Globally, the installed capacity of photovoltaic (PV) power plants are undergoing rapid growth. The fluctuation of PV output power caused by changes of meteorological factors will significantly affect the reliability and stability of power grid. This paper proposes an inverter-data-driven method to achieve the second-level PV power forecasting. In specific, the PV array model will be firstly built to get the training date set according to the target PV plant and then multi-layer feed-forward artificial neural network (ANN) based on error back propagation algorithm is applied to establish the offline mapping relations from shading conditions of PV array to its corresponding output power. Combining with output power of neighboring PV arrays as well as their geographical location information, the shading conditions of those shaded PV array can be reversely deduced online. Subsequently, the deduced shading conditions of all PV arrays will be converted into one virtual cloud image to identify the instant cloud characteristics. In consequence, the consecutively generated virtual cloud images can depict the moving direction, speed and shape of shading cloud and then help easily realize the second-level power forecasting of PV plant. Case study and experimental results verified the performance of the proposed method.

Grid-forming converters have been identified as an enabling technology in more-electronic power systems. Nevertheless, the reactive power control of grid-forming converters has often been ignored during transients. This article innovatively points out the transient angle and/or voltage instability of grid-forming converters with typical reactive power control schemes, comprising voltage, reactive power droop, and reactive power PI control schemes. We disclose that the grid-forming converter with the voltage, reactive power droop, or other zero-order reactive control schemes faces only the transient angle stability problem yet with a limited capacity for reactive power/power factor regulation. In contrast, the reactive power PI control scheme, as a first-order reactive control scheme, allows for constant reactive power/power factor operation. However, this control scheme or other first-order reactive control schemes may destabilize grid-forming converters through three transient stability problems—transient angle, voltage, and mixed stability problems. Moreover, we disclose the mechanisms behind these transient stability problems. Furthermore, we present a thorough comparison of typical reactive power control schemes. Finally, experimental results verify theoretical analyses.

This paper proposes a two-layer global synchronous pulsewidth modulation (GSPWM) method for attenuating the circulating leakage current of a photovoltaic (PV) station with many inverters connected in parallel, and meanwhile allowing the simple two-level string inverters instead of the commonly used three-level string inverters to directly connect in parallel without the transformer isolation in order to save cost and increase efficiency. The first-layer GSPWM could calculate the optimal phase-shift angles of triangular carriers among the parallel-connected two-level inverters to effectively attenuate their circulating leakage current. The second-layer GSPWM collects the equivalent variable per group provided by the first-layer GSPWM and can still significantly attenuate the total current harmonics at the point of common coupling. Experimental results of a small-scale prototype are presented to verify the performance of the proposed two-layer GSPWM.

In principle, the location of global maximum power point (GMPP) of partially shaded photovoltaic (PV) array highly correlates with the shape of $P - V$ curve, but such correlation is complex and hard to be specified using the traditional mathematic expressions. This article, therefore, first explores the nonlinear characteristics of $P - V$ curve under partial shading conditions (PSCs) using an artificial neural network (ANN), which could help locate the GMPP after scanning several selected characteristic points. Then, the perturb and observe (P&O) algorithm will be further integrated to achieve fast GMPP tracking based on the obtained location of GMPP. Being different from the existing ANN-based GMPPT methods, meteorological information, such as irradiance and temperature, is not mandatory. In implementation, both PV array model construction and ANN training process are performed offline, which will not complicate the online GMPP tracking algorithm. Compared with several GMPPT algorithms in recent references, the proposed method achieves higher tracking speed and meanwhile guarantees a high tracking efficiency. The performance of the proposed method was verified in simulation and experiment, where it achieves an average tracking efficiency of 99.32% and an average tracking time of 0.8 s.

This article proposes a thyristor-based T-type converter with modular multilevel dc-link (TTC-MMD), where the thyristor-based T-type converter provides the basic phase commutation capability and two arms of cascaded submodules in dc-link generate the multilevel switching line voltages. To guarantee the successful turn- on and turn- off of thyristors, one full-bridge submodule (FBSM) is embedded in each arm. In addition, the FBSM can also be employed to participate in the multilevel switching modulation. For real applications, the TTC-MMD can be employed as a medium-/high-voltage static synchronous compensator, which features triplen frequency fluctuation of capacitor voltages and easy balancing of capacitor voltage under various operation conditions due to the common dc-link shared for three phases. It can also be implemented in other medium-/high-voltage applications. Finally, simulation and experimental results verified the performance of the proposed converter.