Turbine vibration fault diagnosis method based on deep learning artificial neural network

Abstract

The invention belongs to the technical field of turbine vibration fault diagnosis, and relates to turbine vibration fault diagnosis adopting deep learning, namely a turbine vibration fault diagnosis method based on a deep learning artificial neural network. The invention discloses a turbine vibration fault diagnosis method based on a deep learning artificial neural network. Parameter characteristics and connection among parameters when the turboset system vibrates are comprehensively considered. Data are preprocessed through multivariate analysis, so that an independence relation among variousdata is established, redundant data are reduced, a basic feature table is constructed, then deep learning artificial neural network training is carried out, and the trained deep learning artificial neural network is obtained to diagnose the vibration fault of the steam turbine. The method is based on the deep learning artificial neural network, and is characterized in that the deep learning artificial neural network is used, so that the construction of a calculation model is avoided, and meanwhile, the dimension reduction is carried out by using multivariate analysis during data preprocessing.

Description

technical field [0001] The invention belongs to the technical field of steam turbine vibration fault diagnosis, and relates to a steam turbine vibration fault diagnosis using deep learning, that is, a steam turbine vibration fault diagnosis method based on a deep learning artificial neural network. Background technique [0002] At the moment when high-tech is advancing by leaps and bounds, the data of steam turbine vibration fault detection has entered the era of big data. Massive data not only provides sufficient analysis sources for the steam turbine vibration fault diagnosis system, but also brings interference to the system from redundant data. The vibration fault of the steam turbine has the characteristics of many types of detection data, a large amount of data, and high collection density. If traditional diagnosis methods are used, it will lead to adverse consequences such as huge workload and long working hours. How to efficiently carry out steam turbine fault diagn...

AI Technical Summary

Problems solved by technology

[0004] At present, there are few attempts to use deep learning in the field of steam turbine vibration fault diagnosis at home and abroad, and there is a lack of expe...

Images