An optimal solution of a nonlinear stochastic control problem generally has a dual-control effect. A guidance law with dual-control effect can achieve a correct balance between maintaining good guidance performance and small estimation errors. However, it is impossible to obtain a closed-loop solution of the nonlinear stochastic control problem.
In this dissertation, practical dual-control guidance laws are proposed to achieve ultimately good interception performance for passive homing missiles. First, adaptive intermittent maneuver strategy is proposed. The adaptive intermittent maneuver strategy has switching threshold levels to balance between guidance error and target observability. Second, sliding mode control is used to obtain the dual-control guidance law. A manifold based on the target observability is used as a sliding manifold. Last, direct stochastic optimization is performed by co-evolutionary augmented Lagrangian method. Because the solutions are open-loop solutions, a feedback guidance law implemented a neural network is proposed for the case of application.
Each proposed guidance law is evaluated by Monte Carlo simulations. Simulation results show that the proposed guidance laws are superior to the conventional guidance law without the dual-control effect. The adaptive intermittent maneuver strategy is the most practical guidance law among the proposed dual-control laws in view of simplicity to interception performance.