Discrete-time weigh feeder control using extremum-seeking method

This paper proposes a new method for designing a weigh feeder control system. The model of a weigh feeder cannot be obtained accurately because its dynamic characteristics are time varying and its high frequency vibration modes are not easily identified. In this paper, a weigh feeder is assumed to be a first-order plus integrator system, and a control law consists of a constant feed-forward control input and a feedback loop to obtain a simple control method. The objective of this study is to design a control law to be actually employed in industry. To this end, a control law is designed using proportional gain for control error and steady-state control input. However, the proportional gain must be finely tuned. Hence, it is tuned using a discrete-time extremum-seeking method. In the proposed method, the proportional gain is tuned using an iterative feedback tuning. Therefore, the proposed method has the advantage that the control performance is not deteriorated due to gain fluctuation because the proportional gain is not changed in an experiment. As a result, the proposed control law has a simple structure and its control parameters can be intuitively understood. Consequently, the proposed design method can be easily adopted in industry. Numerical and experimental results demonstrate its effectiveness.