Tracking the dynamic evolution of lithium-ion battery recycling technology using natural language processing

Recycling technology for lithium-ion batteries (LIBs) is vital for reducing resource wastage and environmental pollution caused by spent LIBs. Although enterprises and Research and Development (R&D) personnel have worked hard to research and develop related technologies, they have not analyzed the entire process and future development directions of LIB technologies. Therefore, this study proposes a technical topic evolution analysis framework that combines change point detection and natural language processing technology to analyze LIB recycling technology patents. First, the change point detection method is used to quantitatively divide the period of technology development, allowing for an accurate analysis of the evolution process of technology topics. Next, using Latent Dirichlet allocation (LDA), technical topics existing in each period are identified. Furthermore, the Doc2vec model is used to obtain technical topic vectors while calculating the cosine similarity between topic vectors for constructing evolution paths. Finally, a two-dimensional evaluation model is defined to identify future research and development directions of the technology. Using the constructed framework, we systematically trace the dynamic evolution process of LIB recycling technology and further highlight the future development direction of the technology. This study contributes to developing a better understanding of the highly dynamic field of LIB recycling technology, and the framework constructed here can help analyze the evolution of technology-related topics in other fields.