【摘要】 在空间引力波探测任务中，超高精度惯性传感器在入轨之前必须经过地面测试与评价，悬丝扭秤是地面测试的首选装置。为了获得悬丝扭秤的精确动力学模型，以减小其固有系统误差，提出了一种基于改进储备池计算（RC）的扭秤动力学预测模型。所提模型利用注意力机制强化时间序列数据的长期依赖特征，并通过贝叶斯优化算法定位模型最优超参数空间，提高了动力学模型的预测精度。在实验室收集的扭秤转角时序数据上验证了所提方法的有效性，相较于传统RC的模型预测误差平均降低40%以上，可以为惯性传感器的地面测试提供可靠的动力学参考。更多还原

【Abstract】 In the space gravitational wave detection mission, high-precision inertial sensors must undergo ground testing and evaluation before being placed in orbit. The torsion balance is the preferred device for ground testing. In order to obtain an accurate dynamical model of the torsion balance, to reduce its inherent systematic error, a torsion balance dynamic prediction model based on reservoir computing(RC) is proposed.The proposed model utilizes attention mechanisms to enhance the long-term dependence features of time-series data, and employs Bayesian optimization algorithms to locate the optimal hyperparameter space, thereby improving the predictive accuracy of the dynamical model. The effectiveness of the proposed method is validated on time-series data of torsion balance angular displacement collected in the laboratory. Compared with the traditional RC methods, the prediction error of the proposed method is reduced by more than 40% on average, which can provide a reliable dynamic reference for ground testing of inertial sensors.更多还原