Thesis or Dissertation Explicit Model Predictive Control for Fluctuation Suppression on Mechanical Systems

Sungwan, Boksuwan  ,  Sungwan, Boksuwan

pp.1 - 98 , 2015-03-25 , The University of Electro-Communications
The thesis proposes a control scheme based on an explicit model predictive control (EMPC) for suppressing fluctuation in time responses of mechanical systems. The control objective is not only to achieve robust tracking performance and rejecting disturbances, but also to suppress the fluctuation of the mechanism. Two kinds of fluctuation systems are considered in this thesis, i.e., torsional fluctuation system which a measured output contains an outlier and active magnetic micromanipulator which the operating point changes, or holding angle changes. The torsional fluctuation system can be represented by two-mass system which typically consists of a driving motor and load, both of which connect through a flexible shaft. Consequently, there is the difference between the motor and load speed, which results in the torsional fluctuation inevitably. In addition, the measured output, the motor speed, contains an outlier. The requirement of the high speed servo operation, tracking the desired motor speed, has to carefully design. The robust EMPC is proposed in order to achieve not only good tracking performance and load-change effect rejection, but also low torsional fluctuation whereas the measurement noise contains outliers. The control structure is based fundamentally on the combination of EMPC and an estimator where the well-known Kalman filter is replaced by the estimator to deal with the outlier phenomena. The effectiveness of the proposed method is compared with a PID control scheme by means of the simulation validations. The active magnetic micromanipulator having two dimensional degree of freedom that is able to move along x and y-axis in micro scale. The proposed micromanipulator's structure consists of two decoupling links, namely the top and bottom link are able to move along the x and y-axis, independently. Each link has identically parallel leaf spring mechanisms. For a steering force, the combination of permanent magnets and electric coils is utilized as double driving. The hybrid control scheme is proposed which is a combination between EMPC and PID controllers. The PID controller is suitable for handling the holding angle changes, or the initial displacements. The EMPC controller provides an excellent tracking performance. The control objectives are to achieve the robust tracking performance and to suppress the fluctuation of the flexible structure. Root mean square error is less than 4μm whereas the active magnetic micromanipulator held by the user's hand. The experimental results obtained indicate the effectiveness of the hybrid control. The results in this thesis reveal the effectiveness of both control schemes, robust EMPC and hybrid control, for suppressing the fluctuation in mechanical systems. For the torsional fluctuation system which the measured output contains an outlier, the results also show combining different norms. For the active magnetic micromanipulator, the results also show combining two controllers to handle the nonlinear system collectively. The contributions of the hybrid control enable a user to accomplish tracking reference tasks beyond human dexterity with the active magnetic micromanipulator.

Number of accesses :  

Other information