Gaussian Mixture Model Research Articles

Model-based multifacet clustering with high-dimensional omics applications.

High-dimensional omics data often contain intricate and multifaceted information, resulting in the coexistence of multiple plausible sample partitions based on different subsets of selected features. Conventional clustering methods typically yield only one clustering solution, limiting their capacity to fully capture all facets of cluster structures in high-dimensional data. To address this challenge, we propose a model-based multifacet clustering (MFClust) method based on a mixture of Gaussian mixture models, where the former mixture achieves facet assignment for gene features and the latter mixture determines cluster assignment of samples. We demonstrate superior facet and cluster assignment accuracy of MFClust through simulation studies. The proposed method is applied to three transcriptomic applications from postmortem brain and lung disease studies. The result captures multifacet clustering structures associated with critical clinical variables and provides intriguing biological insights for further hypothesis generation and discovery.

Strength, deformation and crack evolution behaviour of cemented tailings backfill composite structures with different strength ratios

ABSTRACT This work comprehensively investigated cemented tailings backfill composite structures (BCS) mechanics and crack evolution mechanisms. A new crack classification method combining Gaussian Mixture Modeling (GMM) and Moving Average Filtering is proposed for the BCS acoustic emission RA-AF dataset. The findings indicate that the upper CTB did not incur significant damage following BCS destruction, enabling BCS to maintain a certain bearing capacity post-failure. The compressive strength of BCS often approximates that of CTB monomers with a lower cement-tailings ratio (this trend diminishes as the upper CTB’s strength increases), and the loss of bearing capacity in BCS is attributed to the failure of CTB monomers with lower strength. The deformation damage in the upper and lower CTB displays a cooperative characteristic throughout the loading-induced damage process, with the axial and transverse strains in the lower CTB surpassing those in the upper CTB. Analysis results from GMM reveal that the compression failure process of BCS is predominantly characterized by tensile failure, with minimal shear failure. The propagation of cracks in the lower CTB exerts a certain influence on the crack evolution of the upper CTB. The findings offer a novel perspective on discerning the fracture evolution behavior of materials.

Multispecies discrimination of New World Warblers (Family Parulidae) using Hidden Markov models (HMMS)

Bioacoustics allows researchers to study animals through their vocalizations with non-invasive methods. The analysis of the recordings is a difficult task that is best handled by machine learning methods. Hidden Markov Models (HMMs), which are machine learning methods in human speech processing, were developed and implemented for the discrimination of 11 genera and 43 species of the New World Warblers (family Parulidae). Based on the CLO-43SD database, the fundamental goal of the experiments was to determine the classification accuracy on the specific genus and species of birds. Through Mel-Frequency Cepstral Coefficients (MFCCs) along with log energy and time derivative features extracted from the vocalizations, HMMs containing 2 states with single underlying Gaussian Mixture Models (GMMs) generated classification accuracies 91.55% across 11 genera of birds and 63.92% for 43 species of birds. From the results, the framework could be applied to analysis of other birds for both classification and detection of vocalizations.

Parsimonious airfoil Parameterisation: A deep learning framework with Bidirectional LSTM and Gaussian Mixture models

The choice of airfoil parameterisation method significantly influences the overall wing optimisation performance by affecting the flexibility and computational efficiency of the process. Ideally, one should be able to intuitively constrain airfoil shape and structural characteristics as input to the optimisation process. Current parameterisation techniques lack the flexibility to generate airfoils efficiently by specifying parsimonious shape and structural features. To address this limitation, a deep learning framework is proposed, enabling conditional airfoil generation from an airfoil’s shape and structural feature definition.Specifically, we demonstrate the application of Bidirectional Long Short Term Memory models and Bayesian Gaussian Mixture models to derive airfoil coordinates from a compact set of shape and structural characteristics that we define. The proposed framework is shown to achieve favorable airfoil performance optimisation due to improved exploration and exploitation of the design space, compared to traditional approaches. Overall, the proposed optimisation framework is able to realise a 9.04% performance improvement over an airfoil design optimised with traditional parameterisation techniques.

Enhancing the Performance of Landslide Susceptibility Mapping with Frequency Ratio and Gaussian Mixture Model

Enhancing the Performance of Landslide Susceptibility Mapping with Frequency Ratio and Gaussian Mixture Model

BPT-PLR: A Balanced Partitioning and Training Framework with Pseudo-Label Relaxed Contrastive Loss for Noisy Label Learning.

While collecting training data, even with the manual verification of experts from crowdsourcing platforms, eliminating incorrect annotations (noisy labels) completely is difficult and expensive. In dealing with datasets that contain noisy labels, over-parameterized deep neural networks (DNNs) tend to overfit, leading to poor generalization and classification performance. As a result, noisy label learning (NLL) has received significant attention in recent years. Existing research shows that although DNNs eventually fit all training data, they first prioritize fitting clean samples, then gradually overfit to noisy samples. Mainstream methods utilize this characteristic to divide training data but face two issues: class imbalance in the segmented data subsets and the optimization conflict between unsupervised contrastive representation learning and supervised learning. To address these issues, we propose a Balanced Partitioning and Training framework with Pseudo-Label Relaxed contrastive loss called BPT-PLR, which includes two crucial processes: a balanced partitioning process with a two-dimensional Gaussian mixture model (BP-GMM) and a semi-supervised oversampling training process with a pseudo-label relaxed contrastive loss (SSO-PLR). The former utilizes both semantic feature information and model prediction results to identify noisy labels, introducing a balancing strategy to maintain class balance in the divided subsets as much as possible. The latter adopts the latest pseudo-label relaxed contrastive loss to replace unsupervised contrastive loss, reducing optimization conflicts between semi-supervised and unsupervised contrastive losses to improve performance. We validate the effectiveness of BPT-PLR on four benchmark datasets in the NLL field: CIFAR-10/100, Animal-10N, and Clothing1M. Extensive experiments comparing with state-of-the-art methods demonstrate that BPT-PLR can achieve optimal or near-optimal performance.

Automated Analysis of Fluency Behaviors in Aphasia.

This study explored the use of an automated language analysis tool, FLUCALC, for measuring fluency in aphasia. The purpose was to determine whether CLAN's FLUCALC command could produce efficient, objective outcome measures for salient aspects of fluency in aphasia. The FLUCALC command was used on CHAT transcripts of Cinderella stories from people with aphasia (PWA; n = 281) and controls (n = 257) in the AphasiaBank database. PWA produced significantly fewer total words, fewer words per minute, more pausing, more repetitions, more revisions, and more phonological fragments than controls, with only one exception: The Wernicke's group was similar to the control group in percentage of filled pauses. Individuals with Broca's aphasia had significantly longer inter-utterance pauses and fewer total words than all other aphasia groups. Both the Broca's and conduction aphasia groups had higher percentages of phrase repetitions than the NABW (NotAphasicByWAB) group. The conduction aphasia group also had a higher percentage of phrase revisions than the NABW and the anomic aphasia groups. Principal components analysis revealed two principal components that accounted for around 60% of the variance and related to quantity of output, rate of speech, and quality of output. The Gaussian mixture models showed that the participants clustered in three groups, which corresponded predominantly to the controls, the nonfluent aphasia group, and the remaining aphasia groups (all classically fluent aphasia types). FLUCALC is an efficient way to measure objective fluency behaviors in language samples in aphasia. Automated analyses of objective fluency behaviors on large samples of adults with and without aphasia can produce measures that can be used by researchers and clinicians to better understand and track salient aspects of fluency in aphasia. https://doi.org/10.23641/asha.25979863.

Active Bidirectional Self-Training Network for Cross-Domain Segmentation in Remote-Sensing Images

Semantic segmentation with cross-domain adaptation in remote-sensing images (RSIs) is crucial and mitigates the expense of manually labeling target data. However, the performance of existing unsupervised domain adaptation (UDA) methods is still significantly impacted by domain bias, leading to a considerable gap compared to supervised trained models. To address this, our work focuses on semi-supervised domain adaptation, selecting a small subset of target annotations through active learning (AL) that maximize information to improve domain adaptation. Overall, we propose a novel active bidirectional self-training network (ABSNet) for cross-domain semantic segmentation in RSIs. ABSNet consists of two sub-stages: a multi-prototype active region selection (MARS) stage and a source-weighted class-balanced self-training (SCBS) stage. The MARS approach captures the diversity in labeled source data by introducing multi-prototype density estimation based on Gaussian mixture models. We then measure inter-domain similarity to select complementary and representative target samples. Through fine-tuning with the selected active samples, we propose an enhanced self-training strategy SCBS, designed for weighted training on source data, aiming to avoid the negative effects of interfering samples. We conduct extensive experiments on the LoveDA and ISPRS datasets to validate the superiority of our method over existing state-of-the-art domain-adaptive semantic segmentation methods.

Compositions of the Hercules–Aquila Cloud and Virgo Overdensity

ABSTRACT Based on a sample of K giant from Large sky Area Multi-Object fibre Spectroscopic Telescope Data Release 8 and a sample of RR Lyrae (RRL) from Gaia Data Release 3, we investigate the compositions of the Hercules–Aquila Cloud (HAC) and Virgo Overdensity (VOD) and their collective contribution to the tilt and triaxiality of the stellar halo ($r\, \lt \, 40\, {\rm kpc}$) as well as two breaks at $\approx 15\, {\rm kpc}$ and 30 kpc. We apply the Gaussian mixture model to divide the stellar halo into the isotropic component and the radially biased anisotropic component, namely Gaia–Sausage–Enceladus (GSE), and find that both HAC and VOD are dominated by the GSE debris stars with weights of $0.67^{+0.09}_{-0.07}$ and $0.57^{+0.07}_{-0.06}$, respectively. In addition, using the K giants with orbital parameters, we identify the member stars of known substructures, including GSE, Sagittarius (Sgr), Helmi Streams, Sequoia, Thamnos, Pontus, Wukong, and Metal-weak Thick Disc, to probe the compositions of low-eccentricity stars in the HAC and VOD regions. In density fittings of the RRL sample, we note that the absence of HAC and VOD has a weak effect on the shape of halo. Finally, we find that the radially biased anisotropic halo contributes majorly to the stellar halo that can be modelled with a tilted triaxial ellipsoid and a doubly broken power law with breaking radii at $18.08^{+2.04}_{-3.22}\, {\rm kpc}$ and $33.03^{+1.30}_{-1.21}\, {\rm kpc}$. This has important significance for understanding the status of large diffuse overdensities in the Milky Way.

ASOD: an adaptive stream outlier detection method using online strategy

In the current era of information technology, blockchain is widely used in various fields, and the monitoring of the security and status of the blockchain system is of great concern. Online anomaly detection for the real-time stream data plays vital role in monitoring strategy to find abnormal events and status of blockchain system. However, as the high requirements of real-time and online scenario, online anomaly detection faces many problems such as limited training data, distribution drift, and limited update frequency. In this paper, we propose an adaptive stream outlier detection method (ASOD) to overcome the limitations. It first designs a K-nearest neighbor Gaussian mixture model (KNN-GMM) and utilizes online learning strategy. So, it is suitable for online scenarios and does not rely on large training data. The K-nearest neighbor optimization limits the influence of new data locally rather than globally, thus improving the stability. Then, ASOD applies the mechanism of dynamic maintenance of Gaussian components and the strategy of dynamic context control to achieve self-adaptation to the distribution drift. And finally, ASOD adopts a dimensionless distance metric based on Mahalanobis distance and proposes an automatic threshold method to accomplish anomaly detection. In addition, the KNN-GMM provides the life cycle and the anomaly index for continuous tracking and analysis, which facilities the cause analysis and further interpretation and traceability. From the experimental results, it can be seen that ASOD achieves near-optimal F1 and recall on the NAB dataset with an improvement of 6% and 20.3% over the average, compared to baselines with sufficient training data. ASOD has the lowest F1 variance among the five best methods, indicating that it is effective and stable for online anomaly detection on stream data.

O-257 Comparing the accuracy of the novel clustering model versus phenotype classification in predicting live birth in polycystic ovary syndrome women undergoing In-Vitro fertilization

Abstract Study question Is the clustering model more accurate than phenotype classification in predicting live birth in polycystic ovary syndrome (PCOS) women undergoing in vitro fertilization (IVF)? Summary answer Novel PCOS clusters, identified by patients’ endocrine and metabolic profiles, more accurately predict live birth rates in women undergoing IVF than the traditional phenotype classification What is known already PCOS is a heterogeneous condition that affects fertility and IVF success. It is classified based on three factors: hyperandrogenism (HA), ovulatory dysfunction (OD), and polycystic ovaries (PCOM). The ESHRE/ASRM classification system identifies four phenotypes: A (including all three factors), B (HA and OD), C (HA and PCOM), and D (OD and PCOM). Although this classification is widely used, its focus solely on clinical signs and symptoms, while neglecting underlying endocrine and metabolic factors, may limit its effectiveness in predicting fertility outcomes in PCOS patients. Therefore, classifying PCOS patients may require a more comprehensive and inclusive assessment approach. Study design, size, duration This prospective cohort study was conducted from June 2020 to August 2022 at a tertiary IVF center in Ho Chi Minh City, Vietnam. Clustering models were built based on the endocrine and metabolic profiles of 731 PCOS patients. Patients were grouped into novel clusters based on their endocrine and metabolic profiles and compared with traditional phenotypes classification. The primary outcome measured was the live birth rate. Participants/materials, setting, methods Patients aged 18–45, diagnosed with PCOS according to the Rotterdam 2003 criteria, and undergoing IVF treatment were included in the study. Patients with hypothyroidism, Cushing’s syndrome, congenital adrenal hyperplasia, or contraindications to hormonal treatment were excluded. We employed Gaussian Mixture Models for clustering, using endocrine and metabolic data as the key features. We then compared the accuracy of the clustering model with phenotype classification in predicting IVF outcomes using a logistic regression model. Main results and the role of chance We identified three distinct clusters among polycystic ovary syndrome (PCOS) patients. Cluster 0 (n = 298) was characterized by significant insulin resistance and dyslipidemia. Cluster 1 (n = 219) displayed the lean PCOS phenotype, and Cluster 2 (n = 214) had moderate insulin resistance with severe hyperandrogenism. Our findings indicated that Clusters 1 and 2 had a higher cumulative live birth rate (CLBR) (39.7% and 37.9% versus 28.5% in Cluster 0, p = 0.015) and lower gestational diabetes rates (8.68% and 6.07%, respectively) compared to Cluster 0 (12.4%, p = 0.048). Additionally, the prevalence of hypertensive disorders during pregnancy showed differences across clusters (3.69% in Cluster 1, 0.46% in Cluster 2, and 2.8% in Cluster 0, p = 0.059). In contrast, the traditional phenotype classification did not demonstrate significant differences in CLBR, gestational diabetes, and hypertensive disorders. The cluster-based method provided a more accurate prediction of CLBR using logistic regression, with an AUC of 0.67 (95% CI: 0.57-0.76) compared to 0.49 (95% CI: 0.41-0.59) for traditional classifications (p = 0.001) Limitations, reasons for caution This is a single-center study conducted on Vietnamese patients. Therefore, the generalizability may be limited. The novel clustering’s predictive value for IVF outcomes in PCOS patients requires further validation in prospective studies. Wider implications of the findings This study suggests a potential shift towards personalized IVF treatment strategies for PCOS, using machine learning-based PCOS clustering identified by patients’ endocrine and metabolic profiles. Trial registration number NCT04364087

Using Learning from Demonstration (LfD) to perform the complete apple harvesting task

Learning from Demonstration (LfD) can be used to make it easier for robots to learn to perform tasks such as apple harvesting. The required apple harvesting motions can be split into four steps: 1. Approaching, 2. Grasping, 3. Detaching, 4. Placing. So far, only parts of the apple harvesting task have been learned using LfD. In our work, we learned an abstracted version of the complete apple harvesting task using a cube. We used two methods and tested the sensitivity to model parameters and the number of demonstrations. We selected (1) the Gaussian Mixture Model (GMM) with Gaussian Mixture Regression (GMR) and (2) the Hidden Markov Model (HMM) with Linear Quadratic Regression (LQR). Both methods make use of Task Parameterization (TP), which allows models to combine multiple phases from the apple harvesting task. Combining multiple harvesting phases is important for effective LfD. We analysed the performance of these algorithms on four performance metrics, namely the grasp pose, the detachment motion, the place pose, and the success rate when using the real robot. A high grasp pose accuracy was achieved, specifically a position error of 0.004 and an orientation error of 0.004 for the GMM with GMR, trained with 100 demonstrations and 9 Gaussian components. The detachment motion was performed but with a reduced range. The model that performed the best was the GMM with GMR, trained with 60 demonstrations and 5 Gaussian components. A high place pose accuracy was achieved, specifically a position error of 0.003 and an orientation error of 0.003 for the GMM with GMR, trained with 100 demonstrations and 15 Gaussian components. The task was executed by a real robot successfully, resulting in a success rate of 67% for HMM with LQR, trained with 40 demonstrations, 5 states, and a scaling factor of 1.0. For the number of components, no clear relation was found. The same was true for the number of states in the HMM with LQR. For the scaling factor, a clear relation could be found. For this task, a value in the range of -1.0 to 1.0 should be used. For the number of demonstrations, there was an optimum of 40 demonstrations. Further improvements are needed to deal with challenges such as apple shift and detecting detachment of fruits.

Accelerating analysis of Boltzmann equations using Gaussian mixture models: Application to quantum Bose-Fermi mixtures

The Boltzmann equation is a powerful theoretical tool for modeling the collective dynamics of quantum many-body systems subject to external perturbations. Analysis of the equation gives access to linear response properties including collective modes and transport coefficients, but often proves intractable due to computational costs associated with multidimensional integrals describing collision processes. Here, we present a method to resolve this bottleneck, enabling the study of a broad class of many-body systems that appear in fundamental science contexts and technological applications. Specifically, we demonstrate that a Gaussian mixture model can accurately represent equilibrium distribution functions, thereby allowing efficient evaluation of collision integrals. Inspired by cold atom experiments, we apply this method to investigate the collective behavior of a quantum Bose-Fermi mixture of cold atoms in a cigar-shaped trap, a system that is particularly challenging to analyze. We focus on monopole and quadrupole collective modes above the Bose-Einstein transition temperature, and find a rich phenomenology that spans interference effects between bosonic and fermionic collective modes, dampening of these modes, and the emergence of hydrodynamics in various parameter regimes. These effects are readily verifiable experimentally. Published by the American Physical Society 2024

Optimizing Patient Stratification in Healthcare: A Comparative Analysis of Clustering Algorithms for EHR Data

Advanced data analytics are increasingly being employed in healthcare research to improve patient classification and personalize medicinal therapies. In this paper, we focus on the critical problem of clustering electronic health record (EHR) data to enable appropriate patient categorization. In the era of personalized medicine, optimizing patient classification is critical to healthcare analytics. This research presents a comparative assessment of different clustering algorithms for Electronic Health Record (EHR) data, with the goal of improving the efficacy and productivity of patient clustering methods. Our study focuses on Fuzzy Technique for Order of Preference by Similarity to Ideal Solution (Fuzzy TOPSIS) as a Multi-Criteria Decision-Making (MCDM) strategy, includes an in-depth assessment of eight clustering algorithms: K-Means, DBSCAN, Hierarchical Clustering, Mean Shift, Affinity Propagation, Spectral Clustering, Gaussian Mixture Models (GMM), as well as Self-Organizing Maps. The evaluation factors used for evaluation in this research are Cluster Quality Metrics, Scalability, Robustness to Noise, Cluster Shape and Density, Interpretability, Cluster Number, Dimensionality, and Consistency and Stability. These criteria and alternatives were chosen after conducting a thorough assessment of the literature and consulting with domain experts. All participated specialists actively engaged in the decision-making process, bringing unique insights into the best clustering algorithms for healthcare data. The results of this study illustrate each algorithm’s strengths and weaknesses in the setting of patient stratification, providing insight into their performance across multiple dimensions. The fuzzy TOPSIS MCDM strategy is a reliable instrument for synthesizing expert opinions and methodically evaluating the found clustering alternatives. This study advances healthcare analytics by giving practitioners and researchers with informative perspectives on the selection of clustering algorithms designed to address the unique problems of patient stratification utilizing EHR data.

A mapped dataset of surface ocean acidification indicators in large marine ecosystems of the United States

Mapped monthly data products of surface ocean acidification indicators from 1998 to 2022 on a 0.25° by 0.25° spatial grid have been developed for eleven U.S. large marine ecosystems (LMEs). The data products were constructed using observations from the Surface Ocean CO2 Atlas, co-located surface ocean properties, and two types of machine learning algorithms: Gaussian mixture models to organize LMEs into clusters of similar environmental variability and random forest regressions (RFRs) that were trained and applied within each cluster to spatiotemporally interpolate the observational data. The data products, called RFR-LMEs, have been averaged into regional timeseries to summarize the status of ocean acidification in U.S. coastal waters, showing a domain-wide carbon dioxide partial pressure increase of 1.4 ± 0.4 μatm yr−1 and pH decrease of 0.0014 ± 0.0004 yr−1. RFR-LMEs have been evaluated via comparisons to discrete shipboard data, fixed timeseries, and other mapped surface ocean carbon chemistry data products. Regionally averaged timeseries of RFR-LME indicators are provided online through the NOAA National Marine Ecosystem Status web portal.

Detection of the irrotational boundary using machine learning methods

Four machine learning methods, i.e., self-organizing map (SOM), Gaussian mixture model (GMM), eXtreme gradient boosting (XGBoost), and contrastive learning (CL), are used to detect the irrotational boundary (IB), which represents the outer edge of the turbulent and non-turbulent interface layer. To accurately evaluate the detection methods, high-resolution databases from direct numerical simulations of a temporally evolving turbulent plane jet are used. It is found that except for the SOM method, the general contour of the IB appears to be effectively captured using the GMM, XGBoost, and CL methods, which indicate the turbulent and non-turbulent regions can be roughly recognized. Furthermore, the intrinsic features of the detected IB using the GMM, XGBoost, and the CL methods are quantitatively evaluated. Unlike the conventional vorticity norm method, the three machine learning methods do not rely on a single threshold of vorticity magnitude to separate the turbulent and non-turbulent regions. A small part of the detected IB using the three machine learning methods is characterized by the rotational motions, which are expected to be only found inside the turbulent sublayer and turbulent core region. Compared to the vorticity norm and XGBoost methods, the fractal dimensions of the IB detected by the GMM and CL methods are relatively small, which are related to the missing detection of some highly contorted elements. With the three machine learning methods, a large part of the detected IB is characterized by a convex shape, similarly as with the vorticity norm. However, the probability density function profiles of the local curvature of the detected IB differ greatly between the three machine learning methods and the vorticity norm. A mild variation of the mean conditional distributions of the vorticity magnitude can be observed across the detected IB by the three machine learning methods. This study first implies that using the machine learning methods the turbulent and non-turbulent regions can be roughly distinguished, but it is still challenging to obtain the intrinsic features of the detected IB.

Frictional process, AE characteristics, and permeability evolution of rough fractures on Longmaxi shale with specific roughness under water injection

Studying the frictional process, acoustic emission (AE) characteristics and permeability evolution of fractures under water injection is fundamental for proposing theories and methods to control the stability of geological structures. In this work, the shear-flow test was conducted on Longmaxi shale fractures with different specific roughness to analyze the frictional deformation and mechanical properties of shale fractures during the frictional slip process, and the influence of effective normal stress and surface morphology was investigated. AE monitoring technology was also utilized to reveal the AE evolution during the frictional slip process. Based on the AF/RA values, AE signals were subjected to cluster analysis using the Gaussian Mixture Model (GMM) to distinguish and quantify the crack types and proportions during frictional slip. Combining the development of cracks and considering the ratio of worn area, a theoretical analytical expression was established to reasonably describe the friction-damage evolution, classifying the friction-damage during the frictional slip process into tensile and shear damage, respectively. Meanwhile, the permeability evolution of Longmaxi shale fractures was also studied in detail. A theoretical model capable of describing and predicting the permeability evolution was proposed under the condition that the permeability evolution and damage process of Longmaxi shale fractures correspond to each other. Finally, the interaction between asperities on shale fractures with different surface morphologies during the frictional slip process was discussed, and the influence mechanism of surface morphology on permeability evolution was revealed.

A UAV communication strategy based on user distribution time series prediction

Abstract To maximize the role of unmanned aerial vehicles (UAVs) as mobile relays to assist communication, while meeting the requirements of service quality and energy constraints, it is necessary to fully consider the motion trajectories and spatiotemporal distribution of ground users in UAV wireless communication systems. Therefore, an algorithm that combines Long Short-Term Memory (LSTM) and Gaussian Mixture Model (GMM) is proposed to predict user distribution. The LSTM, with its capability of handling temporal data, enables the UAV to refer to the preceding information during flight decision-making and make action outputs based on predictions of the dynamic environment. Simulation results demonstrate that the proposed approach can accurately predict the position information of mobile users, achieve faster convergence, and obtain better performance in dynamic scenarios. The system performance has improved by 5.8 percent.

Mechanical behaviors and progressive fracture processes of dry basalt after Martian cryogenic freeze-thaw cycles

The steep Martian terrain features numerous notches and alcoves that could potentially be adapted as temporary shelters. By sealing their entrances and modifying the interior space, these alcoves can be transformed into habitable structures. Given their prolonged exposure to Martian temperature fluctuations, it is crucial to understand the mechanical properties of alcove rocks to design appropriate shelter parameters. This study investigates the mechanical behavior of basalt specimens subjected to Martian cryogenic freeze-thaw (F-T) cycles (−143∼35 °C) and subsequent uniaxial cyclic loading tests. The test results indicate that the longitudinal wave velocity, peak stress, and elastic modulus of basalt specimens degrade to varying degrees with an increasing number of F-T cycles. At higher stress levels, the ratio of dissipated energy density to elastic energy density in basalt specimens subjected to different F-T treatments converges to approximately 0.25. During the uniaxial cyclic loading tests, both cumulative acoustic emission (AE) counts and average maximum principal strains of the specimens increase stepwise. For specimens after F-T treatment, the spectral-domain characteristics can be classified into three frequency bands, with the middle and high frequencies predominating. The application of the Gaussian Mixture Model (GMM) clustering algorithm enhances the analysis of AE results, facilitating the identification of tensile and shear cracks. As the number of F-T cycles increases, the percentage of shear cracks in basalt specimens decreases from 76.5% (0 cycle) to 53.5% (7 cycles).

Guided-wave Dynamic Probability Model-based Crack Quantification Method for Metal Structures

In the process of applying guided-wave-based aircraft online structural health monitoring, the characteristics of guided wave signals used to represent structural status exhibit random uncertainty due to the interference of loads on guided wave propagation. The trend of signal feature changes caused by damage is often obscured, making the quantification of cracks under loads one of the challenges in online structural health monitoring for aircraft. To address this challenge, this paper proposes a crack quantification method using an adaptive Gaussian mixture model (GMM) and optimized probability migration distance measurement. This method effectively mitigates the uncertainty caused by loads by constructing an adaptive Gaussian mixture model to track the probability distribution changes of guided wave signal features under different structural health conditions. Subsequently, an optimized probability migration distance measurement method is employed to characterize the structural state, further enhancing the correlation between features and the extent of structural damage. Finally, an optimized probability migration distance relationship with crack length is established to estimate the crack length of new structural specimens. The effectiveness of this method is validated using load-induced crack propagation experiments on metal structures.