Research On Fault Detection,Location,and Isolation In The Power Networks Using Wide-area Information

The wide-area measurement system provides real-time wide-area information for online status monitoring,protection,and control of power systems.The acquisition of wide-area information provides important technical support for the decision-making system to coordinate and cooperate between the local and global power networks,and is conducive to making decisions that are beneficial to the entire power system.Based on this,the fault detection,location,and isolation methods in the large-scale power systems using wide-area measurement information are proposed to provide systemwide,cost-effective,and robust wide-area schemes in large-scale power systems.The main research works of this paper can be summarized as:(1)To develop a robust fault detection method for transmission lines based on waveform similarity.A robust fault detection method based on Kendall’s Tau coefficient is proposed to effectively suppress the influence of abnormal data on faulted line identification.Kendall’s Tau coefficient is employed to measure the waveform difference between the faulted line and the healthy line.The algorithm can constrain the influence of the distorted current amplitude on the protection criterion,and is also effective for the interference of multiple abnormal data.The simulation verification carried out on the IEEE 39 node test system shows that the proposed protection algorithm does not require strict data synchronization,and has a fast speed of action and resistance to fault resistance.It also exhibits good protection performance in CT saturation,noise,and abnormal data interference.(2)To propose a wide-area fault detection method with generating samples in power grids based on the ACWGAN-gp network.The node voltage signals at each PMU are encoded as grayscale images,and thus the problem of faulted line identification is transformed into an image classification problem.The fault area is first determined in the proposed scheme,and then the faulted line is identified by the ACWGAN-gp-based faulted line detector at a second stage.A significant aspect of the proposed method is that it is capable of generating new high-quality samples with specified class labels to extend the original dataset consisting of a limited number of real samples and thus improving the accuracy of the proposed fault detection classifiers.(3)A robust wide-area fault location method for the transmission line is proposed considering traveling wave velocity and measurement uncertainty.The traveling wave velocity is considered as a quantity to be solved in the fault location formulation,thus eliminating the effect of wave velocity error on the fault location accuracy.Also,the coefficient matrix is no longer a constant,but an error-affected matrix after considering the uncertainty of traveling wave velocity.Thus,a classical least squares(LS)model for the fault location is replaced with an Errors-in-variables(EIV)Model.The iteratively reweighted Total Least Squares(TLS)method is introduced as a robust estimation of fault location to restrict the influence of outliers and errors in the EIV models.Simulation tests were examined on the Lab VIEW platform designed for fault location,which can send fault location results to remote clients.(4)A novel fault isolation scheme for power grids with dynamic topology changes based on wide-area information is proposed.The flexible and dynamic operating characteristics of the power system drive the fault isolation scheme to have sufficient adaptability.To do this,the wide-area switch status information is used to develop a fault isolation scheme systemwide.Based on graph theory,the real-time adjacency matrices for electrical components and circuit breakers are constructed to reflect dynamic changes in networks topology.The Floyd-Warshall algorithm is employed to search the tripping paths of circuit breakers to obtain the fault isolation set.The proposed scheme performs fault isolation from a wide-area perspective,which is immune to system coordination and selectivity compared to local information-based methods.The effectiveness of the scheme is verified on a power system with typical electrical main wiring under a single fault,multiple faults,circuit breaker failure,the outage of substation DC power,and non-trip scenarios.The tripping paths of circuit breakers are determined before the fault occurs;thus,it can minimize the fault isolation zone with low fault clearing time even if the topology of the network changes dynamically.(5)The identification method of the last circuit breakers in the hybrid AC/DC power grids based on the improved Tarjan algorithm is proposed.To protect electrical equipment from overvoltage threats caused by unexpected tripping of circuit breakers,the last circuit breaker protection is equipped in hybrid AC/DC power grids.The proposed improved Tarjan algorithm solves the problem that the traditional Tarjan algorithm cannot obtain the path’s cut vertices between any two vertices in an undirected connected graph.And then,the proposed algorithm is effectively employed for identifying the last circuit breaker in the AC field instead of formulating a look-up table for various network topologies in advance.The feature of the proposed algorithm is that it can identify the last circuit breaker of each converter valve at the same time when all the vertices in the undirected graph are traversed only once,which improves the operation efficiency.The real-time verification is carried out on the developed closed-loop simulation platform based on Lab VIEW and MATLAB,and the results show that the proposed method has a lower and more stable running time compared with the method based on depthfirst search,showing its application prospect in fast real-time protection.