Research On Low-frequency Oscillation Model Recognition Of Power System Based On Frequency-Domain Decoupling And Reducing-order Model

With the development of power system, and the wide spread of Internet and huge machine group power system excitation, Low-frequency Oscillation has become a threat to the power system. In the mutli-machine power system, there exits such Low-frequency oscillation model as local and regional oscillation models. Therefore, It is of great importance to classify the oscillation model, to identify the related-machines, and thus to protect the sound running of power system.This paper, based on frequency-domain decoupling and reducing-order model, attempts to do the following work: classifying coherent group recognition with the a method of coherent group fuzzy clustering, providing a new method of recognizing low-frequency oscillation model according to the result of fuzzy clustering and lastly providing a new method of defining the character of low-frequency oscillation model.Based on the linearized model C₁～C₁₈ considering the effect of damping windings, this paper provides a way of identifying coherent group's fuzzy cluster analysis. The fuzzy equivalent matrices with a proper threshold ofαcan present information of coherent machine group, and thus achieve reasonable recognition coherent group.Based on the low-frequency oscillation model, the paper puts forward a new method of identifying multi-machine group's low-frequency oscillation model. Through analysis of related machines information provided from the low-frequency oscillation model, the paper finds that the machine or machines with the same low frequency oscillation model belong to the related machine group.According to the result of fuzzy clustering of coherent group and identification of related machines under the model of low-frequency oscillation, the paper provides a new means of recognizing low frequency oscillation model. If related machines with low frequency oscillation model belong to the same coherent group, which indicates that the model is guided by machines inside the coherent group, and it can be classified into the local model; if it is distributed in several various coherent group, it indicates that this model is guided by two or more machines among the coherent group coherent group, and it can be classified into regional group.Taking the detailed visual linearized model considering the effect of damping windings into consideration, this paper analyses the classical four-machine system and regional ten-machine by self-created recognizing procedure of low-frequency oscillation model, and finds the same results as the one recognized by the traditional methods, which proves that our method is valid and correct.Lastly, this recognizing method is applied in recognizing the low-frequency oscillation model in a certain district power net.