自主智能系统(英文)最新文献

Service Composition and Optimization Selection (SCOS) is crucial in Cloud Manufacturing (CMfg), but the uncertainties in service states and working environments pose challenges for existing QoS-based methods. Recently, digital twins have gained prominence in CMfg due to their predictive capabilities, enhancing the reliability of service composition. Heuristic algorithms are widely used in this field for their flexibility and compatibility with uncertain environments. This paper proposes the Whale-Goshawk Optimization Algorithm (WGO), which combines the Whale Optimization Algorithm (WOA) and Northern Goshawk Optimization Algorithm (NGO). A novel similar integer coding method, incorporating spatial feature information, addresses the limitations of traditional integer coding, while a whale-optimized prey generation strategy improves NGO’s global optimization efficiency. Additionally, a local search method based on similar integer coding enhances WGO’s local search ability. Experimental results demonstrate the effectiveness of the proposed approach.

The explainability of recommendation systems refers to the ability to explain the logic that guides the system’s decision to endorse or exclude an item. In industrial-grade recommendation systems, the high complexity of features, the presence of embedding layers, the existence of adversarial samples and the requirements for explanation accuracy and efficiency pose significant challenges to current explanation methods. This paper proposes a novel framework AdvLIME (Adversarial Local Interpretable Model-agnostic Explanation) that leverages Generative Adversarial Networks (GANs) with Embedding Constraints to enhance explainability. This method utilizes adversarial samples as references to explain recommendation decisions, generating these samples in accordance with realistic distributions and ensuring they meet the structural constraints of the embedding module. AdvLIME requires no modifications to the existing model architecture and needs only a single training session for global explanation, making it ideal for industrial applications. This work contributes two significant advancements. First, it develops a model-independent global explanation method via adversarial generation. Second, it introduces a model discrimination method to guarantee that the generated samples adhere to the embedding constraints. We evaluate the AdvLIME framework on the Behavior Sequence Transformer (BST) model using the MovieLens 20 M dataset. The experimental results show that AdvLIME outperforms traditional methods such as LIME and DLIME, reducing the approximation error of real samples by 50% and demonstrating improved stability and accuracy.

This paper investigates resilient consensus control for teleoperation systems under denial-of-service (DoS) attacks. We design resilient controllers with auxiliary systems based on sampled positions of both master and slave robots, enhancing robustness during DoS attacks. Additionally, we establish stability conditions on DoS attack duration and frequency by applying multivariate small-gain methods to ensure closed-loop stability without the need to solve linear matrix inequalities. Finally, the effectiveness of the controllers is validated through the simulation results, demonstrating that the master-slave synchronization is achieved.

Road damage detection is an important aspect of road maintenance. Traditional manual inspections are laborious and imprecise. With the rise of deep learning technology, pavement detection methods employing deep neural networks give an efficient and accurate solution. However, due to background diversity, limited resolution, and fracture similarity, it is tough to detect road cracks with high accuracy. In this study, we offer a unique, efficient and accurate road crack damage detection, namely YOLOv8-ES. We present a novel dynamic convolutional layer(EDCM) that successfully increases the feature extraction capabilities for small fractures. At the same time, we also present a new attention mechanism (SGAM). It can effectively retain crucial information and increase the network feature extraction capacity. The Wise-IoU technique contains a dynamic, non-monotonic focusing mechanism designed to return to the goal-bounding box more precisely, especially for low-quality samples. We validate our method on both RDD2022 and VOC2007 datasets. The experimental results suggest that YOLOv8-ES performs well. This unique approach provides great support for the development of intelligent road maintenance systems and is projected to achieve further advances in future applications.

Electromagnetic jamming is a critical countermeasure in defense interception scenarios. This paper addresses the complex electromagnetic game involving multiple active jammers and radar systems by proposing a multi-agent reinforcement learning-based cooperative jamming decision-making method (MA-CJD). The proposed approach achieves high-quality and efficient target allocation, jamming mode selection, and power control. Mathematical models for radar systems and active jamming are developed to represent a multi-jammer and multi-radar electromagnetic confrontation scenario. The cooperative jamming decision-making process is then modeled as a Markov game, where the QMix multi-agent reinforcement learning algorithm is innovatively applied to handle inter-jammer cooperation. To tackle the challenges of a parameterized action space, the MP-DQN network structure is adopted, forming the basis of the MA-CJD algorithm. Simulation experiments validate the effectiveness of the proposed MA-CJD algorithm. Results show that MA-CJD significantly reduces the time defense units are detected while minimizing jamming resource consumption. Compared with existing algorithms, MA-CJD achieves better solutions, demonstrating its superiority in cooperative jamming scenarios.

Bearings are critical components in machinery, and accurately predicting their remaining useful life (RUL) is essential for effective predictive maintenance. Traditional RUL prediction methods often rely on manual feature extraction and expert knowledge, which face specific challenges such as handling non-stationary data and avoiding overfitting due to the inclusion of numerous irrelevant features. This paper presents an approach that leverages Continuous Wavelet Transform (CWT) for feature extraction, a Channel-Temporal Mixed MLP (CT-MLP) layer for capturing intricate dependencies, and a dynamic attention mechanism to adjust its focus based on the temporal importance of features within the time series. The dynamic attention mechanism integrates multi-head attention with innovative enhancements, making it particularly effective for datasets exhibiting non-stationary behaviour. An experimental study using the XJTU-SY rolling bearings dataset and the PRONOSTIA bearing dataset revealed that the proposed deep learning algorithm significantly outperforms other state-of-the-art algorithms in terms of RMSE and MAE, demonstrating its robustness and accuracy.

Lately, there has been a lot of interest in game-theoretic approaches to the trajectory planning of autonomous vehicles (AVs). But most methods solve the game independently for each AV while lacking coordination mechanisms, and hence result in redundant computation and fail to converge to the same equilibrium, which presents challenges in computational efficiency and safety. Moreover, most studies rely on the strong assumption of knowing the intentions of all other AVs. This paper designs a novel autonomous vehicle trajectory planning approach to resolve the computational efficiency and safety problems in uncoordinated trajectory planning by exploiting vehicle-to-everything (V2X) technology. Firstly, the trajectory planning for connected and autonomous vehicles (CAVs) is formulated as a game with coupled safety constraints. We then define the interaction fairness of the planned trajectories and prove that interaction-fair trajectories correspond to the variational equilibrium (VE) of this game. Subsequently, we propose a semi-decentralized planner for the vehicles to seek VE-based fair trajectories, in which each CAV optimizes its individual trajectory based on neighboring CAVs’ information shared through V2X, and the roadside unit takes the role of updating multipliers for collision avoidance constraints. The approach can significantly improve computational efficiency through parallel computing among CAVs, and enhance the safety of planned trajectories by ensuring equilibrium concordance among CAVs. Finally, we conduct Monte Carlo experiments in multiple situations at an intersection, where the empirical results show the advantages of SVEP, including the fast computation speed, a small communication payload, high scalability, equilibrium concordance, and safety, making it a promising solution for trajectory planning in connected traffic scenarios. To the best of our knowledge, this is the first study to achieve semi-distributed solving of a game with coupled constraints in a CAV trajectory planning problem.

We discover a “less-is-more” effect that adding local antagonistic interactions (negative edge weights) can enhance the overall synchronizability of a dynamical network system. To explain this seemingly counterintuitive phenomenon, a condition is established to identify those edges the weight reduction of which improves the synchronizability index of the underlying network. We further reveal that this condition can be interpreted from the perspective of resistance distance and network community structure. The obtained result is also verified via numerical experiments on a 14-node network and a 118-node network. Our finding brings new thoughts and inspirations to the future directions of optimal network design problems.

This paper presents decentralized solutions for pursuit-evasion problems involving high-order integrators with intracoalition cooperation and intercoalition confrontation. Distinct error variables and hyper-variables are introduced to ensure the control strategies to be independent of the relative velocities, accelerations and higher order information of neighbors. Consequently, our approach only requires agents to exchange position information or to measure the relative positions of the neighbors. The distributed strategies take into consideration the goals of intracoalition cooperation or intercoalition confrontation of the players. Furthermore, after establishing a sufficient and necessary condition for a class of high-order integrators, we present conditions for capture and formation control with exponential convergence for three scenarios: one-pursuer-one-evader, multiple-pursuer-one-evader, and multiple-pursuer-multiple-evader. It is shown that the conditions depend on the structure of the communication graph, the weights in the control law, and the expected formation configuration. Finally, the effectiveness of the proposed algorithm is demonstrated through simulation results.

Machining quality prediction based on cutting big data is the core focus of current developments in intelligent manufacturing. Presently, predictions of machining quality primarily rely on process and signal analyses. Process-based predictions are generally constrained to the development of rudimentary regression models. Signal-based predictions often require large amounts of data, multiple processing steps (such as noise reduction, principal component analysis, modulation, etc.), and have low prediction efficiency. In addition, the accuracy of the model depends on tedious manual parameter tuning. This paper proposes a convolutional neural network quality intelligent prediction model based on automatic feature extraction and adaptive data fusion (CNN-AFEADF). Firstly, by processing signals from multiple directions, time-frequency domain images with rich features can be obtained, which significantly benefit neural network learning. Secondly, the corresponding images in three directions are fused into one image by setting different fusion weight parameters. The optimal fusion weight parameters and window length are determined by the Particle Swarm Optimization algorithm (PSO). This data fusion method reduces training time by 16.74 times. Finally, the proposed method is verified by various experiments. This method can automatically identify sensitive data features through neural network fitting experiments and optimization, thereby eliminating the need for expert experience in determining the significance of data features. Based on this approach, the model achieves an average relative error of 2.95%, reducing the prediction error compared to traditional models. Furthermore, this method enhances the intelligent machining level.