Simulation of two-rate adaptive hybrid control with neural and neuro-fuzzy networks for stochastic model of missile autopilot

This paper describes a two-rate stochastic control system as state-space (SS) type decomposed and discretized models of multi-input/multi-output (MIMO) stochastic subsystems with the "fast" and "slow" control neural networks (NNs) and with the "fast" and "slow" neuro-fuzzy networks (NFNs). The block diagrams both the original system with linear-quadratic-Gaussian (LQG) regular and decomposed subsystems with two-rate NNs and NFNs hybrid adaptive control were designed. An illustrative example - two-rate NN and NFN hybrid control of decomposed stochastic model of a rigid guided missile over different operating conditions was carried out using the proposed two-rate SS decomposition technique. This example demonstrates that this research technique results in simplified low-order autonomous control subsystems with various discretization periods and with various speeds of actuation, and shows the quality of the proposed technique. The obtained results show that the control tasks for the autonomous subsystems can be solved more qualitatively than for the original system. This simulation and animation results with use of software package Simulink demonstrate that this research technique would work for real-time stochastic systems.