H/sub /spl infin// estimation for nonlinear systems

An output estimator for a nonlinear system (the extended H/spl infin/ estimator) is developed based on suboptimal H/spl infin/ estimation theory. The performance bound, used in generating the estimator gain, depends on the model obtained by linearizing the plant about the current estimator state. Therefore, this model and the performance bound vary in an unpredictable manner. Computation of the minimal performance bound at each iteration is untractable, in general. A computationally reasonable adaptive algorithm for generating the performance bound is proposed. The use of this adaptive performance bound is illustrated by application to an example.