Adaptive Network Research Articles

A Novel Faster Fixed-Time Adaptive Control for Robotic Systems With Input Saturation

In this paper, an adaptive anti-saturation fixed-time control method with a faster convergence rate is studied for uncertain robotic systems. Firstly, a new segmental sliding variable is constructed to solve the singularity problem brought by the terminal sliding mode control (TSMC) and achieve a faster convergence rate. Secondly, to approximate and compensate for the model uncertainty and the viscous friction parameter, an adaptive neural network (ANN) is employed. Then, a novel auxiliary system is constructed to mitigate the effects of input saturation. Based on this, a novel non-singular TSMC algorithm integrated with the ANN and the auxiliary system is designed, so that the trajectory tracking errors of the robotic system can converge within a faster fixed time with actuator saturation. Finally, the superiority and practicability of the present method are verified by comparative experiments.

A Transformer-Guided Cross-Modality Adaptive Feature Fusion Framework for Esophageal Gross Tumor Volume Segmentation

Background and ObjectiveAccurate segmentation of esophageal gross tumor volume (GTV) indirectly enhances the efficacy of radiotherapy for patients with esophagus cancer. In this domain, learning-based methods have been employed to fuse cross-modality positron emission tomography (PET) and computed tomography (CT) images, aiming to improve segmentation accuracy. This fusion is essential as it combines functional metabolic information from PET with anatomical information from CT, providing complementary information. While the existing three-dimensional (3D) segmentation method has achieved state-of-the-art (SOTA) performance, it typically relies on pure-convolution architectures, limiting its ability to capture long-range spatial dependencies due to convolution's confinement to a local receptive field. To address this limitation and further enhance esophageal GTV segmentation performance, this work proposes a transformer-guided cross-modality adaptive feature fusion network, referred to as TransAttPSNN, which is based on cross-modality PET/CT scans. MethodsSpecifically, we establish an attention progressive semantically-nested network (AttPSNN) by incorporating the convolutional attention mechanism into the progressive semantically-nested network (PSNN). Subsequently, we devise a plug-and-play transformer-guided cross-modality adaptive feature fusion model, which is inserted between the multi-scale feature counterparts of a two-stream AttPSNN backbone (one for the PET modality flow and another for the CT modality flow), resulting in the proposed TransAttPSNN architecture. ResultsThrough extensive four-fold cross-validation experiments on the clinical PET/CT cohort. The proposed approach acquires a Dice similarity coefficient (DSC) of 0.76 ± 0.13, a Hausdorff distance (HD) of 9.38 ± 8.76 mm, and a Mean surface distance (MSD) of 1.13 ± 0.94 mm, outperforming the SOTA competing methods. The qualitative results show a satisfying consistency with the lesion areas. ConclusionsThe devised transformer-guided cross-modality adaptive feature fusion module integrates the strengths of PET and CT, effectively enhancing the segmentation performance of esophageal GTV. The proposed TransAttPSNN has further advanced the research of esophageal GTV segmentation.

Learning spatio-temporal dynamics on mobility networks for adaptation to open-world events

As a decisive part in the success of Mobility-as-a-Service (MaaS), spatio-temporal dynamics modeling on mobility networks is a challenging task particularly considering scenarios where open-world events drive mobility behavior deviated from the routines. While tremendous progress has been made to model high-level spatio-temporal regularities with deep learning, most, if not all of the existing methods are neither aware of the dynamic interactions among multiple transport modes on mobility networks, nor adaptive to unprecedented volatility brought by potential open-world events. In this paper, we are therefore motivated to improve the canonical spatio-temporal network (ST-Net) from two perspectives: (1) design a heterogeneous mobility information network (HMIN) to explicitly represent intermodality in multimodal mobility; (2) propose a memory-augmented dynamic filter generator (MDFG) to generate sequence-specific parameters in an on-the-fly fashion for various scenarios. The enhanced event-aware spatio-temporal network, namely EAST-Net, is evaluated on several real-world datasets with a wide variety and coverage of open-world events. Both quantitative and qualitative experimental results verify the superiority of our approach compared with the state-of-the-art baselines. What is more, experiments show generalization ability of EAST-Net to perform zero-shot inference over different open-world events that have not been seen.

Health status monitoring of high-speed train brake pads considering noise under variable working conditions

The brake pads of high-speed trains operate under complex and variable conditions, and the collected brake signals are easily affected by noise, making monitoring the health status of brake pads more difficult. A multi-representation adaptation network for online monitoring the health status of high-speed train brake pads, which are affected by noise under variable working conditions, is proposed in this study. First, a parameter-sharing deep residual network is used to extract the friction block features of the source and target domain data. Then, the features are mapped to different low-dimensional feature spaces through the inception adaptation module, and multiple representations are obtained. The network applies conditional maximum mean discrepancy to align representations of the source and target domains, thus learning multiple domain-invariant representations. Hence, the network acquires more knowledge of the friction block status and attenuates the interference of noise signals on the status monitoring. The friction block vibration data were collected from various brake disc speeds, and variable working condition-transfer experiments under the influence of noise were performed on the brake friction and bearing datasets. The results show that the proposed network outperforms other transfer methods, which can better extract and identify the status features of the friction block under the noise interference.

Fully bio-sourced catalyst-free covalent adaptable networks from epoxidized soybean oil and L-tartaric acid

Epoxidized vegetable oil (EVO)-based epoxy systems offer a promising avenue for replacing non-recyclable petroleum-based thermoset elastomers. Looking towards future prospects, they also hold potential to evolve into sustainable covalent adaptable networks (CANs). Typically, transesterification-based CANs require catalysts to achieve crosslinked structures that are reprocessable at relatively high temperatures (T ≥ 150 °C). In this study, epoxidized soybean oil (ESBO) was cross-linked with a eutectic hardener composed of L-tartaric acid (TAR) and ethyl lactate (in an aqueous solution), resulting in the development of a fully bio-based 3D covalent network capable of reprocessing without the need for an exogenous catalyst. In addition, it was demonstrated that the peculiar structure of L-tartaric acid, with –OH groups linked to its backbone, plays a prominent role in the reactivity of the system. These free hydroxyl functions facilitate both the ring-opening of epoxides and transesterification reactions, thus enhancing both curing and covalent exchange kinetics.

Multi-source domain adaptive network based on local kernelized higher-order moment matching for rotating machinery fault diagnosis

Unsupervised domain adaptation has been extensively researched in rotating-machinery cross-domain fault diagnosis. A multi-source domain adaptive network based on local kernelized higher-order moment matching is constructed in this research for rotating-machinery fault diagnosis. Firstly, a multi-branch network is designed to map each source-target pair to a domain-specific shared space and to extract domain-invariant features using domain adversarial thought. Then, a local kernelized higher-order moment matching algorithm is proposed to perform fine-grained matching in shared category subspace. Finally, a feature fusion strategy based on the local domain distribution deviation is applied to synthesize the output features of multiple classifiers to obtain diagnostic results. The experimental validation of two-branch and three-branch networks on two public datasets is carried out and average diagnostic accuracies both exceed 99%. The results demonstrate the effectiveness and superiority of the approach.

Aligning Correlation Information for Domain Adaptation in Action Recognition.

Domain adaptation (DA) approaches address domain shift and enable networks to be applied to different scenarios. Although various image DA approaches have been proposed in recent years, there is limited research toward video DA. This is partly due to the complexity in adapting the different modalities of features in videos, which includes the correlation features extracted as long-range dependencies of pixels across spatiotemporal dimensions. The correlation features are highly associated with action classes and proven their effectiveness in accurate video feature extraction through the supervised action recognition task. Yet correlation features of the same action would differ across domains due to domain shift. Therefore, we propose a novel adversarial correlation adaptation network (ACAN) to align action videos by aligning pixel correlations. ACAN aims to minimize the distribution of correlation information, termed as pixel correlation discrepancy (PCD). Additionally, video DA research is also limited by the lack of cross-domain video datasets with larger domain shifts. We, therefore, introduce a novel HMDB-ARID dataset with a larger domain shift caused by a larger statistical difference between domains. This dataset is built in an effort to leverage current datasets for dark video classification. Empirical results demonstrate the state-of-the-art performance of our proposed ACAN for both existing and the new video DA datasets.

Remaining Useful Life Prediction Under Variable Operating Conditions via Multisource Adversarial Domain Adaptation Networks

Domain adaptation (DA) methods have been extensively applied for intelligent remaining useful life prediction (RUL) tasks, addressing domain shift issues under variable operating conditions. Almost all existing methods focus on exploiting the single-source unsupervised domain adaptation (SUDA) approach which only learns prognostic knowledge from a single domain. Nevertheless, SUDA methods have significant limitations due to the disregard for the valuable information contained in historical data from numerous sources. Compared to conventional SUDA approaches, multisource unsupervised domain adaptation (MUDA) approaches can acquire invariant domain representation from a variety of sources and offer higher generalization in transfer tasks. Therefore, a novel MUDA framework multisource adversarial domain adaptation network (MADAN) is proposed for regression tasks. MADAN can efficiently organize and utilize multiple source domain data through a two-stage adversarial domain alignment, of which adversarial domain distribution alignment and inter-domain regression alignment modules are designed to minimize domain discrepancy and regressor discrepancy between each pair of source and target domains. Experimental results indicate that MADAN achieves accurate cross-domain predictions by using multiple source domains, and a comparison demonstrates that it performs better than the state-of-the-art approaches.

Adaptive RBF Neural Network Tracking Control of Stochastic Nonlinear Systems with Actuators and State Constraints

This paper investigates the adaptive neural network (NN) tracking control problem for stochastic nonlinear systems with multiple actuator constraints and full-state constraints. The issue of system full-state constraints is tackled by a generalized barrier Lyapunov function (GBLF), and the output constraints of the system are considered to be in the form of time-varying functions, which are more in line with the needs of real physical systems. The NN approximation technique is utilized to overcome the influence of the uncertainty term on controller design due to randomness. Based on the backstepping technique, a neural adaptive fixed-time tracking control strategy is designed. Under the designed control strategy, the tracking accuracy of the controlled system can reach the expectation in a fixed time. The multi-actuator constraints are converted into a generalized mathematical model to simplify the controller design process. Using the characteristics of the hyperbolic tangent function, a new function called practical virtual control signal is designed using the virtual control signal as the input. Due to the saturation constraint property of the hyperbolic tangent function, it is theoretically ensured that no state of the system exceeds the constraints through to the new form of the virtual controller. Using the adaptive controller constructed in this paper, the controlled system is semi-global fixed-time stabilized in probability (SGFSP). Finally, the effectiveness of the proposed control strategy is further verified by simulation examples.

Fixed‐time adaptive neural network tracking control for output‐constrained high‐order systems using command filtered strategy

SummaryThis article proposes a fixed‐time adaptive neural command filtered controller for a category of high‐order systems based on adding a power integrator technique. Different from existing research, the presented controller has the following distinguishing advantages: (i) a fixed‐time control framework is extended to the tracking control problem of high‐order systems. (ii) The error compensation mechanism eliminates filter errors that arise from dynamic controllers. (iii) Growth assumptions about unknown functions are relaxed with the help of adaptive neural networks. (iv) More general systems: the developed controller can apply to high‐order systems subject to uncertain dynamics, unknown gain functions and asymmetric constraints. Stability analysis shows that all states are semi‐globally uniformly ultimately bounded, and the convergence rate of tracking error is independent of initial conditions. Finally, simulation results validate the advantages and efficacy of the developed control scheme.

Enhancing Efficiency and Security in MTC Environments: A Novel Strategy for Dynamic Grouping and Streamlined Management

This study presents a new strategy to improve security and efficiency in Machine-Type Communication (MTC) networks, addressing the drawbacks of the existing Adaptive Hierarchical Group-based Mutual Authentication and Key Agreement (AHGMAKA) protocol. The AHGMAKA protocol, crucial for securing communication within groups of devices with limited resources, has been found to cause significant operational delays and inefficiencies. Our proposed solution integrates advanced cryptographic methods, including an optimized Authenticated Message Authentication Code (AMAC) and lightweight encryption, sophisticated optimization algorithms for dynamic grouping, and an efficient, lightweight group management protocol. It also introduces adaptive network management strategies to customize performance according to the needs of MTC networks. The effectiveness of this approach has been validated through empirical analysis, showing considerable improvements in operational performance and energy efficiency. These improvements mark a significant step toward achieving an optimal balance between efficiency and security for MTC networks. However, the research acknowledges ongoing challenges, including the trade-off between security and efficiency and the issue of compatibility with older devices, suggesting these as areas for future study. The paper outlines potential research paths, including using machine learning for better group management, adopting post-quantum cryptographic methods, applying hardware acceleration, and pushing to standardize these technologies. This work significantly advances the field of secure and efficient communication in MTC, a critical component of the growing Internet of Things (IoT) landscape, setting the stage for future breakthroughs.

Fatigue Crack and Residual Life Prediction Based on an Adaptive Dynamic Bayesian Network

Monitoring the health status of aerospace structures during their service lives is a critical endeavor, aimed at precisely evaluating their operational condition through observation data and physical modeling. This study proposes a probabilistic assessment approach utilizing Dynamic Bayesian Networks (DBNs), enhanced by an improved adaptive particle filtering technique. This approach combines physical modeling with various predictive sources, encompassing cognitive uncertainties inherent in stochastic predictions and crack propagation forecasts. By employing crack observation data, it facilitates predictions of crack growth and the residual life of metal structure. To demonstrate the efficacy of this method, the research leverages data from three-point bending and single-edge tension fatigue tests. It gathers data on crack length during the fatigue crack progression, integrating these findings with digital twin theory to forecast the residual fatigue life of the specimens. The outcomes show that the adaptive DBN model can precisely predict fatigue crack propagation in test specimens, offering a potential tool for the online health assessment and life evaluation for aerospace structures.

Fault diagnosis of three-cylinder mud pump based on transfer learning

Mud pumps serve as vital components within the circulating system of oil drilling platforms, primarily facilitating the circulation of drilling fluid. With the rapid advancement of deep learning technology, there has been a growing focus on fault diagnosis techniques for mud pumps based on deep learning methodologies. However, existing deep learning approaches often struggle with fault diagnosis of mud pumps under varying operational conditions, as adjustments to the working conditions are necessary in real-time based on drilling depth. To address this challenge, this study introduces an enhanced transfer learning method for diagnosing faults in mud pumps across different operating conditions. Initially, the collected vibration data undergoes resampling to standardize frequency, followed by the utilization of the short-term autocorrelation method to discern phase information of signal impact. Leveraging this phase information, the signal is segmented into distinct segments with uniform phases, thereby minimizing distribution discrepancies between the source and target domains. Subsequently, the transformer is employed as a feature extractor for the model. Finally, a deep sub-domain adaptation network is employed to facilitate transfer from the source domain to the target domain. Validation of the proposed method was conducted using an experimental dataset, with results demonstrating its efficacy compared to other contemporary approaches.

Feature disentanglement based domain adaptation network for cross-scene coastal wetland hyperspectral image classification

At present, domain adaptation (DA) methods have made noteworthy advancements in cross-scene hyperspectral image (HSI) classification. Their success largely hinges on the alignment of distributions between source and target domains, which is a critical step in extracting domain-invariant features. However, this intense focus on domain-invariant feature extraction frequently leads to the neglect of class-discriminative features, limiting their utility in cross-scene coastal wetland classification, where the nuanced identification of different classes is crucial. In this paper, a feature disentanglement based domain adaptation network (FDDAN) is proposed to disentangle and exclude domain-specific features and class-invariant features, thereby obtaining class-specific domain-invariant features for classification tasks. Specifically, a transformer and convolution fusion-based feature extraction network is designed to capture global–local mixed features. To align domain distributions and learn shared features, two corresponding disentanglers separate domain-invariant features and domain-specific features from mixed features. Furthermore, to allow domain-invariant features containing purer category discriminative information, class-invariant features are also segregated. In addition, an adversarial learning strategy between three features is utilized to simultaneously enhance the transferability and discriminability of domain-invariant features. The effectiveness of FDDAN is demonstrated by the experimental results obtained from three cross-scene unmanned aerial vehicle datasets collected in coastal wetlands and a publicly available dataset. We will make our datasets and source code publicly available upon acceptance of the paper.

Investigation of mechanical properties and self‐healing efficiency of carbon fibers composites infused with unsaturated polyester vitrimer

AbstractUnsaturated polyester resins (UPR) are widely in composite manufacturing, particularly in the energy, wind, and construction sectors. When these composites sustain damage, defects or cracks, several methodologies are employed to heal and repair them to extend their lifespan while reducing environmental impacts. In this work, the use of vitrimeric UPR (UPRv) as a matrix for carbon fiber composite, was investigated in terms of healing efficiency by mechanical and thermal properties. Flexural and interlaminar shear strength tests revealed healing efficiencies exceeding 40% and 90%, respectively. These results were attributed to the transesterification exchange occurring within the crosslinked networks of UPRv. As described in our previous work, the incorporation of “Covalent Adaptable Networks (CANs)” enables the conversion of commercial thermoset (UPR) into vitrimeric resins, thereby providing reprocessing and healing capabilities. Furthermore, a comparison between pristine and healed composites with Mode I DCB and Mode II ENF tests, confirm how the introduction of vitrimer in composite facilitates the manufacturing of a self‐healing product, with a significant recovery of the original properties.Highlights Application of a previous synthetized vitrimeric unsaturated polyester resin (UPRv)—fabrication of CF's composites reinforced with UPRv—investigation of the self‐healing property of the developed laminates—potential application as green and sustainable composites.

Event‐triggered adaptive neural‐network control of nonlinear MIMO systems

SummaryThis article investigates an adaptive neural networks (NNs) tracking control design issue for nonlinear multi‐input and multi‐output (MIMO) systems involving the sensor‐to‐controller event‐triggered mechanism (ETM). In the design, NNs are utilized to approximate the unknown nonlinear functions. A sensor‐to‐controller ETM is designed to save unnecessary transmission and communication resources. Subsequently, a first‐order filter technique is presented to solve the problem that the virtual control function is not differentiable. Furthermore, an event‐triggered adaptive NNs control strategy is presented by constructing Lyapunov functions and using adaptive backstepping recursive design. It is demonstrated that the presented scheme can ensure the whole closed‐loop signals are uniformly ultimately bounded without exhibiting the Zeno behavior. Finally, a numerical simulation example confirms the effectiveness of the presented adaptive event‐triggered control (ETC) approach.

Single particle mass spectral signatures from on-road and non-road vehicle exhaust particles and their application in refined source apportionment using deep learning

With advances in vehicle emission control technology, updating source profiles to meet the current requirements of source apportionment has become increasingly crucial. In this study, on-road and non-road vehicle particles were collected, and then the chemical compositions of individual particles were analyzed using single particle aerosol mass spectrometry. The data were grouped using an adaptive resonance theory neural network to identify signatures and establish a mass spectral database of mobile sources. In addition, a deep learning-based model (DeepAerosolClassifier) for classifying aerosol particles was established. The objective of this model was to accomplish source apportionment. During the training process, the model achieved an accuracy of 98.49 % for the validation set and an accuracy of 93.36 % for the testing set. Regarding the model interpretation, ideal spectra were generated using the model, verifying its accurate recognition of the characteristic patterns in the mass spectra. In a practical application, the model performed hourly source apportionment at three specific field monitoring sites. The effectiveness of the model in field measurement was validated by combining traffic flow and spatial information with the model results. Compared with other machine learning methods, our model achieved highly automated source apportionment while eliminating the need for feature selection, and it enables end-to-end operation. Thus, in the future, it can be applied in refined and online source apportionment of particulate matter.

Prediction of monkeypox infection from clinical symptoms with adaptive artificial bee colony-based artificial neural network

AbstractIn 2022, the World Health Organization declared an outbreak of monkeypox, a viral zoonotic disease. With time, the number of infections with this disease began to increase in most countries. A human can contract monkeypox by direct contact with an infected human, or even by contact with animals. In this paper, a diagnostic model for early detection of monkeypox infection based on artificial intelligence methods is proposed. The proposed method is based on training the artificial neural network (ANN) with the adaptive artificial bee colony algorithm for the classification problem. In the study, the ABC algorithm was preferred instead of classical training algorithms for ANN because of its effectiveness in numerical optimization problem solutions. The ABC algorithm consists of food and limit parameters and three procedures: employed, onlooker and scout bee. In the algorithm standard, artificial onlooker bees are produced as much as the number of artificially employed bees and an equal number of limit values are assigned for all food sources. In the advanced adaptive design, different numbers of artificial onlooker bees are used in each cycle, and the limit numbers are updated. For effective exploitation, onlooker bees tend toward more successful solutions than the average fitness value of the solutions, and limit numbers are updated according to the fitness values of the solutions for efficient exploration. The performance of the proposed method was investigated on CEC 2019 test suites as examples of numerical optimization problems. Then, the system was trained and tested on a dataset representing the clinical symptoms of monkeypox infection. The dataset consists of 240 suspected cases, 120 of which are infected and 120 typical cases. The proposed model's results were compared with those of ten other machine learning models trained on the same dataset. The deep learning model achieved the best result with an accuracy of 75%. It was followed by the random forest model with an accuracy of 71.1%, while the proposed model came third with an accuracy of 71%.

Position adaptive residual block and knowledge complement strategy for point cloud analysis

Due to the sparsity, irregularity and disorder of the point cloud, the tasks related to it are full of challenges. Exploring local geometric patterns and multi-scale features is effective for point cloud understanding, and promising results have been achieved. In this paper, we present a Position Adaptive Residual Block, namely PARB, for the first time. It can carry out powerful geometric signal description and feature learning. Starting from this module, we propose two extensions. First, a Position Adaptive Residual Network, called PARNet, is derived by utilizing PARB. Second, PARB can be regarded as a plug-and-play module embedded in MLP-based networks, which can remarkably enhance the performance of the backbone. We also introduce an efficient Knowledge Complement Strategy, which is part of the PARNet architecture, to make the framework perform better. Extensive experimental results on challenging benchmarks demonstrate that our PARNet delivers the new state-of-the-art on ShapeNet-Part and achieves competitive performance on ModelNet40.

Network Anomaly Traffic Analysis

This paper rigorously analyzes two principal methodologies in network traffic anomaly detection: feature detection and anomaly detection. Each methodology exhibits distinct strengths and confronts specific challenges. The study elucidates how the integration of deep learning with artificial immune systems could potentially transform feature detection. Moreover, it illustrates the enhancement of anomaly detection through the synthesis of machine learning techniques with traditional methods. Looking ahead, the paper delineates research trajectories that concentrate on merging deep learning, artificial intelligence, and behavioral analysis. This integration aims to augment the precision, efficiency, and adaptability of network anomaly traffic monitoring systems. Proposed future strategies include advanced methods in data preprocessing, model development, pattern recognition, and adaptive adjustments. These strategies are directed towards fortifying network defenses in response to the dynamically changing spectrum of cyber threats.