| The power system is developing toward the direction of national network interconnection,whose dynamic security and stability problem is becoming increasingly prominent.The interaction of accidental factors may trigger a cascading failure and cause a large-scale power outage.Islanding control which belong to the third defensive line is the last step to ensure the safe and stable operation of a power system.Cutting the fault system into a number of power islands at the splitting surfaces helps to prevent the accident from further expanding.The identification of the coherent generators and the location of the oscillation center are the key technologies to searching of power system controlled islanding surfaces,which is of great significance for developing a strategy of islanding control.Firstly,a new method based on wavelet fuzzy entropy and GG fuzzy clustering is proposed to deal with the contradiction between comprehensive feature extraction and computational complexity in the coherency identification.An identification method for coherent generators based on fast S-transform and two-dimensional principal component analysis(2DPCA)is proposed.The real-time power angle of the generator is calculated by the electrical operating variables which are measured by phasor measurement unit(PMU).The power angle signal of each generator is converted into the time-frequency characteristic matrix by using the fast S-transform.The matrix feature is extracted by 2DPCA and grouped by self-organizing neural network.The example of IEEE-39 node system and Ghana’s actual grid system show that this method can identify coherent generators accurately with the characteristics of eliminating the influence of noise effectively and extracting the time-frequency characteristics of power angle information fully.Secondly,a new locating method of oscillation centers is proposed based on the measured impedance motion trajectory for oscillation center location of a system with impedance irregularity and potential amplitude inequality.The model of measured impedance is established and then the measured impedance trajectory is obtained.Based on measurement of the impedance trajectory,a new out-of-step oscillation recognizing method is proposed.The locating method of oscillation center position is presented by using the vector product to construct the oscillation center position discriminant formula.The position of the out-of-step center is determined by comparing the relationship between the radius of the impedance track and the distance between the center and the special point.The simulation results validate effectiveness of the presented method.Finally,an interconnected large-scale power grids islanding scheme which takes into account the coherent generators,the location of the oscillation center and the unbalanced power of the islands is given.For islanding of inter-regional interconnected power grids,monitoring the oscillation center position of inter-regional contact line,splitting the power system at the cut surface composed of the inter-regional tie-lines where the oscillation center stays can generally eliminate oscillation quickly.For islanding of a regional power grid,the reactance value between the nodes is used as the line weight,and the Dijkstra algorithm is used to calculate the minimum reactance value between the generator in the coherent generators group and get the coherent generators connection diagram.The prim algorithm is then used to get the minimum spanning tree of the coherent generators connection diagram to ensure connectivity of the generators in the island.By equivalizing the minimum spanning tree to a node,the Dinic algorithm is used to obtain the minimum power flow cutset between non-coherent generators groups and the optimal solution of the controlled islanding surfaces.The simulation of the IEEE-39 bus system verify that the presented method can determine the position of islanding surfaces quickly and accurately. |
PDF Full Text Request