Industrial process performance evaluation method of mixed deep residual shrinkage network and XGBoost algorithm

Abstract

The invention belongs to the field of automatic process control, and discloses an industrial process performance evaluation method of a mixed deep residual shrinkage network and an XGBoost algorithm. The invention develops a novel fault diagnosis method mixing an XGBoost algorithm and a DRSN. The residual learning in the DRSN can effectively avoid the problem of model degradation, and the soft threshold operation can effectively reduce the influence of noise and redundant information on feature learning. In addition, a Nadam optimization algorithm with a better effect is used to update network parameters instead of a traditional Adam algorithm. In the final classification stage, an XGBoost classification algorithm is adopted to carry out fault identification and classification on the extracted feature information instead of a traditional softmax classifier. Experimental results show that the XGBoost has better classification performance.

Description

technical field [0001] The invention belongs to the field of automatic process control, and in particular relates to an industrial process performance evaluation method combining deep residual shrinkage network and XGBoost algorithm. Background technique [0002] The emergence of modern fault diagnosis technology has played an important role in ensuring the production safety of industrial processes and reducing resource waste. It has successively developed model-based methods, knowledge-based reasoning methods and data-driven methods. Due to their own limitations, model-based methods and knowledge-based reasoning methods cannot achieve satisfactory results for complex industrial process data that is characterized by high dimensionality, nonlinearity, intermittent and dynamic characteristics. The data-driven method has been better developed and applied due to its small limitations and only relying on past fault data. [0003] Data-driven methods can be further divided into m...

AI Technical Summary

Problems solved by technology

[0005] The extracted features contain a lot of redundant information, which affects the final diagnostic accuracy

[0006] The common softmax classifier cannot further improve the diagnostic accuracy

Images