科研之友
微信
新浪微博
Facebook
分享链接
摘要
It is well known that Kalman filtering is an optimal state estimation method for GPS/INS integration navigation. However, it can perform perfectly only in the condition that transition models are well defined. The Extended Kalman filter (EKF), as its nonlinear vision, gives a better accuracy which is considered as a common method for practical integrated navigation application for error estimation and compensation. While it still has difficulties to be implemented and achieve stability, adaptability and observability. To solve this problem, particle filter (PF) based on Bayesian Optimal Estimation (BOE) has shown its advantages, even so, it has been bounded by the complex calculation for the application. Meanwhile, it is found that wavelet transform featured with multi-resolution which can effectively reduce the undesirable noise of sensors and improve the Signal-to-Noise Ratio (SNR), which broadly used in denoising the unwanted data of the sensor signal. Therefore a Hybrid Wavelet Multiresolution Particle Filter(HWMPF) solution is proposed: firstly, wavelet filter is employed to handle the inertial sensor noise, and then velocity and acceleration detection are applied to judge whether the system is in high speed and dynamic condition or not. High dynamic system is implemented in using wavelet-based particle filter to deal with strong nonlinear situation with less computational complexity, while EKF is applied for low velocity and dynamic system. In addition, Zero Velocity Updates (ZUPT) carried out in zero velocity situations. The hybrid filter solution can deal with the most part of cases in practical navigation, and simulation and experimental results have confirmed feasibility and rationality of the proposed solution.
- 出版日期2012
