An improved method for estimating propeller speed

Abstract

The present invention provides an improved method for estimating the rotational speed of a propeller. The estimation method includes the following steps: Step 1: DEMON demodulation spectrum analysis: Perform DEMON analysis on the received ship radiated noise signal x(n) to obtain the DEMON demodulation spectrum analysis result X(f); Step 2: Extract DEMON modulation line spectra; Step 3: Determine alternative fundamental frequencies; Step 4: Extract line spectrum harmonic sequences; Step 5: Extract energy characteristics of the alternative fundamental frequency line spectrum harmonic sequences and estimate the rotational speed of the propeller.

Description

Technical Field

[0001] The present invention relates to the technical field of underwater acoustic signal processing, and in particular to an improved method for estimating the rotational speed of a propeller. Background Art

[0002] Passive sonar target classification and recognition, as the main function of a sonar system, is a key technology field developed by various countries and has important application value. Traditional passive target classification and recognition is mainly based on the physical properties of the target and its radiated noise characteristics. By analyzing the target radiated noise and extracting features such as line spectra, continuous spectra, and demodulation spectra (DEMON spectra) that characterize the physical properties of the target, the quality of feature extraction directly affects the accuracy of the recognition result.

[0003] The demodulation spectrum feature is one of the most commonly used features in passive target classification and recognition and is also one of the most important types of features. Through demodulation spectrum analysis, information on target propeller parameters can be obtained, mainly including propeller rotational speed, blade rate, number of blades, number of shafts, etc. These information are important parameters for characterizing the physical properties of the target and are also one of the most effective information for target classification and recognition.

[0004] When the structure of the target radiated noise demodulation spectrum is clear and complete, information such as its rotational speed (fundamental frequency) and number of blades can be effectively extracted. However, during the experiment, the structure of the target noise demodulation spectrum is often unclear and the harmonic structure is incomplete, such as the absence of the fundamental frequency, the absence of the blade rate, only the fundamental frequency and some harmonics, etc. This brings great difficulties to the estimation of propeller parameter information. It is difficult to accurately estimate the target rotational speed, and it is even more difficult to estimate the number of blades and the number of shafts. Incorrect estimation results of propeller parameter information may directly lead to incorrect recognition results, and the accuracy of propeller rotational speed estimation is relatively low. Summary of the Invention

[0005] The problem to be solved by the present invention is to provide an improved method for estimating the rotational speed of a propeller in view of the above deficiencies in the prior art, which solves the problem of relatively low accuracy of propeller rotational speed estimation in the prior art.

[0006] The above object of the present invention is achieved by the following technical solutions: An improved method for estimating the rotational speed of a propeller, the estimation method comprising the following steps:

[0007] Step 1: DEMON demodulation spectrum analysis: Perform DEMON analysis on the received ship radiated noise signal x(n) to obtain the DEMON demodulation spectrum analysis result X(f);

[0008] Step 2: Extract DEMON modulation line spectra;

[0009] Step 3: Determine alternative fundamental frequencies;

[0010] Step 4: Line spectrum harmonic sequence extraction;

[0011] Step 5: Energy feature extraction of alternative fundamental frequency line spectrum harmonic sequences and propeller speed estimation.

[0012] The present invention is further configured such that: the said Step 1 further includes:

[0013] 1.1 Band-pass filtering: According to the modulation strength change of the target in different frequency bands, a band-pass filter is selected to perform band-pass filtering on the received ship radiated noise signal x(n) to obtain a band-pass filtered signal xf(n), n = 1, 2,..., N, that is

[0014]

[0015] In the formula, represents the convolution operation; h BP (k1) is the band-pass filter coefficient, k1 = 1, 2,..., N BP ; N BP is the order of the band-pass filter;

[0016] 1.2 Detection: Square-law detection is used for the band-pass filtered signal to obtain a detected signal y1(n) = |xf(n)| 2 , n = 1, 2,..., N;

[0017] 1.3 Low-pass filtering: The detected signal y1(n), n = 1, 2,..., N is subjected to low-pass filtering to obtain a low-pass filtered signal

[0018]

[0019] Among them, represents the circular convolution operation; h LP (k2) is the low-pass filter coefficient, k = 1, 2,..., N LP ; N LP is the order of the low-pass filter;

[0020] 1.4 Decimation: The low-pass filtered signal y2(n), n = 1, 2,..., N is decimated to obtain a decimated signal

[0021] y3(i) = y2(1:D:N), i = 1, 2,..., N / D

[0022] In the formula, D is the decimation factor;

[0023] 1.5 Power spectrum analysis: The data is accumulated in K batches to NFFT points, NFFT = KN / D, denoted as y4(i), i = 1, 2,..., KN / D, and the accumulated data is subjected to power spectrum analysis,

[0024] Wherein, X(f) is the result of the DEMON demodulation spectrum analysis, and K is the cumulative batch number.

[0025] 9. The present invention is further configured that: the step 2 further includes:

[0026] 2.1 Analyze the result X(f) of the DEMON demodulation spectrum analysis. If the spectral line at the frequency point f j satisfies the following two constraint conditions, it is determined as a line spectrum:

[0027] ① The energy of the spectral line at f j is the maximum value in the neighborhood, that is, X(f j ) > X(f j -1) and X(f j ) > X(f j +1). At the same time

[0028] X(f j ) > X(f j -2) and X(f j ) > X(f j +2);

[0029] ② The energy of the spectral line at f j is β dB higher than the background of this spectral line, that is,

[0030]

[0031] wherein, M B is the background sliding window length; XC(f) is the calculation result of the demodulation spectrum background;

[0032] 2.2 Set to zero the spectral line positions that do not satisfy the above constraint conditions, and retain the spectral line positions that satisfy the conditions. After traversing all frequency points, the DEMON modulation line spectrum extraction result P L (f) is obtained, that is,

[0033]

[0034] The present invention is further configured that: the β takes 6-10 dB.

[0035] The present invention is further configured that: the step 3 further includes:

[0036] 3.1 Suppose a total of M L root line spectra are extracted and their corresponding frequency point positions are k i , i = 1, 2,..., M L , and calculate the frequency differences between every two line spectra respectively to obtain the difference frequency calculation result △f i , i = 1, 2,..., M, M is the number of difference frequencies, that is

[0037]

[0038] 3.2 Set the percentage of the set frequency error as δ, then [Δf i -δΔf i , Δf i +δΔf i The difference frequencies within the range are considered to be the same difference frequency, and count the number of occurrences of each difference frequency. Select the line spectra and difference frequencies within the preset frequency range as the alternative fundamental frequencies sf i , i = 1, 2,..., P, where P is the number of alternative fundamental frequencies.

[0039] The present invention is further configured such that: the preset frequency range is greater than 0.5 Hz and less than 40 Hz.

[0040] The present invention is further configured such that: step 4 further includes: for the alternative fundamental frequency sf i , i = 1, 2,..., P, analyze the relationship between the extracted line spectra and the alternative fundamental frequencies, and set a frequency error threshold Δ. When |f i - round(f i / sf i ) × sf i | ≤ Δ, it is considered that the line spectrum is the harmonic of the current alternative fundamental frequency. And so on, obtain the harmonics corresponding to all alternative fundamental frequencies. For the alternative fundamental frequency sf i , the set of its line spectrum harmonic sequences is represented as

[0041] ZX i = {f j ||f j - round(f j / sf i ) × sf i | ≤ Δ},

[0042] In the formula, f j ∈F, F is the set of extracted line spectra, i = 1, 2,..., N.

[0043] The present invention is further configured such that: step 5 further includes:

[0044] 5.1 Calculate the energy values of all line spectrum harmonic sequences respectively, where the set of energy values of the line spectrum harmonic sequences is represented as

[0045] E i = ∑ j {X(f j )||f j - round(f j / sf i ) × sf i | ≤ Δ},

[0046] In the formula, X(f j ) represents the amplitude corresponding to f in the DEMON demodulation spectrum, f j ∈F, where F is the set of extracted line spectra, and i = 1, 2,..., N; j ∈F, where F is the set of extracted line spectra, and i = 1, 2,..., N;

[0047] 5.2 Select the line spectrum sequence with the largest energy and the number of harmonics not less than the preset value, and the corresponding fundamental frequency is the propeller fundamental frequency f shaft , and the estimated result of the propeller speed is rn = f shaft × 60 r / min.

[0048] In summary, the beneficial technical effects of the present invention are as follows:

[0049] (1) Aiming at the problem of inaccurate propeller speed estimation in the case of missing fundamental frequency or incomplete DEMON modulation harmonic structure in practice, an improved propeller speed estimation method is given, which can effectively realize the speed estimation under the conditions of missing fundamental frequency or incomplete demodulation spectrum structure.

[0050] (2) This method uses the extracted DEMON modulation line spectra and line spectrum difference frequencies as alternative fundamental frequencies, and estimates the propeller speed through harmonic search and energy comparison. On the one hand, it can effectively overcome the problem of speed estimation in the case of missing modulation fundamental frequency, and on the other hand, it can also effectively solve the problem of propeller speed estimation in the case of clear fundamental frequency but incomplete harmonic structure.

[0051] (3) The simulation and actual data analysis results show that: the method proposed in the present invention effectively realizes the speed estimation under the condition of unclear or incomplete DEMON modulation line spectrum structure. Description of the Drawings

[0052] Figure 1 is the flow chart of the improved propeller speed estimation method in the present invention;

[0053] Figure 2 is the speed estimation result diagram under the condition of complete harmonic structure of the propeller in the simulation data of the present invention;

[0054] Figure 3 is the speed estimation result diagram under the condition of missing fundamental frequency of the propeller in the simulation data of the present invention;

[0055] Figure 4 is the speed estimation result diagram under the condition of incomplete harmonic structure of the propeller in the simulation data of the present invention;

[0056] Figure 5 is the speed estimation result diagram under the condition of missing fundamental frequency of the propeller in the actual data of the present invention. Detailed Embodiments

[0057] In order to make the technical means, creative features, achieved objectives and functions of the present invention clearer and easier to understand, the present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0058] (1) Implementation process

[0059] As Figure 1 shown, the present invention proposes an improved method for estimating the propeller speed. The estimation method includes the following steps:

[0060] Step 1: DEMON demodulation spectrum analysis: Perform DEMON analysis on the received ship radiated noise signal x(n) to obtain the DEMON demodulation spectrum analysis result X(f).

[0061] 1.1 Band-pass filtering: According to the modulation strength change of the target in different frequency bands, select a band-pass filter to perform band-pass filtering on the received ship radiated noise signal x(n) to obtain the band-pass filtered signal xf(n), n = 1, 2,..., N, that is

[0062]

[0063] where denotes the convolution operation; h BP (k1) is the band-pass filter coefficient, k1 = 1, 2,..., N BP ; N BP is the order of the band-pass filter;

[0064] 1.2 Detection: Use the square law to detect the band-pass filtered signal to obtain the detected signal y1(n) = |xf(n)| 2 , n = 1, 2,..., N;

[0065] 1.3 Low-pass filtering: Perform low-pass filtering on the detected signal y1(n), n = 1, 2,..., N to obtain the low-pass filtered signal

[0066]

[0067] where denotes the circular convolution operation; h LP (k2) is the low-pass filter coefficient, k = 1, 2,..., N LP ; N LP is the order of the low-pass filter;

[0068] 1.4 Downsampling: Perform downsampling on the low-pass filtered signal y2(n), n = 1, 2,..., N to obtain the downsampled signal

[0069] y3(i) = y2(1:D:N), i = 1, 2,..., N / D

[0070] Where D is the downsampling multiple;

[0071] 1.5 Power spectrum analysis: Accumulate K batches of data to NFFT points, where NFFT = KN / D, denoted as y4(i), i = 1, 2,..., KN / D, and perform power spectrum analysis on the accumulated data.

[0072] Where X(f) is the result of the DEMON demodulation spectrum analysis, and K is the number of accumulated batches.

[0073] In step 1, in sonar signal processing, the algorithm that demodulates the received broadband signal to calculate the low-frequency demodulation spectrum is called DEMON analysis. The demodulated low-frequency time-domain signal is called the envelope signal, and its power spectrum is called the DEMON demodulation spectrum. By analyzing and processing the DEMON demodulation spectrum, the DEMON demodulation spectrum analysis result X(f) can be obtained. Through DEMON analysis, ship physical characteristics such as propeller speed can be obtained, which is of great value for the automatic and intelligent signal processing of underwater equipment.

[0074] Step 2: Extract DEMON modulation line spectra.

[0075] 2.1 Analyze the result X(f) of the DEMON demodulation spectrum analysis. If the spectral line at the frequency point f j satisfies the following two constraint conditions, it is determined as a line spectrum:

[0076] ① The energy of the spectral line at f j is the maximum value in the neighborhood, that is, X(f j ) > X(f j -1) and X(f j ) > X(f j +1), and at the same time

[0077] X(f j ) > X(f j -2) and X(f j ) > X(f j +2);

[0078] ② The energy of the spectral line at f j is β dB higher than the background of this spectral line, that is,

[0079]

[0080] Where M B is the background sliding window length; XC(f) is the calculation result of the demodulation spectrum background; β takes 6 - 10 dB;

[0081] 2.2 Set the spectral line positions that do not meet the above constraint conditions to zero, and retain the spectral line positions that meet the conditions. After traversing all frequency points, the DEMON modulation line spectrum extraction result P is obtained. L (f), that is,

[0082]

[0083] In step 2, combined with the characteristic representation of the line spectrum, the DEMON modulation line spectrum is extracted through maximum value search and signal-to-noise ratio screening. The above-mentioned DEMON modulation line spectrum can reflect the rhythm information of the physical properties related to the ship, such as the propeller rotation speed and the number of propeller blades.

[0084] Step 3: Determine the candidate fundamental frequencies.

[0085] 3.1 Suppose a total of M L spectral lines are extracted and their corresponding frequency point positions are k i , i = 1, 2,..., M L , and calculate the frequency differences between every two spectral lines respectively to obtain the difference frequency calculation result △f i , i = 1, 2,..., M, M is the number of difference frequencies, that is

[0086]

[0087] 3.2 Set the frequency error percentage δ, then the difference frequencies within the range of [△f i - δ△f i , △f i + δ△f i are considered to be the same difference frequency, and count the number of occurrences of each difference frequency. Select the spectral lines and difference frequencies within the frequency range greater than 0.5 Hz and less than 40 Hz as the candidate fundamental frequencies sf i , i = 1, 2,..., P, P is the number of candidate fundamental frequencies.

[0088] In step 3, by comprehensively using the extracted DEMON modulation line spectrum, the DEMON modulation line spectrum and the line spectrum difference frequency are used as candidate fundamental frequencies to solve the problem of target shaft frequency estimation when the shaft frequency is missing or the harmonic structure of the DEMON modulation is incomplete.

[0089] Step 4: Extract the line spectrum harmonic sequence. For the candidate fundamental frequencies sf i , i = 1, 2,..., P, analyze the relationship between the extracted line spectrum and the candidate fundamental frequencies, and set a frequency error threshold △. When |f i - round(f i / sf i ) × sf iWhen |≤△, this line spectrum is considered as the harmonic of the current alternative fundamental frequency. By analogy, all the harmonics corresponding to the alternative fundamental frequencies are obtained. For the alternative fundamental frequency sf i , the set representation of its line spectrum harmonic sequence is

[0090] ZX i ={f j ||f j -round(f j / sf i )×sf i |≤△},

[0091] where f j ∈F, F is the set of extracted line spectra, and i = 1, 2,..., N.

[0092] In step 4, by setting the frequency error threshold to determine the harmonic line spectra of each alternative fundamental frequency, the tolerance of harmonic search can be guaranteed.

[0093] Step 5: Extract the energy characteristics of the alternative fundamental frequency line spectrum harmonic sequence and estimate the propeller speed.

[0094] 5.1 Calculate the energy values of all line spectrum harmonic sequences respectively. The set representation of the energy values of the line spectrum harmonic sequences is

[0095] E i =∑ j {X(f j )||f j -round(f j / sf i )×sf i |≤△},

[0096] where X(f j ) represents the amplitude corresponding to f j in the DEMON demodulation spectrum, f j ∈F, F is the set of extracted line spectra, and i = 1, 2,..., N;

[0097] 5.2 Select the line spectrum sequence with the largest energy and the number of harmonics not less than the preset value. Its corresponding fundamental frequency is the propeller fundamental frequency f shaft , and the estimated result of the propeller speed is rn = f shaft ×60 revolutions per minute.

[0098] In step 5, by analyzing the energy relationship and the number of harmonic line spectra of each alternative fundamental frequency line spectrum harmonic group, the propeller shaft frequency is determined, and the credibility of the propeller shaft frequency estimation is improved.

[0099] The improved propeller speed estimation method uses the extracted DEMON modulation line spectrum and line spectrum difference frequency as alternative fundamental frequencies, and estimates the propeller speed through harmonic search and energy comparison. On the one hand, it can effectively overcome the problem of speed estimation in the case of missing modulation fundamental frequency, and on the other hand, it can also effectively solve the problem of propeller speed estimation in the case of clear fundamental frequency but incomplete harmonic structure.

[0100] Aiming at the problem of inaccurate propeller speed estimation in practical situations such as missing fundamental frequency or incomplete DEMON modulation harmonic structure, an improved propeller speed estimation method is proposed, which can effectively realize the speed estimation under the conditions of missing fundamental frequency or incomplete demodulation spectrum structure.

[0101] (2) Simulation and sea trial data test results

[0102] Based on the “(1) Implementation process”, through the analysis and processing of simulation and actual sea trial data, the processing results of the present invention are given. The processing results of simulation data are as Figures 2 to 4 shown, and the processing results of the actual sea trial data of a certain merchant ship target are as Figure 5 shown.

[0103] Computer simulation: The simulation target has a fundamental frequency of 4.2 Hz, a speed of 252 revolutions per minute, and a 4-blade propeller. The simulation is divided into 3 cases. In the first case, the target propeller has a complete harmonic structure. In the second case, the target propeller is missing the fundamental frequency. In the third case, there is a target fundamental frequency but the harmonic structure is incomplete.

[0104] It can be seen from the simulation processing results that under the conditions of complete harmonic structure, missing fundamental frequency, and having a fundamental frequency but incomplete harmonic structure of the target propeller, the proposed improved propeller speed estimation method can accurately and effectively extract and estimate the target speed.

[0105] Sea trial data processing: The actual data is a certain merchant ship target recorded during the sea trial. Its fundamental frequency is 2.28 Hz, and the speed is 137 revolutions per minute. The target fundamental frequency is missing in the recorded data.

[0106] It can be seen from the sea trial data processing results that for the actual sea trial target data, in the case of a certain DEMON harmonic structure but missing fundamental frequency of the target, the proposed improved propeller speed estimation method can accurately estimate the target propeller speed.

[0107] The above simulation analysis and sea trial data processing results show that: By adopting the proposed improved propeller speed estimation method, it is possible to accurately and effectively realize the propeller speed estimation under the conditions of complete propeller harmonic structure, missing propeller fundamental frequency, and having a fundamental frequency but incomplete harmonic structure, thereby providing feature information support for target classification and recognition. This method has good application prospects.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the purpose and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. An improved method for estimating the rotational speed of a propeller, characterized in that: The estimation method includes the following steps: Step 1: DEMON demodulation spectrum analysis: Perform DEMON analysis on the received ship radiated noise signal x(n) to obtain the DEMON demodulation spectrum analysis result X(f); Step 2: DEMON modulation line spectrum extraction; Step 3: Determine the alternative fundamental frequency; Step 4: Line spectrum harmonic sequence extraction; Step 5: Extract the energy characteristics of the alternative fundamental frequency line spectrum harmonic sequence and estimate the propeller speed; The said Step 3 further includes: 3.1 Suppose a total of M L root line spectra are extracted, and their corresponding frequency point positions are k i1 , i1 = 1, 2,..., M L , and calculate the frequency differences between every two line spectra respectively to obtain the difference frequency calculation result Δf i2 , i2 = 1, 2,..., M, where M is the number of difference frequencies, that is Δf i2 = P L (f km ) - P L (f kn ), m = 1, 2, …, M L ; n = 1, 2, …, M L and m ≠ n; 3.2 Set the frequency error percentage δ, then [Δf i2 -δΔf i2 , Δf i2 +δΔf i2 The difference frequencies within the range are considered the same difference frequency, and count the number of occurrences of each difference frequency. Select the line spectrum and difference frequency within the preset frequency range as the alternative fundamental frequencies sf i3 , i3 = 1, 2, …, P, where P is the number of alternative fundamental frequencies; The said step 4 further includes: for the alternative fundamental frequency sf i3 , i3 = 1, 2, …, P, analyze the relationship between the extracted line spectra and the alternative fundamental frequency, and set a frequency error threshold Δ. When is satisfied, it is considered that this line spectrum is the harmonic of the current alternative fundamental frequency multiplied by an integer. And so on, the harmonics multiplied by an integer corresponding to all alternative fundamental frequencies are obtained. For the alternative fundamental frequency sf i3 , the set representation of its line spectrum harmonic sequence is In the formula, F is the set of extracted line spectra, where i1 = 1, 2, …, M L , and i3 = 1, 2, …, P; The said Step 5 further includes: 5.1 Calculate the energy values of all line spectrum harmonic sequences respectively, where the set of line spectrum harmonic sequence energy values is expressed as In the formula, represents the amplitude corresponding to in the DEMON demodulation spectrum, F is the set of extracted line spectra, and i3 = 1, 2, …, P; 5.2 Select the line spectrum sequence with the largest energy and the number of harmonics not less than the preset value, and its corresponding fundamental frequency is the propeller fundamental frequency f shaft , and the estimated result of the propeller speed is rn = f shaft × 60 r / min.

2. The improved method for estimating the rotational speed of a propeller according to claim 1, characterized in that: The said Step 1 further includes: 1.1 Band-pass filtering: According to the modulation strength change of the target in different frequency bands, select a band-pass filter to perform band-pass filtering on the received ship radiated noise signal x(n) to obtain the band-pass filtered signal xf(n), n = 1, 2, …, N, that is In the formula, denotes the convolution operation; h BP (k1) are the band-pass filter coefficients, where k1 = 1, 2,..., N BP ; N BP is the order of the band-pass filter; 1.2 Detection: Perform detection on the band-pass filtered signal using the square law to obtain the detected signal y1(n) = |xf(n)| 2 , n = 1, 2, ..., N; 1.3 Low-pass filtering: Perform low-pass filtering on the detected signal y1(n), n = 1, 2,..., N to obtain the low-pass filtered signal Among them, represents the circular convolution operation; h LP (k2) are the low-pass filter coefficients, k = 1, 2,..., N LP ; N LP is the order of the low-pass filter; 1.4 Downsampling: Perform downsampling on the low-pass filtered signal y2(n), n = 1, 2,..., N to obtain the downsampled signal y3(i) = y2(1:D:N), i = 1, 2,..., N / D In the formula, D is the downsampling factor; 1.5 Power spectrum analysis: Accumulate the data in K batches to NFFT points, where NFFT = KN / D, denoted as y4(i), i = 1, 2, ..., KN / D, and perform power spectrum analysis on the accumulated data. In the formula, X(f) is the DEMON demodulation spectrum analysis result, and K is the cumulative batch.

3. The improved method for estimating the rotational speed of a propeller according to claim 2, characterized in that: The said Step 2 further includes: 2.1 Analyze the DEMON demodulation spectrum analysis result X(f). If the spectral line at the frequency point f j satisfies the following two constraint conditions, it is determined as a line spectrum: ①f j The spectral line energy at this point is the maximum value within the neighborhood, that is, X(f j ) > X(f j -1) and X(f j ) > X(f j +1). At the same time, X(f j ) > X(f j -2) and X(f j ) > X(f j +2); ②f j The spectral line energy at this point is β dB higher than the background of this spectral line, that is, Where M B is the background sliding window length; XC(f) is the calculation result of the demodulation spectrum background; 2.2 Set the spectral line positions that do not meet the above constraints to zero, and retain the spectral line positions that meet the conditions. After traversing all frequency points, the DEMON modulated line spectrum extraction result P L (f), that is, 4. An improved propeller rotational speed estimation method according to claim 3, characterized in that: The said β takes 6 - 10 dB.

5. An improved propeller rotational speed estimation method according to claim 1, characterized in that: The said preset frequency range is greater than 0.5 Hz and less than 40 Hz.

Images