数字孪生（Digital Twin, DT）通过创建物理实体的数字化虚拟副本，实现对系统当前状态的重现及未来行为的预测，在非线性物理实验教学中具有显著应用潜力. 本研究针对大学物理实验课程中的非线性力学系统——混沌扭摆，开发了一套基于储备池计算（Reservoir Computing, RC）的扭摆数字孪生系统. 通过对比系统数字副本的相图重现相似度δR及轨迹预测时长TP，验证了该系统具备重现混沌扭摆实时状态与预测其运动规律这两大能力. 本研究开发的数字孪生系统对其它实验教学存在着一定借鉴意义，同时为数智化实验教育的实践提供了一种可行途径.

In this paper, a digital twin (DT) system of the chaotic torsion pendulum based on reservoir computing (RC) is developed.By calculating the reconstruction similarity of phase portrait δR and the prediction length of trajectory LP, the capability of the digital copy of this DT system to accurately capture and generate the current and future dynamic states of the experimental device is verified. The DT system developed in this work provides students with an effective auxiliary tool for the chaotic pendulum experiment in collegelevel physics.