In oil and mineral exploration, gravity gradient tensor data include higher- frequency signals than gravity data, which can be used to delineate small-scale anomalies. However, full-tensor gradiometry （FTG） data are...In oil and mineral exploration, gravity gradient tensor data include higher- frequency signals than gravity data, which can be used to delineate small-scale anomalies. However, full-tensor gradiometry （FTG） data are contaminated by high-frequency random noise. The separation of noise from high-frequency signals is one of the most challenging tasks in processing of gravity gradient tensor data. We first derive the Cartesian equations of gravity gradient tensors under the constraint of the Laplace equation and the expression for the gravitational potential, and then we use the Cartesian equations to fit the measured gradient tensor data by using optimal linear inversion and remove the noise from the measured data. Based on model tests, we confirm that not only this method removes the high- frequency random noise but also enhances the weak anomaly signals masked by the noise. Compared with traditional low-pass filtering methods, this method avoids removing noise by sacrificing resolution. Finally, we apply our method to real gravity gradient tensor data acquired by Bell Geospace for the Vinton Dome at the Texas-Louisiana border.展开更多
When the initial position error or the altimeter measurement noise is large,the BUAA Inertial Terrain-Aided Navigation (BITAN) algorithm based on extended Kalman filtering can not be located accurately.To solve this p...When the initial position error or the altimeter measurement noise is large,the BUAA Inertial Terrain-Aided Navigation (BITAN) algorithm based on extended Kalman filtering can not be located accurately.To solve this problem,we propose a modified BITAN algorithm based on nonlinear optimal filtering.The posterior probability density correction is obtained by using the prior probability density of the system's state transition model and the most recent observations.Hence,the local unobservable system caused by the measurement equation through terrain linearization is avoided.This algorithm is tested by using the digital elevation model and flight data,and is compared with BITAN.Results show that the accuracy of the proposed algorithm is higher than BITAN,and the robustness of the system is improved.展开更多
The globally optimal recursive filtering problem is studied for a class of systems with random parameter matrices,stochastic nonlinearities, correlated noises and missing measurements. The stochastic nonlinearities ar...The globally optimal recursive filtering problem is studied for a class of systems with random parameter matrices,stochastic nonlinearities, correlated noises and missing measurements. The stochastic nonlinearities are presented in the system model to reflect multiplicative random disturbances, and the additive noises, process noise and measurement noise, are assumed to be one-step autocorrelated as well as two-step cross-correlated.A series of random variables is introduced as the missing rates governing the intermittent measurement losses caused by unfavorable network conditions. The aim of the addressed filtering problem is to design an optimal recursive filter for the uncertain systems based on an innovation approach such that the filtering error is globally minimized at each sampling time. A numerical simulation example is provided to illustrate the effectiveness and applicability of the proposed algorithm.展开更多
基金financially supported by the SinoProbe-09-01(201011078)
文摘In oil and mineral exploration, gravity gradient tensor data include higher- frequency signals than gravity data, which can be used to delineate small-scale anomalies. However, full-tensor gradiometry （FTG） data are contaminated by high-frequency random noise. The separation of noise from high-frequency signals is one of the most challenging tasks in processing of gravity gradient tensor data. We first derive the Cartesian equations of gravity gradient tensors under the constraint of the Laplace equation and the expression for the gravitational potential, and then we use the Cartesian equations to fit the measured gradient tensor data by using optimal linear inversion and remove the noise from the measured data. Based on model tests, we confirm that not only this method removes the high- frequency random noise but also enhances the weak anomaly signals masked by the noise. Compared with traditional low-pass filtering methods, this method avoids removing noise by sacrificing resolution. Finally, we apply our method to real gravity gradient tensor data acquired by Bell Geospace for the Vinton Dome at the Texas-Louisiana border.
基金supported by the National Natural Science Foundation of China (Grant No.61039003)the Aeronautical Science Foundation of China (Grant Nos.20090818004 and 20100851018)the National Key Laboratory Foundation
文摘When the initial position error or the altimeter measurement noise is large,the BUAA Inertial Terrain-Aided Navigation (BITAN) algorithm based on extended Kalman filtering can not be located accurately.To solve this problem,we propose a modified BITAN algorithm based on nonlinear optimal filtering.The posterior probability density correction is obtained by using the prior probability density of the system's state transition model and the most recent observations.Hence,the local unobservable system caused by the measurement equation through terrain linearization is avoided.This algorithm is tested by using the digital elevation model and flight data,and is compared with BITAN.Results show that the accuracy of the proposed algorithm is higher than BITAN,and the robustness of the system is improved.
基金supported by the National Natural Science Foundation of China(61233005)the National Basic Research Program of China(973 Program)(2014CB744200)
文摘The globally optimal recursive filtering problem is studied for a class of systems with random parameter matrices,stochastic nonlinearities, correlated noises and missing measurements. The stochastic nonlinearities are presented in the system model to reflect multiplicative random disturbances, and the additive noises, process noise and measurement noise, are assumed to be one-step autocorrelated as well as two-step cross-correlated.A series of random variables is introduced as the missing rates governing the intermittent measurement losses caused by unfavorable network conditions. The aim of the addressed filtering problem is to design an optimal recursive filter for the uncertain systems based on an innovation approach such that the filtering error is globally minimized at each sampling time. A numerical simulation example is provided to illustrate the effectiveness and applicability of the proposed algorithm.
