<p>As one of the essential carriers of knowledge, patents can provide rich cross-domain knowledge support for the product innovation design process. However, the main challenge facing most patent-assisted innovation design systems is identifying and delivering patents relevant to design requirements from the vast number of cross-domain patents. Given this problem, this paper proposes a new patent classification and recommendation method based on a multi-dimensional product life cycle dictionary. Firstly, the method defines a multi-dimensional product life cycle dictionary, which includes six main categories: function, structure, material, process, transportation, and recycling (FSMPTR), and supplements the feature keywords of each subcategory based on the Claude model. Secondly, the FSMPTR classification of patent data is completed based on the constructed dictionary, and the patent data set under the functional category is extracted. Thirdly, the TF-IDF algorithm is used to extract the features of the functional patent data set, and the similarity calculation is carried out with the synonym extension technology to accurately identify the patents related to the design requirements to form the candidate patent set related to the design requirements. Finally, a patent classification and recommendation prototype system is developed using the proposed method. Based on the system, a design example of the electric toothbrush is completed, which verifies the feasibility and practicability of the method and system. This study can provide theoretical and methodological support for product innovation design.</p>