Removing fiber orientation uncertainty from the finite element model of a composite lamina with direct updating algorithm

In this paper, a direct updating algorithm is proposed to remove the uncertainties present in the simulated/analytical finite element (FE) model of a composite material lamina. There are number of possible uncertainties present in the composite materials such as its constituent properties and its orientations, boundary conditions and its assumed dimensions etc. It is observed from this analytical study that the uncertainty present in the fiber orientation in the matrix put its direct effect on the modal-model (natural frequencies and corresponding mode shapes) of the composite material lamina. The direct updating algorithm has been already used for many isotropic structures. However, for anisotropic structures like composite materials, the application to accurate the simulated-finite element model by using finite element model updating techniques is a new area of research. In this regard, to remove these uncertainties from the simulated-finite element model of a composite lamina, the application of direct updating algorithm is proposed. It is observed from the present study that by updating the mass and the stiffness matrices through direct updating algorithm, the vibration pattern of the mode shapes are updated. It is found that the maximum percentage error in the constituent properties and in the fiber orientation is 22.58% and 100% respectively that are reduced to 0% in the modal-model of the lamina by the application of direct updating algorithm. This represents the novelty of the application of direct updating method for composite lamina structures. The overlay of frequency response function (FRF) curves are plotted to authenticate the results. Also, it is found that the application of the direct updating algorithm increases the tracking performance of the simulated FE model response when excited at different resonant frequencies.

An Approximation-Based Design Optimization Approach to Eigenfrequency Assignment for Flexible Multibody Systems

The use of flexible multibody simulation has increased significantly over recent years due to the increasingly lightweight nature of mechanical systems. The prominence of lightweight engineering design in mechanical systems is driven by the desire to require less energy in operation and to reach higher speeds. However, flexible lightweight systems are prone to vibration, which can affect reliability and overall system performance. Whether such issues are critical depends largely on the system eigenfrequencies, which should be correctly assigned by the proper choice of the inertial and elastic properties of the system. In this paper, an eigenfrequency assignment method for flexible multibody systems is proposed. This relies on a parametric modal model which is a Taylor expansion approximation of the eigenfrequencies in the neighborhood of a configuration of choice. Eigenfrequency assignment is recast as a quadratic programming problem which can be solved with low computational effort. The method is validated by assigning the lowest eigenfrequency of a two-bar linkage by properly adding point masses. The obtained results indicate that the proposed method can effectively assign the desired eigenfrequency.

Preparing For The Next Pandemic: Transfer Learning From Existing Diseases Via Hierarchical Multi-Modal BERT Models to Predict COVID-19 Outcomes

Developing prediction models for emerging infectious diseases from relatively small numbers of cases is a critical need for improving pandemic preparedness. Using COVID-19 as an exemplar, we propose a transfer learning methodology for developing predictive models from multi-modal electronic healthcare records by leveraging information from more prevalent diseases with shared clinical characteristics. Our novel hierarchical, multi-modal model (TransMED) integrates baseline risk factors from the natural language processing of clinical notes at admission, time-series measurements of biomarkers obtained from laboratory tests, and discrete diagnostic, procedure and drug codes. We demonstrate the alignment of TransMED's predictions with well-established clinical knowledge about COVID-19 through univariate and multivariate risk factor driven sub-cohort analysis. TransMED's superior performance over state-of-the-art methods shows that leveraging patient data across modalities and transferring prior knowledge from similar disorders is critical for accurate prediction of patient outcomes, and this approach may serve as an important tool in the early response to future pandemics.

An Experimentally Aided Operational Virtual Prototyping to Obtain the Best Spindle Speed during Face Milling of Large-Size Structures

The paper presents an original method concerning the problem of vibration reduction in the general case while milling large-size and geometrically complex details with the use of an innovative approach to the selection of spindle speed. A computational model is obtained by applying the so-called operational approach to identify the parameters of the workpiece modal model. Thanks to the experimental modal analysis results, modal subsystem identification was performed and reliable process data for simulation studies were obtained. Next, simulations of the milling process, for successive values of the spindle speed, are repeated until the best vibration state of the workpiece is obtained. For this purpose, the root mean square values of the time plots of vibration displacements are examined. The effectiveness of the approach proposed for reducing vibrations in the process of face milling is verified on the basis of the results of appropriate experimental investigations. The economic profitability of the implementation of the operational technique in the production practice of enterprises dealing with mechanical processing is demonstrated as well.

Identification of radiant source in an enclosure by reduced model

We propose an original method to recover from a few measurement points the integrity of the temperature field of a furnace heated by a radiant thermal source. The radiant thermal source is first identified via a low order reduced model based on based on AROMM (Amalgam Reduced Order Modal Model) method which preserves the integrity of the geometry. The minimization is performed via a trust-region reflective least squares algorithm implemented in MATLAB “lsqcurvefit” function. From that identified heat flux, the integrity of the thermal field is then recovered by direct simulation thanks to a reduced model of higher rank to have a better precision. The treated application is a complex titanium piece heated by two radiant panels placed in a furnace. With four measuring points, the temperature of the whole thermal scene is retrieved at all times with an average error around 1 K on the studied object.

FEATURE FUSION FOR CROSS-MODAL SCENE CLASSIFICATION OF REMOTE SENSING IMAGE

Scene classification plays an important role in remote sensing field. Traditional approaches use high-resolution remote sensing images as data source to extract powerful features. Although these kind of methods are common, the model performance is severely affected by the image quality of the dataset, and the single modal (source) of images tend to cause the mission of some scene semantic information, which eventually degrade the classification accuracy. Nowadays, multi-modal remote sensing data become easy to obtain since the development of remote sensing technology. How to carry out scene classification of cross-modal data has become an interesting topic in the field. To solve the above problems, this paper proposes using feature fusion for cross-modal scene classification of remote sensing image, i.e., aerial and ground street view images, expecting to use the advantages of aerial images and ground street view data to complement each other. Our cross- modal model is based on Siamese Network. Specifically, we first train the cross-modal model by pairing different sources of data with aerial image and ground data. Then, the trained model is used to extract the deep features of the aerial and ground image pair, and the features of the two perspectives are fused to train a SVM classifier for scene classification. Our approach has been demonstrated using two public benchmark datasets, AiRound and CV-BrCT. The preliminary results show that the proposed method achieves state-of-the-art performance compared with the traditional methods, indicating that the information from ground data can contribute to aerial image classification.

Rethinking Label-Wise Cross-Modal Retrieval from A Semantic Sharing Perspective

The main challenge of cross-modal retrieval is to learn the consistent embedding for heterogeneous modalities. To solve this problem, traditional label-wise cross-modal approaches usually constrain the inter-modal and intra-modal embedding consistency relying on the label ground-truths. However, the experiments reveal that different modal networks actually have various generalization capacities, thereby end-to-end joint training with consistency loss usually leads to sub-optimal uni-modal model, which in turn affects the learning of consistent embedding. Therefore, in this paper, we argue that what really needed for supervised cross-modal retrieval is a good shared classification model. In other words, we learn the consistent embedding by ensuring the classification performance of each modality on the shared model, without the consistency loss. Specifically, we consider a technique called Semantic Sharing, which directly trains the two modalities interactively by adopting a shared self-attention based classification model. We evaluate the proposed approach on three representative datasets. The results validate that the proposed semantic sharing can consistently boost the performance under NDCG metric.

Quantifying the structural uncertainty of the aerosol mixing state representation in a modal model

Aerosol mixing state is an important emergent property that affects aerosol radiative forcing and aerosol-cloud interactions, but it has not been easy to constrain this property globally. This study aims to verify the global distribution of aerosol mixing state represented by modal models. To quantify the aerosol mixing state, we used the aerosol mixing state indices for submicron aerosol based on the mixing of optically absorbing and non-absorbing species (χo), the mixing of primary carbonaceous and non-primary carbonaceous species (χc), and the mixing of hygroscopic and non-hygroscopic species (χh). To achieve a spatiotemporal comparison, we calculated the mixing state indices using output from the Community Earth System Model with the modal MAM4 aerosol module, and compared the results with the mixing state indices from a benchmark machine-learned model trained on high-detail particle-resolved simulations from the particle-resolved stochastic aerosol model PartMC-MOSAIC. The two methods yielded very different spatial patterns of the mixing state indices. In some regions, the yearly-averaged χ value computed by the MAM4 model differed by up to 70 percentage points from the benchmark values. These errors tended to be zonally structured, with the MAM4 model predicting a more internally mixed aerosol at low latitudes, and a more externally mixed aerosol at high latitudes, compared to the benchmark. Our study quantifies potential model bias in simulating mixing state in different regions, and provides insights into potential improvements to model process representation for a more realistic simulation of aerosols.