A Method for Estimating Parameters of Ship Hyperbolic Frequency Modulation Signal Based on Direction Pre-Decision

Abstract

This invention discloses a method for estimating parameters of ship hyperbolic frequency modulation signals based on direction pre-decision, comprising the following steps: (1) acquiring the ship hyperbolic frequency modulation signal data sequence to be processed; (2) initializing the residual time-frequency distribution of the ship hyperbolic frequency modulation signal data sequence to be processed; (3) making signal direction determination based on the residual time-frequency distribution and calculating the flipped time frame index set; (4) calculating the separation strip parameter based on the residual time-frequency distribution and the flipped time frame index set; (5) calculating the time parameter estimate based on the separation strip parameter; (6) calculating the frequency parameter estimate based on the time parameter estimate and updating the residual time-frequency distribution; (7) determining whether there are still unseparated signal components based on the residual time-frequency distribution. If there are still unseparated signal components, return to step (3); otherwise, proceed to step (8); (8) calculating and outputting the signal component number estimate, time parameter estimate result, and frequency parameter estimate result.

Description

Technical Field

[0001] This invention belongs to the field of signal processing, specifically relating to a method for estimating parameters of ship hyperbolic frequency modulation signals based on direction pre-decision. Background Technology

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

[0003] Currently, there are four main types of commonly used methods for estimating parameters of ship hyperbolic frequency modulation (FM) signals: (1) mapping the ship hyperbolic FM signal to a two-dimensional spike, such as the maximum likelihood estimation method and the HST method; (2) extracting the time-frequency ridge of the ship hyperbolic FM 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; and (4) analyzing the spectrum of the ship hyperbolic FM signal and deriving the relationship between the signal spectrum and the parameters. However, the above research on ship hyperbolic FM signal parameter estimation methods mainly targets single-component ship hyperbolic FM signals or only focuses on the estimation of frequency parameters, without the ability to estimate time parameters. This makes it difficult to solve the problem of estimating both time and frequency parameters of multi-component ship hyperbolic FM signals with partial or complete time overlap, which has limitations. In addition, the traditional Radon transform method is only applicable to ship linear frequency modulation (FM) signals and has poor focusing on ship hyperbolic FM signals, i.e., there is a spike ambiguity problem, which affects the performance of parameter estimation. Summary of the Invention

[0004] The technical problem is that the purpose of this invention is to estimate the number of components, time and frequency parameters of a multi-component ship hyperbolic frequency modulation signal by separating the time-frequency distribution of the single-component ship hyperbolic frequency modulation signal one by one.

[0005] Technical Solution: To solve the above-mentioned technical problems and achieve the above-mentioned objectives, this invention proposes a method for estimating parameters of ship hyperbolic frequency modulation signals based on direction pre-decision. This method includes the following steps:

[0006] (1) Obtain the hyperbolic frequency modulation signal data sequence of the ship to be processed;

[0007] (2) Initialize the residual time-frequency distribution of the hyperbolic frequency modulated signal data sequence of the ship to be processed;

[0008] (3) Determine the signal direction based on the residual time-frequency distribution and calculate the set of flipped time frame indices;

[0009] (4) Calculate the separation strip parameters based on the residual time-frequency distribution and the set of flipped time frame indices;

[0010] (5) Calculate the estimated value of the time parameter based on the separation strip parameters;

[0011] (6) Calculate the estimated value of the frequency parameter based on the estimated value of the time parameter, and update the residual time-frequency distribution;

[0012] (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);

[0013] (8) Calculate and output the estimated value of the number of signal components, the estimation result of the time parameter, and the estimation result of the frequency parameter.

[0014] Furthermore, in step (1), the following method is used to obtain the data sequence of the ship's double chirp signal to be processed, which specifically includes the following steps:

[0015] Receive the real-time acquisition data of N sampling points from the ship's underwater acoustic sensor as the data sequence x(n) of the ship's double chirp signal to be processed, or extract the data of N sampling points starting from the signal detection moment from the memory as the data sequence x(n) of the ship's double chirp signal to be processed, where n = 0, 1,..., N - 1, and the N is the number of sampling points corresponding to the pulse width length of the detected ship's double chirp signal, and the value is N = 2 a , and a is a positive integer greater than or equal to 9.

[0016] Furthermore, in step (2), the following method is used to initialize the residual time-frequency distribution, which specifically includes the following steps:

[0017] (2-1) Initialize the parameters of the initialized residual time-frequency distribution, specifically including the initialization of the following parameters:

[0018] The window length N of the sliding time window (5) Calculate the estimated value of the time parameter based on the separation strip parameters;

[0011] (6) Calculate the estimated value of the frequency parameter based on the estimated value of the time parameter, and update the residual time-frequency distribution;

[0012] (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);

[0013] (8) Calculate and output the estimated value of the number of signal components, the estimation result of the time parameter, and the estimation result of the frequency parameter.

[0014] Furthermore, in step (1), the following method is used to obtain the data sequence of the ship's double chirp signal to be processed, which specifically includes the following steps:

[0015] Receive the real-time acquisition data of N sampling points from the ship's underwater acoustic sensor as the data sequence x(n) of the ship's double chirp signal to be processed, or extract the data of N sampling points starting from the signal detection moment from the memory as the data sequence x(n) of the ship's double chirp signal to be processed, where n = 0, 1,..., N - 1, and the N is the number of sampling points corresponding to the pulse width length of the detected ship's double chirp signal, and the value is N = 2 a , and a is a positive integer greater than or equal to 9.

[0016] Furthermore, in step (2), the following method is used to initialize the residual time-frequency distribution, which specifically includes the following steps:

[0017] (2-1) Initialize the parameters of the initialized residual time-frequency distribution, specifically including the initialization of the following parameters:

[0018] The window length N of the sliding time window w Initialize to: 16 < N w A positive integer of <floor{N / 6}, where floor{·} is the floor function;

[0019] The step step of the sliding time window is initialized to: 1 ≤ step ≤ floor{N w / 4} positive integer;

[0020] The total number of time frames N of the residual time-frequency distribution t Initialize to: N t = floor{(N - N w

[0011] (6) Calculate the estimated value of the frequency parameter based on the estimated value of the time parameter, and update the residual time-frequency distribution;

[0012] (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);

[0013] (8) Calculate and output the estimated value of the number of signal components, the estimation result of the time parameter, and the estimation result of the frequency parameter.

[0014] Furthermore, in step (1), the following method is used to obtain the data sequence of the ship's double chirp signal to be processed, which specifically includes the following steps:

[0015] Receive the real-time acquisition data of N sampling points from the ship's underwater acoustic sensor as the data sequence x(n) of the ship's double chirp signal to be processed, or extract the data of N sampling points starting from the signal detection moment from the memory as the data sequence x(n) of the ship's double chirp signal to be processed, where n = 0, 1,..., N - 1, and the N is the number of sampling points corresponding to the pulse width length of the detected ship's double chirp signal, and the value is N = 2 a , and a is a positive integer greater than or equal to 9.

[0016] Furthermore, in step (2), the following method is used to initialize the residual time-frequency distribution, which specifically includes the following steps:

[0017] (2-1) Initialize the parameters of the initialized residual time-frequency distribution, specifically including the initialization of the following parameters:

[0018] The window length N of the sliding time window w Initialize to: 16 < N w A positive integer of <floor{N / 6}, where floor{·} is the floor function;

[0019] The step step of the sliding time window is initialized to: 1 ≤ step ≤ floor{N w / 4} positive integer;

[0020] The total number of time frames N of the residual time-frequency distribution t Initialize to: N t = floor{(N - N w ) / step} - 1;

[0021] The total number of frequency points N of the residual time-frequency distributionf Initialized to: N f =floor{N w / 2}+1;

[0022] The discrete index i of the signal component is initialized to: i = 1;

[0023] (2-2) Calculate the time frame index n of the time-frequency distribution t Corresponding segmented data sequence

[0024]

[0025] Where, n t For the time-frequency distribution time frame index, n t =0,1,…,N t -1, r is the time discrete index of the segmented data sequence;

[0026] (2-3) Calculate the residual time-frequency distribution of the hyperbolic frequency modulated signal data sequence x(n) of the ship to be processed.

[0027]

[0028] Where, n f For the frequency point index of the time-frequency distribution, n f =0,1,…,N f -1, |·| represents the modulo value, where j is the imaginary unit, i.e.

[0029] Furthermore, in step (3), the direction decision of the i-th signal component is made using the following method, and the flip-time frame index set of the i-th signal component is calculated, specifically including the following steps:

[0030] (3-1) Initialize the parameters of the frame index set when making the direction decision for the i-th signal component and calculating the flip of the i-th signal component, specifically including the initialization of the following parameters:

[0031] Slope distance search starting value ρ start Initialize to: Where round{·} is the rounding function;

[0032] Number of slant distance searches N ρ Initialize to:

[0033] The slant range search set ρ is initialized as follows: Where, n ρ For the slant distance index, n ρ =0,1,…,N ρ -1, For the nth ρ There are search slope distance values, and there are

[0034] Angle search starting value θ start Initialized to: θ start =0°;

[0035] Angle search termination value θ end Initialized to: θ end =179°;

[0036] Number of angles searched N θ Initialize to: 90≤N θ Positive integers ≤180;

[0037] The angle search set θ is initialized as follows: Where, n θ For the angle index, n θ =0,1,…,N θ -1, For the nth θ There are search angle values, and there are

[0038] (3-2) Calculate the linear Radon transform result R of the i-th signal component. i (n ρ ,n θ ):

[0039]

[0040] Where δ{·} is the impulse function, i.e.:

[0041]

[0042] (3-3) Calculate the peak slant index of the i-th signal component. and peak angle index

[0043]

[0044] in, R represents the linear Radon transform result of the i-th signal component. i (n ρ ,n θ The slope index n corresponding to the maximum value ρ and angle index n θ The set that constitutes;

[0045] (3-4) Calculate the frame index set at the time of flipping The ntht element for:

[0046]

[0047] in, For the first Each search angle value, when When, it indicates that the i-th signal component is a ship hyperbolic up-modulated signal, when When, it indicates that the i-th signal component is a ship hyperbolic down-modulated signal.

[0048] Furthermore, in step (4), the separation strip parameters of the i-th signal component are calculated using the following method, specifically including the following steps:

[0049] (4-1) Initialize the parameters for calculating the separation strip parameters of the i-th signal component, specifically including the initialization of the following parameters:

[0050] Hyperbolic Radon transform real semi-axis length search starting value a start Initialized to: 0 start <f s / 2 positive real numbers, where f s The signal sampling frequency;

[0051] Hyperbolic Radon transform real semi-axis length search termination value a end Initialized as: a start end ≤f s Positive real numbers of 2 / 2;

[0052] Number of searches for the real half-axis length of the hyperbolic Radon transform N a Initialize to: 100≤N a Positive integers less than 250;

[0053] The hyperbolic Radon transform real semi-axis length search set a is initialized as follows: Where, n a n is the real semi-axis long index. a =0,1,…,N a -1, For the nth a Each search term represents a real semi-axis length value, and has

[0054] Starting value b for searching the imaginary semi-axis length of the hyperbolic Radon transform start Initialized as: b start Positive real numbers greater than 0;

[0055] Hyperbolic Radon transform imaginary semi-axis length search termination value b end Initialized to: b​​start end Positive real numbers less than +∞;

[0056] Number of searches for the imaginary semi-axis length of the hyperbolic Radon transform, N b Initialize to: 100≤N b Positive integers less than 250;

[0057] The hyperbolic Radon transform imaginary semi-axis length search set b is initialized as follows: Where, n b n is the long index of the virtual semi-axis. b =0,1,…,N b -1, For the nth b There are search values ​​for the length of the imaginary half-axis, and there are

[0058] The time interval Δt of the time-frequency distribution is initialized as: Δt = step / f s ;

[0059] The time-frequency distribution frequency interval Δf is initialized as follows: Δf = f s / N w ;

[0060] Strip hyperbolic Radon transform strip width search starting value d start Initialized to: d start =0;

[0061] Number of strip widths searched for N in the hyperbolic Radon transform of strips d Initialized to: N d =N a ;

[0062] The strip hyperbolic Radon transform strip width search set D is initialized as follows: Where, n d n is the stripe width index. d =0,1,…,N d -1, For the nth d Each search bar width value, and has

[0063] Optimal strip width decision threshold TH d Initialized to: 0 <TH d Positive real numbers less than 0.3;

[0064] (4-2) Calculate the hyperbolic Radon transform result HR of the i-th signal component. i (n a ,n b ):

[0065] ​

[0066] (4-3) Calculate the index of the optimal real semi-axis length of the i-th signal component. and the optimal virtual half-axis length index

[0067]

[0068] in, The hyperbolic Radon transform result HR of the i-th signal component is represented by... i (n a ,n b The index n of the real semi-axis length when the maximum value is obtained. a and the long index of the virtual half axis n b The set that constitutes;

[0069] (4-4) Calculate the maximum value of the hyperbolic Radon transform result of the i-th signal component.

[0070]

[0071] (4-5) Hyperbolic Radon transform result for the i-th signal component HR i (n a ,n b Normalization is performed to obtain the normalized hyperbolic Radon transform result of the i-th signal component.

[0072]

[0073] (4-6) Calculate the optimal strip hyperbolic Radon transform result for the i-th signal component.

[0074]

[0075] Where max(·,·) represents taking the maximum value of the two, min(·,·) represents taking the minimum value of the two, and η is the index of the newly added strip width;

[0076] (4-7) Calculate the difference function Diff for the i-th signal component. i EHR (m d ):

[0077]

[0078] Where, m d Index for the difference stripe width;

[0079] (4-8) Filter all cases that satisfy the difference function Diff. i EHR (m d The decision threshold TH is less than the optimal strip width. d The set of indexes for the difference strip width

[0080]

[0081] (4-9) Calculate the optimal strip width index for the i-th signal component.

[0082]

[0083] Where Ф is the empty set, and == indicates whether the left and right sets contain exactly the same elements. Represents the set of indices for the difference in strip width. The first element.

[0084] Furthermore, in step (5), the estimated time parameter value of the i-th signal component is calculated using the following method, specifically including the following steps:

[0085] (5-1) Initialize the parameters for calculating the time parameter estimate of the i-th signal component, specifically including the initialization of the following parameters:

[0086] The time frame signal has a decision threshold coefficient. Initialize to: Positive real numbers;

[0087] Maximum frequency interval threshold Initialize to: Positive integers;

[0088] Maximum number of frames when there is no signal frequency Initialize to: Positive integers;

[0089] Total number of frames N when there is no signal frequency non Initialized to: N non =0;

[0090] The total number of frames N for signal components total Initialized to: N total =0;

[0091] Current decision frame index q i Initialized to: q i =0;

[0092] Frame index set when signal component frequency points exist Initialize to:

[0093] (5-2) Calculate the set of frequency indexes on the hyperbolic stripe and hyperbolic strip sub-frequency index set The nth t element and The nth t element They are respectively:

[0094]

[0095]

[0096] in, and The first The search bar width value, the first The search for the real semi-axis length value and the first... Each search term represents the length of the virtual half-axis.

[0097] (5-3) Calculate the set of upper limits of hyperbolic strip frequencies F for the i-th signal component. i high and the lower limit set of hyperbolic strip frequencies F i low F i high The nth t Elements F i high (n t ) and F i low The nth t Elements F i low (n t They are respectively:

[0098]

[0099] Among them, F i high and F i low All are of length N t ;

[0100] (5-4) Calculate the pre-separated time-frequency distribution of the i-th signal component.

[0101]

[0102] (5-5) Calculate the maximum modulus V of the pre-separated time-frequency distribution. i max :

[0103]

[0104] Where max{·} is the function for finding the maximum value;

[0105] (5-6) A decision threshold exists in the frame signal during calculation.

[0106]

[0107] (5-7) Calculate the separation frequency index

[0108]

[0109] in, The frame index q represents the i-th signal component at the current decision time. i Pre-separation time-frequency distribution The index of the time-frequency distribution point corresponding to the maximum value;

[0110] (5-8) Determine the time-frequency distribution of pre-separation At the current judgment time frame index q i Does a signal frequency point exist? That is, determine whether the following conditions are met:

[0111]

[0112] If true, proceed to step (5-9); otherwise, set the total number of frames N when there is no signal frequency. non =N non +1, and proceed to step (5-12);

[0113] (5-9) Determine the frame index set when the frequency points of the signal components exist. To determine if a set is empty, we need to check if the following conditions are true:

[0114]

[0115] If true, proceed to step (5-11); otherwise, proceed to step (5-10).

[0116] (5-10) Determine whether the following conditions are true:

[0117]

[0118] If true, proceed to step (5-11); otherwise, set the total number of frames N when there is no signal frequency. non =N non +1, and proceed to step (5-12);

[0119] (5-11) Let N be the total number of time frames for the signal components. total =N total +1, the total number of frames N when there is no signal frequency. non =0, and update the set of frame indices when the frequency point of the i-th signal component exists.

[0120]

[0121] (5-12) Determine whether the i-th signal component has reached the termination frame, i.e., determine whether the following conditions are met:

[0122] and

[0123] If true, proceed to step (5-14); otherwise, proceed to step (5-13).

[0124] (5-13) Update the current judgment time frame index q i =q i +1, determine if the following conditions are true:

[0125] q i ≤N t -1

[0126] If true, return to step (5-8); otherwise, proceed to step (5-14).

[0127] (5-14) Calculate the starting time frame index of the i-th signal component. Termination Frame Index

[0128]

[0129] in, This represents the set of frame indices when the frequency points of the signal components exist. The first element, This represents the set of frame indices when the frequency points of the signal components exist. The last element.

[0130] Furthermore, in step (6), the frequency parameter estimate of the i-th signal component is calculated using the following method, and the residual time-frequency distribution is updated, specifically including the following steps:

[0131] (6-1) Calculate the set of frame indices T when the signal components of the i-th signal component are continuously present. i exist :

[0132]

[0133] (6-2) Calculate the total number of consecutive time frames for the i-th signal component.

[0134]

[0135] (6-3) Calculate the separated time-frequency distribution of the i-th signal component.

[0136]

[0137] (6-4) Based on the separation time-frequency distribution of the i-th signal component Start-up frame index and the frame index at termination Extract the instantaneous frequency point index set f of the i-th signal component i f i The element for:

[0138]

[0139] Among them, each signal component of the i-th signal component has a time frame index. place, This indicates that the time-frequency distribution is separated. The index of the time-frequency distribution point corresponding to the maximum value;

[0140] (6-5) The set of instantaneous frequency indexes f for the i-th signal component i By performing least-squares linear fitting on the reciprocal of the vector, the frequency estimation parameter vector θ of the i-th signal component is obtained. i :

[0141] θ i =(X T X) -1 X T G i

[0142] Where X is the least squares linear fitting auxiliary matrix. G i Let be the reciprocal vector of the instantaneous frequency point index set of the i-th signal component. The superscript T indicates matrix transpose;

[0143] (6-6) Obtain the estimated period slope value of the i-th signal component. and initial frequency estimate

[0144]

[0145] Where, θ i (1) and θi (2) represents the frequency estimation parameter vector θ of the i-th signal component. i The first and second elements;

[0146] (6-7) Update the residual time-frequency distribution

[0147]

[0148] Furthermore, in step (7), the following method is used to determine whether there are still unseparated signal components, specifically including the following steps:

[0149] (7-1) Decision threshold TH for determining whether there are still unseparated signal components S Initialized as: 80≤TH S Positive real numbers ≤160;

[0150] (7-2) Calculate the residual time-frequency distribution maximum value

[0151]

[0152] (7-3) Determine whether there are still unseparated signal components, that is, determine whether the following conditions are met:

[0153]

[0154] If the condition is met, let the discrete index of the signal component be i = i + 1, and return to step (3); otherwise, proceed to step (8).

[0155] Furthermore, in step (8), the estimated number of signal components and the estimated time and frequency parameters are calculated and output using the following method, specifically including the following steps:

[0156] (8-1) Calculate the estimated number of signal components And output:

[0157]

[0158] (8-2) Output the time parameter estimation results:

[0159]

[0160] (8-3) Output frequency parameter estimation results:

[0161]

[0162]

[0163] Among them, the discrete index of signal components

[0164] Beneficial effects: Compared with the prior art, the technical solution of the present invention has the following beneficial technical effects:

[0165] 1. This invention fully utilizes the characteristics that the angle corresponding to the peak of the linear Radon transform of the ship's hyperbolic up-modulated signal is an acute angle and the angle corresponding to the peak of the linear Radon transform of the ship's hyperbolic down-modulated signal is an obtuse angle, and realizes the direction pre-determination of the ship's hyperbolic up-modulated signal, as shown in step (3). This decision process can provide prior estimation information for subsequent hyperbolic Radon transform.

[0166] 2. This invention makes full use of the instantaneous frequency curve characteristics of the ship hyperbolic frequency modulation signal, and fits the time-frequency distribution of the ship hyperbolic frequency modulation signal with the standard hyperbolic equation, as shown in step (4). This fitting process realizes the pre-estimation of the frequency center ridge line required to separate the signal components, which provides a basis for the subsequent separation process of the time-frequency distribution of the current signal components.

[0167] 3. This invention fully utilizes the focusing characteristics of the ship's hyperbolic frequency modulation 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 to separate 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 components and provide accuracy for subsequent time parameter estimation of the signal components.

[0168] 4. By separating the signal components one by one, the present invention transforms the problem of detection and parameter estimation of multi-component signals into the problem of detection and parameter estimation of single-component signals, as shown in step (6). This 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 signal frequency parameter estimation.

[0169] 5. This invention fully utilizes the time-frequency domain distribution characteristics of ship hyperbolic frequency modulation signals to realize the automatic judgment of whether signal components are completely detected. As shown in step (7), by using the peak value of the residual time-frequency distribution, it is possible to detect signal components that have not yet been parameter estimated, thereby providing a basis for subsequent time and frequency parameter estimation of multi-component ship hyperbolic frequency modulation signals. Attached Figure Description

[0170] Figure 1 This is a schematic flowchart of the method of the present invention;

[0171] Figure 2 The residual time-frequency distribution of the first signal component in Example 1;

[0172] Figure 3 The result of the hyperbolic Radon transform of the first signal component in Example 1;

[0173] Figure 4 The time-frequency distribution of the separated first signal component in Example 1;

[0174] Figure 5 The residual time-frequency distribution of the second signal component in Example 1;

[0175] Figure 6 The result of the hyperbolic Radon transform of the second signal component in Example 1;

[0176] Figure 7 The time-frequency distribution of the separated second signal component in Example 1;

[0177] Figure 8 This is the residual time-frequency distribution of the third signal component in Example 1. Detailed Implementation

[0178] To better understand the purpose, structure, and function of this invention, the invention will be further described below with reference to the accompanying drawings.

[0179] like Figure 1 As shown, this invention proposes a method for estimating parameters of a ship's hyperbolic frequency modulated signal based on direction pre-decision, comprising the following steps:

[0180] (1) Obtain the hyperbolic frequency modulation signal data sequence of the ship to be processed;

[0181] (2) Initialize the residual time-frequency distribution of the hyperbolic frequency modulated signal data sequence of the ship to be processed;

[0182] (3) Determine the signal direction based on the residual time-frequency distribution and calculate the set of flipped time frame indices;

[0183] (4) Calculate the separation strip parameters based on the residual time-frequency distribution and the flipped time frame index set;

[0184] (5) Calculate the estimated value of the time parameter based on the separation strip parameter;

[0185] (6) Calculate the estimated value of the frequency parameter based on the estimated value of the time parameter, and update the residual time-frequency distribution;

[0186] (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, proceed to step (8).

[0187] (8) Calculate and output the estimated number of signal components, the estimated time parameter and the estimated frequency parameter.

[0188] Further, in step (1), the following method is used to obtain the dual-tone frequency modulation signal data sequence of the ship to be processed, which specifically includes the following steps: Receive the real-time acquisition data of N sampling points from the ship's underwater acoustic sensor as the dual-tone frequency modulation signal data sequence x(n) of the ship to be processed, or extract the data of N sampling points starting from the signal detection moment from the memory as the dual-tone frequency modulation signal data sequence x(n) of the ship to be processed, where n = 0, 1,..., N-1, and the N is the number of sampling points corresponding to the pulse width length of the detected dual-tone frequency modulation signal of the ship, and the value is N = 2 a , where a is a positive integer greater than or equal to 9.

[0189] Further, in step (2), the following method is used to initialize the residual time-frequency distribution, which specifically includes the following steps:

[0190] (2-1) Initialize the parameters of the initialized residual time-frequency distribution, which specifically includes the initialization of the following parameters:

[0191] The length N of the sliding time window w Initialize to: 16 < N w a positive integer of <floor{N / 6}, where floor{·} is the floor function;

[0192] The step step of the sliding time window is initialized to: 1 ≤ step ≤ floor{N w / 4} a positive integer;

[0193] The total number of time frames N of the residual time-frequency distribution t Initialize to: N t = floor{(N - N w ) / step} - 1;

[0194] The total number of frequency points N of the residual time-frequency distribution f Initialize to: N f = floor{N w / 2} + 1;

[0195] The discrete index i of the signal component is initialized to: i = 1; No content here, seems to be a formatting issue in the original text.

[0196] No content here, seems to be a formatting issue in the original text. (2-2) Calculate the segmented data sequence corresponding to the time frame index n t corresponding segmented data sequence

[0197]

[0198] where n t is the time frame index of the time-frequency distribution, n t = 0, 1,..., N t -1, and r is the time discrete index of the segmented data sequence;

[0199] (2-3) Calculate the residual time-frequency distribution of the hyperbolic frequency modulated signal data sequence x(n) of the ship to be processed.

[0200]

[0201] Where, n f For the frequency point index of the time-frequency distribution, n f =0,1,…,N f -1, |·| represents the modulo value, where j is the imaginary unit, i.e.

[0202] Furthermore, in step (3), the direction decision of the i-th signal component is made using the following method, and the flip-time frame index set of the i-th signal component is calculated, specifically including the following steps:

[0203] (3-1) Initialize the parameters of the frame index set when making the direction decision for the i-th signal component and calculating the flip of the i-th signal component, specifically including the initialization of the following parameters:

[0204] Slope distance search starting value ρ start Initialize to: Where round{·} is the rounding function;

[0205] Number of slant distance searches N ρ Initialize to:

[0206] The slant range search set ρ is initialized as follows: Where, n ρ n is the slant distance index. ρ =0,1,…,N ρ -1, For the nth ρ There are search slope distance values, and there are

[0207] Angle search starting value θ start Initialized to: θ start =0°;

[0208] Angle search termination value θ end Initialized to: θ end =179°;

[0209] Number of angles searched N θ Initialize to: 90≤N θ Positive integers ≤180;

[0210] The angle search set θ is initialized as follows: Where, n θ For the angle index, nθ =0,1,…,N θ -1, For the nth θ There are search angle values, and there are

[0211] (3-2) Calculate the linear Radon transform result R of the i-th signal component. i (n ρ ,n θ ):

[0212]

[0213] Where δ{·} is the impulse function, i.e.:

[0214]

[0215] (3-3) Calculate the peak slant index of the i-th signal component. and peak angle index

[0216]

[0217] in, R represents the linear Radon transform result of the i-th signal component. i (n ρ ,n θ The slope index n corresponding to the maximum value ρ and angle index n θ The set that constitutes;

[0218] (3-4) Calculate the frame index set at the time of flipping The nth t element for:

[0219]

[0220] in, For the first Each search angle value, when When, it indicates that the i-th signal component is a ship hyperbolic up-modulated signal, when When, it indicates that the i-th signal component is a ship hyperbolic down-modulated signal.

[0221] Furthermore, in step (4), the separation strip parameters of the i-th signal component are calculated using the following method, specifically including the following steps:

[0222] (4-1) Initialize the parameters for calculating the separation strip parameters of the i-th signal component, specifically including the initialization of the following parameters:

[0223] Hyperbolic Radon transform real semi-axis length search starting value a start Initialized to: 0 start <f s / 2 positive real numbers, where f s The signal sampling frequency;

[0224] Hyperbolic Radon transform real semi-axis length search termination value a end Initialized as: a start end ≤f s Positive real numbers of 2 / 2;

[0225] Number of searches for the real half-axis length of the hyperbolic Radon transform N a Initialize to: 100≤N a Positive integers less than 250;

[0226] The hyperbolic Radon transform real semi-axis length search set a is initialized as follows: Where, n a n is the real semi-axis long index. a =0,1,…,N a -1, For the nth a Each search term represents a real semi-axis length value, and has

[0227] Starting value b for searching the imaginary semi-axis length of the hyperbolic Radon transform start Initialized to: b start Positive real numbers greater than 0;

[0228] Hyperbolic Radon transform imaginary semi-axis length search termination value b end Initialized to: b start end Positive real numbers less than +∞;

[0229] Number of searches for the imaginary semi-axis length of the hyperbolic Radon transform, N b Initialize to: 100≤N b Positive integers less than 250;

[0230] The hyperbolic Radon transform imaginary semi-axis length search set b is initialized as follows: Where, n b n is the long index of the virtual semi-axis. b =0,1,…,N b -1, For the nth b There are search values ​​for the length of the imaginary half-axis, and there are

[0231] ​​​The time interval Δt of the time-frequency distribution is initialized as: Δt = step / f s ;

[0232] The time-frequency distribution frequency interval Δf is initialized as follows: Δf = f s / N w ;

[0233] Strip hyperbolic Radon transform strip width search starting value d start Initialized to: d start =0;

[0234] Number of strip widths searched for N in the hyperbolic Radon transform of strips d Initialized to: N d =N a ;

[0235] The strip hyperbolic Radon transform strip width search set D is initialized as follows: Where, n d n is the stripe width index. d =0,1,…,N d -1, For the nth d Each search bar width value, and has

[0236] Optimal strip width decision threshold TH d Initialized to: 0 <TH d Positive real numbers less than 0.3;

[0237] (4-2) Calculate the hyperbolic Radon transform result HR of the i-th signal component. i (n a ,n b ):

[0238]

[0239] (4-3) Calculate the index of the optimal real semi-axis length of the i-th signal component. and the optimal virtual half-axis length index

[0240]

[0241] in, The hyperbolic Radon transform result HR of the i-th signal component is represented by... i (n a ,n b The index n of the real semi-axis length when the maximum value is obtained. a and the long index of the virtual half axis n b The set that constitutes;

[0242] (4-4) Calculate the maximum value of the hyperbolic Radon transform result of the i-th signal component.

[0243]

[0244] (4-5) Hyperbolic Radon transform result for the i-th signal component HR i (n a ,n b Normalization is performed to obtain the normalized hyperbolic Radon transform result of the i-th signal component.

[0245]

[0246] (4-6) Calculate the optimal strip hyperbolic Radon transform result for the i-th signal component.

[0247]

[0248] Where max(·,·) represents taking the maximum value of the two, min(·,·) represents taking the minimum value of the two, and η is the index of the newly added strip width;

[0249] (4-7) Calculate the difference function Diff for the i-th signal component. i EHR (m d ):

[0250]

[0251] Where, m d Index for the difference stripe width;

[0252] (4-8) Filter all cases that satisfy the difference function Diff. i EHR (m d The decision threshold TH is less than the optimal strip width. d The set of indexes for the difference strip width

[0253]

[0254] (4-9) Calculate the optimal strip width index for the i-th signal component.

[0255]

[0256] Where Ф is the empty set, and == indicates whether the left and right sets contain exactly the same elements. Represents the set of indices for the difference in strip width. The first element.

[0257] Furthermore, in step (5), the estimated time parameter value of the i-th signal component is calculated using the following method, specifically including the following steps:

[0258] (5-1) Initialize the parameters for calculating the time parameter estimate of the i-th signal component, specifically including the initialization of the following parameters:

[0259] The time frame signal has a decision threshold coefficient. Initialize to: Positive real numbers;

[0260] Maximum frequency interval threshold Initialize to: Positive integers;

[0261] Maximum number of frames when there is no signal frequency Initialize to: Positive integers;

[0262] Total number of frames N when there is no signal frequency non Initialized to: N non =0;

[0263] The total number of frames N for signal components total Initialized to: N total =0;

[0264] Current decision frame index q i Initialized to: q i =0;

[0265] Frame index set when signal component frequency points exist Initialize to:

[0266] (5-2) Calculate the set of frequency indexes on the hyperbolic stripe and hyperbolic strip sub-frequency index set The nth t element and The nth t element They are:

[0267]

[0268] in, and The first The search bar width value, the first The search for the real semi-axis length value and the first search... Each search term represents the length of the virtual half-axis.

[0269] (5-3) Calculate the set of upper limits of hyperbolic strip frequencies F for the i-th signal component. i high and the lower limit set of hyperbolic strip frequencies F i low F i high The nth t Elements F i high (n t ) and F i low The nth t Elements F i low (n t They are respectively:

[0270]

[0271] Among them, F i high and F i low All are of length N t ;

[0272] (5-4) Calculate the pre-separated time-frequency distribution of the i-th signal component.

[0273]

[0274] (5-5) Calculate the maximum modulus V of the pre-separated time-frequency distribution. i max :

[0275]

[0276] Where max{·} is the function for finding the maximum value;

[0277] (5-6) A decision threshold exists in the frame signal during calculation.

[0278]

[0279] (5-7) Calculate the separation frequency index

[0280]

[0281] in, The frame index q represents the i-th signal component at the current decision time. i Pre-separation time-frequency distribution The index of the time-frequency distribution point corresponding to the maximum value;

[0282] (5-8) Determine the time-frequency distribution of pre-separation At the current judgment time frame index q i Does a signal frequency point exist? That is, determine whether the following conditions are met:

[0283]

[0284] If true, proceed to step (5-9); otherwise, set the total number of frames N when there is no signal frequency. non =N non +1, and proceed to step (5-12);

[0285] (5-9) Determine the frame index set when the frequency points of the signal components exist. To determine if a set is empty, we need to check if the following conditions are true:

[0286]

[0287] If true, proceed to step (5-11); otherwise, proceed to step (5-10).

[0288] (5-10) Determine whether the following conditions are true:

[0289]

[0290] If true, proceed to step (5-11); otherwise, set the total number of frames N when there is no signal frequency. non =N non +1, and proceed to step (5-12);

[0291] (5-11) Let N be the total number of time frames for the signal components. total =N total +1, the total number of frames N when there is no signal frequency. non =0, and update the set of frame indices when the frequency point of the i-th signal component exists.

[0292]

[0293] (5-12) Determine whether the i-th signal component has reached the termination frame, i.e., determine whether the following conditions are met:

[0294] and

[0295] If true, proceed to step (5-14); otherwise, proceed to step (5-13).

[0296] (5-13) Update the current judgment time frame index qi =q i +1, determine if the following conditions are true:

[0297] q i ≤N t -1

[0298] If true, return to step (5-8); otherwise, proceed to step (5-14).

[0299] (5-14) Calculate the starting time frame index of the i-th signal component. Termination Frame Index

[0300]

[0301] in, This represents the set of frame indices when the frequency points of the signal components exist. The first element, This represents the set of frame indices when the frequency points of the signal components exist. The last element.

[0302] Furthermore, in step (6), the frequency parameter estimate of the i-th signal component is calculated using the following method, and the residual time-frequency distribution is updated, specifically including the following steps:

[0303] (6-1) Calculate the set of frame indices when the signal components of the i-th signal component are continuously present.

[0304]

[0305] (6-2) Calculate the total number of consecutive time frames for the i-th signal component.

[0306]

[0307] (6-3) Calculate the separated time-frequency distribution of the i-th signal component.

[0308]

[0309] (6-4) Based on the separation time-frequency distribution of the i-th signal component Start-up frame index and the frame index at termination Extract the instantaneous frequency point index set f of the i-th signal component i f i The element for:

[0310]

[0311] Among them, each signal component of the i-th signal component has a time frame index. place, This indicates that the time-frequency distribution is separated. The index of the time-frequency distribution point corresponding to the maximum value;

[0312] (6-5) The set of instantaneous frequency indexes f for the i-th signal component i By performing least-squares linear fitting on the reciprocal of the vector, the frequency estimation parameter vector θ of the i-th signal component is obtained. i :

[0313] θ i =(X T X) -1 X T G i

[0314] Where X is the least squares linear fitting auxiliary matrix. G i Let be the reciprocal vector of the instantaneous frequency point index set of the i-th signal component. The superscript T indicates matrix transpose;

[0315] (6-6) Obtain the estimated period slope value of the i-th signal component. and initial frequency estimate

[0316]

[0317] Where, θ i (1) and θ i (2) represents the frequency estimation parameter vector θ of the i-th signal component. i The first and second elements;

[0318] (6-7) Update the residual time-frequency distribution

[0319]

[0320] Furthermore, in step (7), the following method is used to determine whether there are still unseparated signal components, specifically including the following steps:

[0321] (7-1) Decision threshold TH for determining whether there are still unseparated signal components S Initialized as: 80≤TH S Positive real numbers ≤160;

[0322] (7-2) Calculate the residual time-frequency distribution maximum value

[0323]

[0324] (7-3) Determine whether there are still unseparated signal components, that is, determine whether the following conditions are met:

[0325]

[0326] If the condition is met, let the discrete index of the signal component be i = i + 1, and return to step (3); otherwise, proceed to step (8).

[0327] Furthermore, in step (8), the estimated number of signal components and the estimated time and frequency parameters are calculated and output using the following method, specifically including the following steps:

[0328] (8-1) Calculate the estimated number of signal components And output:

[0329]

[0330] (8-2) Output the time parameter estimation results:

[0331]

[0332] (8-3) Output frequency parameter estimation results:

[0333]

[0334] Among them, the discrete index of signal components

[0335] In the embodiments of the present invention, the simulated received ship hyperbolic frequency modulated signal model x(t) is:

[0336]

[0337] Where I is the number of signal components, i is the discrete index of the signal component, and τ 0,i τ is the start time of the i-th signal component. i Let A be the pulse width of the i-th signal component. i f 1,i f 2,i k corresponds to the amplitude, start frequency, and end frequency of the i-th signal component, respectively. 0,i The periodic slope of the i-th signal component is defined as: k 0,i =(f 2,i -f 1,i ) / (τ i f 1,i f 2,iw(t) is a function with a mean of zero and a variance of σ. 2 Gaussian white noise, variance σ 2 The magnitude depends on the signal-to-noise ratio (SNR):

[0338] With signal sampling frequency f s Discrete sampling is performed on the hyperbolic frequency modulated signal x(t) received in the above simulation to obtain the sampled data sequence x(n) of the hyperbolic frequency modulated signal of the ship:

[0339]

[0340] in, N i =round{τ i f s}

[0341] Example 1:

[0342] The parameters for the simulated ship hyperbolic frequency modulated signal are set as follows: number of signal components I = 2, signal sampling frequency f s =2kHz, signal-to-noise ratio (SNR) = -1dB, start time τ of the first signal component 0,1 =0.15s, pulse width τ1=0.35s, amplitude A1=1, starting frequency f 1,1 =700Hz, termination frequency f 2,1 =500Hz, period slope k 0,1 = -0.0016, meaning the first signal component is the ship's hyperbolic down-modulated frequency signal; the start time τ of the second signal component 0,2 =0.15s, pulse width τ2=0.25s, amplitude A2=1, starting frequency f 1,2 =230Hz, termination frequency f 2,2 =430Hz, period slope k 0,2 =0.0081, meaning the second signal component is the ship's hyperbolic upmodulation signal.

[0343] The following section describes the detection and frequency parameter estimation of the simulated ship hyperbolic frequency modulated signal:

[0344] According to step (1), extract the data of N = 1024 sampling points starting from the moment the signal is detected from the memory as the data sequence x(n) of the ship hyperbolic frequency modulation signal to be processed, n = 0, 1, ..., N-1;

[0345] Based on step (2), set the sliding time window length N. w =128, sliding time window step=32, total number of frames N for residual time-frequency distribution t =27, the total number of residual time-frequency distribution frequency points N f=65, signal component discrete index i=1; use STFT to obtain the residual time-frequency distribution of the ship hyperbolic frequency modulated signal data sequence x(n) to be processed. like Figure 2 As shown, proceed to step (3);

[0346] Based on step (3), set the initial value ρ for the slant distance search. start = -35, number of slant distance searches N ρ =71, starting value θ for angle search start =0°, the angle search termination value θ end =179°, number of angles searched N θ =180; for residual time-frequency distribution Calculate the peak angle index of the first signal component. because Therefore, the first signal component is the ship's hyperbolic down-modulated signal, and the calculated flip-time frame index set is obtained at this time. Proceed to step (4);

[0347] Based on step (4), set the initial value a for searching the real semi-axis length of the hyperbolic Radon transform. start =0.001, the termination value for the search of the real semi-axis length of the hyperbolic Radon transform. end =1000, the number of searches for the real half-axis length of the hyperbolic Radon transform N a =200, starting value b for searching the imaginary semi-axis length of the hyperbolic Radon transform. start =0.001, the termination value for searching the imaginary semi-axis length of the hyperbolic Radon transform. end =5, Number of searches for the imaginary half-axis length of the hyperbolic Radon transform N b =200, time-frequency distribution time interval Δt = 0.016, time-frequency distribution frequency interval Δf = 15.625, strip hyperbolic Radon transform strip width search starting value d start =0, the number of strip widths N searched in the hyperbolic Radon transform of the strip is 0. d =200, optimal strip width decision threshold TH d =0.1813; Calculate the hyperbolic Radon transform result HR1(n) of the first signal component. a ,n b ),like Figure 3 As shown, the optimal strip width index for calculating the first signal component is... Optimal real half-axis length index and the optimal virtual half-axis length index Proceed to step (5);

[0348] According to step (5), set the decision threshold coefficient for the existence of the time frame signal. Maximum frequency interval threshold Maximum number of frames when there is no signal frequency Total number of frames N when there is no signal frequency non =0, the total number of frames N when the signal components are 0 total =0, the current decision frame index q1=0, the set of frame indices when the signal component frequency point exists. Calculate the frame index at the start of the first signal component. Termination Frame Index Proceed to step (6);

[0349] Based on 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 period. and initial frequency estimates And update the residual time-frequency distribution like Figure 5 As shown, proceed to step (7);

[0350] Based on step (7), set a decision threshold TH to determine whether there are still unseparated signal components. S =80; Calculate the residual time-frequency distribution maximum value because This indicates that there are still unseparated signal components. Let the discrete index of the signal component i = 2, and return to step (3).

[0351] Based on step (3), set the initial value ρ for the slant distance search. start = -35, number of slant distance searches N ρ =71, starting value θ for angle search start =0°, the angle search termination value θ end =179°, number of angles searched N θ =180; for residual time-frequency distribution Calculate the peak angle index of the second signal component. because Therefore, the second signal component is the ship's hyperbolic up-modulated signal, and the calculated flip-time frame index set is obtained at this time. Proceed to step (4);

[0352] Based on step (4), set the initial value a for searching the real semi-axis length of the hyperbolic Radon transform. start =0.001, the termination value for the search of the real semi-axis length of the hyperbolic Radon transform. end =1000, the number of searches for the real half-axis length of the hyperbolic Radon transform N a =200, starting value b for searching the imaginary semi-axis length of the hyperbolic Radon transform. start=0.001, the termination value for searching the imaginary semi-axis length of the hyperbolic Radon transform. end =5, Number of searches for the imaginary half-axis length of the hyperbolic Radon transform N b =200, time-frequency distribution time interval Δt = 0.016, time-frequency distribution frequency interval Δf = 15.625, strip hyperbolic Radon transform strip width search starting value d start =0, the number of strip widths N searched in the hyperbolic Radon transform of the strip is 0. d =200, optimal strip width decision threshold TH d =0.1813; Calculate the hyperbolic Radon transform result HR2(n) of the second signal component. a ,n b ),like Figure 6 As shown, the optimal strip width index for calculating the second signal component is... Optimal real half-axis length index and the optimal virtual half-axis length index Proceed to step (5);

[0353] According to step (5), set the decision threshold coefficient for the existence of the time frame signal. Maximum frequency interval threshold Maximum number of frames when there is no signal frequency Total number of frames N when there is no signal frequency non =0, the total number of frames N when the signal components are 0 total =0, the current decision frame index q2=0, the set of frame indices when the signal component frequency point exists. Calculate the frame index at the start of the second signal component. Termination Frame Index Proceed to step (6);

[0354] Based on 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 initial frequency estimates And update the residual time-frequency distribution like Figure 8 As shown, proceed to step (7);

[0355] Based on step (7), set a decision threshold TH to determine whether there are still unseparated signal components. S =80; Calculate the residual time-frequency distribution maximum value because This indicates that there are no unseparated signal components, proceed to step (8);

[0356] Based on step (8), the estimated number of output signal components and the estimated results of time and frequency parameters are as follows:

[0357]

[0358] The test results indicate that the ship's hyperbolic frequency modulation signal is a multi-component hyperbolic frequency modulation signal, with the number of signal components consistent with the simulation settings, and the period slope... and starting frequency The relative errors are as follows:

[0359]

Claims

1. A method for estimating parameters of a ship's hyperbolic frequency modulated signal based on direction pre-decision, characterized in that, Includes the following steps: (1) Obtain the hyperbolic frequency modulation signal data sequence of the ship to be processed; (2) Initialize the residual time-frequency distribution of the hyperbolic frequency modulated signal data sequence of the ship to be processed; (3) Determine the signal direction based on the residual time-frequency distribution and calculate the set of flipped time frame indices; (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 parameter; (6) Calculate the estimated value of the frequency parameter based on the estimated value of the time parameter, 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, proceed to step (8). (8) Calculate and output the estimated number of signal components, the estimated time parameter, and the estimated frequency parameter; In step (3), the direction decision of the i-th signal component is made using the following method, and the flip frame index set of the i-th signal component is calculated, specifically including the following steps: (3-1) Initialize the parameters of the frame index set when making the direction decision for the i-th signal component and calculating the flip of the i-th signal component, specifically including the initialization of the following parameters: Slope distance search starting value ρ start Initialize to: Where round{·} is the rounding function; Number of slant distance searches N ρ Initialize to: The slant range search set ρ is initialized as follows: Where, n ρ For the slant distance index, n ρ =0,1,…,N ρ -1, For the nth ρ There are search slope distance values, and there are Angle search starting value θ start Initialized to: θ start =0°; Angle search termination value θ end Initialized to: θ end =179°; Number of angles searched N θ Initialize to: 90≤N θ Positive integers ≤180; The angle search set θ is initialized as follows: Where, n θ For the angle index, n θ =0,1,…,N θ -1, For the nth θ There are search angle values, and there are (3-2) Calculate the linear Radon transform result R of the i-th signal component. i (n ρ ,n θ ): Where δ{·} is the impulse function, i.e.: (3-3) Calculate the peak slant index of the i-th signal component. and peak angle index in, R represents the linear Radon transform result of the i-th signal component. i (n ρ ,n θ The slope index n corresponding to the maximum value ρ and angle index n θ The set that constitutes; (3-4) Calculate the frame index set at the time of flipping The nth t element for: in, For the first Each search angle value, when When, it indicates that the i-th signal component is a ship hyperbolic up-modulated signal, when When, it indicates that the i-th signal component is a ship hyperbolic down-modulated signal.

2. The method for estimating parameters of a ship's hyperbolic frequency modulated signal based on direction pre-decision as described in claim 1, characterized in that, In step (1), the hyperbolic frequency modulated signal data sequence of the ship to be processed is obtained using the following method, specifically including the following steps: The data can be obtained from real-time sampling points of N points received from the ship's acoustic sensors as the data sequence x(n) of the ship's hyperbolic frequency modulated signal to be processed, or the data from N sampling points starting from the moment the signal was detected can be extracted from the memory as the data sequence x(n) of the ship's hyperbolic frequency modulated signal to be processed, where n = 0, 1, ..., N-1, and N is the number of sampling points corresponding to the pulse width of the detected ship's hyperbolic frequency modulated signal, with a value of N = 2. a , where a is a positive integer greater than or equal to 9.

3. The method for estimating parameters of a ship's hyperbolic frequency modulated signal based on direction pre-decision as described in claim 2, characterized in that, In step (2), the residual time-frequency distribution is initialized using the following method, specifically including the following steps: (2-1) Initialize the parameters of the initial residual time-frequency distribution, specifically including the initialization of the following parameters: Sliding time window length N w Initialized as: 16 < N w Positive integer of <floor{N / 6}>, where floor{·} is the floor function; The sliding time window step is initialized as follows: 1 ≤ step ≤ floor{N} w Positive integers of size 4; Total number of time frames N in residual time-frequency distribution t Initialized to: N t =floor{(NN w ) / step}-1; Total number of residual time-frequency distribution frequency points N f Initialized to: N f =floor{N w / 2}+1; The discrete index i of the signal component is initialized to: i = 1; (2-2) Calculate the time frame index n of the time-frequency distribution t Corresponding segmented data sequence Where, n t For the time-frequency distribution time frame index, n t =0,1,…,N t -1, r is the time discrete index of the segmented data sequence; (2-3) Calculate the residual time-frequency distribution of the hyperbolic frequency modulated signal data sequence x(n) of the ship to be processed. Where, n f For the frequency point index of the time-frequency distribution, n f =0,1,…,N f -1, |·| represents the modulo value, where j is the imaginary unit, i.e.

4. The method for estimating parameters of a ship's hyperbolic frequency modulated signal based on direction pre-decision as described in claim 1, characterized in that, In step (4), the separation strip parameters of the i-th signal component are calculated using the following method, specifically including the following steps: (4-1) Initialize the parameters for calculating the separation strip parameters of the i-th signal component, specifically including the initialization of the following parameters: Hyperbolic Radon transform real semi-axis length search starting value a start Initialized to: 0 start <f s / 2 positive real numbers, where f s The signal sampling frequency;​ Hyperbolic Radon transform real semi-axis length search termination value a end Initialized as: a start end ≤f s Positive real numbers of 2 / 2;​ Number of searches for the real half-axis length of the hyperbolic Radon transform N a Initialize to: 100≤N a Positive integers less than 250; The hyperbolic Radon transform real semi-axis length search set a is initialized as follows: Where, n a n is the real semi-axis long index. a =0,1,…,N a -1, For the nth a Search for the real semi-axis length value, and have Starting value b for searching the imaginary semi-axis length of the hyperbolic Radon transform start Initialized to: b start Positive real numbers greater than 0; Hyperbolic Radon transform imaginary semi-axis length search termination value b end Initialized to: b start end Positive real numbers less than +∞;​ Number of searches for the imaginary semi-axis length of the hyperbolic Radon transform, N b Initialize to: 100≤N b Positive integers less than 250; The hyperbolic Radon transform imaginary semi-axis length search set b is initialized as follows: Where, n b n is the long index of the virtual semi-axis. b =0,1,…,N b -1, For the nth b There are search values ​​for the length of the imaginary half-axis, and there are The time interval Δt of the time-frequency distribution is initialized as: Δt = step / f s ; The time-frequency distribution frequency interval Δf is initialized as follows: Δf = f s / N w ; Strip hyperbolic Radon transform strip width search starting value d start Initialized to: d start =0; Number of strip widths searched for N in the hyperbolic Radon transform of strips d Initialized to: N d =N a ; The strip hyperbolic Radon transform strip width search set D is initialized as follows: Where, n d n is the stripe width index. d =0,1,…,N d -1, For the nth d Each search bar width value, and has Optimal strip width decision threshold TH d Initialized to: 0 <TH d Positive real numbers less than 0.3; (4-2) Calculate the hyperbolic Radon transform result HR of the i-th signal component. i (n a ,n b ): (4-3) Calculate the index of the optimal real semi-axis length of the i-th signal component. and the optimal virtual half-axis long index in, The hyperbolic Radon transform result HR of the i-th signal component is represented by... i (n a ,n b The index n of the real semi-axis length when the maximum value is obtained. a And the long index of the virtual half axis n b The set that constitutes; (4-4) Calculate the maximum value of the hyperbolic Radon transform result of the i-th signal component. (4-5) Hyperbolic Radon transform result for the i-th signal component HR i (n a ,n b Normalization is performed to obtain the normalized hyperbolic Radon transform result of the i-th signal component. (4-6) Calculate the optimal strip hyperbolic Radon transform result for the i-th signal component. Where max(·,·) represents taking the maximum value of the two, min(·,·) represents taking the minimum value of the two, and η is the index of the newly added strip width; (4-7) Calculate the difference function Diff for the i-th signal component. i EHR (m d ): Where, m d Index for the difference stripe width; (4-8) Filter all cases that satisfy the difference function Diff. i EHR (m d The decision threshold TH is less than the optimal strip width. d The set of indexes for the difference strip width (4-9) Calculate the optimal strip width index for the i-th signal component. Where Ф is the empty set, and == indicates whether the left and right sets contain exactly the same elements. Represents the set of indices for the difference in strip width. The first element.

5. The method for estimating parameters of a ship's hyperbolic frequency modulated signal based on direction pre-decision as described in claim 4, characterized in that, In step (5), the estimated time parameter value of the i-th signal component is calculated using the following method, specifically including the following steps: (5-1) Initialize the parameters for calculating the time parameter estimate of the i-th signal component, specifically including the initialization of the following parameters: The time frame signal has a decision threshold coefficient. Initialize to: Positive real numbers; Maximum frequency interval threshold Initialize to: Positive integers; Maximum number of frames when there is no signal frequency Initialize to: Positive integers; Total number of frames N when there is no signal frequency non Initialized to: N non =0; The total number of frames N for signal components total Initialized to: N total =0; Current decision frame index q i Initialized to: q i =0; Frame index set when signal component frequency points exist Initialize to: (5-2) Calculate the set of frequency indexes on the hyperbolic stripe and hyperbolic strip sub-frequency index set The nth t element and The nth t element They are: in, and Respectively The search bar width value, the first The search for the real semi-axis length value and the first search... Each search term represents the length of the virtual half-axis. (5-3) Calculate the set of upper limits of hyperbolic strip frequencies F for the i-th signal component. i high and the lower limit set of hyperbolic strip frequencies F i low F i high The nth t Elements F i high (n t ) and F i low The nth t Elements F i low (n t They are respectively: Among them, F i high and F i low All are of length N t A set; (5-4) Calculate the pre-separated time-frequency distribution of the i-th signal component. (5-5) Calculate the maximum modulus V of the pre-separated time-frequency distribution. i max : Where max{·} is the function for finding the maximum value; (5-6) A decision threshold exists in the frame signal during calculation. (5-7) Calculate the separation frequency index in, The frame index q represents the i-th signal component at the current decision time. i Pre-separation time-frequency distribution The index of the time-frequency distribution point corresponding to the maximum value; (5-8) Determine the time-frequency distribution of pre-separation At the current judgment time frame index q i Does a signal frequency point exist? That is, determine whether the following conditions are met: If true, proceed to step (5-9); otherwise, set the total number of frames N when there is no signal frequency. non =N non +1, and proceed to step (5-12); (5-9) Determine the frame index set when the frequency points of the signal components exist. To determine if a set is empty, we need to check if the following conditions are true: If true, proceed to step (5-11); otherwise, proceed to step (5-10). (5-10) Determine whether the following conditions are true: If true, proceed to step (5-11); otherwise, set the total number of frames N when there is no signal frequency. non =N non +1, and proceed to step (5-12); (5-11) Let N be the total number of time frames for the signal components. total =N total +1, the total number of frames N when there is no signal frequency. non =0, and update the set of frame indices when the frequency point of the i-th signal component exists. (5-12) Determine whether the i-th signal component has reached the termination frame, i.e., determine whether the following conditions are met: and If true, proceed to step (5-14); otherwise, proceed to step (5-13). (5-13) Update the current judgment time frame index q i =q i +1, determine if the following conditions are true: q i ≤N t -1 If true, return to step (5-8); otherwise, proceed to step (5-14). (5-14) Calculate the starting time frame index of the i-th signal component. Termination Frame Index in, This represents the set of frame indices when the frequency points of the signal components exist. The first element, This represents the set of frame indices when the frequency points of the signal components exist. The last element.

6. The method for estimating parameters of a ship's hyperbolic frequency modulated signal based on direction pre-decision as described in claim 5, characterized in that, In step (6), the frequency parameter estimate of the i-th signal component is calculated using the following method, and the residual time-frequency distribution is updated. Specifically, the steps include: (6-1) Calculate the set of frame indices when the signal components of the i-th signal component are continuously present. (6-2) Calculate the total number of consecutive time frames for the i-th signal component. (6-3) Calculate the separated time-frequency distribution of the i-th signal component. (6-4) Based on the separation time-frequency distribution of the i-th signal component Start-time frame index and the frame index at termination Extract the instantaneous frequency point index set f of the i-th signal component i f i The element for: Among them, each signal component of the i-th signal component has a time frame index. Place, This indicates that the time-frequency distribution is separated. The index of the time-frequency distribution point corresponding to the maximum value; (6-5) The set of instantaneous frequency indexes f for the i-th signal component i By performing least-squares linear fitting on the reciprocal of the vector, the frequency estimation parameter vector θ of the i-th signal component is obtained. i : θ i =(X T X) -1 X T G i Where X is the least squares linear fitting auxiliary matrix. G i Let be the reciprocal vector of the instantaneous frequency point index set of the i-th signal component. The superscript T indicates matrix transpose; (6-6) Obtain the estimated period slope value of the i-th signal component. and initial frequency estimate Where, θ i (1) and θ i (2) represents the frequency estimation parameter vector θ of the i-th signal component. i The first and second elements; (6-7) Update the residual time-frequency distribution 7. The method for estimating parameters of a ship's hyperbolic frequency modulated signal based on direction pre-decision as described in claim 6, characterized in that, In step (7), the following method is used to determine whether there are still unseparated signal components, specifically including the following steps: (7-1) Decision threshold TH for determining whether there are still unseparated signal components S Initialized as: 80≤TH S Positive real numbers ≤160; (7-2) Calculate the residual time-frequency distribution 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 discrete index of the signal component be i = i + 1, and return to step (3); otherwise, proceed to step (8).

8. The method for estimating parameters of a ship's hyperbolic frequency modulated signal based on direction pre-decision as described in claim 7, characterized in that, In step (8), the estimated number of signal components and the estimated time and frequency parameters are calculated and output using the following method, specifically including the following steps: (8-1) Calculate the estimated number of signal components And output: (8-2) Output the time parameter estimation results: (8-3) Output frequency parameter estimation results: Among them, the discrete index of signal components

Images