Intelligent Missile using Artificial Neural Networks

Abstract

Missile guidance systems using the Proportional Navigation (PN) guidance law is limited in performance in supporting wide class of engagement scenarios with varying mission and target parameters. For surpassing this limitation, an Artificial Neural Network (ANN) to substitute the PN guidance is proposed by the authors. The ANN based system enables learning, adaptation, and faster throughput and thus equips the guidance system with capability akin to intelligent biological organisms. This improvement could remove the barrier of limitations with allowable mission scope. In this paper, a Multi-Layer Perceptron (MLP) has been selected to implement the ANN based approach for replacing PN guidance. Attempts to replace PN guidance using MLP are limited in the literature and warrant greater attention due to significant theoretical development with the MLP field in recent times. It is shown in this paper, that the MLP based guidance law can effectively substitute PN for a wide range of engagement scenarios with variations in initial conditions. A foundational argument to justify using an MLP for substituting PN is provided. Besides this, the design, training and simulation based testing approach for an MLP to replace PN has been devised and described. The potential for faster throughput is possible as the MLP nodes process information in parallel when generating PN like guidance commands. The results clearly demonstrate the potential of MLP in future applications to effectively replace and thus upgrade a wide spectrum of modern missile guidance laws.