Artificial intelligence (AI) has rapidly become a core enabling technology in photovoltaic (PV) power systems, supporting improvements in forecasting accuracy, operational control, fault diagnosis, and system-level energy management. Despite the rapid growth of this field, a comprehensive understanding of its intellectual structure, thematic evolution, and emerging methodological directions remains fragmented. To address this gap, this study develops an integrated bibliometric-thematic analysis framework to systematically map the knowledge structure, research trajectories, and methodological frontiers of AI applications in PV power systems. The analysis is based on 4752 peer-reviewed journal articles indexed in Scopus (2006–2025). It combines performance analysis, co-citation analysis, keyword co-occurrence analysis, and bibliographic coupling to answer five structured research questions. The results demonstrate that PV power forecasting constitutes the central intellectual backbone of AI-based PV research, with the highest citation concentration and the strongest thematic connectivity across clusters. Thematic evolution analysis reveals a clear methodological transition from conventional machine learning models toward hybrid deep learning architectures, uncertainty-aware prediction frameworks, and physics-based AI integration. Furthermore, emerging research frontiers are characterized by generative learning models, multi-source data fusion strategies, and resilience-oriented fault diagnostics, while critical gaps persist in benchmarking standardization, uncertainty quantification, system-level integration, and large-scale industrial deployment. Unlike prior reviews that focus on isolated technical applications, this study provides the first integrated performance analysis and science-mapping synthesis that connects intellectual foundations, thematic evolution, and frontier innovations across the entire AI-based PV ecosystem. The findings offer a structured research roadmap and actionable guidance for researchers, PV plant operators, and policymakers aiming to design intelligent, scalable, and resilient PV energy systems that support the global low-carbon transition.
Altyeb et al. (Thu,) studied this question.