A time delay estimation method based on quadratic correlation optimal weighting

Abstract

The present invention relates to a time delay estimation method based on quadratic correlation optimal weighting. By constructing an array signal based on the quadratic correlation function and a steering vector based on frequency and time delay, an optimal weight vector is obtained using a constrained optimization method. The optimal weight vector is used to perform optimal weighting on the time delay search to obtain the time delay difference between any two signals. The time delay estimation method combining quadratic correlation and optimal weight vector beamforming time delay estimation of the present invention improves the time delay estimation accuracy under low signal-to-noise ratio conditions; quadratic correlation can effectively suppress noise, and the optimal weight vector time delay estimation can greatly improve the time delay estimation accuracy without upsampling. The combination of the two points can effectively improve the time delay estimation accuracy under low signal-to-noise ratio conditions.

Description

Technical Field

[0001] The present invention relates to the technical field of computing; reckoning or counting, and particularly relates to a time delay estimation method based on quadratic correlation optimal weighting in the field of sonar signal processing. Background Art

[0002] Using the noise radiated by underwater targets to detect and locate targets is an important means for submarines to detect and attack covertly.

[0003] Currently, the main passive positioning methods include the triple array passive positioning method, the target motion analysis positioning method, and the matched field processing positioning method. Among them, the triple array passive positioning method is still the passive positioning method commonly used in submarine sonar equipment. The key is the estimation of two time delays between the signals received by three sub-arrays. The accuracy of the time delay estimation directly determines the positioning accuracy of the target.

[0004] Currently, the commonly used time delay estimation methods mainly include the direct correlation method, the generalized correlation method, the phase spectrum method, etc. The above methods cannot improve the positioning accuracy under the condition of a limited sampling rate. It is necessary to adopt upsampling or digital interpolation technology. At the same time, in practical applications, the problem of phase ambiguity is very likely to occur. The passive ranging ability of the triple array in the actual environment is poor, which is not conducive to the positioning of the target, the tolerance is relatively poor, and the accuracy of the time delay difference estimation is not high. Summary of the Invention

[0005] The present invention solves the problems existing in the prior art, and provides an optimized time delay estimation method based on quadratic correlation optimal weighting, which combines quadratic correlation and optimal weight vector beamforming time delay estimation to improve the time delay estimation accuracy.

[0006] The technical solution adopted by the present invention is a time delay estimation method based on quadratic correlation optimal weighting, which constructs an array signal based on a quadratic correlation function and a steering vector based on frequency and time delay, and uses a constrained optimization method to obtain an optimal weight vector, and performs optimal weighting on the time delay search with the optimal weight vector to obtain the time delay difference between any two signals.

[0007] The method includes the following steps:

[0008] Step 1: Construct the autocorrelation function R 11 and the cross-correlation function R 12 of any two signals x1 and x2;

[0009] Step 2: Use R 11 and R 12 as new input signals to construct a signal vector, decompose R 11 and R 12 into several narrowbands respectively, and calculate the frequency domain cross-spectrum matrix R of each narrowband based on the signal vector1112 (f k ), construct the driving vector V, where f k is the frequency point within the narrowband;

[0010] Step 3: Calculate the delay search spectrum P 1112 (f k , t) for each narrowband within the set search delay vector range, where t is the narrowband delay of the array at f k band;

[0011] Step 4: Aggregate the delay search spectra of multiple narrowbands to obtain the wideband delay spectrum P 12 (t);

[0012] Step 5: Search for the peak position of the delay spectrum of P 12 to determine the delay difference τ 12 between signal x1 and signal x2.

[0013] Preferably, in the said Step 2, the signal vector X 11 of the autocorrelation function R 11 = [R 11 ; R 11 , and the signal vector X 12 of the cross-correlation function R 12 = [R 11 ; R 12 .

[0014] Preferably, R 1112 (f k ) = E|X 11 (f k ) · X 12 H (f k )|, V = [a(f k , t)].

[0015] Preferably, in the said Step 3,

[0016] Preferably,

[0017] The present invention relates to an optimized delay estimation method based on quadratic correlation optimal weighting. By constructing an array signal based on the quadratic correlation function and a driving vector based on frequency and delay, an optimal weight vector is obtained using a constrained optimization method, and the delay search is optimally weighted with the optimal weight vector to obtain the delay difference between any two signals.

[0018] The delay estimation method combining quadratic correlation and optimal weight vector beamforming delay estimation in the present invention improves the delay estimation accuracy under low signal-to-noise ratio conditions.

[0019] The advantages of the present invention are as follows:

[0020] (1) Quadratic correlation can effectively suppress noise;

[0021] (2) The optimal weight vector time delay estimation can greatly improve the time delay estimation accuracy without upsampling;

[0022] (3) The combination of the above two points can effectively improve the time delay estimation accuracy under low signal-to-noise ratio conditions.

[0023] The present invention shows through theoretical analysis and simulation results that under low signal-to-noise ratio conditions, it can effectively suppress noise, the time delay estimation accuracy is greatly improved compared with the conventional cross-correlation method, and it can better locate the target. Description of the Drawings

[0024] Figure 1 is the flowchart of the present invention;

[0025] Figure 2 is the comparison of the time delay estimation result in the present invention with the conventional cross-correlation time delay estimation. It can be seen from the figure that the main lobe width and sidelobe height of the time delay estimation method with optimal quadratic correlation weighting are significantly lower than those of the conventional cross-correlation, and the time delay estimation error can be effectively reduced. Detailed Embodiments

[0026] The following further describes the present invention in detail in conjunction with embodiments, but the protection scope of the present invention is not limited thereto.

[0027] As Figure 1 shown, the present invention relates to a time delay estimation method based on optimal quadratic correlation weighting, constructs an array signal based on the quadratic correlation function and a steering vector based on frequency and time delay, and obtains an optimal weight vector by using a constrained optimization method, and optimally weights the time delay search with the optimal weight vector to obtain the time delay difference between any two signals.

[0028] The method includes the following steps:

[0029] Step 1: Construct the autocorrelation function R 11 and cross-correlation function R 12 of any two signals x1 and x2;

[0030] Step 2: Take R 11 and R 12 as new input signals, construct a signal vector, decompose R 11 and R 12 into several narrow bands respectively, and calculate the frequency-domain cross-spectrum matrix R 1112 (f k ) of each narrow band based on the signal vector, and construct a steering vector V, where f k is the frequency point within the narrow band;

[0031] In step 2, the autocorrelation function R 11 of the signal vector X 11 =[R 11 ; R 11 , and the cross-correlation function R 12 of the signal vector X 12 =[R 11 ; R 12 .

[0032] R 1112 (f k ) = E|X 11 (f k )·X 12 H (f k )|, V = [a(f k , t)].

[0033] Step 3: Calculate the delay search spectrum P 1112 (f k , t) for each narrowband within the set search delay vector range, where t is the narrowband delay of the array at the f k frequency band;

[0034] In step 3,

[0035] Step 4: The delay search spectra of multiple narrowbands are aggregated to obtain the wideband delay spectrum P 12 (t);

[0036]

[0037] Step 5: Search for the peak position of the delay spectrum of P 12 to determine the delay difference τ 12 between signal x1 and signal x2.

[0038] As Figure 2 shown is the comparison of the delay estimation result in the present invention with the conventional cross-correlation delay estimation. It can be seen from the figure that the main lobe width and sidelobe height of the delay estimation method with quadratic correlation optimal weighting are significantly lower than those of the conventional cross-correlation, and the delay estimation error can be effectively reduced.

Claims

1. A time delay estimation method based on quadratic correlation optimal weighting, characterized in that: Construct an array signal based on the quadratic correlation function and a steering vector based on frequency and time delay, use the constrained optimization method to obtain an optimal weight vector, perform optimal weighting on the time delay search with the optimal weight vector to obtain the time delay difference between any two signals; the steps include: Step 1: Construct the autocorrelation function R of any two signals x1 and x2 11 and the cross-correlation function R 12 ; Step 2: Using R 11 and R 12 as the new input signals, construct signal vectors. The signal vector X 11 of the autocorrelation function R 11 = [R 11 ; R 11 , and the signal vector X 12 of the cross-correlation function R 12 = [R 11 ; R 12 . Decompose R 11 and R 12 into several narrow bands respectively, and calculate the frequency-domain cross-spectrum matrix R 1112 (f k ) based on the signal vectors. , construct the driving vector V, V = [a(f k , t)], where f k is the frequency point within the narrow band. Step 3: Calculate the delay search spectrum P for each narrowband within the set search delay vector range 1112 (f k , t), , where t is the narrowband delay of the array at frequency band f k ; Step 4: After the time-delay search spectra of multiple narrow bands are aggregated, the time-delay spectrum P 12 (t) of the wide band is obtained, ; Step 5: Search for the peak position of the delay spectrum of P 12 to determine the delay difference between signal x1 and signal x2 .

Images