Ship hyperbolic frequency modulation signal parameter estimation method based on direction pre-judgment

Abstract

The invention discloses a ship hyperbolic frequency modulation signal parameter estimation method based on direction pre-judgment. The method comprises the following steps: (1) obtaining a to-be-processed ship hyperbolic frequency modulation signal data sequence; (2) initializing residual time-frequency distribution of the hyperbolic frequency-modulated signal data sequence of the ship to be processed; (3) performing signal direction judgment according to the residual time-frequency distribution, and calculating a frame index set during overturning; (4) calculating a separation stripe parameter according to the residual time-frequency distribution and the flipping time frame index set; (5) calculating a time parameter estimation value according to the separation strip parameter; (6) calculating a frequency parameter estimation value according to the time parameter estimation value, and updating residual time-frequency distribution; (7) judging whether unseparated signal components still exist or not according to the residual time-frequency distribution, if the unseparated signal components still exist, returning to the step (3), and otherwise, entering the step (8); and (8) calculating and outputting a signal component number estimation value, a time parameter estimation result and a frequency parameter estimation result.

Description

Technical Field

[0001] The invention belongs to the field of signal processing, and in particular relates to a method for estimating parameters of ship hyperbolic frequency modulation signals based on direction prejudgment. Background Art

[0002] Hyperbolic frequency modulation signals have good pulse compression characteristics and Doppler invariance, and are widely used in actual complex marine environments. Their time and frequency parameters can characterize the basic characteristics of the signal and provide prior information for subsequent analysis and processing of the signal. Therefore, it is of great significance to study the time and frequency parameter estimation problem of ship hyperbolic frequency modulation signals.

[0003] At present, there are four main types of ship HFM signal parameter estimation methods: (1) mapping the ship HFM signal into two-dimensional peaks, such as the maximum likelihood estimation method and the HST method; (2) extracting the time-frequency ridges of the ship HFM signal using transform domain features, such as the STFT-IRLS method and the STFRFT method; (3) combining time-frequency domain features with image processing, such as the WHT method; (4) analyzing the spectrum of the ship HFM signal and deriving the relationship between the signal spectrum and the parameters. However, the above-mentioned ship HFM signal parameter estimation methods are mainly aimed at single-component ship HFM signals, or only focus on the estimation of frequency parameters, but do not have the ability to estimate time parameters. It is difficult to solve the problem of dual parameter estimation of time and frequency of multi-component ship HFM signals with partial or complete time overlap, and has limitations. In addition, the traditional Radon transform method is only applicable to ship linear FM signals, and has poor focusing on ship HFM signals, that is, there is a peak fuzzy problem, which affects the performance of parameter estimation. Summary of the Invention

[0004] Technical problem, the purpose of the present invention is to estimate the number of components, time and frequency parameters of multi-component ship hyperbolic frequency modulation signals by separating the time-frequency distribution of single-component ship hyperbolic frequency modulation signals one by one.

[0005] Technical solution, in order to solve the above technical problems and achieve the above objectives, the present invention proposes a method for estimating the parameters of a ship's hyperbolic frequency modulation signal based on direction pre-judgment, the method comprising the following steps: (1) Obtain the data sequence of the ship's hyperbolic frequency modulation signal to be processed; (2) Initialize the residual time-frequency distribution of the ship's hyperbolic frequency modulation signal data sequence to be processed; (3) Determine the signal direction based on the residual time-frequency distribution and calculate the flip frame index set; (4) Calculate the separation strip parameters based on the residual time-frequency distribution and the flipped time frame index set; (5) Calculate the estimated value of the time parameter based on the separation strip parameters; (6) Calculate the frequency parameter estimate based on the time parameter estimate and update the residual time-frequency distribution; (7) Based on the residual time-frequency distribution, determine whether there are still unseparated signal components. If there are still unseparated signal components, return to step (3); otherwise, go to step (8); (8) Calculate and output the estimated value of the number of signal components, the estimated results of time parameters and the estimated results of frequency parameters.

[0006] Furthermore, in step (1), the following method is used to obtain the data sequence of the ship hyperbolic frequency modulation signal to be processed, which specifically includes the following steps: Received from the ship's hydroacoustic sensor N The real-time data collected from the sampling points is used as the data sequence of the ship's hyperbolic frequency modulation signal to be processed x ( n ), or extract from the memory the data starting from the moment the signal is detected N The data of the sampling points are used as the data sequence of the ship's hyperbolic frequency modulation signal to be processed x ( n ), n =0,1,…, N -1, described N is the number of sampling points corresponding to the pulse width of the detected ship hyperbolic frequency modulation signal, and its value is N =2 a , a is a positive integer greater than or equal to 9.

[0007] Furthermore, in step (2), the residual time-frequency distribution is initialized using the following method, which specifically includes the following steps: (2-1) Initialize the parameters of the initial residual time-frequency distribution, including the initialization of the following parameters: Sliding time window length N w Initialized to: 16< N w <floor{ N / 6}, where floor{·} is a floor function; Sliding time window stepping steps Initialization: 1≤ steps ≤floor{ N w / 4} positive integer; Total number of time frames of residual time-frequency distribution N t Initialized as: N t =floor{( N - Nw ) / steps}-1; Total number of frequency points in residual time-frequency distribution N f Initialized as: N f =floor{ N w / 2}+1; Signal component discrete index i Initialized as: i =1; (2-2) Calculate the time frame index of the time-frequency distribution n t The corresponding segmented data sequence :

[0008] in, n t is the time frame index of the time-frequency distribution, n t = 0,1, … , N t -1, r It is the time discrete index of the segmented data series; (2-3) Calculate the data sequence of the ship's hyperbolic frequency modulation signal to be processed x ( n )'s residual time-frequency distribution :

[0009] in, n f is the frequency index of time-frequency distribution, n f = 0,1, … , N f -1, |·| represents the modulus value, j is the imaginary unit, that is, .

[0010] Furthermore, in step (3), the following method is used to perform the i The direction of the signal component is determined and the i The flip time frame index set of the signal components specifically includes the following steps: (3-1) i The direction of the signal component is determined and the i Initialize the parameters of the flip frame index set of each signal component, including the initialization of the following parameters: Slope range search start value r start Initialized as: , where round{·} is the rounding function; Number of slant range searches N ρ Initialized as: ; The slant range search set ρ is initialized as: ,in, n ρ is the slope distance index, n ρ = 0, 1,…, N ρ -1, For the n ρ Search slope range values, and there are ; Angle search starting value i start Initialized as: i start =0°; Angle search end value i end Initialized as: i end =179°; Number of angle searches N θ Initialization: 90≤ N θ A positive integer ≤180; The angle search set θ is initialized as: ,in, n θ is the angle index, n θ = 0,1,…, N θ -1, For the n θ Search angle values, and there are ; (3-2) Calculate the i The linear Radon transform results of the signal components R i ( n ρ , n θ ):

[0011] in, d {·} is the impulse function, that is: ; (3-3) Calculate the i The peak slope distance index of the signal component and peak angle index :

[0012] in, Indicates that when i The linear Radon transform results of the signal components R i ( n ρ , n θ ) The corresponding slope distance index when the maximum value is obtained n ρ and angle index n θ the collection it consists of; (3-4) Calculate the flip frame index set , No. n t Elements for:

[0013] in, For the Search angle value, when When i The signal component is the ship's hyperbolic up-frequency modulation signal. When i The first signal component is the ship's hyperbolic down-modulated frequency signal.

[0014] Furthermore, in step (4), the following method is used to calculate the i The separation strip parameters of the signal components include the following steps: (4-1) To calculate the i Initialize the parameters of the separated strip parameters of each signal component, including the initialization of the following parameters: Hyperbolic Radon transform real semi-axis length search starting value a start Initialized to: 0< a start < f s / 2, where f s is the signal sampling frequency; Hyperbolic Radon transform real semi-axis length search end value a end Initialized as: astart < a end ≤ f s / 2 positive real number; Hyperbolic Radon transform real semi-axis length search number N a Initialization: 100≤ N a Positive integer <250; The real semi-axis length search set a of the hyperbolic Radon transform is initialized as: ,in, n a is the real semi-axis length index, n a = 0, 1, … , N a -1, For the n a Search for the real semi-axis length value, and there is ; Hyperbolic Radon transform imaginary semi-axis length search starting value b start Initialized as: b start Positive real numbers > 0; Hyperbolic Radon transform imaginary semi-axis length search end value b end Initialized as: b start < b end Positive real numbers <+∞; Hyperbolic Radon transform imaginary semi-axis length search number N b Initialization: 100≤ N b Positive integer <250; The imaginary semi-axis length search set b of the hyperbolic Radon transform is initialized as: ,in, n b is the imaginary semi-axis length index, n b = 0, 1, … , N b -1, For the n b Search for the imaginary semi-axis length value, and there is ; Time-frequency distribution time interval Δ t Initialized to: Δ t = steps / f s ; Time-frequency distribution frequency interval Δ f Initialized to: Δ f = f s / N w ; Strip hyperbolic Radon transform strip width search starting value d start Initialized as: d start =0; Strip hyperbolic Radon transform strip width search number N d Initialized as: N d = N a ; The strip width search set D of the strip hyperbolic Radon transform is initialized as: ,in, n d is the stripe width index, n d = 0,1, … , N d -1, For the n d Search stripe width values, and ; Optimal stripe width decision threshold TH d Initialized to: 0< TH d Positive real number < 0.3; (4-2) Calculate the i The hyperbolic Radon transform results of the signal components HR i ( n a , n b ): ; (4-3) Calculate the i The optimal real semi-axis length index of the signal component and the optimal imaginary semi-axis length index :

[0015] in, Indicates that when i The hyperbolic Radon transform results of the signal components HR i ( n a , n b ) corresponds to the real semi-axis length index when the maximum value is obtained n a and imaginary semi-axis length index n b the collection it consists of; (4-4) Calculate the i The maximum value of the hyperbolic Radon transform results of the signal components : ; (4-5) i The hyperbolic Radon transform results of the signal components HR i ( n a , n b ) is normalized to obtain i The normalized hyperbolic Radon transform results of the signal components : ; (4-6) Calculate the i The optimal strip hyperbolic Radon transform result of the signal components :

[0016] Among them, max(·,·) means taking the maximum value of the two, and min(·,·) means taking the minimum value of the two. or Added new stripe width index; (4-7) Calculate the i The difference function of the signal components :

[0017] in, m d is the difference stripe width index; (4-8) Screen all the values ​​that satisfy the difference function Less than the optimal stripe width decision threshold TH d The difference strip width index set : ; (4-9) Calculate the i The optimal stripe width index of the signal component :

[0018] Among them, Ф is an empty set, == means judging whether the elements contained in the left and right sets are exactly the same. Indicates the difference strip width index set The first element of .

[0019] Furthermore, in step (5), the following method is used to calculate the i The time parameter estimation value of a signal component specifically includes the following steps: (5-1) To calculate i Initialize the parameters of the time parameter estimation value of each signal component, including the initialization of the following parameters: Time frame signal has decision threshold coefficient Initialized as: A positive real number; Maximum frequency interval threshold Initialized as: A positive integer; Upper limit of total number of frames at no signal frequency Initialized as: A positive integer; Total number of frames at no-signal frequency points N non Initialized as: N non =0; Total number of signal component time frames N total Initialized as: N total =0; Current decision frame index q i Initialized as: q i =0; The frame index set where the signal component frequency exists Initialized as: ; (5-2) Calculate the frequency index set on the hyperbolic strip And the frequency index set under the hyperbolic strip , No. n t Elements and No. n t Elements They are:

[0020]

[0021] in, 、 and Respectively The search strip width value, The search real semi-axis length value and the Search for the imaginary semi-axis length value; (5-3) Calculate the i The set of hyperbolic band frequency upper limits for signal components and the set of lower frequency limits of hyperbolic strips , No. n t Elements and No. n t Elements They are:

[0022]

[0023] in, and The length is N t vector; (5-4) Calculate the i Pre-separation time-frequency distribution of signal components : ; (5-5) Calculate the maximum modulus of the pre-separation time-frequency distribution :

[0024] Among them, max{·} is the maximum value function; (5-6) Calculation of the frame signal existence judgment threshold : ; (5-7) Calculate the separation frequency index :

[0025] in, Indicates the i The signal component is indexed at the current decision frame q i Pre-separated time-frequency distribution The frequency index of the time-frequency distribution corresponding to the maximum value; (5-8) Determine the pre-separation time-frequency distribution Frame index at the current decision time q i Whether there is a signal frequency point, that is, whether the following conditions are met:

[0026] If it is established, go to step (5-9), otherwise, let the total number of time frames with no signal frequency be N non = N non +1 and proceed to steps (5-12); (5-9) Determine whether the signal component frequency point exists in the time frame index set Is it an empty set? That is, determine whether the following conditions are met:

[0027] If so, go to step (5-11), otherwise, go to step (5-10); (5-10) Determine whether the following conditions are met:

[0028] If it is established, go to step (5-11), otherwise, let the total number of time frames with no signal frequency be N non = N non +1 and proceed to steps (5-12); (5-11) Let the total number of signal component time frames be N total = N total +1, total number of frames with no signal frequency N non =0, and update the i The signal component frequency points of the signal components exist in the time frame index set : ; (5-12) Judgement i Whether the signal component reaches the end time frame, that is, whether the following conditions are met:

[0029] If so, go to step (5-14), otherwise, go to step (5-13); (5-13) Update the current decision time frame index q i = qi +1, check whether the following conditions are met:

[0030] If so, return to step (5-8), otherwise, go to step (5-14); (5-14) Calculate the i The starting time frame index of the signal component , End time frame index :

[0031]

[0032] in, Indicates the time frame index set where the signal component frequency exists The first element of Indicates the time frame index set where the signal component frequency exists The last element of .

[0033] Furthermore, in step (6), the following method is used to calculate the i The frequency parameter estimation value of each signal component and the update of the residual time-frequency distribution are obtained, which specifically includes the following steps: (6-1) Calculate the i The frame index set of the signal components exists continuously : ; (6-2) Calculate the i The total number of consecutive time frames of signal components : ; (6-3) Calculate the i Separation time-frequency distribution of signal components : ; (6-4) According to i Separation time-frequency distribution of signal components , starting time frame index and the ending time frame index , extract the i The instantaneous frequency index set of signal components , No. Elements for:

[0034] Among them, in iEach signal component has a time frame index Department, Represents the separation of time-frequency distribution The frequency index of the time-frequency distribution corresponding to the maximum value; (6-5) i The instantaneous frequency index set of signal components The inverse of the least squares linear fitting is performed to obtain the i The frequency estimation parameter vector of the signal components :

[0035] Among them, X is the least squares linear fitting auxiliary matrix, , G i For the i The reciprocal vector of the instantaneous frequency index set of the signal components, , the superscript T indicates the matrix transpose; (6-6) Get the i The estimated value of the periodic slope of the signal components and starting frequency estimates :

[0036] Among them, θ i (1) and θ i (2) indicates the i The frequency estimation parameter vector θ of the signal components i The first and second elements of ; (6-7) Update the residual time-frequency distribution : .

[0037] Furthermore, in step (7), the following method is used to determine whether there are still unseparated signal components, which specifically includes the following steps: (7-1) Decision threshold for determining whether there are still unseparated signal components TH S Initialization: 80≤ TH S Positive real numbers ≤ 160; (7-2) Calculate the residual time-frequency distribution The maximum value : ; (7-3) Determine whether there are still unseparated signal components, that is, determine whether the following conditions are met:

[0038] If the condition is met, let the signal component discrete index i = i +1 and return to step (3); otherwise, go to step (8).

[0039] Furthermore, in step (8), the following method is used to calculate and output the estimated value of the number of signal components and the estimation results of time and frequency parameters, which specifically includes the following steps: (8-1) Calculate the estimated number of signal components And output: ; (8-2) Output time parameter estimation results:

[0040]

[0041] (8-3) Output frequency parameter estimation results:

[0042]

[0043] Among them, the signal component discrete index .

[0044] Beneficial effects: Compared with the prior art, the technical solution of the present invention has the following beneficial technical effects: 1. The present invention makes full use of the characteristics that the angle corresponding to the peak of the linear Radon transform of the ship's hyperbolic up-FM signal is an acute angle, and the angle corresponding to the peak of the linear Radon transform of the ship's hyperbolic down-FM signal is an obtuse angle, thereby realizing the direction pre-determination of the ship's hyperbolic FM signal, as shown in step (3). This judgment process can provide prior estimation information for the subsequent hyperbolic Radon transform.

[0045] 2. The present invention makes full use of the instantaneous frequency curve characteristics of the ship's hyperbolic FM signal and fits the time-frequency distribution of the ship's hyperbolic FM signal with the standard hyperbola equation, as shown in step (4). The fitting process realizes the pre-estimation of the frequency center ridge required for separating the signal components, providing a basis for the subsequent separation process of the time-frequency distribution of the current signal component.

[0046] 3. The present invention makes full use of the focusing characteristics of the ship's hyperbolic FM signal in the hyperbolic Radon transform domain to achieve pre-separation of the time-frequency distribution of the signal components. As shown in step (4), this process obtains the optimal strip width required for separating the time-frequency distribution of the signal components. This pre-separation process can reduce the influence of noise and other signal components on the current signal component and improve the accuracy of the estimation of the time parameters of subsequent signal components.

[0047] 4. The present invention converts the detection and parameter estimation problem of multi-component signals into the detection and parameter estimation problem of single-component signals by separating the signal components one by one, as shown in step (6). The separation process can obtain the time-frequency distribution of a single signal component, and different traditional frequency parameter estimation methods can be applied on this basis to complete the parameter estimation of the signal component, thereby improving the accuracy of the signal frequency parameter estimation.

[0048] 5. The present invention makes full use of the time-frequency domain distribution characteristics of the ship's hyperbolic FM signal and realizes the automatic judgment of whether the signal components are completely detected. As shown in step (7), the peak value of the residual time-frequency distribution is used to detect the signal components whose parameters have not been estimated, thereby providing a basis for the subsequent time and frequency parameter estimation of the multi-component ship's hyperbolic FM signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 Schematic diagram of the process of the present invention; Figure 2 is the residual time-frequency distribution of the first signal component of Example 1; Figure 3 is the hyperbolic Radon transform result of the first signal component of Example 1; Figure 4 is the separated time-frequency distribution of the first signal component of Example 1; Figure 5 is the residual time-frequency distribution of the second signal component in Example 1; Figure 6 is the hyperbolic Radon transform result of the second signal component of Example 1; Figure 7 is the separated time-frequency distribution of the second signal component of Example 1; Figure 8 is the residual time-frequency distribution of the third signal component in Example 1. DETAILED DESCRIPTION

[0050] In order to better understand the purpose, structure and function of the present invention, the present invention is further described below with reference to the accompanying drawings.

[0051] like Figure 1 As shown, the present invention proposes a method for estimating parameters of a ship's hyperbolic frequency modulation signal based on direction pre-judgment, comprising the following steps: (1) Obtain the data sequence of the ship's hyperbolic frequency modulation signal to be processed; (2) Initialize the residual time-frequency distribution of the ship's hyperbolic frequency modulation signal data sequence to be processed; (3) Determine the signal direction based on the residual time-frequency distribution and calculate the flip frame index set; (4) Calculate the separation strip parameters based on the residual time-frequency distribution and the flipped time frame index set; (5) Calculate the estimated value of the time parameter based on the separation strip parameters; (6) Calculate the frequency parameter estimate based on the time parameter estimate and update the residual time-frequency distribution; (7) Based on the residual time-frequency distribution, determine whether there are still unseparated signal components. If there are still unseparated signal components, return to step (3); otherwise, go to step (8); (8) Calculate and output the estimated value of the number of signal components, the estimated results of time parameters and the estimated results of frequency parameters.

[0052] Furthermore, in step (1), the following method is used to obtain the data sequence of the ship's hyperbolic frequency modulation signal to be processed, which specifically includes the following steps: receiving the data sequence of the ship's underwater acoustic sensor from the ship's underwater acoustic sensor; N The real-time data collected from the sampling points is used as the data sequence of the ship's hyperbolic frequency modulation signal to be processed x ( n ), or extract from the memory the data starting from the moment the signal is detected N The data of the sampling points are used as the data sequence of the ship's hyperbolic frequency modulation signal to be processed x ( n ), n =0,1,…, N -1, described N is the number of sampling points corresponding to the pulse width of the detected ship hyperbolic frequency modulation signal, and its value is N =2 a , a is a positive integer greater than or equal to 9.

[0053] Furthermore, in step (2), the residual time-frequency distribution is initialized using the following method, which specifically includes the following steps: (2-1) Initialize the parameters of the initial residual time-frequency distribution, including the initialization of the following parameters: Sliding time window length N w Initialized to: 16< N w <floor{ N / 6}, where floor{·} is a floor function; Sliding time window stepping steps Initialization: 1≤ steps ≤floor{ N w / 4} positive integer; Total number of time frames of residual time-frequency distribution N t Initialized as: Nt =floor{( N - N w ) / steps}-1; Total number of frequency points in residual time-frequency distribution N f Initialized as: N f =floor{ N w / 2}+1; Signal component discrete index i Initialized as: i =1; (2-2) Calculate the time frame index of the time-frequency distribution n t The corresponding segmented data sequence :

[0054] in, n t is the time frame index of the time-frequency distribution, n t = 0,1, … , N t -1, r It is the time discrete index of the segmented data series; (2-3) Calculate the data sequence of the ship's hyperbolic frequency modulation signal to be processed x ( n )'s residual time-frequency distribution :

[0055] in, n f is the frequency index of time-frequency distribution, n f = 0,1, … , N f -1, |·| represents the modulus value, j is the imaginary unit, that is, .

[0056] Furthermore, in step (3), the following method is used to perform the i The direction of the signal component is determined and the i The flip time frame index set of the signal components specifically includes the following steps: (3-1) i The direction of the signal component is determined and the i Initialize the parameters of the flip frame index set of each signal component, including the initialization of the following parameters: Slope range search start value r start Initialized as: , where round{·} is the rounding function; Number of slant range searches N ρ Initialized as: ; The slant range search set ρ is initialized as: ,in, n ρ is the slope distance index, n ρ = 0, 1,…, N ρ -1, For the n ρ Search slope range values, and there are ; Angle search starting value i start Initialized as: i start =0°; Angle search end value i end Initialized as: i end =179°; Number of angle searches N θ Initialization: 90≤ N θ A positive integer ≤180; The angle search set θ is initialized as: ,in, n θ is the angle index, n θ = 0,1,…, N θ -1, For the n θ Search angle values, and there are ; (3-2) Calculate the i The linear Radon transform results of the signal components R i ( n ρ , n θ ):

[0057] in, d{·} is the impulse function, that is: ; (3-3) Calculate the i The peak slope distance index of the signal component and peak angle index :

[0058] in, Indicates that when i The linear Radon transform results of the signal components R i ( n ρ , n θ ) The corresponding slope distance index when the maximum value is obtained n ρ and angle index n θ the collection it consists of; (3-4) Calculate the flip frame index set , No. n t Elements for:

[0059] in, For the Search angle value, when When i The signal component is the ship's hyperbolic up-frequency modulation signal. When i The first signal component is the ship's hyperbolic down-modulated frequency signal.

[0060] Furthermore, in step (4), the following method is used to calculate the i The separation strip parameters of the signal components include the following steps: (4-1) To calculate the i Initialize the parameters of the separated strip parameters of each signal component, including the initialization of the following parameters: Hyperbolic Radon transform real semi-axis length search starting value a start Initialized to: 0< a start < f s / 2, where f s is the signal sampling frequency; Hyperbolic Radon transform real semi-axis length search end value a end Initialized as: a start < a end ≤ f s / 2 positive real number; Hyperbolic Radon transform real semi-axis length search number N a Initialization: 100≤ N a Positive integer <250; The real semi-axis length search set a of the hyperbolic Radon transform is initialized as: ,in, n a is the real semi-axis length index, n a = 0, 1, … , N a -1, For the n a Search for the real semi-axis length value, and there is ; Hyperbolic Radon transform imaginary semi-axis length search starting value b start Initialized as: b start Positive real numbers > 0; Hyperbolic Radon transform imaginary semi-axis length search end value b end Initialized as: b start < b end Positive real numbers < +∞; Hyperbolic Radon transform imaginary semi-axis length search number N b Initialization: 100≤ N b Positive integer <250; The imaginary semi-axis length search set b of the hyperbolic Radon transform is initialized as: ,in, n b is the imaginary semi-axis length index, n b = 0, 1, … , N b -1, For the n b Search for the imaginary semi-axis length value, and there is ; Time-frequency distribution time interval Δ t Initialized to: Δ t = steps / f s ; Time-frequency distribution frequency interval Δ f Initialized to: Δ f = f s / N w ; Strip hyperbolic Radon transform strip width search starting value d start Initialized as: d start =0; Strip hyperbolic Radon transform strip width search number N d Initialized as: N d = N a ; The strip width search set D of the strip hyperbolic Radon transform is initialized as: ,in, n d is the stripe width index, n d = 0,1, … , N d -1, For the n d Search stripe width values, and ; Optimal stripe width decision threshold TH d Initialized to: 0< TH d Positive real number < 0.3; (4-2) Calculate the i The hyperbolic Radon transform results of the signal components HR i ( n a , n b ): ; (4-3) Calculate the i The optimal real semi-axis length index of the signal component and the optimal imaginary semi-axis length index :

[0061] in, Indicates that when i The hyperbolic Radon transform results of the signal components HR i ( n a , n b ) corresponds to the real semi-axis length index when the maximum value is obtained n a and imaginary semi-axis length index n b the collection it consists of; (4-4) Calculate the i The maximum value of the hyperbolic Radon transform results of the signal components : ; (4-5) i The hyperbolic Radon transform results of the signal components HR i ( n a , n b ) is normalized to obtain i The normalized hyperbolic Radon transform results of the signal components : ; (4-6) Calculate the i The optimal strip hyperbolic Radon transform result of the signal components :

[0062] Among them, max(·,·) means taking the maximum value of the two, and min(·,·) means taking the minimum value of the two. or Added new stripe width index; (4-7) Calculate the i The difference function of the signal components :

[0063] in, m d is the difference stripe width index; (4-8) Screen all the values ​​that satisfy the difference function Less than the optimal stripe width decision threshold TH d The difference strip width index set : ; (4-9) Calculate the i The optimal stripe width index of the signal component :

[0064] Among them, Ф is an empty set, == means judging whether the elements contained in the left and right sets are exactly the same. Indicates the difference strip width index set The first element of .

[0065] Furthermore, in step (5), the following method is used to calculate the i The time parameter estimation value of a signal component specifically includes the following steps: (5-1) To calculate the i Initialize the parameters of the time parameter estimation value of each signal component, including the initialization of the following parameters: Time frame signal has decision threshold coefficient Initialized as: A positive real number; Maximum frequency interval threshold Initialized as: A positive integer; Upper limit of total number of frames at no signal frequency Initialized as: A positive integer; Total number of frames at no-signal frequency points N non Initialized as: N non =0; Total number of signal component time frames N total Initialized as: N total =0; Current decision frame index q i Initialized as: q i =0; The frame index set where the signal component frequency exists Initialized as: ; (5-2) Calculate the frequency index set on the hyperbolic strip And the frequency index set under the hyperbolic strip , No. n t Elements and No. n t Elements They are:

[0066]

[0067] in, 、 and Respectively The search strip width value, The search real semi-axis length value and the Search for the imaginary semi-axis length value; (5-3) Calculate the i The set of hyperbolic band frequency upper limits for signal components and the set of lower frequency limits of hyperbolic strips , No. n t Elements and No. n t Elements They are:

[0068]

[0069] in, and The length is N t vector; (5-4) Calculate the i Pre-separation time-frequency distribution of signal components : ; (5-5) Calculate the maximum modulus of the pre-separation time-frequency distribution :

[0070] Among them, max{·} is the maximum value function; (5-6) Calculation of the frame signal existence judgment threshold : ; (5-7) Calculate the separation frequency index :

[0071] in, Indicates the i The signal component is indexed at the current decision frameq i Pre-separated time-frequency distribution The frequency index of the time-frequency distribution corresponding to the maximum value; (5-8) Determine the pre-separation time-frequency distribution Frame index at the current decision time q i Whether there is a signal frequency point, that is, whether the following conditions are met:

[0072] If it is established, go to step (5-9), otherwise, let the total number of time frames with no signal frequency be N non = N non +1 and proceed to steps (5-12); (5-9) Determine whether the signal component frequency point exists in the time frame index set Is it an empty set? That is, determine whether the following conditions are met:

[0073] If so, go to step (5-11), otherwise, go to step (5-10); (5-10) Determine whether the following conditions are met:

[0074] If it is established, go to step (5-11), otherwise, let the total number of time frames with no signal frequency be N non = N non +1 and proceed to steps (5-12); (5-11) Let the total number of signal component time frames be N total = N total +1, total number of frames with no signal frequency N non =0, and update the i The signal component frequency points of the signal components exist in the time frame index set : ; (5-12) Judgement i Whether the signal component reaches the end time frame, that is, whether the following conditions are met:

[0075] If so, go to step (5-14), otherwise, go to step (5-13); (5-13) Update the current decision time frame indexq i = q i +1, check whether the following conditions are met:

[0076] If so, return to step (5-8), otherwise, go to step (5-14); (5-14) Calculate the i The starting time frame index of the signal component , End time frame index :

[0077]

[0078] in, Indicates the time frame index set where the signal component frequency exists The first element of Indicates the time frame index set where the signal component frequency exists The last element of .

[0079] Furthermore, in step (6), the following method is used to calculate the i The frequency parameter estimation value of each signal component and the update of the residual time-frequency distribution are obtained, which specifically includes the following steps: (6-1) Calculate the i The frame index set of the signal components exists continuously : ; (6-2) Calculate the i The total number of consecutive time frames of signal components : ; (6-3) Calculate the i Separation time-frequency distribution of signal components : ; (6-4) According to i Separation time-frequency distribution of signal components , starting time frame index and the ending time frame index , extract the i The instantaneous frequency index set of signal components , No. Elements for:

[0080] Among them, in i Each signal component has a time frame index Department, Represents the separation of time-frequency distribution The frequency index of the time-frequency distribution corresponding to the maximum value; (6-5) i The instantaneous frequency index set of signal components The inverse of the least squares linear fitting is performed to obtain the i The frequency estimation parameter vector of the signal components :

[0081] Among them, X is the least squares linear fitting auxiliary matrix, , G i For the i The reciprocal vector of the instantaneous frequency index set of the signal components, , the superscript T indicates the matrix transpose; (6-6) Get the i The estimated value of the periodic slope of the signal components and starting frequency estimates :

[0082] Among them, θ i (1) and θ i (2) indicates the i The frequency estimation parameter vector θ of the signal components i The first and second elements of ; (6-7) Update the residual time-frequency distribution : .

[0083] Furthermore, in step (7), the following method is used to determine whether there are still unseparated signal components, which specifically includes the following steps: (7-1) Decision threshold for determining whether there are still unseparated signal components TH S Initialization: 80≤ TH S Positive real numbers ≤ 160; (7-2) Calculate the residual time-frequency distribution The maximum value : ; (7-3) Determine whether there are still unseparated signal components, that is, determine whether the following conditions are met:

[0084] If the condition is met, let the signal component discrete index i = i +1 and return to step (3); otherwise, go to step (8).

[0085] Furthermore, in step (8), the following method is used to calculate and output the estimated value of the number of signal components and the estimation results of time and frequency parameters, which specifically includes the following steps: (8-1) Calculate the estimated number of signal components And output: ; (8-2) Output time parameter estimation results:

[0086]

[0087] (8-3) Output frequency parameter estimation results:

[0088]

[0089] Among them, the signal component discrete index .

[0090] In the embodiment of the present invention, the model of the ship's hyperbolic FM signal received by the simulation x ( t )for:

[0091] in, I is the number of signal components, i is the discrete index of the signal component, t 0,i For the i The starting time of each signal component, t i For the i The pulse width of each signal component, A i , f 1,i , f 2,i Corresponding to the i The amplitude, start frequency and end frequency of each signal component, k 0,i For the i The periodic slope of a signal component is defined as: k 0,i=( f 2,i - f 1,i ) / ( t i f 1,i f 2,i ); w ( t ) has a mean of zero and a variance of s 2 Gaussian white noise with variance s 2 The size depends on the signal-to-noise ratio SNR : .

[0092] Signal sampling frequency f s The hyperbolic FM signal of the ship received by the above simulation x ( t ) to perform discrete sampling and obtain the ship's hyperbolic frequency modulation signal sampling data sequence x ( n )for:

[0093] in, , .

[0094] Example 1: The parameters of the simulated ship hyperbolic FM signal are set to: number of signal components I =2, signal sampling frequency f s =2 kHz, SNR SNR =-1 dB, the start time of the first signal component t 0,1 =0.15 s, pulse width t 1=0.35 s, amplitude A 1=1, starting frequency f 1,1 =700 Hz, end frequency f 2,1 =500 Hz, period slope k 0,1 =-0.0016, that is, the first signal component is the ship's hyperbolic down-modulation signal; the starting time of the second signal component is t 0,2 =0.15 s, pulse width t 2=0.25 s, amplitude A 2=1, starting frequency f1,2 =230 Hz, end frequency f 2,2 =430 Hz, period slope k 0,2 =0.0081, that is, the second signal component is the ship's hyperbolic up-frequency modulation signal.

[0095] Next, we will detect and estimate the frequency parameters of the simulated ship hyperbolic FM signal: According to step (1), extract the data from the memory starting from the moment the signal is detected N = 1024 sampling points of data as the ship hyperbolic frequency modulation signal data sequence to be processed x ( n ), n = 0, 1, … , N -1; According to step (2), set the sliding time window length N w =128, sliding time window step steps =32, the total number of time frames of the residual time-frequency distribution N t =27, the total number of frequency points in the residual time-frequency distribution N f =65, discrete index of signal component i =1; use STFT to obtain the data sequence of the ship's hyperbolic frequency modulation signal to be processed x ( n )'s residual time-frequency distribution ,like Figure 2 As shown, go to step (3); According to step (3), set the starting value of the slope range search r start =-35, number of slant range searches N ρ =71, angle search starting value i start =0 ° , angle search end value i end =179 ° , angle search number N θ =180; for the residual time-frequency distribution , calculate the peak angle index of the first signal component ,because , so the first signal component is the ship's hyperbolic down-modulated frequency signal, and the calculated flipped time frame index set is , go to step (4); According to step (4), set the starting value of the real semi-axis length search of the hyperbolic Radon transforma start =0.001, hyperbolic Radon transform real semi-axis length search end value a end =1000, the number of real semi-axis length searches for hyperbolic Radon transform N a =200, the starting value for the search of the imaginary semi-axis length of the hyperbolic Radon transform b start =0.001, hyperbolic Radon transform imaginary semi-axis length search end value b end =5, the number of searches for the imaginary semi-axis length of the hyperbolic Radon transform N b =200, time-frequency distribution time interval Δ t =0.016, time-frequency distribution frequency interval Δ f =15.625, the starting value of the strip width search for the hyperbolic Radon transform d start =0, the number of strip width searches for the hyperbolic Radon transform N d =200, optimal stripe width decision threshold TH d =0.1813; calculate the hyperbolic Radon transform result of the first signal component HR 1( n a , n b ),like Figure 3 As shown, calculate the optimal strip width index of the first signal component , optimal real semi-axis length index and the optimal imaginary semi-axis length index , go to step (5); According to step (5), set the time frame signal existence judgment threshold coefficient , maximum frequency interval threshold , the upper limit of the total number of frames when there is no signal frequency , total number of frames with no signal frequency N non =0, total number of signal component time frames N total =0, current decision frame index q 1=0, the signal component frequency point exists in the time frame index set ; Calculate the starting time frame index of the first signal component , End time frame index , go to step (6); According to step (6), the separated time-frequency distribution of the first signal component is calculated ,like Figure 4 As shown, calculate the estimated slope of the first signal component cycle and the starting frequency estimate , and update the residual time-frequency distribution ,like Figure 5 As shown, go to step (7); According to step (7), set the decision threshold to determine whether there are still unseparated signal components TH S =80; calculate the residual time-frequency distribution The maximum value ,because , indicating that there are still unseparated signal components, let the signal component discrete index i =2, return to step (3); According to step (3), set the starting value of the slope range search r start =-35, number of slant range searches N ρ =71, angle search starting value i start =0 ° , angle search end value i end =179 ° , angle search number N θ =180; for the residual time-frequency distribution ; Calculate the peak angle index of the second signal component ,because , so the second signal component is the ship's hyperbolic up-frequency modulation signal, and the calculated flip time frame index set is , go to step (4); According to step (4), set the starting value of the real semi-axis length search of the hyperbolic Radon transform a start =0.001, hyperbolic Radon transform real semi-axis length search end value a end =1000, the number of real semi-axis length searches for hyperbolic Radon transform N a =200, the starting value for the search of the imaginary semi-axis length of the hyperbolic Radon transform b start =0.001, hyperbolic Radon transform imaginary semi-axis length search end value b end =5, the number of searches for the imaginary semi-axis length of the hyperbolic Radon transform N b =200, time-frequency distribution time interval Δ t=0.016, time-frequency distribution frequency interval Δ f =15.625, the starting value of the strip width search for the hyperbolic Radon transform d start =0, the number of strip width searches for the hyperbolic Radon transform N d =200, optimal stripe width decision threshold TH d =0.1813; calculate the hyperbolic Radon transform result of the second signal component HR 2( n a , n b ),like Figure 6 As shown, calculate the optimal strip width index of the second signal component , optimal real semi-axis length index and the optimal imaginary semi-axis length index , go to step (5); According to step (5), set the time frame signal existence judgment threshold coefficient , maximum frequency interval threshold , the upper limit of the total number of frames when there is no signal frequency , total number of frames with no signal frequency N non =0, total number of signal component time frames N total =0, current decision frame index q 2=0, the signal component frequency point exists in the time frame index set ; Calculate the starting time frame index of the second signal component , End time frame index , go to step (6); According to step (6), the separated time-frequency distribution of the second signal component is calculated ,like Figure 7 As shown, calculate the estimated slope of the second signal component period and the starting frequency estimate , and update the residual time-frequency distribution ,like Figure 8 As shown, go to step (7); According to step (7), set the decision threshold to determine whether there are still unseparated signal components TH S =80; calculate the residual time-frequency distribution The maximum value ,because , indicating that there is no unseparated signal component, and proceed to step (8); According to step (8), the estimated value of the number of signal components and the estimated results of time and frequency parameters are output: ; The test results show that the ship's hyperbolic FM signal is a multi-component hyperbolic FM signal. The number of signal components is consistent with the simulation settings, and the periodic slope is and starting frequency The relative errors are: .

Claims

1. A method for estimating the parameters of ship hyperbolic frequency modulation signals based on direction prejudgment, characterized in that: The following steps are involved: (1) Obtain the data sequence of the ship's hyperbolic frequency modulation signal to be processed; (2) Initialize the residual time-frequency distribution of the ship's hyperbolic frequency modulation signal data sequence to be processed; (3) Determine the signal direction based on the residual time-frequency distribution and calculate the flip frame index set; (4) Calculate the separation strip parameters based on the residual time-frequency distribution and the flipped time frame index set; (5) Calculate the estimated value of the time parameter based on the separation strip parameters; (6) Calculate the frequency parameter estimate based on the time parameter estimate and update the residual time-frequency distribution; (7) Based on the residual time-frequency distribution, determine whether there are still unseparated signal components. If there are still unseparated signal components, return to step (3); otherwise, go to step (8); (8) Calculate and output the estimated value of the number of signal components, the time parameter estimation results and the frequency parameter estimation results.

2. A method for estimating ship hyperbolic frequency modulation signal parameters based on direction pre-judgment according to claim 1, characterized in that: In step (1), the following method is used to obtain the data sequence of the ship hyperbolic frequency modulation signal to be processed, which specifically includes the following steps: Received from the ship's hydroacoustic sensor N The real-time data collected from the sampling points is used as the data sequence of the ship's hyperbolic frequency modulation signal to be processed x ( n ), or extract from the memory the data starting from the moment the signal is detected N The data of the sampling points are used as the data sequence of the ship's hyperbolic frequency modulation signal to be processed x ( n ), n =0,1,…, N -1, described N is the number of sampling points corresponding to the pulse width of the detected ship hyperbolic frequency modulation signal, and its value is N =2 a , a is a positive integer greater than or equal to 9.

3. The method for estimating ship hyperbolic frequency modulation signal parameters based on direction prejudgment according to claim 2, characterized in that: In step (2), the residual time-frequency distribution is initialized using the following method, which specifically includes the following steps: (2-1) Initialize the parameters of the initial residual time-frequency distribution, including the initialization of the following parameters: Sliding time window length N w Initialized to: 16< N w <floor{ N / 6}, where floor{·} is a floor function; Sliding time window stepping step Initialization: 1≤ step ≤floor{ N w / 4} positive integer; Total number of time frames of residual time-frequency distribution N t Initialized as: N t =floor{( N - N w ) / step }-1; Total number of frequency points in residual time-frequency distribution N f Initialized as: N f =floor{ N w / 2}+1; Signal component discrete index i Initialized as: i =1; (2-2) Calculate the time frame index of the time-frequency distribution n t The corresponding segmented data sequence : in, n t is the time frame index of the time-frequency distribution, n t = 0,1, … , N t -1, r It is the time discrete index of the segmented data series; (2-3) Calculate the data sequence of the ship's hyperbolic frequency modulation signal to be processed x ( n )'s residual time-frequency distribution : in, n f is the frequency index of time-frequency distribution, n f = 0,1, … , N f -1, |·| represents the modulus value, j is the imaginary unit, that is, .

4. A method for estimating ship hyperbolic frequency modulation signal parameters based on direction prejudgment according to claim 3, characterized in that: In step (3), the following method is used to perform the i The direction of the signal component is determined and the i The flip time frame index set of the signal components specifically includes the following steps: (3-1) i The direction of the signal component is determined and the i Initialize the parameters of the flip frame index set of each signal component, including the initialization of the following parameters: Slope range search start value ρ start Initialized as: , where round{·} is the rounding function; Number of slant range searches N ρ Initialized as: ; The slant range search set ρ is initialized as: ,in, n ρ is the slope distance index, n ρ = 0, 1,…, N ρ -1, For the n ρ Search slope range values, and there are ; Angle search starting value θ start Initialized as: θ start =0°; Angle search end value θ end Initialized as: θ end =179°; Number of angle searches N θ Initialization: 90≤ N θ A positive integer ≤180; The angle search set θ is initialized as: ,in, n θ is the angle index, n θ = 0,1,…, N θ -1, For the n θ Search angle values, and there are ; (3-2) Calculate the i The linear Radon transform results of the signal components R i ( n ρ , n θ ): in, δ {·} is the impulse function, that is: ； (3-3) Calculate the i The peak slope distance index of the signal component and peak angle index : in, Indicates that when i The linear Radon transform results of the signal components R i ( n ρ , n θ ) The corresponding slope distance index when the maximum value is obtained n ρ and angle index n θ the collection it consists of; (3-4) Calculate the flip frame index set , No. n t Elements for: in, For the Search angle value, when When i The signal component is the ship's hyperbolic up-frequency modulation signal. When i The first signal component is the ship's hyperbolic down-modulated frequency signal.

5. The method for estimating ship hyperbolic frequency modulation signal parameters based on direction prejudgment according to claim 4, characterized in that: In step (4), the following method is used to calculate the i The separation strip parameters of the signal components include the following steps: (4-1) To calculate the i Initialize the parameters of the separated strip parameters of each signal component, including the initialization of the following parameters: Hyperbolic Radon transform real semi-axis length search starting value a start Initialized to: 0< a start < f s / 2, where f s is the signal sampling frequency; Hyperbolic Radon transform real semi-axis length search end value a end Initialized as: a start < a end ≤ f s / 2 positive real number; Hyperbolic Radon transform real semi-axis length search number N a Initialization: 100≤ N a Positive integer <250; The real semi-axis length search set a of the hyperbolic Radon transform is initialized as: ,in, n a is the real semi-axis length index, n a = 0, 1, … , N a -1, For the n a Search for the real semi-axis length value, and there is ; Hyperbolic Radon transform imaginary semi-axis length search starting value b start Initialized as: b start Positive real numbers > 0; Hyperbolic Radon transform imaginary semi-axis length search end value b end Initialized as: b start < b end Positive real numbers <+∞; Hyperbolic Radon transform imaginary semi-axis length search number N b Initialization: 100≤ N b Positive integer <250; The imaginary semi-axis length search set b of the hyperbolic Radon transform is initialized as: ,in, n b is the imaginary semi-axis length index, n b = 0, 1, … , N b -1, For the n b Search for the imaginary semi-axis length value, and there is ; Time-frequency distribution time interval Δ t Initialized to: Δ t = step / f s ; Time-frequency distribution frequency interval Δ f Initialized to: Δ f = f s / N w ; Strip hyperbolic Radon transform strip width search starting value d start Initialized as: d start =0; Strip hyperbolic Radon transform strip width search number N d Initialized as: N d = N a ; The strip width search set D of the strip hyperbolic Radon transform is initialized as: ,in, n d is the stripe width index, n d = 0,1, … , N d -1, For the n d Search stripe width values, and ; Optimal stripe width decision threshold TH d Initialized to: 0< TH d Positive real number < 0.3; (4-2) Calculate the i The hyperbolic Radon transform results of the signal components HR i ( n a , n b ): ； (4-3) Calculate the i The optimal real semi-axis length index of the signal component and the optimal imaginary semi-axis length index : in, Indicates that when i The hyperbolic Radon transform results of the signal components HR i ( n a , n b ) corresponds to the real semi-axis length index when the maximum value is obtained n a and imaginary semi-axis length index n b the collection it consists of; (4-4) Calculate the i The maximum value of the hyperbolic Radon transform results of the signal components : ； (4-5) i The hyperbolic Radon transform results of the signal components HR i ( n a , n b ) is normalized to obtain i The normalized hyperbolic Radon transform results of the signal components : ； (4-6) Calculate the i The optimal strip hyperbolic Radon transform result of the signal components : Among them, max(·,·) means taking the maximum value of the two, and min(·,·) means taking the minimum value of the two. η Added new stripe width index; (4-7) Calculate the i The difference function of the signal components : in, m d is the difference stripe width index; (4-8) Screen all the values ​​that satisfy the difference function Less than the optimal stripe width decision threshold TH d The difference strip width index set : ； (4-9) Calculate the i The optimal stripe width index of the signal component : Among them, Ф is an empty set, == means judging whether the elements contained in the left and right sets are exactly the same. Indicates the difference strip width index set The first element of .

6. A method for estimating ship hyperbolic frequency modulation signal parameters based on direction pre-judgment according to claim 5, characterized in that: In step (5), the following method is used to calculate the i The time parameter estimation value of a signal component specifically includes the following steps: (5-1) To calculate the i Initialize the parameters of the time parameter estimation value of each signal component, including the initialization of the following parameters: Time frame signal has decision threshold coefficient Initialized as: A positive real number of Maximum frequency interval threshold Initialized as: A positive integer; Upper limit of total number of frames at no signal frequency Initialized as: A positive integer; Total number of frames at no-signal frequency points N non Initialized as: N non =0; Total number of signal component time frames N total Initialized as: N total =0; Current decision frame index q i Initialized as: q i =0; The frame index set where the signal component frequency exists Initialized as: ; (5-2) Calculate the frequency index set on the hyperbolic strip and the frequency index set under the hyperbolic strip , No. n t Elements and No. n t Elements They are: in, 、 and Respectively The search strip width value, The search real semi-axis length value and the Search for the imaginary semi-axis length value; (5-3) Calculate the i The set of hyperbolic band frequency upper limits for signal components and the set of lower frequency limits of hyperbolic strips , No. n t Elements and No. n t Elements They are: in, and The length is N t A collection of (5-4) Calculate the i Pre-separation time-frequency distribution of signal components : ； (5-5) Calculate the maximum modulus of the pre-separation time-frequency distribution : Among them, max{·} is the maximum value function; (5-6) Calculation of the frame signal existence judgment threshold : ； (5-7) Calculate the separation frequency index : in, Indicates the i The signal component is indexed at the current decision frame q i Pre-separated time-frequency distribution The frequency index of the time-frequency distribution corresponding to the maximum value; (5-8) Determine the pre-separation time-frequency distribution Frame index at the current decision time q i Whether there is a signal frequency point, that is, whether the following conditions are met: If it is established, go to step (5-9), otherwise, let the total number of time frames with no signal frequency be N non = N non +1 and proceed to steps (5-12); (5-9) Determine whether the signal component frequency point exists in the time frame index set Is it an empty set? That is, determine whether the following conditions are met: If so, go to step (5-11), otherwise, go to step (5-10); (5-10) Determine whether the following conditions are met: If it is established, go to step (5-11), otherwise, let the total number of time frames with no signal frequency be N non = N non +1 and proceed to steps (5-12); (5-11) Let the total number of signal component time frames be N total = N total +1, total number of frames with no signal frequency N non =0, and update the i The signal component frequency points of the signal components exist in the time frame index set : ； (5-12) Judgement i Whether the signal component reaches the end time frame, that is, whether the following conditions are met: If so, go to step (5-14), otherwise, go to step (5-13); (5-13) Update the current decision time frame index q i = q i +1, check whether the following conditions are met: If so, return to step (5-8), otherwise, go to step (5-14); (5-14) Calculate the i The starting time frame index of the signal component , End time frame index : in, Indicates the time frame index set where the signal component frequency exists The first element of Indicates the time frame index set where the signal component frequency exists The last element of .

7. A method for estimating ship hyperbolic frequency modulation signal parameters based on direction pre-judgment according to claim 6, characterized in that: In step (6), the following method is used to calculate the i The frequency parameter estimation value of each signal component and the update of the residual time-frequency distribution are obtained, which specifically includes the following steps: (6-1) Calculate the i The signal components of the signal component are continuously present in the frame index set : ； (6-2) Calculate the i The total number of consecutive time frames of signal components : ； (6-3) Calculate the i Separation time-frequency distribution of signal components : ； (6-4) According to i Separation time-frequency distribution of signal components , starting time frame index and the ending time frame index , extract the i The instantaneous frequency index set of signal components , No. Elements for: Among them, in i Each signal component has a time frame index Department, Represents the separation of time-frequency distribution The frequency index of the time-frequency distribution corresponding to the maximum value; (6-5) i The instantaneous frequency index set of signal components The inverse of the least squares linear fitting is performed to obtain the i frequency estimation parameter vector of the signal components : Among them, X is the least squares linear fitting auxiliary matrix, , G i For the i The reciprocal vector of the instantaneous frequency index set of the signal components, , the superscript T indicates the matrix transpose; (6-6) Get the i The estimated value of the periodic slope of the signal components and starting frequency estimates : Among them, θ i (1) and θ i (2) indicates the i The frequency estimation parameter vector θ of the signal components i The first and second elements of ; (6-7) Update the residual time-frequency distribution : 。 8. A method for estimating ship hyperbolic frequency modulation signal parameters based on direction pre-judgment according to claim 7, characterized in that: In step (7), the following method is used to determine whether there are still unseparated signal components, which specifically includes the following steps: (7-1) Decision threshold for determining whether there are still unseparated signal components TH S Initialization: 80≤ TH S Positive real numbers ≤ 160; (7-2) Calculate the residual time-frequency distribution The maximum value : ； (7-3) Determine whether there are still unseparated signal components, that is, determine whether the following conditions are met: If the condition is met, let the signal component discrete index i = i +1 and return to step (3); otherwise, go to step (8).

9. A method for estimating ship hyperbolic frequency modulation signal parameters based on direction pre-judgment according to claim 8, characterized in that: In step (8), the following method is used to calculate and output the estimated value of the number of signal components and the estimation results of time and frequency parameters, which specifically includes the following steps: (8-1) Calculate the estimated number of signal components And output: ； (8-2) Output time parameter estimation results: (8-3) Output frequency parameter estimation results: Among them, the signal component discrete index .

Images