Error Modeling Method for BeiDou Satellite Receivers Based on Long Short-Term Memory Neural Networks

In response to the influence of errors in the BeiDou satellite receiver on the precision correction fuze in two-dimensional correction munitions during accurate weapon strikes, this study proposes a BeiDou satellite receiver positioning error compensation model based on Long Short-Term Memory (LSTM) networks. By analyzing radar data from flight tests and the measurement data retrieved from recovered projectiles, the receiver error sequence is isolated. Considering the relationship between the receiver and its dynamic characteristic parameters, an error prediction model based on LSTM is established. The model takes the dynamic characteristic parameters of Beidou satellite receiver as input and the error sequence of each dynamic parameter as output to supplement the missing error sequence and compensate the error. The results show that the model can effectively predict receiver errors and reduce the maximum errors of position and speed. Through Monte Carlo simulation, error compensation effectively improves the accuracy of projectile correction, and verifies the effectiveness of the model. This error compensation method can effectively improve the accuracy and combat effectiveness of two-dimensional trajectory correction control