Design And Implementation Of Multi Mass Elastic Torsion System Controller Based On Model Predictive Control

The elastic torsion drive is widely used in intaglio printing driven, spinning machinery in theroller drive, and in the CNC machine tools, electric ball screw drive structure that widely used ininjection molding machine etc.. In addition, the situation of load inertia changes or motor inertiaand load inertia mismatch will produce similar effects.One of the key properties of elastic torsional drive is the existence of torsional vibration,torsional oscillation makes the fatigue life of machine equipment and control precision is greatlyreduced, what s more, may cause mechanical accident. How to restrain the torsional oscillationsand make the system meet the need of dynamic and steady-state performance is the core problemof elastic torsion drive.The paper selected the The multi mass elastic torsion system device which designed anddeveloped by B&R company as the research object. We designed the control system to achieve thespeed controlling of the second mass. Specific contents are as follows:In the thesis, two approaches were used to get the system model: mechanism modeling andexperimental modeling. Through the mechanism modeling, motor, encoder, spring and two inertialsystem were analyzed the mechanism characteristics separately, then we established the equationsof each part and selected the state variables to get the system state space equation. Based on thebuilt model, the thesis simulated on Matlab/Simulink platform. Experimental modeling collectedthe step response data from the Automation Studio. The sampling data is collected and curves areproduced by the CFTOOL in Matlab in order to get time-domain expression and then work out thetransfer function of the system. We analysis frequency domain characteristics, and study the affectof the system on different hardware changes. Model predictive control algorithm has been used in the thesis based on object properties.Then we analysis the affect of the system controlling in different control interval, predictionhorizon and control horizon by simulating MPC algorithms in Matlab/Simulink. The thesisrealizes the seamless connection of Matlab/Simulink and Automation Studio in order to changeMPC algorithms to C language code, then download the code to PCC to achieve MPC algorithmsdesign.Based on hardware platform, the MPC algorithm is used for the speed controlling of thesecond mass to get a better performance than the conventional PID algorithm. We experiment thecontrol system through extending to three wheels, human disturbance and other ways, and theresults show that the scheme can effectively suppress the torsional oscillations. In addition, thesystem shows good dynamic and steady-state performance and robustness, which has somepractical value.