EAI Endorsed Transactions on Energy Web

Enhancing Power Grid Reliability with AGC and PSO: Insights from the Timimoun Photovoltaic Park

Ali Abderrazak Tadjeddine — 2024-11-22

This article investigates the impact of integrating Variable Renewable Energy (VRE), specifically solar energy from the Timimoun Photovoltaic Park, on the PIAT electrical grid stability in southern Algeria. The study focuses on how fluctuations in power demand and changes in weather conditions can affect grid frequency control, potentially leading to transient stability issues. To address these challenges, the research proposes the implementation of an Automatic Generation Control (AGC) system combined with the Particle Swarm Optimization (PSO) algorithm to optimize solar energy distribution. This approach effectively regulates real-time frequency deviations resulting from VRE integration, ensuring balanced supply and demand, and controllable power factor injection. The findings demonstrate that the integration of AGC and PSO stabilizes the frequency at the Timimoun Photovoltaic Park and reduces total active losses in the PIAT network by 13.88%. Additionally, strategic power factor control at the injection buses ensures optimal power quality and maximizes the utilization of the photovoltaic park, leading to a 4.84% reduction in the PIAT grid's reliance on gas turbines. This approach contributes to lowering operational costs, reducing carbon emissions, and supporting a transition to greener energy.

Construction of a Fast Monitoring System for Electric Energy Equipment Status Based on Data Mining

Fusheng Wei — 2024-12-20

In modern power system operation, it is crucial to achieve fast and accurate monitoring of the electrical equipment status. To achieve this fast and accurate detection, this study proposes a generative adversarial network that combines edge features to amplify and recognize infrared images of devices, aiming to improve the model’s training effect. This model extracted edge features from infrared images to eliminate background noise in infrared images to achieve the goal of improving the accurate monitoring of the status of electrical equipment. The results showed that on the balanced dataset, the recognition accuracy of the model could reach about 96%, and the recognition effect of the model was relatively stable. On imbalanced datasets, the highest model recognition accuracy was around 89%, and the model recognition accuracy fluctuated greatly. The constructed model effectively improves the accuracy of monitoring the operating status of electric energy equipment, achieving fast and accurate monitoring of this state. This study can achieve rapid monitoring of the operating status of electric energy equipment, effectively reducing the operation and maintenance costs of the power system.

Improving Fault Classification Accuracy Using Wavelet Transform and Random Forest with STATCOM Integration

Shradha Umathe — 2024-11-05

INTRODUCTION: Fault detection in transmission lines is critical for keeping the grid stable and reliable. This research offers a new methodology, the Wavelet Transform-Enhanced Random Forest Fault Classification System with STATCOM Integration (WERFCS-SI), to solve the shortcomings of existing fault detection approaches.

OBJECTIVES: The integration of STATCOM-compensated transmission lines improves fault detection capabilities. The Wavelet Transform finds faults by analysing approximation and detail coefficients, allowing for multiresolution analysis and exact fault localisation.

METHODS: Feature selection approaches, such as information gain, are used to discover and keep relevant features, increasing classification accuracy.

RESULTS: Due to its ability to process complex, high-dimensional data and identify minute feature connections, Random Forest (RF) is utilised for classification tasks. The proposed approach improves RF model performance while maintaining precision.

CONCLUSION: The integrated technique simplifies fault categorisation, increasing accuracy and efficiency by detecting problems in the transmission line system.

CNN Based Fault Classification and Predition of 33kw Solar PV System with IoT Based Smart Data Collection Setup

K. Punitha — 2024-12-12

A Solar Photovoltaic (PV) System is an energy conversion system that uses the photovoltaic effect to convert sunlight into electricity. A fault in a Solar Photovoltaic (PV) system refers to any abnormal condition or defect that disrupts the normal operation and performance of the solar system. These faults can arise from a variety of factors, including environmental conditions, manufacturing defects, installation errors, and wear and tear of the components. Fault diagnosis in solar PV systems involves the detection, identification, and rectification of faults or abnormalities that can occur due to various reasons. By detecting and addressing faults early, systems can maintain optimal performance levels. Machine Learning (ML) in Solar Photovoltaic (PV) systems refers to the application of algorithms and statistical models that enable computers to perform specific tasks without using explicit instructions, relying instead on patterns and inference. In the context of solar PV systems, ML is used to analyse and interpret vast amounts of data generated by these systems to enhance their efficiency, predict energy production, detect and diagnose faults, and optimize maintenance and operation. By analysing data from sensors and system logs, ML algorithms can identify patterns indicative of faults or inefficiencies, such as shading, soiling, or equipment malfunctions, often before they become serious issues. Convolutional Neural Networks (CNNs) are a class of deep learning algorithms most commonly applied. They are particularly powerful for tasks involving data recognition, classification, and analysis due to their ability to automatically and adaptively learn spatial hierarchies of features. This research presents a unique machine learning model based fault diagnosis and detection method for a 33 KW solar PV system at P.S.R. Engineering College, Sivakasi. The real-time data from the PV system for five years, covering 23,000 instances of eight types of faults such as Cell Cracks or Hot Spots, Partial Shading, sensor fault, Module failure, Ground Faults, Communication Errors, Environmental Factors, Grid Connectivity Issues are collected. CNN is applied to the data and analysed their performance in terms of accuracy, precision, and standard deviation (SD)-score. It is found that CNN achieved the best results, with an accuracy of 98.7% a precision of 95%, a recall of 98%, and an F1 score of 96.5%. Therefore, CNN is used as the fault prediction also. The model is implemented using Python programming language and demonstrated its effectiveness on test cases. The smart data gathering system was achieved utilizing an ESP32 node with several sensors. The obtained data was stored in an authorized Google Sheet and compared to predetermined threshold ranges. When any parameter deviates from its threshold value, the ESP32 node starts a cooling and dust cleaning procedure with a water pump and drip pipe configuration. If the divergence persists, the ESP32 node activates a camera to capture an image of the panel and sends it to the Google Sheet via a connection for further analysis and fault correction.

Enhancing Torque Smoothness in BLDC Motors with Built-in DC-DC Converter via Bitterling Fish Optimization Algorithm

K. Balamurugan — 2025-01-30

Brushless DC (BLDC) motors are efficient and robust electric motors with fewer moving parts, but their application is often limited by torque ripple, which arises from current variations between the entering and exiting phases during commutation. This study aims to minimize torque ripple in BLDC motors integrated with a DC-DC converter. The proposed optimization method utilizes the Bitterling Fish Optimization (BFO) Algorithm to effectively control torque error and speed, addressing the torque ripple caused by current variations during commutation. The proposed method is implemented using the MATLAB working environment and compared with various existing methods like Spider Web Algorithm (SWA), Improved Tunicate Swarm Optimization Algorithm (ITSA), and Harris Hawks Optimizer with Black Widow Optimization (HHO-BWO). The results indicate that the proposed method achieves a reduced torque ripple rate of 9.64, significantly lower than the rates of 17.32, 11.20 and 22.19 for ITSA, HHO-BWO and SWA respectively. Additionally, the proposed approach exhibits low error of 0.168, outperforming the existing methods errors of 0.287, 0.195 and 0.311. These findings demonstrate that the BFO algorithm effectively minimizes torque ripple more than existing optimization techniques, providing a promising solution for enhancing the performance of BLDC motors.

Investigating Safe and Economic Adjustment of Power Balance in Smart Grids Based on Integration of Renewable Energy

Hongyan Zhang — 2024-12-19

The present pace of integration of renewable sources into the electrical grid is insufficient, failing to fulfill the expectations of producers or coincide with sustainable national objectives. Furthermore, sustainable national policies are not being executed. Despite the growth of the solar and wind energy industry and the installation of decentralized energy production systems, this scenario has emerged. Several factors contribute to this scenario, including advancements in administration, forecasting, and oversight, along with enhancements in infrastructure. These issues may arise notwithstanding the decentralized nature of renewable energy sources. The integration rate of renewable energy sources into networks, along with the efficiency of these networks, is clearly hindered as a result of this. Furthermore, we will examine the problems associated with the implementation of this network. We will focus on the low injection rate and the balance between supply and demand. Subsequently, we will examine the impact they have on the operation of the interconnected system. We will provide management solutions tailored to each detected issue, along with the suggested cures for any recognized concerns. The aim is to discover the structures, procedures, and tools that will enhance the network's reliability and energy efficiency while simultaneously reducing installation costs and fortifying the network. The findings indicate that the interruptions in voltage, frequency, and power have been mitigated due to the dynamic simulations using the proposed method. The calculations were predicated on an integration of solar and wind energy, with twenty percent of the energy derived from wind.

Research on Energy-Efficient Building Design Using Target Function Optimization and Genetic Neural Networks

Youxiang Huan — 2025-01-21

OBJECTIVES: This research aims to provide data for decision-makers to achieve sustainability in building construction projects.

METHODS: A multi-objective optimization method, using the non-sorting genetic algorithm (NSGA-II), assesses energy efficiency by determining optimal wall types, insulation thickness, and insulation type. This paper utilizes the EnergyPlus API to directly call the simulation engine from within the optimization algorithm. The genetic neural network algorithm iteratively modifies design parameters (e.g., building orientation, insulation levels etc) and evaluates the resulting energy performance using EnergyPlus.

RESULTS: This reduces energy consumption and life cycle costs. The framework integrates Matlab-based approaches with traditional simulation tools like EnergyPlus. A data-driven technology compares the framework's effectiveness.

CONCLUSION: The study reveals that optimal design configurations can reduce energy consumption by 30% and life cycle costs by 20%, suggesting changes to window fenestration and envelope insulation are necessary. The framework's accuracy and simplicity make it valuable for optimizing building performance.

Fuzzy Allocation Optimization Algorithm for High-Density Storage Locations with Low Energy Consumptions

Ziyi Gao — 2024-11-04

The global demand for stored and processed data has surged due to the development of IoTs and similar computational structures, which has led to further energy consumption by concentrated data storage facilities and thus the demands of global energy and environmental needs. The current paper introduces Fuzzy Allocation Optimization Algorithm to mitigate energy consumption in high storage density settings. It uses the principles of Fuzzy logic to determine the best way to assign the tasks in relation to storage density necessity, urgency and energy consumption. Thus, the proposed approach incorporates fuzzy inference systems with multi-objective optimization methods where location of storage is dynamically assessed and assigned according to energy efficiency parameters. The findings of the simulation and case study prove that the algorithm is successful in saving energy while at the same time lowering storage I/O response time, which provides a viable solution to energy issues in evolving data centres. This work satisfies the lack of energy efficient algorithms in high density storage areas and responds to the recent calls for green technology and smart utilization of resources in the energy field. The findings are used in the promotion of significant IT infrastructures towards developing the next generation of energy efficient data centers with respect to Future Internet and evolving energy web environments.

Investigation of the anchor chain tension distribution and six-degree-of-freedom motion characteristics of a floating fan platform in a wind-wave basin

Xi Zhang — 2025-04-24

INTRODUCTION: Because of its versatility and adaptability, floating wind turbine platforms have emerged as the go-to foundation type for deep sea wind power as offshore wind power production steadily expands into deeper waters.

OBJECTIVES: But the floating platform's six-degree-of-freedom motion and the anchor chain system's mechanical reaction to the combined force of wind, wave, and current are incredibly intricate, and this directly affects the platform's stability and safety.

METHODS: Under various wind, wave, flow incidence angle, and chain length conditions, the platform's six-degree-of-freedom motion characteristics and anchor chain tension distribution are carefully studied using potential flow theory, the finite element method, and the fluid-structure coupling model. The numerical simulation combined JONSWAP wave spectrum and NPD wind spectrum to conduct multi-condition analysis.

RESULTS: the results show that the incidence Angle and anchor chain configuration have significant effects on the dynamic response of platform pitching and pitching.

CONCLUSION: This paper deeply discusses the platform motion response under the broken anchor chain, and puts forward the corresponding optimal design scheme.

Distributed Photovoltaic Power Energy Generation Prediction Based on Improved Multi-objective Particle Algorithm

Yuanzheng Xiao — 2025-03-13

Accurate prediction of distributed photovoltaic (DPV) power generation is crucial for stable grid operation, yet existing methods struggle with the non-linear, intermittent nature of solar power, and traditional machine learning models face hyperparameter selection and overfitting challenges. This study developed a highly accurate DPV power prediction method by optimizing a Long Short-Term Memory (LSTM) network's hyperparameters using an improved Multi-Objective Particle Swarm Optimization (MO-PSO) algorithm. A hybrid LSTM-PSO model was created, where the LSTM network served as the core prediction model, and the improved MO-PSO algorithm optimized its hyperparameters, enhancing generalization and avoiding overfitting. The LSTM-PSO model significantly improved prediction accuracy compared to traditional methods. Key results from two power stations included a maximum deviation of 6.2 MW at Power Station A, a peak time deviation of less than 0.1 MW at Power Station B, and a prediction interval error controlled below 30 MW at an 80% confidence level. The optimized LSTM-PSO model effectively captures DPV power generation dynamics, and the superior performance metrics demonstrate its potential for intelligent grid management. However, limitations include prediction accuracy under extreme weather and computational efficiency for large datasets. Future work will focus on broader applicability and more efficient algorithm variants.

Two-stage high-frequency switching power supply device design study

Lijuan Zhang — 2025-02-06

The current volume and efficiency of high-frequency switching power supplies in power supply system cannot meet practical requirements. Therefore, a modular equipment was studied to optimize the design of PWM rectifiers and DC-DC converters, and corresponding control strategies were adopted. At the same time, experimental verification was performed. The experimental results show that before using the control, there are two large secondary voltage ripples in the PWM rectifier, with an amplitude of approximately 8 V; After using the control, the amplitude was approximately 1 V, a decrease of 87.5%. In addition, the DC-DC converter module may have fluctuations when it is lifted, but the amplitude of the voltage wave remains basically 62 A after steady state. In practical applications, the ripple is controlled at 1 V through the proposed control method, and the actual displayed current is a relatively standard sine wave with a low distortion rate. Meanwhile, compared with other methods, the efficiency of the studied method is as high as 89.110%. Overall, the control strategy proposed by the research institute can effectively control the front and rear modules in theory. In practical applications, it can effectively improve the power output and reduce the pollution in the power grid. It has high effectiveness and feasibility in practical industrial application.

Energy management system design for high energy consuming enterprises integrating the Internet of Things and neural networks

Zhaolin Wang — 2025-05-02

INTRODUCTION: High energy consuming enterprises continue to pay increasing attention to energy consumption. Therefore, designing an energy management system is significant.

OBJECTIVES: To improve the management level and economic benefits of enterprises, a high energy consuming enterprise energy management system design based on Internet of Things technology and neural network algorithms is proposed.

METHODS: Internet of Things devices are used for data collection and transmission. The combination of neural network model prediction and optimization algorithms can achieve real-time monitoring, prediction, and optimization control of energy consumption.

RESULTS: The research results indicated that the response time of the high energy consuming enterprise energy management system proposed in the study was 80.2 ms when the number of people was 600. The fluctuation range of CPU usage within 24 hours was 14% to 45%.

CONCLUSION: A high energy consuming enterprise energy management system that integrates the Internet of Things and neural networks can manage energy more efficiently and intelligently, thereby improving the production efficiency and economic benefits of the enterprise. This helps companies gain greater advantages in fierce market competition.

Intelligent Equipment Scheduling Optimization Model for Transmission Lines Based on Improved BFO Algorithm

Wulue Zheng — 2025-04-22

INTRODUCTION: In modern power systems, the optimization of intelligent equipment scheduling for transmission lines is a key task.

OBJECTIVES: To improve the effectiveness of scheduling optimization, this study introduces an intelligent equipment scheduling optimization model for transmission lines on the ground of the improved Bacterial Foraging Optimization algorithm.

METHODS: This model achieves global and local search capabilities through an improved Bacterial Foraging Optimization algorithm, maintaining the diversity of equipment states and effectively improving the optimization level of scheduling results.

RESULTS: At 3000 iterations, the model was able to reach its optimal state, and its optimization results showed excellent performance in terms of convergence and uniformity, which was very close to the optimal solution. In practical applications, the performance of the intelligent equipment scheduling optimization model for transmission lines on the ground of the improved Bacterial Foraging Optimization algorithm is also excellent. The average line usage rate of the scheduling scheme proposed by the model reached 70.69%, while the average line usage rate of the manual scheduling scheme was only 64.63%. In addition, the optimal relative error percentage of this model is less than 2.1%, while the BRE of other algorithms reaches around 10%.

CONCLUSION: The intelligent equipment scheduling optimization model for transmission lines on the ground of improved Bacterial Foraging Optimization algorithm has important practical significance for improving the operational efficiency of the power system, reducing operating costs, and making sure the stable and reliable operation of the power system.

Construction and Application Analysis of an Intelligent Distribution Network Identification System Based on Deep Neural Networks

Yu Ma — 2024-11-21

INTRODUCTION: At present, the communication between measuring data and network topology in the distribution system cannot be accurately established. Therefore, deep neural networks were utilized to learn the mapping relationship between the measurement data and network topology, achieving topology structure discrimination under different working conditions.

OBJECTIVES: This study aims to establish a machine learning-based Intelligent Distribution Network (IDN) online topology recognition model to address the limited measurement equipment in distribution networks and improve the accuracy and efficiency of network topology recognition.

METHODS: First, light GBM was used for feature selection to reduce computational complexity and improve learning efficiency. Then, a DNN model was constructed for topological identification and enhances the model scalability through incremental and transfer learning mechanisms. In addition, the Cross-Validation Grid Search Algorithm (GSA) was used to optimize the hyperparameters to ensure that the model can achieve the optimal performance on different data sets. Finally, a new intelligent distribution network identification model (Intelligent Distribution Electricity Network Identification System, IDENIS) was constructed.

RESULTS: The study was experimentally verified on the distribution system of IEEE 33 and PG&E 69. The experimental results showed that the accuracy of the DNN-based model reached 0.9817 on the test set, while the accuracy after feature selection only decreased by 1.3%, and the features decreased by 81.8%. In the PG&E 69 node system, the features were reduced by 85.5%, while the identification accuracy was decreased by only 0.51%. These results demonstrated that the proposed method maintained high identification accuracy while reducing the computational resource consumption.

CONCLUSION: Its efficient computing speed fully meets the real-time requirements in practical applications. This paper provides new ideas and methods for achieving intelligent distribution network topology recognition of high proportion distributed power sources.

Risk prediction method for power Internet of Things operation based on ensemble learning

Chao Hong — 2025-02-26

INTRODUCTION: The power Internet of Things is an important strategic support for the State Grid Corporation of China to build an international leading energy internet enterprise. However, the operating environment of the power Internet of Things is complex and varied, which has serious implications for the safe operation of the power Internet of Things.

OBJECTIVES: To timely predict the various risk.

METHODS: A data set is fused based on time series. The training set is over-sampled using an adaptive synthetic oversampling method. Then, by jointly considering the contribution of features to classification and the correlation between features, a risk prediction method ground on ensemble learning is established.

RESULTS: From the results, the accuracy of predicting 5 risk categories increased by 7.00%, 1.10%, 2.20%, 2.30%, and 0.60%, respectively, reducing the features from the original 118 columns to 60 columns and reducing the data dimension by 49.00%. Compared with traditional models, the accuracy was 98.61%, and the overall accuracy was improved by 0.60%.

CONCLUSION: This risk prediction scheme can quickly and accurately predict the risk categories that affect its operation. It has high prediction accuracy and fast speed than other algorithms. This research can provide strong assistance for security decision-making in the power Internet of Things.

Reinforcement Learning Data-Driven Optimal Load-Frequency Control for Power Systems

Yi Zhao — 2025-03-04

INTRODUCTION: Power systems are complex due to their time-varying and uncertain parameters, challenging conventional control methods.

OBJECTIVES: This study proposes an adaptive dynamic programming (ADP) controller to address this limitation. The ADP controller eliminates the need for pre-existing knowledge of the system dynamics, a significant advantage in real-world applications.

METHODS: By iteratively solving the Riccati equation using only system state and input data, the controller learns an approximate optimal control strategy. In this study, we use an iterative computational approach with an online adaptive optimal controller designed for unknown power system dynamics.

RESULTS: Utilizing real-time collected system states and input information, even in the absence of knowledge about the power system matrix, we achieve iterative solutions for the algebraic Riccati equation, enabling the computation of an optimal controller. Simulation results demonstrate the ease of implementation of this approach in power system load frequency control (LFC).

CONCLUSION: The proposed ADP controller exhibits good control performance of grid stability, making it a valuable reference for LFC, especially in scenarios with unknown system parameters.

Overcoming Weak Grid Challenges: A Combined Approach to VSI Stability with Impedance Adjustment, Control Optimization, and Microgrid Integration

Harendra Pal Singh — 2025-02-06

This paper addresses the challenges in Voltage Source Inverter (VSI) systems connected to weak grids, where frequent impedance changes lead to instability and power quality issues. This research studies how changing grid impedance affects current distortion and the stability of a VSI. It proposes the stability analysis of a single loop controller and optimize its settings using various techniques (ZN-method, PSO, GA) to ensure VSI stability and meet current distortion limits (THD compliance), when grid impedance varies. The primary focus revolves around addressing two key challenges: managing impedance variations at the PCC and enhancing the tracking performance of the PI controller. The VSI-based system connected to the weak grid and in standalone mode is simulated on Typhoon HIL, to validate the effectiveness of obtained optimized controller parameters by changing various conditions like, the output power regulation and sudden load change in a standalone distribution network. The MATLAB/SIMULINK with m-files is utilized for the parameters optimization and controller model simulation purposes. This research is important for developing more reliable and resilient power systems, specifically by investigating the transient behaviour of VSI frequency and voltage under sudden changes, to ensure an uninterruptible power supply to critical loads.

Power optimization and control strategy for new energy hybrid power generation system based on deep learning

Fei Li — 2025-04-29

The continuous shortage of non-renewable energy and the increasingly serious environmental pollution have made clean renewable energy represented by wind and solar energy become the focus of attention. This paper mainly studies the power optimization and control strategy of a new energy hybrid power generation system based on deep learning. This paper introduces the basic principle and structure of a new energy hybrid power generation system and the application of deep learning technology in power optimization and control strategy. In this paper, a power optimization method based on deep learning is proposed, which realizes real-time optimization of power generation system powers by training neural network models. A control strategy based on deep learning is designed to improve the stability and efficiency of the power generation system. The effectiveness of the proposed method in practical application is verified by experiments.

Research on Fault Diagnosis Method for Photovoltaic Array Based on XGBoost Algorithm

Zongyu Zhang — 2024-11-19

INTRODUCTION: Photovoltaic (PV) energy sources frequently experience issues, including fragmentation, open-circuit, short-circuiting, and other common and hazardous problems. The current focus of PV research is on fault detection within solar arrays. Traditional models encounter challenges in identifying errors due to uncertainties in panel settings and the complex nature of the actual PV structure.

OBJECTIVES: This study aims to introduce a novel Extreme Gradient Boosting (XGBoost) approach for fault diagnosis in PV arrays.

METHODS: The XGBoost algorithm is trained using collected PV array defect data samples. Data preprocessing is performed to manage missing values and remove noisy data. Feature extraction is conducted using Linear Discriminant Analysis (LDA) to improve detection accuracy. To further enhance XGBoost’s performance, the World Cup Optimization (WCO) approach is applied to select optimal features from the extracted data. Fault detection is then conducted using the XGBoost algorithm on the processed data. Various indicators are utilized for performance assessment within the Python environment.

RESULTS: The comparative analysis demonstrates that this research improves fault detection efficiency in PV arrays compared to existing methodologies.

CONCLUSION: The study presents an effective method for enhancing fault detection in PV systems, showcasing the advantages of the XGBoost and WCO-based approach over conventional methods.

Research on intelligent detection method of new energy vehicle power battery based on improved ViBe algorithm

Lei Gu — 2025-03-11

Background: Traditional foreground detection methods for new energy vehicles using the ViBe algorithm often suffer from ghosting effects, which can obscure the accurate detection of moving targets.

Aims: This study enhances foreground detection accuracy by addressing ghosting issues in the ViBe algorithm and improving the battery pack state detection system for new energy vehicles.

Method: The method includes analyzing global light changes before foreground detection and updating the background model using the three-frame difference method. The system integrates hardware and software to process data with the ViBe algorithm, measuring voltage from twelve 18650-type lithium batteries.

Results: The battery management system prototype exhibits an absolute measurement error within -1.2 mV compared to the high-precision multimeter. The system maintains measurement accuracy across varying temperatures, demonstrating effective environmental adaptability.

Conclusion: The enhanced system successfully reduces ghosting in foreground detection and provides reliable battery state monitoring. It is robust under extreme conditions, contributing to improved diagnostic capabilities and enhanced traffic safety.

Research on a New Maximum Power Tracking Algorithm for Photovoltaic Power Generation Systems

Lei Shi — 2024-11-19

INTRODUCTION: Significant advances have been made in photovoltaic (PV) systems, resulting in the development of new Maximum Power Point Tracking (MPPT) methods. The output of PV systems is heavily influenced by the varying performance of solar-facing PV panels under different weather conditions. Partial shading (PS) conditions pose additional challenges, leading to multiple peaks in the power-voltage (P-V) curve and reduced output power. Therefore, controlling MPPT under partial shading conditions is a complex task.

OBJECTIVES: This study aims to introduce a novel MMPT algorithm based on the ant colony incorporated bald eagle search optimization (AC-BESO) method to enhance the efficiency of PV systems.

METHODS: The effectiveness of the proposed MPPT algorithm was established through a series of experiments using MATLAB software, tested under various levels of solar irradiance.

RESULTS: Compared to existing methods, the proposed AC-BESO algorithm stands out for its simplicity in implementation and reduced computational complexity. Furthermore, its tracking performance surpasses that of conventional methods, as validated through comparative analyses.

CONCLUSION: This study confirms the efficacy of the AC-BESO method over traditional strategies. It serves as a framework for selecting an MPPT approach when designing PV systems.

Harmonic Measurement Algorithm of Power System Integrating Wavelet Transform and Deep Learning

Hanshu Jiang — 2024-12-05

INTRODUCTION: The problem of low accuracy in harmonic measurement is a significant challenge in power systems. Traditional methods often exhibit higher measurement errors, leading to unreliable detection of harmonics. To address this, the author proposes a new approach that integrates wavelet transform and deep learning techniques for enhanced harmonic measurement accuracy.

OBJECTIVES: The primary goal of this study is to develop a more accurate harmonic measurement algorithm by combining full phase fast Fourier transform (FFT) and adaptive neural networks. The research aims to automatically detect power system harmonics with minimal error and improve upon the limitations of traditional methods.

METHODS: The study implemented a harmonic measurement method using full phase FFT integrated with an adaptive neural network. This approach calculates harmonic amplitudes based on the fundamental component and its amplitude, while determining the precise start and end times of harmonics. The system also incorporates mean filtering for automatic detection of harmonics. The effectiveness of the proposed method was evaluated through experiments that compared it to traditional harmonic measurement techniques.

RESULTS: Experimental results demonstrated that the proposed method achieved an average measurement error of 0.02V, with a maximum error of 0.03V, both of which are below the acceptable error limit. In contrast, traditional methods exhibited significantly higher average errors of 3.31V and a maximum error of 5.17V. The new method consistently showed higher accuracy in harmonic detection compared to conventional approaches.

CONCLUSION: The study concludes that the proposed harmonic measurement algorithm significantly improves accuracy compared to traditional methods. With its lower measurement error and effective automatic detection capabilities, the method proves to be highly suitable for harmonic measurement in power systems.

Research on anti-error operation warning of power grid dispatching based on deep bidirectional gated recurrent neural network

Jinglong He — 2025-04-11

To improve the security and overall efficiency of grid scheduling work and accurately optimize scheduling decisions, a grid scheduling error-proof operation warning method based on a deep bidirectional gated recurrent neural network is proposed. This paper combines the principle of hierarchical data construction, summarizes the structured data of metadata operation tickets and maintenance plans of CIM model and OMS network frame model, and constructs the data warehouse of grid dispatching error prevention; based on the natural language processing (NLP) technology, key information and knowledge entities related to grid dispatching error prevention are automatically identified and extracted from the data warehouse. Based on the deep bidirectional gated recurrent neural network, the extracted information sequence is used as input to construct the grid scheduling operation state reconstruction model, and the error prevention warning is carried out according to the output prediction results. The experimental results show that: the data docking speed in different scheduling phases is fast with the fastest speed of 71.254MB/s, and the convergence speed of the analysis and calculation is within 0.01MB/s, indicating that the overall analysis efficiency is high, the application performance is good, and it can determine whether there is any misoperation in the process of grid scheduling and carry out highly efficient, accurate, and fast early warning.

Distributed energy storage hierarchical partition dispatch control of virtual power plant based on SaDE-BBO algorithm

Tianyi Yu — 2025-04-14

To improve the response ability of the virtual power plant during operation and the adjustment ability when the load fluctuates, and ensure its stable operation, a virtual power plant distributed energy storage hierarchical partition dispatch control method based on the SaDE-BBO algorithm is proposed. This method is based on the operation structure of the virtual power plant, analyzes the operating characteristics of the distributed energy storage system and the output of uncertainty factors, considers the grid load, renewable energy and distributed energy storage on the time scale, and constructs hierarchical partitions of the virtual power plant. The dispatch model determines the day-ahead and day-in-day hierarchical partition dispatch control objective functions, and sets corresponding constraints; the dispatch control model based on the solution of the SaDE-BBO algorithm outputs the virtual power plant distributed energy storage hierarchical partition dispatch control optimization plan. The test results show that the maximum load peak value after dispatch control through this method is 40.9 MW; the active power loss results are all below 10 MW, real-time response to control instructions ensures the safety and stability of the voltage of the virtual power plant under the access of renewable energy, and the nodal voltage fluctuated within the permissible range of 0.95 to 1.05 p.u.

Research on short-term power load forecasting based on deep reinforcement learning with multiple intelligences

Tianyun Luo — 2025-04-14

A reliable supply of power systems is critical for industry, commerce, and residential life. Improving the accuracy and reliability of short-term electricity load forecasting plays a crucial role in ensuring the satisfaction of electricity demand and the stable operation of the power system. Therefore, to realize accurate and efficient prediction of short-term power loads, a short-term power load prediction method based on multi-intelligence deep reinforcement learning is proposed to address the complex nonlinear characteristics of load data. In this paper, we analyze the multi-intelligence application architecture in power load forecasting, and analyze the function of each intelligent unit applied to short-term power load forecasting; based on clarifying the interaction relationship of each intelligent unit in short-term power load forecasting, we model short-term power load forecasting as a distributed and partially observable Markov decision-making process, which is suitable for multi-intelligence deep reinforcement learning; based on the MATD3 algorithm, a centralized training-distributed execution framework is used to train multiple intelligences within the model to achieve short-term power load forecasting. The experimental results show that in the August short-term electricity load forecasting using the design method, the obtained MAE value is 35.94 kW, MAPE value is 4.05%, and RMSE value is 32.71 kW. In the short-term power load forecasting evaluation conducted for December, the average absolute error (MAE) value obtained was 36.75 kilowatts, the average absolute percentage error (MAPE) value was 4.51%, and the root mean square error (RMSE) value was 34.82 kilowatts. These evaluation results fully demonstrate that the design method adopted has high prediction accuracy and forecast precision. This method has demonstrated good practical value and broad application prospects in practical applications due to its high-precision prediction performance and strong prediction stability.

Power grid inspection based on multimodal foundation models

Jingbo Hao — 2025-04-14

INTRODUCTION: With the development of large foundation models, power grid inspection is transmitting from traditional deep learning to multimodal foundation models.

OBJECTIVES: This paper aims to boost the application of multimodal foundation models for power grid inspection.

METHODS: Current research on foundation models and multimodal large language models (LLMs) is introduced respectively. Three application forms of multimodal foundation models in power grid inspection are explored. The reliability of these models is discussed as well.

RESULTS: These techniques can significantly reduce the time and cost of inspection by automating the analysis of large amounts of sensor data. They can also improve the accuracy and reliability of inspection by leveraging the understanding and reasoning abilities of LLMs.

CONCLUSION: These advanced techniques have shown great application potential in power grid inspection. But it is important to note that they should not entirely replace human inspectors who can validate automatic findings and address possible issues not captured by these models alone.

Research on multi-stage optimization planning of power internet of things based on seagull optimization algorithm

Peng Ye — 2025-04-15

The Power Internet of Things (PIoT) is a significant technology for realizing the transformation of future energy systems, with the Integrated Energy System (IES) playing a crucial role in realizing the value of PIoT. Traditional IES planning methods typically focus on a single-stage planning approach and involve complex solution models, often resulting in inefficient equipment configurations and resource wastage. This study proposes a multi-stage IES planning method aimed at enhancing both energy efficiency and the economic performance of IES. The method models the IES based on electric, gas, and thermal buses, considering the coupling, storage, and conversion of multiple energy sources. A range of constraints, such as energy coupling, equipment capacity, and energy purchases, are considered. The planning cycle is divided into multiple stages, and an economic model is developed that accounts for both system investment and operating costs. Given the complexity of the multi-stage planning model, the Seagull Optimization Algorithm (SOA) is introduced to solve the problem. The SOA leverages its strong global and local search capabilities to determine the optimal capacity configuration at each stage. The comparison of the single-stage planning method by a calculation example proves the economic advantage of the multi-stage planning scheme and effectiveness of SOA.

Cloud model-based unconventional risk assessment method for flexible distribution system

Shaotao Guo — 2025-04-16

The integration of high-penetration distributed resources has led to increased complexity and uncertainty in the unconventional risks of distribution networks, posing higher demands on the risk assessment of distribution networks. This paper proposes an unconventional risk assessment method for flexible distribution system based on cloud model. Firstly, an unconventional risk assessment system for distribution networks is constructed by considering the probability of unconventional risk occurrence and the severe consequences, and an improved AHP-entropy weight method for index weighting is proposed. Then, the cloud model for risk assessment is used to quantitatively evaluate the risk level of the distribution system. The variable weight cloud model is employed to replace the traditional cloud model to provide risk indicator evaluation information. The inverse cloud generator is used to infer and correct the risk cloud model parameters, and the assessment is completed by comparing with the digital characteristics of the standard cloud model. Finally, the effectiveness of the proposed assessment method is verified through an example analysis of a certain region in China.

Modeling and simulation of control strategy for voltage source converter multi-terminal DC system

Xingchao Zhang — 2025-04-17

An improved strategy for controlling DC voltage droop is proposed to solve the problems of conventional DC voltage droop control strategy, such as the received power of converter stations is distributed according to fixed proportion, the power variation amplitude is large, the power regulation ability is easy to lose when reaching the upper limit, and there is obvious voltage deviation. A three-terminal VSC-MTDC system model including wind farm was constructed by Matlab/Simulink. According to different conditions of power fluctuation caused by wind power output variation, simulation and analysis of VSC active power and DC voltage variation were carried out. Simulation results showed that the improved strategy for controlling DC voltage droop proposed in this paper made the converter station retain more power regulation margin when the converter station absorbed more active power, and could prevent the converter station from losing its power regulation ability when the power reached the upper limit, and realize reasonable power distribution among converter stations. Meanwhile, the proposed control strategy could maintain DC voltage constant, which verified the feasibility of its application in wind power grid-connected flexible DC transmission system.