A hybrid optimization algorithm for GWO Fine-tuning GRU-Aided AKF during GPS outage

•The Grey Wolf Optimization algorithm is utilized to fine-tune the hyperparameters in the GRU network, which solves the problem that the manually selected neural network hyperparameters are not optimal solutions. It has been verified that the network structure after hyperparameter optimization has the advantage of high prediction accuracy.•Sage-Husa-AKF is introduced to estimate the state vector, and the adaptive filtering strategy is used as a supplement to the GWO-GRU network, which solves the problem that outliers are difficult to deal with.•Experiments on land vehicles show that the model proposed in this paper (GWO-GRU + AKF) can also predict complex trajectory curves well, and can successfully predict different trajectory curves within the GPS outage time, such as two bends in different directions in the track.