In this paper, we propose a distributed online voltage control algorithm for distribution networks with multiple photovoltaic (PV) systems based on dual ascent method. Conventional distributed algorithms implement voltage control only when the algorithms converge. However, our proposed algorithm is able to carry out voltage control immediately. In particular, we derive a closed-form solution for PV controllers to locally update the active and reactive power set-points aiming at minimizing the total loss and maintaining bus voltages within the acceptable ranges. The optimality is guaranteed and the convergence is established analytically. Moreover, our proposed algorithm only requires the information exchange between neighboring PV systems, thus reducing communication complexity. Finally, numerical tests on IEEE 37-bus distribution system verify the effectiveness and robustness of our proposed algorithm.

High penetration of wind energy in the modern power system exposes the need of smoothing the fluctuating output power in an effective and conducive way. In this context, this paper proposes two novel control strategies that utilize the self-capability of permanent magnet synchronous generator based wind turbine to realize power smoothing. The first strategy pursues to offer power smoothing support via simultaneous utilization of DC-link voltage control, rotor speed control and pitch angle control. The second control strategy seeks to coordinate the three concerned individual control schemes in a hierarchical manner, where the power smoothing tasks are allocated to individual control modules or their combinations dynamically in line with WT's operation status. Both two strategies are able to provide power smoothing support by fully exploiting wind turbine's self-capability, whereas the second strategy has the merits on 1) reducing the activation frequency of pitch angle control, and 2) enhancing wind energy harvesting. Case studies of the proposed control strategies are carried out to compare and verify their effectiveness in achieving power smoothing.

The identification of the distribution system topology is the key concern in distribution system state estimation and the precondition for its energy management. However, lacking sufficient measurement devices, full-scale identification of entire distribution grid can hardly be achievable in practice. The frequent topology changes in distribution systems impose challenges for topology identification. This paper proposes a novel topology identification method by deeply mining the data obtained from gird terminals and smart meters at end-users premises. The proposed method starts with data processing, followed by nodal correlation analysis and topology modeling based on the Markov Random Field (MRF) method, where the pseudo-likelihood method and L2 regularization theory are introduced to improve the computation efficiency while preventing the over-fitting problem. Then the iterative screening method is developed to generate the distribution system topology of medium/low-voltage distribution systems. Finally, the incremental learning and parallel programming models are proposed to implement the algorithms on single/multi-terminal. The effectiveness of the proposed model is validated on IEEE 33-node, IEEE 123-node and actual distribution systems.

The universal use of open communication networks in interconnected power systems has changed the constant delay in the formerly dedicated channels into time-varying delay. The delay will deteriorate the dynamic performance of the load frequency control and lead to system frequency instability. This paper presents a fractional-order sliding mode controller for interconnected power systems with time-varying delay and performs stability analysis based on the Lyapunov direct method. Firstly, fractional-order integral sliding mode control is used to improve the stability of interconnected power systems containing time-varying delay. This expands the delay margin by taking advantage of the scalability of the fractional-order to the sliding mode surface. Then, the Riemann-Liouville fractional-order Lyapunov functional is introduced to stability analysis. This reduced the conservatism of the stability analysis by adjusting the upper limit of the functional integral. Lastly, combining Lyapunov's theory and integral inequality to obtain the stability conditions of the time-varying delay power system described by linear matrix inequality, and bring the stability margin of the system. The simulation results show that the designed controller can guarantee that the maximum delay time of the system is less than a predefined upper limit. The proposed control strategy obtains a larger stability margin than the other four scenarios. Additional simulation study was also conducted to verify its robustness to load variations and wind fluctuations.

Due to the limited presence of monitoring and measurement devices, timely identification of distribution grid topology has been challenging. Therefore, this article proposes a power grid topological generative adversarial network (Gridtopo-GAN) model to identify the distribution grid topology of either meshed or radial structure with limited measurements. By leveraging the topology preserved node embedding architecture, this model can efficiently handle large-scale systems with different topological configurations. Because of the generative capability of GAN, the model is robust enough when fed with bad measurement data, including missing data, commonly encountered in practical applications. Numerical simulations are carried out on the IEEE 33-node system, 118-node, 415-node, and real 76-node distribution systems to demonstrate the effectiveness and efficiency of the proposed topology identification model.

Distributed generation (DG) can locally provide energies in distribution network and thus help to reduce the negative impact caused by large-scale electric vehicle (EV) charging demand. This paper proposes a two-stage robust distributed generation investment planning model capable of accommodating uncertainty of EV charging demand and renewable generations. The DG investment location and size are optimized in the first stage and the distribution network operation feasibility in the worst-case scenario is checked in the second stage to avoid constraint violations. The EV charging demand uncertainty is modeled by sampling from real-life scenarios of travel behavior. Finally, the column-and-constraint generation (C&CG) algorithm is adopted to solve the proposed problem. Simulations on modified IEEE 123-node distribution network demonstrate the effectiveness of the proposed model.