Peng Li

Distributed photovoltaic (PV) power forecasting remains difficult because existing data-driven methods mainly learn statistical correlations from historical samples and often produce physically inconsistent outputs under rapidly changing weather. From a measurement perspective, this task requires reliable inference of an unobserved future power quantity from heterogeneous and uncertain inputs, including historical power measurements and numerical weather predictions. To address this issue, this paper proposes a forecasting framework that dynamically fuses physical priors and forecast meteorological information. Its novelty lies in treating the physical power envelope and baseline as interpretable measurement references rather than merely concatenating them with data-driven features. A physical power upper bound and a physical baseline power are first constructed from irradiance, temperature, wind speed, and humidity. A dual-branch TimeMixer encoder separately models the forecast meteorological and physical-prior sequences, while multi-scale cross attention and dynamic gating adaptively fuse weather evolution with physical power evolution. Residual correction around the physical baseline is further combined with physics-constrained losses to suppress over-bound and nighttime-positive predictions. Experiments on three distributed PV sites show that the proposed method achieves R 2 values of 0.874, 0.898, and 0.766 and root mean square error (RMSE) values of 5.603, 9.854, and 6.423 kW, respectively, with a false alarm rate (FAR) of 0 across all sites. Compared with the best-performing baseline at each site, it reduces RMSE by up to 35.8% under normal conditions and by up to 37.7% under complex weather; even in complex-weather scenarios, it maintains R 2 values of 0.679–0.801 and a FAR of 0. These results demonstrate improved accuracy, traceability to physically meaningful reference quantities, and robustness under challenging weather, with broader relevance to measurement-informed monitoring, grid operation, and energy management.

Partial discharge (PD) is one of key manifestations of early insulation faults in power equipment. Under various application environments, PD fault features typically exhibit high aggregation, indistinct separability, and density imbalance. To address these challenges, an adaptive Multivariate Optimization Algorithm for PD Defect Feature Classification in Multiple Scenarios (AMO-PDDFC) is proposed by this paper. A maximum-density deviation data-driven strategy is introduced to adaptively determine effective scanning radii and identify high-frequency PD regions as initial search targets. Global and local search atoms are constructed based on the historical memory search information. Furthermore, an adaptive iteration termination criterion is designed to improve convergence efficiency, and a dynamic nearest-distance strategy is employed to handle residual outliers. Extensive experiments on real and noise-mixed PD datasets demonstrate that AMO-PDDFC achieves superior recognition performance, with an overall accuracy of 99.52%.

Advanced Metering Infrastructure (AMI) of the power consumption side, serving as the neural hub of modern power systems, orchestrates real time energy data flows between distributed smart meters and centralized grid control centers, yet this pervasive connectivity introduces critical vulnerabilities to cyber attacks. Traditional intrusion detection systems struggle with the high-dimensional traffic patterns and severe class imbalance inherent in AMI networks, particularly in identifying handful attack variants amidst overwhelming normal traffic. To address these challenges, this paper proposes a detection model incorporating Random Fourier Features (RFF), Adaptive Weighted Focal Loss (AWFL), and a dual-scale convolutional neural network. The model first employs RFF to construct a low-dimensional explicit mapping, replacing the implicit kernel function in kernel PCA. This transforms the original nonlinear eigenvalue problem into linear PCA on random features, thereby reducing computational complexity significantly while preserving the capability of nonlinear dimensionality reduction. Subsequently, AWFL dynamically adjusts the model’s focus on different class samples during training, significantly improving class imbalance handling in network intrusion datasets. Finally, a dual-scale depthwise separable convolution module (DDSCM) is adopted to extract spatial features from input information, enabling more accurate identification of intrusion types. Evaluations on the NSL-KDD dataset and TON_IoT dataset demonstrate that the proposed model achieves superior intrusion detection performance, outperforming various comparative models in multiclass classification tasks regarding accuracy, precision, recall and F1-score.

A novel dynamic inner manifold learning algorithm based on forward selection component variables (FSV-DiML) is proposed for industrial process monitoring. The method begins with forward selection component analysis (FSCA) to select monitoring variables, effectively removing redundancies while preserving critical information. DiML is then used to construct a new objective function designed to extract dynamic latent variables with manifold structures from high-dimensional and nonlinear data. This approach enables the separation of process data into dynamic and static subspaces. However, since high-dimensional data often contains low-dimensional manifold structures, neighborhood preserving embedding (NPE) is applied to reduce the dimensionality of each subspace while preserving its manifold structure. Finally, Bayesian inference and fusion techniques are employed to merge various monitoring statistics into a unified monitoring index, further improving fault detection performance. The proposed FSV-DiML model is validated on the Tennessee Eastman and Multiphase Flow Facility processes, demonstrating significant improvements over existing methods in accurately detecting faults in complex industrial processes.

Accurate photovoltaic power (PVP) prediction is a prerequisite for the efficient and stable operation of new power systems. While existing research has extensively explored the relationship between global factors such as temperature, irradiance, and photovoltaic power, the local dynamic impacts of these factors are often overlooked, which may reduce the accuracy of predictions. To address this issue, this paper considers the dynamic interrelationships among multiple factors and proposes a dynamic locally featured embedding-based broad learning system (DLFE-BLS) algorithm for PVP prediction. Firstly, a novel dynamic phase space reconstruction method (DPSR) is proposed to characterize the dynamic properties of multivariate data. Furthermore, a dynamic local featured embedding (DLFE) algorithm is introduced to extract local dynamic features from multivariate data. Finally, by integrating the dynamic reconstruction and dynamic feature extraction processes into the broad learning system (BLS) framework, we propose the DLFE-BLS algorithm to improve the accuracy of PVP prediction. Case studies have shown that DLFE-BLS outperforms other models in terms of prediction accuracy. Additionally, it has the highest accuracy when applied to transfer prediction.