Detection And Recognition Algorithm On Power Quality Disturbances Based On Mathematical Morphology

With the development of society, the degree of automation in every business isincreasing more and more rapidly, various types of electrical equipment in use is alsoincreasing more and more rapidly. Therefore, the impact on power will cause a variety ofvoltage sags, voltage swells, voltage fluctuation, oscillatory transients and other power qualityproblems. Operation security of power systems will be caused incalculable harm by Thesedisturbances. Therefore, the quickly detection for the issue of power quality disturbances willenable us to make effective action to stabilize our power system immediately when thedisturbance occurs. To make effective action, We have to know what kind of disturbingsignals it is firstly. Therefore, the recognition and classification of the disturbance are alsovery important. Mathematical Morphology is a mathematical theory, which is newly used innonlinear signal processing, with the advantages of simple and effective and fast operation.Therefore, the focus of this paper is to study power quality disturbance signal detection andrecognition based on mathematical morphology.Firstly, the power quality disturbances classification signal were introduced, and itsmathematical model is simulated digitally in Matlab software. Corresponding analog signalswere obtained for post-processing, and the theory of mathematical morphology is introduced.At the time as a de-noising tool, the theory is used for constructing an adaptive combinationof morphological filters, and the filtering effect is verified for the signal filter.Secondly, the mathematical morphology theory is applied for the detection of startingand ending time of transient power quality disturbances. Firstly an adaptive combined filteris constructed, noisy signal disturbance has been filtered. Secondly, the disturbance signal ischanged for morphological transformation, therefore, morphology edge detection operator isobtained.The operator is transformed by TOP-HAT. we get the characteristic signal bythresholds processing. The method was validated in different contexts, such as temporarydecline in value, harmonics, noise. The simulation results show that this method has a highcomputing speed, which can quickly locate the starting and ending time the disturbance.Furthermore, the combination of mathematical morphology and entropy theory is usedfor the detection of transient power quality disturbance signal. Firstly, morphological filter isused for signal denoising. Then a simplified concept of entropy is proposed. The filteredsignal is converted to Strike eigenvalues by the theory of simplify entropy. And The finalresult can be obtained by threshold processing, which can Charactery the start and end time ofthe disturbance. The method was validated in different contexts,such as temporary decline invalue, harmonics, noise. The simulation results show that This method can quickly and effectively locate the disturbance start and end time.Finally, the combination of mathematical morphology and dq transform and Euclideandistance and dynamic time warping algorithm is used for single disturbance recognition andclassification. First, the noisy disturbance signal is processed by morphological filters, Signalcomponent is configured to form of dq component, which is referring to the power system. Bythe expressions we can find that dq components is only related to the magnitude and the noiseat the moment. When the noise is filtered out, the components is only related to the magnitudecomponent. Therefore, the square of the amplitude of the dq components can be certaincharacterized for the characteristics of the signal. In order to identify the type of signal,Euclidean distance is used for istance calculation of dq-converted squared amplitude for themeasured signal and the reference signal. Optimal path is obtained by dynamic time warpingalgorithm. The matched corresponding reference signal is the category of the measured signal.The disturbance signal is recognized in the circumstances of various amplitude and differentbackground noise. The simulation results show that this method can accurately identify alltypes of single disturbance.