An anti-interference sound signal data extraction method based on energy data perception

Abstract

The application provides an anti-interference acoustic signal data extraction method based on energy data sensing, and belongs to the technical field of underwater acoustic signal processing. The method comprises the following steps: A / D sampling and digital quadrature demodulation are performed on a received acoustic signal to obtain a baseband signal; low-pass filtering and down-sampling are performed on the baseband signal to obtain a pretreated signal; the pretreated signal is matched filtered by using a reference signal; a key improvement lies in that a local energy estimation function of the pretreated signal is calculated, and the matched filtering result is normalized with the energy estimation function to obtain a normalized detection signal. Through energy sensing normalization, the application effectively suppresses the shielding effect of strong interference signals on weak target signals due to non-quadrature reference signals and the "near-end strong and far-end weak" scene, significantly reduces the false recognition probability, and improves the accuracy and reliability of extracting a long-distance and low signal-to-noise ratio target beacon signal in a complex underwater acoustic environment. The application is suitable for underwater navigation, positioning and communication systems.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of underwater acoustic signal data extraction, and in particular to an anti-interference acoustic signal data extraction method based on energy data perception. BACKGROUND

[0002] Current underwater beacon signal extraction mainly relies on three technical routes of signal processing algorithm optimization 、 Encoding modulation joint design and Intelligent noise reduction method. However, affected by the complex environment such as multipath effect, dynamic noise and channel time-varying characteristics, the existing technology still faces significant bottlenecks in practical application. The current mainstream system takes signal processing algorithm as the core architecture, for example, a real-time noise suppression scheme based on adaptive filtering technology can effectively deal with steady-state interference, but it is sensitive to burst pulse interference, and the time-frequency analysis algorithm such as wavelet transform has the problem of incompatible calculation complexity and real-time performance. In terms of anti-interception, the encoding modulation technology can reduce the bit error rate caused by multipath effect through LDPC-GMSK and other redundancy encoding mechanisms, but the inherent limitations of the preset static channel model make it difficult to adapt to dynamic propagation environments such as the thermocline. In recent years, the intelligent noise reduction model has shown advantages in feature extraction in low signal-to-noise ratio scenarios, but its model training relies on large-scale labeled data sets, and the deployment cost caused by its large number of parameters limits its application in embedded devices.

[0003] Because the reference signals between different beacons are not completely orthogonal, even after matched filtering processing, the beacon signal close to the receiving device still occupies a large amplitude in the matching result, and thus there is a possibility of covering the true detection result of the target beacon signal. This situation leads to misjudgment of the remote beacon signal. In view of the above problems, the present application provides an anti-interference acoustic signal data extraction method based on energy data perception to solve the above problems. SUMMARY

[0004] Therefore, the purpose of the present application is to provide an anti-interference acoustic signal data extraction method based on energy data perception to at least solve the above problems.

[0005] The technical scheme adopted by the present application is as follows:

[0006] An anti-interference acoustic signal data extraction method based on energy data perception, the method comprising the following steps:

[0007] S1, A / D sampling is performed on the received acoustic signal of each period to obtain a sampling sequence signal ; the sampling sequence signal carries out digital quadrature baseband demodulation to obtain a baseband signal ;

[0008] S2, the baseband signal is expressed as a frequency domain form, wherein the frequency domain form is based on a demodulation carrier frequency , a sampling period , a phase correction factor , and an index of a received signal sampling point

[0009] S3, the baseband signal is subjected to low-pass filtering and down-sampling processing to obtain a pre-processed signal ;

[0010] S4, the pre-processed signal is subjected to matched filtering processing using a reference signal to obtain a filtered signal ; ;

[0011] S5, based on the pre-processed signal , an energy estimation function is calculated ;

[0012] S6, according to the filtered signal and the energy estimation function , normalization processing is performed to obtain a normalized detection signal , which is used to identify a target signal.

[0013] Further, in step S1, the sampling rate of the A / D sampling is set to 80 kS / s; the digital quadrature baseband demodulation is implemented through the following formula:

[0014]

[0015] wherein the carrier frequency is 7.5 kHz, the phase correction factor is used to compensate for the initial phase offset of the carrier; and each cycle of the received acoustic signal is 10 seconds.

[0016] Further, in step S2, the baseband signal is expressed as a frequency domain form, specifically, it is expressed as a frequency domain spectrum form through a mathematically equivalent transformation:

[0017]

[0018] The frequency domain expression is used to guide the parameter design of the low-pass filter, so as to realize the translation of the original signal spectrum around ± to the vicinity of direct current, so as to realize sidelobe suppression and bandwidth limitation by using a low-order low-pass filter.

[0019] Further, in step S3, the low-pass filter is implemented by a finite impulse response filter or an infinite impulse response filter, and the passband frequency range is 0-5 kHz, which is used to reserve the baseband effective information and suppress the high-frequency noise component; the down-sampling processing is integer times down-sampling, and the down-sampling factor is 4, which is performed after the low-pass filtering to obtain the down-sampled signal As the preprocessed signal, it is expressed as:

[0020]

[0021] wherein, is the impulse response of the low-pass filter.

[0022] Further, in step S4, the reference signal is generated by using the same signal processing procedure as S1 and S3 on the known transmitted signal, which is the baseband template of the transmitted signal; the matched filtering processing is implemented by calculating the cross-correlation between the preprocessed signal and the reference signal , and the specific formula is:

[0023]

[0024] wherein, is the length of the matched filter, which is equal to the length of the reference signal , is the index of the matched filter.

[0025] Further, the energy estimation function is obtained by calculating the local energy of the preprocessed signal in a sliding window with a length of ; and the specific calculation formula is:

[0026]

[0027] An energy operator is introduced, and the energy estimation function is obtained:

[0028]

[0029] wherein, the length of the sliding window is consistent with the length of the matched filter in step S4.

[0030] Further, in step S6, the normalization processing is implemented by dividing the filtered signal by the energy estimation function To achieve, get normalized detection signal:

[0031]

[0032] The normalized detection signal For suppressing the amplitude masking effect caused by adjacent strong interference beacon signals.

[0033] Further, step S6 also includes a target signal identification step: based on the normalized detection signal An adaptive detection threshold is set; when The peak value of the peak value exceeds the adaptive detection threshold, it is determined as the effective peak value of the target signal; the adaptive detection threshold is dynamically set based on the noise statistical characteristics of the normalized detection signal in the target signal-free section, specifically the noise mean plus K times the standard deviation, wherein K is a preset constant.

[0034] Further, the target signal identification step further includes: if the peak value exceeding the adaptive detection threshold is identified in the normalized detection signal The largest peak value is selected as the main target signal peak value; and the consistency check is performed on other peak values, the consistency check includes judging whether the peak value interval is greater than the minimum time interval threshold or the cross-cycle peak position is stable.

[0035] Compared with the prior art, the beneficial effects of the present application are:

[0036] The present application introduces a normalization link based on local signal energy, in the presence of multiple beacons, especially in the typical scenario of using non-orthogonal or similar reference signals, working at the same time and having different distances from the receiver, the absolute amplitude of the traditional matched filter output will be significantly lifted due to the strong signal of the near-end beacon, resulting in the far-end weak signal peak being submerged. The present application converts the matched filter output Divided by the local energy estimate From "absolute correlation amplitude" to "relative signal-to-noise ratio strength", which makes the far-end weak target peak value stand out in the presence of near-end strong interference background, fundamentally solving the technical problem of "near-end masking far-end". BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only preferred embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0038] Figure 1 is a flowchart of the embodiments of the present application.

[0039] Figure 2 is the original sequence diagram of an embodiment of the present application.

[0040] Figure 3 is the matched filter output result diagram of an embodiment of the present application

[0041] Figure 4 is the normalized final signal result diagram of an embodiment of the present application. DETAILED DESCRIPTION

[0042] The principles and features of the present application are described below in conjunction with the accompanying drawings, in which the enumerated embodiments are only used to explain the present application and are not intended to limit the scope of the present application.

[0043] With reference to Figure 1 , the present application provides an anti-interference acoustic signal data extraction method based on energy data sensing, which comprises the following steps:

[0044] S1, A / D sampling is performed on the received acoustic signal of each cycle to obtain a sampling sequence signal ; the sampling sequence signal is subjected to digital quadrature baseband demodulation to obtain a baseband signal ;

[0045] S2, the baseband signal is expressed in a frequency domain form, wherein the frequency domain form is based on a demodulation carrier frequency , a sampling period , a phase correction factor , and an index of a received signal sampling point

[0046] S3, low-pass filtering and down-sampling processing are performed on the baseband signal to obtain a preprocessed signal ;

[0047] S4, a reference signal is used to perform matched filtering processing on the preprocessed signal to obtain a filtered signal ;

[0048] S5, based on the preprocessed signal , an energy estimation function is calculated;

[0049] S6, according to the filtered signal and the energy estimation function , normalization processing is performed to obtain a normalized detection signal , which is used to identify a target signal.

[0050] In step S1, the sampling rate of the A / D sampling is set to 80 kS / s; the digital quadrature baseband demodulation is realized by the following formula:

[0051]

[0052] wherein the carrier frequency is 7.5 kHz, the phase correction factor is used to compensate the initial phase offset of the carrier; each cycle of the received acoustic signal is 10 seconds.

[0053] In step S2, the baseband signal is expressed in the frequency domain, specifically expressed in the form of frequency domain spectrum by mathematical equivalent transformation:

[0054]

[0055] The frequency domain expression is used to guide the parameter design of the low-pass filter, so as to realize the translation of the original signal spectrum around ± to the vicinity of direct current, so as to realize sidelobe suppression and bandwidth limitation by using a low-order low-pass filter.

[0056] In step S3, the low-pass filtering is realized by using a finite impulse response filter or an infinite impulse response filter, the passband frequency range of which is 0 to 5 kHz, which is used to retain the effective information of the baseband and suppress the high-frequency noise components; the down-sampling processing is integer times down-sampling, and the down-sampling factor is 4, which is performed after the low-pass filtering, so that the down-sampled signal is used as the preprocessed signal, which is expressed as:

[0057]

[0058] wherein is the impulse response of the low-pass filter.

[0059] In step S4, the reference signal is generated by using the same signal processing procedure as S1 and S3 on the known transmitted signal, which is the baseband template of the transmitted signal; the matched filtering processing is realized by calculating the cross-correlation between the preprocessed signal and the reference signal , and the specific formula is:

[0060]

[0061] wherein, is the length of the matched filter, which is equal to the length of the reference signal , is the index of the matched filter.

[0062] The energy estimation function The local energy within a sliding window of length is obtained; the specific calculation formula is:

[0063]

[0064] An energy operator is introduced , and the energy estimation function is obtained:

[0065]

[0066] The length of the sliding window is consistent with the length of the matched filter in step S4.

[0067] In step S6, the normalization processing is realized by dividing the filtered signal by the energy estimation function , and the normalized detection signal is obtained:

[0068]

[0069] The normalized detection signal is used to suppress the amplitude masking effect caused by adjacent strong interference beacon signals.

[0070] After step S6, a target signal identification step is further included: an adaptive detection threshold is set based on the normalized detection signal ; when the peak value of the normalized detection signal exceeds the adaptive detection threshold, it is determined as a valid peak value of the target signal; the adaptive detection threshold is dynamically set based on the noise statistical characteristics of the normalized detection signal in the target signal-free section, specifically, the noise mean value plus K times the standard deviation, wherein K is a preset constant.

[0071] The target signal identification step further includes: if a peak value exceeding the adaptive detection threshold is identified in the normalized detection signal , the largest peak value is selected as the main target signal peak value; and consistency verification is performed on other peak values, the consistency verification including judging whether the peak value interval is greater than a minimum time interval threshold or whether the cross-period peak position is stable.

[0072] For example, the present embodiment is applied to an underwater beacon positioning test in the northern sea area of the South China Sea, aiming to verify the effectiveness of the present application in extracting a far-end weak beacon signal in the presence of strong near-end interference.

[0073] ​For the test environment and layout, in the sea environment: the test sea area is more than 3500 meters deep, the environmental noise is mainly low-frequency ship noise and marine environmental noise, and the channel has obvious multipath effect and time-varying characteristics; in the equipment and layout: three underwater acoustic beacons (beacon1, beacon2, beacon3) are laid on the seabed, and a hydrophone array is dragged by an investigation ship for mobile reception, all the beacons and the shipborne receiving system are time-synchronized by high-precision atomic clocks; in the scene setting: the three beacons use the same spread spectrum modulation system, but use non-orthogonal spread spectrum code sequences, among them, beacon1 and beacon2 are close to the receiving ship (about 2-5 kilometers), and beacon3 is far away from the receiving ship (about 10-12 kilometers), forming a typical "near-end strong, far-end weak" test scene.

[0074] For signal parameters and front-end settings, in the signal frame structure: each beacon transmits acoustic signals in a fixed time slot with a period of 10 seconds, in the receiving end parameters, the A / D sampling rate is 80 kS / s (thousand samples per second); the receiving carrier frequency is 7.5 kHz; the analog front-end passband effectively covers the working frequency band, and provides not less than 20 dB suppression to out-of-band noise (especially 0-1 kHz and above 40 kHz), and the equivalent input noise of the receiving system is lower than 1.7 μV.

[0075] In the signal processing flow:

[0076] Taking the extraction of the signal of the far-end beacon beacon3 as an example, the processing flow is as follows:

[0077] For step S1, signal acquisition and quadrature demodulation, the received analog acoustic signal is A / D converted to get a discrete sequence , then digital quadrature demodulation is performed, using the carrier frequency =7.5 kHz and the phase correction factor obtained by initial synchronization estimation, down-converts to baseband according to the formula to get the baseband signal , this step moves the signal spectrum to near zero frequency.

[0078] For step S2, low-pass filtering and downsampling, the baseband signal is filtered by an FIR low-pass filter with a passband of 0-5 kHz to retain valid information and suppress high-frequency noise, and after filtering, 4 times downsampling is performed to get the preprocessed signal with lower data rate and more conducive to subsequent processing, thereby reducing the data amount of subsequent operation.

[0079] ​For step S3, Matched Filtering, a baseband reference signal of beacon3 is generated in advance according to its known transmitted chips , whose length M is consistent with the code length, and the pre-processed signal is cross-correlated with , i.e. matched filtering, to obtain the correlation output . Matched filtering can compress the signal energy and provide processing gain.

[0080] For step S4, Local Energy Estimation, a sliding window with the same length M as the matched filter can be used to calculate the local energy of the pre-processed signal , and the energy estimation function is obtained, which reflects the local energy background of the signal at each time point, including the energy sum of the target signal, noise, and potential strong interference signals.

[0081] For step S5, Energy Normalization Detection, the matched filtering output can be divided by the local energy estimation to obtain the normalized detection signal: .

[0082] For step S6, Target Identification and Decision, a peak search can be performed on the normalized detection signal . An adaptive detection threshold is set, which is based on the noise mean and standard deviation in the time period when no target signal is confirmed (e.g. threshold = noise mean + 3 times standard deviation), and when the peak of exceeds this threshold, it is determined that a target signal is detected. If there are multiple peaks exceeding the threshold, the highest peak is selected as the arrival time of beacon3, and a time interval consistency check (such as checking whether the peak interval conforms to the transmission timing of the beacon) is used to exclude false peaks.

[0083] For effect comparison, Figure 2 shows a baseband signal in a receiving period, where the peaks of the near-end beacons beacon1 and beacon2 are significant, and the peak of the far-end beacon beacon3 is weak, with background noise mixed in the environment.

[0084] Figure 3 , 4 shows a baseband signal in a receiving period or , and it can be observed that the strong impulse responses of beacon1 and beacon2 are much higher in amplitude than the weak response of beacon3, and the signal of beacon3 is almost submerged in the background noise and near-end interference.

[0085] Figure 3The output of the conventional method, i.e. only to the bad walk S3 matched filter It can be seen that the correlation peaks of beacon1 and beacon2 are very sharp and have very high amplitude, while the peak of beacon3 is almost invisible and is very easy to be mistaken as noise or directly shielded by the strong peaks.

[0086] Figure 4 The normalized detection signal obtained by applying the method of the present application is shown After energy normalization processing, the strong peaks of beacon1 and beacon2 are significantly suppressed (relative amplitude is reduced), while the peak value of beacon3 is clearly highlighted, which can be accurately identified and extracted.

[0087] Therefore, the present embodiment fully proves the effectiveness of the method of the present application. In the harsh environment of strong near-end interference, the present application successfully solves the problem of shielding of the weak target signal at a long distance by using the energy-aware normalization technology, and significantly improves the accuracy and reliability of underwater acoustic signal detection.

[0088] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for extracting anti-interference acoustic signal data based on energy data sensing, characterized in that, The method includes the following steps: S1. Perform A / D sampling on the received acoustic signal for each cycle to obtain the sampled sequence signal. ; for the sampled sequence signal Digital quadrature baseband demodulation is performed to obtain the baseband signal. ; S2, the baseband signal Represented in frequency domain form, wherein the frequency domain form is based on the demodulation carrier frequency. Sampling period Phase correction factor Index of the received signal sampling point ; S3, regarding the baseband signal Low-pass filtering and downsampling are performed to obtain the preprocessed signal. ; S4. Use reference signal For the preprocessed signal Perform matched filtering to obtain the filtered signal. The reference signal The baseband template for the transmitted signal is generated by applying the same signal processing procedure as S1 and S3 to the known transmitted signal; the matched filtering process calculates the preprocessed signal. With the reference signal The cross-correlation is achieved, and the specific formula is as follows: in, The length of the matched filter is equal to the reference signal. Length, For the index of the matched filter; S5. Based on the preprocessed signal Calculate the energy estimation function The energy estimation function By calculating the preprocessed signal In length The local energy within the sliding window is obtained; the specific calculation formula is: Introducing energy operators The energy estimation function is obtained. : Wherein, the length of the sliding window The length of the matched filter described in step S4 should be consistent with the length of the matched filter. S6. Based on the filtered signal and the energy estimation function Normalization is performed to obtain the normalized detection signal. It is used to identify target signals.

2. The method for extracting anti-interference acoustic signal data based on energy data sensing according to claim 1, characterized in that, In step S1, the sampling rate of the A / D sampling is set to 80 kS / s; the digital quadrature baseband demodulation is achieved through the following formula: Among them, carrier frequency 7.5 kHz, phase correction factor Used to compensate for the initial phase offset of the carrier; each period of the received acoustic signal is 10 seconds.

3. The method for extracting anti-interference acoustic signal data based on energy data perception according to claim 1 or 2, characterized in that, In step S2, the baseband signal Represented in frequency domain form, specifically through mathematical equivalent transformations to express it in frequency domain spectral form: Frequency domain representation is used to guide the parameter design of low-pass filters, so as to achieve the refraction of the original signal spectrum around ± The filter is shifted to the vicinity of DC, thereby using a low-order low-pass filter to achieve sidelobe suppression and bandwidth limitation.

4. The method for extracting anti-interference acoustic signal data based on energy data perception according to claim 1, characterized in that, In step S3, the low-pass filtering is implemented using a finite impulse response (FIR) filter or an infinite impulse response (IR) filter, with a passband frequency range of 0 to 5 kHz, used to retain effective baseband information and suppress high-frequency noise components; the downsampling process is an integer multiple downsampling with a downsampling factor of 4, performed after the low-pass filtering, to obtain the downsampled signal. As a preprocessed signal, it is represented as: in, The impulse response of the low-pass filter is given.

5. The method for extracting anti-interference acoustic signal data based on energy data sensing according to claim 1, characterized in that, In step S6, the normalization process is performed by changing the filtered signal. Divided by the energy estimation function To achieve this, we obtain the normalized detection signal: The normalized detection signal Used to suppress amplitude masking effects caused by nearby strong interfering beacon signals.

6. A method for extracting anti-interference acoustic signal data based on energy data sensing according to claim 1 or 5, characterized in that, Step S6 is followed by a target signal recognition step: based on the normalized detection signal Set an adaptive detection threshold; when When the peak value exceeds the adaptive detection threshold, it is determined to be a valid peak value of the target signal; the adaptive detection threshold is dynamically set based on the noise statistical characteristics of the normalized detection signal in the segment without target signal, specifically the noise mean plus K times the standard deviation, where K is a preset constant.

7. The method for extracting anti-interference acoustic signal data based on energy data perception according to claim 6, characterized in that, The target signal identification step further includes: if the normalized detection signal If a peak value exceeding the adaptive detection threshold is identified, the largest peak value is selected as the main target signal peak value; and other peak values ​​are subjected to consistency verification, which includes determining whether the peak interval is greater than the minimum time interval threshold or whether the peak position across cycles is stable.

Images