Tracking and predicting technological knowledge interactions between artificial intelligence and wind power: Multimethod patent analysis

Abstract

To track the dynamics of AI and wind power technology knowledge interaction and predict future interaction directions, this study proposes a multiview and multilayer patent analysis framework based on three data-driven methods: DMC co-occurrence networks, LDA, and link prediction. The framework is applied to collate and analyse patents related to wind power technologies using artificial intelligence from 2010 to 2021. We find that the number of AI and wind power technology knowledge interactions increases significantly over time, but the network is sparse overall and still has much room for improvement. Second, the AI and wind power technology knowledge interaction patterns show a shift from machine learning models (generation-side wind power technology) to deep learning models (generation-side and transmission- and distribution-side wind power technology) to hybrid AI models (generation, transmission, distribution, and power consumption in the whole process of wind power technology). Finally, possible future directions of interaction between AI and wind power are predicted. The proposed framework is expected to yield a new empirical perspective on green energy technology development. Additionally, the obtained results provide a comprehensive understanding of AI application research in wind power generation.