To address prominent problems in college physics teaching for science and engineering students, such as difficulty in understanding abstract concepts, inefficient classroom interaction, and insufficient differentiation in students' ability development, our teaching team constructed a three-stage teaching model of “Precise Planning-Flexible Interaction-In-depth Inquiry.” This model integrates the strengths of traditional teaching with the advantages of digital-intelligent teaching. Before class, learning tasks are designed around teaching objectives to guide students in integrating digital resources, while teaching content is reconstructed and iteratively optimized based on learning analytics. During class, core problems drive a “questioning-inquiry-verification” learning process. Role reversal is used to stimulate students’ active participation, while intelligent platforms collect real-time learning data, dynamically adjust groupings, deliver tiered tasks, and provide precise support through personalized guidance. After class, differentiated extension tasks are assigned, and personalized learning paths are generated. Taking the teaching of “Conservation of Angular Momentum” as an example, this paper presents the teaching practice of the three-stage model in detail. Results from questionnaires, anonymous teaching evaluations, and students’ assignment completion show that the model has gained broad recognition. It effectively stimulates students’ interest in learning and enhances their ability to analyze and solve problems, providing a replicable teaching model and practical experience for cultivating talents under the background of emerging engineering education.