Chih-Hsien Kung

Fuzzy-based adaptive digital power metering using a genetic algorithm

This paper describes an innovative, fuzzy-based, adaptive approach to the metering of power and rms voltage and current employing a genetic algorithm. The fuzzy-based adaptive metering engine adjusts the number of points per cycle to be processed and the location of these points. Adjustments are based on the optimal fuzzy rules constructed by a genetic algorithm to satisfy overall metering-error criteria under different operating environments while minimizing the number of points actually employed in the metering computation. This results in a reduction in the metering-computation effort, which frees up the processor for other tasks such as communication or power quality measurements. The fuzzy-based adaptive metering algorithm has been implemented on a microcontroller-based power metering system that employs a multitasking operating system which exploits the efficiencies achieved by the reduced metering rate. The fuzzy-based adaptive metering algorithm has been tested with a variety of actual and synthesized power-system waveforms and the experimental evaluations have demonstrated excellent accuracy in the metered power system quantities.

An adaptive power system load forecasting scheme using a genetic algorithm embedded neural network

The ability of load monitoring instrumentation to predict the onset of peak power demands, based on the acquired voltage and current measurements, is essential to most load management strategies. This paper describes an innovative load forecasting scheme employing the genetic algorithm (GA) embedded neural network. The new load forecasting technique is compared with the conventional artificial neural network approaches, which sometime suffer from the local minima optimization problem. Employing genetic algorithms on the design and training of artificial neural networks (ANNs) allows parameters to be easily optimized. Furthermore, the artificial neural network requires a heavy trial-and-error design and training procedure which impairs the performances of the resulting load forecasting schemes. Using the genetic algorithm approach to construct the ANN results in significantly reduced trial-and-error effort in the training phase and produces a load forecasting scheme which is more efficient, adaptive, and optimized than that provided by traditional artificial neural network based approaches. The proposed genetic algorithm embedded artificial forecasting scheme has been tested with data obtained from a sample study performed on the Taiwan power system and the experimental evaluations have demonstrated adaptability and effectiveness of the proposed load forecasting scheme.

Power source scheduling and adaptive load management via a genetic algorithm embedded neural network

This paper describes in detail the development of a decision support information system which would assist industrial managers in scheduling their equipment investment, as well as, in formulating the operational strategies for the various power sources. Such a system would also help dealing with various load management scenarios such that the objective of both minimizing the demand charge as well as maximizing the operational capability of facilities could be achieved. The proposed object-oriented management information system provides a user-friendly user interface has been developed and tested with data obtained from a sample study performed on the Taiwan power system. The experimental evaluations have demonstrated the feasibility, adaptability and effectiveness of the proposed power source scheduling and load management strategies.

The VLSI implementation of an artificial neural network scheme embedded in an automated inspection quality management system

This paper describes in detail the VLSI implementation of an innovative artificial neural network scheme embedded in an automated inspection system for the manufacturing industry employing the digital image processing techniques. An FPGA-based prototype which has been implemented and integrated with the recently developed automated inspection system, and a standard cell-based prototype which is under development in the laboratory utilizing several high performance integrated circuit design tools are also described in this paper.

Vision system motion tracking control with high performance motion estimation

In this paper, a vision system motion tracking controller with high performance motion estimation mechanism is developed, With the emergence of the information era, the users demands on video communication become more and more sophisticated, especially the requirements for both the functionality of video communication and quality. Take the video-meeting system for example, conventionally the camera can only be aimed on a particular direction. Through analyzing the movement of the object, we can estimate the new motion direction of objects. Then, according to the estimated value, the shooting direction of camera can be changed with the motion of object. In addition, the proposed method can be employed by the user to estimate the motion of his/herself to control the modon of focal point. Hence, the objective of this research is focused on the approaches to estimate the motion direction and locus by computing the motion vector of object.

Support vector machine for image based automated inspection system

ut -This paper focused on the actual applications of suppon vector machines (SVW connected with uutomated inspection scheme by utilizing the digital imageprocessing techniques. Such a system is capable of performing real-time image processing tasks and identifies the size and location of thefinished components on manufacturedproducts as well as theflaw and scrutches on surface ofproducts during the manufacturing process. The SVM embedded quality management system provides a user-friendly user interface that has been implemented and tested on a case study from the surface of credit card inspection. The experimental results have demonstruted the functionality and superiority of the developed SVM embedded inspection system.

The Design of an Innovative Method for Digital Video Surveillance System with Watermarking and Error Control Codes

This paper describes in detail the design of an innovative digital video surveillance watermark embedding and error control system. The proposed approach includes two sub-modules, one is watermark embedding and the other is error control coding. By these embedding and error control schemes, the locations where was attacked in video would be detected and corrected such that the integrity of the embedded watermark can be maintained and the attempts of manipulation of video contents can be discouraged

The development of an artificial neural network embedded automated inspection quality management system

This paper describes in detail the development of an innovative artificial neural network embedded automated inspection scheme for the manufacturing industry employing digital image processing techniques. Such a system is capable of performing real-time image processing tasks and identifies the size and location of the finished components on manufactured products as well as the flaws and scratches on surface of products during the manufacturing process. The proposed artificial neural network embedded quality management system provides a user-friendly user interface that has been implemented and tested on a case study from a printed circuit board manufacture. The experimental results have demonstrated the functionality and superiority of the developed artificial neural network embedded inspection system.

High Performance Real-time Object Detecting and Tracking System for Multiple Moving Targets

In this paper, high performance real-time object detecting and tracking system for multiple moving targets are proposed. The robot vision system which employs the proposed object detection and tracking scheme could identify color and compute the positions of object in real-time. The position information is calculated by using the color information embedded in the image and the proposed color models matching algorithm. Along with the proposed multiple targets tracking algorithm, the implemented system could detect and track up to seven targets in the field simultaneously and compute the position, velocity, and motion orientation of the objects in a speed of 1/30 seconds. The calculated information could then be provided to the decision making system and enables the system to make a proper decision of the strategy and control the soccer robots accordingly. Experimental results demonstrate that the propose method improve the effect of multiple moving targets detection, recognize target contour clearly.

MPEG-7 Based Intellectual Digital Album

In recent years the digital camera has become an essential personal accessory. The increased storage space of the memory cards result in difficulties of finding the desired photos in the traditional digital camera for the users. The described intellectual digital album is an innovative scheme to this problem. In this paper, several algorithms that can process digital content easily are employed for the implementation of the intellectual digital album prototype. The developed system can perform the feature extraction operation on the pictures in the database of photos and compare the acquired features automatically with the desired feature selected by the users to obtain the desired pictures. The proposed approach follows the MPEG-7s color, texture and shape descriptor and defines five feature selections