Intelligent adaptive optimal control using incremental model-based global dual heuristic programming subject to partial observability

The scarcity of information regarding dynamics and full-state feedback increases the demand for a model-free control technique that can cope with partial observability. To deal absence prior knowledge system perfect measurements, this paper develops novel intelligent scheme by combining global dual heuristic programming an incremental model-based identifier. An augmented consisting unknown nonlinear plant varying references is identified online using locally linear regression technique. actor–critic implemented artificial neural networks, actuator saturation constraint addressed exploiting symmetrical sigmoid activation function in output layer actor network. Numerical experiments are conducted applying proposed method to adaptive optimal tasks aerospace system. results reveal developed observability performance comparable control, while outperforming stability adaptability.