Axis control using model predictive control: identification and friction effect reduction

International audienceThis paper treats the identification and control of a machining center by means of predictive control, specifically focusing on the aspect of reducing friction effect. The machine tool is a five-axis CNC Mikron machine, in the context of HSM "High Speed Machining", with open control architecture. The axes are internally controlled by current and speed PI controllers in a classical cascade framework. In an external position loop, a predictive controller is considered instead of a classical position proportional controller with a feed forward action. The novelties stressed in the paper are the identification and the tuning of the predictive controller in order to reduce the impact of the frictions. The two-degree of freedom controller obtained using predictive strategy permits to adjust separately the tracking performance and the disturbance rejection. The tracking performance is tuned to reduce the contour error and the disturbance rejection is tuned by means of a disturbance model in order to reduce the friction impact. First, based on a nonlinear simulation model considering the frictions in the axis, a numerical model is derived by least square identification. Afterwards this numerical model is used to synthetize a predictive GPC controller reducing the impact of the friction. The benefit of the proposed structure is analyzed by means of experimental tests and a comparison with the classical position loop control with speed feed-forward. The experimental results are obtained for a two-axis trajectory, showing that the resulting experimental contour errors are smaller using the predictive controller. As perspective the paper proposes to use a control structure including only an internal current controller and external predictive position loop, without velocity loop