CN-121995355-A - Signal processing method of anti-interference acoustic phased array Doppler log

CN121995355ACN 121995355 ACN121995355 ACN 121995355ACN-121995355-A

Abstract

The invention discloses an anti-interference acoustic phased array Doppler log signal processing method which comprises the steps of constructing a standard echo signal matrix, dividing the constructed standard echo signal matrix into K non-overlapping frames, constructing a beam output time sequence, executing short-time Fourier transform on the beam output time sequence to obtain a time spectrum matrix, extracting J frequency shift parameters from the time spectrum matrix, clustering G frequency shift parameter clusters, constructing a speed observation vector for carrier speed estimation, inputting the speed observation vector into a pre-trained time sequence denoising network to output an anti-interference speed sequence, performing time integration on the anti-interference speed sequence to construct an accumulated range of a carrier in a measurement period, and continuously outputting reliable range information in a complex underwater sound environment because the accumulated range obtained by time integration is not jumped or interrupted due to individual abnormal frequency shift.

Inventors

LI XIANG
WU QILIN
TIAN TAO

Assignees

巢湖学院

Dates

Publication Date: 20260508
Application Date: 20260227

Claims (8)

1. The signal processing method of the anti-interference acoustic phased array Doppler log is characterized by comprising the following steps of: S1, constructing a standard echo signal matrix of the Doppler log in the direction of expected beam pointing; s2, dividing the standard echo signal matrix into K non-overlapping frames along the sampling time direction of the standard echo signal matrix; s3, constructing beam output sub-sequences corresponding to the K non-overlapping frames respectively, and splicing the beam output sub-sequences into a complete beam output time sequence; S4, performing short-time Fourier transform on the beam output time sequence to obtain a time spectrum matrix; S5, searching local energy maxima along a frequency axis of the time-frequency spectrum matrix, and extracting J frequency shift parameters; s6, performing density clustering on J frequency shift parameters to generate G frequency shift parameter clustering clusters; s7, constructing a speed observation vector for carrier speed estimation according to the G frequency shift parameter clustering clusters; S8, inputting the speed observation vector to a pre-trained time sequence denoising network, and outputting an anti-interference speed sequence; s9, performing time integration on the anti-interference speed sequence to construct an accumulated course of the carrier in the measurement period.
2. The method of claim 1, wherein constructing a standard echo signal matrix for the doppler meter in the desired beam pointing direction comprises: S1-1, synchronously acquiring echo signals at M sampling moments based on an acoustic phased array of N hydrophones; S1-2, arranging echo signals acquired by each hydrophone at M sampling moments in time sequence, and constructing N original echo signal sequences; s1-3, aligning the N original echo signal sequences according to sampling time to form an original echo signal matrix; S1-4, acquiring relative time delay offset among channel signals in an original echo signal matrix; S1-5, performing time delay compensation on each channel signal in the original echo signal matrix based on the time delay offset, and constructing a standard echo signal matrix.
3. The method of claim 1, wherein the stitching is a complete beam output timing sequence, comprising: S3-1, calculating covariance matrixes of the non-overlapped frames; s3-2, calculating a corresponding beam weight vector based on covariance matrixes of the non-overlapping frames; S3-3, respectively acting each beam weight vector on a corresponding non-overlapping frame to generate K beam output subsequences; and S3-4, splicing the K beam output subsequences based on the sampling time sequence to generate the beam output time sequence.
4. The method of claim 3, wherein calculating the corresponding beam weight vector based on the covariance matrix of each non-overlapping frame comprises: S3-2-1, constructing a corresponding guiding vector according to a preset expected beam pointing direction; s3-2-2, calculating MVDR beam weight vectors corresponding to the kth non-overlapping frame based on the covariance matrix and the inverse matrix thereof and combining the steering vectors until K MVDR beam weight vectors are obtained; S3-2-3, normalizing the K MVDR beam weight vectors to generate beam weight vectors corresponding to the non-overlapping frames.
5. The method of claim 4, wherein searching local energy maxima along the frequency axis of the time-frequency spectrum matrix, extracting J frequency shift parameters, comprises: s5-1, calculating the energy value of each time frame in the time-to-frequency spectrum matrix on each frequency point; S5-2, traversing all frequency points in each time frame, and identifying frequency points meeting local maximum conditions; s5-3, taking frequency values corresponding to all frequency points meeting local maximum conditions as candidate frequency shift parameters; and S5-4, summarizing candidate frequency shift parameters of all time frames, and constructing a set containing J frequency shift parameters.
6. The method of claim 5, wherein density clustering J frequency shift parameters to generate G frequency shift parameter clusters, comprising: S6-1, marking access states of J frequency shift parameters as unaccessed, and predefining a neighborhood radius and a minimum neighborhood point number; s6-2, judging core frequency shift parameters of the frequency shift parameters with J access states marked as unaccessed, and constructing a core frequency shift parameter set; S6-3, traversing each core frequency shift parameter in the core frequency shift parameter set; S6-4, if the access state of the core frequency shift parameter is not accessed, a frequency shift parameter cluster is newly built; S6-5, adding the core frequency shift parameter into the newly built frequency shift parameter cluster, and updating the access state of the newly built frequency shift parameter cluster to be accessed; S6-6, recursively searching all frequency shift parameters with reachable density from the core frequency shift parameters, adding all frequency shift parameters with reachable density into the newly built frequency shift parameter cluster, and updating the access state of the frequency shift parameter cluster to be accessed; the density is defined as that if a frequency shift parameter can be connected to the core frequency shift parameter through a series of adjacent frequency shift parameters, and the frequency deviation between any two adjacent frequency shift parameters in the path does not exceed the neighborhood radius, the frequency shift parameter is regarded as the density is reachable; s6-7, outputting a total of G frequency shift parameter clustering clusters after all the core frequency shift parameters are processed.
7. The method of claim 6, wherein performing a core frequency shift parameter decision on the J frequency shift parameters with access status markers as not accessed, constructing a core frequency shift parameter set, comprises: S6-2-1, selecting any frequency shift parameter with access state not accessed from J frequency shift parameters as a target frequency shift parameter; s6-2-2, calculating frequency deviation between the target frequency shift parameter and the rest J-1 frequency shift parameters; S6-2-3, if the frequency deviation is not larger than the neighborhood radius, marking the frequency shift parameter corresponding to the frequency deviation as a neighborhood point, otherwise marking the frequency shift parameter as a non-neighborhood point; S6-2-4, traversing J-1 frequency deviations until all neighborhood points of the target frequency shift parameters are marked; s6-2-5, counting the number of neighborhood points of the target frequency shift parameter; s6-2-6, if the number of the neighborhood points is not smaller than the minimum neighborhood point number, marking the target frequency shift parameter as a core frequency shift parameter; s6-2-7, traversing J frequency shift parameters until all core frequency shift parameters are marked, and obtaining a core frequency shift parameter set.
8. The method of claim 1, wherein constructing a velocity observation vector for carrier velocity estimation from G clusters of frequency shift parameters comprises: S7-1, calculating the center frequency of each cluster, and taking the center frequency as cluster center frequency shift parameters of G candidate frequency shift parameters; s7-2, calculating the effective radial speed corresponding to each cluster center frequency shift parameter according to the cluster center frequency shift parameter; S7-3, splicing G effective radial speeds in time sequence to form a speed observation vector.

Description

Signal processing method of anti-interference acoustic phased array Doppler log Technical Field The invention relates to the field of echo signal processing of Doppler log, in particular to an anti-interference acoustic phased array Doppler log signal processing method. Background Currently, underwater acoustic doppler odometers commonly employ phased arrays to receive echo signals from a seafloor or water scatterer and calculate the radial velocity of the carrier relative to the environment through beamforming and doppler shift analysis. The typical implementation method comprises the steps of forming a receiving beam in a preset direction by utilizing fixed delay summation, performing short-time Fourier transform on beam output to obtain a time spectrum, determining Doppler frequency shift by detecting an energy peak value in the spectrum, converting the Doppler frequency shift into speed according to a Doppler formula, and obtaining an accumulated range through time integration. However, in a typical marine environment, echo signals are often severely affected by multiple sources of interference, strong reverberation causes spectrum broadening, adjacent scatterers induce Doppler shift aliasing, and platform self-vibration introduces spurious shifts that are not motion-dependent. Conventional doppler meters typically employ a fixed directional beam and a single peak detection strategy, which makes it difficult to distinguish between target echoes and interfering components. When the interference energy is close to or exceeds the target signal, the frequency shift estimation is easy to jump or even completely fail, so that the integral navigation path is suddenly changed, drifted or interrupted, and the reliability of underwater long-time navigation is severely restricted. Disclosure of Invention Aiming at the defects of the prior art, the invention provides an anti-interference acoustic phased array Doppler log signal processing method, and solves the technical problems in the background art by constructing an accumulated range which can be integrated by time. In order to achieve the above purpose, the invention is realized by the following technical scheme: The signal processing method of the anti-interference acoustic phased array Doppler log comprises the following steps: S1, constructing a standard echo signal matrix of the Doppler log in the direction of expected beam pointing; s2, dividing the standard echo signal matrix into K non-overlapping frames along the sampling time direction of the standard echo signal matrix; s3, constructing beam output sub-sequences corresponding to the K non-overlapping frames respectively, and splicing the beam output sub-sequences into a complete beam output time sequence; S4, performing short-time Fourier transform on the beam output time sequence to obtain a time spectrum matrix; S5, searching local energy maxima along a frequency axis of the time-frequency spectrum matrix, and extracting J frequency shift parameters; s6, performing density clustering on J frequency shift parameters to generate G frequency shift parameter clustering clusters; s7, constructing a speed observation vector for carrier speed estimation according to the G frequency shift parameter clustering clusters; S8, inputting the speed observation vector to a pre-trained time sequence denoising network, and outputting an anti-interference speed sequence; s9, performing time integration on the anti-interference speed sequence to construct an accumulated course of the carrier in the measurement period. In some specific embodiments, constructing a standard echo signal matrix for the doppler log in the desired beam pointing direction includes: S1-1, synchronously acquiring echo signals at M sampling moments based on an acoustic phased array of N hydrophones; S1-2, arranging echo signals acquired by each hydrophone at M sampling moments in time sequence, and constructing N original echo signal sequences; s1-3, aligning the N original echo signal sequences according to sampling time to form an original echo signal matrix; S1-4, acquiring relative time delay offset among channel signals in an original echo signal matrix; S1-5, performing time delay compensation on each channel signal in an original echo signal matrix based on the time delay offset, and constructing a standard echo signal matrix; in some specific embodiments, the splicing is a complete beam output timing sequence, including: S3-1, calculating covariance matrixes of the non-overlapped frames; s3-2, calculating a corresponding beam weight vector based on covariance matrixes of the non-overlapping frames; S3-3, respectively acting each beam weight vector on a corresponding non-overlapping frame to generate K beam output subsequences; and S3-4, splicing the K beam output subsequences based on the sampling time sequence to generate the beam output time sequence. In some specific embodiments, calculating the corresponding beam weight vector based