As a pivotal energy industry, production safety is of critical importance to coal mining. However, current monitoring approaches mainly rely on manual experience-based analysis of field data, exhibiting limitations such as high operational costs, low efficiency, and inadequate reliability. In contrast, power load data from coal production processes serves as a comprehensive state indicator containing rich operational information. Thus, this paper proposes a time-frequency multi-scale reconstruction anomaly detection approach (TFMSR-AD) leveraging power load data to identify anomalous states during production. The methodology integrates multi-scale Transformer architecture with a compact pyramidal structure for temporal feature extraction, while incorporating Fourier Transform for periodic pattern analysis. This synergistic framework achieves high-fidelity reconstruction to enhance anomaly detection performance. Experimental results demonstrate TFMSR-AD's effectiveness in identifying latent anomaly signatures within power load data, thereby providing a cost-effective, highly robust intelligent monitoring solution for coal mine safety management.

Azimuth thrusters are widely used for controlling marine vehicles, especially, for dynamic positioning and hovering purposes. However, including azimuth thruster makes control allocation a nonlinear non-convex problem which is commonly solved using nonlinear programming methods, simplified by paring azimuth thrusters (e.g., two thrusters will always move at the same angle), or locally linearized using approximation equations such as Taylor series expansions and polynomial functions. In this paper, a new approach is presented to modify the azimuth thruster control allocation problem into a convex quadratic problem with a new force decomposition and linear first-order inequality constraints. As a result, the complexity of the control allocation increases linearly with respect to the number of azimuth thrusters, allowing it to be implemented on the marine vehicles with increased numbers of azimuth thrusters controlled independently and can be solved using constrained quadratic programming solvers. Case studies has been presented to validate the proposed method on simulated Autonomous Underwater Vehicles (AUVs) with two and four azimuth thrusters configured with different azimuth angle limits (±45, ±90, and ±135 degrees). The results shows excellent control performance of the proposed approach in controlling multiple states (surge, pitch, yaw, depth and sway) simultaneously, even when experiencing a cross-track ocean current. Recommendation on hardware implementation is also discussed for real world platform integration.

High-resolution images come with rich color information and texture details. Due to the rapid upgrading of display devices and rendering technologies, high-resolution real-time rendering faces the computational overhead challenge. To address this, the current mainstream solution is to render at a lower resolution and then upsample to the target resolution by supersampling techniques. However, while many prior supersampling approaches have attempted to exploit rich rendered data such as color, depth, motion vectors at low resolution, there is little discussion on how to harness high-frequency information that is readily available in the high-resolution (HR) G-buffers of modern renders. In this article, we seek to investigate how to fully leverage information from HR G-buffers to maximize the visual quality of supersampling results. We propose a neural network for real-time supersampling of rendered content, which is based on several core designs, including gated G-buffers encoder, G-buffers attended encoder and reflection-aware loss. These designs are especially made for the sake of effectively using HR G-buffers, enabling faithful recovery of a variety of high-frequency scene details from low-resolution, highly aliased inputs. Furthermore, a simple occlusion-aware blender is proposed to efficiently rectify dis-occluded features in the warped previous frame, allowing us to better exploit history information to improve temporal stability. The experiments show that our method, equipped with strong ability to harness HR G-buffer information, significantly improves the visual fidelity of high-resolution reconstructions upon previous state-of-the-art methods, even for challenging $4 \times 4$4×4 upsampling, while still being compute-efficient.

Abstract In this paper, we propose a few-shot method for pose transfer of anime characters—given a source image of an anime character and a target pose, we transfer the pose of the target to the source character. Despite recent advances in pose transfer on real people images, these methods typically require large numbers of training images of different person under different poses to achieve reasonable results. However, anime character images are expensive to obtain they are created with a lot of artistic authoring. To address this, we propose a meta-learning framework for few-shot pose transfer, which can well generalize to an unseen character given just a few examples of the character. Further, we propose fusion residual blocks to align the features of the source and target so that the appearance of the source character can be well transferred to the target pose. Experiments show that our method outperforms leading pose transfer methods, especially when the source characters are not in the training set.

Threshold segmentation based on swarm intelligence optimization algorithm is a research hotspot in image processing, because of its good segmentation effect and easy implementation. This paper proposes an image threshold segmentation method based on an improved sparrow search algorithm and 2-D maximum entropy method. In the proposed algorithm, the nonlinear inertia weight is introduced into the entrants' update formula to improve the local exploration ability of the algorithm, and Levy flight is introduced into the vigilant sparrows' update formula to prevent the algorithm from falling into the local optimal solution in the later stage of iteration. In addition, improved sparrow search algorithm is tested on fifteen benchmark functions. The results represent the merit of the proposed algorithm with respect to other algorithms. Finally, the proposed algorithm is applied to entropy based image segmentation. Experiment results on classical images and medical images show that the proposed method improves the segmentation effect in terms of peak signal-to-noise ratio and feature similarity.

This article addresses the cooperative learning formation control problem for multiple homogeneous marine surface vessels (MSVs) subject to external time-varying disturbances and modeling uncertainties under the prescribed performance constraint. The modeling uncertainties, including hydrodynamic damping terms and unmodeled dynamics are identified/learned by the localized radial basis function neural networks (NNs) in a cooperative way. Disturbance observers are incorporated into the formation control design to compensate for the external time-varying disturbances. A novel cooperative learning formation controller is proposed, which is shown to be capable not only of fulfilling the predefined formation pattern with guaranteed prescribed performance but also of identifying/learning the associated uncertain dynamics based on the cooperative deterministic learning theory. Moreover, the learned knowledge on identified uncertain dynamics is stored in NN models with converged constant NN weights. Based on the stored knowledge, an experience-based formation controller is developed, which can improve the control performance including reduction of the computational burden, while guaranteeing prescribed performance of formation tracking errors. Simulation results demonstrate the effectiveness of the proposed formation control protocol.

This article investigates the problem of small fault detection (sFD) for discrete-time nonlinear systems with uncertain dynamics. The faults are considered to be "small" in the sense that the system trajectories in the faulty mode always remain close to those in the normal mode, and the magnitude of fault can be smaller than that of the system's uncertain dynamics. A novel adaptive dynamics learning-based sFD framework is proposed. Specifically, an adaptive dynamics learning approach using radial basis function neural networks (RBF NNs) is first developed to achieve locally accurate identification of the system uncertain dynamics, where the obtained knowledge can be stored and represented in terms of constant RBF NNs. Based on this, a novel residual system is designed by incorporating a newmechanism of absolute measurement of system dynamics changes induced by small faults. An adaptive threshold is then developed for real-time sFD decision making. Rigorous analysis is performed to derive the detectability condition and the analytical upper bound for sFD time. Simulation studies, including an application to a three-tank benchmark engineering system, are conducted to demonstrate the effectiveness and advantages of the proposed approach.

For an uncertain multiagent system, distributed cooperative learning control exerting the learning capability of the control system in a cooperative way is one of the most important and challenging issues. This article aims to address this issue for an uncertain high-order nonlinear multiagent system with guaranteed transient performance and preserved initial connectivity under an undirected and static communication topology. The considered multiagent system has an identical structure and the uncertain agent dynamics are estimated by localized radial basis function (RBF) neural networks (NNs) in a cooperative way. The NN weight estimates are rigorously proven to converge to small neighborhoods of their common optimal values along the union of all agents' trajectories by a deterministic learning theory. Consequently, the associated uncertain dynamics can be locally accurately identified and can be stored and represented by constant RBF networks. Using the stored knowledge on identified system dynamics, an experience-based distributed controller is proposed to improve the control performance and reduce the computational burden. The theoretical results are demonstrated on an application to the formation control of a group of unmanned surface vehicles.

This paper addresses the formation control problem for a group of mechanical systems with nonlinear uncertain dynamics under the virtual leader-following framework. New cooperative deterministic learning-based adaptive formation control algorithms are proposed. Specifically, the virtual leader dynamics is constructed as a linear system subject to unknown bounded inputs, so as to produce more diverse reference signals for formation tracking control. A cooperative discontinuous nonlinear estimation protocol is first proposed to estimate the leader's state information. Based on this, a cooperative deterministic learning formation control protocol is developed using artificial neural networks, such that formation tracking control and locally-accurate nonlinear identification with learning knowledge consensus can be achieved simultaneously. Finally, by utilizing the learned knowledge represented by constant neural networks, an experience-based distributed control protocol is further proposed to enable position-swappable formation control. Numerical simulations using a group of autonomous underwater vehicles have been conducted to demonstrate the effectiveness and usefulness of the proposed results.

This paper presents adaptive neural tracking control of underactuated surface vessels with modeling uncertainties and time-varying external disturbances, where the tracking errors consisting of position and orientation errors are required to keep inside their predefined feasible regions in which the controller singularity problem does not happen. To provide the preselected specifications on the transient and steady-state performances of the tracking errors, the boundary functions of the predefined regions are taken as exponentially decaying functions of time. The unknown external disturbances are estimated by disturbance observers and then are compensated in the feedforward control loop to improve the robustness against the disturbances. Based on the dynamic surface control technique, backstepping procedure, logarithmic barrier functions, and control Lyapunov synthesis, singularity-free controllers are presented to guarantee the satisfaction of predefined performance requirements. In addition to the nominal case when the accurate model of a marine vessel is known a priori, the modeling uncertainties in the form of unknown nonlinear functions are also discussed. Adaptive neural control with the compensations of modeling uncertainties and external disturbances is developed to achieve the boundedness of the signals in the closed-loop system with guaranteed transient and steady-state tracking performances. Simulation results show the performance of the vessel control systems.

In this paper, a new concept of formation learning control is introduced to the field of formation control of multiple autonomous underwater vehicles (AUVs), which specifies a joint objective of distributed formation tracking control and learning/identification of nonlinear uncertain AUV dynamics. A novel two-layer distributed formation learning control scheme is proposed, which consists of an upper-layer distributed adaptive observer and a lower-layer decentralized deterministic learning controller. This new formation learning control scheme advances existing techniques in three important ways: 1) the multi-AUV system under consideration has heterogeneous nonlinear uncertain dynamics; 2) the formation learning control protocol can be designed and implemented by each local AUV agent in a fully distributed fashion without using any global information; and 3) in addition to the formation control performance, the distributed control protocol is also capable of accurately identifying the AUVs' heterogeneous nonlinear uncertain dynamics and utilizing experiences to improve formation control performance. Extensive simulations have been conducted to demonstrate the effectiveness of the proposed results.

This paper presents a new exact-memory delay control scheme for a class of uncertain systems with time-varying state delay under the integral quadratic constraint (IQC) framework. The uncertain system is described as a linear fractional transformation model including a state-delayed linear time-invariant (LTI) system and time-varying structured uncertainties. The proposed exact-memory delay controller consists of a linear state-feedback control law and an additional term that captures the delay behavior of the plant. We first explore the delay stability and the L2-gain performance using dynamic IQCs incorporated with quadratic Lyapunov functions. Then, the design of exact-memory controllers that guarantee desired L2-gain performance is examined. The resulting delay control synthesis conditions are formulated in terms of linear matrix inequalities, which are convex on all design variables including the scaling matrices associated with the IQC multipliers. The IQC-based exact-memory control scheme provides a novel approach for delay control designs via convex optimization, and advances existing control methods in two important ways: 1) better controlled performance and 2) simplified design procedure with less computational cost. The effectiveness and advantages of the proposed approach have been demonstrated through numerical studies.

Abstract In this paper, we extend the deterministic learning theory to sampled-data nonlinear systems. Based on the Euler approximate model, the adaptive neural network identifier with a normalized learning algorithm is proposed. It is proven that by properly setting the sampling period, the overall system can be guaranteed to be stable and partial neural network weights can exponentially converge to their optimal values under the satisfaction of the partial persistent excitation (PE) condition. Consequently, locally accurate learning of the nonlinear dynamics can be achieved, and the knowledge can be represented by using constant-weight neural networks. Furthermore, we present a performance analysis for the learning algorithm by developing explicit bounds on the learning rate and accuracy. Several factors that influence learning, including the PE level, the learning gain, and the sampling period, are investigated. Simulation studies are included to demonstrate the effectiveness of the approach.