Double-attention-guided rotary machine health assessment method

Abstract

The invention relates to a double-attention-guided rotating machine health assessment method, and the method comprises the steps: constructing a double-attention-guided model which comprises a feature extractor, a reconstructor and a regression device, and introducing a deformable convolution thought into a feature extraction method, capturing fine-grained features by deformable convolution and time convolution and by skillfully combining with a double-attention channel; aggregating local key information so that information loss of global feature learning is improved; adopting regression loss and reconstruction loss to update the double-attention guidance model, and outputting optimal training mechanical degradation features; and inputting the test set into the updated feature extractor, obtaining test mechanical degradation features, and minimizing the difference between the optimal training and test mechanical degradation features by using the multi-core maximum mean difference loss. The problem of poor prediction precision of a traditional method can be well solved, the information consistency of the data can be restrained, and high-precision residual life prediction of the test data is realized.

Description

technical field [0001] The invention relates to the technical field of mechanical health feature assessment based on machine learning methods, in particular to a double-attention-guided rotating machinery health assessment method. Background technique [0002] During long-term operation, rotating machinery is prone to failure due to factors such as load, wear and cracks, which in turn leads to problems in equipment operation. Therefore, the fault and health management (PHM) technology can be used to diagnose the health status of the machinery according to the behavior of the machinery, make maintenance plans in advance, shorten unnecessary downtime, and reduce production costs. Prediction of the remaining service life of mechanical systems is a key task of PHM technology. Current machinery health assessment methods can be divided into model-based and data-driven methods. Model-based methods typically employ statistical models of mechanical degradation data to predict lifet...

AI Technical Summary

Problems solved by technology

This shortcoming tends to limit regression performance for many real-world applications and may not work effectively

Generally speaking, the pre-trained model can achieve the best performance under similar working conditions, but for mechanical systems, although the working conditions are similar, the fault types are compound, which leads to differences in feature distribution

Images