Study On Intelligent Detecting Algorithms Of Power Quality And Its Applications

The development of science & technology and the national economy is proposing more demands on electric power as well as higher requirements on power quality (PQ). As an issue that draws extensive world-wide attention, the power quality is of crucial importance for guaranteeing the safe and economical running of the power network and device,elevating the overall efficiency of the economy, promoting product qualities,and maintaining daily life.For improving the power quality, it is necessary to systematically analyze and study the problems in PQ,and find out the causes of the problems before any proper countermeasure is to be developed.The detection and recognition of PQ is the first difficulty to overcome in this direction.It also provides reasonable basis for the power sensitive load programming and PQ evaluation, prediction,maintenance and management.Therefore,it is of significant theoretical and practical importance to research PQ intelligent detection algorithms,and construct PQ detection-analysis system for the detection, evaluation and classification of PQ.This dissertation, sponsored by the National Natural Science Foundation project "Improving FFT analysis method of dynamic signal and the application in electrical harmonic detection"(Grant No.60872128)and Guangxi Electric Power Science & Technology Foundation project "Virtual power parameter automatic test system",is focused on the practical requirements of PQ parameter analysis,and in-depth researches have been carried out on the harmonic analysis algorithm, the classification and recognition of multiple PQ disturbances method,voltage flicker measurement algorithm, and characteristics extraction for voltage sags and the recognition of sag sources.An intelligent PQ detection system using virtual instrument technologies is developed.Main researches are as follows.1) The definition, classification and standards of PQ and PQ parameters are introduced.PQ relevant literatures such as harmonic analysis,classification and recognition of PQ disturbances,measurement of voltage flickers,and voltage sags analysis are reviewed, cons and pros of the current research work are summarized, and the research orientation of the dissertation is given.2) The key factors relevant to the FFT-based harmonic analysis accuracy are analyzed.To suppress the spectral leakage and the picket-fence effects,the Nuttall self-convolution window of high performance is proposed.With theoretical results on frequency characteristics of the Nuttall self-convolution window, the improved FFT-based harmonic analysis algorithm based on the window is proposed.By utilizing least-square polynomial fitting method,the interpolating polynomial for discrete frequency spectrum and the calculation of harmonic parameters are derived. Simulation researchs show that the computation of the fundamental and harmonic parameters using the novel method is simple,and the interferences between harmonics can be effectively suppressed,thus resulting in more accurate measurement of fundamental and harmonic parameters.3) To achieve PQ multiple disturbances recognition,classification of PQ multiple disturbances based on the S-transform and probabilistic neural network is proposed. Basics of the S-transform are introduced.The S-transform is applied for the time-frequency analysis of PQ disturbance signals,from whose results the features of the signal are extracted.Using these features as inputs,a probabilistic neural network is trained for disturbance classification.Simulation results show that the method is correct and effective.Furthermore,a novel PQ multiple disturbance classification method using S-transform data from different stages of the transform is given, which increases classification accuracy without introducing extra computational load.4) To achieve voltage flickers measurement and time-frequency analysis,a new voltage flicker measurement algorithm using the square demodulation method based on S-transform is formulated and realized.Based on simplifing the voltage flicker measurement propocess using the square demodulation method, the mean square root of the voltage signal is submitted to the S-transform,whose frequency domain data is used to calculate the frequency spectrum and later on the magnitude of the voltage flicker. The sum of high frequency part of the transform result gives the moments when the flicker starts and ends.Simulation results show the method is of a reasonable accuracy of flicker measurement, and the starting and ending time of flickers can be determined precisely.5) The characteristics extraction and sag source recognition method of voltage sag based on the generalized S-transform is provided.The fundamental magnitude, fundamental phase matrix and high frequency spectrum magnitude are obtained by applying the generalized S-transform on the voltage signal,and are used to compute the magnitude,the phase jump and the duration of the voltage sag.From the generalized S-transform result, four features,namely, the concavity and convexity of the magnitude curve,harmonics,phase jump,and the number of magnitude step changes,are extracted for categorizing the voltage sag using a decision tree approach. Simulation results show that the method is correct and effective.6) Based on the LabVIEW software development platform and relevant hardware, a PQ intelligent detection system integrating the above new algorithms is designed. The overall structures of the system software and hardware and the system operation flowcharts are given.System modules are analyzed in details.Algorithms and the corresponding implementations for the system modules are provided.In-spot running and tests at clients such as the Measurement Center of Hunan Electric Power Corporation and Nanning Municipal Bureau of Power Supply show that the system has features of user-friendly human-machine interface,convenient operation and reliable measurement. The system can detect the PQ stable-state parameters, demonstrate the ability of being adaptive to the real power network circumference, and capture as well as recognize the PQ disturbances at the test point.