A high-precision data fusion method for robust measurement of highly dynamic non-Gaussian models

Abstract

A high-precision data fusion method for robust measurement of highly dynamic non-Gaussian models, considers the sampling period fluctuation of hardware sensors in a highly dynamic system as the random uncertainty of the system, and establishes a model including UKF filter and a model according to its fluctuation range and trend. The filter model set of the fuzzy inference system, calculates the probability that the UKF filter model matches the current high dynamic system state through Bayes' theorem, updates the matching probability in real time, and uses the updated matching probability as the input of the fuzzy inference system, through the fuzzy inference The system obtains the adaptive estimation probability, and finally fuses multiple state estimates based on the adaptive estimation probability to obtain the final mean and covariance estimation of the state variables of the highly dynamic system. The present invention can not only realize the combination of highly dynamic, strong nonlinear, and non-Gaussian models The data fusion of the system can reduce the number of pre-stored model sets, while improving the computational efficiency of model probability update and the measurement robustness of high dynamic systems.

Description

technical field [0001] The invention relates to a high-precision data fusion method oriented to robust measurement of a high-dynamic non-Gaussian model, and its applicable field is combined navigation and other multi-sensor information fusion fields. Background technique [0002] Global Navigation Satellite System (GNSS) is a navigation system that can provide all-weather precise positioning services, but it is susceptible to human and non-human interference, resulting in poor positioning robustness. Inertial Navigation System (INS) is a completely autonomous navigation system with good anti-interference ability, high accuracy in short-term and low navigation accuracy in long-term work. Combining the two navigation systems can learn from each other and obtain better navigation effects, so it has become a hotspot in navigation research. The data fusion algorithm of multi-sensor output is the focus of integrated navigation research. In recent years, the Kalman filter (KF) and...

AI Technical Summary

Problems solved by technology

Traditional IMM information fusion is mostly used for tracking applications of dynamic targets. Recently, there have been discussions on applying it to integrated navigation data fusion. However, there are still many problems in directly using IMM, such as constructing a large enough model to accurately match the operating state of the system. At the same time, the accuracy of the model set is greatly affected by the prior knowledge of the algorithm designer, while the variable structure multi-mode interaction (VSIMM) algorithm can reduce the number of model sets stored in the system in advance and reduce the probability transfer. The computational load of the matrix, and the algorithm adaptively generates a new model in the limited model set to adapt to the change of the statistical characteristics of the system process noise

Images