BP neural network-based Kalman gain correction method

Abstract

The invention belongs to the field of wireless location technology and relates to a BP neural network-based Kalman gain correction method. According to the method, firstly, a mobile direction finding station, GPS equipment and existing radio transmission equipment of GPS-coordinates are respectively arranged in the downtown area, the suburban area and the rural area to obtain training data. Secondly, theoretical angle values and the variance of the measurement values are calculated, and then the above data are filtered based on the MGEKF algorithm to obtain the gain correction factor and the coordinates of a target estimation location for each iteration process. Based on the above data, a BP neural network is trained, so that the BP neural network can be fitted to the correlativity between the measurement values, the variance of the measurement values, the target estimation location, the location of the direction finding station and the gain correction factor. Therefore, the influence of the measurement error on the gain correction result based on the traditional MGEKF algorithm can be reduced. According to the technical scheme of the invention, the relationship between the fitted measurement value of the BP neural network and the gain correction factor thereof is firstly proposed, so that the gain correction error is lowered. Meanwhile, the positioning accuracy of the filtering algorithm is improved.

Description

technical field [0001] The invention discloses a BP neural network-based Kalman gain correction method and relates to the technical field of wireless positioning. Background technique [0002] Mobile single-station target positioning is more and more used in military and civil fields because of its simple equipment, strong mobility, and easy realization. In recent years, wireless interference sources in cities have become more and more rampant, which has seriously affected people's lives. Vehicle-mounted mobile single-station only direction-finding target positioning is an important means of finding interference sources in cities, and vehicle-mounted mobile stations and airborne mobile stations and elevated Compared with fixed stations, its height is very low and it is difficult to receive direct waves. In addition, the complex urban environment exacerbates this situation. In addition to improving the accuracy of direction finding equipment, a reliable filtering algorithm w...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to solve the problem of using the measured value instead of the real corrected gain in the gain corrected Kalman filter algorithm so that the measured value error is propagated to the gain corrected result, and a Kalman gain corrected method based on BP neural network is proposed , using the BP neural network can fit the characteristics of any nonlinear correlation, fitting the relationship between the measured value, the error variance of the measured value, the estimated position of the target, the position of the direction finder, and the Kalman gain corrected by the true value, suppressing Influence of Measurement Error on Gain Correction Results

Images