Signal Instantaneous Frequency Estimation Method And Its Applications

Instantaneous frequency is a characteristic of the phenomenon essential to express the time-varying non-stationary signals and has been widely applied to radar, sonar, mobile communication and so on. So the research about its methods to estimate will be valuable in theory and practice.In this thesis, two signals are divided into: signals with frequency varying in small range (such as power grid signals) and signals with frequency varying in large range, according to the varying range of frequency and applying fields. First worldwide developments in instantaneous frequency fields are comprehensively summarized including research background, definition, and some concepts. With respect to the two signals, in-depth research about the estimating methods and applications are carried out from different respects.With regard to signals with small varying range, taking power grid signals for example, based on orthogonal transform, a new method of tracking power grids frequency has been provided. Simulations prove the method can tracking frequency accurately and instantly, and have the preferable effectiveness to the abruptly fluctuating-frequency power signals; at the same time, this method overcome the constrains of synchronous sampling, and have the better astringency with different initial frequency. But the effect is not good enough to track the varying slow signals, without the ability of astringency and antinoise.Based on dual orthogonal digital filters and weighted smoothing phase difference method, an approach for tracking power system frequency is proposed. With respect to several typical models of power signals, detailed simulations have been carried out to verify the algorithm's dynamic characteristics, and analyze the impact factors on tracking error and time delay are analyzed accordingly. The research results show that, the algorithm is not subjected to the synchronized sampling criteria and can track the instantaneous frequency of different kinds of power signals accurately and instantly, presenting preferable accuracy and effectiveness to the ones available in previous similar literatures.Three different approaches are adopted for compute phases of three typical power signals to track power grids frequency. The simulation results are detailed analyzed and made into more full-scale comparison. According to synthesizing tracking precision, computing speed and antinoise and so on, the thesis compares and analyzes three different approaches to track power grids frequency and points out approach based on dual orthogonal digital filters has better tracking character and practical value.With respect to signals with big range, MUSIC (Multiple signal classification) method and time-frequency distribution (TFD) method are adopted to estimate then-frequencies, taking frequency modulating signals in the communication, radar fields.Based MUSIC method, a new approach for estimating instantaneous frequency of linearity frequency modulating signals (Chirp signal) and simulations are carried out. The results prove this approach has higher precision and better antinoise. But it only is applied into mono-component signals and demands higher signal-to-noise environment. So, another way of time-frequency distribution method is sought for to research. With respect to the difficulties of estimating multi-component signals, this paper provides a new method to separate every component with crossings. It takes every component in short time for varying linear, takes TFD for image and then detects the components around crossings in Hough transform while Wigner-ville distribution (WVD) with good time-frequency assembling is used to detect the components away from the crossings. Distance peak-orientation and peak threshold methods are adopted in simulations. Results prove this method has better detection and is able to detect all components in low SNR, according to setting off appropriate areas.