Underwater propeller blade number identification method based on shaft blade frequency double harmonic search

Abstract

The invention discloses an underwater propeller blade number identification method based on shaft blade frequency double harmonic search. The method comprises the following steps: (1) collecting an underwater propeller noise signal; (2) calculating the acquired noise signal by using a rapid spectrum-following correlation algorithm to obtain a cyclic spectrum correlation; (3) normalizing the obtained cyclic spectrum correlation to obtain cyclic spectrum coherence, and further constructing an enhanced envelope spectrum by integral average; (4) preprocessing the enhanced envelope spectrum, extracting a line spectrum, and then inputting conditions to carry out peak value search to obtain an alternative axis frequency; (5) inputting detection parameters to construct a search space; and for the combination of each alternative shaft frequency and the blade number, calculating a peak index and obtaining a combination probability table, and finally determining the blade number. The method can reliably extract the propeller blade number characteristics in a complex noise environment, improves the recognition precision, and has important practical application value for target recognition and classification.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of signal processing and target recognition, and particularly relates to a propeller blade number recognition method based on shaft-leaf frequency double harmonic search. BACKGROUND

[0002] Accurate recognition of underwater targets is crucial. As a key power component of underwater vehicles, ships and the like, the propeller blade number information is of great significance to target classification, performance evaluation and behavior prediction. Traditional propeller blade number recognition methods mainly rely on a large number of sample data covering various propeller blade numbers to achieve effective model training and recognition. At the same time, the accurate extraction of shaft frequency is the basis for such methods to accurately identify the propeller blade number.

[0003] A propeller blade number feature extraction method based on radiation noise modulation is disclosed in Chinese patent document CN111160207A, which includes: finding local peak values on a modulation graph and resonance frequencies at the local peak values; determining a shaft frequency and the number of resonance frequencies thereof; determining the values of other resonance frequencies according to the frequency multiplication relationship of the shaft frequency and the blade frequency; determining the average spectral coherence values at the harmonic frequencies; and obtaining the propeller blade number by using the inference method of naive Bayes. This method can analyze various civil ship radiation noise modulation spectra, and conveniently extract shaft frequency, blade frequency information and propeller blade number characteristic quantities.

[0004] However, in the actual marine environment, underwater acoustic signals face many challenges. On the one hand, the marine environment is extremely complex, and there are various background noises such as wind and wave, marine biological activities and noises generated by other ships sailing, etc. These noises seriously interfere with the propeller noise signal, resulting in a significant reduction in the signal-to-noise ratio. At the same time, underwater acoustic signals are affected by factors such as seawater medium characteristics, temperature gradient, salinity change and seabed topography during propagation, and will undergo serious propagation attenuation and distortion, making it difficult to effectively extract signal characteristics.

[0005] Under this complex background, shaft frequency detection is prone to large errors. As the basic frequency of propeller rotation, the detection accuracy of shaft frequency is directly related to the accuracy of propeller blade number recognition. However, noise interference and signal distortion can cause shaft frequency detection algorithms to misjudge, resulting in a significant decrease in the accuracy of propeller blade number recognition methods based on shaft frequency. Moreover, it is extremely difficult to collect actual noise data of propellers with different blade numbers under different working conditions. This not only requires a large amount of manpower, material resources and financial resources to build a large experimental monitoring system, but also is restricted by factors such as the variability of the marine environment and the instability of the target sailing state, making it difficult to obtain comprehensive, accurate and representative noise data, which seriously limits the practical application of traditional methods based on a large number of sample data.

[0006] In addition, the existing classical signal processing methods based on Fourier transform, wavelet transform and the like can analyze the propeller noise signal to a certain extent, but due to the fact that these methods do not deeply mine the complex and unique structural correlation between the shaft frequency and the blade frequency and the harmonics thereof, when facing the complex and changeable marine environment, the effective information in the signal cannot be fully utilized, the number of blades cannot be stably and reliably identified, and there are obvious technical bottlenecks and application limitations. SUMMARY

[0007] The application provides an underwater propeller blade number identification method based on shaft-blade frequency double harmonic search, which can reliably extract the propeller blade number feature in a complex noise environment, improve the identification accuracy, and has important practical application value for target identification and classification.

[0008] An underwater propeller blade number identification method based on shaft-blade frequency double harmonic search, comprising:

[0009] (1) collecting a propeller noise signal under water;

[0010] (2) calculating the collected noise signal by using a fast spectrum correlation algorithm to obtain a cyclic spectrum correlation;

[0011] (3) obtaining a cyclic spectrum coherence after normalizing the obtained cyclic spectrum correlation, and then further integrating and averaging to construct an enhanced envelope spectrum;

[0012] (4) extracting a line spectrum after preprocessing the enhanced envelope spectrum, and then inputting a condition to search for a peak value to obtain a candidate shaft frequency;

[0013] (5) inputting a detection parameter to construct a search space; for each combination of the candidate shaft frequency and the number of blades, calculating a peak value index and obtaining a combination probability table, and finally determining the number of blades.

[0014] The application fully utilizes the harmonic correlation between the shaft frequency and the blade frequency through double harmonic search, and can still stably identify the number of blades under noise interference, thereby providing a reliable technical means for propeller target identification.

[0015] In step (2), the formula for obtaining the cyclic spectrum correlation is:

[0016]

[0017] wherein, is the cyclic spectrum correlation, alpha is a cyclic frequency, f is a spectrum frequency, is a cyclic autocorrelation function of a time domain signal x(t), tau is a time delay, and j is an imaginary unit.

[0018] In step (3), the formula for obtaining the cyclic spectrum coherence is:

[0019]

[0020] where γ x (α,f) is the cyclic spectrum coherence, is the cyclic spectrum correlation at the cyclic frequency of 0.

[0021] In step (3), the formula for constructing the enhanced envelope spectrum is:

[0022]

[0023] where, is the enhanced envelope spectrum, F s is the sampling frequency of the signal x(t).

[0024] The specific process of step (4) is:

[0025] (4-1) Pretreatment and line spectrum extraction: The enhanced envelope spectrum is subjected to detrending processing to eliminate non-periodic interference, that is, the stable line spectrum is extracted by subtracting the first-order fitting curve from the enhanced envelope spectrum;

[0026] (4-2) Peak search and sorting: Set the shaft frequency detection interval, the maximum number of peaks, and the minimum peak interval, search for the peaks, and then arrange them in descending order of amplitude to form the candidate shaft frequency set {f1,f2,…,f n}.

[0027] The specific process of step (5) is:

[0028] (5-1) Constructing the search space: input the value interval of the number of blades r and the detection bandwidth Δf to form a multi-dimensional combination space of shaft frequency and the number of blades;

[0029] (5-2) Calculate the peak index: for each combination of the candidate shaft frequency f i and the number of blades r, calculate the peak index

[0030] (5-3) Generate the combination probability table: normalize the peak index to obtain the combination probability

[0031] (5-4) Determine the final number of blades: select the number of blades corresponding to the maximum probability in the combination probability table as the recognition result.

[0032] In step (5-2), the formula for calculating the peak index is:

[0033]

[0034] where, is the peak amplitude of the candidate shaft frequency f i , m hrThe h-order harmonic peak amplitude of the blade frequency in the detection bandwidth Δf when the number of blades r is h The upper limit of the calculated blade frequency harmonic.

[0035] In step (5-3), the combined probability of the number of blades r is obtained The formula is:

[0036]

[0037] Wherein, The alternative shaft frequency f i The combined probability of the number of blades r, n is the number of elements in the alternative shaft frequency set {f1, f2, …, f n} and r1 and r2 are the minimum and maximum values of the number of blades r, respectively.

[0038] Compared with the prior art, the present application has the following beneficial effects:

[0039] 1. Compared with the traditional blade number recognition method, the method of the present application does not require the shaft frequency to be known as a prerequisite, and by synchronously searching the shaft frequency and the number of blades characteristics, the demodulation spectrum line spectrum structure characteristics are excavated, and the dependence on a large amount of different number of blades data accumulation is eliminated; at the same time, the peak value index calculation and normalization processing form a combined probability table, which effectively avoids the influence of noise interference and peak frequency error, so that the number of blades recognition is more reliable in complex environment.

[0040] 2. The method of the present application can effectively extract the propeller shaft frequency, blade frequency and harmonic characteristics in complex scenes, and the identified number of blades is more consistent with the actual situation. The number of blades obtained by the method can provide technical support for propeller classification, target monitoring and other practical engineering applications. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 It is a flowchart of a kind of underwater propeller blade number identification method based on shaft blade frequency double harmonic search of the present application.

[0042] Figure 2 It is the envelope spectrum of two-blade propeller noise signal in open water.

[0043] Figure 3 It is the combined probability of two-blade propeller noise signal double harmonic search blade number in open water.

[0044] Figure 4 It is the envelope spectrum of three-blade propeller noise signal in open water.

[0045] Figure 5 It is the combined probability of three-blade propeller noise signal double harmonic search blade number in open water.

[0046] Figure 6To open water four-bladed propeller noise signal enhancement envelope spectrum.

[0047] Figure 7 To open water four-bladed propeller noise signal double harmonic search blade array combination probability. DETAILED DESCRIPTION

[0048] The application will be described in further detail below with reference to the drawings and embodiments, it should be pointed out that the following described embodiments are intended to facilitate the understanding of the application, and do not have any limiting effect on it.

[0049] As Figure 1 shown, a kind of underwater propeller blade number identification method based on shaft-blade frequency double harmonic search, comprising the following steps:

[0050] S01, using hydrophone to collect the noise signal of underwater propeller.

[0051] S02, the signal collected is introduced into program, and the cyclic spectrum correlation is calculated by fast spectrum correlation algorithm:

[0052]

[0053] Wherein, α is cyclic frequency, f is spectrum frequency, It is the cyclic autocorrelation function of time domain signal x (t), τ is time delay, j is imaginary unit.

[0054] S03, according to the cyclic spectrum correlation, the cyclic spectrum coherence and enhanced envelope spectrum are calculated:

[0055]

[0056] S04, the shaft frequency search is carried out to enhanced envelope spectrum: first, after the enhanced envelope spectrum is detrended, linear spectrum is extracted;Then, set the shaft frequency search parameters, for example, shaft frequency detection interval 10Hz-25Hz, maximum peak value 6, minimum interval 2 data points, search peak value and arrange in descending order, obtain the candidate shaft frequency set.

[0057] S05, blade number search is carried out based on the candidate shaft frequency obtained in S04: first, set the blade number search parameters, for example, blade number range 2-7 blades, bandwidth Δf=0.5Hz, construct combination space;Then, according to the formula The peak value index of each shaft frequency and blade array combination is calculated;Finally, the combination probability table is obtained by normalization The blade number corresponding to the maximum probability is selected.

[0058] To verify the effectiveness of the method, two-bladed, three-bladed and four-bladed propeller noise signals in open water are selected for analysis:

[0059] In open water, the envelope spectrum of two-bladed propeller with shaft frequency of about 21Hz is as followsFigure 2 As shown, the three circles from left to right represent the shaft frequency, blade frequency, and double blade frequency, which have the highest probability of dual harmonic search combinations. The results of blade number identification are shown below. Figure 3 As shown. According to Figure 2 The envelope spectrum, the open water environment causing a decrease in signal-to-noise ratio, and the Doppler effect generated by the thruster motion, together lead to interference signals in the low-frequency region of the envelope spectrum. However, through analysis of... Figure 3 A detailed analysis of the data reveals that among all combinations of shaft frequency and number of blades, the combination of a shaft frequency of 21.10 Hz and a number of blades of 2 has a probability of 60.66%, showing a clear advantage in statistical probability and consistent with the actual propulsion shaft frequency and number of blades.

[0060] In open water, the noise signal envelope spectrum of a three-bladed propeller with a shaft frequency of approximately 17 Hz is as follows: Figure 4 As shown, the blade count identification results are detailed in [link to documentation]. Figure 5 .Depend on Figure 4 It can be seen that, apart from the shaft frequency, the other harmonics do not exhibit typical blade number structure characteristics. However, due to... Figure 5 It can be seen that among all combinations of shaft frequency and blade number, the combination of 16.80Hz shaft frequency and 3 blades has a probability of 18.83%, which is statistically advantageous and consistent with the actual shaft frequency and blade number of propellers. The analysis results of this signal strongly support the effectiveness and reliability of this method in identifying the blade number of signals with unclear harmonic structure characteristics in open water scenarios.

[0061] Finally, the noise signal of the four-bladed propeller with a shaft frequency of approximately 23.3 Hz was analyzed, and its noise signal envelope spectrum is as follows: Figure 6 The blade count recognition results are recorded in Figure 7 .from Figure 6 It can be observed that the difference in the number of blades exacerbates the line spectrum noise interference around the shaft frequency, posing a challenge to the accurate identification of the number of blades. However, through analysis of... Figure 7 The study found that among all combinations of shaft frequency and number of blades, the combination of 23.31 Hz shaft frequency and 4 blades has a probability as high as 48.71%, showing a significant advantage in the statistical dimension and perfectly matching the actual shaft frequency and number of blades of the propeller.

[0062] The above analysis results fully demonstrate that, in open water environments, facing complex situations such as ambiguous shaft frequency characteristics and insignificant harmonic structures, the dual harmonic search method can stably and effectively complete the identification of different blade numbers, exhibiting good applicability and reliability. Compared with traditional methods, this invention does not require prior shaft frequency information, fully explores harmonic structure characteristics through dual harmonic search, and achieves higher identification accuracy and stronger adaptability in noisy environments.

[0063] The above embodiments describe the technical solutions and advantages of the present application in detail. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the present application. Any modification, supplement and equivalent replacement made within the principle range of the present application shall be included in the protection range of the present application.

Claims

1. A method for identifying the number of blades of an underwater propeller based on shaft-blade frequency double harmonic search, characterized in that, Comprise: (1) Collecting underwater propeller noise signals; (2) Calculating the collected noise signals by using fast spectral correlation algorithm to obtain cyclic spectrum correlation; (3) Obtaining cyclic spectrum coherence after normalizing the obtained cyclic spectrum correlation, and then further integrating and averaging to construct an enhanced envelope spectrum; (4) Extracting line spectrum after preprocessing the enhanced envelope spectrum, then inputting conditions to search for peak values, and obtaining candidate shaft frequencies; (5) Inputting detection parameters to construct a search space; For each combination of candidate shaft frequencies and blade numbers, calculate the peak index and obtain a combination probability table, and finally determine the blade number.

2. The method of claim 1, wherein the number of blades of the underwater propeller is identified based on a shaft-blade frequency double harmonic search. In step (2), the formula for obtaining the cyclic spectrum correlation is: wherein is the cyclic spectral correlation, a is the cyclic frequency, and f is the spectral frequency, is the cyclic autocorrelation function of the time-domain signal x(t), τ is the time delay, and j is the imaginary unit.

3. The method according to claim 2, wherein, In step (3), the formula for obtaining the cyclic spectrum coherence is: where γ x (α, f) is the cyclic spectral coherence, is the cyclic spectral correlation at cyclic frequency 0.

4. The method of claim 3, wherein the number of blades of the underwater propeller is identified based on the shaft frequency and the double harmonic frequency. In step (3), the formula for constructing the enhanced envelope spectrum is: in, To enhance the envelope spectrum, F s Let x(t) be the sampling frequency of the signal.

5. The shaft-blade-frequency and double-harmonic search based underwater thruster blade number identification method according to claim 1, wherein, The specific process of step (4) is: (4-1) Preprocessing and line spectrum extraction: Perform detrending on the enhanced envelope spectrum to eliminate non-periodic interference, that is, extract stable line spectrum by subtracting the first-order fitting curve from the enhanced envelope spectrum; (4-2) Peak search and sorting: Set the shaft frequency detection interval, the maximum number of peaks and the minimum peak interval, search for peaks and arrange them in descending order of amplitude to form the candidate shaft frequency set {f1, f2, …, fN}. n} 6. The shaft-blade-frequency and double-harmonic search based underwater thruster blade number identification method according to claim 1, wherein, The specific process of step (5) is: (5-1) Constructing a search space: Input the blade number r value interval and the detection bandwidth Δf to form a multi-dimensional combination space of shaft frequency and blade number; (5-2) Calculate the peak index: for each candidate shaft frequency f i in combination with the number of blades r, calculate the peak index (5-3) Generating a combination probability table: normalizing the peak indices to obtain a combination probability (5-4) Determining the final blade number: Select the blade number corresponding to the maximum probability in the combination probability table as the recognition result.

7. The shaft-blade-frequency and double-harmonic search based underwater thruster blade number identification method according to claim 6, characterized in that, In step (5-2), the peak index is calculated The formula is: wherein is the peak amplitude of the shaft frequency f i is the peak amplitude of the shaft frequency f hr is the hth harmonic peak amplitude of the blade frequency within the detection bandwidth Δf for a number of blades r, N h is the upper limit of the calculated blade frequency harmonics.

8. The method of claim 6, wherein the number of blades of the underwater propeller is identified based on a shaft-blade frequency double harmonic search. In step (5-3), the combined probability The formula is: wherein, f is an alternative shaft frequency i The combined probability of the number of blades r, n is the number of elements of the set of alternative shaft frequencies {f1, f2, …, f n} and r1 and r2 are the minimum and maximum values of the interval of values of the number of blades r, respectively.

Images