Fault discrimination method, system and device based on small sample self-learning fault migration

Abstract

The invention belongs to the field of industrial equipment fault discrimination, particularly relates to a fault discrimination method, system and device based on small sample self-learning fault migration, and aims to solve the problems that a model in the prior art needs a large amount of label data training and cannot perform online evaluation and updating according to new fault data accumulated by continuous operation of equipment. The method comprises the steps: performing target equipment operation state diagnosis by using a small amount of labeled sample data of target equipment by taking an equipment operation database which does not contain target equipment information as a support; achieving the knowledge migration from the equipment operation database to the model through feature extraction in combination with transfer learning; introducing the fault diagnosis knowledge into model training, verifying new and low-confidence fault types, updating an equipment operation database and a target data set, and realizing the knowledge migration of the fault diagnosis knowledge to the model. When the fault data of the target equipment is insufficient, the equipment operation database and the fault discrimination model are continuously updated, and the fault diagnosis capability of the model is improved.

Description

technical field [0001] The invention belongs to the field of fault discrimination of industrial equipment, and in particular relates to a fault discrimination method, system and device based on small-sample self-learning fault migration. Background technique [0002] The industrial production environment is complex, a device usually has several different working states, and different devices also have similar fault types. The traditional fault identification method requires experts to deeply understand the operation mechanism of equipment, and design mechanism models for different production environments. The design requirements are high and difficult. Sensors and other equipment installed in complex industrial environments collect and store a large amount of production data. These industrial data are large in volume and contain equipment health status information, which can help improve product functions and diagnose product failures in advance. [0003] Deep learning meth...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the above-mentioned problems in the prior art, that is, the prior art model training requires a large amount of label data and cannot perform online evaluation and update of the model based on the new fault data accumulated by the continuous operation of the equipment, the present invention provides a small-sample-based A fault discrimination method for self-learning fault migration, the fault discrimination method includes:

Images