Filter Designing Applied For Power System Oscillation Parameter Identification Using PMU Data

With the continuous expansion of the scale of power grid interconnection,the interconnected power grid plays an active role in improving environment,ensuring power supply,promoting the development of clean energy.However,various kinds of power oscillation accidents such as ultralow frequency oscillations,low frequency oscillations and subsynchronous oscillations are also increasing the threat to the safe and stable operation of power system.Accurate parameter identification of oscillation is the base and premise of suppressing the power oscillation in power system and has important research significance.Prony algorithm in common use can directly calculate oscillation parameter which has preferably fitting result for the ideal power oscillation signal.However,the power oscillation data obtained by PMU contain a certain degree of white Gaussian noise,reducing the degree of accuracy of Prony algorithm for identifying the oscillation parameters.Therefore,this thesis designs filter applied for PMU power oscillation data to improve the accuracy of Prony algorithm for identification of the oscillation parameters based on the digital filter and wavelet transformation.Firstly,based on the micro variations of electric power system,the mathematical models of low oscillator signal is derived and the applicability of Prony algorithm is demonstrated.A pretreatment process is designed for abnormal data and Butterworth low pass filter is designed for white Gaussian noise which cannot be eliminate during pretreatment process.The compensating methods of amplitude and phase is proposed and the affect of filter is verified under 20 dB noise-signal ratio.Then,in order to solve the applicability of Prony algorithm under alternative oscillation parameters,the wavelet transformation is introduced to identify mutation position of oscillation parameters,so as to segment PMU low frequent oscillation data.The filter based on the micro wave transformation can be designed by three different filtering schemes under the circumstance that multiscale analysis alternated from micro wave transformation,combining with different features of power oscillation signals and white Gaussian noise.It can be verified under the certain noise-signal ratio and appropriate filtering parameters.Finally,Butterworth low pass filter and the filter based on micro wave transformation is introduced into the subsynchronous oscillation and ultralow frequency oscillation.Because of the wide band of minor synchronous oscillation,Butterworth band pass filter is used and the simultaneously compensation formula of amplitude and phase is amended.The filtering parameters is also amended in the filters based on micro wave transformation.In addition,the application in low frequency oscillation of Butterworth low pass filter,the identification of mutation position of oscillation parameter from wavelet transformation,the application of low frequency oscillation of filter based on micro wave transformation,all three algorithms have been realized in the software program and examples are given.