A method and system for improving sensor signal strength and recognition

Abstract

The application provides a method and system for improving sensor signal strength and recognition degree, comprising the following steps: S1: obtaining sensor data; S2: using conjugate gradient optimization and orthogonal projection to extract a generalized feature pair of the obtained sensor data; S3: filtering the obtained sensor data using the generalized feature pair, and then performing data enhancement and recognition process. The method effectively solves the problem of online extraction of multiple generalized feature pairs by using the conjugate gradient optimization and orthogonal projection idea, and the extraction process has high numerical stability, fast convergence speed, and is more simple, which can effectively improve the sensor signal strength and recognition degree.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of adaptive signal processing, and relates to a method and system for improving sensor signal strength and recognition degree. BACKGROUND

[0002] In sensor signal processing, the subsequent processing of flow data such as filtering, enhancement and recognition has high requirements for the early preprocessing of data. The improvement of signal strength and recognition degree in the early preprocessing process is helpful for the later application of data. Generalized eigenvalue decomposition belongs to an unsupervised learning method and is a linear dimension reduction technique. Generalized eigenvalue decomposition aims to extract generalized eigenvectors / pairs of a covariance matrix to realize data filtering, linear discrimination, data compression and the like. It plays an important role in statistical signal processing and machine learning. For example, in beamforming, the dominant generalized eigenvector is used to design the weight coefficient of the maximum signal-to-noise ratio filter, so that the output signal has the maximum output signal-to-noise ratio. In Fisher linear discrimination, the dominant generalized eigenvector describes the direction in which the inter-class variance is maximized and the intra-class variance is minimized. The existing generalized eigenvalue pair extraction method usually relies on matrix decomposition technology and is based on direct decomposition operation on batch data. However, the matrix decomposition technology usually requires large computational complexity and memory overhead and is not suitable for the observation-extraction data stream extraction scene. Therefore, it is a common idea to realize the extraction of generalized eigenvalue pairs by means of gradient descent method. However, the gradient descent algorithm usually has the problem of slow convergence. In order to improve the convergence speed, the second-order algorithm such as Newton method and quasi-Newton method is used for the extraction of generalized eigenvectors. These methods involve the calculation of the second-order derivative Hessian matrix. In some special cases, these quantities do not exist, which causes the algorithm to have the problem of calculation stability. SUMMARY

[0003] In view of the problems in the prior art, the present application provides a method and system for improving sensor signal strength and recognition degree, so as to effectively improve the strength and recognition degree of sensor signals.

[0004] The present application is realized by the following technical solutions:

[0005] A method for improving sensor signal strength and recognition degree, comprising the following steps:

[0006] S1: acquiring sensor data;

[0007] S2: extracting generalized eigenvalue pairs of the acquired sensor data by means of conjugate gradient optimization and orthogonal projection;

[0008] S3: filtering the acquired sensor data by means of the generalized eigenvalue pairs, and then performing data enhancement and recognition process.

[0009] Preferably, the step S1 is specifically as follows: at the observation time , two signal data are collected through sensors, respectively recorded as and ,and , .

[0010] Preferably, before extracting the generalized feature pairs of the acquired data, the method further includes:

[0011] S201: Initializing the parameters in the conjugate gradient optimization process;

[0012] S202: Update the covariance matrix pair in the conjugate gradient optimization process.

[0013] Preferably, the step S201 is specifically as follows:

[0014] Initialize the data and The covariance matrix of ;

[0015] Initialize The predicted value of the generalized eigenvalue ;

[0016] Initialize The prediction residuals of the generalized characteristic expressions ;

[0017] Initialize A normalized weight vector ;

[0018] Initialize A non-normalized weight vector ;

[0019] Initialize The update step size of the unnormalized weight vector ;

[0020] Initialize Conjugate search direction of generalized eigenvectors ;

[0021] Initialize The conjugate update step of the generalized eigenvectors ;

[0022] Set the data r of the generalized feature pairs to be extracted, and .

[0023] Preferably, the step S202 is specifically as follows:

[0024]

[0025] wherein,

[0026] is a smoothing factor, taking values between 0 and 1 ;

[0027] and is the data stream vector observed by the sensor at time t, and ;

[0028] is the covariance matrix pair of the data stream at time t, and .

[0029] Preferably, the step S2 is specifically:

[0030] S2001: predicting the updated covariance matrix pair of the generalized eigenvalues;

[0031] S2002: predicting the prediction residual of the updated generalized eigenvector of the current data according to the current generalized eigenvalue expression;

[0032] S2003: calculating the update step of the current weight vector;

[0033] S2004: updating the current weight vector;

[0034] S2005: calculating the update step of the current conjugate search direction;

[0035] S2006: updating the current conjugate search direction;

[0036] S2007: using orthogonal projection, and repeating the steps S2001-S2006 to extract the remaining generalized eigenvalue pairs in turn.

[0037] Preferably, the step S2001 is specifically: updating the prediction of the th generalized eigenvalue at time t:

[0038]

[0039] wherein, is the normalized weight vector corresponding to the th at time t, and ;

[0040] The step S2002 is specifically: updating the prediction residual of the th generalized eigenvalue expression at time t:

[0041] ;

[0042] The step S2003 specifically comprises: calculating The update step of the kth non-normalized weight vector:

[0043]

[0044] wherein,

[0045]

[0046]

[0047]

[0048] wherein is a scalar involved in the calculation intermediate process, and ;

[0049] is a vector involved in the calculation intermediate process, and ;

[0050] , is the kth orthogonal complementary projection matrix, and ;

[0051] is the kth normalized weight vector corresponding to the time t, and ; is the kth conjugate search direction corresponding to the time t, and

[0052] ; The step S2004 specifically comprises: updating The kth non-normalized weight vector corresponding to the time t:

[0053] ; The step S2005 specifically comprises: calculating

[0054] The update step of the kth generalized eigenvector conjugate gradient search direction corresponding to the time t:

[0055]

[0056] ​​​​​​​​

[0057] Among them, the update step size ;

[0058] The step S2006 is specifically as follows: Moment Conjugate gradient search direction of generalized eigenvectors:

[0059] ;

[0060] The step S2007 is specifically as follows: using orthogonal projection to calculate the orthogonal complementary projection matrix

[0061]

[0062] And set the number of iterations , repeat S2001-S2006 to complete the extraction of the remaining generalized feature pairs until .

[0063] A system for improving sensor signal strength and recognition, comprising:

[0064] Data acquisition module, used to obtain signal data from sensors;

[0065] A generalized feature pair extraction module is used to extract the generalized feature pairs of the acquired data using conjugate gradient optimization and orthogonal projection;

[0066] The data processing module uses the extracted generalized features to filter the acquired data, and then performs enhancement and recognition.

[0067] A terminal device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of any one of the above methods when executing the computer program.

[0068] A computer-readable storage medium stores a computer program, wherein the computer program implements the steps of any one of the above methods when executed by a processor.

[0069] Compared with the prior art, the present invention has the following beneficial technical effects:

[0070] A method for improving the strength and recognition of sensor signals effectively solves the problem of online extraction of multiple generalized feature pairs through conjugate gradient optimization and orthogonal projection ideas. The extraction process has high numerical stability, fast convergence speed, and a simpler extraction process, which can effectively improve the strength and recognition of sensor signals.

[0071] Furthermore, in the process of updating the covariance matrix pair, The smoothing factor is between 0 and 1, which can make the iteration process forget the historical data and keep the current statistics to track the new data.

[0072] Further, in the process of calculating the orthogonal complementary projection matrix, the current extracted generalized feature pair component is orthogonally projected from the covariance space, so as to eliminate the component and pave the way for the extraction of the secondary generalized feature pair component. BRIEF DESCRIPTION OF DRAWINGS

[0073] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments will be briefly introduced as follows. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0074] Figure 1 A flow chart of a method for improving sensor signal strength and recognition degree in the embodiments of the present application;

[0075] Figure 2 A structural schematic diagram of a system for improving sensor signal strength and recognition degree provided in the embodiments of the present application;

[0076] Figure 3 A flow chart of a generalized feature pair extraction method in the embodiments of the present application. DETAILED DESCRIPTION

[0077] In order to make the objects, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions of the embodiments of the present application will be described clearly and completely in combination with the drawings of the embodiments of the present application. Obviously, the described embodiments are some embodiments of the present application, but not all the embodiments. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.

[0078] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. All other embodiments obtained by those skilled in the art on the basis of the embodiments in the present application without creative labor are within the scope of protection of the present application.

[0079] It should be noted that: similar reference numerals and letters represent similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.

[0080] In the description of the embodiments of the present application, it should be noted that if the terms "upper", "lower", "horizontal", "inner" and the like indicating the orientation or position relationship are based on the orientation or position relationship shown in the drawings, or the orientation or position relationship when the product of the present application is usually placed, and are only for the convenience of describing the present application and simplifying the description, and therefore cannot be understood as indicating or implying that the device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present application. In addition, the terms "first", "second" and the like are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0081] In addition, if the term "horizontal" appears, it does not mean that the component must be absolutely horizontal, but can be slightly inclined. For example, "horizontal" only means that its direction is relatively more horizontal than "vertical", and does not mean that the structure must be completely horizontal, but can be slightly inclined.

[0082] In the description of the embodiments of the present application, it should be noted that unless otherwise explicitly specified and limited, if the terms "arrangement", "installation", "connection", "connection" appear, they should be understood in a broad sense. For example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium; it can be the communication between the two elements inside. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0083] The present application will be described in further detail below with reference to the accompanying drawings:

[0084] As shown in the figure, a method for improving the signal strength and recognition of a sensor includes the following steps: Figure 1

[0085] S1: Obtain data of the sensor;

[0086] Specifically, at the observation time , two signal data are collected by the sensor, respectively denoted as and , and , .

[0087] S201: Initialize the parameters in the conjugate gradient optimization process;

[0088] Specifically,

[0089] Initialize the covariance matrix of the data and , denoted as ;

[0090] Initialize the first​ a predicted value of the generalized eigenvalue ;

[0091] initialize a predicted residual of the generalized eigenvalue expression ;

[0092] initialize a normalized weight vector ;

[0093] initialize a non-normalized weight vector ;

[0094] initialize a non-normalized weight vector update step size ;

[0095] initialize a conjugate search direction of the generalized eigenvector ;

[0096] initialize a conjugate update step size of the generalized eigenvector ;

[0097] set the data r of the generalized eigenpair to be extracted, and .

[0098] S202: update the covariance matrix pair in the conjugate gradient optimization process.

[0099] Specifically,

[0100]

[0101] wherein,

[0102] is a smoothing factor, and the value is between 0 and 1;

[0103] and is a data flow vector corresponding to the sensor observed at the moment, and ;

[0104] is a covariance matrix pair of the data flow at the moment, and .

[0105] S2: extracting the generalized eigenpair of the obtained sensor data by using the conjugate gradient optimization and the orthogonal projection;

[0106] The specific extraction process is as follows Figure 3 ​​​​​​​​As shown:

[0107] S2001: predict the generalized eigenvalues ​​of the updated covariance matrix pair;

[0108] S2002: predicting the prediction residual of the updated generalized feature vector of the current data according to the current generalized feature expression;

[0109] S2003: Calculate the update step size of the current weight vector;

[0110] S2004: Update the current weight vector;

[0111] S2005: Calculate the update step size of the current conjugate search direction;

[0112] S2006: Update the current conjugate search direction;

[0113] S2007: Using orthogonal projection, repeat steps S2001-S2006 to sequentially extract the remaining generalized feature pairs.

[0114] in,

[0115] Step S2001 is specifically: Update Moment The predicted quantity of generalized eigenvalues:

[0116]

[0117] in, for The moment corresponds to the A normalized weight vector, and ;

[0118] The step S2002 is specifically as follows: Moment Prediction residuals of a generalized characteristic expression:

[0119] ;

[0120] The step S2003 is specifically as follows: Moment The update step size of the unnormalized weight vector is:

[0121]

[0122] in,

[0123]

[0124]

[0125]

[0126] wherein is a scalar involved in the computation of the intermediate process, and ; is a vector involved in the computation of the intermediate process, and ; , is the k-th orthonormal complementary projection matrix, and ; ; corresponds to the k-th normalized weight vector at the moment , and ; ; corresponds to the k-th conjugate search direction at the moment , and ; ;

[0127] The step S2004 specifically comprises updating the k-th non-normalized weight vector at the moment ;

[0128] ;

[0129] The step S2005 specifically comprises calculating the update step of the conjugate gradient search direction of the k-th generalized eigenvector at the moment ;

[0130] wherein the update step

[0131] ;

[0132] The step S2006 specifically comprises updating the conjugate gradient search direction of the k-th generalized eigenvector at the moment ;

[0133] ;

[0134] The step S2007 specifically comprises calculating the orthonormal complementary projection matrix

[0135]

[0136] and setting the iteration number , repeating S2001-S2006 to complete the extraction of the remaining generalized eigenpairs, until .

[0137] ​​​S3: filtering the acquired sensor data using the generalized feature pair, and then performing a data enhancement and recognition process.

[0138] As shown in Figure 2 A system for improving sensor signal strength and recognition, comprising:

[0139] A data acquisition module 100 for acquiring signal data of a sensor;

[0140] A generalized feature pair extraction module 200 for extracting a generalized feature pair of the acquired data using conjugate gradient optimization and orthogonal projection;

[0141] A data processing module 300 for filtering the acquired data using the extracted generalized feature pair, and then performing enhancement and recognition.

[0142] The embodiment of the present application provides a schematic diagram of a terminal device. The terminal device of the embodiment comprises a processor, a memory, and a computer program stored in the memory and executable on the processor. The processor implements the steps in S1-S3 when executing the computer program. Alternatively, the processor implements the functions of each module / unit in each of the above device embodiment when executing the computer program.

[0143] The computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present application.

[0144] The terminal device can be a desktop computer, a notebook computer, a palm computer, a cloud server, and the like. The terminal device can include, but is not limited to, a processor and a memory.

[0145] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like.

[0146] The memory can be used to store the computer program and / or modules. The processor realizes various functions of the terminal device by running or executing the computer program and / or modules stored in the memory, and calling data stored in the memory.

[0147] The modules / units integrated in the terminal device, if realized in the form of software function units and sold or used as independent products, can be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. When the computer program is executed by a processor, the steps in S1-S3 can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

[0148] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. The present application can have various modifications and changes for those skilled in the art. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for improving the strength and resolution of a sensor signal, characterized by, The method comprises the following steps: S1: obtaining sensor data; S2: extracting a generalized eigenpair of the obtained sensor data by using conjugate gradient optimization and orthogonal projection; S3: filtering the obtained sensor data by using the generalized eigenpair, and then performing a data enhancement and recognition process; The step S2 specifically comprises: S2001: predicting a generalized eigenvalue of an updated covariance matrix pair; S2002: predicting a prediction residual of a generalized eigenvector updated by current data according to a current generalized eigenexpression; S2003: calculating an update step length of a current weight vector; S2004: updating the current weight vector; S2005: calculating an update step length of a current conjugate search direction; S2006: updating the current conjugate search direction; S2007: extracting a remaining generalized eigenpair by using orthogonal projection and repeating the steps S2001-S2006.

2. The method of claim 1, wherein, The step S1 is specifically: collecting two signal data respectively through sensors at the observation time , denoted as and , and , .

3. The method of claim 1, wherein the method further comprises: Before the step of extracting the generalized eigenpair of the obtained data, the method further comprises: S201: initializing a parameter in the conjugate gradient optimization process; S202: updating a covariance matrix pair in the conjugate gradient optimization process.

4. The method of claim 3, wherein the method further comprises: The step S201 specifically comprises: initializing the data and the covariance matrix of ; initializing the prediction of the first generalized eigenvalue ; initializing the prediction residual of the first generalized feature expression ; initializing a first normalized weight vector ; initializing a first non-normalized weight vector ; initializing a first non-normalized weight vector update step ; initializing a first conjugate search direction of a generalized eigenvector ; initializing the conjugate update step size of the first generalized eigenvector ; Set the data r of the pair of generalized features to be extracted, and .

5. The method of claim 3, wherein the method further comprises: The step S202 specifically comprises: wherein, is a smoothing factor, taking values between 0 and 1 ; and for the moment corresponds to a data stream vector observed by the sensor, and ; To The covariance matrix pair of the time data stream, and .

6. The method of claim 1, wherein the method further comprises: The step S2001 specifically includes: updating The moment of the The prediction of the generalized eigenvalue: wherein, is the time instant corresponds to the th normalized weight vector, and ; The step S2002 specifically includes: updating The moment is the The prediction residual of the i-th generalized feature expression: ； The step S2003 specifically includes: calculating The time point t The update step of the non-normalized weight vector: wherein, wherein is a scalar involved in the calculation of the intermediate process, and ; is a vector involved in the computation of the intermediate process, and ; , is the first orthogonal complement projection matrix, and is the second orthogonal complement projection matrix, and ; For The time instant corresponds to the th normalized weight vector, and ; For the moment corresponds to the th conjugate search direction, and ; The step S2004 specifically includes: updating The time point is the Non-normalized weight vector: ； The step S2005 specifically includes: calculating The updating step of the conjugate gradient search direction of the kth generalized eigenvector The updating step of the conjugate gradient search direction of the kth generalized eigenvector wherein the update step size ; The step S2006 specifically includes: updating The time is the The conjugate gradient search direction of the general eigenvector: ； The step S2007 specifically comprises: calculating an orthogonal complement projection matrix by using orthogonal projection and set the number of iterations S2001-S2006 are repeated to complete the extraction of the remaining pairs of generalized features until .

7. A system for enhancing the strength and discernibility of a sensor signal, the system comprising: The method comprises: a data acquisition module configured to obtain signal data of a sensor; an extraction generalized eigenpair module configured to extract a generalized eigenpair of the obtained data by using conjugate gradient optimization and orthogonal projection; a data processing module configured to filter the obtained data by using the extracted generalized eigenpair, and then perform an enhancement and recognition process. The step of extracting a generalized eigenpair of the obtained data by using conjugate gradient optimization and orthogonal projection specifically comprises: S2001: predicting a generalized eigenvalue of an updated covariance matrix pair; S2002: predicting a prediction residual of a generalized eigenvector updated by current data according to a current generalized eigenexpression; S2003: calculating an update step length of a current weight vector; S2004: updating the current weight vector; S2005: calculating an update step length of a current conjugate search direction; S2006: updating the current conjugate search direction; S2007: extracting a remaining generalized eigenpair by using orthogonal projection and repeating the steps S2001-S2006.

8. A terminal device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor implements the steps of the method according to any one of claims 1-6 when executing the computer program.

9. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 8. The computer program is executed by the processor to implement the steps of the method according to any one of claims 1-6.

Images