Application of Wavelet Transform and Artificial Intelligencefor Identifying Locations of Switching Capacitor Transients

• Main Authors: Po-Yuan Chen, 陳柏元
• Other Authors: Ying-Yi Hong
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/22258706215959711491

碩士 === 中原大學 === 電機工程研究所 === 93 === In recent years, due to the rapid developments of the hi-tech industry as well as much more usages of the precise production equipments and test instruments, the far high power quality (PQ) is demanded nowadays. Hence, improvement of power quality is an important task for utility companies and their customers. Generally, power quality problems include voltage swell, voltage sag, power harmonic, three-phase imbalance, frequency variation and voltage flicker. Besides, the electromagnetic transient phenomenon of the power system, such as capacitor switching, can cause incident of the voltage and current transients that would result in over voltage transient due to the resonance phenomenon. The high voltage and current transient may result in damage of devices in the power systems and malfunction of protection equipment of sensitive loads. Therefore, capacitor switching transient is a serious threat to power electronic equipments in the viewpoint of PQ. Actually, the accurate location and time of PQ problem are useful for responsibility authority and accident correction Therefore, identifying and locating the locations of transient sources have attracted more attention of utility engineers and scholars. This thesis presents a new method for efficiently locating the sources associated for utility capacitor switching transients. The proposed method first combines wavelet transform and Parserval theorem to extract the features of the transients. Then proper location number for metering measurements by fuzzy clustering is determined. Finally, the features and transient source location is trained by neural networks. Diverse patterns of PQ events are simulated by Matlab6.5/NeuroSolutions software Finally, an 18-bus power system is used for testing, Simulation results obtained by using Matlab6.5/NeuroSolutions show that the proposed approach is effective and relatively accurate in comparison with existing approaches.