Kazuhiro Yubai

Fault-tolerant control system of flexible arm for sensor fault by using reaction force observer

In recent years, control system reliability has received much attention with increase of situations where computer-controlled systems such as robot control systems are used. In order to improve reliability, control systems need to have abilities to detect a fault (fault detection) and to maintain the stability and the control performance (fault tolerance). In this paper, we address the vibration suppression control of a one-link flexible arm robot. Vibration suppression is realized by an additional feedback of a strain gauge sensor attached to the arm besides motor position. However, a sensor fault (e.g., disconnection) may degrade the control performance and make the control system unstable at its worst. In this paper, we propose a fault-tolerant control system for strain gauge sensor fault. The proposed control system estimates a strain gauge sensor signal based on the reaction force observer and detects the fault by monitoring the estimation error. After fault detection, the proposed control system exchanges the faulty sensor signal for the estimated one and switches to a fault-mode controller so as to maintain the stability and the control performance. We apply the proposed control system to the vibration suppression control system of a one-link flexible arm robot and confirm the effectiveness of the proposed control system by some experiments.

Design of a PID controller based on H ∞ loop shaping method using frequency responses

Robust control is used for a system with perturbations of a plant and disturbances. Above all, H∞ loop shaping method is known as a method showing a good control performance. However, this method takes designing cost because it needs a mathematical model of the actual plant. In order to reduce the identification cost, we propose a fixed-order controller design method which only uses frequency responses. The proposed method can derive a controller satisfying the better control performance with non-linear optimization.

Iterative Design of the Reduced-Order Weight and Controller for the

The H infin loop-shaping method is known to be an effective control method. However, it has two drawbacks. The first is that it is difficult to select appropriate loop-shaping weights, and the second is that the resulting controller is very complex. For the first drawback, Lanzon has proposed a suboptimal loop-shaping weight design method. It is formulated as a generalized eigenvalue minimization problem (GEVP). This suboptimal loop-shaping weight design method provides high-order weights, exacerbating the second drawback. To resolve these two drawbacks, a reduced-order loop-shaping weight design method is proposed for SISO systems in this paper. In the proposed method, the weight structure is first fixed, and the weight is then decomposed into a frequency-dependent vector and parameter matrices characterizing the loop-shaping weight. Since the open-loop constraints are represented as linear matrix inequalities with respect to the parameter matrices, the proposed reduced-order loop-shaping weight design problem for SISO systems is formulated as a GEVP, as well as Lanzons suboptimal loop-shaping weight design method. The proposed method can reduce the designers burden, although it is only valid for SISO systems. The effectiveness of the proposed method is verified experimentally by velocity control of a belt-driven two-mass system.

H_{\infty}

Generalized Internal Model Control (GIMC) structure resolve the problem of trade-off between nominal control performance and robustness. GIMC structure proposed by Zhou consists of a conditional feedback structure and an outer-loop controller. A conditional feedback structure included in GIMC constructs a robust control system easily, while an outer-loop controller specifies nominal control performance. This paper proposes the design method of the Youla parameter on GIMC by using the input/output data of the dual Youla parameter without identification of the Youla parameter. The effectiveness of the proposed design method is verified experimentally in a belt-driven two-mass system.

Loop-Shaping Method Under Open-Loop Magnitude Constraints for SISO Systems

The purpose of this paper is to demonstrate experimentally the effectiveness of a model-free controller design method for multivariable systems labeled as the Fictitious Correlation-based Tuning (FCbT). It is worth noting that the FCbT tunes controller parameters directly by using a couple of the closed-loop I/O data with no need for any mathematical plant descriptions. This approach is applied to an experimental system with two inputs and two outputs, and useful results are obtained in terms of designing controllers with a fixed structure (e.g., PID controller) encountered in many industrial processes.

A direct design from input/output data of the youla parameter for compensating plant perturbation on GIMC structure

This paper deals with the control algorithm for a reconfigurable robot system. Firstly, the concept of reconfigurable robot is described. Secondly, the distributed control in the system consisting of local-and central intelligence is explained. Then, the authors propose virtual velocity transmission algorithm (VVTA) as the best-suited non Jacobian-based inverse kinematics for this new system, and refer to the possible redundant control with VVTA. Lastly, effectiveness of thus proposed algorithm is confirmed by several simulation results.

Implementation of direct parameters tuning method for multivariable controller using a couple of closed-loop I/O data

This paper deals with a study on reconfigurable robots. Firstly, the concept of the reconfigurable robot is described. Secondly, construction of the distributed robot control system consisting of local- and central intelligence is explained. Then the authors propose virtual velocity transmission algorithm (VVTA) as the best-suited control scheme for this system, and finally confirm by simulation the effectiveness of the proposed Algorithm

Inverse Kinematics and Redundant Control Best-suited for Reconfigurable Robot

Two-mass systems are shown as a model of two masses connected by a spring and often seen in mechanical systems. PID control, resonance ratio control, H/sup /spl infin// control, and etc. have been applied to the two-mass systems. Conventional controllers that use a disturbance observer for two-inertia systems are adopted. A systematic parameter design method for the systems is proposed, where a load variation, Coulomb friction, and fast and precise control are considered. For limit cycle due to friction, a suppression condition of limit cycle is derived by an analysis using a describing function method. For load variation, robust stability for time varying system with structured uncertainty is guaranteed by the quadratic stability. Moreover, nominal performance is improved by maximizing the smallest eigen value of control system under the above restrictions. Effectiveness is confirmed by some simulations and experiments.

A fundamental study on reconfigurable robot system construction with central- and local intelligence

This paper proposes the correlation-based model-free 2DOF controller tuning method for MIMO systems. In the proposed method, a feedback controller parameters is first tuned by using the input/output data sets obtained independently from the controlled plant as often as the number of the input dimension of the controlled plant second, a precompensator is tuned by using the same number of data sets of the achieved closed-loop system. In addition, if the controllers are parameterized linearly with respect to the controller parameters, each optimal controller parameter can be obtained by least squares. The effectiveness of the proposed method is confirmed by the experimental results.

Suppression of limit cycle and improvement of robust performance in two-mass resonant systems with nonlinearity

In general, the actual responses of control systems do not coincide with their nominal ones corresponding to their simulation results due to modeling errors and/or disturbances. For LTI (linear time invariant) systems, a disturbance observer has been proposed to overcome modeling errors and/or disturbances. A disturbance observer estimates disturbances and the effect of modeling error from available signals such as a plant input and a plant output as an input disturbance, and compensates these effects by feedback of the estimated disturbance. However, a traditional disturbance observer can not be applied to LPV (linear parameter varying) systems which vary their dynamics according to their operating point. In this paper, we propose the design method of a gain-scheduled disturbance observer for LPV systems and verify its effectiveness by experiments.