Identification Of Low-Frequency Oscillations Modes Based On Fuzzy Clustering And Dynamic Partition

The wide range electrical network interconnection is a noticeable factor which the low-frequency oscillation is easy to occur. Online oscillation characteristic analysis is the important academic foundation for achieving online detection and damping control of low-frequency oscillations, while the application and development of wide-area measurements based on Global Positioning System (GPS) have provided a new approach for online identification and control of low-frequency oscillation modes. It has already obtained the widespread application analyzing power system oscillation information by the oscillation curve.There are different low-oscillation modes in inter-area system, while the dominant low damping or negative mode should be focused on. In a practical multi-machine power system, a lot of extrinsic disturbance factors, take noise for example, make natural signal identification discommodious. It is needed to select appropriate swing curve to identify the area dominant mode.First of all, the essence and cause of low-frequency oscillation in an interconnected bulk power system are researched from the coherent generators identification. The principle of iterative self-organizing data analysis techniques algorithm (ISODATA) fuzzy clustering method based on fuzzy partition is presented. Then, on the basis of fundamental assumption and necessary system provability the fuzzy set is formed, while power system is partitioned by the iterative method until the satisfying result comes out.Then, dynamic partition theory based on the coherent generators identification in the multi-machine power system is researched. The article mostly introduces basal concepts of dynamic partition theory, then advances the divide and not been partitioned method for the appropriate low-frequency oscillation signal. The theory shows weak link of network, and gives the importance academic value in looking for appropriate swing curve.Finally, Prony method can be accurate to abstract the area dominant oscillation modes from the appropriate low-frequency oscillation signal to reveal the interaction of mode in complex oscillation. For validating the method that is advanced, the feasibility and efficiency of the proposed method are verified on the EPRI 36-bus AC system.