Remarks on feedback loop gain characteristics of adaptive type neural network feedforward feedback controller

This paper presents a discussion of a feedback loop gain characteristics of a feedforward feedback neural network controller. Discussion of its stability condition under linear assumptions is briefly introduced and compared with simulation results. Simulation focuses on two points. One is that an effect of the feedback gain is similar to that of a parameter determining neural network learning speed. However, when the larger feedback gain is selected, its effect is not similar to that of the parameter determining the neural network learning speed. Second is that, when the feedback gain becomes small, the feedforward feedback controller become to be close to the direct controller. By use of this fact, simulation results also confirm that the feedback loop can suppress whole control system becomes to be unstable.