Acquisition precision improving method of multi-sub-band frequency alternating acquisition system

Abstract

The invention discloses a method for improving the acquisition precision of a multi-sub-band frequency alternate acquisition system, and the method comprises the steps: enabling the multi-sub-band frequency alternate acquisition system to be started, and carrying out the initialization, so as to obtain the energy characteristics of background noise; then, switching the multi-sub-band frequency alternate acquisition system to a normal acquisition mode, and then inputting a signal to start normal acquisition; during normal acquisition, the amplitude-frequency response of the data acquired each time is also calculated, the data is converted into energy on each frequency point, the energy on each frequency point is compared with the energy characteristic of noise, so that the energy distribution of the signal in the acquisition is obtained, and a plurality of signal energy distribution results are accumulated to obtain the actual existence position of the signal; a frequency domain filter is designed according to the actual existence position of the signal, and the signal is output to a frequency domain filtering module for coefficient updating, and finally the purposes of automatically reducing noise outside an effective frequency band in the frequency alternating system and intelligently improving the signal acquisition precision are achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of frequency alternating acquisition systems, and more specifically, relates to a method for improving acquisition accuracy of a multi-subband frequency alternating acquisition system. Background Art

[0002] Ultra-wideband acquisition systems are widely used in various cutting-edge industries. For example, 5G and 6G communications require broadband analysis of signals of tens of GHz. However, due to device limitations, there is currently no single acquisition device that can achieve tens of GHz bandwidth and tens or even hundreds of GSPS sampling rates. Therefore, various system-level technologies have been proposed to improve the sampling rate and bandwidth of the acquisition system.

[0003] In order to improve the sampling rate and bandwidth of the system at the same time, the most commonly used technology is to use frequency alternating acquisition technology, such as Figure 1 As shown in the figure, in the frequency alternating acquisition technology, the system is divided into analog conditioning part, digital processing part and control and analysis center with the collector as the dividing line. The input signal first passes through M decomposition filters, each of which decomposes the signal into different sub-bands according to the frequency: the frequency band range of the decomposition filter m is frequency f m-1 To frequency f m , where m ranges from 1 to M, and f0 represents direct current; after passing through the decomposition filter, the second to Mth signals each pass through a mixer with a different local oscillator to down-convert the high-frequency signal to the baseband frequency; each baseband signal passes through a sampler to obtain a baseband digital signal; each baseband digital signal is interpolated M times through each interpolator to obtain a baseband signal at the system sampling rate; finally, the second to Mth baseband signals at the system sampling rate are then passed through an up-converter with the same mixing frequency as the mixer to restore the baseband signal to the original frequency band; finally, all frequency band signals are superimposed to obtain the final acquisition result, which is sent to the control and analysis center for further analysis, display and other operations.

[0004] Based on the frequency alternating acquisition technology, each sub-band is superimposed, which can achieve a doubling of the system bandwidth. However, the additional bandwidth inevitably brings additional noise, which will reduce the amplitude accuracy of the final acquisition result, thereby affecting the subsequent analysis of the signal. There are almost no methods for improving the acquisition accuracy of such systems in existing patents and literature. The accuracy improvement of traditional broadband acquisition systems often relies on filters to suppress out-of-band noise. However, the traditional filter noise suppression method has the following difficulties: First, it often uses various low-pass and band-pass filters to achieve noise suppression. For signals with multiple independent frequency bands, on the one hand, it is difficult to design a single filter with multiple narrow passbands, so as to simultaneously filter signals in multiple frequency bands; on the other hand, if multiple filters are designed, each filter has a single narrow passband, which will greatly increase the resource consumption of the system. In addition, the traditional design method can only process signals with fixed frequency bands, and it is difficult to identify the frequency band characteristics of the signal and quickly reconstruct the filter, so it is impossible to intelligently improve the accuracy for various different test occasions.

[0005] Therefore, in order to reduce the noise in signal testing and improve the accuracy of signal testing, it is necessary to design a new method to automatically identify the frequency band of the signal, and automatically design a multi-band bandpass filter under the low sampling rate condition of single-channel sampling data based on the identification result, so as to reduce the overall noise of the acquisition system as much as possible and realize the intelligent improvement of the signal acquisition accuracy in the frequency alternating acquisition system. Summary of the invention

[0006] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a method for improving the acquisition accuracy of a multi-subband frequency alternating acquisition system, by identifying the frequency band of the acquired signal and automatically designing a band-limited filter that meets the signal frequency band, so as to maximize the effect of improving the signal acquisition accuracy.

[0007] To achieve the above-mentioned object of the invention, the present invention provides a method for improving the acquisition accuracy of a multi-subband frequency alternating acquisition system, characterized by comprising the following steps:

[0008] (1) The multi-subband frequency alternating acquisition system is turned on and first enters the initialization phase to obtain the energy characteristics of the background noise;

[0009] (1.1) Assume that the multi-subband frequency alternating acquisition system includes M collectors;

[0010] (1.2) When the multi-subband frequency alternating acquisition system is turned on and initialized, no signal is input. Suppose the pure noise collected by the mth collector of the system is x m [i], where m = 1, 2, ..., M, i represents the sampling point number;

[0011] (1.3), each channel of pure noise xm [i] The sub-bands are processed by the digital processing part. The processing process is as follows:

[0012] (1.3.1), take each channel of pure noise x m The first N points of [i] are then transformed by N-point FFT to obtain the frequency response X m [n], n = 1, 2, ..., N;

[0013] (1.3.2) Calculate the amplitude-frequency response A m [n];

[0014]

[0015] Among them, real(·) and imag(·) represent taking the real part and taking the imaginary part respectively;

[0016] (1.3.3), the amplitude-frequency response A m [n] and sent to the signal recognition module, first through the frequency energy acquisition module to obtain the frequency energy E m [n] = 20log10(A m [n]); then the frequency energy E m [n] is input to the noise energy calculation module to calculate the background noise energy characteristic E of the mth subband Nm ; Finally, the background noise energy characteristic E of each sub-band Nm Stored in a noise energy storage module;

[0017] or E Nm =E Nmmax =max(E m [n])

[0018] Among them, E Nmavg Represents the average energy at all frequency points, E Nmmax Indicates the maximum value at all frequency points;

[0019] (2) Switch the multi-subband frequency alternating acquisition system to the normal acquisition mode, and then start normal signal acquisition;

[0020] (2.1) Set the coefficient H of the frequency domain filter in each subband m [n]=1,n=1,2,…,N;Set the number of acquisition cycles C required for coefficient update, initialize the current acquisition number c=0,c=1,2,…,C;Set the count value J in the cumulative count module of the signal effective frequency band tm [n] = 0; Set the initial output value of the signal effective frequency band counting module to J Rm [n] = 1; Set the energy threshold δ in frequency band identification e、The number of times the signal is judged to exist is c e , the extended range threshold δ for solving the frequency domain filter coefficients f ;

[0021] (2.2) Input the signal at the input end of the multi-subband frequency alternating acquisition system, and then collect it through each collector in the system. The acquisition signal of the mth collector in the cth acquisition cycle is recorded as

[0022] (2.3) Each channel of collected signal The sub-band is processed by the digital processing part, and the processing process is as follows:

[0023] (2.3.1) Obtain the filter coefficients of the frequency domain filter in the next acquisition in the lower branch of each subband

[0024] (2.3.1.1) Collecting signals First, follow steps (1.3.1)-(1.3.2) to obtain the amplitude-frequency response of each signal.

[0025] (2.3.1.2) The amplitude-frequency response Sent to the signal recognition module, the frequency energy is first obtained through the frequency energy acquisition module

[0026] (2.3.1.3) The frequency energy The signal is input to the signal frequency band identification module, which reads the background noise energy feature E in the noise energy storage module. Nm , and then determine the location of the collected signal and the result for:

[0027]

[0028] Among them, δ e is the energy threshold in frequency band identification;

[0029] (2.3.1.4) The judgment result Input to the signal effective frequency band counting module for accumulation;

[0030] (2.3.1.4.1) When c≤C-1, firstly, the signal effective frequency band counting module is accumulated and summed, and the result is

[0031]

[0032] (2.3.1.4.2), after summing, update the output J Rm[n]: When c<C-1, output J Rm [n] remains unchanged; when c = C-1, the sum result J tm [n] The number of times the signal exists e Compare and make the module output J Rm [n] is:

[0033]

[0034] (2.3.1.4.3) Update the loop count number after updating the output: when c<C-1, let c=c+1; when c=C-1, let c=0;

[0035] (2.3.1.5), the frequency domain filter coefficient solution module is based on J Rm [n] Design the filter coefficients of the frequency domain filter for the next acquisition

[0036] (2.3.2) acquiring the collected data with improved collection accuracy in the upper branch of each sub-band;

[0037] (2.3.2.1) and collect the signal Filtering is performed through a frequency domain filter to obtain the filtered output

[0038] (2.3.2.2) After filtering is completed, the filter coefficients output by the frequency domain filter receiving signal recognition module are and update it;

[0039] (2.3.2.3), in the first subband, the output after filtering After being processed by the interpolator, the output y1[i] of the first subband is obtained;

[0040] In the 2nd to Mth subbands, the filtered output After being processed by the interpolator and up-converter in turn, the output y of each sub-band is obtained m [i], m = 2, 3, ..., M;

[0041] (2.3.2.4), all sub-band output y m [i] are superimposed to obtain the system output y[i];

[0042]

[0043] The object of the invention of the present invention is achieved in this way:

[0044] The present invention discloses a method for improving the acquisition accuracy of a multi-subband frequency alternating acquisition system. The multi-subband frequency alternating acquisition system is powered on and firstly initialized to obtain the energy characteristics of background noise. Then, the multi-subband frequency alternating acquisition system is switched to a normal acquisition mode, and then a signal is input to start normal acquisition. During normal acquisition, the amplitude-frequency response of the data acquired each time is also calculated and converted into energy at each frequency point, and the energy at each frequency point is compared with the energy characteristics of noise, so as to obtain the energy distribution of the signal in the acquisition, and the energy distribution results of the signal are accumulated for multiple times to obtain the actual existence position of the signal. A frequency domain filter is designed according to the actual existence position of the signal, and is output to a frequency domain filter module for coefficient update, so as to finally achieve the purpose of automatically reducing the noise outside the effective frequency band in the frequency alternating system and intelligently improving the signal acquisition accuracy.

[0045] At the same time, the acquisition accuracy improvement method of a multi-subband frequency alternating acquisition system of the present invention also has the following beneficial effects:

[0046] (1) Directly designing real coefficient filters in the frequency domain for filtering reduces the design requirements for filters for multi-band signals. For designing multiple single narrowband filters, this method only needs to use one filter, reducing resource consumption; for designing a single multi-subband filter, this method directly only performs simple 0-1 judgment or slope calculation operations, without the need to use complex algorithms to design filter coefficients, so the design method is simpler.

[0047] (2) This method can adaptively change the filter parameters according to different signal characteristics and different test scenarios. When the test scenario changes, it can automatically measure the frequency band of the signal and quickly reconstruct the filter parameters to reduce the out-of-band noise of the signal in the new test scenario. This process does not require human intervention, so it can intelligently improve the accuracy of various test scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is the architecture diagram of the traditional frequency alternating acquisition system;

[0049] Figure 2 It is a sub-band structure diagram of the digital processing part of the frequency alternating acquisition system;

[0050] Figure 3 It is a structural diagram of the signal identification module in each sub-band in the digital processing part;

[0051] Figure 4 It is the overall structure diagram of the dual sub-band frequency alternating system for collecting signals and intelligently improving them;

[0052] Figure 5 It is a schematic diagram of the frequency domain filter design of brick wall type, semi-ladder form and ladder form;

[0053] Figure 6 This is a schematic diagram of the values ​​when adjacent frequency bands in the half-ladder form overlap. DETAILED DESCRIPTION

[0054] The specific implementation of the present invention is described below in conjunction with the accompanying drawings so that those skilled in the art can better understand the present invention. It should be noted that in the following description, when the detailed description of known functions and designs may dilute the main content of the present invention, these descriptions will be omitted here.

[0055] Example

[0056] In this embodiment, the frequency alternating acquisition system is composed of M sub-bands, numbered 1, 2, ..., m, ..., M; the system analog signal conditioning part is Figure 1 The same. Each sub-band satisfies the Nyquist sampling theorem, that is, the sampling rate is greater than 2 times the input bandwidth of the sub-band. The difference lies in the digital processing part, where the structures of the 2nd to the Mth sub-bands are the same. Take the mth sub-band as an example, Figure 2 As shown, it includes two branches, the upper branch includes a frequency domain filter, an interpolator and an up-converter connected in series in sequence; the lower branch includes an FFT transformation module, an amplitude-frequency response module and a signal recognition module;

[0057] The collected data x for each channel m [i] After entering the digital processing part, it will be divided into two paths, one of which is the normal signal processing path in the upper branch, and the other is the signal recognition path in the lower branch.

[0058] In the upper branch, the collected signal is first filtered by the frequency domain filter, then interpolated by the interpolator, and finally up-converted by the up-converter as the output of the sub-band; in the lower branch, the collected signal is first FFT transformed, then the amplitude-frequency response is obtained by the amplitude-frequency response module, and finally the filter coefficient is obtained by the signal recognition module and fed back to the frequency domain filter as the frequency domain filter coefficient for the next acquisition;

[0059] The structure of the first sub-band is basically the same as that of the remaining M-1 sub-bands, the only difference being that the up-converter is removed from the upper branch and the output of the interpolator is used as the output of the first sub-band.

[0060] In this embodiment, the internal principle structure of the signal recognition module is as follows: Figure 3 As shown, the signal identification module includes two signal processing branches, the first signal processing branch is used for initializing the multi-sub-band frequency alternating acquisition system, and the second signal processing branch is used for the multi-sub-band frequency alternating acquisition system to acquire input signals;

[0061] When the multi-subband frequency alternating acquisition system is initialized, the amplitude-frequency response is calculated through the frequency energy obtaining module to calculate the frequency energy, and then the frequency energy is obtained through the noise energy obtaining module to obtain the background noise energy characteristics and store them in the noise energy storage module;

[0062] When the multi-subband frequency alternating acquisition system collects input signals, the amplitude-frequency response first calculates the frequency energy through the frequency energy acquisition module, and then inputs it into the signal frequency band identification module by reading the background noise energy characteristics, and then obtains the filter coefficient through the signal effective frequency band cumulative counting module and the frequency domain filter coefficient design module in turn.

[0063] like Figure 4 As shown, below we take the overall structure of intelligent signal enhancement of the dual sub-band frequency alternating system as an example to describe the digital processing flow of each sub-band in detail, including the following steps:

[0064] (1) The multi-subband frequency alternating acquisition system is turned on and first enters the initialization phase to obtain the energy characteristics of the background noise;

[0065] (1.1) The multi-subband frequency alternating acquisition system includes M = 2 collectors;

[0066] (1.2) When the multi-sub-band frequency alternating acquisition system is powered on and initialized, no signal is input. At this time, the input of each sub-band digital processing part is pure noise. Suppose the pure noise collected by the mth collector of the system is x m [i], where m = 1, 2, i represents the sampling point number;

[0067] (1.3), each channel of pure noise x m [i] The sub-bands are processed by the digital processing part. The processing process is as follows:

[0068] (1.3.1), take each channel of pure noise x m [i] The first N = 1024 points, and then through the N-point FFT transformation, get the frequency response X m [n], that is, x1[1] to x1

[1024] is subjected to 1024-point FFT to obtain the frequency response X1[n], and x2[1] to x2

[1024] is subjected to 1024-point FFT to obtain the frequency response X2[n];

[0069] (1.3.2) Calculate the amplitude-frequency response A m [n];

[0070]

[0071] Among them, real(·) and imag(·) represent taking the real part and taking the imaginary part respectively;

[0072] (1.3.3), the amplitude-frequency response A m [n] and sent to the signal recognition module, first through the frequency energy acquisition module to obtain the frequency energy E m [n] = 20log10(A m [n]); then the frequency energy E m [n] is input to the noise energy calculation module to calculate the background noise energy characteristic E of the mth subband Nm ; Finally, the background noise energy characteristic E of each sub-band Nm Stored in a noise energy storage module;

[0073]

[0074] Among them, E Nmavg Represents the average energy at all frequency points, E Nmmax Indicates the maximum value at all frequency points;

[0075] (2) Switch the multi-subband frequency alternating acquisition system to the normal acquisition mode, and then start normal signal acquisition;

[0076] (2.1) Set the coefficient H of the frequency domain filter in each subband m [n]=1,n=1,2,…,N;Set the number of acquisition cycles required for coefficient update C=10, initialize the current acquisition number c=0, c=1,2,…,C;Set the count value J in the cumulative count module of the signal effective frequency band tm [n] = 0; Set the initial output value of the signal effective frequency band counting module to J Rm [n] = 1; Set the energy threshold δ in frequency band identification e =20dB, the number of times the signal is judged to exist is c e =8, extended range threshold δ for solving the frequency domain filter coefficients f =10;

[0077] (2.2) Input the signal at the input end of the multi-subband frequency alternating acquisition system, and then collect it through each collector in the system. Record the collected signal of the mth collector in the cth acquisition cycle:

[0078] (2.3) Each channel of collected signal The sub-band is processed by the digital processing part, and the processing process is as follows:

[0079] (2.3.1) Obtain the filter coefficients of the frequency domain filter in the next acquisition in the lower branch of each subband

[0080] (2.3.1.1) Collecting signals First, follow steps (1.3.1)-(1.3.2) to obtain the amplitude-frequency response of each signal.

[0081] (2.3.1.2) The amplitude-frequency response And sent to the signal recognition module, the frequency energy is first obtained by the frequency energy acquisition module

[0082] (2.3.1.3) The frequency energy The signal is input to the signal frequency band identification module, which reads the background noise energy feature E in the noise energy storage module. Nm , and then determine the location of the collected signal and the result for:

[0083]

[0084] Among them, δ e is the energy threshold in frequency band identification;

[0085] (2.3.1.4) The judgment result Input to the signal effective frequency band counting module for accumulation;

[0086] (2.3.1.4.1) When c≤C-1, firstly, the signal effective frequency band counting module is accumulated and summed, and the result is

[0087]

[0088] (2.3.1.4.2), after summing, update the output J Rm [n]: When c<C-1, output J Rm [n] remains unchanged; when c = C-1, the sum result J tm [n] The number of times the signal exists e Compare and make the module output J Rm [n] is:

[0089]

[0090] (2.3.1.4.3) Update the loop count number after updating the output: when c<C-1, let c=c+1; when c=C-1, let c=0;

[0091] (2.3.1.5), the frequency domain filter coefficient solution module is based on J Rm [n] Design the filter coefficients of the frequency domain filter for the next acquisition

[0092] In this embodiment, the frequency domain filter coefficient design module designs the filter coefficient H m [n] methods include brick wall, half ladder and ladder, such as Figure 5 Due to the symmetry of the spectrum and frequency domain filters, only filters from 1 to N / 2 are drawn here. Assume that the signal in the mth subband has a total of S frequency bands, and the range of the sth frequency band is n sL to n sR , n sL <n sR , s=1,2,…,S;

[0093] The specific process of the brick wall design method is:

[0094] If the judgment result J Rm [n] in n sL to n sR If the interval is 1, then H m [n] in max(1, n sL -δ f ) to min(n sR +δ f , N) is 1;

[0095] The specific process of the half-ladder design method is:

[0096] If the judgment result J Rm [n] in n sL to n sR If the interval is 1, then H m [n] from n sL Start to decrease linearly to max(1, n sL -δ f ) is 0, and from n sR Start to decrease linearly to min(n sR +δ f , N) is 0. If the values ​​of two adjacent frequency bands overlap, the larger value at the same point is selected as the final value H m [n], such as Figure 6 As shown, after the crossover, the final H m [n] Take the solid line as the coefficient value:

[0097] The specific process of the ladder form design method is:

[0098] If the judgment result J Rm [n] in n sL to n sR If the interval is 1, then:

[0099] When s = 1, H m [n] from n 1L It starts to decrease linearly and reaches 0 at n=1.1R Start to decrease linearly to n 1R +floor((n 2L -n 1R ) / 2) is 0, where floor() means rounding down;

[0100] When s=2,3,…,S-1, H m [n] from n sL Start to decrease linearly to n (s-1)R +floor((n sL -n (s-1)R ) / 2)+1 is 0, and from n sR Start to decrease linearly to n sR +floor((n (s+1)L -n sR ) / 2) is 0;

[0101] When s=S, H m [n] from n sL Start to decrease linearly to n (S-1)R +floor((n SL -n (S-1)R ) / 2)+1 is 0, and from n SR It starts to decrease linearly and reaches 0 at N / 2.

[0102] In this embodiment, the brick wall type is taken as an example. Assuming that the frequency band of the input signal in this scenario is in the second subband, the positions of the existing signals are numbered 5 to 15, 20 to 70, 233, 509 to 510. According to the symmetry of the spectrum, the positions of the existing signals in the second Nyquist domain are numbered 516 to 517, 793, 956 to 1006, 1011 to 1021, that is, J Rm [n] is:

[0103] J R1 [n]=0n=1,2,...,1024

[0104] as well as

[0105]

[0106] Since all n J on the first subband R1 [n] is 0, so H1[n] is also 0 for all n, that is,

[0107] H1[n]=0n=1,2,...,1024

[0108] For the second subband where there is a position number, H2[n] is in max(1, n sL -δ f ) to min(nsR +δ f , N) is 1. By assumption, it can be obtained that the signal has a total of 8 frequency bands, namely n 1L =5,n 1R =15; n 2L =20,n 2R =70; n 3L =233,n 3R =233; n 4L =509,n 4R =510;n 5L =516,n 5R =517;n 6L =793,n 6R =793; n 7L =956,n 7R =1006;n 8L =1011,n 1R = 1021. Calculate n for each frequency band sL -10 and n sR +10, and limit the result to [1,1024], we get:

[0109]

[0110] H1[n] and H2[n] are received by the frequency domain filters of subband 1 and subband 2, respectively, and are used for the frequency domain filtering operation during the next acquisition.

[0111] (2.3.2) acquiring the collected data with improved collection accuracy in the upper branch of each sub-band;

[0112] (2.3.2.1) and collect the signal Filtering is performed through a frequency domain filter to obtain the filtered output

[0113] (2.3.2.2) After filtering is completed, the filter coefficients output by the frequency domain filter receiving signal recognition module are and update it;

[0114] (2.3.2.3), in the first subband, the output after filtering After being processed by the interpolator, the output y1[i] of the first subband is obtained;

[0115] In the 2nd to Mth subbands, the filtered output After being processed by the interpolator and up-converter in turn, the output y of each sub-band is obtained m [i], m = 2, 3, ..., M;

[0116] (2.3.2.4), all sub-band output ym [i] are superimposed to obtain the system output y[i];

[0117]

[0118] Although the above describes the illustrative specific embodiments of the present invention to facilitate those skilled in the art to understand the present invention, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations using the concept of the present invention are protected.

Claims

1. A method for improving the acquisition accuracy of a multi-subband frequency alternating acquisition system, characterized in that: The following steps are involved: (1) The multi-subband frequency alternating acquisition system is turned on and first enters the initialization phase to obtain the energy characteristics of the background noise; (1.1) Assume that the multi-subband frequency alternating acquisition system includes M collectors; (1.2) When the multi-subband frequency alternating acquisition system is turned on and initialized, no signal is input. Suppose the pure noise collected by the mth collector of the system is x m [i], where m = 1, 2, ..., M, i represents the sampling point number; (1.3), each channel of pure noise x m [i] The sub-bands are processed by the digital processing part. The processing process is as follows: (1.3.1), take each channel of pure noise x m The first N points of [i] are then transformed by N-point FFT to obtain the frequency response X m [n], n = 1, 2, ..., N; (1.3.2) Calculate the amplitude-frequency response A m [n]; Among them, real(·) and imag(·) represent taking the real part and taking the imaginary part respectively; (1.3.3), the amplitude-frequency response A m [n] and sent to the signal recognition module, first through the frequency energy acquisition module to obtain the frequency energy E m [n] = 20log10(A m [n]); then the frequency energy E m [n] is input to the noise energy calculation module to calculate the background noise energy characteristic E of the mth subband Nm ; Finally, the background noise energy characteristic E of each sub-band Nm Stored in a noise energy storage module; Among them, E Nmavg Represents the average energy at all frequency points, E Nmmax Indicates the maximum value at all frequency points; (2) Switch the multi-subband frequency alternating acquisition system to the normal acquisition mode, and then start normal signal acquisition; (2.1) Set the coefficient H of the frequency domain filter in each subband m [n]=1,n=1,2,…,N;Set the number of acquisition cycles C required for coefficient update, initialize the current acquisition number c=0,c=1,2,…,C;Set the count value J in the cumulative count module of the signal effective frequency band tm [n] = 0; Set the initial output value of the signal effective frequency band counting module to J Rm [n] = 1; Set the energy threshold δ in frequency band identification e 、The number of times the signal is judged to exist is c e , the extended range threshold δ for solving the frequency domain filter coefficients f ; (2.2) Input the signal at the input end of the multi-subband frequency alternating acquisition system, and then collect it through each collector in the system. The acquisition signal of the mth collector in the cth acquisition cycle is recorded as (2.3) Each channel of collected signal The sub-band is processed by the digital processing part, and the processing process is as follows: (2.3.1) Obtain the filter coefficients of the frequency domain filter in the next acquisition in the lower branch of each subband (2.3.1.1) Collecting signals First, follow steps (1.3.1)-(1.3.2) to obtain the amplitude-frequency response of each signal. (2.3.1.2) The amplitude-frequency response Sent to the signal recognition module, the frequency energy is first obtained through the frequency energy acquisition module (2.3.1.3) The frequency energy The signal is input to the signal frequency band identification module, which reads the background noise energy feature E in the noise energy storage module. Nm , and then determine the location of the collected signal and the result for: Among them, δ e is the energy threshold in frequency band identification; (2.3.1.4) The judgment result Input to the signal effective frequency band counting module for accumulation; (2.3.1.4.1) When c≤C-1, firstly, the signal effective frequency band counting module is accumulated and summed, and the result is (2.3.1.4.2), after summing, update the output J Rm [n]: When c<C-1, output J Rm [n] remains unchanged; when c = C-1, the sum result J tm [n] The number of times the signal exists e Compare and make the module output J Rm [n] is: (2.3.1.4.3) Update the loop count number after updating the output: when c<C-1, let c=c+1; when c=C-1, let c=0; (2.3.1.5), the frequency domain filter coefficient solution module is based on J Rm [n] Design the filter coefficients of the frequency domain filter for the next acquisition (2.3.2) acquiring the collected data with improved collection accuracy in the upper branch of each sub-band; (2.3.2.1) and collect the signal Filtering is performed through a frequency domain filter to obtain the filtered output (2.3.2.2) After filtering is completed, the filter coefficients output by the frequency domain filter receiving signal recognition module are and update it; (2.3.2.3), in the first subband, the output after filtering After being processed by the interpolator, the output y1[i] of the first subband is obtained; In the 2nd to Mth subbands, the filtered output After being processed by the interpolator and up-converter in turn, the output y of each sub-band is obtained m [i], m = 2, 3, ..., M; (2.3.2.4), all sub-band output y m [i] are superimposed to obtain the system output y[i]; 2. The method for improving acquisition accuracy of a multi-subband frequency alternating acquisition system according to claim 1, characterized in that: The frequency domain filter coefficient design module designs the filter coefficient H m [n] Methods include brick wall, half-staircase and ladder forms; Assume that the signal in the mth subband has a total of S frequency bands, and the range of the sth frequency band is n sL to n sR , n sL <n sR , s=1,2,…,S; The specific process of the brick wall design method is: If the judgment result J Rm [n] in n sL to n sR If the interval is 1, then H m [n] in max(1, n sL -δ f ) to min(n sR +δ f , N) is 1; The specific process of the half-ladder design method is: If the judgment result J Rm [n] in n sL to n sR If the interval is 1, then H m [n] from n sL Start to decrease linearly to max(1, n sL -δ f ) is 0, and from n sR Start to decrease linearly to min(n sR +δ f , N) is 0. If the values ​​of two adjacent frequency bands overlap, the larger value at the same point is selected as the final value H m [n]; The specific process of the ladder form design method is: If the judgment result J Rm [n] in n sL to n sR If the interval is 1, then: When s = 1, H m [n] from n 1L It starts to decrease linearly and reaches 0 at n=1. 1R Start to decrease linearly to n 1R +floor((n 2L -n 1R ) / 2) is 0, where floor() means rounding down; When s=2,3,…,S-1, H m [n] from n sL Start to decrease linearly to n (s-1)R +floor((n sL -n (s-1)R ) / 2)+1 is 0, and from n sR Start to decrease linearly to n sR +floor((n (s+1)L -n sR ) / 2) is 0; When s=S, H m [n] from n sL Start to decrease linearly to n (S-1)R +floor((n SL -n (S-1)R ) / 2)+1 is 0, and from n SR It starts to decrease linearly and reaches 0 at N / 2.

3. The method for improving acquisition accuracy of a multi-subband frequency alternating acquisition system according to claim 1, characterized in that: The digital processing part includes M sub-bands, numbered 1, 2, ..., m, ..., M; The structures of the 2nd to Mth sub-bands are the same, including upper and lower branches, the upper branch includes a frequency domain filter, an interpolator and an up-converter connected in series in sequence; the lower branch includes an FFT transformation module, an amplitude-frequency response module and a signal recognition module; The upper and lower branches simultaneously receive the acquisition signal of the collector. In the upper branch, the acquisition signal is first filtered by the frequency domain filter, then interpolated by the interpolator, and finally up-converted by the up-converter as the output of the sub-band; in the lower branch, the acquisition signal is first FFT transformed, then the amplitude-frequency response is obtained by the amplitude-frequency response module, and finally the filter coefficient is obtained by the signal recognition module and fed back to the frequency domain filter as the frequency domain filter coefficient for the next acquisition; The structure of the first sub-band is basically the same as that of the remaining M-1 sub-bands, the only difference being that the up-converter is removed from the upper branch and the output of the interpolator is used as the output of the first sub-band.

4. The method for improving acquisition accuracy of a multi-subband frequency alternating acquisition system according to claim 3, characterized in that: The signal identification module includes two signal processing branches, the first signal processing branch is used for initializing the multi-sub-band frequency alternating acquisition system, and the second signal processing branch is used for the multi-sub-band frequency alternating acquisition system to acquire input signals; When the multi-subband frequency alternating acquisition system is initialized, the amplitude-frequency response is calculated through the frequency energy obtaining module to calculate the frequency energy, and then the frequency energy is obtained through the noise energy obtaining module to obtain the background noise energy characteristics and store them in the noise energy storage module; When the multi-subband frequency alternating acquisition system collects input signals, the amplitude-frequency response first calculates the frequency energy through the frequency energy acquisition module, and then inputs it into the signal frequency band identification module by reading the background noise energy characteristics, and then obtains the filter coefficient through the signal effective frequency band cumulative counting module and the frequency domain filter coefficient design module in turn.

Images