Electric Power Research Institute

Deep reinforcement learning (DRL), graph neural network (GNN) techniques, and large models are applied to the dynamic path planning of substation equipment operation and maintenance (O and M), aiming to overcome the limitations of static scheduling and rule-based methods. Achieving adaptive optimization under uncertainty, this study employs a Markov decision model to simulate the O and M process. To provide high-fidelity representations of equipment status and network configuration, the framework utilizes graph neural network technology together with a pre-trained large model for the power industry to extract correlation features, semantic information, and topological characteristics from both sensor and unstructured data. An optimized maintenance path is created using a dual-depth Q-network strategy, with the large model providing contextual understanding and decision augmentation to balance cost, efficiency, and risk in complex substation environments. This method enhances maintenance efficiency, reduces operational costs, and strengthens risk management capabilities compared to traditional algorithms and baseline deep learning models. The IEEE 118 node system validated this approach. Results demonstrate its outstanding performance in high-dimensional dynamic scenarios, along with scalability and robustness. This study establishes a framework for intelligent, large-model-driven, data-driven O and M of modern power systems.

The penetration rate of distributed photovoltaic (PV) systems and electric vehicles (EVs) in the distribution network continues to increase, posing challenges to the voltage stability in the distribution area. In response to the high proportion of PV access, inverter regulation is difficult to completely solve the problems of voltage exceeding limits, disorderly charging of EVs during peak hours, node voltage falling below the lower limit and exacerbating load peak valley differences. This paper proposes a collaborative control strategy that integrates PV inverter optimization and orderly charging scheduling of EVs, and uses grey wolf optimization algorithm to solve and verify it. The simulation results of the IEEE 33 node power system show that the proposed strategy is effective. When EVs are charged in an orderly manner with high PV penetration, reasonable power regulation can absorb excess PV power, suppress voltage rise, and alleviate voltage drop at night. The combination of PV inverter regulation can synergistically reduce voltage fluctuations, minimize network losses, and enhance the consumption capacity of renewable energy.

The integration of heat pumps with thermal energy storage (HP-TES) systems is gaining attention as a viable solution for managing peak building demand driven by immense cooling and heating loads. With growing reliance on renewable energy sources, thermal energy storage offers an excellent opportunity to mitigate mismatches in thermal load between energy supply and demand. The use of phase-change material (PCM) TES is especially promising, as PCMs offer significant latent energy storage capacity with smaller temperature glides in smaller volumes compared to other TES technologies. However, challenges arise because current HP-TES architectures can load-shift only cooling or heating, not both, requiring two systems and thus doubling cost, weight, and footprint. Furthermore, current research efforts lack specific tools and techniques to advance integrated systems from concept design to end-user application, focusing on only discharge performance. To address these challenges, this research proposes a dual-mode (heating and cooling) integrated HP-TES system that uses room-temperature PCM-TES as a high-temperature heat source in heating mode and a low-temperature heat sink in cooling mode, thereby reducing temperature lifts and compressor power. Design criteria for PCM-TES heat exchangers were developed, balancing thermal-hydraulic performance with practical constraints such as available building space and weight requirements along with PCM selection considerations such as shipping conditions, moisture exposure, and number of available cycles. A detailed transient model for HP-TES systems was developed to enable rapid annual performance assessments in any US climate zone. Detailed comparisons of HP-TES performance versus a state-of-the-art base heat pump were conducted for all US ASHRAE climate zones. In cooling mode, overall cooling demand reductions in California were around 20%, while Honolulu's tropical climate led to 14% reductions. Heating mode demand reductions were notably higher, especially in very cold-climate regions, ranging from 40% to 65% for Chicago, IL, to Fairbanks, AK, primarily because the new HP-TES system does not require additional backup heating during peak hours. A laboratory-scale HP-TES system was prototyped and tested using a novel test matrix designed to cover a wide range of possible operating modes, e.g., continuous and cyclic heating and cooling modes. Experimental testing showed demand reductions of 0.6%-30% in cooling mode and 40%-60% in heating mode, with the larger reductions observed under more extreme ambient conditions. The experimental data were utilized to validate the transient HP-TES model, which demonstrated excellent agreement in capacity (<4%), power draw (<1%), and compressor suction / discharge conditions. The highest deviations were reported during startup conditions. Finally, a commercialization plan for HP-TES was presented, highlighting key insights on capital expenses and expected location-specific operating cost savings, along with opportunities for California-specific markets. The work presented here provides design guides and procedures to turn such a system from concept to an off-the-shelf product for consumers.

High-voltage circuit breakers (HVCBs) are critical switching devices whose mechanical reliability directly affects power system safety and operational continuity. Accurate fault diagnosis remains challenging due to nonlinear vibration characteristics and the sensitivity of support vector machines (SVMs) to hyperparameter selection. To address this issue, a multi-strategy improved dung beetle optimization–support vector machine (MIDBO–SVM) framework is proposed for vibration-based mechanical fault diagnosis. Frequency-domain features are extracted from vibration signals using the fast Fourier transform to characterize fault-related spectral variations. A multi-strategy improved dung beetle optimization (MIDBO) algorithm incorporating chaotic initialization, adaptive search regulation, and mutation enhancement is developed to improve population diversity, global exploration, and convergence stability. The optimized MIDBO is used to determine the penalty and kernel parameters of the SVM, constructing a robust and well-generalized diagnostic model. Experimental results show that MIDBO–SVM achieves a diagnostic accuracy of 96.67%, outperforming conventional SVM (86.25%) and random forest (89.17%). The proposed method also demonstrates faster convergence and maintains accuracy above 86% under imbalanced sample conditions, confirming its robustness and generalization capability. These advantages contribute to more reliable mechanical condition assessment and improved maintenance decision support for HVCBs.

To mitigate the substantial energy footprint of data centers and exploit their flexibility for low-carbon power scheduling, a low-carbon economic dispatch scheme is developed for virtual power plants (VPPs) that explicitly models the spatiotemporal distribution of data-center workloads. First, the flexibility of data center workload adjustment in temporal shifting and spatial migration is characterized and exploited, and the “source-load-storage” coordinated scheduling architecture, including wind power generation, photovoltaic power generation, energy storage systems and data center, is constructed. Second, the carbon trading mechanism and tenant satisfaction function are introduced to minimize the total operating cost, composed of energy storage, electricity transaction, and carbon trading costs. Considering the tenant satisfaction constraint, power-balance constraint, energy storage system constraint, and electricity transaction constraint, the VPP optimal scheduling model, including data center, is established. Finally, an example is analyzed based on the Institute of Electrical and Electronics Engineers 118-bus system and the operation data of an actual data center in Gansu Province. Simulation studies validate that the scheme leverages workload spatiotemporal shifting to curb system-level carbon emissions and total operating expenditure, while maintaining tenant quality of service and markedly enhancing renewable-energy utilization.

In order to realize the controllable loading of electromagnetic force in electromagnetic forming and improve the forming effect of tube, this paper puts forward an electromagnetic forming technology of tube based on double magnetic field converters. In this technology, two magnetic field converters are introduced between the tube and the driving coil, and the electromagnetic force is regulated by changing the geometric parameters and spacing of the magnetic field converters. Based on the specific 2024- aluminum alloy tube, the simulation research is carried out. The simulation data show that the change of the geometric structure and spacing of the two magnetic field converters can change the induced eddy current and realize the regulation of electromagnetic force. Compared with the traditional electromagnetic forming of tube, the uniform forming range Dr of electromagnetic forming of Dr based on double magnetic field converters is increased by 2.77 times. To sum up, the electromagnetic forming technology based on double magnetic field converters can effectively improve the forming uniformity of tube and promote the wide application of electromagnetic forming technology in industry.