Signal processing method and system for radio frequency transmitting and receiving module

By introducing artificial intelligence-based signal processing methods in the RF transmission and reception module, including local semantic segmentation of signal time and frequency and noise component screening, the problem that traditional methods cannot adapt to in complex communication scenarios is solved, and efficient noise reduction of RF signals is achieved and quality improvement.

CN118646429BActive Publication Date: 2024-11-29SICHUAN XINKEAO ELECTRONIC TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202411116904.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-15
Publication Date
2024-11-29
Estimated Expiration
2044-08-15

AI Technical Summary

Technical Problem

Traditional RF signal processing methods cannot be effectively adapted in complex multi-user, multi-path and multi-frequency modern communication scenarios, resulting in difficulty in ensuring signal quality and signal-to-noise ratio.

Method used

Using local semantic segmentation of signal time-frequency, time-frequency feature extraction, one-way interactive matching network and Softmax function noise component filter, the signal processing methods based on artificial intelligence are used to perform feature extraction and noise component screening on radio frequency echo signals to generate reduction Noisy RF echo signal.

Benefits of technology

It improves the purity and quality of the radio frequency signal, enhances the reliability of the signal, and ensures the accurate transmission of information between different devices.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118646429B_ABST
    Figure CN118646429B_ABST
Patent Text Reader

Abstract

The present application discloses a signal processing method and system for a radio frequency transmitting and receiving module, and relates to the field of radio frequency signal processing technology. It first receives a radio frequency echo signal, and introduces an artificial intelligence-based signal processing and analysis algorithm at the back end to analyze the radio frequency echo signal, so as to capture the correlation relationship between each local time-frequency semantic in the radio frequency echo signal and the probability of the noise component therein, thereby filtering out the noise component to remove unnecessary frequency components to generate a noise-reduced radio frequency echo signal, which can ensure the purity and quality of the received radio frequency echo signal, thereby facilitating the accurate transmission of the received information between different devices and increasing the reliability of the signal.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of radio frequency signal processing, and in particular to a signal processing method and system for a radio frequency transmitting and receiving module. Background Art

[0002] Radio frequency technology is the cornerstone of modern wireless communications, allowing information to be transmitted in space via radio waves. The RF transmitter and receiver module is the core component in the wireless communication system, responsible for converting electrical signals into radio waves for transmission, as well as receiving radio waves from other devices and converting them back into electrical signals. The effectiveness of this process directly affects the quality and reliability of communication.

[0003] In wireless communications, RF signals are affected by various interferences and noises during transmission, so effective signal processing is required to extract useful information and reduce the impact of interference noise. In the RF transmitting and receiving modules, traditional RF signal processing methods usually include steps such as signal amplification, filtering, and demodulation. In these steps, filtering is to remove unnecessary frequency components in the signal to ensure the purity of the signal, thereby improving the signal-to-noise ratio and quality. However, traditional filter design is usually based on fixed rules, and filters based on fixed rules may not be able to adapt to complex and changing communication environments, especially in modern communication scenarios with multiple users, multiple paths, and multiple frequencies.

[0004] Therefore, an optimized signal processing solution for the radio frequency transmitting and receiving module is desired. Summary of the invention

[0005] This application is made in view of the above problems. One object of this application is to provide a signal processing method and system for a radio frequency transmitting and receiving module.

[0006] An embodiment of the present application provides a signal processing method of a radio frequency transmitting and receiving module, which includes:

[0007] receiving a radio frequency echo signal;

[0008] Performing signal time-frequency local semantic segmentation on the radio frequency echo signal to obtain a sequence of radio frequency echo local time-frequency graphs;

[0009] Extracting time-frequency features from the sequence of local time-frequency diagrams of the radio frequency echoes to obtain a sequence of local time-frequency feature vectors of the radio frequency echoes;

[0010] Taking each local time-frequency feature vector of the radio frequency echo in the sequence of local time-frequency feature vectors of the radio frequency echo as a query feature vector, inputting the query feature vector and the sequence of local time-frequency feature vectors of the radio frequency echo into a one-way interactive matching network to obtain a sequence of one-way matching feature vectors;

[0011] Inputting each one-way matching feature vector in the sequence of one-way matching feature vectors into a noise component filter based on a Softmax function to obtain a sequence of noise component probability values;

[0012] Based on the sequence of noise component probability values, feature filtering and signal filtering and restoration are performed on the sequence of echo local time-frequency feature vectors to obtain a noise-reduced radio frequency echo signal.

[0013] For example, according to the signal processing method of the radio frequency transmitting and receiving module of an embodiment of the present application, the radio frequency echo signal is subjected to signal time-frequency local semantic segmentation to obtain a sequence of radio frequency echo local time-frequency diagrams, including:

[0014] Performing wavelet analysis on the radio frequency echo signal to obtain a radio frequency echo time-frequency diagram;

[0015] The radio frequency echo time-frequency diagram is subjected to super-pixel segmentation using a super-pixel segmentation method to obtain a sequence of the radio frequency echo local time-frequency diagrams.

[0016] For example, according to the signal processing method of the RF transmitting and receiving module of the embodiment of the present application, time-frequency feature extraction is performed on the sequence of RF echo local time-frequency diagrams to obtain a sequence of RF echo local time-frequency feature vectors, including: each RF echo local time-frequency diagram in the sequence of RF echo local time-frequency diagrams is passed through a time-frequency feature extractor based on a void convolutional neural network model to obtain a sequence of RF echo local time-frequency feature vectors.

[0017] For example, according to the signal processing method of the radio frequency transmitting and receiving module of the embodiment of the present application, each radio frequency echo local time-frequency feature vector in the sequence of the radio frequency echo local time-frequency feature vector is used as a query feature vector, and the query feature vector and the sequence of the radio frequency echo local time-frequency feature vector are input into a one-way interactive matching network to obtain a sequence of one-way matching feature vectors, including:

[0018] Calculating the semantic association between each local time-frequency feature vector of the radio frequency echo in the sequence of local time-frequency feature vectors of the radio frequency echo and the query feature vector to obtain a sequence of local time-frequency query metric coefficients of the radio frequency echo;

[0019] Normalizing the sequence of the radio frequency echo local time-frequency query metric coefficients to obtain a sequence of radio frequency echo local time-frequency one-way query matching weights;

[0020] Taking each radio frequency echo local time-frequency one-way query matching weight in the sequence of radio frequency echo local time-frequency one-way query matching weights as a weighting coefficient, weighted optimization is performed on each radio frequency echo local time-frequency feature vector in the sequence of radio frequency echo local time-frequency feature vectors to obtain the sequence of one-way matching feature vectors.

[0021] For example, according to the signal processing method of the radio frequency transmitting and receiving module of an embodiment of the present application, the semantic association between each radio frequency echo local time-frequency feature vector in the sequence of the radio frequency echo local time-frequency feature vector and the query feature vector is calculated to obtain a sequence of radio frequency echo local time-frequency query metric coefficients, including:

[0022] Calculating the logarithmic function value with base 2 of each local time-frequency feature vector of the radio frequency echo in the sequence of local time-frequency feature vectors of the radio frequency echo according to the position to obtain a sequence of local time-frequency representation vectors of the radio frequency echo;

[0023] Calculating the point-by-point multiplication between each radio frequency echo local time-frequency representation vector in the sequence of radio frequency echo local time-frequency representation vectors and the query feature vector to obtain a sequence of radio frequency echo local time-frequency-query semantic interaction feature vectors;

[0024] The global mean between each position feature value of each radio frequency echo local time-frequency-query semantic interaction feature vector in the sequence of the radio frequency echo local time-frequency-query semantic interaction feature vector is calculated to obtain the sequence of the radio frequency echo local time-frequency query metric coefficients.

[0025] For example, according to the signal processing method of the radio frequency transmitting and receiving module of an embodiment of the present application, the sequence of the radio frequency echo local time-frequency query metric coefficients is normalized to obtain a sequence of radio frequency echo local time-frequency one-way query matching weights, including:

[0026] Taking each radio frequency echo local time-frequency query metric coefficient in the sequence of radio frequency echo local time-frequency query metric coefficients as a power, calculating an exponential function value with a natural constant as a base to obtain a sequence of radio frequency echo local time-frequency query metric class support coefficients;

[0027] Calculating the sum of each radio frequency echo local time-frequency query metric class support coefficient in the sequence of the radio frequency echo local time-frequency query metric class support coefficient to obtain the total value of the radio frequency echo local time-frequency query metric class support coefficient;

[0028] The division between each RF echo local time-frequency query metric class support coefficient in the sequence of the RF echo local time-frequency query metric class support coefficient and the sum of the RF echo local time-frequency query metric class support coefficients is calculated to obtain the sequence of the RF echo local time-frequency one-way query matching weights.

[0029] For example, according to the signal processing method of the radio frequency transmitting and receiving module of an embodiment of the present application, the sequence of the echo local time-frequency feature vectors is subjected to feature filtering and signal filtering and restoration based on the sequence of the noise component probability values ​​to obtain a noise-reduced radio frequency echo signal, including:

[0030] Based on the comparison between each noise component probability value in the sequence of noise component probability values ​​and a preset threshold, feature filtering is performed on the sequence of local time-frequency feature vectors of the radio frequency echo to obtain a sequence of filtered local time-frequency feature vectors of the radio frequency echo;

[0031] The sequence of the local time-frequency feature vectors of the filtered radio frequency echo is input into a signal filtering restorer based on a generative adversarial network to obtain a noise-reduced radio frequency echo signal.

[0032] For example, according to the signal processing method of the RF transmitting and receiving module of the embodiment of the present application, it also includes a training step for training the time-frequency feature extractor based on the void convolutional neural network model, the one-way interactive matching network, the noise component filter based on the Softmax function, and the signal filtering restorer based on the adversarial generative network.

[0033] For example, according to the signal processing method of the radio frequency transmitting and receiving module of the embodiment of the present application, the training step includes:

[0034] Acquire training data, wherein the training data includes a training radio frequency echo signal and a real noise-reduced radio frequency echo signal;

[0035] Performing signal time-frequency local semantic segmentation on the training radio frequency echo signal to obtain a sequence of local time-frequency graphs of the training radio frequency echo;

[0036] Passing each training radio frequency echo local time-frequency diagram in the sequence of training radio frequency echo local time-frequency diagrams through the time-frequency feature extractor based on the atrous convolutional neural network model to obtain a sequence of training radio frequency echo local time-frequency feature vectors;

[0037] Taking each training radio frequency echo local time-frequency feature vector in the sequence of training radio frequency echo local time-frequency feature vectors as a training query feature vector, inputting the training query feature vector and the sequence of the training radio frequency echo local time-frequency feature vectors into the one-way interactive matching network to obtain a sequence of training one-way matching feature vectors;

[0038] Inputting each training one-way matching feature vector in the sequence of training one-way matching feature vectors into the noise component filter based on the Softmax function to obtain a sequence of training noise component probability values;

[0039] Based on the comparison between each training noise component probability value in the sequence of training noise component probability values ​​and a training preset threshold, feature filtering is performed on the sequence of training radio frequency echo local time-frequency feature vectors to obtain a sequence of training filtered radio frequency echo local time-frequency feature vectors;

[0040] Inputting the sequence of local time-frequency feature vectors of the radio frequency echo after training filtering into the signal filtering restorer based on the adversarial generation network to obtain a training denoised radio frequency echo signal;

[0041] Calculating a cross entropy loss function value between the training denoised radio frequency echo signal and a true denoised radio frequency echo signal to obtain a denoised radio frequency echo semantic cross entropy loss function value;

[0042] Based on the semantic cross entropy loss function value of the denoised radio frequency echo, the time-frequency feature extractor based on the hole convolutional neural network model, the one-way interactive matching network, the noise component filter based on the Softmax function, and the signal filtering restorer based on the adversarial generation network are trained.

[0043] The embodiment of the present application also provides a signal processing system of a radio frequency transmitting and receiving module, which includes:

[0044] A signal receiving module, used for receiving a radio frequency echo signal;

[0045] A signal time-frequency local semantic segmentation module, used for performing signal time-frequency local semantic segmentation on the radio frequency echo signal to obtain a sequence of radio frequency echo local time-frequency graphs;

[0046] A time-frequency feature extraction module, used for extracting time-frequency features from the sequence of local time-frequency images of the radio frequency echo to obtain a sequence of local time-frequency feature vectors of the radio frequency echo;

[0047] A one-way interactive matching module, used to use each radio frequency echo local time-frequency feature vector in the sequence of radio frequency echo local time-frequency feature vectors as a query feature vector, and input the query feature vector and the sequence of radio frequency echo local time-frequency feature vectors into a one-way interactive matching network to obtain a sequence of one-way matching feature vectors;

[0048] A screening module, used for inputting each one-way matching feature vector in the sequence of one-way matching feature vectors into a noise component filter based on a Softmax function to obtain a sequence of noise component probability values;

[0049] The noise reduction generation module is used to perform feature filtering and signal filtering and restoration on the sequence of echo local time-frequency feature vectors based on the sequence of noise component probability values ​​to obtain a noise-reduced radio frequency echo signal.

[0050] The present application has at least the following technical effects: compared with the prior art, the signal processing method and system of the RF transmitting and receiving module of the embodiment of the present application receives the RF echo signal and introduces an artificial intelligence-based signal processing and analysis algorithm at the back end to analyze the RF echo signal, so as to capture the correlation relationship between each local time-frequency semantics in the RF echo signal and the probability of the noise component therein, thereby filtering out unnecessary frequency components by filtering out the noise component to generate a noise-reduced RF echo signal, which can ensure the purity and quality of the received RF echo signal, thereby facilitating the accurate transmission of the received information between different devices and increasing the reliability of the signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application are briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0052] Figure 1 A schematic diagram of the application architecture of the signal processing method of the radio frequency transmitting and receiving module in an embodiment of the present application is shown;

[0053] Figure 2 A flow chart showing a signal processing method of a radio frequency transmitting and receiving module in an embodiment of the present application is shown;

[0054] Figure 3 A flowchart of sub-step S540 of the signal processing method of the radio frequency transmitting and receiving module in an embodiment of the present application is shown;

[0055] Figure 4 A schematic diagram of the structure of the signal processing system of the radio frequency transmitting and receiving module in an embodiment of the present application is shown;

[0056] Figure 5 An application scenario diagram of the signal processing method of the radio frequency transmitting and receiving module in an embodiment of the present application is shown;

[0057] Figure 6 A schematic diagram of a storage medium according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0059] The terms used in this specification are those common terms currently widely used in the art in consideration of the functions of the present application, but these terms may vary according to the intention of a person of ordinary skill in the art, precedents, or new technologies in the art. In addition, specific terms may be selected by the applicant, and in this case, their detailed meanings will be described in the detailed description of the present application. Therefore, the terms used in the specification should not be understood as simple names, but rather as a general description based on the meaning of the terms and the present application.

[0060] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0061] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. At the same time, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0062] Figure 1 A schematic diagram of the application architecture of the signal processing method of the radio frequency transmitting and receiving module in the embodiment of the present application is shown, including a server 100 and a terminal device 200.

[0063] The terminal device 200 and the server 100 can be connected via the Internet to achieve mutual communication. Optionally, the above-mentioned Internet uses standard communication technology and / or protocol. The Internet is usually the Internet, but it can also be any network, including but not limited to any combination of a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a mobile, wired or wireless network, a private network or a virtual private network. In some embodiments, the data exchanged through the network is represented by technologies and / or formats including Hyper Text Markup Language (HTML), Extensible Markup Language (XML), etc. In addition, conventional encryption technologies such as Secure Socket Layer (SSL), Transport Layer Security (TLS), Virtual Private Network (VPN), Internet Protocol Security (IPsec), etc. can also be used to encrypt all or some links. In other embodiments, customized and / or dedicated data communication technologies can also be used to replace or supplement the above-mentioned data communication technologies.

[0064] The server 100 can provide various network services for the terminal device 200, wherein the server 100 can be a single server, a server cluster consisting of several servers, or a cloud computing center. Specifically, the server 100 may include a processor 110 (Center Processing Unit, CPU), a memory 120, an input device 130, and an output device 140, etc. The input device 130 may include a keyboard, a mouse, a touch screen, etc., and the output device 140 may include a display device, such as a liquid crystal display (Liquid Crystal Display, LCD), a cathode ray tube (Cathode Ray Tube, CRT), etc.

[0065] The memory 120 may include a read-only memory (ROM) and a random access memory (RAM), and provides the processor 110 with program instructions and data stored in the memory 120. In the embodiment of the present application, the memory 120 may be used to store the program of the signal processing method of the RF transmitting and receiving module in the embodiment of the present application.

[0066] The processor 110 calls the program instructions stored in the memory 120, and the processor 110 is used to execute the steps of the signal processing method of any radio frequency transmitting and receiving module in the embodiments of the present application according to the obtained program instructions.

[0067] In addition, the application architecture diagram in the embodiment of the present application is intended to more clearly illustrate the technical solution in the embodiment of the present application, and does not constitute a limitation on the technical solution provided in the embodiment of the present application. Of course, for other application architectures and business applications, the technical solution provided in the embodiment of the present application is also applicable to similar problems.

[0068] The following is a non-limiting description of the signal processing method of the radio frequency transmitting and receiving module provided according to at least one embodiment of the present application through several examples or embodiments. As described below, different features in these specific examples or embodiments can be combined with each other without conflicting with each other to obtain new examples or embodiments, and these new examples or embodiments also belong to the scope of protection of the present application.

[0069] In response to the above technical problems, the technical concept of the present application is to receive the RF echo signal in the RF transmitting and receiving module, and introduce an artificial intelligence-based signal processing and analysis algorithm at the back end to analyze the RF echo signal, so as to capture the correlation relationship between the local time-frequency semantics in the RF echo signal and the probability of the noise component therein, and then eliminate unnecessary frequency components by filtering out the noise component to generate a noise-reduced RF echo signal, which can ensure the purity and quality of the received RF echo signal, thereby facilitating the accurate transmission of the received information between different devices and increasing the reliability of the signal.

[0070] Figure 2 The flowchart of the signal processing method of the radio frequency transmitting and receiving module in the embodiment of the present application is shown. For example, the signal processing method of the radio frequency transmitting and receiving module can be executed by a server, and the server can be Figure 1 The server 100 shown in FIG. Figure 2As shown, according to the signal processing method of the RF transmitting and receiving module of the embodiment of the present application, the method includes the following steps: S510, receiving a RF echo signal; S520, performing signal time-frequency local semantic segmentation on the RF echo signal to obtain a sequence of RF echo local time-frequency diagrams; S530, performing time-frequency feature extraction on the sequence of RF echo local time-frequency diagrams to obtain a sequence of RF echo local time-frequency feature vectors; S540, using each RF echo local time-frequency feature vector in the sequence of RF echo local time-frequency feature vectors as a query feature vector, and inputting the query feature vector and the sequence of RF echo local time-frequency feature vectors into a one-way interactive matching network to obtain a sequence of one-way matching feature vectors; S550, respectively inputting each one-way matching feature vector in the sequence of one-way matching feature vectors into a noise component filter based on a Softmax function to obtain a sequence of noise component probability values; S560, performing feature filtering and signal filtering restoration on the sequence of echo local time-frequency feature vectors based on the sequence of noise component probability values ​​to obtain a de-noised RF echo signal.

[0071] Specifically, in the technical solution of the present application, first, a radio frequency echo signal is received. In a radio frequency communication system, the radio frequency signal often has a time-varying characteristic during the transmission process, that is, the frequency of the signal changes with time, so the radio frequency echo signal often contains rich time domain and frequency domain information. Based on this, in order to more clearly observe the changing law of the signal in the time domain and frequency domain, and help to better understand the characteristics of the signal; in the technical solution of the present application, the radio frequency echo signal is subjected to wavelet analysis to obtain a radio frequency echo time-frequency diagram. Among them, wavelet analysis is an effective signal processing method that can help to better understand and describe the time-frequency characteristics of the signal. That is, wavelet analysis has the characteristics of multi-resolution and can provide information about the signal in the time domain and frequency domain at the same time. This characteristic makes wavelet analysis more flexible and efficient in capturing the instantaneous characteristics and frequency characteristics of the signal, and helps to more accurately analyze the time-frequency structure of the signal.

[0072] In order to be able to more accurately identify the useful part and the noise part of the signal in a complex communication environment, so as to more specifically perform signal enhancement and noise suppression in subsequent processing steps, in the technical solution of the present application, the superpixel segmentation method is further used to perform superpixel segmentation on the RF echo time-frequency diagram to obtain a sequence of RF echo local time-frequency diagrams. It should be understood that the superpixel segmentation method can divide the image into multiple areas with similar attributes according to the texture and distance characteristics of the RF echo time-frequency diagram. These areas are visually coherent and have similar textures or colors. In the RF echo time-frequency diagram, this means that different parts of the signal can be effectively distinguished, thereby providing clearer and more accurate data information for feature extraction.

[0073] Correspondingly, in step S520, the radio frequency echo signal is subjected to local semantic time-frequency segmentation to obtain a sequence of radio frequency echo local time-frequency maps, including: performing wavelet analysis on the radio frequency echo signal to obtain a radio frequency echo time-frequency map; and performing super-pixel segmentation on the radio frequency echo time-frequency map using a super-pixel segmentation method to obtain a sequence of radio frequency echo local time-frequency maps.

[0074] Furthermore, each local time-frequency map of the radio frequency echo in the sequence of local time-frequency maps of the radio frequency echo is passed through a time-frequency feature extractor based on a dilated convolutional neural network model to obtain a sequence of local time-frequency feature vectors of the radio frequency echo. It should be understood that the dilated convolutional neural network is a special convolution operation that increases the receptive field by adding holes to the convolution kernel (i.e., inserting zero values ​​into the convolution kernel) without increasing the size of the convolution kernel. This enables the network to capture feature information in the local time-frequency map of the radio frequency echo at different scales, i.e., simultaneously consider the local details of the signal and the broader contextual information to provide richer time-frequency features. Therefore, through the processing of the time-frequency feature extractor based on the dilated convolutional neural network model, representative features can be effectively extracted from the time-frequency map to help understand the time-frequency characteristics of the signal.

[0075] Correspondingly, in step S530, time-frequency features are extracted from the sequence of local time-frequency diagrams of the radio frequency echo to obtain a sequence of local time-frequency feature vectors of the radio frequency echo, including: each local time-frequency diagram of the radio frequency echo in the sequence of local time-frequency diagrams of the radio frequency echo is passed through a time-frequency feature extractor based on a hole convolutional neural network model to obtain a sequence of local time-frequency feature vectors of the radio frequency echo.

[0076] Correspondingly, each RF echo local time-frequency feature vector in the sequence of RF echo local time-frequency feature vectors respectively contains local time-frequency semantic features related to the RF echo map, and these different RF echo local time-frequency semantics have correlation relationships and hidden association patterns. The association relationship between these local time-frequency features can well highlight the essential dynamic mode and change of the signal, which is conducive to screening and filtering noise components. Based on this, in the technical solution of the present application, each RF echo local time-frequency feature vector in the sequence of RF echo local time-frequency feature vectors is further used as a query feature vector, and the query feature vector and the sequence of the RF echo local time-frequency feature vector are input into a one-way interactive matching network to obtain a sequence of one-way matching feature vectors. It should be understood that the core function of the one-way interactive matching network is matching features. It can automatically learn and capture the one-way matching features and hidden associations between each RF echo local time-frequency feature and all other RF echo local time-frequency features, so as to help identify and compare the similarities and differences between different RF echo local time-frequency features. This matching is crucial for understanding the intrinsic structure and characteristics of the signal, helps to distinguish the differential feature information brought by the noise component, and identifies the noise component, which is further beneficial for the subsequent filtering and restoration of these noise components.

[0077] Accordingly, if Figure 3 As shown, in step S540, each radio frequency echo local time-frequency feature vector in the sequence of radio frequency echo local time-frequency feature vectors is used as a query feature vector, and the query feature vector and the sequence of radio frequency echo local time-frequency feature vectors are input into a one-way interactive matching network to obtain a sequence of one-way matching feature vectors, including: S541, calculating the semantic association between each radio frequency echo local time-frequency feature vector in the sequence of radio frequency echo local time-frequency feature vectors and the query feature vector to obtain a sequence of radio frequency echo local time-frequency query metric coefficients; S542, normalizing the sequence of radio frequency echo local time-frequency query metric coefficients to obtain a sequence of radio frequency echo local time-frequency one-way query matching weights; S543, using each radio frequency echo local time-frequency one-way query matching weight in the sequence of radio frequency echo local time-frequency one-way query matching weights as a weighting coefficient, and performing weighted optimization on each radio frequency echo local time-frequency feature vector in the sequence of radio frequency echo local time-frequency feature vectors to obtain the sequence of one-way matching feature vectors.

[0078] Wherein, in step S541, the semantic association between each RF echo local time-frequency feature vector in the sequence of RF echo local time-frequency feature vectors and the query feature vector is calculated to obtain a sequence of RF echo local time-frequency query measurement coefficients, including: calculating the position-based logarithmic function value with base 2 of each RF echo local time-frequency feature vector in the sequence of RF echo local time-frequency feature vectors to obtain a sequence of RF echo local time-frequency representation vectors; calculating the position-based point multiplication between each RF echo local time-frequency representation vector in the sequence of RF echo local time-frequency representation vectors and the query feature vector to obtain a sequence of RF echo local time-frequency-query semantic interaction feature vectors; calculating the global mean between the positional feature values ​​of each RF echo local time-frequency-query semantic interaction feature vector in the sequence of RF echo local time-frequency-query semantic interaction feature vectors to obtain a sequence of RF echo local time-frequency query measurement coefficients.

[0079] Specifically, calculating the semantic association between each local time-frequency feature vector of the radio frequency echo in the sequence of local time-frequency feature vectors of the radio frequency echo and the query feature vector to obtain a sequence of local time-frequency query measurement coefficients of the radio frequency echo, including: calculating the semantic association between each local time-frequency feature vector of the radio frequency echo in the sequence of local time-frequency feature vectors of the radio frequency echo and the query feature vector using the following semantic association measurement formula to obtain a sequence of local time-frequency query measurement coefficients of the radio frequency echo; wherein the semantic association measurement formula is:

[0080]

[0081] in, and are the query feature vector and the value of each position in each local time-frequency feature vector of the radio frequency echo, respectively, is the scale of the query feature vector and each of the local time-frequency feature vectors of the radio frequency echo, represents the local time-frequency query metric coefficient of the radio frequency echo at each position in the sequence of the local time-frequency query metric coefficient of the radio frequency echo, Represents the value of the logarithmic function with base 2.

[0082] Among them, in step S542, the sequence of the radio frequency echo local time-frequency query metric coefficients is normalized to obtain a sequence of radio frequency echo local time-frequency one-way query matching weights, including: taking each radio frequency echo local time-frequency query metric coefficient in the sequence of the radio frequency echo local time-frequency query metric coefficient as a power, calculating an exponential function value with a natural constant as the base to obtain a sequence of radio frequency echo local time-frequency query metric class support coefficients; calculating the sum of each radio frequency echo local time-frequency query metric class support coefficient in the sequence of the radio frequency echo local time-frequency query metric class support coefficient to obtain a total value of the radio frequency echo local time-frequency query metric class support coefficients; calculating the division between each radio frequency echo local time-frequency query metric class support coefficient in the sequence of the radio frequency echo local time-frequency query metric class support coefficient and the total value of the radio frequency echo local time-frequency query metric class support coefficients to obtain the sequence of radio frequency echo local time-frequency one-way query matching weights.

[0083] Specifically, the sequence of the local time-frequency query metric coefficients of the radio frequency echo is normalized to obtain a sequence of the local time-frequency one-way query matching weights of the radio frequency echo, including: normalizing the sequence of the local time-frequency query metric coefficients of the radio frequency echo according to the following normalization formula to obtain a sequence of the local time-frequency one-way query matching weights of the radio frequency echo; wherein the normalization formula is:

[0084]

[0085] in, is the number of the sequence of the local time-frequency one-way matching semantic metric coefficients of the radio frequency echo The local time-frequency one-way matching semantic metric coefficient of the RF echo, is the position of the semantic metric coefficient of the local time-frequency one-way matching of the RF echo in the sequence, is the total number of semantic metric coefficients of local time-frequency one-way matching of RF echo, is an exponential operation with a natural constant as the base, is the first in the sequence of the semantic metric weights of the local time-frequency one-way matching of the radio frequency echo The semantic metric weight of the local time-frequency one-way matching of RF echoes.

[0086] After one-way matching and difference measurement are performed between the local time-frequency features of each RF echo, in order to effectively identify the noise components therein, in the technical solution of the present application, each one-way matching feature vector in the sequence of the one-way matching feature vector is further input into the noise component filter based on the Softmax function to obtain a sequence of noise component probability values. It should be understood that the Softmax function is a commonly used probability distribution function, which can convert the input feature vector into a probability distribution. In this way, a probability value can be assigned to each one-way matching feature vector, indicating the possibility that it belongs to the noise component. That is, through the noise component screening process of the Softmax function, it is possible to identify which local time-frequency features of the RF echo are more likely to be noise, thereby providing a basis for subsequent noise suppression.

[0087] Further, after the probability of noise component is calculated and screened for each local time-frequency feature of the radio frequency echo, the sequence of local time-frequency feature vectors of the radio frequency echo is further subjected to feature filtering based on the comparison between each noise component probability value in the sequence of the noise component probability value and the preset threshold value to obtain a sequence of local feature vectors of the radio frequency echo after filtering. That is, by comparing the noise component probability value of each local time-frequency feature of the radio frequency echo with the preset threshold value, it can be determined which feature vectors contain higher noise components, thereby performing corresponding filtering processing to reduce the interference of noise on signal processing. In particular, in response to each noise component probability value in the sequence of the noise component probability value being greater than the preset threshold value, the radio frequency echo time-frequency feature corresponding to the noise component probability value is filtered. Further, the sequence of local time-frequency feature vectors of the radio frequency echo after filtering is input into a signal filtering restorer based on a confrontation generation network to obtain a noise-reduced radio frequency echo signal. In this way, the purity and quality of the received radio frequency echo signal can be ensured, so that the received information can be accurately transmitted between different devices and the reliability of the signal is increased.

[0088] Correspondingly, in step S560, the sequence of echo local time-frequency feature vectors is feature filtered and signal filtered and restored based on the sequence of noise component probability values ​​to obtain a noise-reduced radio frequency echo signal, including: based on a comparison between each noise component probability value in the sequence of noise component probability values ​​and a preset threshold, feature filtering is performed on the sequence of radio frequency echo local time-frequency feature vectors to obtain a sequence of filtered radio frequency echo local time-frequency feature vectors; and the sequence of filtered radio frequency echo local time-frequency feature vectors is input into a signal filtering restorer based on an adversarial generative network to obtain a noise-reduced radio frequency echo signal.

[0089] Furthermore, in the technical solution of the present application, the signal processing method of the RF transmitting and receiving module also includes a training step for training the time-frequency feature extractor based on the void convolutional neural network model, the one-way interactive matching network, the noise component filter based on the Softmax function, and the signal filtering restorer based on the adversarial generative network.

[0090] Wherein, the training step includes: obtaining training data, wherein the training data includes a training RF echo signal and a real noise-reduced RF echo signal; performing signal time-frequency local semantic segmentation on the training RF echo signal to obtain a sequence of training RF echo local time-frequency graphs; passing each training RF echo local time-frequency graph in the sequence of training RF echo local time-frequency graphs through the time-frequency feature extractor based on the void convolutional neural network model to obtain a sequence of training RF echo local time-frequency feature vectors; using each training RF echo local time-frequency feature vector in the sequence of training RF echo local time-frequency feature vectors as a training query feature vector, inputting the sequence of the training query feature vector and the training RF echo local time-frequency feature vector into the one-way interactive matching network to obtain a sequence of training one-way matching feature vectors; and inputting each training one-way matching feature vector in the sequence of training one-way matching feature vectors into the noise vector based on the Softmax function. A component filter is used to obtain a sequence of training noise component probability values; based on the comparison between each training noise component probability value in the sequence of training noise component probability values ​​and a preset training threshold, the sequence of training RF echo local time-frequency feature vectors is feature filtered to obtain a sequence of training filtered RF echo local time-frequency feature vectors; the sequence of training filtered RF echo local time-frequency feature vectors is input into the signal filter restorer based on the adversarial generation network to obtain a training denoised RF echo signal; the cross entropy loss function value between the training denoised RF echo signal and the true denoised RF echo signal is calculated to obtain a semantic cross entropy loss function value of the denoised RF echo; based on the semantic cross entropy loss function value of the denoised RF echo, the time-frequency feature extractor based on the void convolutional neural network model, the one-way interactive matching network, the noise component filter based on the Softmax function and the signal filter restorer based on the adversarial generation network are trained.

[0091] Specifically, in the technical solution of the present application, each training radio frequency echo local time-frequency feature vector in the sequence of training radio frequency echo local time-frequency feature vectors respectively represents the time-frequency semantic association features of each training radio frequency echo local time-frequency graph in the sequence of training radio frequency echo local time-frequency graphs. Considering that each training radio frequency echo local time-frequency graph in the sequence of training radio frequency echo local time-frequency graphs is different, this results in a more significant inconsistency in the distribution of time-frequency semantic features between each training radio frequency echo local time-frequency feature vector in the sequence of training radio frequency echo local time-frequency feature vectors. In this way, when each training radio frequency echo local time-frequency feature vector in the sequence of the training radio frequency echo local time-frequency feature vector is used as a query feature vector and a one-way interactive matching is performed on the training query feature vector and the sequence of the training radio frequency echo local time-frequency feature vector, the feature distribution difference between the training query feature vector and each training radio frequency echo local time-frequency feature vector in the sequence of the training radio frequency echo local time-frequency feature vector will cause a difference in the distribution of feature semantic interaction weights, so that each of the training one-way matching feature vectors will also have the problem of insufficient aggregation of fusion semantic feature distribution, thereby affecting its classification convergence efficiency through the noise component filter based on the Softmax function, that is, affecting the efficiency of classification training and the accuracy of the noise component probability value.

[0092] Furthermore, each training one-way matching feature vector in the sequence of training one-way matching feature vectors is respectively input into the noise component filter based on the Softmax function to obtain a sequence of training noise component probability values, including: performing feature distribution aggregation optimization on the training one-way matching feature vectors according to the following formula to obtain corrected training one-way matching feature vectors; and inputting the corrected training one-way matching feature vectors into the noise component filter based on the Softmax function to obtain training noise component probability values.

[0093] The process of optimizing the aggregation of feature distribution can be expressed as follows:

[0094]

[0095] in, represents the training one-way matching feature vector, and represents the first norm and second norm of the training one-way matching feature vector, It means to calculate the value of natural exponential function. is the length of the training one-way matching feature vector, It means adding by position. represents dot product, A trained one-way matching feature vector representing the correction.

[0096] Based on this, the present application targets the training one-way matching feature vector as a feature cluster, wherein the dynamic semantic expression is based on the feature attribute of the training one-way matching feature vector. In order to integrate the comprehensive semantics constituted by different dynamic semantic expressions of the training one-way matching feature vector in real time without ignoring independent semantic changes, the independent features of the training one-way matching feature vector and the collective expression of its cluster representation are used as the total feature range and the partial feature range, and the stages of different dimensions of the secondary negative correlation of the feature cluster of the training one-way matching feature vector are adjusted to flexibly adjust the semantic change connection of the training one-way matching feature vector, thereby improving the overall semantic information expression integration of the feature cluster of the training one-way matching feature vector, improving the classification training speed of the training one-way matching feature vector through the noise component filter based on the Softmax function, that is, improving the training efficiency and the accuracy of the noise probability estimation, and being able to more accurately remove unnecessary RF echo time-frequency components to generate a noise-reduced RF echo signal, thereby ensuring the purity and quality of the received RF echo signal, which is conducive to the accurate transmission of information between different devices and increasing the reliability of the signal.

[0097] Based on the above embodiments, see Figure 4 As shown, it is a structural schematic diagram of a signal processing system 800 of a radio frequency transmitting and receiving module in an embodiment of the present application. The signal processing system 800 of the radio frequency transmitting and receiving module includes: a signal receiving module 810, which is used to receive radio frequency echo signals; a signal time-frequency local semantic segmentation module 820, which is used to perform signal time-frequency local semantic segmentation on the radio frequency echo signals to obtain a sequence of radio frequency echo local time-frequency diagrams; a time-frequency feature extraction module 830, which is used to extract time-frequency features from the sequence of radio frequency echo local time-frequency diagrams to obtain a sequence of radio frequency echo local time-frequency feature vectors; a one-way interactive matching module 840, which is used to extract the time-frequency features of each radio frequency echo local time-frequency feature vector in the sequence of radio frequency echo local time-frequency feature vectors. As a query feature vector, the query feature vector and the sequence of the local time-frequency feature vector of the RF echo are input into a one-way interactive matching network to obtain a sequence of one-way matching feature vectors; a screening module 850 is used to input each one-way matching feature vector in the sequence of the one-way matching feature vector into a noise component filter based on a Softmax function to obtain a sequence of noise component probability values; a noise reduction generation module 860 is used to perform feature filtering and signal filtering and restoration on the sequence of local time-frequency feature vectors of the echo based on the sequence of noise component probability values ​​to obtain a noise-reduced RF echo signal.

[0098] Those skilled in the art will appreciate that the specific functions and operations of each module in the signal processing system 800 of the radio frequency transmitting and receiving module have been described in detail above. Figure 2 to Figure 3The signal processing method of the RF transmitting and receiving module has been described in detail, and therefore, its repeated description will be omitted.

[0099] Figure 5 FIG. 1 is an application scenario diagram of the signal processing method of the radio frequency transmitting and receiving module according to the embodiment of the present application. Figure 5 As shown, in this application scenario, the RF signal is first transmitted by the RF transmitting module 300, and then the RF receiving module 400 obtains the RF echo signal, and then the RF echo signal is input into the server 100 which is equipped with the signal processing algorithm of the RF transmitting and receiving module, wherein the server can use the signal processing algorithm of the RF transmitting and receiving module to process the RF echo signal to obtain a noise-reduced RF echo signal.

[0100] Based on the above embodiments, another exemplary embodiment of an electronic device is also provided in the embodiments of the present application. In some possible implementations, the electronic device in the embodiments of the present application may include a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor can implement the steps of the signal processing method of the radio frequency transmitting and receiving module in the above embodiments when executing the program.

[0101] For example, in the case of electronic equipment Figure 1 Taking the server 100 in the example as an example, the processor in the electronic device is the processor 110 in the server 100, and the memory in the electronic device is the memory 120 in the server 100.

[0102] An embodiment of the present application also provides a computer-readable storage medium. Figure 6 1 shows a schematic diagram of a computer-readable storage medium 1000 according to an embodiment of the present application. Figure 6 As shown, the computer-readable storage medium 1000 stores computer-executable instructions 1001. When the computer-executable instructions 1001 are executed by the processor, the signal processing method of the radio frequency transmitting and receiving module according to the embodiment of the present application described with reference to the above figures can be executed. The computer-readable storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. The volatile memory may include, for example, a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may include, for example, a read-only memory (ROM), a hard disk, a flash memory, etc.

[0103] The embodiment of the present application also provides a computer program product or a computer program, which includes computer executable instructions, and the computer executable instructions are stored in a computer readable storage medium. The processor of the computer device reads the computer executable instructions from the computer readable storage medium, and the processor executes the computer executable instructions, so that the computer device executes the signal processing method of the radio frequency transmitting and receiving module according to the embodiment of the present application.

[0104] Those skilled in the art will appreciate that the contents disclosed in this application may be subject to various variations and improvements. For example, the various devices or components described above may be implemented by hardware, or by software, firmware, or a combination of some or all of the three.

[0105] Those skilled in the art will appreciate that all or part of the steps in the above method can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk or an optical disk. Optionally, all or part of the steps in the above embodiment can also be implemented using one or more integrated circuits. Accordingly, each module / unit in the above embodiment can be implemented in the form of hardware or in the form of software function modules. The present application is not limited to any particular form of combination of hardware and software.

[0106] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined as such herein.

[0107] The above is an explanation of the present application and should not be considered as a limitation thereof. Although several exemplary embodiments of the present application have been described, it will be readily understood by those skilled in the art that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present application. Therefore, all of these modifications are intended to be included within the scope of the present application as defined by the claims. It should be understood that the above is an explanation of the present application and should not be considered to be limited to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present application is defined by the claims and their equivalents.

Claims

1. A signal processing method for a radio frequency transmitting and receiving module, characterized in that: include: receiving a radio frequency echo signal; Performing signal time-frequency local semantic segmentation on the radio frequency echo signal to obtain a sequence of radio frequency echo local time-frequency graphs; Extracting time-frequency features from the sequence of local time-frequency diagrams of the radio frequency echoes to obtain a sequence of local time-frequency feature vectors of the radio frequency echoes; Taking each local time-frequency feature vector of the radio frequency echo in the sequence of local time-frequency feature vectors of the radio frequency echo as a query feature vector, inputting the query feature vector and the sequence of local time-frequency feature vectors of the radio frequency echo into a one-way interactive matching network to obtain a sequence of one-way matching feature vectors; Inputting each one-way matching feature vector in the sequence of one-way matching feature vectors into a noise component filter based on a Softmax function to obtain a sequence of noise component probability values; Based on the sequence of noise component probability values, feature filtering and signal filtering and restoration are performed on the sequence of echo local time-frequency feature vectors to obtain a noise-reduced radio frequency echo signal; The method of performing local time-frequency semantic segmentation on the radio frequency echo signal to obtain a sequence of local time-frequency graphs of the radio frequency echo includes: Performing wavelet analysis on the radio frequency echo signal to obtain a radio frequency echo time-frequency diagram; Performing super-pixel segmentation on the radio frequency echo time-frequency diagram using a super-pixel segmentation method to obtain a sequence of the radio frequency echo local time-frequency diagrams; The method comprises: taking each local time-frequency feature vector of the radio frequency echo in the sequence of the local time-frequency feature vector of the radio frequency echo as a query feature vector, inputting the query feature vector and the sequence of the local time-frequency feature vector of the radio frequency echo into a one-way interactive matching network to obtain a sequence of one-way matching feature vectors, including: Calculating the semantic association between each local time-frequency feature vector of the radio frequency echo in the sequence of local time-frequency feature vectors of the radio frequency echo and the query feature vector to obtain a sequence of local time-frequency query metric coefficients of the radio frequency echo; Normalizing the sequence of the radio frequency echo local time-frequency query metric coefficients to obtain a sequence of radio frequency echo local time-frequency one-way query matching weights; Taking each radio frequency echo local time-frequency one-way query matching weight in the sequence of radio frequency echo local time-frequency one-way query matching weights as a weighting coefficient, weighted optimization is performed on each radio frequency echo local time-frequency feature vector in the sequence of radio frequency echo local time-frequency feature vectors to obtain the sequence of one-way matching feature vectors; The step of calculating the semantic association between each local time-frequency feature vector of the radio frequency echo in the sequence of local time-frequency feature vectors of the radio frequency echo and the query feature vector to obtain a sequence of local time-frequency query metric coefficients of the radio frequency echo comprises: Calculating the logarithmic function value with base 2 of each local time-frequency feature vector of the radio frequency echo in the sequence of local time-frequency feature vectors of the radio frequency echo according to the position to obtain a sequence of local time-frequency representation vectors of the radio frequency echo; Calculating the point-by-point multiplication between each radio frequency echo local time-frequency representation vector in the sequence of radio frequency echo local time-frequency representation vectors and the query feature vector to obtain a sequence of radio frequency echo local time-frequency-query semantic interaction feature vectors; Calculating the global mean between the feature values ​​of each position of each radio frequency echo local time-frequency-query semantic interaction feature vector in the sequence of the radio frequency echo local time-frequency-query semantic interaction feature vector to obtain the sequence of the radio frequency echo local time-frequency query metric coefficients; The method of performing feature filtering and signal filtering restoration on the sequence of local time-frequency feature vectors of the echo based on the sequence of probability values ​​of the noise component to obtain a noise-reduced radio frequency echo signal includes: Based on the comparison between each noise component probability value in the sequence of noise component probability values ​​and a preset threshold, feature filtering is performed on the sequence of local time-frequency feature vectors of the radio frequency echo to obtain a sequence of filtered local time-frequency feature vectors of the radio frequency echo; The sequence of the local time-frequency feature vectors of the filtered radio frequency echo is input into a signal filtering restorer based on a generative adversarial network to obtain a noise-reduced radio frequency echo signal.

2. The signal processing method of the radio frequency transmitting and receiving module according to claim 1, characterized in that: The method comprises: performing time-frequency feature extraction on the sequence of local time-frequency diagrams of the radio frequency echo to obtain a sequence of local time-frequency feature vectors of the radio frequency echo, including: passing each local time-frequency diagram of the radio frequency echo in the sequence of local time-frequency diagrams of the radio frequency echo through a time-frequency feature extractor based on a hole convolutional neural network model to obtain a sequence of local time-frequency feature vectors of the radio frequency echo.

3. The signal processing method of the radio frequency transmitting and receiving module according to claim 2, characterized in that: Normalizing the sequence of the radio frequency echo local time-frequency query metric coefficients to obtain a sequence of radio frequency echo local time-frequency one-way query matching weights, including: Taking each radio frequency echo local time-frequency query metric coefficient in the sequence of radio frequency echo local time-frequency query metric coefficients as a power, calculating an exponential function value with a natural constant as a base to obtain a sequence of radio frequency echo local time-frequency query metric class support coefficients; Calculating the sum of each radio frequency echo local time-frequency query metric class support coefficient in the sequence of the radio frequency echo local time-frequency query metric class support coefficient to obtain the total value of the radio frequency echo local time-frequency query metric class support coefficient; The division between each RF echo local time-frequency query metric class support coefficient in the sequence of the RF echo local time-frequency query metric class support coefficient and the sum of the RF echo local time-frequency query metric class support coefficients is calculated to obtain the sequence of the RF echo local time-frequency one-way query matching weights.

4. The signal processing method of the radio frequency transmitting and receiving module according to claim 3, characterized in that: It also includes a training step for training the time-frequency feature extractor based on the hole convolutional neural network model, the one-way interactive matching network, the noise component filter based on the Softmax function, and the signal filtering restorer based on the adversarial generation network.

5. The signal processing method of the radio frequency transmitting and receiving module according to claim 4, characterized in that: The training step comprises: Acquire training data, wherein the training data includes a training radio frequency echo signal and a real noise-reduced radio frequency echo signal; Performing signal time-frequency local semantic segmentation on the training radio frequency echo signal to obtain a sequence of local time-frequency graphs of the training radio frequency echo; Passing each training radio frequency echo local time-frequency diagram in the sequence of training radio frequency echo local time-frequency diagrams through the time-frequency feature extractor based on the atrous convolutional neural network model to obtain a sequence of training radio frequency echo local time-frequency feature vectors; Taking each training radio frequency echo local time-frequency feature vector in the sequence of training radio frequency echo local time-frequency feature vectors as a training query feature vector, inputting the training query feature vector and the sequence of the training radio frequency echo local time-frequency feature vectors into the one-way interactive matching network to obtain a sequence of training one-way matching feature vectors; Inputting each training one-way matching feature vector in the sequence of training one-way matching feature vectors into the noise component filter based on the Softmax function to obtain a sequence of training noise component probability values; Based on the comparison between each training noise component probability value in the sequence of training noise component probability values ​​and a training preset threshold, feature filtering is performed on the sequence of training radio frequency echo local time-frequency feature vectors to obtain a sequence of training filtered radio frequency echo local time-frequency feature vectors; Inputting the sequence of local time-frequency feature vectors of the radio frequency echo after training filtering into the signal filtering restorer based on the adversarial generation network to obtain a training denoised radio frequency echo signal; Calculating a cross entropy loss function value between the training denoised radio frequency echo signal and a true denoised radio frequency echo signal to obtain a denoised radio frequency echo semantic cross entropy loss function value; Based on the semantic cross entropy loss function value of the denoised radio frequency echo, the time-frequency feature extractor based on the hole convolutional neural network model, the one-way interactive matching network, the noise component filter based on the Softmax function, and the signal filtering restorer based on the adversarial generation network are trained.

6. A signal processing system for a radio frequency transmitting and receiving module, wherein the signal processing system applies the signal processing method for a radio frequency transmitting and receiving module as claimed in claim 5, characterized in that: include: A signal receiving module, used for receiving a radio frequency echo signal; A signal time-frequency local semantic segmentation module, used for performing signal time-frequency local semantic segmentation on the radio frequency echo signal to obtain a sequence of radio frequency echo local time-frequency graphs; A time-frequency feature extraction module, used for extracting time-frequency features from the sequence of local time-frequency images of the radio frequency echo to obtain a sequence of local time-frequency feature vectors of the radio frequency echo; A one-way interactive matching module, used to use each radio frequency echo local time-frequency feature vector in the sequence of radio frequency echo local time-frequency feature vectors as a query feature vector, and input the query feature vector and the sequence of radio frequency echo local time-frequency feature vectors into a one-way interactive matching network to obtain a sequence of one-way matching feature vectors; A screening module, used for inputting each one-way matching feature vector in the sequence of one-way matching feature vectors into a noise component filter based on a Softmax function to obtain a sequence of noise component probability values; The noise reduction generation module is used to perform feature filtering and signal filtering and restoration on the sequence of echo local time-frequency feature vectors based on the sequence of noise component probability values ​​to obtain a noise-reduced radio frequency echo signal.

Citation Information

Patent Citations

  • Power line carrier noise testing device and method, equipment and storage medium

    CN116032324A

  • Wireless communication signal detection method

    CN116866129A