From Patents to Foresight: Identifying Dual-Carbon Technologies in the Automotive Sector With LDA and Co-Occurrence Networks

Abstract

This study traces the development of dual carbon oriented technologies in the automotive industry by constructing a patent based intelligence and analytics framework. The methodology combines Latent Dirichlet Allocation topic modelling with co-occurrence networks derived from keywords and International Patent Classification codes. Using a dataset of 10,808 invention patents filed with China's national patent office between 1986 and 2023, we identify major thematic clusters, uncover cross domain recombination patterns, and depict their temporal evolution. On this basis, technologies are located along a dynamic lifecycle and the associated opportunity spaces are grouped into prospective, emerging, established, and declining fields. The empirical results show that electric mobility remains the dominant trajectory, with charging infrastructure and battery technologies at its core, while vehicle control systems, data and cyber security functions, lightweight materials, and solutions for thermal management and powertrain efficiency become increasingly prominent. The paper contributes by proposing an integrated lifecycle and network based lens to link knowledge recombination and structural positions to sectoral transformation, by building a reusable early warning pipeline for continuous monitoring and foresight, and by offering practical guidance for steering research portfolios and technology roadmaps toward dual carbon objectives.