Non-contact radar tide observation wave absorbing method, device, equipment and medium

Abstract

The invention discloses a non-contact radar tide observation wave absorbing method, device and equipment and a medium, and relates to the field of tide observation, and the method comprises the steps: obtaining an original non-contact radar tide signal; de-noising processing is carried out on the original non-contact radar tidal signal to obtain a comprehensive feature vector; a pre-trained deep learning model is adopted to carry out wave absorption on the comprehensive feature vector, and a non-contact radar tidal signal after wave absorption is obtained; wherein the deep learning model comprises a first convolutional neural network, a second convolutional neural network, an attention mechanism layer, a third convolutional neural network and a full-connection network which are connected in sequence; and the output of the first convolutional neural network is also connected with the output residual error of the attention mechanism layer. According to the method, a traditional denoising method is combined with deep learning, and an attention mechanism and residual connection are introduced into the deep learning model, so that the precision of the non-contact radar tidal signal is improved, and the long-term stability and reliability of observation data are further guaranteed.

Description

Technical Field

[0001] The present application relates to the field of tidal observation, and in particular to a non-contact radar tidal observation wave elimination method, device, equipment and medium based on deep learning. Background Art

[0002] As an important part of marine environmental monitoring, tidal observation plays an indispensable and fundamental role in marine disaster warning, sea level change monitoring and marine engineering design. Accurate tidal observation data is not only an important basis for demarcating warning tide levels and formulating disaster prevention and mitigation measures, but also a core support for improving the accuracy of climate forecasts in coastal areas.

[0003] At present, there are various tidal observation technologies, mainly including float, pressure, acoustic, radar and laser tidal sensors. These technologies can be divided into two categories according to the contact with the sea surface: contact and non-contact. Among them, contact sensors have high observation accuracy and stability, but they rely on the physical structure of tide wells, and have high requirements for site and construction costs, which limits their widespread application in harsh marine environments. Non-contact observation technologies such as radar tide gauges have gradually become a research hotspot for tidal observation due to their advantages such as no need to build wells and easy installation. However, when facing complex marine environments, the accuracy of observation data of current non-contact observation technologies is easily affected by factors such as wave reflection and environmental noise, which reduces its reliability in practical applications.

[0004] A major challenge of non-contact radar tidal observation technology is how to effectively eliminate noise interference in the observation data. In a complex ocean environment, multipath reflection signals and wave interference blur the characteristics of tide signals, directly affecting the accuracy of tidal observation data. Most of the current signal processing methods rely on traditional filtering techniques or simple signal enhancement algorithms, but they are limited in effectiveness when faced with dynamic noise such as waves. At the same time, these methods are usually difficult to adapt to the characteristics of non-stationary signals, and cannot deeply mine the key feature information in radar observation data, resulting in insufficient data processing capabilities. Summary of the invention

[0005] The purpose of this application is to provide a non-contact radar tidal observation wave elimination method, device, equipment and medium, which can solve the problems of poor adaptability and low accuracy of traditional tidal observation technology under noise interference, thereby ensuring the long-term stability and reliability of observation data.

[0006] To achieve the above objectives, this application provides the following solutions: In a first aspect, the present application provides a non-contact radar tidal observation wave elimination method, comprising: Obtain raw non-contact radar tidal signals; De-noising the original non-contact radar tidal signal to obtain a comprehensive feature vector; Using a pre-trained deep learning model, the comprehensive feature vector is clipped to obtain a non-contact radar tidal signal after clipping; Among them, the deep learning model includes a first convolutional neural network, a second convolutional neural network, an attention mechanism layer, a third convolutional neural network and a fully connected network connected in sequence; the output of the first convolutional neural network is also connected to the output residual of the attention mechanism layer.

[0007] In a second aspect, the present application provides a non-contact radar tidal observation and wave elimination device, comprising: A signal acquisition module, used for acquiring original non-contact radar tidal signals; A denoising module, used for denoising the original non-contact radar tidal signal to obtain a comprehensive feature vector; A wave elimination module, used for eliminating the comprehensive feature vector by using a pre-trained deep learning model to obtain a non-contact radar tidal signal after wave elimination; Among them, the deep learning model includes a first convolutional neural network, a second convolutional neural network, an attention mechanism layer, a third convolutional neural network and a fully connected network connected in sequence; the output of the first convolutional neural network is also connected to the output residual of the attention mechanism layer.

[0008] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned non-contact radar tidal observation wave elimination method.

[0009] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned non-contact radar tidal observation and wave elimination method.

[0010] According to the specific embodiments provided by the present application, the present application has the following technical effects: The present application provides a non-contact radar tidal observation wave elimination method, device, equipment and medium, which obtains a comprehensive feature vector by denoising the original non-contact radar tidal signal, and eliminates the comprehensive feature vector in combination with a deep learning model, combines the traditional denoising method with deep learning, and improves the accuracy of the non-contact radar tidal signal after wave elimination. And introduces the attention mechanism and residual connection into the deep learning model, and utilizes its intelligent and high robustness advantages in feature extraction to achieve the key focus on effective signals and the suppression of redundant interference, thereby improving the confidence of feature extraction and the observation accuracy, and then ensuring the long-term stability and reliability of the observation data. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0012] Figure 1 This is a diagram of the application environment of a non-contact radar tidal observation and wave elimination method in one embodiment of the present application.

[0013] Figure 2 A schematic flow chart of a non-contact radar tidal observation and wave elimination method provided in one embodiment of the present application.

[0014] Figure 3 Schematic diagram of the signal denoising and wave elimination process in one embodiment of the present application.

[0015] Figure 4 A schematic diagram of the functional modules of a non-contact radar tidal observation and wave-breaking device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0017] Non-contact radar tidal observation plays an important role in ocean monitoring, but in complex ocean environments, radar tidal signals are susceptible to noise interference such as multipath effects and dynamic waves, resulting in data loss and background noise, which affects the continuity and accuracy of observations. In response to the above problems, this application proposes a non-contact radar tidal observation wave elimination method based on deep learning technology, combining traditional signal processing methods with deep learning models to optimize the completion, feature extraction and denoising process of radar tidal signals, effectively enhancing the accuracy and continuity of tidal observations, providing solid technical support for the intelligent development of ocean observation networks, and contributing theoretical basis and practical reference to the sustainable development of the marine economy and the improvement of disaster prevention and mitigation capabilities.

[0018] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0019] The non-contact radar tidal observation wave elimination method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the original non-contact radar tidal signal to the server 104. After the server 104 receives the original non-contact radar tidal signal, it denoises the original non-contact radar tidal signal to obtain a comprehensive feature vector; using a pre-trained deep learning model, the comprehensive feature vector is eliminated to obtain a non-contact radar tidal signal after elimination. The server 104 can feedback the obtained non-contact radar tidal signal after elimination to the terminal 102. In addition, in some embodiments, the non-contact radar tidal observation elimination method can also be implemented by the server 104 or the terminal 102 alone.

[0020] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart vehicle-mounted devices, etc. The portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers, or may be a cloud server.

[0021] In an exemplary embodiment, Figure 2 As shown, a non-contact radar tidal observation wave elimination method is provided, which is executed by a computer device, and can be specifically executed by a computer device such as a terminal or a server alone, or can be executed by a terminal and a server together. In the embodiment of the present application, the method is applied to Figure 1 The server 104 in the example is used for explanation, and the steps include the following steps 201 to 204.

[0022] Step 201, obtaining an original non-contact radar tidal signal.

[0023] Specifically, the original non-contact radar tidal signal includes multiple time points and the high and low tide heights corresponding to each time point, which are represented by a curve graph, with the horizontal axis being the time point and the vertical axis being the high and low tide heights.

[0024] Step 202: Use piecewise linear interpolation to fill missing values ​​in the original non-contact radar tidal signal. Filling missing values ​​ensures signal continuity.

[0025] In a specific application example, for any missing data point in the original non-contact radar tidal signal, according to the time interval of the time point corresponding to the missing data point, , the following formula is used to determine the high and low tide heights of the missing data points to fill the missing values ​​of the original non-contact radar tidal signal.

[0026] .

[0027] in, y is the high and low tide heights of the missing data points, and are the time points at both ends of the time interval corresponding to the missing data point. and They are the high and low tide heights at the two end points of the time interval corresponding to the time point of the missing data point.

[0028] Missing data points are time points where there is no high or low tide height in the original non-contact radar tide signal.

[0029] By determining the interval of adjacent known data points where each missing data point is located, the linear interpolation formula is used to calculate the ordinate value of the missing data point according to the abscissa values ​​of the two end points of the interval, thereby completing the missing data and ensuring the continuity of the signal. The normalization and standardization techniques are further used to uniformly process the amplitude of the signal.

[0030] Step 203: De-noising the original non-contact radar tidal signal to obtain a comprehensive feature vector.

[0031] In a specific application example, step 203 includes the following steps 31 to 34.

[0032] Step 31, low-pass filtering the original non-contact radar tidal signal to obtain a low-pass filtered signal. Specifically, the low-frequency characteristics of the original non-contact radar tidal signal are extracted by the low-pass filter, the high-frequency noise components are removed, and the background interference is suppressed.

[0033] Step 32, performing wavelet denoising on the original non-contact radar tidal signal to obtain a wavelet denoised signal. Specifically, the original non-contact radar tidal signal is decomposed into different frequency ranges using wavelet decomposition technology to achieve multi-scale time-frequency analysis.

[0034] Step 33: Perform short-time Fourier transform on the original non-contact radar tidal signal to obtain a time-frequency domain signal. The background noise is initially isolated and suppressed by short-time Fourier transform.

[0035] Step 34, concatenating the low-pass filtered signal, the wavelet denoised signal and the time-frequency domain signal to obtain a comprehensive feature vector.

[0036] Among them, the low-pass filtered signal, wavelet denoising signal and time-frequency domain signal are all curve graphs.

[0037] Step 204: Use a pre-trained deep learning model to clip the comprehensive feature vector to obtain a clipped non-contact radar tidal signal, and then perform tidal observation based on the clipped non-contact radar tidal signal.

[0038] Deep learning technology has shown great potential in solving non-stationary signal processing problems. The non-contact radar tidal observation wave elimination method based on deep learning can automatically extract the key features of the tidal signal by building a multi-layer neural network, effectively eliminating dynamic wave interference and environmental noise. Compared with traditional methods, deep learning technology can not only adapt to complex observation environments, but also realize the intelligent fusion of multi-dimensional features, and improve the robustness and accuracy of data processing. In addition, the deep learning model has good adaptive ability and can be optimized and adjusted for different observation scenarios, providing a guarantee for the long-term stability of tidal observation data.

[0039] This application targets non-contact radar tidal signals in complex backgrounds. By combining low-pass filtering, wavelet denoising, short-time Fourier transform and deep learning models, it can decompose and suppress the background noise of non-contact radar tidal signals, and solve the problem of decreased accuracy of tidal signals in complex environments such as dynamic waves, multipath reflections and random noise.

[0040] Among them, Figure 3 As shown, the deep learning model includes a first convolutional neural network, a second convolutional neural network, an attention mechanism layer, a third convolutional neural network and a fully connected network connected in sequence. The output of the first convolutional neural network is also connected to the output residual of the attention mechanism layer.

[0041] Specifically, the residual noise is refined and optimized through the deep learning model, and the high-dimensional features of the signal are extracted through the convolutional neural network. At the same time, the residual connection mechanism is introduced to enhance the training efficiency of the deep network and alleviate the problem of gradient disappearance. The attention mechanism is further introduced to dynamically weight the feature channels, highlight the key feature areas, and suppress irrelevant interference signals. Through the joint analysis of multi-dimensional information, the robustness and comprehensiveness of the non-contact radar tidal signal are significantly improved, and a non-contact radar tidal signal with higher signal-to-noise ratio and richer effective features is generated.

[0042] In a specific application example, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network are all 3×3 convolution operations, and the attention mechanism layer is a compression and excitation module (Squeeze-and-ExcitationBlock, SE-Block). SE-Block is a lightweight attention mechanism that enhances important channels in the feature map by adaptively adjusting the weights of each channel and suppresses irrelevant channels. The basic idea of ​​SE-Block is to compress the spatial information of each channel into a single value through global average pooling to obtain a global feature description of each channel. The weights of each channel are learned through a fully connected network, and each channel is weighted and adjusted. The feature map is readjusted using weighted information to enhance the features of important channels and suppress unimportant channels.

[0043] When training the deep learning model, the signal-to-noise ratio and root mean square error are used to quantitatively analyze the performance of the deep learning model.

[0044] This application further combines simulation data and measured data to conduct a comprehensive evaluation of various background noise conditions. In the simulation experiment, the MATLAB platform is used to simulate various complex noise scenarios to verify the performance of the deep learning model in signal denoising and feature extraction. The signal-to-noise ratio and root mean square error are used as the main evaluation indicators to quantitatively analyze the performance of the deep learning model.

[0045] The signal-to-noise ratio is the ratio of signal power to noise power, usually expressed in decibels, and is used to measure the strength of the signal relative to the noise. The signal-to-noise ratio can indicate the clarity of the signal compared to the background noise, and the higher the value, the better. The formula is: ;in, is the signal-to-noise ratio, is the signal power, is the noise power.

[0046] The root mean square error is used to measure the deviation between the result of signal processing and the original signal, reflecting the size of the error. The root mean square error represents the error of the signal, and the smaller the value, the better. Its formula is: ;in, is the root mean square error, N is the number of data points, is the first wave in the non-contact radar tidal signal after wave elimination. j The value of the data point, The first j The value of the data point.

[0047] In terms of model optimization, this application improves the generalization ability and processing efficiency of the deep learning model by adjusting hyperparameters such as learning rate and network depth and combining techniques such as cross-validation. The optimized deep learning model shows higher robustness and can achieve efficient tidal signal analysis and processing under a variety of observation conditions.

[0048] This application introduces residual connections and attention mechanisms to train deep learning models to further improve the robustness and denoising effects of signal processing. The expected results will significantly improve the signal processing capabilities of non-contact radar tidal observation technology and provide a theoretical basis and technical demonstration for the intelligent development of ocean monitoring.

[0049] Compared with the traditional wave elimination method, the beneficial effects of the present application include the following points.

[0050] (1) In a complex ocean environment, non-contact radar tidal signals are susceptible to interference from waves, multipath reflections, and random noise, which results in reduced signal integrity and observation accuracy. To address this problem, this application uses an interpolation method in the data preprocessing phase to complete the missing data and convert it into a continuous vector form that is suitable for the deep learning model, thereby ensuring signal integrity and the adaptability of the input model.

[0051] (2) Traditional wavelet elimination methods are mostly based on second-order statistics to extract low-order features such as time-frequency spectrum or power spectrum, which makes it difficult to deal with complex background noise and fully and accurately describe the multi-dimensional information characteristics of non-contact radar signals. This application combines the wavelet decomposition characteristics of traditional filtering methods with the high-dimensional feature extraction capabilities of deep learning. Through the synchronous fusion of low-order and high-order information, it fully captures the multi-dimensional information of the signal in the time domain, frequency domain and space domain, and improves the feature extraction accuracy after signal elimination.

[0052] (3) The actual observation data samples of non-contact radar tidal signals often contain strong background noise and irrelevant redundant signals. These interference factors will significantly reduce the accuracy and reliability of feature extraction. This application introduces the residual connection and attention mechanism in the deep learning model, and uses its intelligence and high robustness advantages in feature extraction to achieve the focus on effective signals and suppress redundant interference, thereby improving the high confidence of feature extraction and observation accuracy.

[0053] (4) The adaptability and generalization ability of traditional wave-breaking algorithms under multiple environmental conditions are weak, and it is difficult to ensure the observation accuracy and stability in different scenarios. This application introduces strategies such as learning rate adjustment and hyperparameter optimization through the optimization link of the deep learning model, combines simulation experiments with measured data to verify the model performance, and constructs a deep learning model with strong adaptability to ensure its universality and reliability in different marine environments.

[0054] In summary, this application constructs a set of efficient and intelligent non-contact radar tidal observation wave elimination methods, providing a new technical path and theoretical support for improving the processing efficiency and observation accuracy of tidal signals.

[0055] Based on the same inventive concept, the embodiment of the present application also provides a non-contact radar tidal observation wave elimination device for implementing the non-contact radar tidal observation wave elimination method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more non-contact radar tidal observation wave elimination device embodiments provided below can refer to the limitations of the non-contact radar tidal observation wave elimination method above, and will not be repeated here.

[0056] In an exemplary embodiment, Figure 4 As shown, a non-contact radar tidal observation and wave elimination device is provided, including: a signal acquisition module 401, a preprocessing module 402, a denoising module 403 and a wave elimination module 404.

[0057] The signal acquisition module 401 is used to acquire the original non-contact radar tidal signal.

[0058] The preprocessing module 402 is used to fill missing values ​​of the original non-contact radar tidal signal using a piecewise linear interpolation method.

[0059] The denoising module 403 is used to perform denoising processing on the original non-contact radar tidal signal to obtain a comprehensive feature vector.

[0060] The wave clipping module 404 is used to use a pre-trained deep learning model to clip the comprehensive feature vector to obtain a non-contact radar tidal signal after wave clipping.

[0061] The deep learning model includes a first convolutional neural network, a second convolutional neural network, an attention mechanism layer, a third convolutional neural network and a fully connected network connected in sequence. The output of the first convolutional neural network is also connected to the output residual of the attention mechanism layer.

[0062] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0063] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0064] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0065] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0066] In this application, all actions to obtain signals, information or data are carried out in compliance with the relevant data protection laws and policies of the country where they are located and with the authorization given by the owner of the corresponding device.

[0067] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0068] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0069] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0070] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A non-contact radar tidal observation wave elimination method, characterized in that: The non-contact radar tidal observation wave elimination method comprises: Obtain raw non-contact radar tidal signals; De-noising the original non-contact radar tidal signal to obtain a comprehensive feature vector; Using a pre-trained deep learning model, the comprehensive feature vector is clipped to obtain a non-contact radar tidal signal after clipping; Among them, the deep learning model includes a first convolutional neural network, a second convolutional neural network, an attention mechanism layer, a third convolutional neural network and a fully connected network connected in sequence; the output of the first convolutional neural network is also connected to the output residual of the attention mechanism layer.

2. The non-contact radar tidal observation and wave elimination method according to claim 1, characterized in that: Before the original non-contact radar tidal signal is subjected to denoising, the non-contact radar tidal observation wave elimination method further includes: The piecewise linear interpolation method is used to fill the missing values ​​of the original non-contact radar tidal signal.

3. The non-contact radar tidal observation and wave elimination method according to claim 2, characterized in that: The original non-contact radar tidal signal includes a plurality of time points and the high and low tide heights corresponding to each time point; The missing values ​​of the original non-contact radar tidal signal are filled using a piecewise linear interpolation method, specifically including: For any missing data point in the original non-contact radar tidal signal, the high and low tide height of the missing data point is determined by the following formula according to the time interval of the time point corresponding to the missing data point, so as to fill the missing value of the original non-contact radar tidal signal: ； in, y is the high and low tide heights of the missing data points, and are the time points at both ends of the time interval corresponding to the missing data point. and They are the high and low tide heights at the two end points of the time interval corresponding to the time point of the missing data point.

4. The non-contact radar tidal observation and wave elimination method according to claim 1, characterized in that: The original non-contact radar tidal signal is subjected to denoising to obtain a comprehensive feature vector, which specifically includes: Performing low-pass filtering on the original non-contact radar tidal signal to obtain a low-pass filtered signal; Performing wavelet denoising on the original non-contact radar tidal signal to obtain a wavelet denoised signal; Performing short-time Fourier transform on the original non-contact radar tidal signal to obtain a time-frequency domain signal; The low-pass filtered signal, the wavelet denoised signal and the time-frequency domain signal are concatenated to obtain a comprehensive feature vector.

5. The non-contact radar tidal observation and wave elimination method according to claim 1, characterized in that: The attention mechanism layer is a compression and excitation module.

6. The non-contact radar tidal observation and wave elimination method according to claim 1, characterized in that: When training the deep learning model, the signal-to-noise ratio and root mean square error are used to quantitatively analyze the performance of the deep learning model.

7. The non-contact radar tidal observation and wave elimination method according to claim 1, characterized in that: The non-contact radar tidal observation and wave elimination method also includes: Tidal observation is carried out based on the non-contact radar tidal signal after wave elimination.

8. A non-contact radar tidal observation wave elimination device, applied to the non-contact radar tidal observation wave elimination method according to any one of claims 1 to 7, characterized in that: The non-contact radar tidal observation and wave elimination device comprises: A signal acquisition module, used for acquiring original non-contact radar tidal signals; A denoising module, used for denoising the original non-contact radar tidal signal to obtain a comprehensive feature vector; A wave elimination module, used for eliminating the comprehensive feature vector by using a pre-trained deep learning model to obtain a non-contact radar tidal signal after wave elimination; Among them, the deep learning model includes a first convolutional neural network, a second convolutional neural network, an attention mechanism layer, a third convolutional neural network and a fully connected network connected in sequence; the output of the first convolutional neural network is also connected to the output residual of the attention mechanism layer.

9. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the non-contact radar tidal observation wave elimination method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the non-contact radar tidal observation wave elimination method described in any one of claims 1 to 7 is implemented.

Images