Research on low-dimensional multivariate information fusion prediction based on space battlefield situation information.

Abstract

Aiming to address the dynamic prediction challenge of space battlefield targets, this paper proposes a hybrid prediction model integrating fuzzy cognitive map (FCM) and echo state networks (ESN). The model first constructs a fuzzy relation map of enemy target combat situations through hierarchical fusion of multivariate time series data, then optimizes node associations using genetic algorithms, and finally enhances prediction robustness through a deviation feedback mechanism. Experimental results demonstrate that compared with conventional methods, the proposed model achieves significantly improved prediction accuracy in dynamic environments (the average error of genetic algorithm is 8.99%, the average error of particle swarm algorithm is 10.39%, the average error of LSTM algorithm is 63.78%, and the average error of our algorithm is 5.10%), providing effective support for real-time space battlefield decision-making. We will release the source code for peer reference¹.