Clouds moving in front or away from the sun are the leading cause of irradiance variability. These variations have a repercussion on the electricity production of photovoltaic systems. Predicting such changes is essential for proper control of these systems and for maintaining grid stability. Images from the sky have proven to help with short-term solar irradiance forecasting, especially when combined with artificial intelligence. Nevertheless, these models tend to smooth the irradiance fluctuations. We propose a forecasting model to predict the clear-sky index in a forecast horizon of 20 min with a 1-minute resolution. Our model, based on a classifier to determine the sky conditions and, on an optical flow, applies an artificial intelligence model explicitly trained on each class of sky conditions. This strategy has an equivalent performance to an unclassified model and a forecast skill between 5 and 20% with respect to the smart persistence model for most classes of sky conditions while requiring considerably less training data. Although our model reduces the overall predicting error, it still has difficulties predicting irradiance changes and mainly overcast days. Our classifying strategy can be applied to other models targeting different objectives to predict sudden changes in either irradiance or power related to photovoltaic systems.

Solar photovoltaic (PV) energy is variable. The output power can change considerably in a matter of minutes, imposing challenges on the control of systems connected downstream. The power from these systems can be smoothed using electric storage, potentially increasing the system cost. An alternative is to deliberately curtail the power before it starts to change. This strategy relies on ultra-short-term forecasting to determine the curtailment point. Unfortunately, forecasting is prone to errors and high uncertainty even in the very short-term, leading to control errors. We propose an active power curtailment control strategy for a stand-alone solar photovoltaic system powering an electrolyzer. Our work's novelty relies on the controller's ability to deal with large forecasting errors and high uncertainty, combining artificial intelligence for predicting the power ramps and fuzzy logic to account for imperfect prediction. We validated our approach using a hardware emulator of the photovoltaic system, power converter and electrolyzer. Under clear sky conditions, the curtailment results in unnecessary energy loss, while under variable irradiance, the controller successfully smooths the power ramps within 10% of the PV system's nominal power. Although our approach was designed for a stand-alone system, its concept can be directly applied to grid-connected systems as well.

The development of clean hydrogen and photovoltaic (PV) systems is lagging behind the goals set in the Net Zero Emissions scenario of the International Energy Agency. For this reason, efficient hydrogen production systems powered from renewable energy need to be deployed faster. This work presents an optimization procedure for a stand-alone, fully PV-powered alkaline electrolysis system. The approach is based on the Particle Swarm Optimization algorithm to obtain the best configuration of the PV plant that powers the electrolyzer and its compressor. The best configuration is determined with one of three indicators: cost, efficiency, or wasted energy. The PV plant needs to be oversized 2.63 times with respect to the electrolyzer to obtain minimum cost, while for high efficiency, this number increases by 2%. Additionally, the configuration that minimizes cost, wasted energy or maximizes efficiency does not correspond to the configuration that maximizes the annual PV yield. Optimizing for cost results also leads to the best operation of the electrolyzer at partial loads than optimizing for efficiency or wasted energy.

We present the analysis of the size and angle configuration of a solar photovoltaic (PV) plant connected to an alkaline electrolyzer without electrical storage nor support from the grid. Our approach is based on using the available energy in the most efficient way. To reach our objective, we allocated the available PV power to the stack and auxiliary components, then determined the produced hydrogen in one year. By defining 4 indicators, we determine that the PV plant should be oversized 1.6 times the rated power of the electrolysis stack. The orientation analysis of the PV modules shows that the optimal angles determined for maximum PV yield are not the same if other indicators are used and many of them are in conflict.