things can be mined from the inspection image data.By studying the short text similarity model,the semantic matching between the work order and the rule field in the knowledge base is realized,which assists the update of the rules of the operation and maintenance knowledge base.This research provides ideas for the application of artificial intelligence in the field of industrial operation and maintenance.