Solar and Wind Quantity 24 h—Series Prediction Using PDE-Modular Models Gradually Developed according to Spatial Pattern Similarity

The design and implementation of efficient photovoltaic (PV) plants wind farms require a precise analysis definition specifics in the region interest. Reliable Artificial Intelligence (AI) models can recognize long-term spatial temporal variability, including anomalies solar patterns, which are necessary to estimate generation capacity configuration parameters PV panels turbines. proposed 24 h planning renewable energy (RE) production involves an initial reassessment optimal day data records based on pattern similarity latest hours their follow-up statistical AI learning. Conventional measurements comprise larger territory allow development robust representing unsettled meteorological situations significant changes from comprehensive aspect, becomes essential middle-term time horizons. Differential learning is new unconventionally designed neurocomputing strategy that combines differentiated modules composed selected binomial network nodes as output sum. This approach, solutions partial differential equations (PDEs) defined nodes, enables us high uncertainty nonlinear chaotic contingent upon RE local potential, without undesirable reduction dimensionality. form back-produced modular compounds PDE directly related complexity large-scale patterns used training avoid problem simplification. preidentified day-sample series reassessed secondary applicability, one by one, better characterize progress. Applicable phase or frequency (e.g., azimuth, temperature, radiation, etc.) amplitudes at each determine solve particular node PDEs complex periodic sine/cosine components. improvements contribute performance concept models, cable represent dynamics systems. results compared with recent deep strategy. Both methods show approximation ability radiation ramping events, often power supply; moreover, provides more stable gust predictions alterations errors, namely over-break frontal fluctuations. Their average percentage correlation real 87.8 88.1% global day-cycles 46.7 36.3% speed h. series. A parametric C++ executable program complete metadata for month available free enable another comparative evaluation conducted experiments.