Comparison of Feedforward Control Schemes for Real-Time Hybrid Substructuring (RTHS)

Abstract

In order to meet the high demands in testing, actuators must be able to follow their desired displacement with high precision. Feedforward control enables high tracking performance of actuators. In combination with feedback controllers, an actuator can follow a prescribed trajectory quickly, stably and robustly under varying conditions. In Real-Time Hybrid Substructuring (RTHS), a method where parts can be tested under realistic boundary conditions, high tracking performance of the actuator is vital. It not only increases fidelity of the RTHS test outcome—meaning that the test replicates the environment and boundary conditions of the test specimen well—but it also prevents the RTHS loop from becoming unstable. Hence, research is carried out in the field of control schemes being applied to RTHS systems. In this work, the existing cascaded feedback control of the position controlled Stewart Platform is expanded by three different feedforward control schemes: model-based dynamic feedforward, modeling-free iterative learning control and velocity feedforward. The tracking performances are compared and discussed using a commanded sine trajectory. Results reveal that modeling-free iterative learning control and velocity feedforward outperform model-based dynamic feedforward and follow the desired trajectory with high amplitude and phase accuracy. Velocity feedforward is simple and requires almost no implementation effort. Thus it is recommended for applications with stiff actuators. In contrast, modeling-free iterative learning control is recommended for tasks where the actuator is not stiff compared to the test specimen. As all these feedforward control schemes improve the tracking performance compared to feedback control, the fidelity of the RTHS test will improve using them.