Neural network autopilot to enhance the performance of a semiactive homing guided missile

An autopilot is the nerve of a missile guidance system and has the objective of stabilizing the system with good tracking and rejecting disturbances while minimizing the effects due to measurement noises. This objective is quiet complex task due to the nonlinearity and time varying nature of the system in addition to different random/stochastic sources of uncertainty. However, the intelligent design techniques among which is the neural network offer a promising solution to those problems and satisfy the required objective. Therefore, this paper is devoted to present an online neural network scheme which provides the benefits of adaptation with only minor modification to the existing control architecture yielding a substantial advantage over other approaches that require complete redesign. The advantages of the proposed technique are demonstrated through an application to autopilot design within a semiactive guided missile. The design requirement is for lateral acceleration control such that robust trajectory following is achieved. Structured uncertainties due to errors in the computation of aerodynamic coefficients, linearization and unmodeled dynamics, as well as effects of unknown disturbances are included in the control synthesis and compensated for by the neural network autopilot.