Research On The Harmonics Analyzing Algorithm Of High-Frequency Electrical Power Signal

Because of the high-speed characteristic,the usual operating frequency of high-speed motors is nearly as high as the swiching frequency of power devices,leading to lower carrier ratio.Therefore,harmonics related to the carrier frequency can be introduced into the whole system,which will increase the electromagnetic loss substantially and interfere the original magnetic field distribution.Harmonics analysis on voltage signals will make sense in looking for the specific connection with inverters and harmonics analysis on current signals will help strengthen the evalution of motor design,indicating the importance of harmonics analysis.This paper aims at launching researches on harmonics analysis algorithm whose objectives are electrical power signals of high-speed motors.The shortcomes of traditional Fourier analysis method is discussed theoritically in this paper,and a new method based on variable sampling frequency is proposed to tackle spectrum leakage effectively.To achieve better performance when analyzing harmonics and inter-harmonics,modern spectral estimation is adopted.Semi-simulation and high-speed moto test are carried out to obtain data for harmonics analysis,so that the validity of algorithms can be proved.Firstly,the traditional Fourier analysis method is discussed in theory,and the mechanisms of spectrum leakage and fence effect are analyzed in details.Differernt window functions are introduced to show the performance of different features in harmonics analysis.Interpolation method is adopted to correct the analysis results of windowed sample set.Secondly,this paper puts forward a new method based on adaptive variable sampling frequency and FFT.The parameterization model of the adaptive process is established to obtain the searching region of optimal sampling frequency.A fast searching method on the basis of dichotomy is proposed to reach the optimal sampling frequency in few steps.This method can reduce the impact of spectrum leakage to the most extent.Autoregressive spectral estimation,the most typical method in m odern spectral estimation is introduced to increase the resolution ability due to its high resolution.Through building parameter model of the signal,this method can gain the accurate frequencies of all the components in the signal.Combined with adaptive linear neural network,amplitude of components can be obtained fast and accurately.Ultimately,to verify the validity of algorithms in this paper,semi-simulation and high-speed motor test are carried out in details.Using function generator to output signals with specific parameters,and driving high-speed motor to operate at different speed,we can collect semi-simulation signals and actual power signals of the motor,respectively with JYTEK measuring instrument,and compare the performances on harmonics analysis when adopting different methods.