Previous Abstract Return to Session A1

Session A1: High Performance Inertial Sensor Technologies

Time Series Modeling of the Accelerometer’s bias Considering the Temperature Delay Effect
Wenfeng Tan, Wei Wu, Dongkai Dai, Xingshu Wang and Shiqiao Qin, College of Opto-electronic Science and Engineering, National University of Defense Technology, China
Location: Big Sur
Alternate Number 4

As the core sensor of strapdown inertial navigation systems (SINSs), the accelerometer’s bias caused by temperature variation affects the system navigation accuracy greatly. The accelerometer built-in temperature sensors can only measure the temperature of its outer surface. Due to the heat conduction and convection effect, the actual temperature which affects the accelerometer’s bias directly, is delayed and different from the temperature sensed by the temperature sensors. Thus, there exists a response delay between the accelerometer’s output and the temperature of the accelerometer’s outer surface. Moreover, different ambient temperature conditions (the mathematical boundary conditions) result in different delays. Therefore the model coefficients vary with the temperature varying environment in conventional bias temperature model, which means the temperature model may fail to work in complex temperature varying environments.
Based on the basic heat transfer law, the heat transfer process is introduced into the modeling process in our research. It is theoretically proved that the model coefficients are determined by the environmental temperature varying pattern in conventional bias temperature model due to temperature delay effect. It limits the universality of the conventional model in complex temperature varying environments. We deduced a novel temperature model according to the temperature delay effect, called the time series model. Since the proposed model is based on the temperature delay effect and the model coefficients are determined by the accelerometer’s physical parameters and independent of the environmental temperature variation. Therefore, it is a more physically realistic temperature compensation model for accelerometer’s bias in complex temperature varying environments. Experiments under different temperature varying environments were also designed to verify its effectiveness.
The time series model has the following advantages: Firstly, the modeling experiments can be simplified. The time series modeling method can achieve high prediction accuracy over a wide temperature range with the modeling experimental data over a small temperature range, thus yielding a decrease in the complexity of the experiments. Secondly, the system components can be protected. A large temperature-varying rate is possible to cause some damages to the internal components of the system. The time series model can achieve high prediction accuracy over a large temperature-varying rate range through the modeling experiments over a small temperature-varying rate range, thus avoiding the possible damage caused by experimental environments. Therefore, the proposed model has important application in complex temperature-varying environments.



Previous Abstract Return to Session A1