Strategic Patenting and Technological Evolution in Energy Utilization and Recovery

This study investigates the strategic evolution of patent technologies in energy utilization and recovery, focusing on current technological landscapes and predicting patent development trends. We identify key insights into the sector's growth and future potential by analyzing 30 196 patents from the Google Patents database (2008–2023) and applying machine learning techniques. Patents were retrieved using International Patent Classification (IPC) codes specific to energy utilization and recovery, then deduplicated and screened by the main patent office. Patent quality was estimated by backward-citation impact. The findings show that the energy recovery field is expanding, with a lifecycle coefficient expected to reach maturity by 2025. In addition, the results indicate that technological development accelerated after the 2015 Paris Agreement, peaking in 2016 and 2020. However, notable disparities exist across technological subfields, with uneven distribution, highlighting the need for increased collaboration and cross-disciplinary innovation. The “drum-type steam boilers” sector emerges as a key area for future growth, with high potential for cross-technology applications, but receives disproportionately low funding. A convolutional neural network model trained with the Adam optimizer achieved 73.6% accuracy in predicting patent citation trends, demonstrating the effectiveness of machine learning in identifying high-quality patents and automating large-scale quality assessments. Furthermore, increasing patent citations and diversifying patent categories significantly enhance patent quality, with collaboration among patent teams playing a critical role in advancing cross-technology fields. These findings help optimize technological strategies, identify innovation gaps, and develop models for high-quality patent identification in energy utilization and recovery.