Real-time midcourse missile guidance robust against launch conditions

Abstract Two real-time suboptimal midcourse missile guidance laws, called the γ -correction guidance law and the σ -feedback guidance law, respectively, are proposed in this paper. These guidance laws are based on the neural-network training of optimal trajectory data, and are designed to be robust against missile launch conditions. Using the optimal flight-path angle for the nominal launch condition as a reference signal, the γ -correction guidance law allows the missile to follow a nominal flight trajectory closely, even under perturbed launch conditions. The σ -feedback guidance law, which is motivated by proportional navigation, produces small terminal miss distances, although the resulting trajectory may deviate from the optimal one. By combining these two guidance laws, a new real-time guidance law, which is robust in the face of variations in the launch condition, is derived. Substantial performance improvements over the neural-net guidance law previously proposed by the authors are confirmed by computer simulations.