Research On New Protection Algorithms And Development Of SoC Based Digital Relay For Transmission Lines

With the rapid development of power system, significant changes have been made in manyaspects, such as the type of power sources, network size, voltage levels, control technologies,and load characteristics, which put more stringent requirements on the performance ofprotection system. The conventional protection schemes, which although have been developedand improved for quite a long time, have many shortcomings when facing the new changes inpower system. Therefore, there is a need for the next generation relay, which will be morereliable and faster than the conventional protection relays.Fourier transform have been widely applied to the phasor measurement and frequencyanalysis in power system for more than half a century. Nowadays, it is still the most commonsignal processing technique in practical application. However, Fourier transform is a periodicfunction based transform, whose performance suffers from the presence of the aperiodiccomponent in the input signal. In addition, it needs to be implemented in a long data window,which leads to heavy computational load and slow response speed. This thesis is concernedwith the development of a new generation of protection algorithms for power transimissionlines, based on Mathematical Morphology (MM), Singular Value Decomposition (SVD) andHigher Order Statistics (HOS). The research work focus on finding solutions to several issuesemerged in high voltage transmission lines, such as the distributed capacitive current, samplingmisalignment, current transformer saturation.Single-pole autoreclosure (SPAR) can be used to eliminate the temporary fault, and issignificant for improving the stability of power system. For the application of SPAR, the faultedphase should be identified as fast and accurate as possible firstly. Therefore, a reliable faultedphase selector is the basis for the troubleshooting. A new signal processing method, morphologysingular entropy (WSE), which has the merits of mathematical morphology and singular valuedecomposition, is developed for phase selection. The technique calculates the entropycorresponding to the samples of voltage signals in a very short window. The change in the valueof entropy represents the changes in the components of the imput voltage singal. Consideringthe coupling effect between phases in transmission line, a set of phase selection indices aredefined to achieve more reliable phase selection results. Data generated by using PSCAD/ EMTDC are used to test the performance of the proposed phase selector. The simulation resultshave demonstrated that the phase selector is of high sensitivity and acts very fast for single-phase faults and phase-to-phase faults. Even in the extreme case of high impedance faultoccurring at a low inception angle, the phase selector also has a good selectivity. Extensivesimulation studies with random fault parameters have showed that the proposed phase selectoroutperforms the conventional superimposed voltage phase selection scheme in terms ofaccuracy and response speed.Apart from the method mentioned above, a novel differential protection scheme ispresented for the protection of high voltage long distance transmission line. The proposedscheme detects faults in transmission lines by monitoring the changes in the skewness of thedifferential current, and discriminants the external faults from the internal faults based on theratio of the absolute values of the skewnesses of the differential current and restraining current.The new method is immune to the distributed capacitive current and Gaussian noise. Therefore,it does not need to compensate the distributed capacitive current, and can achieve a highreliability without the sacrifice of the sensitivity. Numerical simulations have verified that theproposed differential protection scheme has high sensitivity and fast response speed. Comparedwith the conventional phasor-based differential protection scheme, the new method has a betterperformance in detecting high impedance ground fault, and is less affected by the samplingmisalignment.The core of the current transformer may saturate during an external fault, which could leadsto the mal-operation of differential protection. To deal with this issue, a new algorithm basedon morphological gradient (MG) is presented for saturation detection. Two saturation detectorsare employed to detect the saturation located at the positive half cycle and the negative halfcycle, respectively. Two different unsymmetrical flat structuring elements (SE) are used toperform the morphological gradient in the two detectors, respectively. The beginning and theending of the saturation are determined by the local maximum values of the outputs. Afterwards,the parameters of the secondary current are estimated by applying Least Error Square (LES) tothe samples of the unsaturated portion, and the samples of the saturated portion can bereconstructed by using the parameters.This thesis also pays attention to the realization technologies of the protection algorithm. A system-on-a-chip (SoC) based platform is designed and built for the development ofprotection relay. The proposed skewness-based differential protection algorithm is realized intothe platform and the experimental results have verified that the design meets the requirements.At the end of the thesis, a systematic summary is given, and further work is suggested. Themajor contribution of the work in this thesis is the utilization of the advanced performance ofMM, SVD, HOS and SoC, for feature enhancement, extraction and noise suppression in signalsof power systems, which may lead to the development of the next generation of powerprotection relaying system.