PHOTOVOLTAIC FLUCTUATION SMOOTHING: MULTI-CRITERION OPTIMIZATION OF LOW-PASS FILTERS

Abstract

The integration of photovoltaic (PV) systems into power grids faces inherent challenges posed by power fluctuations driven by variable weather conditions, potentially destabilizing grid operations. This study, based on real data from the CDER network in Algeria, proposes a systematic comparative analysis of a discrete first-order low-pass filter (LPF) for attenuating these fluctuations by evaluating three representative time constants (TC = 10, 20, 30 min). Using a novel multi-criteria evaluation framework, we assess both smoothing performance and battery stress effects, and introduce an innovative SVM-based heatmap classification method for ramp rate (RR %) pattern analysis. Key findings reveal fundamental trade-offs: while TC = 30 min achieves an 80% reduction in high fluctuations (RR% > 10%), it concurrently increases battery energy throughput by 35%, potentially accelerating long-term degradation. The TC = 20 min configuration emerges as the optimal balance, delivering 60% power variance reduction while limiting the mean squared error (MSE) increase to just 8%. Our results provide: (1) quantitative evidence for LPF parameter optimization, (2) visual analytics for fluctuation pattern recognition, and (3) practical guidelines for maintaining grid stability while extending battery lifespan in PV-integrated systems.