To satisfy the growing demand for traffic prediction induced by urbanization, the intelligent transportation system integrated various cutting-edge artificial intelligence technologies, with large language models (LLMs) as a representative, has been developed. However, existing methods are mostly confined by shallow LLMs utilization, where the semantic capacity of LLMs is ignored and the traffic data are directly fed in. Furthermore, the modality diversity of different traffic prediction scenarios (e.g., flow, speed, and demanding) remains to be underexplored, which restricts the model flexibility towards downstream applications. To mitigate these limitations, we propose a Mixture of Semantic and Spatial Experts (SS-MoE) for traffic prediction along with the human-intelligible post-hoc result explanation. Specifically, to enlighten the traffic predictor with abundant semantic information, we design hierarchically coarse- and fine-grained prompts including role assignments, dataset descriptions, and background supplements, which serves as the auxiliary knowledge for downstream prediction. Afterwards, considering the diversity of real-world traffic scenarios, we construct the MoE framework consisting of a spatial expert, a semantic expert, and a general expert, which accounts for the node-level features, the semantic representations, and the overall generalization, respectively. At last, we instruct the LLM to explain and analyze the final prediction, which is able to provide insightful conclusions and support intelligent transportation decisions, forming a unified prediction-explanation pipeline. Extensive experiments on five public traffic datasets demonstrate the superiority of SS-MoE across three traffic prediction tasks. Experimental results indicate that the MAE and RMSE values of SS-MoE are reduced by up to 4.04% and 3.20% compared with that of the runner-up, respectively.

Accurate traffic flow prediction is a key aspect of building data-driven intelligent transportation systems (ITSs) which relies on the Internet of Things (IoT) sensors deployed along roads, and dynamic spatial-temporal dependencies mining is a major area of interest in traffic flow prediction. Existing methods, however, overlook the diversities of traffic flow patterns from the perspectives of temporal and spatial dimensions. To this end, this article presents a multiview spatial-temporal adaptive transformer-GRU (MST-ATG) framework based on the encoder-decoder architecture to capture complex spatial-temporal dependencies from various perspectives. Specifically, a multiview embedding layer (MEL) containing original traffic data and spatial-temporal correlated features is designed to enrich the feature encoding. Then, based on the inherent characteristics of traffic flow, we introduce a periodicity-trend decomposition (PTD) method to fully consider the periodic- and trend-oriented features of time series. Finally, we propose a spatial-temporal adaptive transformer-GRU (ST-ATG) to dynamically extract spatial-temporal dependencies and adaptively choose computation steps in which a temporal adaptive stacked-GRU module (T-AGM) is proposed to extract correlations in temporal dimension and spatial dependencies captured by a spatial adaptive transformer module (S-ATM). Experimental results on six large-scale real-world datasets demonstrate that our MST-ATG framework outperforms the benchmarks in prediction accuracy. For instance, the average root-mean-square error of MST-ATG on PeMS08 is reduced by 48.3%, 41.09%, 12.95%, 17.67%, 18.64%, 2.4%, 14.67%, 9.15%, 1.1%, 2.4%, 2.51%, and 1.2% compared to that of autoregressive integrated moving average, long short-term memory (LSTM), DCRNN, STGCN, ASTGCN, GWNet, STSGCN, AGCRN, Bi-STAT, STAEformer, PDFormer, and STPGNN, respectively.

The widespread application of AI-generated content (AIGC) services has driven demand for efficient computational resources, making effective task scheduling and computation offloading in edge computing (EC) environments a critical research topic. However, the high computational requirements and low latency demands of AIGC services, combined with the limitations of EC, present challenges for existing offloading methods, such as unstable decision making in dynamic task environments and resource overloading. Here, we propose a decentralized AIGC task offloading architecture within an IoT-assisted EC network to optimize the quality of AIGC services. In this architecture, we define a multiobjective joint optimization problem for AIGC task offloading, aiming to simultaneously optimize key performance metrics, such as task latency, energy efficiency, and load balancing. To address this problem, we introduce an improved proximal policy optimization (PPO)-based deep reinforcement learning (DRL) algorithm, named TOPPO. By incorporating a policy update step size constraint and a clipping mechanism, TOPPO significantly enhances the stability of the training process and reduces fluctuations during policy updates. Additionally, the algorithm integrates an LSTM model to improve its ability to handle temporal dependencies. Through continuous interaction between the model and the environment, the offloading strategy is iteratively updated to ensure that diverse AIGC tasks are efficiently executed on IoT devices or edge servers. Extensive simulations and performance evaluations demonstrate that the proposed method achieves significant improvements in task latency, energy consumption, and load management during AIGC task processing.

With the rapid development of edge computing, devices now offer powerful computing capabilities and diverse applications. However, the surge in smart devices accessing the Internet overwhelms edge servers, which have limited and unevenly distributed resources. This results in challenges like energy management, load balancing (LB), real-time performance, and system complexity. Existing research fails to comprehensively consider these challenges’ combined impact, making it difficult to maximize performance when facing real complex scenarios. To address the above issues, this article proposes an edge cloud resource scheduling scheme based on evolutionary-weighted clustering and transformer-augmented reinforcement learning (EC-TRL). First, server nodes are deployed at the center of user clusters, based on user device locations, to optimize communication delay and evenly distribute resources. Second, the multiobjective scheduling optimization problem under delay constraints is converted into a Markov decision problem, and a deep reinforcement learning method based on soft actor-critic (SAC) is proposed. Finally, actor transformer (AT) and critic transformer (CT) are proposed to improve the network structure of SAC, capture long-term dependencies and complex patterns in long task scheduling sequences, and improve the model’s adaptability and generalization performance in complex dynamic environments. Through comparison experiments with round robin, random, proximal policy optimization, dueling double deep Q-learning network, SAC-L, and SAC-M, the results show that the proposed method improves the optimization performance of energy consumption, LB, and rejection rate of edge cloud resource scheduling by at least 9.57%, 10.90%, and 5.05%.

Anomaly detection for flight data is crucial in maintaining the safety and stability of unmanned aerial vehicles (UAVs), making it a topic of significant research and attention. However, existing anomaly detection methods often ignore the random noise of UAV flight data and lack effective parameter selection, resulting in inadequate anomaly detection performance. Furthermore, current methods generally face the problem of insufficient feature extraction capability. In this paper, a spatio-temporal correlation based on one-dimensional convolutional neural network (1D CNN), bidirectional long short-term memory (BiLSTM), and attention mechanism (AM) hybrid neural network with residual filtering (STC-1D CBiAM-RF) data-driven multiple regression framework is proposed for anomaly detection and recovery of UAV flight data. First, a correlation analysis method is used for parameter selection to reduce the dependence on expert knowledge. Second, a multiple regression model fusing attention mechanism is designed. It utilizes 1D CNN-BiLSTM as a feature extractor, guided by the attention mechanism, to enhance the learning of crucial information from UAV flight data. Then, to effectively mitigate the impact of random noise, a residual filtering method is introduced to smooth the residuals, thereby improving anomaly detection performance. Finally, anomaly detection is achieved by comparing the square of the smoothed residuals with the statistical threshold, and data recovery is achieved by replacing the anomalous data with the predicted data. The effectiveness of the proposed method is verified through a series of experiments using real UAV flight data injected with different anomaly types.

Meta-heuristic algorithms play an essential role in solving real-world optimization problems. However, their performance is limited by the complexity and variability of the problems. Hence, various efficient algorithms are being actively explored. The exponential distribution optimizer (EDO), having attracted attention for its efficient search performance, has been extended to several applications. However, it suffers from falling into local optima and weak exploitation. Meanwhile, it cannot be directly applied to solve binary optimization problems. To address these challenges, this paper proposes an enhanced EDO called BOMLDEDO. The Bernstein-assisted oppositional-multiple learning strategy is proposed to avoid falling into local optimality. The Bernstein-based adaptive differential strategy is developed to improve exploitation capability. Moreover, by introducing a transfer function, repair method, and binary-to-real operation, BOMLDEDO is extended to a binary version. The IEEE (Institute of Electrical and Electronics Engineers) CEC (Congress on Evolutionary Computation) test functions and engineering problems are used to evaluate BOMLDEDO's optimization performance for continuous problems. Compared to its competitors, BOMLDEDO ranks first on more than 8 out of 10 IEEE CEC 2020 functions and more than 10 out of 12 IEEE CEC 2022 functions. Meanwhile, it achieves the global optimum in 91% of engineering problems. Furthermore, the 0–1 knapsack problems are applied to verify BOMLDEDO's binary optimization capabilities, and the results show that BOMLDEDO is successfully utilized in 14 knapsack instances. The above results demonstrate that incorporating multiple strategies helps improve the performance of BOMLDEDO, making it more reliable and applicable in solving continuous optimization problems and 0–1 knapsack problems.

Resource management challenges frequently manifest in systems and networks as tough online decision tasks, for which the proper solution is dependent on an understanding of the workload and environment and facilitates smooth use of mobile edge and cloud resources. Due to the geographical dispersion of resources, constrained resource capacity, unpredictable nature of tasks, and network hierarchy present in such contexts, it is difficult to efficiently schedule jobs in edge environments. Unfortunately, existing heuristic-based methods lack generality and fast adaptability and thus cannot optimally solve such problems. The advantage actor–critic (A2C) method, on the one hand, can quickly adapt to dynamic circumstances based on relatively few data, and deep reinforcement learning (DRL) agents can on the other hand rapidly learn from their experience of environmental interactions to make better judgments. Therefore, we present an A2C-DRL real-time task scheduling technique for stochastic edge–cloud environments that enables decentralized learning and simultaneous work scheduling across multiple servers. With the aim of producing efficient scheduling decisions, we develop reward values for various resources and model the update policy, server resource scheduling method, and policy learning method. The model is adaptive and includes various hyperparameters that can be adjusted in accordance with the application requirements. We evaluate the load balancing capability of the model by introducing a load balancing factor. Experiments on real datasets show that the proposed A2C-DRL method outperforms seven state-of-the-art algorithms in terms of the reward value, task rejection, and the load balancing factor.

Semantic communication is an emerging paradigm for digital twin (DT) synchronization in unmanned aerial vehicle (UAV)-assisted edge computing environments, where machine learning (ML) models are deployed on edge servers and UAVs as semantic encoders and decoders to perform real-time synchronization. However, with limited system resources, additional computation workloads are still brought to all participants for semantic information extraction and recovery. In this work, we propose an optimized tiny ML-based DT synchronization framework to minimize the synchronization latency in UAV-assisted edge computing environments, considering time-average constraints on virtual energy deficit queue stability. Due to the coexistence of tiny ML-based semantic communications, a semantic extraction factor is introduced to formulate the DT synchronization problem as a time-average time minimization problem. By leveraging the Lyapunov optimization framework, the multi-stage DT synchronization problem is transformed into several per-slot resource allocation problems. To solve the per-slot optimization problem efficiently, a deep reinforcement learning-based synchronization (DRLS) algorithm is proposed, where an actor-critic structure is adopted to generate synchronization actions with low time complexity. Finally, we conduct simulation experiments to evaluate the performance of the proposed DRLS scheme. Numerical results demonstrate that our DRLS algorithm can reduce 8.23% of DT synchronization delay and 15.31% of synchronization data dropping rates on average by comparing it with the UAV-edge collaborative synchronization scheme without semantic communications. Besides, the DRLS algorithm can achieve up to 57.14% synchronization energy reduction compared with representative synchronization policies.

Abstract Using convolutional neural networks (CNNs) in classifying hyperspectral images (HSIs) has achieved quite good results in recent years. It is widely used in agricultural remote sensing, geological exploration, environmental monitoring, and marine remote sensing. Unfortunately, the complexity of network structures used for hyperspectral image classification challenges the efficient delivery of HSI data extremely, and existing methods suffer from a large amount of redundancy in the network weight parameters during training, as they either require huge computational resources or make inefficient use of storage space when designing the network structure, and many of the parameters that waste computational resources contribute less to the rich spectral and spatial information transfer in HSI. So we introduce LCTCS, a better low-memory and less-parametric network approach. LCTCS aims to improve the efficiency of computational resource utilization with advanced classification performance and lower levels of computational resources. Unlike the conventional 2D and 3D convolution used previously, we use simple and efficient 3D grouped convolution as a vehicle to convey the semantic features of HSIs. More specifically, we design a novel two-channel sparse network to classify HSIs since grouped 3D convolution conveys the properties of hyperspectral data well in the time and space domains.We have compared LCTCS with eight widely used network methods on four publicly available hyperspectral datasets for learning HSI information. A series of experiments shows that the model architecture designed has $$65.89 \%$$ 65.89 % less storage space than the DBDA method, consumes $$67.36 \%$$ 67.36 % fewer computational resources than the SSRN method on the IP dataset, and accomplishes a highly accurate classification task with the number of parameters accounting for only $$1.99 \%$$ 1.99 % that of the DBMA method.

Compositional Zero-Shot Learning (CZSL) has been applied to various scenarios, including scene understanding, visual-language representation, and domain adaptation. Despite numerous endeavours and significant advancements, the crucial issues of fuzzy conceptualization of visual attributes and insufficient inter-class connectivity, have remained insufficiently addressed. To address these issues, we propose Learning Visual Attributes Representation for Compositional Zero-Shot Learning (LVAR-CZSL), which has the ability to learn visual attributes and inter-class dependencies. LVAR-CZSL is mainly composed of two key components: the Visual Attribute Representation Module (VARM) and the Connected Learning Module (CLM). Specifically, VARM extracts detailed attributes and object visual features from global visual features, resolving the issue of fuzzy visual attribute concepts. Moreover, CLM endows LVAR-CZSL with the capability to perceive connectivity between different attributes and objects, effectively enhancing inter-class connectivity. To establish a close connection between VARM and CLM and minimize the gap between image and text features, we introduce the composition-attribute-object Joint Scoring Function (JSF). Additionally, we propose Joint Loss Function (JLF) to optimize the learning process of VARM and CLM. The experiment results on four datasets show that LVAR-CZSL achieves state-of-the-art performance. The code is available athttps://github.com/mxjmxj1/LVAR-CZSL.

High‐performance tracking control is essential for permanent magnet synchronous motors in the perturbed environment. Given this, a new hybrid controller is proposed in this study for a permanent magnet synchronous motor with load disturbances as well as time delays. First, a new prescribed performance method is proposed to achieve the full‐state performance constraints with load disturbances. Second, a time‐varying filter is proposed for the first time to avoid the “complexity explosion” problem of the backstepping method while guaranteeing the convergence of the filtering error. Third, by combining Lyapunov–Krasovskii functionals with adaptive neural networks, the time‐delay disturbance and unknown nonlinear dynamics of the control system have been solved. The stability analysis proves that all signals in the closed‐loop system are bounded. To show the effectiveness of the intelligent controller, the comparison simulations are given to confirm the advantages of the proposed adaptive neural control scheme.

The manufacturing processes of unmanned aerial vehicle (UAV) power systems generate large amounts of data and knowledge. The extraction of useful information or patterns from redundant data and knowledge texts has become a challenge in intelligent manufacturing. Unfortunately, graphics processing unit (GPU)‐based parallel computing is limited, and the inference speeds of the available named entity recognition (NER) models for Chinese text datasets are low because they are mainly based on the long short‐term memory (LSTM) algorithm. Herein, first, the flat‐lattice transformer (FLAT) model was optimized by using a stochastic gradient descent with momentum (SGDM) optimizer and adjusting the model hyperparameters. Compared with the existing NER methods, the proposed optimization algorithm achieved better performance on the available dataset. Then, an NER method named the TL_FLAT model based on transfer learning and the abovementioned optimization model was introduced. Finally, a Chinese text dataset from a UAV power system created by the authors was used to validate the proposed method. The F1 score was 76.26%, the precision value was 76.98%, and the recall value was 75.56%, indicating that the TL_FLAT model was suitable for Chinese text entity recognition for UAV power systems.

In order to improve the comprehensive performance of energy dispatching between different sites, the optimization research of particle swarm optimization (PSO) algorithm and ant colony optimization (ACO) algorithm is carried out. We proposed a new improved PSO‐ACO algorithm based on the idea of hybrid algorithm to solve the problem of poor energy dispatching efficiency between sites. First, the multiobjective performance indicators were introduced to transform the sites’ energy dispatching problem into a multiobjective optimization problem. Second, the vitality factor was introduced into the PSO strategy to solve the local optimal problem, and in the PSO‐ACO fusion strategy, the PSO routes were transformed into the ant colony enhancement pheromone to accelerate the accumulation speed of the ACO initial pheromone. Then, the angle guidance function was introduced into the state transition probability of the ACO strategy to improve the global search capability, and a high‐quality pheromone update rule was proposed to improve the convergence speed of the algorithm. Finally, simulation experiments were carried out on the improved PSO‐ACO algorithm, Min–Max Ant System (MMAS) algorithm, ACO algorithm, PSO algorithm, and PSO update algorithm in a variety of complex site scenarios. The simulation results show that the improved PSO‐ACO algorithm can plan a site energy dispatching route with shorter route, less time‐consuming, and higher security and realize the comprehensive and global optimization of energy dispatching.

The development of object detection networks has reached a high point, and there have been significant improvements in accuracy and detection speed. Object detection is widely used in intelligent robots, self‐driving cars, and other edge‐intelligent terminals. Unfortunately, when a detector is allowed to learn new objects in an unfamiliar environment, it can catastrophically forget the objects it has already learned. In particular, reliable and stable knowledge cannot be extracted from old models. Based on this, a new multinetwork mean distillation loss function for open‐world domain incremental object detection is presented. To better extract reliable and stable knowledge from old models, we enhanced the distillation output of the detector with a ResNet50 backbone and an output RoI head. The distillation output of the intermediate RPN is softened by adaptive distillation. To obtain more stable results, the ResNet50 backbone and RPN on the channel are zero‐averaged. Various incremental steps and stability experiments are performed on two benchmark datasets, PASCAL VOC and MS COCO. The experimental results show the excellent performance of our method in different experimental scenarios, and it is superior to the most advanced methods. For example, in the setting of the batch task, incremental object detection on the PASCAL VOC and MS COCO datasets is improved by 3.4% and 2.1%, respectively.

Fixed‐wing unmanned aerial vehicles (FW‐UAVs) play an essential role in many fields, but the faults of FW‐UAV components lead to severe accidents frequently; so, there is a need to continuously explore more intelligent fault detection methods to improve the safety and reliability of FW‐UAVs. Deep learning provides advanced solution ideas for future UAV fault detection, but the current lack of UAV monitoring data limits the advantages of deep learning in UAV fault detection, which are both a challenge and an opportunity. In this paper, we mainly consider the data availability of deep learning under various practical flight conditions of FW‐UAVs and propose a fault detection framework based on hybrid deep domain adaptation BiLSTM networks and the Hampel filter (HDBNH), the main purpose of which is to learn the knowledge of acquired data for detecting FW‐UAV faults in other unknown operating conditions. HDBNH consists of three modules: feature extractor, domain adaptor, and fault detector. The feature extractor is two BiLSTM networks constructed to extract the past and future state features from the time‐series flight data. The discrepancy of feature distribution between different domains is effectively reduced in the domain adaptor by a hybrid adversarial and the maximum mean discrepancy (MMD) domain adaptation method. The fault detector consists of a fault classification module and a Hampel filter. According to the continuous and dynamic characteristics of FW‐UAV state changes, the Hampel filter is used to detect and correct the predicted values of the fault classification module. Meanwhile, a new state sample preparation strategy is proposed to support the work of HDBNH better. Finally, the effectiveness of HDBNH is confirmed by conducting extensive experiments in real FW‐UAV flight data.

This article focuses on dynamic analysis and the fuzzy fixed-time optimal synchronization control problem of unidirectionally coupled fractional-order (FO) permanent magnet synchronous generator (PMSG) system. The synchronization model between FO master and slave PMSGs with capacitive and resistive couplings is built. The dynamic analysis fully reveals its abundant dynamical behaviors including chaotic oscillations and gives stability/instability boundaries with the designed numerical method. In controller design, the hierarchical type-2 fuzzy neural network (HT2FNN) with a transformation is designed to approximate unknown functions, the fixed-time command filter matched up with the compensating signal is proposed to achieve precise estimate and fast convergence, and a fixed-time preconfigured performance function integrated with a smooth and invertible function is built to realize fixed-time convergence and performance constraint. Then a fuzzy fixed-time optimal synchronization control scheme fusing with the HT2FNN, filter, performance function and optimal control is developed under the FO backstepping theory. The stability analysis proves that all signals of the closed-loop system are bounded along with the cost function being minimized. Finally, numerical simulation results verify the feasibility and advantages of our scheme.

Recently, the enhanced frequency band entropy (EFBE) was proposed based on the replacement of short-time Fourier transform with wavelet packet transform. In view of the shortcomings of EFBE, a feasible solution is provided, namely modified frequency band envelope kurtosis (MFBEK), which can be described as follows: First, the modified adaptive resonance bandwidth (MARB) based on the bearing inner-race fault frequency is proposed. The kurtosis of the envelope signal as an available indicator and the MARB are used to determine the optimal depth of MFBEK. Second, this special case, that is, the resonant frequency occurs at the junction of two adjacent subbands is considered, and a corresponding effective solution is provided to determine the optimal subband(s). Then, the reconstructed signal can be obtained. And then, if necessary, a band-pass filter is designed to process the reconstructed signal to enhance the noise reduction performance. Finally, envelope power spectrum analysis is performed on the reconstructed signal or the filtered signal to extract the fault characteristic frequency. In addition, a modified indicator is proposed to measure the analysis results. Analysis results on simulated and vibration signals measured from actual bearing have revealed that the MFBEK can obtain more robust performance.

The singular value decomposition (SVD) based on the Hankel matrix is commonly used in signal processing and fault diagnosis. The noise reduction performance of SVD based on the Hankel matrix is affected by three factors: the reconstruction component(s), the structure of the Hankel matrix, and the point number of the analysis data. In this article, the three influencing factors are systematically studied, and a method based on correlated SVD (C-SVD) is proposed and successfully applied to bearing fault diagnosis. First, perform SVD analysis on the collected original signal. Then, the reconstructed component(s) determination method of SVD based on the combination of singular value ratio (SVR) and correlation coefficient is proposed. Then, based on the SVR, using the envelope kurtosis as the indicator, the optimal structure of the Hankel matrix (number of rows and columns) is studied. Then, the number of data points of the analysis signal is discussed, and the constraint range is given. Finally, the envelope power spectrum analysis is performed on the reconstructed signal to extract the fault features. The proposed C-SVD method is compared with the existing typical methods and applied to the simulated signal and the actual bearing fault signal, and its superiority is verified.