Neuro-fuzzy Model for Estimating the Release Envelope for an LGB

Abstract

An unguided bomb defines a unique ballistic trajectory for specific release conditions from a fixed point in the space to a defined target. For a Laser Guided Bomb, the point of drop lies in an envelope that provides enough reaction time for the pilot to act. The release envelope gives a minimum and a maximum point from which the guided bomb can be dropped for damaging the defined target. These minimum/maximum release points depend on the guidance and control system components like sensor instantaneous field of view (IFOV), in which sensor can sense the target so that guidance can be applied effectively and bomb maneuverability, which depends on the limitations of the control surfaces of the bomb. The problem can be attempted with the ballistic equations using numerical techniques. It requires large amount of design data. From the firing data one can get the release envelope with linear / nonlinear regression but the error is very large. A neural model helps to increase this accuracy in most of the release conditions. In this work, a neuro-fuzzy model has been developed to obtain the minimum/maximum range for given altitude, dive angle and velocity of the aircraft. The optimum parameters of a fuzzy inference system are extracted from the firing data using neural network. This neuro-fuzzy model not only improves the accuracy but improves over the outliers also. The results from numerical, neural model and neuro-fuzzy model are obtained and compared. It is observed that the estimates obtained using neuro-fuzzy model are in very close agreement with the experimental values.