Publications
Sort:
Issue
AI-DRIVEN PHYSICS RESEARCH: A TEACHING CASE ON KEPLER’S ELLIPTICAL LAW OF PLANETARY MOTION
Physics and Engineering 2024, 34(5): 198-203
Published: 02 February 2026
Abstract PDF (1.3 MB) Collect
Downloads:1

AI-driven scientific research has made significant advancements in various fields such as physics, chemistry, and biology. Integrating AI tools into college physics education to cultivate students’ familiarity with their principles and methods is a cutting-edge topic in current educational reform. This paper takes Kepler’s law of elliptical planetary motion as an example to demonstrate the role and limitations of symbolic regression in discovering physical laws. It guides students to start from data, combine artificial intelligence tools with physical imagery, and re-derive Kepler’s laws of planetary elliptical motion. This paper demonstrates the positive role of artificial intelligence tools in scientific exploration and emphasizes the importance of physical reasoning when using AI tools. It provides valuable teaching examples for integrating AI-driven physics research into physics education.

Issue
AN AI-DRIVEN PHYSICS RESEARCH CASE: MAGNETIC-MECHANICAL OSCILLATOR
Physics and Engineering 2025, 35(2): 250-256
Published: 07 August 2025
Abstract PDF (3.2 MB) Collect
Downloads:32

This paper is based on the 6th question of the 2023 Chinese Physics Academic Competition, exploring the characteristics of a magnetic mechanical oscillator. This oscillator consists of two leaf springs fixed on a non-magnetic base and magnets attached to the upper ends of the springs. Due to the repulsive force between the magnets, the vibration of the two leaf springs is coupled. Through experimental observation and data collection, this paper thoroughly analyzes the oscillation phenomenon of the system. With the assistance of artificial intelligence (AI) algorithms, we attempt to directly derive the system's equations of motion from experimental data and compare them with theoretical analysis results. The results show that the AI algorithm can quickly and accurately fit the equations of motion without requiring complex theoretical analysis, demonstrating its potential in dealing with complex physical systems and greatly improving research efficiency. This validates the feasibility of using datadriven methods to study complex oscillatory systems and provides valuable insights for introducing AI-assisted teaching in university physics experiments.

Total 2