Switching regression is a data mining tool for revealing intrinsic non-linear dependencies among exploratory and objective variables, and Fuzzy c-Regression Models (FCRM) is an FCM-type switching regression algorithm. In this paper, FCRM is applied to a residential solar electric power analysis. Because meteorological observation of actual sunshine duration is often underestimated from the actual quantity of solar radiation, we often fail to reveal intrinsic relation among solar electric power and sunshine duration. In this paper, FCRM is modied for handling direction-dependent uncertain observations in an element-wise manner and its applicability is demonstrated with several experimental results.
This paper considers synthesis of gain-scheduled (GS) controllers for discrete-time linear parameter-varying systems. We propose a new parameter-dependent LMI condition for synthesis whose solution gives a GS controller. Derivation of the proposed LMI does not depend on a sufficient condition that the previous paper introduced via matrix augmentation and forcing some of the variables to be indenpendent of the scheduling parameter to obtain an LMI for synthesis. Numerical examples are provided to illustrate that the proposed method can synthesize a gain-scheduled controller with a smaller upper bound of closed-loop l2 gain.
This paper describes fuzzy average difference imaging for ultrasonic nondestructive testing. In our experiment, we employ a piece of wind turbine blade as a specimen. The specimen has holes on back side as artificial damages. We acquire ultrasonic waveforms from scanning lines on surface of the specimen using an ultrasonic single probe. We make cross-section images of the specimen by correcting the each scanning line wave data. We set scanning lines so that specimen constructions under the lines are same each other. Therefore the images show same construction of inside of the specimen, we can enhance the damage echoes by using average difference imaging. To extract the damage echoes from the images, we applied damage extraction method aided by fuzzy logic and average difference imaging. As the results, we found the line image with all damage portions, and we estimated depth of damage surface with high accuracy. Therefore fuzzy average difference imaging showed effectiveness for extracting difference potions on similar images.
In this paper, we propose a new design method of RKF via l1 regression for multi output systems. Parameters of conventional RKF are designed by heuristic methods, so the parameters have no physical meanings. It is shown that statistics of Gaussian measurement noise determine the parameters of RKF via a primal and dual problem of l1 optimization problem. We discuss a covariance matrix of updated state estimation error. The proposed parameters can design the parameters systematically. In addition, the parameters have physical meanings, and we need no prior information except Gaussian measurement noise. RKF with the proposed design method is applied to a two-wheeled vehicle control with outliers, and the effectiveness is demonstrated by numerical simulations.
In this paper, the dynamic model of ’credibility’, which is used as main parameter of judgment in immunity-based system diagnosis, is extended so that diagnosis system can quickly judge fault if a specific fault mode occurs in a sensor. The amount of change of credibility changes according to size of similarity between actual output of a sensor and estimated output of it. The estimated output is calculated by using relationship among outputs of other sensors if a specific fault mode occurs in the sensor. The effectiveness of the proposed method is evaluated by experiments.
Many control system design methods are developed based on the dynamics of the accurate plant model. However, desired control performance cannot be achieved by the designed controller if the modeling error is not small. To overcome this problem, we have proposed a compensator to minimize the gap between an actual plant and its nominal model in SISO setting. The compensator design problem has been reduced to the standard LMI design problem and the modeling error is minimized. Consequently, the desired control performance can be achieved if we use the designed controller with our proposed compensator. The design method is extended to MIMO systems. In particular, not only error between compensated system and model, but also magnitude of the compensation input are handled in the compensator design. The effectiveness is shown by the numerical examples.