Unknown signal recognition method and system in complex environment

Abstract

The application provides a kind of unknown signal identification method and system under complex environment, comprising: step 1: by least mean square adaptive filtering, the interference spectrum of complex environment background noise is obtained, the interference spectrum of complex environment background noise is realized tracking;Step 2: by the interference spectrum of measured environmental noise, unknown signal is spectrum shifted, so that unknown signal spectrum is in the frequency band lower than the amplitude of interference spectrum of background noise;Step 3: after unknown signal after spectrum shift passes through complex environment, after measurement and narrowband filtering, spectrum subtraction is used to obtain noise reduction signal, unknown signal identification under unknown complex environment is realized;Step 4: repeat step 1 to step 3, real-time unknown signal adaptive spectrum shift and identification under complex environment are realized.The application can realize adaptive spectrum shift of unknown signal according to the interference spectrum determined, and real-time interference suppression of background noise is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of signal sensing, in particular to an unknown signal identification method and system in a complex environment. BACKGROUND

[0002] In the field of sensing, signal-to-noise ratio is one of the important factors affecting the performance of sensing. In the complex environment, it is of great significance to realize the identification or interference suppression of unknown signals under the condition of extremely low signal-to-noise ratio and slowly varying non-stationary background noise. In the complex environment, the environmental noise is mostly additive wideband noise, and changes with the environment and time.

[0003] Patent document CN112528774A (application number: CN202011359966.6) discloses an unknown radar signal intelligent sorting system and method in a complex electromagnetic environment, which forms a closed loop of radar signal sorting process by designing radar signal intra-pulse information sorting module, radar signal time-frequency aliasing separation module, radar signal inter-pulse information sorting module and radar behavior cognition module.

[0004] Most of the existing methods need some characteristics or statistical properties of the environmental noise. It is very difficult to separate the signal from the noisy signal without prior knowledge of the environmental noise. SUMMARY

[0005] In view of the defects in the prior art, the purpose of the present application is to provide an unknown signal identification method and system in a complex environment.

[0006] The unknown signal identification method in a complex environment provided by the present application comprises:

[0007] Step 1: obtain the interference spectrum of the complex environment background noise by least mean square adaptive filtering, and realize the tracking of the interference spectrum of the complex environment background noise;

[0008] Step 2: shift the spectrum of the unknown signal by using the measured interference spectrum of the environmental noise, so that the unknown signal spectrum is located in the frequency band with lower amplitude of the interference spectrum of the background noise;

[0009] Step 3: after the spectrum-shifted unknown signal passes through the complex environment, after measurement and narrowband filtering, the noise reduction signal is obtained by using the spectrum subtraction method, and the unknown signal identification in the unknown complex environment is realized;

[0010] Step 4: repeat steps 1 to 3 to realize the adaptive spectrum shifting and identification of the unknown signal in the complex environment in real time.

[0011] Preferably, the step 1 comprises:

[0012] Step 1.1: Select a single frequency sinusoidal signal, measure the value at each time after passing through the complex environment;

[0013] Step 1.2: Process the measured value by using the least mean square adaptive filter to obtain the real-time tap coefficient at different time, and then obtain the interference spectrum of the complex environment background noise.

[0014] Preferably, the step 1.2 comprises:

[0015] Step 1.2.1: Set the tap number M and step size μ of the least mean square error filter, and set the tap weight vector to 0, wherein, represents a vector, represents transposition, and n is the number of signals;

[0016] Step 1.2.2: Take the estimate value of the time-varying parameter as the input signal vector Take the ideal output of the known signal as the expected signal d[n], and the output signal is represented as is a random variable;

[0017] Step 1.2.3: The error signal is the difference between the expected signal and the output signal e[n] = d[n] - y[n], and the threshold value δ of E{e 2 [n]} is set as the convergence evaluation criterion, and the tap weight vector

[0018] Step 1.2.4: Set the least mean square error adaptive filter to start converging at N1 time and end converging at N2 time, keep the tap weight vector estimate at N1 time, and calculate the output noise vector wherein M+1 < n < N1, perform windowed Fourier transform to obtain the noise spectrum at M+1 ~ N2 time, and repeat the above steps starting from N2+1 time data to obtain the real-time background noise interference spectrum.

[0019] Preferably, the step 2 comprises:

[0020] Step 2.1: According to the noise spectrum at the current time, determine the frequency band f1 ~ f2 with low noise spectrum amplitude;

[0021] Step 2.2: Considering that the prior information of the unknown signal spectrum range is known, i.e. s(t) may be located in the frequency band f L ~ f H , move the unknown signal to the frequency band f m(t) = s(t) x A cos(2πΔft), so that it is located in the above target frequency band f1~f2, i.e. f L +Δf~f H +Δf is located in f1~f2, wherein A and Δf are the amplitude and frequency of the single-frequency cosine signal used for frequency shifting, respectively.

[0022] Preferably, the step 3 comprises:

[0023] Step 3.1: measuring the unknown signal after frequency shifting to obtain the value at each time;

[0024] Step 3.2: band-pass filtering the measurement result to obtain the signal and the narrow-band interference with low amplitude, and performing windowed Fourier transform to obtain the frequency spectrum R(f);

[0025] Step 3.3: subtracting the interference spectrum value N(f) of the background noise in the corresponding time period by using the spectrum subtraction method to obtain the frequency spectrum S(f) of the to-be-detected signal, wherein S(f) = R(f)-N(f), obtaining the estimated value of the to-be-detected unknown signal, and realizing the identification of the unknown signal in the corresponding time period through inverse windowed Fourier transform.

[0026] The unknown signal identification system in a complex environment provided by the application comprises:

[0027] Module M1: obtaining the interference spectrum of the background noise in the complex environment by using the least mean square adaptive filtering, and realizing the tracking of the interference spectrum of the background noise in the complex environment;

[0028] Module M2: performing the frequency spectrum shifting on the unknown signal by using the measured interference spectrum of the environmental noise, so that the frequency spectrum of the unknown signal is located in the frequency band with low amplitude of the interference spectrum of the background noise;

[0029] Module M3: after the unknown signal after the frequency spectrum shifting passes through the complex environment, the noise reduction signal is obtained by using the spectrum subtraction method after the measurement and the narrow-band filtering, and the identification of the unknown signal in the complex environment is realized;

[0030] Module M4: repeatedly calling the modules M1 to M3 to realize the adaptive frequency spectrum shifting and identification of the unknown signal in the complex environment in real time.

[0031] Preferably, the module M1 comprises:

[0032] Module M1.1: selecting a single-frequency sinusoidal signal, measuring the signal after passing through the complex environment to obtain the measurement value at each time;

[0033] Module M1.2: processing the measurement value by using the least mean square adaptive filtering to obtain the real-time tap coefficient at different times, and then obtaining the interference spectrum of the background noise in the complex environment.

[0034] Preferably, the module M1.2 comprises:

[0035] Module M1.2.1: Set the tap number M and step size μ of the minimum mean square error filter, and set the tap weight vector to 0, where, represents the vector, represents the transpose, and n is the number of signals;

[0036] Module M1.2.2: Take the estimated value of the time-varying parameter as the input signal vector Take the ideal output of the known signal as the expected signal d[n], and the output signal is represented as is a random variable;

[0037] Module M1.2.3: The error signal is the difference between the expected signal and the output signal e[n] = d[n] - y[n], and the threshold value δ of E{e 2 [n]} is set as the evaluation criterion for convergence, and the tap weight vector

[0038] Module M1.2.4: Set the minimum mean square error adaptive filter to start converging at time N1 and end converging at time N2, retain the tap weight vector estimate at time N1 , and calculate the output noise vector where M+1 < n < N1, perform windowed Fourier transform to obtain the noise spectrum at time M+1 ~ N2, and start repeating the above modules for data at time N2+1 to continuously obtain the real-time interference spectrum of the background noise.

[0039] Preferably, the module M2 comprises:

[0040] Module M2.1: According to the noise spectrum at the current time, determine the frequency band f1 ~ f2 with low noise spectrum amplitude;

[0041] Module M2.2: Considering that the prior information of the frequency spectrum range of the unknown signal is known, i.e., s(t) can be located within the frequency band f L ~ f H , perform frequency shift on the unknown signal s m (t) = s(t) × A cos(2πΔft) to make it located within the above target frequency band f1 ~ f2, i.e., f L + Δf ~ f H + Δf is located within f1 ~ f2, where A and Δf are the amplitude and frequency of the single-frequency cosine signal used for frequency shift, respectively.

[0042] Preferably, the module M3 comprises:

[0043] Module M3.1: measuring the unknown signal after spectrum shift, obtaining the value at each time;

[0044] Module M3.2: band-pass filtering the measurement result, obtaining the signal and low-amplitude narrow-band interference, and performing window Fourier transform, obtaining the spectrum R(f);

[0045] Module M3.3: subtracting the interference spectrum value N(f) of the background noise in the corresponding time period by using spectrum subtraction, obtaining the spectrum S(f) of the signal to be measured, wherein S(f)=R(f)-N(f), obtaining the estimated value of the unknown signal to be measured, and realizing the identification of the unknown signal in the corresponding time period through window inverse Fourier transform.

[0046] Compared with the prior art, the present application has the following beneficial effects:

[0047] (1) The present application has reasonable structure and is convenient to use, can overcome the defects of the prior art, can realize the real-time identification of unknown signals under the conditions of extremely low signal-to-interference ratio and slowly-varying non-stationary background noise, and conforms to the development trend of the sensing field;

[0048] (2) The present application increases the window function for the signal to be measured through the duration of the tap coefficient of the least mean square adaptive filter, realizes the function of time-frequency analysis on the basis of clearly obtaining the frequency components contained in the signal;

[0049] (3) The present application can realize the adaptive spectrum shift of unknown signals according to the measured interference spectrum, and realize the interference suppression of the background noise in real time. BRIEF DESCRIPTION OF DRAWINGS

[0050] Other features, objects and advantages of the present application will become more apparent from the following detailed description of non-limiting embodiments with reference to the attached drawings:

[0051] Figure 1 Fig. 1 is a schematic diagram of the principle of the present application;

[0052] Figure 2 Fig. 2 is a flow chart of the determination of the interference spectrum based on the least mean square adaptive filter in the present application. DETAILED DESCRIPTION

[0053] The present application will be described in detail below with reference to specific embodiments. The following embodiments will help those skilled in the art to further understand the present application, but do not limit the present application in any form. It should be pointed out that those skilled in the art can make several changes and improvements without departing from the concept of the present application. These all belong to the protection scope of the present application.

[0054] Embodiment:

[0055] This invention employs least mean square adaptive filtering to track and obtain the interference spectrum of an unknown and complex environment in real time. Then, it adaptively shifts the spectrum of the unknown signal so that it is located in the spectral band with a lower background noise amplitude. Finally, it uses spectral subtraction to achieve interference suppression and identification of the unknown signal.

[0056] like Figure 1 As shown, the method for identifying unknown signals under conditions of extremely low signal-to-interference ratio and slowly varying non-steady-state background noise provided by the present invention includes:

[0057] Step 1: Tracking the interference spectrum of background noise in complex environments. The interference spectrum of background noise in complex environments is obtained through least mean square adaptive filtering, thus achieving the tracking of the interference spectrum of background noise in complex environments.

[0058] Step 2: Adaptive Spectrum Shifting Process for Unknown Signals. Using the measured interference spectrum of the ambient noise, the spectrum of the unknown signal is shifted so that its frequency band falls within the lower amplitude range of the background noise interference spectrum.

[0059] Step 3: Identification of Unknown Signals. After the spectrum-shifted unknown signal passes through a complex environment, it undergoes measurement and narrowband filtering. Then, spectral subtraction is used to obtain a denoised signal, thus achieving the identification of unknown signals in complex environments.

[0060] Step 4: Repeat steps 1-3 to achieve real-time adaptive spectrum shifting and identification of unknown signals in complex environments.

[0061] Step 1 includes the following steps:

[0062] Step 1.1: Measurement Procedure. A single-frequency sinusoidal signal is selected and measured after passing through a complex environment to obtain the measurement values ​​at various times.

[0063] Step 1.2: Adaptive Filtering Step. The measured values ​​are processed using least mean square adaptive filtering to obtain the real-time tap coefficients at different times, thereby obtaining the interference spectrum of the background noise in the complex environment.

[0064] like Figure 2 As shown, step 1.2 includes the following steps:

[0065] Step 1.2.1: Parameter Setting Sub-step. Set the number of taps M and step size μ of the minimum mean square error filter. Also, set the tap weight vector... Set to 0. Wherein, Represents a vector. This represents transposition.

[0066] Step 1.2.2: Filtering sub-step. The estimated values ​​of the time-varying parameters are used as the input signal vector. The ideal output of the known signal is taken as the desired signal d[n], and the output signal can be expressed as

[0067] Step 1.2.3: Tap weight vector adaptive sub-step. The error signal is the difference between the desired signal and the output signal e[n] = d[n] - y[n], and the least mean square error criterion is adopted. A threshold value δ of E{e 2 [n]} is set as the criterion for judging whether convergence is achieved, and the tap weight vector is updated in real time

[0068] Step 1.2.4: Interference spectrum calculation sub-step of background noise. It is assumed that the least mean square error adaptive filter starts converging at time N1 and ends converging at time N2. The tap weight vector estimate at time N1 is retained and the output noise vector is calculated where M+1 < n < N1, and the window Fourier transform is performed, so that the noise spectrum at time M+1 to N2 can be obtained. The above steps are repeated starting from the data at time N2+1, so that the real-time interference spectrum of the background noise is continuously obtained.

[0069] The step 2 comprises the following steps:

[0070] Step 2.1: Target spectrum band judgment. In the vicinity of two groups of sensing elements, it can be considered that the overall distribution of the noise is approximately the same. According to the noise spectrum at the current time, the frequency band f1 to f2 with a lower noise spectrum amplitude is determined.

[0071] Step 2.2: Considering that the prior information of the frequency spectrum range of the unknown signal is known, i.e., s(t) can be located in the frequency band f L ~ f H , the unknown signal is frequency-shifted s m (t) = s(t) x A cos(2πΔft) to be located in the above target frequency band f1 to f2 (f2-f1 > f H -f L ), i.e., f L +Δf ~ f H +Δf is located in f1 to f2. Wherein, A and Δf are the amplitude and frequency of the single-frequency cosine signal used for frequency shifting, respectively.

[0072] The step 3 comprises the following steps:

[0073] Step 3.1: Measurement step. The frequency-shifted unknown signal is measured to obtain the value at each time.

[0074] Step 3.2: Filtering step. The measurement result is band-pass filtered to obtain the signal and the narrow-band interference with a lower amplitude, and the window Fourier transform is performed to obtain the frequency spectrum R(f).

[0075] Step 3.3: spectrum subtraction step. The spectrum of the unknown signal S(f) is obtained by subtracting the interference spectrum N(f) of the background noise of the corresponding time period, that is, S(f) = R(f) - N(f). The estimated value of the unknown signal is obtained. The unknown signal of the corresponding time period is identified by inverse windowed Fourier transform.

[0076] The unknown signal identification system in a complex environment according to the present application comprises: module M1: obtaining the interference spectrum of the background noise in the complex environment by minimum mean square adaptive filtering, to realize tracking of the interference spectrum of the background noise in the complex environment; module M2: performing spectrum shift on the unknown signal by the measured interference spectrum of the environmental noise, to make the spectrum of the unknown signal in a frequency band lower than the amplitude of the interference spectrum of the background noise; module M3: after the spectrum-shifted unknown signal passes through the complex environment and is measured and filtered in a narrow band, a noise-reduced signal is obtained by spectrum subtraction, to realize identification of the unknown signal in the unknown complex environment; and module M4: repeatedly calling modules M1 to M3, to realize adaptive spectrum shift and identification of the unknown signal in the complex environment in real time.

[0077] The module M1 comprises: module M1.1: selecting a single-frequency sinusoidal signal, measuring the signal after passing through the complex environment, and obtaining the measurement value at each time; module M1.2: processing the measurement value by minimum mean square adaptive filtering, to obtain real-time tap coefficients at different times, and then obtain the interference spectrum of the background noise in the complex environment.

[0078] The module M1.2 comprises: module M1.2.1: setting the tap number M and step size μ of the minimum mean square error filter, and setting the tap weight vector to 0, wherein, represents a vector, represents transposition, and n is the number of signals; module M1.2.2: taking the estimate value of the time-varying parameter as an input signal vector Taking the ideal output of the known signal as an expected signal d[n], the output signal is represented as , which is a random variable; module M1.2.3: the error signal is the difference e[n] = d[n] - y[n] between the expected signal and the output signal, and the threshold value δ of E{e 2 [n]} is set as a judgment standard for convergence, and the tap weight vector is updated in real time. Module M1.2.4: setting the minimum mean square error adaptive filter to start converging at N1 time and end converging at N2 time, retaining the tap weight vector estimate at N1 time, and calculating the output noise vector Wherein, M+1<n<N1, the window Fourier transform is carried out, the noise spectrum of M+1~N2 time is obtained, the above module is repeatedly called starting from N2+1 time data, and real-time background noise interference spectrum is continuously obtained.

[0079] The module M2 comprises: module M2.1: judging the frequency band f1~f2 with lower noise spectrum amplitude according to the noise spectrum of the current time; module M2.2: considering that the prior information of the unknown signal spectrum range is known, namely, s(t) can be located in the frequency band f L ~f H , the unknown signal is frequency shifted s m (t) = s(t) * A * cos(2 * pi * Delta * f * t), so as to be located in the above target frequency band f1~f2, namely, f L + Delta * f ~ f H + Delta * f is located in f1~f2, wherein, A and Delta * f are the amplitude and frequency of the single-frequency cosine signal used for frequency shifting, respectively.

[0080] The module M3 comprises: module M3.1: measuring the unknown signal after frequency shifting to obtain the value at each time; module M3.2: carrying out band-pass filtering on the measurement result to obtain the signal and the narrow-band interference with lower amplitude, and carrying out window Fourier transform to obtain the spectrum R(f); module M3.3: subtracting the interference spectrum value N(f) of the background noise in the corresponding time period by using the spectrum subtraction method to obtain the spectrum S(f) of the signal to be measured, wherein, S(f) = R(f)-N(f), the estimated value of the unknown signal to be measured is obtained, and the unknown signal in the corresponding time period is identified through the inverse window Fourier transform.

[0081] Those skilled in the art know that, in addition to implementing the system, device and each module thereof provided by the present application in the form of pure computer readable program code, the same program can also be realized by logically programming the method steps in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers and embedded microcontrollers. Therefore, the system, device and each module thereof provided by the present application can be considered as a hardware component, and the modules included therein for realizing various programs can also be considered as structures in the hardware component; the modules for realizing various functions can also be considered as both software programs for realizing methods and structures in the hardware component.

[0082] The specific embodiments of the present application are described above. It should be understood that the present application is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essential content of the present application. The embodiments of the present application and the features in the embodiments can be arbitrarily combined with each other without conflict.

Claims

1. A method for unknown signal identification in complex environments, characterized in that, The method comprises the following steps: Step 1: obtaining the interference spectrum of the complex environmental background noise by least mean square adaptive filtering, and realizing tracking of the interference spectrum of the complex environmental background noise; Step 2: performing spectrum shift on the unknown signal by using the measured interference spectrum of the environmental noise, so that the spectrum of the unknown signal is located in a frequency band with a lower amplitude of the interference spectrum of the background noise; The step 2 comprises: Step 2.1: judging the frequency band f1-f2 with a lower noise spectrum amplitude according to the noise spectrum at the current time; Step 2.2: considering the prior information of the unknown signal's spectral range is known, i.e. s(t) can be located in the frequency band f L ~ f H , the unknown signal is frequency shifted s m (t) = s(t) x A cos(2pAf t) to be located in the above target frequency band f1~f2, i.e. f L + Af ~ f H + Af is located in f1~f2, where A and Af are the amplitude and frequency of the single-frequency cosine signal used for frequency shifting, respectively. Step 3: after the spectrum-shifted unknown signal passes through the complex environment and is measured and narrow-band filtered, a noise-reduced signal is obtained by using spectrum subtraction, and the unknown signal in the unknown complex environment is identified; Step 4: steps 1 to 3 are repeatedly executed, and adaptive spectrum shift and identification of the unknown signal in the complex environment are realized in real time.

2. The method of claim 1, wherein, The step 1 comprises: Step 1.1: selecting a single-frequency sinusoidal signal, measuring the signal after passing through the complex environment, and obtaining the measurement value at each time; Step 1.2: processing the measurement value by using least mean square adaptive filtering, obtaining the real-time tap coefficient at different times, and then obtaining the interference spectrum of the complex environmental background noise.

3. The method of claim 2, wherein, The step 1.2 comprises: Step 1.2.1: Set the number of taps M and the step size μ of the MMSE filter and set the tap weight vector to zero, where denotes the vector T represents the transpose and n is the number of signals; Step 1.2.2: The estimate of the time-varying parameter is used as an input signal vector The output signal is denoted as is a random variable; Step 1.2.3: The error signal is the difference between the desired signal and the output signal e[n] = d[n] - y[n], and the threshold value δ of E{e 2 [n]} is set as the criterion of whether to converge or not, and the tap weight vector is updated in real time Step 1.2.4: Assuming the minimum mean square error adaptive filter converges at time N1 and ends at time N2, the tap weight vector estimate at time N1 is retained and the output noise vector is calculated where M+1 < n < N1, the windowed Fourier transform is performed to obtain the noise spectrum at time M+1 ~ N2, and the above steps are repeated for data at time N2+1 to continuously obtain the real-time interference spectrum of the background noise.

4. The method of claim 1, wherein, The step 3 comprises: Step 3.1: measuring the spectrum-shifted unknown signal to obtain the value at each time; Step 3.2: performing band-pass filtering on the measurement result to obtain the signal and the narrow-band interference with a lower amplitude, and performing windowed Fourier transform to obtain the spectrum R(f); Step 3.3: subtracting the interference spectrum value N(f) of the background noise corresponding to the time period by using spectrum subtraction to obtain the spectrum S(f) of the signal to be measured, wherein S(f)=R(f)-N(f), obtaining the estimated value of the unknown signal to be measured, and realizing identification of the unknown signal corresponding to the time period by inverse windowed Fourier transform.

5. An unknown signal recognition system in a complex environment, characterized by, The method comprises the following steps: Module M1: obtaining the interference spectrum of the complex environmental background noise by least mean square adaptive filtering, and realizing tracking of the interference spectrum of the complex environmental background noise; Module M2: performing spectrum shift on the unknown signal by using the measured interference spectrum of the environmental noise, so that the spectrum of the unknown signal is located in a frequency band with a lower amplitude of the interference spectrum of the background noise; The module M2 comprises: Module M2.1: judging the frequency band f1-f2 with a lower noise spectrum amplitude according to the noise spectrum at the current time; Module M2.2: considering the prior information known that the spectrum range of the unknown signal is known, i.e. s(t) can be located in the frequency band f L ~ f H Within the above target frequency band f1~f2, the unknown signal is frequency shifted s m (t) = s(t) x A cos(2pAf t) to be located in the above target frequency band f1~f2, i.e. f L + Af ~ f H + Af is located in f1~f2, wherein A and Af are the amplitude and frequency of the single-frequency cosine signal used for frequency shifting, respectively. Module M3: after the spectrum-shifted unknown signal passes through the complex environment and is measured and narrow-band filtered, a noise-reduced signal is obtained by using spectrum subtraction, and the unknown signal in the unknown complex environment is identified; Module M4: repeatedly calling modules M1 to M3, and realizing adaptive spectrum shift and identification of the unknown signal in the complex environment in real time.

6. The unknown signal recognition system in complex environments according to claim 5, wherein, The module M1 comprises: Module M1.1: selecting a single-frequency sinusoidal signal, measuring the signal after passing through the complex environment, and obtaining the measurement value at each time; Module M1.2: processing the measurement value by using least mean square adaptive filtering, obtaining the real-time tap coefficient at different times, and then obtaining the interference spectrum of the complex environmental background noise.

7. The unknown signal recognition system in complex environments according to claim 6, characterized in that, The module M1.2 comprises: Module M1.2.1 : Set the tap number M and the step size μ of the minimum mean square error filter and set the tap weight vector to zero, where denotes the vector, T denotes the transpose and n is the number of signals. Module M1.2.2: The estimated values of the time-varying parameters are taken as input signal vector The output signal is denoted as is a random variable; Module M1.2.3: The error signal is the difference e[n] = d[n] - y[n] between the desired signal and the output signal. Using the least mean square error criterion, the threshold value δ of E{e 2 [n]} is set as the criterion for judging whether to converge or not, and the tap weight vector is updated in real time Module M1.2.4: assuming that the minimum mean square error adaptive filter converges at time N1 and ends at time N2, retaining the tap weight vector estimate at time N1 and calculating the output noise vector where M+1 < n < N1, performing a windowed Fourier transform to obtain the noise spectrum at times M+1 to N2, and repeating the above module for data at time N2+1 and onwards to obtain the real-time interference spectrum of the background noise.

8. The unknown signal recognition system in complex environments according to claim 5, wherein, The module M3 comprises: Module M3.1: measuring the spectrum-shifted unknown signal to obtain the value at each time; Module M3.2: band-pass filtering the measurement results to obtain a signal and a narrow-band interference with a low amplitude, and performing a windowed Fourier transform to obtain a frequency spectrum R(f); Module M3.3: subtracting an interference spectrum value N(f) of the background noise corresponding to the time period by using a spectrum subtraction method to obtain a frequency spectrum S(f) of the signal to be measured, wherein S(f)=R(f)-N(f), obtaining an estimated value of the unknown signal to be measured, and performing an inverse windowed Fourier transform to identify the unknown signal corresponding to the time period.

Images