CONTROL OF NONLINEAR-SYSTEMS WITH MARKOVIAN PARAMETER CHANGES

The control of hybrid systems is considered where a totally observed state process is generated by a nonlinear stochastic differential equation whose parameters are functions of a random jump process, called the plant regime. Regime transitions typically represent target maneuvers in a tracking system. The control task is split between a jump detection and identification (JDI) filter and a discrete adaptation, or reconfiguration. A design is presented that takes into account the errors of the JDI device and the plant nonlinearities to produce a near optimal passively adaptive control law with an easily implemented affine structure.