The deployment of heterogeneous Automated Guided Vehicles (AGVs) in smart manufacturing requires control strategies that can accommodate distinct actuation characteristics and constraints. This paper proposes a Multi-Factor Coupled Parameter-Adaptive Model Predictive Control (MFCP-AMPC) framework. Unlike conventional approaches requiring vehicle-specific tuning, this framework unifies differential-drive, dual-steer, and mecanum-wheel platforms under a single parameter-varying state-space model that respects the specific actuation limits of each topology. A key contribution is the multi-factor coupling mechanism that dynamically adjusts the prediction horizon and weighting matrices based on path curvature, vehicle speed, and tracking error. Experiments on industrial AGV prototypes demonstrate that the framework achieves robust tracking precision under varying payloads. Crucially, by acknowledging physical limits, the framework achieves strict millimeter-level accuracy (RMSE < 7 mm) in quasi-static low-speed complex maneuvers (v <= 0.3 m/s), and maintains highly competitive industrial precision (RMSE approximate to 15 similar to 25 mm) under aggressive high-speed tracking (v >= 1.0 m/s). Crucially, the proposed method significantly improves the control input smoothness (Smoothness Index > 0.75), thereby reducing mechanical wear and preventing actuator saturation. Real-time validation (12 ms average solve time on an Intel i7 IPC) confirms its suitability for resource-constrained industrial controllers.

Accurate wind power forecasting is essential for the reliable and efficient operation of renewable energy–dominated power systems, as it directly affects power system scheduling and grid stability. Nevertheless, the inherent intermittency, strong nonlinearity, and stochastic characteristics of wind power generation pose significant challenges to short-term forecasting accuracy. To tackle these issues, this paper develops a hybrid forecasting framework that integrates Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN), Variational Mode Decomposition (VMD), and a deep learning predictor based on Convolutional Neural Networks and Long Short-Term Memory networks (CNN–LSTM). Specifically, CEEMDAN is first employed to decompose the original wind power time series into multiple intrinsic mode functions at different temporal scales. Subsequently, VMD is applied to the high-frequency components to further suppress noise and enhance signal stationarity. Each decomposed subsequence is then modeled and predicted using a CNN–LSTM architecture, and the final wind power forecast is obtained through reconstruction of all predicted components. Extensive experiments conducted on real-world wind farm data that the proposed hybrid model consistently outperforms several benchmark methods in terms of forecasting accuracy, thereby verifying its effectiveness and practical applicability for short-term wind power forecasting.

Unmatched disturbances pose significant challenges to the control performance of unmanned surface vehicles (USVs), necessitating effective mitigation control strategies. This study proposes composite disturbance observer-based sliding mode control (SMC) strategies to address the trajectory tracking control problem of USVs under the influence of unmatched and matched disturbances, and input saturation. First, a composite disturbance observer is designed to estimate and compensate for the effects of both unmatched and matched disturbances. Subsequently, an SMC strategy is proposed to eliminate input discrepancies and ensure system state convergence within a designated region. Moreover, to mitigate the reliance on prior knowledge of the upper bounds of lumped disturbances, an improved SMC strategy is further proposed that incorporates a positive semi-definite barrier (PSDB) function. This strategy enhances the steady-state performance of the system and prevents excessive estimation of control gains. Finally, numerical simulations validate the effectiveness of the proposed control strategies.

Accurate photovoltaic (PV) power forecasting is essential for renewable energy integration and power system scheduling. This study proposes a hybrid framework for short-term point and interval PV power forecasting. The framework combines complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), partial least squares regression (PLS), Pearson correlation coefficient (PCC) analysis, a Kalman-Infused Adaptive Attention Transformer (KAAformer), bidirectional long short-term memory (BiLSTM), and conformal prediction (CP). CEEMDAN is first used to decompose the PV power sequence into intrinsic mode functions, while PLS and PCC are used to select informative components and relevant meteorological variables. Based on the selected features, the KAAformer-BiLSTM model is developed to enhance noise-aware temporal feature learning by em bedding Kalman-filter-based state correction into the attention mechanism. CP is then introduced to construct distribution-free prediction intervals. The proposed framework is evaluated using PV data from the Yulara Solar System in Australia and further validated on an additional PV dataset from China. Experimental results show that the proposed method achieves lower prediction errors, stronger correlation with observed values, and more compact prediction intervals than representative baseline models. These results confirm the effectiveness and generalization ability of the proposed framework for reliable short-term PV power forecasting and uncertainty quantification.

This paper presents a fixed-time composite prescribed performance control law to address the trajectory-tracking control problem for unmanned surface vehicles (USVs) subjected to external disturbances, error constraints, and time constraints. Initially, a fixed-time disturbance observer employing the super-twisting algorithm is designed to estimate time-varying disturbances, ensuring that the estimation error converges within a fixed time. Subsequently, by formulating an error performance function alongside an error transformation function, the prescribed performance control strategy is introduced, thus assuring that the USVs’ trajectory-tracking error satisfies predefined performance criteria. Then a composite control law is proposed by integrating a non-singular terminal sliding mode (NSTSM) surface with the disturbance observer and the prescribed performance control strategy, which guarantees that the system stabilization time is independent of the initial state. Finally, numerical simulations conducted on a fully actuated USV demonstrate the efficacy and advantages of the proposed control law.

With the increasing demand for large-scale maritime security and ocean surveillance, infrared ship target detection systems are required to process massive infrared data streams with high accuracy and real-time performance. However, the small size of ship targets, low contrast in infrared imagery, and interference from complex sea backgrounds pose significant challenges to both detection accuracy and computational efficiency, limiting the practical deployment of existing methods in high-performance or distributed computing environments. To address these challenges, this paper proposes an efficient infrared ship detection framework, termed AAF-YOLO (ADown+AKConv-SE+FLAHead-YOLO), which is designed to achieve a favorable balance between detection performance and computational cost, making it suitable for real-time inference and parallel acceleration on modern supercomputing platforms. Experimental results on an infrared ship dataset demonstrate that AAF-YOLO achieves a mAP of 90.9

With the continuous development of society and the continuous update of science and technology, people's dependence on energy is becoming higher and higher, resulting in the disharmony between energy and social development. Therefore, the study of building cooling load forecasting is of great significance for energy conservation. The traditional building cooling load forecasting methods mostly use a single forecasting model, and the prediction accuracy is not very high. In order to solve these problems, this paper puts forward a method which uses genetic algorithm to improve the relevant parameters of neural network to forecast the building cooling load. The experiment shows that the prediction accuracy of the improved model has been improved, which lays a certain foundation for energy saving equipment.

The increasing global demand for renewable energy poses significant challenges to grid stability due to the fluctuation and unpredictability of photovoltaic (PV) power generation. To enhance the accuracy of short-term PV power prediction, this study proposes an innovative integrated model that combines Convolutional Neural Networks (CNN) and Bidirectional Long Short-Term Memory (BiLSTM), optimized using the Starfish Optimization Algorithm (SFOA) and integrated with a multi-method data processing framework. To reduce input feature redundancy and improve prediction accuracy under different conditions, the K-means clustering algorithm is employed to classify past data into three typical weather scenarios. Empirical Mode Decomposition is utilized for multi-scale feature extraction, while Kernel Principal Component Analysis is applied to reduce data redundancy by extracting nonlinear principal components. A hybrid CNN-BiLSTM neural network is then constructed, with its hyperparameters optimized using SFOA to enhance feature extraction and sequence modeling capabilities. The experiments were carried out with historical data from a Chinese PV power station, and the results were compared with other existing prediction models. The results demonstrate that the Root Mean Square Error of PV power generation prediction for three scenarios are 9.8212, 12.4448, and 6.2017, respectively, outperforming all other comparative models.

The randomness and volatility of wind power pose significant challenges for short-term forecasting, requiring the model to capture both temporal dynamics and the spatial correlations among turbines. To address this issue, this paper proposes a Spatio-Temporal Adaptive Graph Convolutional Recurrent Network (STAGCRN). The proposed method dynamically constructs and updates the spatial relationship graph through node adaptive parameter learning (NAPL) and a data adaptive graph generation (DAGG) module, enabling more accurate modeling of spatio-temporal dependencies in wind power data. In addition, a spatio-temporal self-attention mechanism is introduced to enhance the model’s ability to capture both short-term fluctuations and long-term temporal patterns. By stacking multiple spatio-temporal adaptive graph convolutional recurrent layers, the model is capable of extracting complex nonlinear characteristics in wind power sequences. Experimental results based on real wind farm data demonstrate that the proposed method achieves significantly improved prediction accuracy and robustness compared with existing approaches in short-term wind power forecasting tasks.

Addressing the challenge of detecting small floating target in complex water environments, where it is difficult to balance real-time performance and end-to-end capabilities, this study introduces a specialized model called the Enhancing Real-Time Detection Transformer (ERT-DETR). This model integrates a Dynamic Feature Pyramid Network (DFPNet) with a Micro-Attention Module (MAM) to achieve refined feature extraction and enhanced object recognition mechanisms. Additionally, it incorporates an Inner Intersection over Union (Inner-IoU) auxiliary bounding box loss function, which accelerates model convergence and improves bounding box accuracy. Experiments conducted on the Floating Object in Water Image Dataset (FloW-Img) demonstrate that the ERT-DETR model achieves a 92.6% Average Precision (AP) and 118.84 Frames Per Second (FPS), outperforming the baseline Real-Time Detection Transformer (RT-DETR) by 4.3% and 19.8%, respectively. The ERT-DETR model's precision and real-time performance in dynamic water surface environments surpass those of the most advanced You Only Look Once (YOLO) and Detection Transformer (DETR) series detectors. This model is significant for enhancing small floating target detection capabilities in complex water environments and can be extended to applications in marine management, environmental monitoring, and vessel tracking.

Accurate power load forecasting provides critical data support for energy management departments, guiding power production sectors to adjust generation output appropriately and reduce energy consumption. However, the high volatility inherent in load data poses significant challenges to achieving precise predictions. This paper first elucidates the fundamental principles of the Informer model for load forecasting and evaluates its prediction performance using real-world power load data from a district in Chongqing, China. Experimental results demonstrate that the Informer model achieves relatively accurate predictions, showcasing its robust forecasting capabilities.

This paper proposes an optimal scheduling algorithm based on particle swarm optimization (PSO) for a water-wind-solar-storage complementary system. The goal is to enhance the utilization of hydropower, wind, and solar generation while mitigating output volatility during grid integration. The output model for the system is established, with constraints and output equations for hydropower, wind, solar, and pumped storage. The total output power fluctuation is chosen as the objective function. An improved PSO algorithm is applied to optimize the system. The results demonstrate that the proposed method significantly increases renewable energy absorption and reduces the grid impact of multiple energy sources when integrated, confirming its robust optimization performance.