Deep learning has been employed to prognostic and health management of automotive and aerospace with promising results. Literature in this area has revealed that most contributions regarding deep learning is largely focused on the model’s architecture. However, contributions regarding improvement of different aspects in deep learning, such as custom loss function for prognostic and health management are scarce. There is therefore an opportunity to improve upon the effectiveness of deep learning for the system’s prognostics and diagnostics without modifying the models’ architecture. To address this gap, the use of two different dynamically weighted loss functions, a newly proposed weighting mechanism and a focal loss function for prognostics and diagnostics task are investigated. A dynamically weighted loss function is expected to modify the learning process by augmenting the loss function with a weight value corresponding to the learning error of each data instance. The objective is to force deep learning models to focus on those instances where larger learning errors occur in order to improve their performance. The two loss functions used are evaluated using four popular deep learning architectures, namely, deep feedforward neural network, one-dimensional convolutional neural network, bidirectional gated recurrent unit and bidirectional long short-term memory on the commercial modular aero-propulsion system simulation data from NASA and air pressure system failure data for Scania trucks. Experimental results show that dynamically-weighted loss functions helps us achieve significant improvement for remaining useful life prediction and fault detection rate over non-weighted loss function predictions.
There is an increasing need for the use of additive manufacturing (AM) to produce improved critical application engineering components. However, the materials manufactured using AM perform well below their traditionally manufactured counterparts, particularly for creep and fatigue. Research has shown that this difference in performance is due to the complex relationships between AM process parameters which affect the material microstructure and consequently the mechanical performance as well. Therefore, it is necessary to understand the impact of different AM build parameters on the mechanical performance of parts. Machine learning (ML) models are able to find hidden relationships in data using iterative statistical analyses and have the potential to develop process–structure–property–performance relationships for manufacturing processes, including AM. The aim of this work is to apply ML techniques to materials testing data in order to understand the effect of AM process parameters on the creep rate of additively built nickel-based superalloy and to predict the creep rate of the material from these process parameters. In this work, the predictive capabilities of ML and its ability to develop process–structure–property relationships are applied to the creep properties of laser powder bed fused alloy 718. The input data for the ML model included the Laser Powder Bed Fusion (LPBF) build parameters used—build orientation, scan strategy and number of lasers—and geometrical material descriptors which were extracted from optical microscope porosity images using image analysis techniques. The ML model was used to predict the minimum creep rate of the Laser Powder Bed Fused alloy 718 samples, which had been creep tested at $$650\,^\circ $$ 650 ∘ C and 600 MPa. The ML model was also used to identify the most relevant material descriptors affecting the minimum creep rate of the material (determined by using an ensemble feature importance framework). The creep rate was accurately predicted with a percentage error of $$1.40\%$$ 1.40 % in the best case. The most important material descriptors were found to be part density, number of pores, build orientation and scan strategy. These findings show the applicability and potential of using ML to determine and predict the mechanical properties of materials fabricated via different manufacturing processes, and to find process–structure–property relationships in AM. This increases the readiness of AM for use in critical applications.
When machine learning supports decision-making in safety-critical systems, it is important to verify and understand the reasons why a particular output is produced. Although feature importance calculation approaches assist in interpretation, there is a lack of consensus regarding how features' importance is quantified, which makes the explanations offered for the outcomes mostly unreliable. A possible solution to address the lack of agreement is to combine the results from multiple feature importance quantifiers to reduce the variance of estimates. Our hypothesis is that this will lead to more robust and trustworthy interpretations of the contribution of each feature to machine learning predictions. To assist test this hypothesis, we propose an extensible Framework divided in four main parts: (i) traditional data pre-processing and preparation for predictive machine learning models; (ii) predictive machine learning; (iii) feature importance quantification and (iv) feature importance decision fusion using an ensemble strategy. We also introduce a novel fusion metric and compare it to the state-of-the-art. Our approach is tested on synthetic data, where the ground truth is known. We compare different fusion approaches and their results for both training and test sets. We also investigate how different characteristics within the datasets affect the feature importance ensembles studied.
High loads and bearing life requirements make journal bearings a potential choice for use in high power, epicyclic gearboxes in jet engines. Particularly in a planetary configuration the kinematic conditions are complex. With the planet gears rotating about their own axis and orbiting around the sun gear, centrifugal forces generated by both motions interact with each other and affect the external flow behavior of the oil exiting the journal bearing. Computational Fluid Dynamics (CFD) simulations using the Volume of Fluid (VoF) method are carried out in ANSYS Fluent [1] to numerically model the two-phase flow behavior of the oil exiting the bearing and merging into the air surrounding the bearing. This paper presents an investigation of two numerical schemes that are available in ANSYS Fluent to track or capture the air-oil phase interface: the geometric reconstruction scheme and the compressive scheme. Both numerical schemes are used to model the oil outflow behavior in the most simplistic approximation of a journal bearing: a representation, rotating about its own axis, with a circumferentially constant, i.e. concentric, lubricating gap. Based on these simplifications, a three dimensional (3D) CFD sector model with rotationally periodic boundaries is considered. A comparison of the geometric reconstruction scheme and the compressive scheme is presented with regards to the accuracy of the phase interface reconstruction and the time required to reach steady state flow field conditions. The CFD predictions are validated against existing literature data with respect to the flow regime, the direction of the predicted oil flow path and the oil film thickness. Based on the findings and considerations of industrial requirements, a recommendation is made for the most suitable scheme to be used. With a robust and partially validated CFD model in place, the model fidelity can be enhanced to include journal bearing eccentricity. Due to the convergent-divergent gap and the resultant pressure field within the lubricating oil film, the outflow behavior can be expected to be very different compared to that of a concentric journal bearing. Naturally, the inlet boundary conditions for the oil emerging from the journal bearing into the external environment must be consistent with the outlet conditions from the bearing. The second part of this paper therefore focuses on providing a method to generate appropriate inlet boundary conditions for external oil flow from an eccentric journal bearing.
When machine learning supports decision-making in safety-critical systems, it is important to verify and understand the reasons why a particular output is produced. Although feature importance calculation approaches assist in interpretation, there is a lack of consensus regarding how features’ importance is quantified, which makes the explanations offered for the outcomes mostly unreliable. A possible solution to address the lack of agreement is to combine the results from multiple feature importance quantifiers to reduce the variance in estimates and to improve the quality of explanations. Our hypothesis is that this leads to more robust and trustworthy explanations of the contribution of each feature to machine learning predictions. To test this hypothesis, we propose an extensible model-agnostic framework divided in four main parts: (i) traditional data pre-processing and preparation for predictive machine learning models, (ii) predictive machine learning, (iii) feature importance quantification, and (iv) feature importance decision fusion using an ensemble strategy. Our approach is tested on synthetic data, where the ground truth is known. We compare different fusion approaches and their results for both training and test sets. We also investigate how different characteristics within the datasets affect the quality of the feature importance ensembles studied. The results show that, overall, our feature importance ensemble framework produces 15% less feature importance errors compared with existing methods. Additionally, the results reveal that different levels of noise in the datasets do not affect the feature importance ensembles’ ability to accurately quantify feature importance, whereas the feature importance quantification error increases with the number of features and number of orthogonal informative features. We also discuss the implications of our findings on the quality of explanations provided to safety-critical systems.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.