IDENTIFICATION OF MINESNIPER’S DAMPING PARAMETERS USING NEURAL NETWORKS Experimental Results

In this article experimental work is presented on the identification of the damping parameters in a new defence system, called Minesniper. Simulations have revealed that the standard identification model for the damping parameters with a linear and a quadratic term yields inaccurate results. Therefore, neural networks are used to represent the damping. As the available, noisy training data set is too short for a full identification of the dynamics, feedback neural networks encounter problems in representing the damping in regions that have not been visited during training. To alleviate this problem radial basis functions have been applied successfully.