Aeromagnetic anomaly signal matching detection algorithm

Abstract

The present invention discloses an aeromagnetic anomaly signal matching detection algorithm, comprising the following steps: S1, in the process of magnetic anomaly detection, obtaining real-time signals and corresponding background field signals, performing similarity analysis and matching processing on the real-time signals and the background field signals, and obtaining a similarity function between the real-time signals and the background field signals; S2, performing denoising processing on the similarity function obtained in step S1; S3, inputting the processed data in step S2 into a preset orthogonal basis function detector, and comparing the output value of the orthogonal basis function detector with a preset threshold value, if the output value is greater than the preset threshold value, the magnetic anomaly signal is successfully matched, and it is determined that the target generating the magnetic anomaly signal exists. The aeromagnetic anomaly signal matching detection algorithm provided by the present invention has strong noise resistance and good weak magnetic detection capability; the detection rate of the algorithm under low signal-to-noise ratio is higher than the existing difference matching-orthogonal basis function detection, and has better detection performance.

Description

Technical Field

[0001] The invention relates to the field of magnetic anomaly signal detection, and in particular to an aeromagnetic anomaly signal matching detection algorithm. Background Art

[0002] Ferromagnetic objects are magnetized by the Earth's magnetic field to generate additional magnetic fields, which disturb the magnetic field in the surrounding space. The technology of detecting and locating targets by observing this disturbance is called magnetic anomaly detection. Aeromagnetic survey is one of the main forms of magnetic anomaly detection. It has a wide scanning area, is less affected by natural interference, and has strong maneuverability. It is widely used in anti-submarine reconnaissance. However, the detection platform flies at a high altitude, and the target magnetic anomaly is relatively weak, making it difficult to detect.

[0003] The time and frequency domain characteristics of magnetic anomaly signals are the main basis for detecting targets by existing methods. However, the characteristics of interference targets such as seabed magnetic rocks and shipwrecks are very similar to the target signals. In the absence of prior information, it is difficult to distinguish them, which can easily lead to false alarms. At present, the research on this technology is mostly focused on the compensation of carrier background magnetic interference and the construction of high-precision geomagnetic reference maps, while there is less research on matching detection algorithms. In existing studies, the difference between the detection signal and the background field signal is often used for detection, which is less effective under low signal-to-noise ratio conditions. Summary of the invention

[0004] In order to solve the problems raised in the above background technology, the object of the present invention is to provide an aeromagnetic anomaly signal matching detection algorithm.

[0005] To achieve the above object, the technical solution adopted by the present invention is: an aeromagnetic anomaly signal matching detection algorithm, comprising the following steps:

[0006] S1. During the magnetic anomaly detection process, a real-time signal and a corresponding background field signal are obtained, and similarity analysis and matching processing is performed on the real-time signal and the background field signal to obtain a similarity function between the real-time signal and the background field signal;

[0007] S2, performing denoising processing on the similarity function obtained in step S1;

[0008] S3. Input the data processed in step S2 into a preset orthogonal basis function detector, and compare the output value of the orthogonal basis function detector with a preset threshold. If the output value is greater than the preset threshold, the magnetic anomaly signal is successfully matched, and it is determined that the target that generates the magnetic anomaly signal exists.

[0009] In some embodiments, in step S1, the similarity function is obtained by specifically adopting the following steps:

[0010] First, the real-time signal collected by the sensor within a period of time is recorded as X current, as shown in formula (1), and the corresponding background field signal is extracted from the geomagnetic reference map according to the corresponding position information and recorded as X background , as shown in formula (2):

[0011]

[0012]

[0013] In equations (1) and (2), the value of M is determined by time and the sampling rate of the sensor;

[0014] Next, a sliding window with a width of s and a step size of 1 along the time axis is used to smooth the real-time signal in equation (1) and the background field signal in equation (2), and the two are converted into M-s+1 sub-signals respectively. Any sub-signal after the conversion of equation (1) is shown in equation (3), and any sub-signal after the conversion of equation (2) is shown in equation (4):

[0015]

[0016]

[0017] In formula (3) and formula (4), i represents the window index, i = 1, 2, 3, ..., M-s+1;

[0018] Finally, for each given window i of length s, the similarity function is established and obtained using the following formula (5):

[0019]

[0020] In some embodiments, in step S1, s=3.

[0021] In some embodiments, in step S2, a wavelet packet threshold denoising algorithm is used to perform denoising processing.

[0022] In some embodiments, in step S2, the denoising process using the wavelet packet threshold denoising algorithm specifically includes the following steps:

[0023] First, wavelet basis functions and decomposition levels are selected, and the similarity function obtained in step S1 is decomposed by wavelet packets;

[0024] Then, threshold quantization of wavelet packet decomposition coefficients is performed;

[0025] Finally, the similarity function is reconstructed based on the wavelet packet coefficients after threshold quantization.

[0026] Compared with the prior art, the present invention has the following beneficial effects:

[0027] The aeromagnetic anomaly signal matching detection algorithm provided by the present invention has strong noise resistance and good weak magnetic detection capability; the detection rate of the algorithm under low signal-to-noise ratio is higher than that of the existing difference matching (filtering)-orthogonal basis function detection, and has better detection performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A schematic diagram of the steps of the aeromagnetic anomaly signal matching detection algorithm provided by the present invention;

[0029] Figure 2 It is a schematic diagram of the relationship between the signal-to-noise ratio and the signal window width;

[0030] Figure 3 It is a schematic diagram of the relationship between time delay and signal window width;

[0031] Figure 4 is a schematic diagram of a target signal in a specific embodiment;

[0032] Figure 5 is a schematic diagram of a background signal in a specific embodiment;

[0033] Figure 6 is a schematic diagram of a difference matching result in a specific embodiment;

[0034] Figure 7 is a schematic diagram of a similarity analysis and matching processing result in a specific embodiment;

[0035] Figure 8 For Figure 7 Schematic diagram of the result of wavelet packet threshold denoising;

[0036] Fig. 9 is a schematic diagram of a difference matching-orthogonal basis function detection result in a specific embodiment;

[0037] Fig.10 It is a schematic diagram of a difference matching-orthogonal basis function detection result after filtering and denoising in a specific embodiment;

[0038] Fig.11 A schematic diagram of a detection result obtained by using the algorithm provided by the present invention in a specific embodiment;

[0039] Fig.12 The detection performance comparison chart of the three algorithms. DETAILED DESCRIPTION

[0040] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the following further describes how the present invention is implemented in conjunction with the accompanying drawings and specific implementation methods.

[0041] Reference Figure 1 As shown, the present invention provides an aeromagnetic anomaly signal matching detection algorithm, comprising the following steps:

[0042] S1. During the magnetic anomaly detection process, a real-time signal and a corresponding background field signal are obtained, and similarity analysis and matching processing is performed on the real-time signal and the background field signal to obtain a similarity function between the real-time signal and the background field signal;

[0043] S2, performing denoising processing on the similarity function obtained in step S1;

[0044] S3. Input the data processed in step S2 into a preset orthogonal basis function detector, and compare the output value of the orthogonal basis function detector with a preset threshold. If the output value is greater than the preset threshold, the magnetic anomaly signal is successfully matched, and it is determined that the target that generates the magnetic anomaly signal exists.

[0045] In a specific embodiment, the complete process of the aeromagnetic anomaly signal matching detection algorithm provided by the present invention is as follows:

[0046] First, in step S1, magnetic anomaly signal matching processing is performed based on similarity analysis. When a target appears in the detection flight area, the main difference between the real-time signal and the corresponding background field signal is that the real-time signal contains the target signal. Therefore, the key to target detection is to identify this difference to the greatest extent through matching analysis of magnetic anomaly signals. The present invention uses a sliding window to perform similarity analysis on the real-time signal and the corresponding background field signal. The window with poor similarity is the mismatch area of ​​the two measurement data. This information can be used to detect and locate the target.

[0047] Specifically, the magnetic anomaly signal is a non-periodic signal with a limited duration. The real-time signal collected by the sensor within a period of time is recorded as X current , as shown in formula (1), and the corresponding background field signal is extracted from the geomagnetic reference map according to the corresponding position information and recorded as X background , as shown in formula (2):

[0048]

[0049]

[0050] In equations (1) and (2), the value of M is determined by the time and the sampling rate of the sensor.

[0051] If the similarity between each matching point is directly calculated, the result will be greatly affected by noise. Therefore, a sliding window is added in the time domain to obtain a fixed-width signal sequence and then perform similarity analysis to improve the robustness of the algorithm. Specifically, a sliding window with a width of s and a step size of 1 along the time axis is used to smooth the real-time signal in equation (1) and the background field signal in equation (2), and the two are converted into M-s+1 sub-signals respectively. Any sub-signal after the conversion of equation (1) is shown in equation (3), and any sub-signal after the conversion of equation (2) is shown in equation (4):

[0052]

[0053]

[0054] In equations (3) and (4), i represents the window index, i=1, 2, 3, ..., M-s+1.

[0055] Since the magnetic anomaly signal collected by the sensor is a discrete time series, it is considered to match the magnetic anomaly signal through a sequence similarity measurement algorithm. Commonly used similarity measurement algorithms include Euclidean distance method, edit distance method, Hausdorff distance method, etc. Euclidean distance is an overall difference and is very sensitive to noise. The edit distance method eliminates the influence of noise by quantizing the distance between elements to 0 and 1, but this processing method easily ignores the details, resulting in inaccurate judgment results. Hausdorff distance cannot judge the similarity well when the curve trends are different. These algorithms are not suitable for direct use in the local similarity analysis of magnetic anomaly signals.

[0056] In the present invention, a new characteristic function is constructed as a measure of the local similarity between matching signals. For each given window i of length s, the similarity function between the real-time signal and the corresponding background field signal is defined as shown in formula (5):

[0057]

[0058] From the definition, we can see that the smaller the absolute value of the function is, the higher the similarity of the matching signals is; the window with a larger absolute value of the function indicates that the matching signals are very different and are more likely to contain target information.

[0059] The selection of smoothing window size s has a great influence on the matching detection effect. When s is small, the time domain resolution is high, but it is sensitive to noise and the result is easily affected by transient interference; when s is large, the output result is smoother and the robustness of the algorithm is improved, but more noise will be introduced, and the delay will increase, and the target positioning error will increase. Therefore, the selection of the optimal window size needs to comprehensively consider the signal-to-noise ratio and delay changes. The corresponding relationship between the simulation calculation signal-to-noise ratio and the window size s is as follows: Figure 2 As shown; the corresponding relationship between delay and window size s is as follows Figure 3 shown.

[0060] Depend on Figure 2 and Figure 3 It can be seen that as s increases, the signal-to-noise ratio shows an upward trend, especially when s increases to 3, the signal-to-noise ratio is significantly improved; after exceeding 3, the signal-to-noise ratio changes little, but as the window width increases, the delay continues to increase. To obtain a higher signal-to-noise ratio and a smaller delay, the optimal smoothing window width selected in this embodiment is 3, that is, s=3.

[0061] Next, enter step S2 to denoise the similarity function obtained in step S1. In step S1, the real-time signal and the background field signal are converted into a time series represented by a similarity function after matching analysis. In order to extract more stable signal features, it is necessary to filter the time series. Considering the algorithm's adaptability to non-Gaussian noise, in this implementation, a wavelet packet threshold denoising algorithm is used for denoising.

[0062] Wavelet packet analysis is essentially a multi-resolution time-frequency analysis method. By dividing the frequency band into multiple levels, a higher time-frequency resolution is obtained. Therefore, wavelet packet threshold denoising can better preserve the local characteristics of the signal and is a more sophisticated denoising method. In addition, since the energy of the target signal is mainly concentrated in the wavelet packet coefficients with larger amplitudes in the characteristic frequency band, while the noise energy is distributed in the entire wavelet domain, its coefficient amplitude is smaller. Therefore, some noise coefficients can be set to zero and the target signal characteristics can be retained by setting a threshold. Wavelet packet threshold denoising usually includes the following steps: ① Select appropriate wavelet basis functions and decomposition levels according to signal characteristics, and perform wavelet packet decomposition on the original signal. ② Threshold quantization of wavelet packet decomposition coefficients. ③ Reconstruct the signal based on the wavelet packet coefficients after threshold quantization.

[0063] Finally, enter step S3, input the processed data in step S2 into the preset orthogonal basis function detector, and compare the output value of the orthogonal basis function detector with the preset threshold value. If the output value is greater than the preset threshold value, the magnetic anomaly signal is successfully matched, and it is determined that the target generating the magnetic anomaly signal exists. In the process of aerial magnetic survey, due to the long detection distance, which is usually greater than twice the target size, the target magnetic field can be equivalent to the magnetic field generated by the magnetic dipole. Orthogonal basis function (OBF) detection is to decompose the magnetic dipole model into a linear combination of three standard orthogonal bases, and use the characteristics that the background noise is unrelated to the orthogonal base to square and sum the weighted coefficients of each base, and compare the result with the threshold value to achieve target detection.

[0064] The effectiveness of the present invention is verified by a specific simulation experiment.

[0065] In a specific simulation test, the detection flight area was located at 23.45° north latitude and 141.70° east longitude. The local magnetic declination was about -0.8° and the magnetic inclination was about 1.3°.

[0066] The existing difference matching algorithm is used to process the magnetic survey data and Gaussian white noise is mixed into the real-time magnetic anomaly signal of the simulated detection. When the peak value of the target magnetic anomaly signal is 0.28nT (the signal-to-noise ratio is -15dB), the target signal and the background signal in the experiment are respectively referred to Figure 4 and Figure 5 As shown in the figure, the result after processing by the difference matching algorithm is as follows Figure 6 shown. Figure 4-Figure 6 In the figure, the horizontal axis has been converted to the distance flown. It can be seen that due to the long detection distance and small target magnetic moment, the target signal is very weak compared to the background magnetic anomaly of several hundred nT; in addition, during the difference matching process, the target signal and the interference are offset or the interference is superimposed, which reduces the amplitude of the useful signal and enhances the interference. It is not effective in the case of low signal-to-noise ratio. Figure 6 In the figure, the target signal is in the circled area. It can be seen from the figure that the target signal is masked by the noise.

[0067] By using step S1 of the algorithm provided by the present invention, similarity analysis and matching processing is performed, and the similarity function obtained by calculation is further obtained as the similarity analysis and matching processing result. Figure 7 As shown, the target signal is also in the circled area. Figure 6 It can be seen that by using the method provided by the present invention, the signal peak at the target position can be clearly observed. Calculation shows that the algorithm result improves the signal-to-noise ratio by 5.22 dB compared with the difference signal, indicating that the method provided by the present invention can suppress noise well.

[0068] Furthermore, the output value of the similarity matching process is decomposed into three layers using the sym6 wavelet packet, and the wavelet packet coefficients are quantized using a hard threshold function to reconstruct the signal. Figure 8 Obviously, the signal after wavelet packet threshold denoising is better than Figure 7 The result is smoother and the peak of the target signal is more obvious, which shows that this method can suppress noise while retaining the characteristics of the target signal. Under the aviation matching detection model, different signal-to-noise ratio simulation conditions are set and the matching detection preprocessing method proposed in the present invention is used for processing. The results are shown in Table 1.

[0069] Table 1 Simulation data processing results with different signal-to-noise ratios

[0070]

[0071] It can be seen from Table 1 that the signal-to-noise ratio of the matching processing results based on similarity analysis is significantly improved compared with the difference signal, and the wavelet packet threshold denoising further suppresses the noise interference; the effect of wavelet packet threshold denoising is weakened when the input signal-to-noise ratio is high. This is because the information is lost when the threshold of the wavelet packet coefficients of each layer is quantized, but this situation will basically not affect the detection result of the target.

[0072] It can be seen that after being processed according to the algorithm provided by the present invention, a higher quality signal can be provided for subsequent links, greatly improving the robustness of the matching detection algorithm.

[0073] Next, the difference signal, the difference signal after wavelet packet threshold denoising and Figure 8 The signals shown are input into the orthogonal basis function (OBF) detector respectively. The length of the OBF detection sliding window is selected as 6. When SNR = -15dB, the output energy signal is as follows: Figure 9-11 As shown, Fig. 9 Corresponding to the difference signal input, Fig.10 Corresponding to the difference signal input after wavelet packet threshold denoising, Fig.11 Corresponding to the algorithm obtained according to the present invention Figure 8 Signal input shown.

[0074] from Figure 9-11 It can be seen that the target signal energy is not obvious in the difference matching-orthogonal basis function detection. Even if the denoised difference signal is input into the OBF detector, the target cannot be accurately detected by setting the threshold. This fully illustrates the limitation of difference matching. However, by using the method provided by the present invention, the target signal energy is very obvious and the time delay is smaller under similarity analysis-orthogonal basis function matching detection. The existence of the target can be accurately detected by the threshold value. This shows that the algorithm provided by the present invention has stronger anti-noise ability, more concentrated energy, better target detection effect, and is superior to the existing algorithms.

[0075] Finally, the Neyman-Pearson (NP) criterion is widely used in the detection of magnetic anomaly signals. It minimizes the probability of missed alarms under a given false alarm probability and is a commonly used criterion for hypothesis testing. In the process of aeromagnetic detection, since auxiliary means such as satellites and sonars can effectively identify false alarm targets, it is acceptable for the magnetometer to have less than or equal to 3 false alarms per hour. According to the simulation conditions, the allowable false alarm rate of the detection is set to 0.42%, and the detection threshold under different signal-to-noise ratios is calculated by the NP criterion. The Monte Carlo method is used to analyze the performance of the detection algorithm. The input signals of the three algorithms (the algorithm provided by the present invention, the difference matching-orthogonal basis function detection algorithm, and the difference matching-orthogonal basis function detection algorithm denoised by wavelet packet threshold filtering) are the same. The simulation is performed 1000 times under each signal-to-noise ratio condition, and the performance curve of the detection algorithm is obtained as shown in the figure. Fig.12 shown.

[0076] It can be seen that as the signal-to-noise ratio gradually decreases, the detection rates of the three algorithms continue to decrease, which is consistent with the actual situation; when the signal-to-noise ratio is <-3dB, the detection rate of the difference matching-orthogonal basis function detection decreases significantly. When the signal-to-noise ratio is -9dB, it is basically impossible to distinguish between the target and the noise, while the detection rate calculated by the algorithm provided by the present invention is still around 90%; even if the difference signal is denoised and then input into the OBF detector, the effect is still worse than the algorithm provided by the present invention, which once again illustrates the advantage of similarity matching analysis.

[0077] In summary, the aeromagnetic anomaly signal matching detection algorithm provided by the present invention has strong anti-noise ability and good weak magnetic detection ability; the detection rate of the algorithm under low signal-to-noise ratio is higher than that of the existing difference matching (filtering)-orthogonal basis function detection, and has better detection performance.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. An aeromagnetic anomaly signal matching detection algorithm, characterized in that: The steps include: S1. During the magnetic anomaly detection process, a real-time signal and a corresponding background field signal are obtained, and similarity analysis and matching processing is performed on the real-time signal and the background field signal to obtain a similarity function between the real-time signal and the background field signal; S2, performing denoising processing on the similarity function obtained in step S1; S3, inputting the data processed in step S2 into a preset orthogonal basis function detector, and comparing the output value of the orthogonal basis function detector with a preset threshold value. If the output value is greater than the preset threshold value, the magnetic anomaly signal is successfully matched, and it is determined that the target generating the magnetic anomaly signal exists; In step S1, the similarity function is obtained by specifically adopting the following steps: First, the real-time signal collected by the sensor within a period of time is recorded as X current , as shown in formula (1), and the corresponding background field signal is extracted from the geomagnetic reference map according to the corresponding position information and recorded as X background , as shown in formula (2): In equations (1) and (2), the value of M is determined by time and the sampling rate of the sensor; Next, a sliding window with a width of s and a step size of 1 along the time axis is used to smooth the real-time signal in equation (1) and the background field signal in equation (2), and the two are converted into M-s+1 sub-signals respectively. Any sub-signal after the conversion of equation (1) is shown in equation (3), and any sub-signal after the conversion of equation (2) is shown in equation (4): In formula (3) and formula (4), i represents the window index, i = 1, 2, 3, ..., M-s+1; Finally, for each given window i of length s, the similarity function is established and obtained using the following formula (5): In step S2, a wavelet packet threshold denoising algorithm is used to perform denoising, which specifically includes the following steps: First, wavelet basis functions and decomposition levels are selected, and the similarity function obtained in step S1 is decomposed by wavelet packets; Then, threshold quantization of wavelet packet decomposition coefficients is performed; Finally, the similarity function is reconstructed based on the wavelet packet coefficients after threshold quantization.

2. The aeromagnetic anomaly signal matching detection algorithm according to claim 1, characterized in that: In step S1, s=3.

Images