The method of Support Vector Machine (SVM) based on Dissolved Gas Analysis (DGA) has been studied in the field of power transformer fault diagnosis. However, there are still some shortcomings, such as the fuzzy boundaries of DGA data, and SVM parameters are difficult to determine. Therefore, this paper proposes a power transformer fault diagnosis method based on Kernel Principal Component Analysis (KPCA) and a hybrid improved Seagull Optimization Algorithm to optimize the SVM (TISOA-SVM). Firstly, KPCA is used to extract features from DGA feature quantities. In addition, TISOA is further proposed to optimize the SVM parameters to build the optimal diagnosis model based on SVM. For the SOA, three improvement methods are proposed. An improved tent map is used to replace the original population initialization to improve population diversity. In addition, the nonlinear inertia weight and random double helix formula are proposed to improve the optimization accuracy and efficiency of the SOA. Then, benchmarking functions are used to test the optimization performance of TISOA and six algorithms, and the results show that TISOA has the best optimization accuracy and convergence speed. Finally, the fault diagnosis method based on KPCA and TISOA-SVM is obtained, and it is noteworthy that three examples are tested to verify the diagnostic performance of the proposed method. These results show that the proposed method has higher diagnostic accuracy, shorter diagnosis time, stronger significance and validity than other methods. Therefore, a research idea is provided for solving practical engineering problems in the field of fault diagnosis.
26 January 2022 in IET Generation, Transmission & Distribution
In recent years, support vector machine has been widely used in the field of transformer fault diagnosis, but its fault diagnosis performance is still to be improved. A power transformer fault diagnosis model based on hybrid improved grey wolf optimization optimized least squares‐support vector machine will be proposed as an attempt to improve the efficiency and accuracy of its fault diagnosis performance. First, the hybrid improvement method will be introduced with a view to solving the three problems that grey wolf optimization is troubled with, that is, premature convergence, slow convergence, and low population diversity. Then, six benchmark functions and Wilcoxon rank sum test are used to test hybrid improved grey wolf optimization and the other optimization algorithms, and the results show that hybrid improved grey wolf optimization has better optimization performance and significance. Finally, dissolved gas analysis data will be used for example simulation which will be further examined from three aspects in order to testify the fault diagnosis performance of the proposed method. The results show that the validity of the proposed method is much higher than that of the other models. Therefore, it can be concluded that the proposed method can greatly improve the performance of transformer fault diagnosis, and has great practical significance in engineering.
In general, microgrids have a high renewable energy abandonment rate and high grid construction and operation costs. To improve the microgrid renewable energy utilization rate, the economic advantages, and environmental safety of power grid operation, we propose a hybrid energy storage capacity optimization method for a wind–solar–diesel grid-connected microgrid system, based on an augmented ε- constraint method. First, the battery is coupled with a seasonal hydrogen energy storage system to establish a hybrid energy storage model that avoids the shortcomings of traditional microgrid systems, such as a single energy storage mode and a small capacity. Second, by considering the comprehensive cost and carbon emissions of the power grid within the planning period as the objective function, the abandonment rate of renewable energy as the evaluation index, and the electric energy storage and seasonal hydrogen energy storage system operating conditions as the main constraints, the capacity allocation model of the microgrid can be constructed. Finally, an augmented ε- constraint method is implemented to optimize the model above; the entropy–TOPSIS method is used to select the configuration scheme. By comparative analysis, the results show that the optimization method can effectively improve the local absorption rate of wind and solar radiation, and significantly reduce the carbon emissions of microgrids.
2 April 2026 in IEEE Transactions on Industrial Electronics
To address the drift of equivalent parameters in lithium-ion batteries during operation due to varying operating conditions and aging, which degrades model adaptability, and the difficulty of accurately tracking multiple coupled states simultaneously, this article proposes a digital twin–driven joint state estimation method to enable coordinated updating of electrothermal parameters and multiple states in a battery pack. First, an online-oriented digital twin–driven joint state estimation framework is established, where real-time measured voltage, current, and temperature data are used for model-parameter updating and state correction, ensuring real-time consistency between the twin model and the physical counterpart. Second, an embeddable electrothermal coupled equivalent model is developed, and an online parameter identification strategy based on an improved metaheuristic algorithm is proposed. By using a sliding window and a parameter-variation penalty term, parameter fluctuations induced by noise are suppressed, thereby improving the tracking performance of time-varying parameters. Finally, a multitime-scale joint state estimation algorithm is designed to achieve coordinated state of charge (SOC)–state of health (SOH) updating at the sampling scale and the cycle scale, enabling accurate joint estimation under multistate coupling. Experimental results show that the proposed method can track parameter variations more stably and improve the accuracy of joint state estimation, while the computational time satisfies the requirements for online operation, providing theoretical and technical support for online battery state sensing and management.
20 November 2025 in IEEE Journal of Emerging and Selected Topics in Power Electronics
In high-voltage brushless DC (BLDC) generator systems for aerospace applications, significant coupling exists between the main generator and the excitation subsystem. During operation, this coupling is highly susceptible to variations in temperature and magnetic saturation, resulting in parameter drift that adversely affects the accuracy of impedance modeling, thereby compromising system stability analysis and control performance. To address these challenges, this paper proposes a frequency-domain impedance modeling and decoupled parameter identification method tailored for strongly coupled high-voltage DC (HVDC) generators. A two-port impedance model is constructed to accurately characterize the dynamic frequency-domain coupling between subsystems. By leveraging impedance sensitivity spectra, a frequency-band decoupling strategy is designed to explicitly identify the characteristic coupling intervals between the main generator and the exciter, thereby enhancing the observability and identifiability of the coupled parameters. To mitigate the degradation in identification accuracy caused by strong multi-parameter coupling, a multi-stage hierarchical identification strategy based on an improved slime mold optimization algorithm is developed. This approach enables hierarchical convergence within the parameter space and facilitates refined local identification. Experimental results demonstrate that, compared with conventional identification methods, the proposed strategy reduces the maximum identification error by 7.5% and improves accuracy by 5.2% over other optimization algorithms. Furthermore, the method effectively compensates for parameter drift caused by thermal variations and magnetic saturation, maintaining the output impedance magnitude error within ±2 dB. These results confirm the enhanced accuracy and engineering applicability of the proposed impedance modeling approach for aerospace high-voltage BLDC generator systems.
28 April 2026 in IEEE Transactions on Transportation Electrification
Conventional energy systems can hardly satisfy the operational requirements of hypersonic vehicles under high Mach numbers and high-altitude conditions. To investigate the system-level performance optimization and evaluation methods for hypersonic vehicle energy systems, this study takes the Ram Air Turbine Power System (RATPS) as a typical research object and constructs an integrated optimization and evaluation framework by establishing the coupling relationship between key design parameters and performance penalty (PFP). Firstly, PFP theory is combined with the physical characteristics of RATPS to establish a bi-objective model aiming at minimizing PFP of bleed aerodynamic drag and system mass, thereby realizing the quantitative mapping between design variables and system-level performance compensation. Then, to address strong nonlinearity of PFP model and computational burden caused by multi-objective optimization, a surrogate model based on deep neural network (DNN) is constructed, and evolutionary-neural architecture search (ENAS) is adopted to optimize its hyperparameters, and a dynamic adaptive dual-population cooperative algorithm (DADPCA) is adopted to efficiently solve the Pareto frontier. Meanwhile, Robust Nash Compromise Model (RNCM) decision-making model is introduced to identify compromise solutions objectively and stably. Finally, multi-objective optimization and evaluation are carried out under four representative flight profiles. The results show that the proposed framework can efficiently obtain reasonable compromise design parameters, reveal the nonlinear sensitivity laws of key design variables under different cruise conditions, and achieve a significant reduction in the total PFP compared with the baseline design.