CN-122020141-A - Signal frequency analysis method and system for unmanned aerial vehicle racing motion

CN122020141ACN 122020141 ACN122020141 ACN 122020141ACN-122020141-A

Abstract

The invention relates to the technical field of wireless communication, in particular to a signal frequency analysis method and a system for racing movement of an unmanned aerial vehicle, wherein the method comprises the steps of establishing a three-dimensional coordinate system of a track and collecting historical data of multiple training flights to construct a historical window set; dividing the historical windows with similar characteristics into the same area according to the motion response characteristics of each historical window in the historical window set, traversing candidate wavelet decomposition layers for each area and calculating a denoising effect score to determine the optimal decomposition layer number of each area, collecting flight data of the current window in real time and calling the corresponding optimal decomposition layer number to carry out self-adaptive wavelet denoising after matching the area, and extracting instantaneous frequency to complete signal frequency analysis. The invention improves the accuracy and adaptability of instantaneous frequency extraction and overcomes the interference problem caused by the non-stable change of signals in the complex channel environment.

Inventors

LU SHIDONG
YU LIMIN
NI GUANGYAO

Assignees

浙江云朵航空科技集团股份有限公司
浙江比翼智慧科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (8)

1. The signal frequency analysis method for the racing motion of the unmanned aerial vehicle is characterized by comprising the following steps of: establishing a three-dimensional coordinate system of the track and collecting historical data of training flights for a plurality of times to construct a historical window set; dividing the history windows with similar motion response characteristics into the same area according to the motion response characteristics of each history window in the history window set; traversing the candidate wavelet decomposition layer number for each region and calculating a denoising effect score to determine the optimal decomposition layer number of each region; Collecting flight data of a current window in real time, matching the flight data with the area, then calling the corresponding optimal decomposition layer number to perform adaptive wavelet denoising, and extracting instantaneous frequency to complete signal frequency analysis; Wherein, the determination of the optimal decomposition layer number is as follows: For each area, carrying out wavelet denoising on the received signals of each history window in the area by adopting the candidate wavelet decomposition layer number and restoring the instantaneous speed of the unmanned aerial vehicle, calculating the denoising effect score of each candidate wavelet decomposition layer number according to the statistical result of the speed restoring error between the restored instantaneous speed and the real speed, and determining the candidate wavelet decomposition layer number with the largest denoising effect score as the optimal decomposition layer number of the area.
2. The method of claim 1, wherein collecting historical data of a plurality of training flights to construct a set of historical windows comprises: Setting sampling frequency, synchronously acquiring a real flight speed sequence, a receiving signal sequence, a triaxial angular speed sequence and a triaxial acceleration sequence of the unmanned aerial vehicle in multiple training flights, and carrying out normalization processing on acquired data; Setting the length of a sliding window and the sliding step length, and sliding and intercepting a plurality of history windows on the time sequence of each training flight to construct a history window set.
3. The method of claim 2, wherein dividing the history window of similar characteristics into the same region comprises: Extracting a triaxial angular velocity sequence and a triaxial acceleration sequence of each historical window, calculating the mean value and standard deviation of each axial angular velocity sequence and the mean value and standard deviation of each axial acceleration sequence, and splicing the calculated statistical features to form a motion feature vector of the historical window; and classifying by using the motion characteristic vectors of all the history windows as samples and adopting a clustering algorithm, merging the history windows with similar motion response characteristics into the same cluster, wherein each cluster corresponds to a race track area.
4. A method of signal frequency analysis for a racing motion of a drone according to claim 3, wherein calculating a denoising effect score comprises: Performing wavelet decomposition on the received signal sequence of the history window by using the candidate wavelet decomposition layer numbers to obtain a low-frequency component and each layer of high-frequency component, and calculating the energy confusion of the received signal according to the energy duty ratio of each component; determining a judging threshold value based on the energy chaos degree, performing hard threshold processing on each layer of high-frequency components by utilizing the judging threshold value, performing wavelet inverse transformation on the processed high-frequency components and original low-frequency components to obtain a denoised received signal sequence, extracting instantaneous frequency through Hilbert transformation, converting the instantaneous frequency into a restored instantaneous speed, and calculating a speed restoration error between the restored instantaneous speed and the real speed; and calculating the mean value and standard deviation of the speed reduction errors of all the history windows in the statistical region under the corresponding candidate wavelet decomposition layers, and calculating according to the mean value and standard deviation to obtain the denoising effect score.
5. The method of claim 4, wherein determining the decision threshold based on the energy confusion comprises: and determining the lower limit and the upper limit of the judging threshold, wherein the lower limit of the judging threshold is set to be a value corresponding to the low-order digits of all the high-frequency component absolute values, the upper limit of the judging threshold is set to be a value corresponding to the high-order digits of all the high-frequency component absolute values, and the judging threshold takes a larger value of the lower limit of the judging threshold and the upper limit of the judging threshold multiplied by the negative energy confusion degree power based on a natural constant.
6. A method of analyzing the frequency of a signal of a racing movement of an unmanned aerial vehicle according to claim 3, wherein collecting the flight data of the current window in real time and matching the area comprises: And acquiring a triaxial angular velocity sequence, a triaxial acceleration sequence and a received signal sequence of a current window formed by continuous time periods ending at the current moment in real time, extracting motion feature vectors according to the triaxial angular velocity sequence and the triaxial acceleration sequence of the current window, calculating the distance between the extracted motion feature vectors and central vectors of all areas, and selecting the area with the minimum distance as a matching area of the current window.
7. The signal frequency analysis method for the racing motion of the unmanned aerial vehicle according to claim 1, wherein the candidate wavelet decomposition layer number is generated after determining a maximum decomposable layer number according to sampling points in a history window, and the maximum decomposable layer number is formed by rounding down the logarithm of the sampling points based on 2.
8. A signal frequency analysis system of a racing movement of a drone, comprising a processor and a memory storing computer program instructions which, when executed by the processor, implement the signal frequency analysis method of a racing movement of a drone according to any one of claims 1 to 7.

Description

Signal frequency analysis method and system for unmanned aerial vehicle racing motion Technical Field The invention relates to the technical field of wireless communication. More particularly, the invention relates to a signal frequency analysis method and system for a racing motion of an unmanned aerial vehicle. Background The unmanned aerial vehicle race is a high-speed and high-dynamic competitive motion, when the unmanned aerial vehicle runs in a complex race track at a high speed, a received signal is obviously influenced by Doppler effect, and the instantaneous frequency of the signal is closely related to the real-time speed of the unmanned aerial vehicle. Therefore, the Doppler frequency shift information is extracted by carrying out frequency analysis on the received signals, and the method is a key technical means for restoring the instantaneous speed and the position of the unmanned aerial vehicle. The existing signal frequency analysis method mostly adopts wavelet transformation to carry out denoising treatment on a received signal, and then combines Hilbert transformation to extract instantaneous frequency. However, the motion characteristics and the channel environment differences of different areas of the unmanned aerial vehicle racing track are large, and in the signal frequency analysis process of the existing method, the fixed wavelet decomposition layer number is adopted to uniformly process the full-track signal, so that the accuracy and the stability of frequency extraction are difficult to be compatible. Therefore, how to improve the adaptive capacity of signal frequency analysis in the racing motion of the unmanned aerial vehicle is a technical problem to be solved in the art. Disclosure of Invention In order to solve the technical problem of how to improve the self-adaptive capability of signal frequency analysis in the racing motion of the unmanned aerial vehicle, the invention provides a scheme in the following aspects. In a first aspect, a method for analyzing signal frequency of racing motion of a drone includes: establishing a three-dimensional coordinate system of the track and collecting historical data of training flights for a plurality of times to construct a historical window set; dividing the history windows with similar motion response characteristics into the same area according to the motion response characteristics of each history window in the history window set; traversing the candidate wavelet decomposition layer number for each region and calculating a denoising effect score to determine the optimal decomposition layer number of each region; Collecting flight data of a current window in real time, matching the flight data with the area, then calling the corresponding optimal decomposition layer number to perform adaptive wavelet denoising, and extracting instantaneous frequency to complete signal frequency analysis; Wherein, the determination of the optimal decomposition layer number is as follows: For each area, carrying out wavelet denoising on the received signals of each history window in the area by adopting the candidate wavelet decomposition layer number and restoring the instantaneous speed of the unmanned aerial vehicle, calculating the denoising effect score of each candidate wavelet decomposition layer number according to the statistical result of the speed restoring error between the restored instantaneous speed and the real speed, and determining the candidate wavelet decomposition layer number with the largest denoising effect score as the optimal decomposition layer number of the area. Optionally, collecting historical data for a plurality of training flights to construct a set of historical windows includes: Setting sampling frequency, synchronously acquiring a real flight speed sequence, a receiving signal sequence, a triaxial angular speed sequence and a triaxial acceleration sequence of the unmanned aerial vehicle in multiple training flights, and carrying out normalization processing on acquired data; Setting the length of a sliding window and the sliding step length, and sliding and intercepting a plurality of history windows on the time sequence of each training flight to construct a history window set. Optionally, dividing the history window of similar characteristics into the same region includes: Extracting a triaxial angular velocity sequence and a triaxial acceleration sequence of each historical window, calculating the mean value and standard deviation of each axial angular velocity sequence and the mean value and standard deviation of each axial acceleration sequence, and splicing the calculated statistical features to form a motion feature vector of the historical window; and classifying by using the motion characteristic vectors of all the history windows as samples and adopting a clustering algorithm, merging the history windows with similar motion response characteristics into the same cluster, wherein each cluster corresponds to a