This paper is concerned with the identification of turntable servo system through the usage of a reframed multi-innovation least-squares scheme. A Wiener–Hammerstein model is employed in this paper to depict the dynamic characteristics of the turntable system. In the test bed, the stabilized platform can be considered as a linear dynamic subsystem. The motor is also a linear dynamic subsystem. And the major nonlinearity characteristic between motor and platform is captured by a continuously differentiable friction model. A new reframed multi-innovation least-squares approach (RMILS) is proposed to identify the Wiener–Hammerstein model. By introducing the intermediary step updating, the innovation updating is decomposed into sub-innovations updating, which can solve the inverse of covariance matrix and improve the identification performance. Then, the consistency nature of the RMILS method is discussed by using the theoretical analysis. Finally, the simulation and experiment results explain that the developed approach produces an outstanding performance in convergence speed and identification precision comparing to the conventional multi-innovation least-squares approach.

中文翻译：

---

一种改进的转台伺服系统辨识多创新算法

本文涉及通过使用重构的多创新最小二乘方案来识别转台伺服系统。本文采用 Wiener-Hammerstein 模型来描述转盘系统的动态特性。在试验台中，稳定平台可视为一个线性动态子系统。电机也是一个线性动态子系统。电机和平台之间的主要非线性特性由连续可微的摩擦模型捕获。提出了一种新的重构多创新最小二乘法 (RMILS) 来识别 Wiener-Hammerstein 模型。通过引入中间步更新，将创新更新分解为子创新更新，可以求解协方差矩阵的逆，提高识别性能。然后，通过理论分析讨论了RMILS方法的一致性性质。最后，仿真和实验结果表明，与传统的多创新最小二乘法相比，所开发的方法在收敛速度和识别精度方面具有出色的性能。