In the context of intelligent manufacturing, this paper proposes a modeling framework for product value enhancement path driven by industrial design innovation. Based on 1536 product design samples and 18240 innovation behavior records, a multi-modal feature matrix covering morphological and semantic structure, functional configuration, process parameters, user interaction feedback and market response is constructed. Product value labels are generated through innovation execution scores, manufacturing adaptation scores, and feedback acceptance scores. The interpretable value improvement paths were extracted by combining graph association analysis and rule learning, and the path rules were retained under the support threshold of 0.08 and the confidence threshold of 0.67. The rule vectors are clustered into five product innovation patterns with a silhouette coefficient of 0.66. The path with high confidence is embedded into the intelligent design generation module as a constraint control condition. The experimental results achieve 91.4% path recognition accuracy and 88.7% value response consistency, which verifies the effectiveness and cross-scene adaptability of the proposed framework.
