A novel hybrid grey-based strategy for improving the model precision of a dynamically tuned gyroscope

作者：Fan, CL^*; Jin, ZH; Tian, WF

来源：Measurement Science and Technology, 2003, 14(6): 759-765.

DOI：10.1088/0957-0233/14/6/308

• grey model
• drift data
• dynamically tuned gyroscope
• wavelet transform
• linear regression

This paper addresses several grey-based models for reducing the modelling error of a dynamically tuned gyroscope (DTG). Wavelet transform (WT) and linear regression techniques are introduced into the grey model to enhance the modelling capability of the GM (1, 1), which is a single variable first-order grey model. The raw DTG drift data are preprocessed by the WT to eliminate any disturbing impactive noise. The post-processing data are then used to construct the grey model. Due to the inherent errors between the modelling output values and actual data, the grey model is further compensated by the linear regression method. Long-term numerical results that measured x-axis and y-axis drift data of a DTG verify the effectiveness of the proposed hybrid strategies. The neural-network modelling method is also investigated to provide a comparison with the grey-based models. It is shown that the hybrid grey model gives a very satisfactory performance when both the WT and the linear regression methods are employed.

• 出版日期2003-6
• 单位上海交通大学