| Summary: | 碩士 === 國立臺北科技大學 === 電機與能源研究所 === 88 === The technology of the power electronics is used increasefully during recent years,and the facilities of the power electronics are used widely in the power device.The non-linear loading characteristics of are make a lot of harmonic currents put into the power systems and effect the power quality of power systems. To find out the positions of the harmonic sources is the primary study for power quality. This thesis mostly finds out the harmonic in the power systems, and it involves generation,source,effect,power flow analysis of harmonic,finding out the harmonic,and the future direction.
The centrobaric of this research is in finding out the harmonic sources, look through the power flow analysis of harmonic,the choices of the putting positions of the harmonic meters, and using the Kohonen artificial neural network to find out the positions of the harmonic sources. In the sampling choices section, this research takes samples above 100 points in each case after gets these samples. From mapping scatter gram, we can look the proximity groups to get together, because neural network can look back it, and we can input the datas that we want and let it look back,finally we can know the harmonic sources .
After taking samplings, in order to the vectors of the Kohonen network are predigested, and bates its complicacy, this purpose is that the data in the Kohonen network to be calculated faster,and saved on the meters. Finally a easy 5-bus power system and a 14-bus power system of IEEE is used for illustrating the performance of the proposed method.
|
|---|
