Learning in an adaptive backthrough control structure

The paper presents a neural network (NN) based adaptive backthrough control (ABC) scheme for both linear and nonlinear dynamic plants. Unlike other feedforward NN based control schemes the ABC comprises of one neural network which simultaneously acts as both plant model (emulator) and the controller (inverse of the emulator). For linear plants, without noise, the resulting feedforward controller, providing that the order of the plant and plant model are equal, is a perfect adaptive poles-zeros canceller. In the case of a nonlinear dynamic system, and for the monotonic nonlinearity, the proposed ABC control represents the nonlinear predictive controller. The ABC scheme is based on the discrete nonlinear (NARMAX) dynamic model. For such models and for monotonic nonlinearity, the calculation of the desired control signal is the result of the nonlinear optimization procedure with a guaranteed convex search function and consequently with a unique solution.