[Objective] To realize coordinated source-storage-grid planning for expressway charging energy supply systems under intelligent connected environments， and address the spatiotemporal uncertainty of charging loads induced by mixed traffic flow as well as the challenge of coordinated configuration of photovoltaic （PV）-storage capacity and DC corridor topology under divergent resource endowments and grid-connection conditions along roadside stations， this paper proposes a coordinated planning method of PV-storage DC corridors for expressway main and substations considering mixed traffic equilibrium. [Methods] First， a semi-dynamic mixed traffic equilibrium model is developed. A modified Bureau of Public Roads （BPR） road network impedance function is adopted to characterize the difference in link travel cost between human-driven vehicles （HDVs） and connected and autonomous vehicles （CAVs）. Stochastic user equilibrium and system optimum are used to describe the route choices of HDVs and CAVs， respectively. A residual flow mechanism is introduced to depict the temporal evolution of traffic states. On this basis， the mapping from link traffic flow to nodal charging load is established to generate spatiotemporal charging load profiles across multiple typical days. A coordinated energy supply architecture with main and substations equipped with PV and storage is then proposed. A joint planning model for PV-storage capacity and DC corridor topology is formulated to minimize the total annualized cost including capital investment， operation and maintenance cost， and weighted operational cost over multiple typical days. Improved spanning forest constraints are introduced to guarantee the connectivity of substations and the radial topology of the network. Piecewise linearization is applied to convert the model into a mixed-integer linear programming （MILP） problem for solution. [Results] A 25-node expressway network is adopted for case verification. The optimal plan constructs 21 DC corridors and forms 4 independent connected zones. It guarantees power supply accessibility of substations while preventing excessive interconnection. Compared with the fully interconnected scheme and independent operation scheme， the total cost is reduced by 5.26% and 2.17%， respectively， without load shedding. [Conclusions] The integration of charging load generation under mixed traffic equilibrium and energy network optimization for heterogeneous expressway stations facilitates source-load complementarity among stations and achieves a desirable trade-off between load supply capability and grid investment. The proposed method provides viable modeling approaches and decision support for the coordinated development of PV-storage-charging infrastructure and DC corridors along expressways.

[Objective] To address the limitation that local optimization search methods for alternating-current optimal power flow （AC-OPF） generally converge only to local optima and struggle to obtain higher-quality feasible solutions， this paper investigates a method that can deterministically escape from an obtained local optimum and continue searching for better solutions. [Methods] An augmented quotient gradient system （AQGS） bifurcation-based superior solution search method （ABS） was proposed based on the bifurcation behavior of the AQGS. By analyzing the bifurcation-induced disappearance mechanism of stable equilibrium points corresponding to suboptimal solutions in AC-OPF， this study revealed the influence of penalty-parameter adjustment on the equilibrium-point structure of the AQGS. Furthermore， the saddle-node bifurcation between a stable equilibrium point and a type-1 unstable equilibrium point was exploited to eliminate the stable equilibrium point corresponding to the current local optimum， thereby guiding the system trajectory to converge to a new optimal solution. [Results] The proposed method was validated on multi-solution test systems with 2， 3， 9， and 300 buses. Numerical results show that ABS successfully obtains feasible local optima with lower objective function values from given local optima in all tested cases， achieving maximum objective-function improvements of 2.92%， 0.21%， 27.60%， and 2.46%， respectively. [Conclusions] The proposed ABS method uses the bifurcation behavior of the AQGS to achieve deterministic escape from local optima and improve solution quality. It provides a new analytical perspective and computational method for multi-solution search and local optimum improvement in AC-OPF， verifying its effectiveness in complex multi-solution power flow optimization problems.

[Objective] To address power flow congestion and cost allocation problems in distribution networks caused by peer-to-peer （P2P） energy trading among multiple microgrids （MG）， and to promote the local consumption of distributed energy resources， a coordinated scheduling model considering dynamic network usage fees and power flow congestion is proposed. [Methods] First， a quantitative method for determining dynamic network usage fees is developed. The proposed fee consists of an electrical distance cost， a voltage management cost， and a network loss cost， thereby accounting for both the network resources occupied by P2P transactions and their effects on distribution network operation. Second， a bilevel coordinated scheduling framework involving the distribution system operator （DSO） and multiple MGs is established. At the upper level， the DSO performs power flow security verification for the initial P2P transaction schedules submitted by the MGs. When line power flow or nodal voltage limits are violated， a congestion management mechanism that minimizes the penalty associated with transaction curtailment is implemented to adjust the P2P transactions. At the lower level， a multi-MG scheduling model incorporating the dynamic network usage fees is formulated to coordinate internal resource scheduling and energy transactions among the MGs. Finally， the Karush-Kuhn-Tucker conditions， strong duality theory， and the Fortuny-Amat transformation are employed to reformulate the bilevel model as a tractable single-level mixed-integer second-order cone programming problem. [Results] Simulation studies based on a modified IEEE 33-bus system demonstrate that the proposed model can promote the local consumption of distributed energy resources， reduce distribution network losses and overall social costs， and effectively mitigate nodal voltage limit violations. The dynamic network usage fees can reasonably reflect the effects of P2P energy trading among multiple MGs on distribution network operation. [Conclusions] The proposed coordinated scheduling model therefore improves the economic efficiency of multi-MG energy trading while maintaining the secure operation of the distribution network.

[Objective] To address the challenges of extracting non-stationary features from fault signals at the Point of Common Coupling in active distribution networks， and to overcome the redundancy of existing deep learning models that hinders real-time deployment on edge intelligent terminals， a fault diagnosis method based on a lightweight time-frequency dual-branch pyramid attention residual network （TF-PARNet） is proposed. [Methods] First， a time-frequency dual-branch parallel architecture was constructed， incorporating a differentiated multi-head self-attention mechanism to capture transient waveform mutations and global spectral dependencies， respectively. Second， an intelligent adaptive fusion module based on a gating mechanism was designed to achieve dynamic weighting and recalibration of heterogeneous time-frequency features. Finally， the pyramid residual unit was reconstructed using depthwise separable convolutions to decouple spatial and channel correlations， achieving extreme model compression. [Results] Simulations were conducted on an IEEE 13-node system using the PSCAD/EMTDC platform. The test results demonstrate that under complex working conditions with superimposed Gaussian noise and temporal offsets， TF-PARNet achieves a mean accuracy of 96.14%. The model contains only 151.0 × 10³ parameters， and the single-sample inference time in a pure CPU environment is as low as 3.8 ms. [Conclusion] The proposed method reduces computational overhead while ensuring high-fidelity feature analysis. It combines high diagnostic accuracy with edge computing efficiency， providing a reliable engineering solution for the localized intelligent protection of resource-constrained distribution networks.

[Objective] As a critical national infrastructure， power systems are vital to national security and socioeconomic development， and their secure and stable operation is of great significance. In recent years， the international security environment has become increasingly complex， and unconventional extreme risks such as external strikes have posed growing threats to power systems. Nevertheless， few studies have focused on characterizing such extreme risks and developing targeted defensive countermeasures. Accordingly， this paper proposes a scenario deduction method for attack-defense games of power systems facing unconventional extreme events. [Methods] First， models for node attack failure and optimal dispatch response are constructed. A coupled node load-centrality index is adopted to screen candidate critical nodes so as to reduce the search dimension of attack-defense strategies. Second， multiple scenarios with different attack scales are established to account for the uncertainty of attack resources. On this basis， a sequential attack-defense-system-response game model is formulated and solved via the column-and-constraint generation （C&CG） algorithm. Finally， simulation verifications are carried out on the IEEE 30-bus and IEEE 57-bus test systems. [Results] The results show that for the IEEE 30-bus system， the optimal defended nodes derived from the attack-defense game are {8，12}， and the expected weighted load shedding is reduced by 24.51% compared with the no-defense scenario. For the IEEE 57-bus system， the optimal defended nodes are {12，8，9，1}， and the expected weighted load shedding is reduced by 77.46% relative to the no-defense scenario. [Conclusions] The proposed method can comprehensively take into account attack-target selection， attack-resource uncertainty and system dispatch response. It helps identify critical nodes with high protection benefits under limited defense resources， and provides a basis for risk scenario deduction and defense-resource allocation of power systems under unconventional extreme events.

[Objective] To address the lack of a unified quantitative criterion for evaluating supplementary damping control schemes in renewable energy units and the difficulty of jointly selecting feedback signals and injection locations， a strategy selection method based on the damping reshaping effect index （DREI） is proposed. [Methods] Candidate control combinations were constructed by combining voltage， current， power， DC-link voltage and phase-locked loop （PLL） angular-frequency signals with typical injection locations in the converter control loops. A positive-sequence impedance model considering machine-side dynamics was established through dq-domain impedance modeling and sequence transformation. DREI was defined as the integral of the negative region of the real part of the positive-sequence impedance， and was used to quantify the ability of each control combination to weaken negative damping and to rank candidate schemes. [Results] Impedance analysis， Nyquist stability analysis and time-domain simulation show that the impedance reshaping effect varies significantly with feedback signal and injection location. The scheme using the grid-side q-axis current as the feedback signal and injecting it into the q-axis current inner loop achieves the minimum DREI and provides the best negative-damping suppression and stability-margin improvement. [Conclusions] DREI can uniformly characterize the impedance reshaping capability of supplementary damping control over the full frequency range， providing a quantitative basis for supplementary damping strategy design and wideband oscillation stability analysis of renewable energy units.

[Objective] To address severe net load fluctuations， high renewable curtailment， and voltage deviations in active distribution networks caused by penetration of distributed generation （DG）， this paper proposes a hierarchical coordinated multi-objective optimization method that integrates energy storage system （ESS） scheduling with active distribution network operational optimization. [Methods] A hierarchical optimization framework based on an iterative coordination mechanism was established. In the upper layer， a composite objective considering ESS economic benefits and net load peak-shaving performance was formulated. A dynamic weighting adjustment mechanism based on load fluctuation characteristics was introduced， and an improved whale optimization algorithm （IWOA） was employed to determine the day-ahead charging and discharging strategies of the ESS. In the lower layer， a multi-objective model aimed at minimizing network power losses， node voltage deviations， and renewable energy curtailment was constructed. An improved multi-objective particle swarm optimization （IMOPSO） algorithm was adopted to coordinate reactive power compensation devices， distributed generation outputs， and ESS power allocation. Coordination between the two layers was achieved through alternating iterations and information feedback. [Results] Simulation results on an improved IEEE 33-node active distribution network system using MATLAB demonstrate that the proposed method yields a daily energy storage revenue of RMB 427.4 and a peak-shaving and valley-filling rate of 15.1%. It reduces total network loss to 1，796.3 kWh， total renewable energy curtailment to 1，725.2 kWh， and the sum of squared voltage limit violations to 0.001， thereby effectively alleviating voltage deviations. The method exhibits strong adaptability and outstanding comprehensive benefits under diverse load fluctuation conditions. [Conclusions] The proposed hierarchical coordinated optimization framework effectively addresses the operational challenges introduced by high DG penetration， significantly enhancing the operational efficiency and renewable energy accommodation capability in active distribution networks. The method features satisfactory computational efficiency and practical applicability， providing robust technical support for optimal operation under complex conditions.

[Objective] The large-scale integration of high-proportion renewable energy into power systems has intensified seasonal variations in system net load. The joint clearing mechanism of the electricity spot energy market and ancillary service market further exacerbates market uncertainty， imposing substantial tail risks on the bidding decision-making of pumped storage power plants. To address this issue， this paper proposes a bidding framework for pumped storage power plants that adapts to multi-seasonal operational scenarios and fully considers market risk constraints. [Methods] First， three typical seasonal scenarios， including the spring-autumn transition season， summer， and winter， are classified based on the evolutionary characteristics of system net load. Second， a novel seasonal joint strategy index is constructed by integrating the arbitrage space characterized by the net load peak-valley difference and the tail risk characteristics of operational revenue. A 15-minute resolution bi-level bidding model is established for pumped storage power plants participating in the integrated spot market. The lower-level model replicates the joint clearing process of the electricity spot market， while the upper-level model embeds the proposed seasonal joint strategy index into the profit function and constraint system， so as to characterize the differentiated bidding behaviors of power plants under diverse seasonal scenarios. [Results] The proposed method can adaptively optimize the capacity allocation between the energy market and ancillary service market according to seasonal net load profiles and dynamic tail risks， effectively improving the tail profit level of pumped storage power plants. Moreover， the derived risk-return frontier enables adjustable bidding strategies to accommodate the diverse risk preferences of market participants. [Conclusions] The incorporation of the seasonal joint strategy index and the corresponding constraint response mechanism in the developed model provides robust technical support for the optimal bidding of pumped storage power plants in the complex integrated electricity spot market.

[Objective] In the context of coordinated medium- and long-term and spot electricity markets， clean energy producers with hybrid pumped-storage plants face the challenge that maintenance arrangements and spot market risks jointly compress their profit margins. To fully exploit the regulation capability of hybrid pumped-storage plants， a multi-time-scale rolling scheduling method for clean energy producers with hybrid pumped storage considering the dynamic adjustment of medium- and long-term water levels is proposed. [Methods] The upper-level model simulated the annual medium- and long-term market and took the week as the optimization time scale. Considering the remaining contract energy and unit maintenance arrangements， it maximized the comprehensive revenue of the clean energy producer over the remaining scheduling horizon， determining weekly contract energy， spot energy， maintenance plans， and weekly terminal reference water levels. The lower-level model simulated the intra-week spot market with an hourly optimization time scale. Based on the boundaries determined by the upper-level scheduling plan， wind power output and inflow uncertainties were considered， and conditional value-at-risk was introduced to maximize the comprehensive intra-week electricity sales revenue， thereby optimizing the intra-week bidding curve and the terminal water level of the hybrid pumped-storage plant. The actual contract completion， actual terminal water level， and executed maintenance status obtained from the lower-level model were fed back to subsequent rolling periods to update the remaining contract energy， weekly terminal reference water levels， and remaining maintenance tasks. [Results] Case-study results show that all three scheduling scenarios satisfy annual contract fulfillment requirement. Compared with the traditional static scheduling method with fixed maintenance plans and fixed weekly terminal reference water levels， the proposed method increases spot energy completion by 10.23% and net revenue by 6.57%. [Conclusions] The proposed method improves the water-level coordination between medium- and long-term scheduling plans and intra-week spot market operation through the rolling correction of weekly terminal reference water levels. Furthermore， it reduces capacity losses in high-value operating weeks through the rolling optimization of maintenance windows， thereby enhancing the spot-market generation support capability and revenue level of clean energy producers with hybrid pumped storage.

[Objective] Driven by dual-carbon goals and an 80% green electricity requirement， data centers are transitioning from passive loads to flexible resources. However， a unified framework is still lacking to characterize the impact of flexible computing workloads on system planning and operation， as well as to quantitatively compare them with physical energy storage. This study develops an equivalent energy storage characterization method for flexible workloads to quantify their value in renewable energy accommodation and system regulation. [Methods] An integrated planning and operation optimization model for a data center-power grid-energy storage system is formulated across 8760 h of a year. Aiming to minimize the annualized total cost， the model jointly optimizes wind， photovoltaic， and energy storage capacities alongside workload shifting， subject to constraints on green electricity share， power balance， and workload service requirements. Furthermore， a renewable-energy-accommodation equivalent energy storage metric is proposed to evaluate the substitution effect of flexible computing workloads for physical energy storage in terms of equivalent capacity and unit economic value. [Results] For a data center with a maximum load of 0.36 MW， under a 1 h flexibility window and a 2 h storage benchmark， increasing the flexible workload share from 0% to 40% yields an equivalent storage capacity of 0.13 MWh with a unit value of 142 600 yuan/MWh. When the flexible workload share reaches 100%， the equivalent capacity increases to 0.21 MWh， while its unit value decreases to 126 700 yuan/MWh， indicating diminishing marginal returns. Sensitivity analysis reveals that reducing delay compensation and widening the flexibility window can enhance this substitution effect. In terms of storage duration， flexible computing workloads tend to substitute for long-duration energy storage at the system level. [Conclusions] The proposed metric provides a unified energy and economic basis for evaluating flexible loads against physical energy storage. It supports coordinated generation-load-storage planning and flexibility valuation for data centers under high wind and solar penetration.