Non-stationary signal processing method and device, electronic equipment and storage medium

Abstract

According to the non-stationary signal processing method and device, the electronic equipment and the storage medium, the non-stationary signal is accurately captured through the signal collector and converted into the discretized signal, and a high-quality data basis is provided for subsequent processing. And then the fractional Fourier transform algorithm is combined to effectively improve the resolution of the signal in the time-frequency domain and enhance the stability of the signal. And further mining the signal features by using an embedded track matrix algorithm, and constructing a track matrix reflecting the internal structure of the signal. Through singular value decomposition, effective extraction and dimensionality reduction of signal features are realized, and noise interference is effectively suppressed. And finally, based on processing of a singular value matrix algorithm, a processing result corresponding to the non-stationary signal is accurately determined, the accuracy and stability of non-stationary signal processing are remarkably improved, and the application effect of non-stationary signal processing is guaranteed.

Description

Technical Field

[0001] The present application relates to the field of signal processing technology, and in particular to a non-stationary signal processing method, device, electronic device and storage medium. Background Art

[0002] The statistical characteristics of non-stationary signals, such as frequency and amplitude, will change over time. This time-varying characteristic often contains important information. By processing non-stationary signals, these time-varying characteristics hidden in the signal can be revealed, thereby extracting useful information. For example, in a communication system, the received signal may be affected by various interferences and noises, which are often non-stationary. By processing these non-stationary signals, useful information can be extracted to improve the performance and reliability of the system. Therefore, the processing of non-stationary signals is crucial.

[0003] Based on the above situation, the non-stationary signal processing method in the prior art has poor performance when processing non-stationary signals, thereby affecting the application effect of non-stationary signal processing. Summary of the invention

[0004] In view of this, the purpose of the present application is to propose a non-stationary signal processing method, device, electronic device and storage medium to solve the above-mentioned technical problems.

[0005] Based on the above purpose, the present application provides a non-stationary signal processing method, comprising:

[0006] Using a signal collector to collect a non-stationary signal, and determining a first discretized signal corresponding to the non-stationary signal, wherein the non-stationary signal includes a radar signal, a voice signal, an electrocardiogram signal, an image signal, and a temperature signal;

[0007] Processing the first discretized signal using a fractional Fourier transform algorithm to obtain a second discretized signal;

[0008] Processing the second discretized signal by an embedding trajectory matrix algorithm to obtain an embedding trajectory matrix;

[0009] Performing singular value decomposition on the embedded trajectory matrix to obtain a decomposition matrix, and determining a singular value matrix, a first singular vector matrix, and a second singular vector matrix from the decomposition matrix;

[0010] Processing the singular value matrix, the first singular vector matrix and the second singular vector matrix by a singular value matrix algorithm to obtain a matrix corresponding to the singular value;

[0011] A processing result corresponding to the non-stationary signal is determined based on the matrix corresponding to the singular values.

[0012] Optionally, the processing the first discretized signal by using a fractional Fourier transform algorithm to obtain a second discretized signal includes:

[0013] Using a digital signal processor to receive set fractional Fourier transform parameters;

[0014] Based on the first discretized signal and the translation operator of the fractional Fourier transform under the fractional Fourier transform parameter, the second discretized signal is obtained by the following formula:

[0015]

[0016] Among them, x α (n) represents the second discretized signal, x(n) represents the first discretized signal, α represents the fractional Fourier transform parameter, j represents the imaginary unit, n represents the index of the sequence, Represents the translation operator of the fractional Fourier transform.

[0017] Optionally, the processing the second discretized signal by using an embedded trajectory matrix algorithm to obtain an embedded trajectory matrix includes:

[0018] The second discretized signal is embedded into the preset trajectory matrix by the following formula to obtain the embedded trajectory matrix:

[0019]

[0020] Among them, H α represents the embedded trajectory matrix, K represents the number of rows of the preset trajectory matrix, and P represents the number of columns of the preset trajectory matrix.

[0021] Optionally, performing singular value decomposition on the embedded trajectory matrix to obtain a decomposition matrix includes:

[0022] Based on the embedded trajectory matrix, the decomposition matrix is ​​obtained by the following formula:

[0023]

[0024] Among them, H α represents the embedding trajectory matrix, represents the decomposition matrix, U represents the first singular vector matrix, Denotes the embedding trajectory matrix H α The conjugate transposed matrix of represents the second singular vector matrix, Second singular vector matrix is the co-transposed matrix of the matrix V, Λ represents the singular value matrix, and the elements in the singular value matrix Λ are The square roots of the eigenvalues ​​of .

[0025] Optionally, the processing based on the singular value matrix, the first singular vector matrix and the second singular vector matrix by a singular value matrix algorithm to obtain a matrix corresponding to the singular values ​​includes:

[0026] Based on the singular value matrix, the first singular vector matrix and the second singular vector matrix, a matrix corresponding to the singular values ​​is obtained by the following formula:

[0027]

[0028] in, represents the matrix corresponding to the singular values, u l represents the lth first singular column vector in the first singular vector matrix, represents the lth second singular column vector in the second singular vector matrix, λ l Represents the lth singular value in the singular value matrix.

[0029] Optionally, determining a processing result corresponding to the non-stationary signal based on a matrix corresponding to the singular value includes:

[0030] Based on the anti-diagonal elements in the matrix corresponding to the singular values, the following formula is used to perform averaging to obtain the first time domain time series:

[0031]

[0032] in, represents the first time domain time series, represents the matrix corresponding to the singular value, n represents the index of the reconstructed sequence, k represents the index of the column of the matrix corresponding to the singular value, K represents the number of rows of the preset trajectory matrix, and P represents the number of columns of the preset trajectory matrix;

[0033] Based on the first time-domain time series and the preset fractional Fourier transform parameters, the demodulation process is performed by the following formula to obtain the second time-domain time series:

[0034]

[0035] in, represents the second time domain time series, represents the first time domain time series, α represents the fractional Fourier transform parameter, j represents the imaginary unit, and n represents the index of the sequence;

[0036] Based on the first time domain time series and the second time domain time series, a normalized correlation function value is determined by the following formula:

[0037]

[0038] in, represents the normalized correlation function value, represents the second time domain time series, represents the transposed matrix of the second time domain time series, represents the first time domain time series, represents the transposed matrix of the first time-domain time series;

[0039] Based on the comparison result between the normalized correlation function value and a preset threshold value, the time domain time series is grouped to obtain a plurality of time domain time groups, each time domain time group including a plurality of time domain time signals;

[0040] The multiple time domain time signals in each time domain time group are summed up to obtain the time domain time signal component corresponding to each time domain time group, and the time domain time signal component corresponding to each time domain time group is used as the processing result corresponding to the non-stationary signal.

[0041] Optionally, determining a first discretized signal corresponding to the non-stationary signal includes:

[0042] The non-stationary signal is discretized according to a preset sampling interval to obtain a first discretized signal.

[0043] Based on the same inventive concept, the second aspect of the present application provides a non-stationary signal processing device, comprising:

[0044] A discretized signal determination module is configured to collect a non-stationary signal using a signal collector and determine a first discretized signal corresponding to the non-stationary signal, wherein the non-stationary signal includes a radar signal, a voice signal, an electrocardiogram signal, an image signal, and a temperature signal;

[0045] A fractional Fourier transform module, configured to process the first discretized signal using a fractional Fourier transform algorithm to obtain a second discretized signal;

[0046] an embedding trajectory matrix determination module, configured to process the second discretization signal through an embedding trajectory matrix algorithm to obtain an embedding trajectory matrix;

[0047] A singular value decomposition module is configured to perform singular value decomposition on the embedded trajectory matrix to obtain a decomposition matrix, and determine a singular value matrix, a first singular vector matrix, and a second singular vector matrix from the decomposition matrix;

[0048] A singular value matrix processing module is configured to perform processing based on the singular value matrix, the first singular vector matrix and the second singular vector matrix through a singular value matrix algorithm to obtain a matrix corresponding to the singular value;

[0049] The processing result determination module is configured to determine the processing result corresponding to the non-stationary signal based on the matrix corresponding to the singular value.

[0050] Based on the same inventive concept, the third aspect of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executed by the processor, wherein the processor implements the method described in the first aspect above when executing the computer program.

[0051] Based on the same inventive concept, the fourth aspect of the present application provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable a computer to execute the method described in the first aspect above.

[0052] As can be seen from the above, the non-stationary signal processing method, device, electronic device and storage medium provided by the present application accurately capture non-stationary signals through a signal collector and convert them into discrete signals, providing a high-quality data basis for subsequent processing. Then, the fractional Fourier transform algorithm is combined to effectively improve the resolution of the signal in the time and frequency domain and enhance the stability of the signal. The embedded trajectory matrix algorithm is then used to further explore the signal features and construct a trajectory matrix that reflects the internal structure of the signal. Through singular value decomposition, effective extraction and dimensionality reduction of signal features are achieved, and noise interference is effectively suppressed. Finally, based on the processing of the singular value matrix algorithm, the processing results corresponding to the non-stationary signal are accurately determined, which significantly improves the accuracy and stability of non-stationary signal processing and ensures the application effect of non-stationary signal processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the present application or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0054] Figure 1 This is a flowchart of a non-stationary signal processing method according to an embodiment of the present application;

[0055] Figure 2 A schematic diagram of a non-stationary signal processing process according to an embodiment of the present application;

[0056] Figure 3 A first schematic diagram of a signal corresponding to a processing result obtained by non-stationary signal processing in an embodiment of the present application;

[0057] Figure 4 A second schematic diagram of a signal corresponding to a processing result obtained by non-stationary signal processing in an embodiment of the present application;

[0058] Figure 5 This is a structural block diagram of a non-stationary signal processing device according to an embodiment of the present application;

[0059] Figure 6 A schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0060] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0061] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present application should be the usual meanings understood by people with ordinary skills in the field to which the present application belongs. The "first", "second" and similar words used in the embodiments of the present application do not represent any order, quantity or importance, but are only used to distinguish different components. "Including" or "comprising" and similar words mean that the elements or objects appearing in front of the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0062] It is understandable that before using the technical solutions of each embodiment of the present application, the type, scope of use, usage scenarios, etc. of the personal information involved will be informed to the user in an appropriate manner, and the user's authorization will be obtained.

[0063] For example, in response to receiving an active request from a user, a prompt message is sent to the user to clearly remind the user that the operation requested to be performed will require obtaining and using the user's personal information. Thus, the user can independently choose whether to provide personal information to the electronic device, application, server, storage medium or other software or hardware that performs the operation of the technical solution of the present application according to the prompt message.

[0064] As an optional but non-limiting implementation, in response to receiving the user's active request, the prompt information may be sent to the user in the form of a pop-up window, in which the prompt information may be presented in text form. In addition, the pop-up window may also carry a selection control for the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0065] It is understandable that the above notification and the process of obtaining user authorization are merely illustrative and do not constitute a limitation on the implementation method of this application. Other methods that meet relevant laws and regulations may also be applied to the implementation method of this application.

[0066] Non-stationary signals are a common type of signal encountered in engineering fields such as communications, radar, sonar, and biomedicine. With the development of science and technology and the improvement of people's requirements for signal processing accuracy, non-stationary signal models have gradually become more complex. For example, in the field of radar target detection, with the increase in target movement speed and the complexity and changeability of the movement state, the radar echo signal presents strong non-stationary characteristics; in the field of communications, the high-speed movement of mobile terminals such as high-speed railways and low-orbit satellites and the near-field effects caused by the expansion of extremely large-scale arrays all make the channel model present non-stationary properties. Therefore, algorithms for non-stationary signal processing are constantly updated and developed, and can be roughly divided into two types: non-adaptive and adaptive.

[0067] Non-adaptive non-stationary signal processing algorithms are represented by time-frequency analysis, among which linear time-frequency analysis methods mainly include short-time Fourier transform, wavelet transform and fractional Fourier transform, etc.; nonlinear time-frequency analysis mainly includes Wigner-Wiley distribution, fuzzy function, etc. Nonlinear time-frequency analysis has higher resolution than linear time-frequency analysis, but the nonlinear time-frequency distribution of multi-component non-stationary signals will produce cross terms and form pseudo-signal components, which makes it difficult to identify signals. In addition, the basis function of time-frequency analysis is fixed and cannot change with the non-stationary form of the signal. Only when the non-stationary type of the signal matches the non-stationary type of the basis function can the ideal processing effect be obtained, but there are many types of non-stationary signals, and it is difficult to predict the type of signal and select the appropriate signal processing algorithm before processing.

[0068] Adaptive non-stationary signal processing algorithms include empirical mode decomposition, variational mode decomposition, singular spectrum analysis, etc. These algorithms do not have fixed basis functions and can adaptively obtain different components of the signal according to the waveform of the acquired signal, which is helpful for applications such as signal denoising and multi-component decomposition. Taking singular spectrum analysis as an example, although it can be used to a certain extent in most non-stationary signal processing, its theory originates from the power spectrum of stationary signals, and its theoretical support in non-stationary signal processing is insufficient, so its performance in common frequency modulation non-stationary signal processing is unstable.

[0069] The purpose of this application is to overcome the above technical deficiencies, combine the theory of non-adaptive signal processing, and improve the performance of adaptive signal processing in non-stationary signal processing. Specifically, a non-stationary signal processing method based on lifting singular spectrum analysis is provided to solve the problems in the related art.

[0070] The problems existing in the prior art are as follows:

[0071] (1) Unstable performance when processing non-stationary signals

[0072] The related theories of singular spectrum analysis have theoretical support in stationary signal processing, and singular spectrum analysis has stable performance in stationary signal processing. Although singular spectrum analysis can also be used in non-stationary signal processing, its performance is unstable, which is reflected in the fact that it cannot perform well in certain types of non-stationary signal processing.

[0073] (2) Poor performance in broadband signal denoising applications

[0074] As the name implies, broadband signals occupy a wide range of bandwidth. According to the corresponding relationship between the Hankel matrix singular values ​​and the signal power spectrum, the broadband signal arrangement trajectory matrix has more singular values ​​with larger values. In the denoising application, the noise covers the entire frequency band, that is, the noise is evenly distributed on the singular values. Therefore, when selecting the time series corresponding to more singular values ​​as the denoised sequence, more noise is also selected, affecting the denoising effect.

[0075] (3) Inflexible in processing non-stationary signals

[0076] The types of non-stationary signals vary greatly, but the basis function pattern of the fractional Fourier transform is fixed, which limits the types of non-stationary signal decomposition. When processing non-stationary signals that do not match the basis function form, the performance needs to be improved.

[0077] This application combines the advantages of singular spectrum analysis and fractional Fourier transform, proposes an improved singular spectrum analysis method for non-stationary signal processes, solves the above technical problems, and achieves the following purposes:

[0078] (1) Improve the stability of singular spectrum analysis when processing non-stationary signals

[0079] Using the theoretical basis of fractional Fourier analysis in processing non-stationary signals, a new trajectory matrix construction method is proposed to build the theoretical basis of a new singular spectrum analysis algorithm to maintain its stability in non-stationary signal processing.

[0080] (2) Improving the performance of singular spectrum analysis in broadband signal denoising

[0081] Broadband signal only means that the signal has a large bandwidth in the frequency domain. The fractional Fourier transform has a free parameter, which can provide a representation of the signal from the time domain to the frequency domain. According to the uncertainty principle of time-frequency analysis, frequency domain broadband signals are often narrowband in the time domain. That is, frequency domain broadband signals may be narrowband in a certain fractional Fourier transform domain. For such signals, the singular value distribution corresponding to its trajectory matrix is ​​relatively concentrated or even sparse. In the process of singular value selection, it only contains very few noise components. Therefore, it has a good performance in denoising.

[0082] (3) Improving the flexibility of fractional Fourier analysis in processing non-stationary signals

[0083] The fractional Fourier transform has a free parameter, but the selection of different parameters can only determine the specific representation of the linear frequency modulation basis function, and will not change the essence of its basis function as the linear frequency modulation function. In order to be able to analyze non-stationary signals on a more flexible basis function, the advantage of singular spectrum analysis that it is not restricted by fixed basis functions is borrowed to obtain a flexible method for processing non-stationary signals.

[0084] The embodiment of the present application provides a non-stationary signal processing method, which accurately captures non-stationary signals through a signal collector and converts them into discrete signals, providing a high-quality data basis for subsequent processing. Then, the fractional Fourier transform algorithm is combined to effectively improve the resolution of the signal in the time and frequency domain and enhance the stability of the signal. The embedded trajectory matrix algorithm is then used to further mine the signal features and construct a trajectory matrix that reflects the internal structure of the signal. Through singular value decomposition, effective extraction and dimensionality reduction of signal features are achieved, and noise interference is effectively suppressed. Finally, based on the processing of the singular value matrix algorithm, the processing results corresponding to the non-stationary signal are accurately determined, which significantly improves the accuracy and stability of non-stationary signal processing and ensures the application effect of non-stationary signal processing.

[0085] like Figure 1 As shown, the method of this embodiment includes:

[0086] Step 101, using a signal collector to collect a non-stationary signal, and determining a first discretized signal corresponding to the non-stationary signal, wherein the non-stationary signal includes a radar signal, a voice signal, an electrocardiogram signal, an image signal, and a temperature signal.

[0087] In this step, the signal collector is a device that can capture and record various forms of signals. Non-stationary signals refer to signals whose distribution parameters or distribution laws change over time, such as radar signals, voice signals, electrocardiogram signals, image signals, and temperature signals.

[0088] Radar signals are electromagnetic wave signals transmitted and received by radar systems. They are usually used to detect, track and identify targets. Radar signals are non-stationary because their characteristics (such as frequency, amplitude and phase) may change with time and the change of targets.

[0089] Speech signals are non-stationary because the characteristics of speech such as timbre, pitch, and volume may change with the speaker's emotion, speaking speed, and pronunciation.

[0090] ECG signals are also non-stationary because they vary with factors such as their waveform and characteristic emotions.

[0091] Image signal is a signal that describes information such as image brightness and color. In dynamic images, the non-stationarity of image signals is manifested in the changes in image content, brightness, color and other characteristics.

[0092] The signal collector collects non-stationary signals through sensors or interfaces and converts them into electrical signals. Then, these analog signals need to be sampled and quantized before they can be converted into discrete signals.

[0093] Sampling refers to collecting analog signals at a certain frequency to convert continuous analog signals into discrete sampling values. The sampling theorem stipulates that the sampling frequency must be at least twice the highest frequency of the signal to ensure that the signal is not distorted after sampling.

[0094] Quantization is the process of converting a continuous sampled signal into a discrete digital signal. The quantization process is to classify the sampled values ​​according to a certain precision. Each sampled value is assigned to a corresponding quantization level. In the digital signal output by the quantizer, the size of the sampled value is represented by an integer.

[0095] After sampling and quantization, the non-stationary signal is converted into the corresponding first discretized signal. These discretized signals have finite time points and amplitude values ​​and can be stored and processed in digital circuits and computer systems.

[0096] For example, a non-stationary signal is collected as a radar signal by a signal collector, and a first discretized signal of the radar signal corresponding to the radar signal is determined; or,

[0097] Using a signal collector to collect a non-stationary signal as a speech signal, and determining a first discretized signal of the speech signal corresponding to the speech signal; or,

[0098] Using a signal collector to collect a non-stationary signal as an electrocardiogram signal, and determining a first discretized signal of the electrocardiogram signal corresponding to the electrocardiogram signal; or,

[0099] Using a signal collector to collect a non-stationary signal as an image signal, and determining a first discretized signal of the image signal corresponding to the image signal; or,

[0100] The non-stationary signal is collected as a temperature signal by using a signal collector, and a first discretized signal of the temperature signal corresponding to the temperature signal is determined.

[0101] Step 102: Process the first discretized signal using a fractional Fourier transform algorithm to obtain a second discretized signal.

[0102] In this step, Fourier transform is a way to convert a signal from the time domain (or space domain) to the frequency domain. It is widely used in signal processing, image analysis and other fields to reveal the frequency components of a signal.

[0103] The traditional Fourier transform can be viewed as a complete conversion of a signal between the time domain and the frequency domain (i.e., a rotation of π / 2 radians). The fractional Fourier transform allows the signal to rotate at any angle between the two domains, providing a more flexible time-frequency representation than the traditional Fourier transform. This transform is between the traditional time domain and the frequency domain, and can reveal the characteristics of the signal in different "fractional order" domains.

[0104] In signal processing, signals usually exist in a discretized form, that is, the signal is sampled as a series of numerical points. The "first discretized signal" here refers to the initial, discrete signal to be processed. It may come from sensor data, audio signals, image data, etc.

[0105] Applying the fractional Fourier transform algorithm to the first discretized signal can achieve the extraction of specific frequency components in the signal, enhancement of signal features, denoising, etc., and obtain a new discrete signal (i.e., the second discretized signal). This signal may have different characteristics from the original signal in the time domain, frequency domain, or a certain fractional order domain, depending on the parameter settings of the fractional Fourier transform.

[0106] For example, the first discretized signal of the radar signal is processed using a fractional Fourier transform algorithm to obtain a second discretized signal of the radar signal; or,

[0107] Processing the first discretized signal of the speech signal using a fractional Fourier transform algorithm to obtain a second discretized signal of the speech signal; or,

[0108] Processing the first discretized signal of the electrocardiogram signal by using a fractional Fourier transform algorithm to obtain a second discretized signal of the electrocardiogram signal; or,

[0109] Processing the first discretized signal of the image signal using a fractional Fourier transform algorithm to obtain a second discretized signal of the image signal; or,

[0110] The first discretized signal of the temperature signal is processed by using a fractional Fourier transform algorithm to obtain a second discretized signal of the temperature signal.

[0111] Step 103: Process the second discretized signal by using an embedding trajectory matrix algorithm to obtain an embedding trajectory matrix.

[0112] In this step, the embedding trajectory matrix transforms the one-dimensional time series data into a multi-dimensional matrix form to reveal the underlying structures and patterns in the data.

[0113] Specifically, the embedding trajectory matrix is ​​constructed by combining adjacent points or points at specific intervals in the time series into multi-dimensional vectors. These vectors can be regarded as snapshots of the time series within a certain time window, thus providing richer information about the dynamic behavior of the system.

[0114] The construction of the embedding trajectory matrix can vary depending on the application, including choosing the appropriate embedding dimension (i.e., the number of elements contained in each vector) and embedding delay (i.e., the time interval between elements in adjacent vectors).

[0115] In this process, the second discretized signal is input into the embedding trajectory matrix algorithm. The algorithm converts the signal into an embedding trajectory matrix according to predetermined parameters (such as embedding dimension and embedding delay).

[0116] For example, the second discretized signal of the radar signal is processed by an embedded trajectory matrix algorithm to obtain an embedded trajectory matrix of the radar signal; or,

[0117] Processing the second discretized signal of the speech signal by an embedding trajectory matrix algorithm to obtain an embedding trajectory matrix of the speech signal; or,

[0118] Processing the second discretized signal of the electrocardiogram signal by an embedded trajectory matrix algorithm to obtain an embedded trajectory matrix of the electrocardiogram signal; or,

[0119] Processing the second discretized signal of the image signal by an embedding trajectory matrix algorithm to obtain an embedding trajectory matrix of the image signal; or,

[0120] The second discretized signal of the temperature signal is processed by an embedded trajectory matrix algorithm to obtain an embedded trajectory matrix of the temperature signal.

[0121] Step 104 , performing singular value decomposition on the embedded trajectory matrix to obtain a decomposition matrix, and determining a singular value matrix, a first singular vector matrix, and a second singular vector matrix from the decomposition matrix.

[0122] In this step, singular value decomposition (SVD) is a way to decompose a matrix into the product of three specific matrices. For a given matrix A (here is the embedding trajectory matrix), SVD can decompose it into:

[0123] A=∪ΣV T

[0124] where ∪ and V are orthogonal matrices (i.e., their transpose is equal to their inverse), and ∑ is a diagonal matrix whose elements on the diagonal are called singular values.

[0125] Decomposition matrix: refers to the three matrices U, ∑ and V obtained by SVD T . These three matrices together form a complete representation of the original matrix A.

[0126] The singular value matrix refers to the ∑ matrix. It is a diagonal matrix, and the elements on the diagonal are the singular values. The singular values ​​are non-negative and are arranged in order from large to small (if sorted). The singular values ​​reflect the "energy" or "importance" of the original matrix in different directions.

[0127] The first singular vector matrix refers to the ∪ matrix. In SVD, the column vectors of ∪ are called left singular vectors, and they form a set of orthogonal bases in the column space of the original matrix A. The first singular vector usually refers to the column vector in ∪ corresponding to the largest singular value, which represents the most important column direction of the original matrix.

[0128] The second singular vector matrix is ​​V T Matrix. In SVD, the column vectors of V are called right singular vectors, which form a set of orthogonal bases of the row space of the original matrix A. The second singular vector usually refers to the column vector in V corresponding to the largest singular value, which represents the most important row direction of the original matrix.

[0129] For example, performing singular value decomposition on the embedded trajectory matrix of the radar signal to obtain a decomposition matrix of the radar signal, and determining a singular value matrix of the radar signal, a first singular vector matrix of the radar signal, and a second singular vector matrix of the radar signal from the decomposition matrix of the radar signal; or,

[0130] Performing singular value decomposition on the embedded trajectory matrix of the speech signal to obtain a decomposition matrix of the speech signal, and determining a singular value matrix of the speech signal, a first singular vector matrix of the speech signal, and a second singular vector matrix of the speech signal from the decomposition matrix of the speech signal; or,

[0131] Performing singular value decomposition on the embedded trajectory matrix of the ECG signal to obtain a decomposition matrix of the ECG signal, and determining a singular value matrix of the ECG signal, a first singular vector matrix of the ECG signal, and a second singular vector matrix of the ECG signal from the decomposition matrix of the ECG signal; or,

[0132] Performing singular value decomposition on the embedded trajectory matrix of the image signal to obtain a decomposition matrix of the image signal, and determining a singular value matrix of the image signal, a first singular vector matrix of the image signal, and a second singular vector matrix of the image signal from the decomposition matrix of the image signal; or,

[0133] The embedded trajectory matrix of the temperature signal is subjected to singular value decomposition to obtain a decomposition matrix of the temperature signal, and a singular value matrix of the temperature signal, a first singular vector matrix of the temperature signal and a second singular vector matrix of the temperature signal are determined from the decomposition matrix of the temperature signal.

[0134] Step 105: Process the singular value matrix, the first singular vector matrix and the second singular vector matrix using a singular value matrix algorithm to obtain a matrix corresponding to the singular values.

[0135] In this step, a matrix corresponding to the singular value (ie, a matrix corresponding to the singular value) is calculated based on each singular value and its corresponding left and right singular vectors through a singular value matrix algorithm to obtain a matrix with specific properties.

[0136] For example, a singular value matrix of the radar signal, a first singular vector matrix of the radar signal, and a second singular vector matrix of the radar signal are processed by a singular value matrix algorithm to obtain a matrix corresponding to the singular values ​​of the radar signal; or,

[0137] Processing the singular value matrix of the speech signal, the first singular vector matrix of the speech signal and the second singular vector matrix of the speech signal by a singular value matrix algorithm to obtain a matrix corresponding to the singular values ​​of the speech signal; or,

[0138] Processing the singular value matrix of the electrocardiogram signal, the first singular vector matrix of the electrocardiogram signal and the second singular vector matrix of the electrocardiogram signal by a singular value matrix algorithm to obtain a matrix corresponding to the singular values ​​of the electrocardiogram signal; or,

[0139] Processing the singular value matrix of the image signal, the first singular vector matrix of the image signal and the second singular vector matrix of the image signal by a singular value matrix algorithm to obtain a matrix corresponding to the singular values ​​of the image signal; or,

[0140] The singular value matrix of the temperature signal, the first singular vector matrix of the temperature signal and the second singular vector matrix of the temperature signal are processed by a singular value matrix algorithm to obtain a matrix corresponding to the singular values ​​of the temperature signal.

[0141] Step 106: Determine a processing result corresponding to the non-stationary signal based on the matrix corresponding to the singular value.

[0142] In this step, singular value decomposition can be used as an effective tool when processing non-stationary signals. Specifically, the non-stationary signal can be represented in matrix form and then decomposed using singular value decomposition. By retaining the larger singular values ​​and the corresponding singular vectors, the main information in the signal can be extracted while removing noise and redundant information. This method has a wide range of applications in the field of signal processing, such as signal compression, denoising, feature extraction, etc.

[0143] For example, in image processing, singular value decomposition can be used for image compression and restoration. By performing singular value decomposition on the image matrix and truncating some singular values, the image can be compressed while maintaining the main features of the image. Similarly, in signal processing, singular value decomposition can be used to extract the main components of non-stationary signals, thereby achieving effective representation and processing of the signal.

[0144] The singular value decomposition technique is used to process non-stationary signals, and the main information in the signal is extracted by retaining larger singular values ​​and corresponding singular vectors, so that the processing results can be determined stably and effectively.

[0145] For example, determining a processing result corresponding to the radar signal based on a matrix corresponding to the singular value of the radar signal; or,

[0146] Determine a processing result corresponding to the speech signal based on a matrix corresponding to the speech signal and the singular values; or,

[0147] Determine the processing result corresponding to the electrocardiogram signal based on the matrix corresponding to the electrocardiogram signal and the singular value; or,

[0148] Determine a processing result corresponding to the image signal based on a matrix corresponding to the image signal and the singular values; or,

[0149] A processing result corresponding to the temperature signal is determined based on a matrix corresponding to the singular values ​​with respect to the temperature signal.

[0150] Through the above scheme, the non-stationary signal is accurately captured by the signal collector and converted into a discrete signal, providing a high-quality data basis for subsequent processing. Then, the fractional Fourier transform algorithm is combined to effectively improve the resolution of the signal in the time and frequency domain and enhance the stability of the signal. The embedded trajectory matrix algorithm is then used to further mine the signal features and construct a trajectory matrix that reflects the internal structure of the signal. Through singular value decomposition, the effective extraction and dimensionality reduction of signal features are achieved, and noise interference is effectively suppressed. Finally, based on the processing of the singular value matrix algorithm, the processing results corresponding to the non-stationary signal are accurately determined, which significantly improves the accuracy and stability of non-stationary signal processing.

[0151] In some embodiments, a non-stationary signal is collected as a radar signal using a signal collector, and a first discretized signal of the radar signal corresponding to the radar signal is determined;

[0152] Processing the first discretized signal of the radar signal by using a fractional Fourier transform algorithm to obtain a second discretized signal of the radar signal;

[0153] Processing the second discretized signal of the radar signal by an embedded trajectory matrix algorithm to obtain an embedded trajectory matrix of the radar signal;

[0154] Performing singular value decomposition on the embedded trajectory matrix of the radar signal to obtain a decomposition matrix of the radar signal, and determining a singular value matrix of the radar signal, a first singular vector matrix of the radar signal, and a second singular vector matrix of the radar signal from the decomposition matrix of the radar signal;

[0155] Processing the radar signal based on the singular value matrix, the first singular vector matrix and the second singular vector matrix of the radar signal by a singular value matrix algorithm to obtain a matrix corresponding to the singular values ​​of the radar signal;

[0156] A processing result corresponding to the radar signal is determined based on a matrix corresponding to the singular values ​​regarding the radar signal.

[0157] In some embodiments, step 102 includes:

[0158] Step A1: Receive set fractional Fourier transform parameters using a digital signal processor.

[0159] Step A2: Based on the first discretized signal and the translation operator of the fractional Fourier transform under the fractional Fourier transform parameter, a second discretized signal is obtained by the following formula:

[0160]

[0161] Among them, x α(n) represents the second discretized signal, x(n) represents the first discretized signal, α represents the fractional Fourier transform parameter, j represents the imaginary unit, n represents the index of the sequence, Represents the translation operator of the fractional Fourier transform.

[0162] In the above scheme, the Fractional Fourier Transform (FRFT) is applied. Fractional Fourier Transform is a generalization of the traditional Fourier Transform, which allows the signal to be rotated at any angle between the Fourier domain and the time / space domain, rather than just the π / 2 (i.e. 90 degree) rotation corresponding to the traditional Fourier transform. The following is a detailed explanation of the above content:

[0163] The digital signal processor (DSP) first receives one or more parameters of the fractional Fourier transform. These parameters are usually one or more fractional orders (also called fractional exponents, rotation angles, etc.), which determine the angle by which the signal will be rotated in the fractional Fourier domain.

[0164] The first discretized signal refers to the signal obtained after the original signal has been sampled or discretized. In digital signal processing, the signal usually exists in a discrete form, that is, the signal is sampled into a series of discrete values.

[0165] The translation operator allows the signal to be translated in the fractional Fourier domain. This translation operator is usually related to the parameters of the fractional Fourier transform (such as the fractional order).

[0166] The second discretized signal is obtained by applying the fractional Fourier transform and the translation operator to the first discretized signal. This new signal has different characteristics, such as different behaviors in different domains (time domain, frequency domain or fractional Fourier domain).

[0167] Select the fractional Fourier transform parameter α and use the fractional Fourier transform translation operator under this parameter to transform the time series x(n) into a new time series x α (n);

[0168] Specifically, the fractional Fourier transform builds a bridge between the time domain signal and the transform domain signal representation. It is a generalized form of the Fourier transform, and its specific expression is

[0169]

[0170] Among them, the kernel function K α (u, t) is expressed as

[0171]

[0172] The classic translation algorithm is Lm [x(n)] = x(n+m), where m is the amount of translation. In contrast, the translation operator of the fractional Fourier transform is Thus, the time series x(n) is converted into a fractional Fourier transform translation sequence after taking the starting point n=1

[0173] In some embodiments, step 103 includes:

[0174] The second discretized signal is embedded into the preset trajectory matrix by the following formula to obtain the embedded trajectory matrix:

[0175]

[0176] Among them, H α represents the embedded trajectory matrix, K represents the number of rows of the preset trajectory matrix, and P represents the number of columns of the preset trajectory matrix.

[0177] In the above solution, the second discretized signal is embedded into the preset trajectory matrix to generate a new embedded trajectory matrix.

[0178] The preset trajectory matrix is ​​a matrix with a specific structure and size, which is used to represent or store some expected trajectory or path information. The number of rows and columns of this matrix (represented by and respectively) defines its size, and the elements in the matrix may represent points, speeds, positions or other related parameters on the trajectory.

[0179] The data of the second discretized signal is inserted into the preset trajectory matrix according to a certain rule or method. The specific method of embedding may depend on many factors, such as the characteristics of the signal, the structure of the trajectory matrix, and the purpose or application scenario of embedding.

[0180] The time series x α (n) is embedded into a trajectory matrix of size K×P, denoted as α-trajectory matrix H α , its specific form is:

[0181]

[0182] In some embodiments, in step 104, performing singular value decomposition on the embedded trajectory matrix to obtain a decomposition matrix includes:

[0183] Based on the embedded trajectory matrix, the decomposition matrix is ​​obtained by the following formula:

[0184]

[0185] Among them, H α represents the embedding trajectory matrix, represents the decomposition matrix, U represents the first singular vector matrix, Denotes the embedding trajectory matrix H α The conjugate transposed matrix of represents the second singular vector matrix, Second singular vector matrix is the co-transposed matrix of the matrix V, Λ represents the singular value matrix, and the elements in the singular value matrix Λ are The square roots of the eigenvalues ​​of .

[0186] In the above scheme, the embedded trajectory matrix is ​​processed by singular value decomposition (SVD) and a decomposition matrix is ​​obtained therefrom.

[0187] Decomposition Matrix is obtained from the embedding trajectory matrix H by SVD. α The decomposition matrix may contain some dimensionality reduction or feature extraction representation of the original data.

[0188] In SVD decomposition, the original matrix H α (i.e., the embedding trajectory matrix) is decomposed into the product of three matrices, the first of which contains the left singular vectors of the original data. These vectors capture the main components or features of the data.

[0189] Calculate H α Singular value decomposition of in It refers to the conjugate transpose of V, U=[u1,…,u K ] and V = [v1,…,v P ] are H α The left singular vector u of k , k = 1, ..., K and the right singular vector v p , p = 1, ..., P constitutes a K × K and P × P matrix; Λ = diag[λ1, ..., λ min{K，P} ] is K×P by H α The diagonal matrix is ​​composed of the singular values ​​λ1 of . The singular values ​​are arranged in descending order, and the left and right singular vectors are also changed to the order corresponding to the singular values.

[0190] Specifically, H α The left singular value vector matrix U is given by The characteristic vector of H α The right singular value vector matrix V is given by The characteristic vector of min{K，P} The elements of ] are Square root of the eigenvalue.

[0191] The embedded trajectory matrix is ​​quickly decomposed into the product of three matrices through SVD: the first singular vector matrix, the singular value matrix and the second singular vector matrix.

[0192] In some embodiments, step 105 includes:

[0193] Based on the singular value matrix, the first singular vector matrix and the second singular vector matrix, a matrix corresponding to the singular values ​​is obtained by the following formula:

[0194]

[0195] in, represents the matrix corresponding to the singular values, u l represents the lth first singular column vector in the first singular vector matrix, represents the lth second singular column vector in the second singular vector matrix, λ l Represents the lth singular value in the singular value matrix.

[0196] In the above scheme, each singular value and its corresponding left and right singular vectors are calculated to calculate the matrix corresponding to the singular value l = 1, ..., min{K, P}, where u l is the lth left singular column vector, is the conjugate transpose of the lth right singular column vector, λ l is the lth singular value.

[0197] In some embodiments, step 106 includes:

[0198] Step B1, based on the anti-diagonal elements in the matrix corresponding to the singular values, an averaging process is performed using the following formula to obtain a first time domain time series:

[0199]

[0200] in, represents the first time domain time series, represents the matrix corresponding to the singular value, n represents the index of the reconstructed sequence, k represents the index of the column of the matrix corresponding to the singular value, K represents the number of rows of the preset trajectory matrix, and P represents the number of columns of the preset trajectory matrix.

[0201] Step B2, performing inverse modulation processing based on the first time domain time series and preset fractional Fourier transform parameters through the following formula to obtain a second time domain time series:

[0202]

[0203] in, represents the second time domain time series, represents the first time-domain time series, α represents the fractional Fourier transform parameter, j represents the imaginary unit, and n represents the index of the sequence.

[0204] Step B3, based on the first time domain time series and the second time domain time series, determine the normalized correlation function value by the following formula:

[0205]

[0206] in, represents the normalized correlation function value, represents the second time domain time series, represents the transposed matrix of the second time domain time series, represents the first time domain time series, Represents the transposed matrix of the first time-domain time series.

[0207] Step B4: based on the comparison result between the normalized correlation function value and a preset threshold, grouping the time domain time series to obtain a plurality of time domain time groups, each time domain time group including a plurality of time domain time signals.

[0208] Step B5, summing up the multiple time domain signals in each time domain group to obtain the time domain signal component corresponding to each time domain group, and taking the time domain signal component corresponding to each time domain group as the processing result corresponding to the non-stationary signal.

[0209] In the above scheme, the average is taken in anti-diagonal order and the inverse modulation of the fractional Fourier transform translation operator is performed from H α The matrix corresponding to each singular value of Get the time series in the time domain;

[0210] Specifically, The time series obtained by averaging the anti-diagonal elements of is:

[0211]

[0212] On this basis, Modulate to obtain the components of the original time series

[0213] With two time series and The normalized correlation function between As a measure of distance, real is the operation of taking the real part. According to the given application scenario and signal processing requirements, the distance threshold or the number of groups is determined. Then, according to the size of the distance, the distance less than the given threshold is into a group; or according to a given number of groups, Adaptive grouping is performed based on the principle of close distance. Finally, the time series is divided into q groups

[0214] For example, in the application of signal denoising, q=2 is required, that is, the signal is divided into a clean signal component and a noise component; in the application of multi-target separation, q is required to be the preset number of targets to obtain the signal component corresponding to each target.

[0215] Sum the time series in each group to get q signal components, which are the signal components that meet the actual application requirements. For denoising applications, the signal component and the noise component are obtained separately; for multi-target separation applications, the signal component corresponding to each target is obtained.

[0216] In some embodiments, in step 101, determining a first discretized signal corresponding to the non-stationary signal includes:

[0217] The non-stationary signal is discretized according to a preset sampling interval to obtain a first discretized signal.

[0218] In the above scheme, when processing a non-stationary signal, in order to convert it into a digital form for subsequent digital signal processing (DSP) or analysis, it is necessary to sample the signal at a fixed time interval, which is the preset sampling interval.

[0219] Sampling a signal at a preset sampling interval is to convert a continuously changing signal into a series of discrete data points. This process is called signal discretization. The discretized signal, the first discretized signal, is a digital sequence, and each data point represents the amplitude of the original signal at the corresponding sampling moment.

[0220] After the above sampling and discretization processing, a discrete signal corresponding to the original non-stationary signal is obtained, that is, the first discretization signal. This signal is the basis for subsequent digital signal processing and analysis.

[0221] For example, the received signal x(t) (i.e., a non-stationary signal) is discretized at an appropriate sampling interval to obtain a time series x(n) (i.e., a first discretized signal);

[0222] Specifically, assuming that the sampling interval is Δt, the discretization of the received signal is x(nΔt). To simplify the representation, the discretized signal is recorded as x(n) (ie, the first discretized signal).

[0223] In some embodiments, Figure 2 As shown, the non-stationary signal processing process of the present application can be divided into three parts: data preprocessing, data adaptive processing, and data reconstruction.

[0224] The data preprocessing part includes steps 201 to 202;

[0225] Specifically: Step 201, discretization of received signals;

[0226] Step 202: Under the selected fractional Fourier transform parameters, a new time series is constructed using fractional transform.

[0227] The data adaptive processing includes steps 203 and 204;

[0228] Specifically: Step 203, construct a trajectory matrix, calculate the singular value decomposition of the matrix, and arrange the singular values ​​in descending order, and calculate a matrix with the same size as the trajectory matrix from each singular value and its corresponding left and right singular neighbors;

[0229] Step 204: average the elements of the constructed matrix in anti-diagonal order and perform modulation to obtain a time domain sequence.

[0230] The data reconstruction part includes step 205;

[0231] Specifically, in step 205, the signal is divided into different groups as required, and the components in the groups are superimposed to obtain the desired signal components.

[0232] like Figure 3 and Figure 4 As shown, the real and imaginary parts of a complex-valued non-stationary signal containing three components are shown. The signal model is:

[0233] x(n)=exp(-jπn 2 +j0.025πn)+0.8exp(-jπn 2 +j3πn+j330sin(πn / 300))+0.63exp(-jπn 2 +j0.625πn+j52.5sin(πn / 300))

[0234] The real parts of these three components and the real part of the reconstructed signal obtained by the scheme proposed in this application are Figure 3 In the above example, the imaginary parts of these three components and the real part of the reconstructed signal obtained by the scheme proposed in this patent are Figure 4 middle.

[0235] (1) This application improves the stability of singular spectrum analysis in non-stationary signal processing

[0236] In the past, singular spectrum analysis directly arranged the received signals into a trajectory matrix, and then performed signal analysis according to the matrix analysis method. The new algorithm currently proposed preprocessed the data before singular spectrum analysis, which helps the theoretical basis of singular spectrum analysis in non-stationary signal processing, and thus ensures the performance of the new algorithm in non-stationary signal processing.

[0237] (2) This application improves the effect of singular spectrum analysis in the denoising application of broadband signals

[0238] Convert non-stationary or broadband signals into narrowband signals in the fractional Fourier transform domain; narrowband signals correspond to fewer singular values ​​with larger values. In the denoising process, fewer singular values ​​mean that fewer noise components are retained, so the performance of the original singular spectrum analysis algorithm in broadband signal denoising can be improved.

[0239] (3) This application improves the flexibility of fractional Fourier transform in non-stationary signal processing

[0240] The basis functions of fractional Fourier transform are pre-set, which results in the signal being forcibly expanded on the linear frequency modulation basis functions during signal processing. When the restriction type does not match the signal model, the fractional Fourier transform of the signal cannot obtain a narrowband or concise representation, which brings difficulties to subsequent processing. Combining fractional Fourier transform with singular spectrum analysis can transfer the flexibility of singular spectrum analysis to fractional Fourier analysis.

[0241] It should be noted that the method of the embodiment of the present application can be performed by a single device, such as a computer or server. The method of this embodiment can also be applied to a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only perform one or more steps in the method of the embodiment of the present application, and the multiple devices will interact with each other to complete the described method.

[0242] It should be noted that the above describes some embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0243] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present application also provides a non-stationary signal processing device.

[0244] refer to Figure 5 , the non-stationary signal processing device comprises:

[0245] The discretized signal determination module 501 is configured to collect a non-stationary signal using a signal collector and determine a first discretized signal corresponding to the non-stationary signal, wherein the non-stationary signal includes a radar signal, a voice signal, an electrocardiogram signal, an image signal, and a temperature signal;

[0246] A fractional Fourier transform module 502 is configured to process the first discretized signal using a fractional Fourier transform algorithm to obtain a second discretized signal;

[0247] The embedding trajectory matrix determination module 503 is configured to process the second discretization signal by an embedding trajectory matrix algorithm to obtain an embedding trajectory matrix;

[0248] A singular value decomposition module 504 is configured to perform singular value decomposition on the embedded trajectory matrix to obtain a decomposition matrix, and determine a singular value matrix, a first singular vector matrix, and a second singular vector matrix from the decomposition matrix;

[0249] A singular value matrix processing module 505 is configured to process the singular value matrix, the first singular vector matrix and the second singular vector matrix through a singular value matrix algorithm to obtain a matrix corresponding to the singular value;

[0250] The processing result determination module 506 is configured to determine the processing result corresponding to the non-stationary signal based on the matrix corresponding to the singular value.

[0251] In some embodiments, the fractional Fourier transform module 502 is specifically configured to:

[0252] Using a digital signal processor to receive set fractional Fourier transform parameters;

[0253] Based on the first discretized signal and the translation operator of the fractional Fourier transform under the fractional Fourier transform parameter, the second discretized signal is obtained by the following formula:

[0254]

[0255] Among them, x α (n) represents the second discretized signal, x(n) represents the first discretized signal, α represents the fractional Fourier transform parameter, j represents the imaginary unit, n represents the index of the sequence, Represents the translation operator of the fractional Fourier transform.

[0256] In some embodiments, the embedding trajectory matrix determination module 503 is specifically configured to: embed the second discretized signal into a preset trajectory matrix by using the following formula to obtain an embedded trajectory matrix:

[0257]

[0258] Among them, H α represents the embedded trajectory matrix, K represents the number of rows of the preset trajectory matrix, and P represents the number of columns of the preset trajectory matrix.

[0259] In some embodiments, the singular value decomposition module 504 is specifically configured to:

[0260] Based on the embedded trajectory matrix, the decomposition matrix is ​​obtained by the following formula:

[0261]

[0262] Among them, H α represents the embedding trajectory matrix, represents the decomposition matrix, U represents the first singular vector matrix, Denotes the embedding trajectory matrix H α The conjugate transposed matrix of represents the second singular vector matrix, Second singular vector matrix is the co-transposed matrix of the matrix V, Λ represents the singular value matrix, and the elements in the singular value matrix Λ are The square roots of the eigenvalues ​​of .

[0263] In some embodiments, the singular value matrix processing module 505 is specifically configured as follows:

[0264] Based on the singular value matrix, the first singular vector matrix and the second singular vector matrix, a matrix corresponding to the singular values ​​is obtained by the following formula:

[0265]

[0266] in, represents the matrix corresponding to the singular values, u l represents the lth first singular column vector in the first singular vector matrix, represents the lth second singular column vector in the second singular vector matrix, λ l Represents the lth singular value in the singular value matrix.

[0267] In some embodiments, the processing result determination module 506 is specifically configured to:

[0268] Based on the anti-diagonal elements in the matrix corresponding to the singular values, the following formula is used to perform averaging to obtain the first time domain time series:

[0269]

[0270] in, represents the first time domain time series, represents the matrix corresponding to the singular value, n represents the index of the reconstructed sequence, k represents the index of the column of the matrix corresponding to the singular value, K represents the number of rows of the preset trajectory matrix, and P represents the number of columns of the preset trajectory matrix;

[0271] Based on the first time-domain time series and the preset fractional Fourier transform parameters, the demodulation process is performed by the following formula to obtain the second time-domain time series:

[0272]

[0273] in, represents the second time domain time series, represents the first time domain time series, α represents the fractional Fourier transform parameter, J represents the imaginary unit, and n represents the index of the sequence;

[0274] Based on the first time domain time series and the second time domain time series, a normalized correlation function value is determined by the following formula:

[0275]

[0276] in, represents the normalized correlation function value, represents the second time domain time series, represents the transposed matrix of the second time domain time series, represents the first time domain time series, represents the transposed matrix of the first time-domain time series;

[0277] Based on the comparison result between the normalized correlation function value and a preset threshold value, the time domain time series is grouped to obtain a plurality of time domain time groups, each time domain time group including a plurality of time domain time signals;

[0278] The multiple time domain time signals in each time domain time group are summed up to obtain the time domain time signal component corresponding to each time domain time group, and the time domain time signal component corresponding to each time domain time group is used as the processing result corresponding to the non-stationary signal.

[0279] In some embodiments, the discretization signal determination module 501 is specifically configured to:

[0280] The non-stationary signal is discretized according to a preset sampling interval to obtain a first discretized signal.

[0281] For the convenience of description, the above device is described in terms of functions divided into various modules. Of course, when implementing the present application, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0282] The device of the above embodiment is used to implement the corresponding non-stationary signal processing method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0283] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the non-stationary signal processing method described in any of the above embodiments is implemented.

[0284] Figure 6 A more specific schematic diagram of the hardware structure of an electronic device provided in this embodiment is shown, and the device may include: a processor 601, a memory 602, an input / output interface 603, a communication interface 604, and a bus 605. The processor 601, the memory 602, the input / output interface 603, and the communication interface 604 are connected to each other in communication within the device through the bus 605.

[0285] The processor 601 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0286] The memory 602 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 602 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program codes are stored in the memory 602 and called and executed by the processor 601.

[0287] The input / output interface 603 is used to connect the input / output module to realize information input and output. The input / output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device can include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device can include a display, a speaker, a vibrator, an indicator light, etc.

[0288] The communication interface 604 is used to connect a communication module (not shown) to realize communication interaction between the device and other devices. The communication module can realize communication through a wired mode (such as USB, network cable, etc.) or a wireless mode (such as mobile network, WIFI, Bluetooth, etc.).

[0289] The bus 605 includes a path for transmitting information between various components of the device (eg, the processor 601 , the memory 602 , the input / output interface 603 , and the communication interface 604 ).

[0290] It should be noted that, although the above device only shows the processor 601, the memory 602, the input / output interface 603, the communication interface 604 and the bus 605, in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include the components necessary for implementing the embodiments of the present specification, and does not necessarily include all the components shown in the figure.

[0291] The electronic device of the above embodiment is used to implement the corresponding non-stationary signal processing method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0292] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present application also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the non-stationary signal processing method described in any of the above embodiments.

[0293] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0294] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the non-stationary signal processing method described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0295] A person skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present application is limited to these examples. In line with the concept of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present application as described above, which are not provided in detail for the sake of simplicity.

[0296] In addition, to simplify the description and discussion, and in order not to make the embodiments of the present application difficult to understand, the known power supply / ground connection with the integrated circuit (IC) chip and other components may or may not be shown in the provided drawings. In addition, the device can be shown in the form of a block diagram to avoid making the embodiments of the present application difficult to understand, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform to be implemented in the embodiments of the present application (that is, these details should be fully within the scope of understanding of those skilled in the art). In the case of elaborating specific details (e.g., circuits) to describe exemplary embodiments of the present application, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details or when these specific details are changed. Therefore, these descriptions should be considered to be illustrative rather than restrictive.

[0297] Although the present application has been described in conjunction with specific embodiments of the present application, many replacements, modifications and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.

[0298] The embodiments of the present application are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the present application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present application should be included in the protection scope of the present application.

Claims

1. A non-stationary signal processing method, characterized in that: include: Using a signal collector to collect a non-stationary signal, and determining a first discretized signal corresponding to the non-stationary signal, wherein the non-stationary signal includes a radar signal, a voice signal, an electrocardiogram signal, an image signal, and a temperature signal; Processing the first discretized signal using a fractional Fourier transform algorithm to obtain a second discretized signal; Processing the second discretized signal by an embedding trajectory matrix algorithm to obtain an embedding trajectory matrix; Performing singular value decomposition on the embedded trajectory matrix to obtain a decomposition matrix, and determining a singular value matrix, a first singular vector matrix, and a second singular vector matrix from the decomposition matrix; Processing the singular value matrix, the first singular vector matrix and the second singular vector matrix by a singular value matrix algorithm to obtain a matrix corresponding to the singular value; A processing result corresponding to the non-stationary signal is determined based on the matrix corresponding to the singular values.

2. The method according to claim 1, characterized in that: The step of processing the first discretized signal by using a fractional Fourier transform algorithm to obtain a second discretized signal includes: Using a digital signal processor to receive set fractional Fourier transform parameters; Based on the first discretized signal and the translation operator of the fractional Fourier transform under the fractional Fourier transform parameter, the second discretized signal is obtained by the following formula: Among them, x α (n) represents the second discretized signal, x(n) represents the first discretized signal, α represents the fractional Fourier transform parameter, j represents the imaginary unit, n represents the index of the time series, Represents the translation operator of the fractional Fourier transform.

3. The method according to claim 1, characterized in that The step of processing the second discretized signal by using an embedded trajectory matrix algorithm to obtain an embedded trajectory matrix includes: The second discretized signal is embedded into the preset trajectory matrix by the following formula to obtain the embedded trajectory matrix: Among them, H α represents the embedded trajectory matrix, K represents the number of rows of the preset trajectory matrix, and P represents the number of columns of the preset trajectory matrix.

4. The method according to claim 1, characterized in that: The step of performing singular value decomposition on the embedded trajectory matrix to obtain a decomposition matrix includes: Based on the embedded trajectory matrix, the decomposition matrix is ​​obtained by the following formula: Among them, H α represents the embedding trajectory matrix, represents the decomposition matrix, U represents the first singular vector matrix, Denotes the embedding trajectory matrix H α The conjugate transposed matrix of represents the second singular vector matrix, Second singular vector matrix is the co-transposed matrix of the matrix V, Λ represents the singular value matrix, and the elements in the singular value matrix Λ are The square roots of the eigenvalues ​​of .

5. The method according to claim 1, characterized in that The processing based on the singular value matrix, the first singular vector matrix and the second singular vector matrix by a singular value matrix algorithm to obtain a matrix corresponding to the singular value includes: Based on the singular value matrix, the first singular vector matrix and the second singular vector matrix, a matrix corresponding to the singular values ​​is obtained by the following formula: in, represents the matrix corresponding to the singular values, u l represents the lth first singular column vector in the first singular vector matrix, represents the lth second singular column vector in the second singular vector matrix, λ l Represents the lth singular value in the singular value matrix.

6. The method according to claim 1, characterized in that The determining of the processing result corresponding to the non-stationary signal based on the matrix corresponding to the singular value comprises: Based on the anti-diagonal elements in the matrix corresponding to the singular values, the following formula is used to perform averaging to obtain the first time domain time series: in, represents the first time domain time series, represents the matrix corresponding to the singular value, n represents the index of the reconstructed sequence, k represents the index of the column of the matrix corresponding to the singular value, K represents the number of rows of the preset trajectory matrix, and P represents the number of columns of the preset trajectory matrix; Based on the first time-domain time series and the preset fractional Fourier transform parameters, the demodulation process is performed by the following formula to obtain the second time-domain time series: in, represents the second time domain time series, represents the first time domain time series, α represents the fractional Fourier transform parameter, j represents the imaginary unit, and n represents the index of the sequence; Based on the first time domain time series and the second time domain time series, a normalized correlation function value is determined by the following formula: in, represents the normalized correlation function value, represents the second time domain time series, represents the transposed matrix of the second time domain time series, represents the first time domain time series, represents the transposed matrix of the first time-domain time series; Based on the comparison result between the normalized correlation function value and a preset threshold value, the time domain time series is grouped to obtain a plurality of time domain time groups, each time domain time group including a plurality of time domain time signals; The multiple time domain time signals in each time domain time group are summed up to obtain the time domain time signal component corresponding to each time domain time group, and the time domain time signal component corresponding to each time domain time group is used as the processing result corresponding to the non-stationary signal.

7. The method according to claim 1, characterized in that The determining of a first discretized signal corresponding to the non-stationary signal comprises: The non-stationary signal is discretized according to a preset sampling interval to obtain a first discretized signal.

8. A non-stationary signal processing device, characterized in that: include: A discretized signal determination module is configured to collect a non-stationary signal using a signal collector and determine a first discretized signal corresponding to the non-stationary signal, wherein the non-stationary signal includes a radar signal, a voice signal, an electrocardiogram signal, an image signal, and a temperature signal; A fractional Fourier transform module, configured to process the first discretized signal using a fractional Fourier transform algorithm to obtain a second discretized signal; an embedding trajectory matrix determination module, configured to process the second discretization signal through an embedding trajectory matrix algorithm to obtain an embedding trajectory matrix; A singular value decomposition module is configured to perform singular value decomposition on the embedded trajectory matrix to obtain a decomposition matrix, and determine a singular value matrix, a first singular vector matrix, and a second singular vector matrix from the decomposition matrix; A singular value matrix processing module is configured to perform processing based on the singular value matrix, the first singular vector matrix and the second singular vector matrix through a singular value matrix algorithm to obtain a matrix corresponding to the singular value; The processing result determination module is configured to determine the processing result corresponding to the non-stationary signal based on the matrix corresponding to the singular value.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to enable a computer to execute the method according to any one of claims 1 to 7.

Images