The output of photovoltaic (PV) power generation systems remains uncertain primarily due to the uncontrollable nature of weather conditions, which may introduce disturbances to the power grid upon integrating PV systems. Accurate short-term PV power forecasting is an essential approach for ensuring the stability of the power system. The paper proposes a short-term PV power forecasting model based on improved zebra optimization algorithm (IZOA)-stochastic configuration network (SCN). First, the historical PV data are divided into three weather patterns, effectively reducing the uncertainty of PV power. Second, a prediction model based on SCN is developed. To enhance the forecasting model’s accuracy even further, the IZOA is introduced to optimize the key parameters of the SCN. Finally, IZOA-SCN is employed for short-term PV power through various weather patterns. Experiment results show that the proposed method significantly improves the prediction accuracy in contrast to other comparison models.

As extreme weather events become more frequent, the icing of transmission lines in winter has become more common, causing significant economic losses to power systems and drawing increasing attention. However, owing to the complexity of the conductor icing process, establishing high-precision ice thickness prediction models is vital for ensuring the safe and stable operation of power grids. Therefore, this paper proposes a hybrid model combining an improved snake optimization (ISO) algorithm, deep extreme learning machine (DELM), and hybrid kernel extreme learning machine (HKELM). Firstly, based on the analysis of the factors that influence the icing, the temperature, the humidity, the wind velocity, the wind direction, and the precipitation are selected as the weather parameters for the prediction model of the transmission line icing. Secondly, the HKELM is introduced into the regression layer of DELM to obtain the deep hybrid kernel extreme learning machine (DHKELM) model for ice thickness prediction. The SO algorithm is then augmented by incorporating the Latin hypercube sampling technique, t-distribution mutation strategy, and Cauchy mutation, enhancing its convergence. Finally, the ISO-DHKELM model is applied to the icing data of transmission lines in Sichuan Province for experiments. The simulation results indicate that this model not only performs well, but also enhances the accuracy of ice thickness predictions.

Maximum power point tracking (MPPT) is a pivotal technology for photovoltaic (PV) systems. Due to variations in light intensity and temperature, the output characteristic curve of the photovoltaic system exhibits multi-peak phenomena, and traditional MPPT algorithms perform poorly in complex and changing environments. Therefore, this paper proposes an MPPT method grounded on an improved RIME optimization algorithm (IRIME). This approach enhances the algorithm’s exploratory capabilities by incorporating logistic mapping during the initialization stage. Furthermore, it optimizes the algorithm’s parameters through sequences generated by piecewise mapping, thereby realizing a harmonious balance between global exploration and local exploitation. Additionally, the introduction of an adaptive inertia weight dynamically adjusts the search strategy, thereby increasing the adaptability, convergence speed, and search efficiency of algorithm. Compared to PSO-MPPT and RIME-MPPT, the proposed method reduced the average tracking time by 0.085 s and 0.425 s, respectively. Additionally, in terms of maximum output power, the proposed method achieved an average improvement of 0.97% and 3.48% over the aforementioned methods, respectively. Particularly in PV system simulations under varying irradiance and temperature conditions, the proposed method consistently achieves the best results, verifying its efficient, stable, and fast-converging characteristics in MPPT strategies for PV systems.

To address the voltage over-limit issue in rural distribution networks caused by high-penetration distributed photovoltaic (DPV) integration, as well as the frequent voltage disturbances resulting from frequent load switching and variable operating conditions in agricultural grid systems, this paper proposes a dual-layer optimized configuration and adaptive voltage control method for grid-forming energy storage systems. First, an outer-layer siting and sizing model is established with constraints including voltage stability, deviation, network losses, and economic factors. The configuration scheme is solved using the Multi-Objective Particle Swarm Optimization (MOPSO) algorithm. Then, an inner-layer rolling optimized dispatch model is constructed with Model Predictive Control (MPC) as the core, and an adaptive reactive power–voltage control method based on Virtual Synchronous Generator (VSG) control is proposed to enhance transient voltage support under disturbance conditions. Simulation analysis based on an actual 10 kV rural distribution line verifies that the proposed method can effectively mitigate overvoltage issues and alleviate reverse power flow during typical daily operation, while significantly improving node voltage recovery speed and reactive power support capability under voltage disturbances.

Accurate estimation of the state of health (SOH) and prediction of the remaining useful life (RUL) of lithium-ion batteries (LIBs) are critical for ensuring system reliability and safety across diverse energy storage applications. This paper proposes a hybrid deep learning framework that integrates an axial-embedding Transformer (AxEmbTrans) encoder and a bidirectional LSTM (BiLSTM) module for the joint estimation of SOH and RUL. The AxEmbTrans encoder employs axial attention with abstract embeddings to capture global dependencies among multidimensional health features at reduced computational complexity compared to standard self-attention, while the BiLSTM models local temporal dynamics and short-term degradation fluctuations across consecutive cycles, with its bidirectional structure enhancing robustness against transient noise. Informative health features are extracted from charge–discharge curves, grouped into temporal, energy, and thermal categories, and fused using local linear embedding (LLE) for nonlinear dimensionality reduction. An improved Newton–Raphson-based optimizer (INRBO) is introduced to automatically tune the framework’s key hyperparameters, including the hidden dimension, number of attention heads, number of BiLSTM units, and learning rate, incorporating directional similarity modulation and multi-elite guidance to overcome the convergence instability of the standard NRBO. Extensive experiments on NASA and Maryland datasets demonstrate that the proposed method consistently outperforms baselines in both SOH and RUL prediction, achieving higher accuracy, improved robustness, and better cross-condition generalization.