Search

The diagnosis of blade crack faults is critical to ensuring the safety of turbomachinery. Blade tip timing (BTT) is a non-contact vibration displacement measurement technique, which has been extensively studied for blade vibration condition monitoring recently. The fault diagnosis methods based on deep learning can be summarized as studying the internal logical relationship of data, automatically mining features, and intelligently identifying faults. This research proposes a crack fault diagnostic method based on BTT measurement data and convolutional neural networks (CNNs) for the crack fault detection of blades. There are two main aspects: the numerical analysis of the rotating blade crack fault diagnosis and the experimental research in rotating blade crack fault diagnosis. The results show that the method outperforms many other traditional machine learning models in both numerical models and tests for diagnosing the depth and location of blade cracks. The findings of this study contribute to the real-time online crack fault diagnosis of blades. [ABSTRACT FROM AUTHOR]

Bearing fault diagnosis is an important research field for rotating machinery health monitoring. Recently, many intelligent fault diagnosis methods driven by big data, such as transfer learning, have been studied. However, there are two shortcomings for the prior transfer learning method in industry application. First, it is necessary to design a complex loss function to enhance the similarity between the two domains further. Second, previous studies required big data both in source and target task, without considering the lack of sufficient training samples. Inspired by relevant research work, this article proposes a local joint distribution discrepancy to increase similar features. A sub-domain adaptive transfer learning is designed to detect bearing faults based on the residual network. Two kinds of transfer experiments are designed to verify the method effectiveness. After that, the impact of small training samples and noise on the results is explored. The proposed method reaches high accuracy. [ABSTRACT FROM AUTHOR]

• An experimental dataset of flow boiling heat transfer in mini channels with pin fins was built. • A transfer learning framework was proposed to predict the heat transfer coefficient. • The effects of different factors on heat transfer coefficient prediction accuracy was identified. Flow boiling in mini channels with micro pin fins is a promising heat sink technique to achieve high-efficient aircraft thermal management. The accurate prediction of its heat transfer coefficient is critical for the practical design of two-phase heat exchanger based on mini channels with micro pin fins. Previous investigation shows that heat transfer coefficient prediction accuracy of the machine learning method is generally better than that of conventional empirical correlations. However, the machine learning method cannot guarantee its prediction accuracy among different data domains. To extend the application region of the conventional machine learning method, a transfer learning framework was proposed in present study. First of all, an experimental system was built to acquire test data from different sample domains (i.e., the diamond pin fins with different geometries). Then a conventional machine learning model was developed based on the deep learning method. Furthermore, the developed deep learning model was adjusted with transfer learning process, and the performance of these two kinds of models (i.e., conventional machine learning model and transfer learning model) was comprehensively evaluated. Results showed that the conventional machine learning model had a good prediction accuracy with an overall deviation of 4.11 % but was only confined in the same data domain as training data. Differently, the transfer learning model with 70 % data set from the new domain can achieve appropriate prediction in the new domains with deviation of 4.28 %. The results of this paper demonstrated that the conventional machine learning model can extend into different domains with reasonable prediction accuracy through transfer learning frameworks. It will benefit the practical design of high-efficiency heat exchangers for aircraft thermal management. [ABSTRACT FROM AUTHOR]

The actual operation of turbomachinery is inevitably affected by multi-source uncertainties. Such uncertainties are detrimental to the performance and reliability of energy systems. Based on graph learning methods, this work aims to provide a convenient and effective approach for aerodynamic robust optimization of turbomachinery. A radial inflow turbine is taken as the research target and Dual Graph Neural Network (DGNN) regression model is constructed for flow field prediction and performance discrimination. By comparing the accuracy and time consumption, the advantages of DGNN over classical surrogate models and computational fluid dynamics (CFD) are clarified. The proposed model is integrated into uncertainty quantification and aerodynamic robust optimization. The effect of multi-source uncertainties on performance is quantified. The stochastic response of flow fields is also obtained conveniently through DGNN. Robust optimization is performed for power and efficiency, respectively. The power robust optimization improves the power by 1.52% and reduces the standard deviation of power by 15.45%. The efficiency robust optimization achieves an efficiency improvement of 1.76% (increment) and an efficiency standard deviation reduction of 36.82%. The proposed approach is an efficient and competitive choice for uncertainty quantification and robust optimization. The present work contributes to constructing the digital twin of turbomachinery systems. • A new complete approach for aerodynamic robust optimization. • A graph deep learning model is contained for flow fields and performance prediction. • Our model performs better than classical surrogate models. • The effect of multi-source uncertainties is quantified. • The proposed approach can effectively solve aerodynamic robust optimization problems. [ABSTRACT FROM AUTHOR]