Industrial process small sample fault diagnosis method based on deep optimal feature transmission

Abstract

The invention discloses an industrial process small sample fault diagnosis method based on deep optimal feature transmission, and the method comprises the steps: constructing a reduced source domain and target domain deep adaptation network based on domain adaptation, carrying out the analysis of industrial small sample fault data, and rapidly building an industrial small sample fault diagnosis quantitative model based on deep optimal feature transmission; constructing a fault adaptation task set according to the historical fault adaptation tasks; representing the fault adaptation task set in an embedding space; and carrying out model identification and parameter estimation on the normal form learned by the embedded space, and updating the optimal characteristics of the transmission depth of the model, thereby obtaining a fault classification result under an unknown fault task. According to the method, the fault type of the sparse heterogeneous industrial signals under different working conditions can be accurately predicted, the algorithm complexity is low, the calculation period is short, and the portability is high.

Description

technical field [0001] The invention belongs to the technical field of industrial process fault diagnosis and processing, relates to a small-sample machine learning technology, and in particular relates to a small-sample fault diagnosis method for an industrial process based on deep optimal feature transmission. Background technique [0002] Industry is the lifeblood of the national economy. With the vigorous development of sensor technology, today's industrial field has gradually formed an environment suitable for industrial process data information and sensing technology. Therefore, the data-driven model-free intelligent diagnosis methods have been unprecedentedly developed. The success of these techniques depends on having sufficient labeled data, however, in practical applications, the collection of labeled data for industrial processes is difficult. It is mainly reflected in the heterogeneity and sparseness of industrial data distribution, such as industrial equipment ...

AI Technical Summary

Problems solved by technology

However, because industrial fault signals cannot provide a large number of reliable adaptable source domains, and the current domain adaptation method lacks an incremental learning mechanism, similarity learning is shallow, and there is no fast and accurate diagnostic method for industrial small-sample fault diagnosis. Therefore, It is urgent to effectively combine intelligent technology with industrial fault data analysis to form a complete fault diagnosis method for small sample problems

Images