Research on equipment failure analysis model based on big data
The research focuses on the accuracy of equipment fault diagnosis and adopts a method combining bidirectional long short-term memory network and association rule classifier to improve the accuracy of equipment fault diagnosis.This model is trained on word vectors using Word2vec and applies SMOTE technology to address data imbalance issues.The results showed that the per-formance was best when the support and confidence thresholds were set to 0.28 and 0.55,respectively,achieving an accuracy rate of 84.9％and a recall rate of 83.85％.The bidirectional long short-term memory network also performs well in feature extraction,with accuracy,recall,and F1 values of 95.6％,96.2％,and 95.9％,respectively.It has been proven that the method combining bidi-rectional long short-term memory network and association rule classifier is effective in improving the accuracy of fault diagnosis,pro-viding strong technical support for the maintenance of on-board equipment,and has significant practical application value.
big databidirectional short-duration memory networkequipmentmalfunctionassociation rule
万方数据
维普
