Research On Frequency Measurement Algorithms And Evaluation On Their Measurement Accuracy

Frequency is the most basic and important parameter that can describe the characteristics of a sinusoidal signal.Frequency estimation of a sinusoidal signals in noisy is not only a classical problem in theoretical research fields such as electronic communications and signal processing,but also a necessary tool in some practical engineering applications such as radar detection,monitoring of power systems,early warning of natural disasters and biomedical monitoring.Therefore,algorithms for frequency estimation of a sinusoidal signal has important theoretical research significance and practical application value.According to the development status about algorithms for frequency estimation of a sinusoidal signal,this paper broadly classifies them as maximum likelihood estimation method,spectral estimation method,time domain analysis method and frequency domain analysis method,etc.It selects spectral estimation method,zerocrossing method,nonlinear curve fitting method,interpolation method based on Fourier transform,phase difference method and inverse function method to study and simulate,investigates the effects of noise,frequency range,sampling frequency and phase on the results and performance on frequency estimations and makes individual or whole evaluations on the performance of each method,such as stability,accuracy and precision of the estimation results with the help of a large amount of simulation data,data processing and error analysis methods.It also introduces new parameters into each method,like sampling length of a segment of the signal and sampling interval between multiple segments of the signal,finds that both of them not only influence the results of the frequency estimation on each method,but also make the trend of its performance exhibit a clear digital multiplicative relationship,and then analyzes and explains the causes of these phenomena.In addition,based on zerocrossing method,this paper improves zerocrossing method into peakinterval method,take it to participate in the comparison and evaluation of the estimation performance.