Controlled Islanding Strategy Based On WAMS Information In Large-Scale Interconnected Power Grid

ABSTRACT:With the large-scale interconnected power grid has become the inevitable trend of power systems development at home and abroad, the dynamic security and stability of the power grid is also becoming increasingly serious. When power grid suffers from severe cascading failures, it will face with the risk of blackouts. As the last degense to prevent power systems from blackouts, islanding control can separate the whole power system into several islands for preventing accidents to expand up to widespread blackout. There are three key problems for theoretical research and analysis:on-line identification of coherent generator based on various failure modes, the searching of optimum controlled separation surfaces and fast corrective control in islands. An integrated set of controlled islanding strgetgy based on WAMS in large-scale power grid is proposed. The simulation examples validate the effectiveness and flexibility of this approach. The main work of this dissertation is summarized as follows:In coherency identification research, an improved Laplacian Eigenmap algorithm is proposed to identify coherent generators. The relations between slow mode number and coherency group number are researched and analysed in multiple time-scale system. Based on the optimum coherency group method, the sample learning category information is introduced. It focuses on the disturbed trajectory information post-fault clearance provided by WAMS. Combined with categorization inforation in the kernel space, the kernal space distance measurementare is established. Then the mainifold embedded fearute extraction is obtained by Laplacian Eigenmap algorithm in the low dimensional space. With cosine similarity factor expressed the tightness of the coherence, it can provide automation for identifying coherent generator in the three-dimensional space. Under the different failure modes, the practicality and efficiency of conherency identification are analyzed based on the improved Laplacian Eigenmap algorithm. This approach can fully consider dynamic behavior of different fault modes and time varying characteristic of nonlinear system, which makes the controlled islanding with the coherence and selectivity.In separation surfaces identification research, it is proposed that combined with the optimal segmentation of graph soectral and variable neighborhood heuristic searching algorithm to search the optimum separation surfaces. Based on the coherent generator groups, it is introduced that the aggregation of virtual coherent generator nodes to simplify the complicated network structure. Then based on the simplified network structure, analyzing the multiple partition algorithm of Laplace graph spectrum, it is defined that the laplacian graph spectrum considering the absolute value of the active power and electrical distance. By considering the direction of the power flows, it adopts a variable neighborhood heuristic searching algorithm to search for the optimal separation surfaces so that the net power imbalance of islands is minimized. It takes the preliminary separation surfaces as the initial solution and changes the neighborhood structure setting with parallel grouping method to expand the searching scope so that the total amount of imbalanced active power could be further reduced. It makes the controlled islanding with the coordination.In corrective control of islands research, it is proposed that the bidirectional power flow tracing algorithm of separation surfaces and the DC flow algorithm develop generator tripping and load shedding to maintain the power balance and stable operation. By analyzing the sensitivity of separation surfaces to generator and load power, the bidirectional power flow tracing algorithm is proposed. According to the unbalanced power in each island, the object of need adjustment generators and load nodes and the amount of adjustment can be determined. Then DC flow algorithm for fast generator tripping and load shedding is proposed, which makes the power balance and identify the branch power folw constraints. Furthermore, combined with PQ power flow decomposition method, it is necessary to verify the constraints of transimission lines power and node voltage. Finally the optimal generator tripping and load shedding are determined to guarantee the stability of each island. Based on the New England39node system, the island post controlled can maintain sybchronized operation and power balance. This approach improves the calculating speed and precision of corrective control in the island. It makes the controlled islanding with stability.In summary, this thesis takes full advantage of the global dynamic real-time information by wide area measurement system. Three critical issues of coherent identification, separation surfaces searching and corrective control in island are further researched. An integrated set of controlled islanding strgetgy based on WAMS in large-scale power grid is proposed. It achieves the coherency, selectivity, coordination and stability, which enhances the third defensive line of power system to avoid the blackout accidents.