Penetration overload signal layer crossing feature extraction method based on improved Teager energy operator
By improving the Teager energy operator, analysis of the energy distribution of the frequency band signal and weighted fusion, the problems of invading high-frequency information loss and noise interference in signal processing in multi-layer targets of ammunition are solved, and more accurate layer-through energy feature extraction and resistance Noiseability.
Patent Information
- Application Number
- CN202411074746.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-07
- Publication Date
- 2024-11-22
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
When the prior art deals with multi-layer targets that penetrate the ammunition, it is difficult to effectively separate and identify the layer-through characteristics in the overload signal, resulting in high-frequency information loss and interfering signals affecting the accuracy of the meter layer recognition.
Using the improved Teager energy operator, the characteristic energy value is calculated through the frequency band signal energy distribution analysis, the energy mean calculation in the sliding window and normalized weighted fusion are performed, and the noise threshold is set to filter out the noise between layers, and the overload characteristic information of the entire frequency band is retained. .
It realizes accurate extraction and thoroughness of the energy characteristics of overload signal through layer, improves signal-to-noise ratio and recognition, reduces inter-layer noise interference, and overcomes the problems of high-frequency information loss and noise sensitivity.
Smart Images

Figure CN119004070A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of penetration signal processing, in particular to a method for extracting layer penetration features of a penetration overload signal based on an improved Teager energy operator. Background Art
[0002] In modern warfare, penetrating ammunition is an important weapon for attacking solid targets. As the core component of the ammunition system, the penetration fuze undertakes the important tasks of target identification and detonation control, and plays an important role in maximizing the damage efficiency of the entire ammunition. Layer counting detonation is a detonation control method for attacking multi-layer targets. Its working principle is that the penetration fuze obtains the overload signal of the target during the penetration process according to the acceleration sensor, uses a specific signal processing method and layer counting strategy, extracts the penetration characteristics to identify the number of layers penetrated by the projectile, and controls its detonation at the target layer. At present, multi-layer signal processing and identification have become the focus and difficulty of penetration fuze research.
[0003] The overload signal components of a projectile penetrating a multi-layer target are complex, including both the rigid body overload formed by the resistance when penetrating the target plate and the high-frequency noise formed by the oscillation of the projectile structure. Various signal components overlap each other in the time-frequency domain and are difficult to separate. In addition, the collision between the tail of the projectile and the target plate and the deflection of the flight attitude of the projectile when the projectile leaves the target will also cause abnormal overload signals, which will cause difficulties in the signal analysis of the penetration process and the layer identification. At present, the common signal processing methods based on overload signals mainly include low-pass filtering, wavelet decomposition, empirical mode decomposition, etc., which can achieve the effect of highlighting the low-frequency penetration resistance overload characteristics by filtering out the high-frequency signal components. However, since the high-frequency components often contain the instantaneous change characteristics of the penetration signal frequency, filtering out the high-frequency band components of the signal will cause high-frequency information loss, which reduces the accuracy of the recognition result. In addition, the commonly used signal processing methods have poor recognition effects on the abnormal signals generated by the collision of the tail of the projectile and the stress wave oscillation signals, and cannot effectively eliminate interference, affecting the accuracy of layer identification. Summary of the invention
[0004] In view of this, the purpose of the present invention is to overcome the shortcomings of the above-mentioned signal processing method and proposes a method for extracting the penetration feature of the penetration overload signal based on the improved Teager energy operator. The method effectively retains the overload feature information of the entire frequency band, reduces the interference of inter-layer noise and abnormal overload signals, and achieves the purpose of accurately extracting the penetration energy characteristics of the penetration overload signal.
[0005] To achieve the above object, the technical solution of the present invention is as follows:
[0006] Step 1: Simulate or measure to obtain multi-layer penetration overload data, perform time-frequency analysis on the test signal to determine the signal frequency range and main frequency band, divide the signal outside the main frequency band, and extract the signal components in each frequency band through a bandpass filter.
[0007] Step 2: Set a sliding window to traverse each signal component, calculate the characteristic energy value of the data point in turn according to the energy operator, and calculate the energy mean in the sliding window to obtain the signal energy distribution of each frequency band component;
[0008] Step 3: Normalize each energy signal component and then perform weighted fusion, set the noise threshold to filter out inter-layer noise. Furthermore, the frequency division of the overload signal in step 1 also includes the following sub-steps:
[0009] Step 1.1: Perform fast Fourier transform on the test signal to obtain the single-sided spectrum of the signal. The signal frequency range is: 0~f n Hz, where f s is the signal sampling frequency, and the main frequency band of the signal is determined according to the spectrum diagram (f m-1 ,f m ).
[0010] Step 1.2: Divide the signal within the frequency range into n frequency bands: (0,f1),...,(f m-1 ,f m ),...,(f n-1 ,f n ), and extract the signals of each frequency band through band-pass filters.
[0011] Furthermore, the energy distribution average value of the signal in the short-time sliding window is calculated as follows:
[0012]
[0013] Furthermore, the window slides by one data unit length each time, and the energy distribution average value calculation formula is iteratively calculated as follows:
[0014]
[0015] Among them, x k (i) represents the kth frequency division signal component x k The i-th data point, E k (i) represents the average energy distribution of the signal in the window at the i-th data point of the k-th frequency-divided signal, N s is a fixed time window length, CF k is the characteristic function used by the kth frequency division signal. The calculation of the energy mean at the beginning and end of the signal sequence is used to fill the missing data with zeros.
[0016] Furthermore, the characteristic energy value is calculated according to the improved energy operator CF k calculate:
[0017]
[0018] Furthermore, the energy distribution of the signal in each frequency band is calculated in turn: E1(i), E2(i), ..., E n (i) Passing Energy distribution diagram converted to time domain.
[0019] Further normalize the energy signal components of each frequency band for preprocessing:
[0020]
[0021] Furthermore, the fusion weight of the signal components of each frequency band is solved by the following formula:
[0022]
[0023] Furthermore, P k is the weighted coefficient of variation between the kth signal energy sequence and other sequences, and its calculation formula is:
[0024]
[0025] In the above formula, μ is the main frequency band weight coefficient, which is 0.4. Cor(a,b) is the Charlson correlation coefficient, which is calculated as follows: cov(a,b) is the covariance of the signal sequences a and b, σ a , σ b are the standard deviations of the signal sequences a and b respectively.
[0026] Furthermore, the normalized cross-layer energy feature is obtained by weighted fusion of each component:
[0027]
[0028] The advantages of the present invention are:
[0029] 1. The present invention innovatively proposes to analyze the layer penetration characteristics of overload signals from the perspective of frequency band signal energy distribution. The energy value calculated by the characteristic function adopts to characterize the time-varying characteristics of the signal amplitude and frequency, and retains the amplitude-frequency energy characteristics of the layer penetration signal to the greatest extent. The processed signal has a higher signal-to-noise ratio and good recognition.
[0030] 2. The present invention processes multi-layer penetration overload signals to obtain the time domain energy distribution of signals in each frequency band and the normalized characteristic energy distribution. The processed signals reflect the penetration energy information of the entire frequency band, overcoming the shortcoming of the traditional signal processing method that lacks high-frequency information, and making the processing results more accurate.
[0031] 3. The energy signal processed by the present invention is insensitive to signal noise and pulse signals generated by the tail of the projectile hitting the target, and has good noise resistance. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] Figure 1 Signal processing flow chart
[0033] Figure 2 To simulate overload signal
[0034] Figure 3 is the short-time Fourier transform frequency spectrum
[0035] Figure 4 Energy distribution diagram of each signal component
[0036] Figure 5 The fusion result DETAILED DESCRIPTION
[0037] In order to deepen the understanding of the present invention, the embodiments of the present invention are further described below in conjunction with the accompanying drawings, but the protection scope of the present invention is not limited thereby.
[0038] The signal processing flow chart of the method for extracting the penetration overload signal layer penetration feature based on the improved Teager energy operator in this example is as follows: Figure 1 As shown, the steps include:
[0039] Step 1: Simulate or measure to obtain multi-layer penetration overload data, perform time-frequency analysis on the test signal to determine the signal frequency range and main frequency band, divide the signal outside the main frequency band, and extract the signal components in each frequency band through a bandpass filter.
[0040] Step 2: Set a sliding window to traverse each signal component, calculate the characteristic energy value of the data point in turn according to the energy characteristic function, and calculate the energy mean in the sliding window to obtain the signal energy distribution of each frequency band component; Step 3: Normalize each energy signal component and perform weighted fusion, set the noise threshold to filter out inter-layer noise.
[0041] This embodiment obtains the overload curve of the fuze position where the projectile penetrates ten layers of concrete target plate through simulation, such as Figure 2 As shown, the signal sampling frequency f s is 100000 Hz. According to step 1, the original frequency spectrum is analyzed by short-time Fourier transform, and a Hamming window with a window length of 50 and a time window overlap length of 30 is used to obtain the following: Figure 3 Short-time Fourier transform time-frequency diagram. According to the time-frequency spectrum diagram, the signal has serious adhesion in the time domain below 5000Hz, has a higher power density and better stratification in the range of 7000-15000Hz, and this range is determined as the main frequency band of the signal. The remaining frequency bands are divided into 5 frequency bands using wide-band division, namely (0, 7000) Hz, (7000, 15000) Hz, (15000, 30000) Hz, (30000, 40000) Hz, and (40000, 50000) Hz. The signal components of each frequency band are extracted through a bandpass filter.
[0042] In step 2, a sliding window is set to traverse each signal component, and the characteristic energy values of the data points are calculated in sequence according to the energy operator:
[0043]
[0044] Calculate the mean energy in the sliding window:
[0045]
[0046] The time window length is set to 22, and the signal energy of each frequency band is calculated in turn: E1, E2, ..., E n . Pass-through The energy distribution diagram converted into time domain is as follows: Figure 4 shown.
[0047] In step 3, after normalizing the data of each component, the weighted coefficient of variation is calculated according to the following formula:
[0048]
[0049] The fusion weight of each component is solved by the following formula:
[0050]
[0051] The fusion weight values of each component are calculated and shown in the table below.
[0052] signal component Portion one Serving size 2 Serving size 3 Serving size 4 5 servings Fusion weight 0.23583 0.35778 0.11584 0.14333 0.14719
[0053] The component signals are weighted and fused, and the noise threshold is set to filter the inter-layer noise energy signal, and finally the fusion result is obtained, such as Figure 5 Compared with the original signal, the fusion result has obvious energy characteristics of penetration layer, and the inter-layer interference signal is significantly suppressed, which can be used for post-processing of multi-layer penetration signals.
[0054] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.
Claims
1. A method for extracting layer penetration features of penetration overload signals based on an improved Teager energy operator, characterized in that: The steps include: Step 1: Obtain multi-layer penetration overload data, perform time-frequency analysis on the test signal to determine the signal frequency range and main frequency band, divide the signal outside the main frequency band, and extract the signal components of each frequency band through a bandpass filter. Step 2: Set a sliding window to traverse each signal component, calculate the characteristic energy value of the data point in turn according to the energy operator, and calculate the energy mean in the sliding window to obtain the signal energy distribution of each frequency band component; Step 3: Normalize each energy signal component and perform weighted fusion, and set the noise threshold to filter out inter-layer noise.
2. According to claim 1, a method for extracting penetration overload signal layer penetration features based on improved Teager energy operator is characterized in that: In step 1, the frequency division of the overload signal comprises the following steps: Step 1.1: Perform fast Fourier transform on the test signal to obtain the single-sided spectrum of the signal. The signal frequency range is: 0~f n Hz, where f s is the signal sampling frequency, and the main frequency band of the signal is determined according to the spectrum diagram (f m-1 ,f m ). Step 1.2: Divide the signal within the frequency range into n frequency bands: (0,f1),...,(f m-1 ,f m ),...,(f n-1 ,f n ), and extract the signals of each frequency band through band-pass filters.
3. According to claim 1, a method for extracting layer penetration features of penetration overload signals based on an improved Teager energy operator is characterized in that: The calculation formula for the average value of the energy distribution of the signal in the short-time sliding window in step 2 is: The window slides by one data unit length each time, and its iterative calculation formula is: Among them, x k (i) represents the kth frequency division signal component x k The i-th data point, E k (i) represents the average energy distribution of the signal in the window at the i-th data point of the k-th frequency-divided signal, N s is a fixed time window length, CF k is the characteristic function used by the kth frequency-divided signal, and the calculation of the energy mean at the beginning and end of the signal sequence is used to fill the missing data with zeros.
4. The method for extracting layer penetration features of penetration overload signals based on improved Teager energy operator according to claim 3 is characterized in that: The characteristic energy value in step 2 is calculated based on the improved energy operator CF k calculate: According to the energy mean calculation method, the energy distribution of each frequency band signal is calculated in turn: E1(i), E2(i), ..., E n (i) Passing Energy distribution diagram converted to time domain.
5. The method for extracting layer penetration features of penetration overload signals based on improved Teager energy operator according to claim 1 is characterized in that: In step 3, the signal components of each frequency band are normalized and preprocessed:
6. The method for extracting layer penetration features of penetration overload signals based on improved Teager energy operator according to claim 1 is characterized in that: In step 3, the fusion weight of the signal components of each frequency band is solved by the following formula: Among them, P k is the weighted coefficient of variation between the kth signal energy sequence and other sequences, and its calculation formula is: In the above formula, μ is the main frequency band weight coefficient, which is 0.
4. Cor(a,b) is the Charlson correlation coefficient, which is calculated as follows: cov(a,b) is the covariance of the signal sequences a and b, σ a , σ b are the standard deviations of signal sequences a and b respectively.
7. The method for extracting layer penetration features of penetration overload signals based on improved Teager energy operator according to claim 1 is characterized in that: The normalized cross-layer energy feature in step 3 is obtained by weighted fusion of each component: In the above formula, ω k is the fusion weight of k-band signal components, is the k-band signal component after normalization, and E is the final fusion result.
Citation Information
Patent Citations
Self-adaptive time-frequency synchronous compression method based on Rayleigh entropy
CN107576943A
Anti-noise speech recognition system
CN108564965A
Penetration overload signal feature extraction method
CN111007280A
Pathological voice spectrum reconstruction method based on energy operator
CN114822567A
