Abnormity detection method and device for maritime antenna attenuator, and storage medium

Abstract

The invention discloses an anomaly detection method and device for a maritime antenna attenuator and a storage medium. The method comprises the steps of setting a maintenance node for the attenuator according to communication operation data of a maritime antenna on a ship; when a maintenance node is reached, inputting a sample test signal into an attenuator, and sequentially outputting a first attenuation test signal, a second attenuation test signal and a third attenuation test signal; inputting the sample test signal into a first feature network, and extracting a sample signal feature in a time domain and / or a frequency domain; inputting the first attenuation test signal, the second attenuation test signal and the third attenuation test signal into a second feature network, and jointly extracting attenuation signal features in a time domain and / or a frequency domain; inputting the sample signal features and the attenuation signal features into a classification network to identify the state abnormity probability of the attenuator; and detecting whether the state of the attenuator is abnormal according to the probability. According to the embodiment, the attenuator state detection efficiency can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of marine antennas, and in particular, to an abnormal detection method, device, and storage medium for a marine antenna attenuator. Background Art

[0002] An attenuator is configured in a marine antenna. The attenuator is usually composed of basic electronic components such as resistors, inductors, and capacitors, and can be designed in a form with adjustable attenuation in the marine antenna. It automatically adjusts the attenuation according to an external control signal (such as voltage or current), and adjusts the signal level according to the attenuation.

[0003] During the long-term operation of the attenuator, abnormalities may occur. Therefore, professionals are arranged to carry professional equipment to overhaul the marine antenna at regular intervals. The overhaul method mainly uses professional equipment to measure the transmit power and receive power of the marine antenna, and evaluates the state of the attenuator based on the transmit power and receive power.

[0004] However, this overhaul mode is relatively cumbersome and inefficient, resulting in a low overhaul frequency. There are performance hidden dangers in the attenuator, leading to safety risks in navigation. Summary of the Invention

[0005] In view of this, the present invention provides an abnormal detection method, device, and storage medium for a marine antenna attenuator to improve the overhaul efficiency of the marine antenna attenuator.

[0006] The first aspect of the present invention provides an abnormal detection method for a marine antenna attenuator, including:

[0007] Collect communication operation data of the marine antenna on the ship; the marine antenna has an attenuator;

[0008] Set an overhaul node for the attenuator according to the communication operation data;

[0009] When the overhaul node is reached, input a sample test signal into the attenuator, and sequentially output a first attenuation test signal at a first attenuation position, a second attenuation test signal at a second attenuation position, and a third attenuation test signal at a third attenuation position;

[0010] Determine an antenna attenuation detection model; the antenna attenuation detection model includes a first feature network, a second feature network, and a classification network;

[0011] Input the sample test signal into the first feature network, and extract sample signal features in the time domain and / or frequency domain;

[0012] Input the first attenuation test signal, the second attenuation test signal, and the third attenuation test signal into the second feature network to jointly extract attenuation signal features in the time domain and / or frequency domain;

[0013] Input the sample signal features and the attenuation signal features into the classification network to identify the probability of abnormal state of the attenuator;

[0014] Detect whether the state of the attenuator is abnormal according to the probability.

[0015] The second aspect of the present invention provides an abnormal detection device for a marine antenna attenuator, including:

[0016] A communication operation data acquisition module for collecting communication operation data of a marine antenna on a ship; the marine antenna has an attenuator;

[0017] An overhaul node setting module for setting overhaul nodes for the attenuator according to the communication operation data;

[0018] A test signal attenuation module for inputting a sample test signal into the attenuator when reaching the overhaul node, and sequentially outputting a first attenuation test signal at a first attenuation position, a second attenuation test signal at a second attenuation position, and a third attenuation test signal at a third attenuation position;

[0019] A model determination module for determining an antenna attenuation detection model; the antenna attenuation detection model includes a first feature network, a second feature network, and a classification network;

[0020] A sample signal feature extraction module for inputting the sample test signal into the first feature network to extract sample signal features in the time domain and / or frequency domain;

[0021] An attenuation signal feature extraction module for inputting the first attenuation test signal, the second attenuation test signal, and the third attenuation test signal into the second feature network to jointly extract attenuation signal features in the time domain and / or frequency domain;

[0022] An abnormal probability identification module for inputting the sample signal features and the attenuation signal features into the classification network to identify the probability of abnormal state of the attenuator;

[0023] A state detection module for detecting whether the state of the attenuator is abnormal according to the probability.

[0024] The third aspect of the present invention provides an electronic device, the electronic device includes:

[0025] At least one processor; and

[0026] A memory communicatively connected to the at least one processor; wherein,

[0027] The memory stores a computer program executable by the at least one processor, and when the computer program is executed by the at least one processor, the at least one processor is enabled to execute the abnormal detection method of the maritime antenna attenuator as described in the first aspect above.

[0028] A fourth aspect of the present invention provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the abnormal detection method of the maritime antenna attenuator as described in the first aspect above is implemented.

[0029] A fifth aspect of the present invention provides a computer program product including a computer program, and when the computer program is executed by a processor, the abnormal detection method of the maritime antenna attenuator as described in the first aspect above is implemented.

[0030] In this embodiment, communication operation data of a maritime antenna on a ship is collected; an attenuator is provided in the maritime antenna; maintenance nodes are set for the attenuator according to the communication operation data; when the maintenance node is reached, a sample test signal is input into the attenuator, and a first attenuation test signal is output at a first attenuation position, a second attenuation test signal is output at a second attenuation position, and a third attenuation test signal is output at a third attenuation position in sequence; an antenna attenuation detection model is determined; the antenna attenuation detection model includes a first feature network, a second feature network, and a classification network; the sample test signal is input into the first feature network, and sample signal features are extracted in the time domain and / or frequency domain; the first attenuation test signal, the second attenuation test signal, and the third attenuation test signal are input into the second feature network, and attenuation signal features are jointly extracted in the time domain and / or frequency domain; the sample signal features and the attenuation signal features are input into the classification network to identify the probability of abnormal state of the attenuator; the state of the attenuator is detected according to the probability. In this embodiment, a comparison set is constructed for the attenuation of the signal by the attenuator at different attenuation positions in the maritime antenna, and the features are decomposed in the time domain and / or frequency domain, which can amplify the small fluctuations of the signal in the time domain and frequency domain, ensure the accuracy of detecting the state of the attenuator, realize the self-check of the attenuator, and the self-check process has a high degree of automation, convenient operation, and can effectively improve the efficiency of detecting the state of the attenuator.

[0031] Moreover, the self-check nodes are adaptively set according to the communication operation conditions of the maritime antenna, the maintenance frequency is reasonably increased, and the performance hidden dangers of the attenuator are effectively eliminated while minimizing the impact on the maritime antenna as much as possible, ensuring the safety of navigation.

[0032] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0034] Figure 1 is a flowchart of a method for detecting anomalies of a maritime antenna attenuator provided in Embodiment 1 of the present invention.

[0035] Figure 2 is a schematic structural diagram of an antenna attenuation detection model provided in Embodiment 1 of the present invention.

[0036] Figure 3 is a schematic structural diagram of a device for detecting anomalies of a maritime antenna attenuator provided in Embodiment 2 of the present invention.

[0037] Figure 4 is a schematic structural diagram of an electronic device provided in Embodiment 3 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0038] In order to enable those skilled in the art to better understand the solutions of the present invention, the following clearly and completely describes the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, rather than all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0039] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that such used data can be interchanged under appropriate circumstances so that the embodiments of the present invention described herein can cover sequences other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0040] Embodiment 1

[0041] Refer to Figure 1 , which shows a flowchart of an abnormal detection method for a maritime antenna attenuator provided in Embodiment 1 of the present invention. This method can be executed by an abnormal detection device for a maritime antenna attenuator. The abnormal detection device for a maritime antenna attenuator can be implemented in the form of hardware and / or software, and the abnormal detection device for a maritime antenna attenuator can be configured in an electronic device. As Figure 1 shown, the method includes:

[0042] Step 101: Collect communication operation data for the maritime antenna on the ship.

[0043] In this embodiment, the log file recorded for the maritime antenna can be read from the edge computing node of the ship, and the communication operation data generated by the maritime antenna during communication can be read from the log file.

[0044] Step 102: Set a maintenance node for the attenuator according to the communication operation data.

[0045] There is an attenuator in the maritime antenna, and the computing amount of the attenuator can be roughly evaluated from the overall communication operation data.

[0046] In this embodiment, a self-check mode is provided for the attenuator in the maritime antenna. Each time the self-check mode is applied to self-check the attenuator in the maritime antenna, the attenuator in the maritime antenna will temporarily stop providing external services. Therefore, each time the attenuator in the maritime antenna is self-checked, the next maintenance node for the attenuator in the maritime antenna can be set according to the communication operation data in terms of time.

[0047] In a specific implementation, after querying the communication operation data for the intensity of the received signal during each communication after the last self-check of the attenuator, the intensity is compared with a preset first threshold.

[0048] If the intensity is greater than the preset first threshold, it indicates that the computing amount of the attenuator is relatively high and there is a certain loss, then the communication frequency is incremented by 1.

[0049] Generate an adjustment coefficient according to the communication frequency; among them, the adjustment coefficient is negatively correlated with the communication frequency, that is, the higher the communication frequency, the higher the loss of the attenuator, the smaller the adjustment coefficient, so that the maintenance node is closer. On the contrary, the lower the communication frequency, the lower the loss of the attenuator, the larger the adjustment coefficient, so that the maintenance node is farther.

[0050] Exemplarily, the adjustment coefficient is:

[0051] H = min(2 / e 2x , α);

[0052] Wherein, H is an adjustment coefficient, x is the communication frequency, α is a preset upper limit value, e is the natural number, and min is the minimum value function.

[0053] In this example, α ensures the lowest self-check frequency. Initially, 2 / e 2x ≥α. Based on α, the maintenance nodes are determined. As the communication frequency increases, 2 / e 2x <α, and the self-check frequency increases rapidly to better adapt to the change in the operation volume of the attenuator.

[0054] Multiply the adjustment coefficient by the preset unit time to obtain the maintenance node for the next maintenance of the attenuator.

[0055] Of course, in addition to setting the self-start of the maintenance node to perform self-check on the attenuator of the maritime antenna, relevant users (such as the owner or lessee of the ship, etc.) can also use the application program provided by the manufacturer of the maritime antenna to actively start the self-check on the attenuator of the maritime antenna. This embodiment does not limit this.

[0056] Step 103: When the maintenance node is reached, input the sample test signal into the attenuator, and output the first attenuation test signal at the first attenuation position, the second attenuation test signal at the second attenuation position, and the third attenuation test signal at the third attenuation position in sequence.

[0057] When the time reaches the next maintenance node, a self-maintenance prompt message can be generated into the application program provided by the manufacturer of the maritime antenna. Relevant users (such as the owner or lessee of the ship, etc.) browse the prompt message and confirm. At this time, start the self-check on the attenuator of the maritime antenna.

[0058] During the self-check process, the sample test signal can be input into the attenuator by an external device (such as a shore base station, the maritime antenna of other ships, a tester, etc.) or an internal device (such as configuring a dedicated generator in the maritime antenna). The attenuator adjusts the attenuation amount to attenuate the sample test signal.

[0059] Among them, there is no specific requirement for the waveform of the sample test signal during self-check, and the signal of conventional communication can be used.

[0060] As Figure 2 shown, the sample test signal S1 is attenuated at the first attenuation position to obtain the first attenuation test signal E11, the sample test signal S1 is attenuated at the second attenuation position to obtain the second attenuation test signal E21, and the sample test signal S1 is attenuated at the third attenuation position to obtain the third attenuation test signal E31.

[0061] Generally, the first attenuation position, the second attenuation position, and the third attenuation position cover the attenuation range of the attenuator to comprehensively detect the performance of the attenuator.

[0062] Among them, the first attenuation bit tends to the upper limit value of the attenuation range, such as 90%, the second attenuation bit tends to the middle value of the attenuation range, such as 50%, and the third attenuation bit tends to the lower limit value of the attenuation range, such as 10%.

[0063] Step 104: Determine the antenna attenuation detection model.

[0064] In this embodiment, the antenna attenuation detection model can be pre-constructed, trained, and verified based on deep learning. The antenna attenuation detection model belongs to a binary classification network, which can reduce the generalization requirements of the antenna attenuation detection model, reduce the structural complexity of the antenna attenuation detection model, and ensure a certain self-checking accuracy, facilitating operation on the edge computing nodes on the ship.

[0065] Using the historical sample test signals, the first attenuation test signal, the second attenuation test signal, and the third attenuation test signal as samples, and the operating states (normal, abnormal) labeled for the attenuator as the label Label, and using binary cross-entropy as the loss function, train and verify the antenna attenuation detection model so that the antenna attenuation detection model has the ability to self-check the attenuator in the maritime antenna.

[0066] Step 105: Input the sample test signal into the first feature network to extract sample signal features in the time domain and / or frequency domain.

[0067] The antenna attenuation detection model includes a first feature network, and the first feature network is responsible for extracting features from the sample test signal in the time domain and / or frequency domain. Then, during self-checking, the sample test signal can be input into the first feature network to extract features from the sample test signal in the time domain and / or frequency domain, denoted as sample signal features.

[0068] In an embodiment of the present invention, as Figure 2 shown, the first feature network includes a first time-domain convolutional layer ConvT_1, a first frequency-domain convolutional layer ConvP_1, a first time-frequency convolutional layer ConvTP_1, and a first convolutional layer Conv_1.

[0069] Among them, the convolution kernel in the first time-domain convolutional layer ConvT_1 has a dimension of n1 in the time domain direction and a dimension of 1 in the frequency domain direction, the convolution kernel in the first frequency-domain convolutional layer ConvP_1 has a dimension of 1 in the time domain direction and a dimension of m1 in the frequency domain direction, and the convolution kernel in the first time-frequency convolutional layer ConvTP_1 has a dimension of n1 in the time domain direction and a dimension of m1 in the frequency domain direction.

[0070] In this embodiment, perform continuous wavelet transform (Continuous Wavelet Transform, CWT) on the sample test signal S1 to obtain the sample time-frequency signal S2.

[0071] Input the sample time-frequency signal S2 into the first time-domain convolutional layer ConvT_1 to perform a convolutional operation, and extract the sample time-domain features in the time domain.

[0072] Input the sample time-frequency signal S2 into the first frequency-domain convolutional layer ConvP_1 to perform a convolutional operation, and extract the sample frequency-domain features in the frequency domain.

[0073] Input the sample time-frequency signal S2 into the first time-frequency convolutional layer ConvTP_1 to perform a convolutional operation, and extract the sample time-frequency features in the time domain and the frequency domain.

[0074] Concatenate the sample time-domain features, sample frequency-domain features, and sample time-frequency features into a sample composite feature.

[0075] Input the sample composite feature into the first convolutional layer Conv_1 to perform a convolutional operation, and extract the sample signal features.

[0076] In this embodiment, decomposing the sample time-frequency signal in the time domain, frequency domain, and both the time domain and the frequency domain can fully exploit the features of the sample time-frequency signal, which helps to improve the accuracy of self-checking.

[0077] Step 106: Input the first attenuation test signal, the second attenuation test signal, and the third attenuation test signal into the second feature network to jointly extract attenuation signal features in the time domain and / or the frequency domain.

[0078] If the antenna attenuation detection model includes a second feature network, and the second feature network is responsible for extracting features from the first attenuation test signal, the second attenuation test signal, and the third attenuation test signal in the time domain and / or the frequency domain, then, during self-checking, the first attenuation test signal, the second attenuation test signal, and the third attenuation test signal can be input into the second feature network to jointly extract features from the first attenuation test signal, the second attenuation test signal, and the third attenuation test signal in the time domain and / or the frequency domain, denoted as attenuation signal features.

[0079] In one embodiment of the present invention, as Figure 2 shown, the second feature network includes a first interaction module, a second interaction module, a third interaction module, and a second convolutional layer Conv_2; then, in this embodiment, step 106 may include the following steps:

[0080] Step 1061: Perform continuous wavelet transforms on the first attenuation test signal, the second attenuation test signal, and the third attenuation test signal respectively to obtain the first attenuation time-frequency signal, the second attenuation time-frequency signal, and the third attenuation time-frequency signal.

[0081] In this embodiment, as Figure 2As shown, perform continuous wavelet transform (CWT) on the first attenuation test signal E11 to obtain the first attenuation time-frequency signal E12, perform continuous wavelet transform (CWT) on the second attenuation test signal E21 to obtain the second attenuation time-frequency signal E22, and perform continuous wavelet transform (CWT) on the third attenuation test signal E31 to obtain the third attenuation time-frequency signal E32.

[0082] Step 1062: Input the first attenuation time-frequency signal, the second attenuation time-frequency signal, and the third attenuation time-frequency signal into the first interaction module, and jointly extract interaction time-domain features in the time domain.

[0083] The first interaction module is responsible for extracting features from the first attenuation time-frequency signal, the second attenuation time-frequency signal, and the third attenuation time-frequency signal in the time domain. Then, during self-check, the first attenuation time-frequency signal, the second attenuation time-frequency signal, and the third attenuation time-frequency signal can be input into the first interaction module to jointly extract features from the first attenuation time-frequency signal, the second attenuation time-frequency signal, and the third attenuation time-frequency signal in the time domain, denoted as interaction time-domain features.

[0084] In one design, as Figure 2 shown, the first interaction module includes a second time-domain convolutional layer ConvT_2, a third time-domain convolutional layer ConvT_3, a fourth time-domain convolutional layer ConvT_4, and a third convolutional layer Conv_3.

[0085] Among them, the convolution kernel in the second time-domain convolutional layer ConvT_2 has a dimension of n2 in the time domain and a dimension of 1 in the frequency domain, the convolution kernel in the third time-domain convolutional layer ConvT_3 has a dimension of n3 in the time domain and a dimension of 1 in the frequency domain, and the convolution kernel in the fourth time-domain convolutional layer ConvT_4 has a dimension of n4 in the time domain and a dimension of 1 in the frequency domain.

[0086] In this design, input the first attenuation time-frequency signal E12 into the second time-domain convolutional layer ConvT_2 to extract the first attenuation time-domain features in the time domain.

[0087] Input the second attenuation time-frequency signal E22 into the third time-domain convolutional layer ConvT_3 to perform a convolution operation and extract the second attenuation time-domain features in the time domain.

[0088] Input the third attenuation time-frequency signal E33 into the fourth time-domain convolutional layer ConvT_4 to perform a convolution operation and extract the third attenuation time-domain features in the time domain.

[0089] Concatenate the first attenuation time-domain features, the second attenuation time-domain features, and the third attenuation time-domain features Concat into the fourth attenuation time-domain features.

[0090] Input the fourth decaying time-domain feature into the third convolutional layer Conv_3 to perform a convolutional operation and extract the interactive time-domain feature.

[0091] In this embodiment, when decomposing and mining the time-domain features of the first decaying time-frequency signal, the second decaying time-frequency signal, and the third decaying time-frequency signal, the time-domain features of the first decaying time-frequency signal, the second decaying time-frequency signal, and the third decaying time-frequency signal are aligned in the time domain, which is convenient for comparing the time-domain features within the first decaying time-frequency signal, the second decaying time-frequency signal, and the third decaying time-frequency signal, and is also convenient for comparing with the time-domain features of the sample time-frequency signal, so as to amplify the small features with fluctuations in the time domain.

[0092] Step 1063: Input the first decaying time-frequency signal, the second decaying time-frequency signal, and the third decaying time-frequency signal into the second interactive module to jointly extract the interactive frequency-domain feature in the frequency domain.

[0093] Since the second interactive module is responsible for extracting features from the first decaying time-frequency signal, the second decaying time-frequency signal, and the third decaying time-frequency signal in the frequency domain, during self-checking, the first decaying time-frequency signal, the second decaying time-frequency signal, and the third decaying time-frequency signal can be input into the second interactive module to jointly extract features from the first decaying time-frequency signal, the second decaying time-frequency signal, and the third decaying time-frequency signal in the frequency domain, which is denoted as the interactive frequency-domain feature.

[0094] In one design, as Figure 2 shown, the second interactive module includes the second frequency-domain convolutional layer ConvP_2, the third frequency-domain convolutional layer ConvP_3, the fourth frequency-domain convolutional layer ConvP_4, and the fourth convolutional layer Conv_4.

[0095] Among them, the convolution kernel in the second frequency-domain convolutional layer ConvP_2 has a dimension of 1 in the time domain and a dimension of m2 in the frequency domain, the convolution kernel in the third frequency-domain convolutional layer ConvP_3 has a dimension of 1 in the time domain and a dimension of m3 in the frequency domain, and the convolution kernel in the fourth frequency-domain convolutional layer ConvP_4 has a dimension of 1 in the time domain and a dimension of m4 in the frequency domain.

[0096] In this design, input the first decaying time-frequency signal E12 into the second frequency-domain convolutional layer ConvP_2 to perform a convolutional operation and extract the first decaying frequency-domain feature in the frequency domain.

[0097] Input the second decaying time-frequency signal E22 into the third frequency-domain convolutional layer ConvP_3 to perform a convolutional operation and extract the second decaying frequency-domain feature in the frequency domain.

[0098] Input the third decaying time-frequency signal E32 into the fourth frequency-domain convolutional layer ConvP_4 to perform a convolutional operation and extract the third decaying frequency-domain feature in the frequency domain.

[0099] Concatenate the first attenuation frequency-domain feature, the second attenuation frequency-domain feature, and the third attenuation frequency-domain feature into a fourth attenuation frequency-domain feature.

[0100] Input the fourth attenuation frequency-domain feature into the fourth convolutional layer Conv_4 to perform a convolution operation and extract the interactive frequency-domain feature.

[0101] In this embodiment, when decomposing and mining the features in the frequency domain of the first attenuation time-frequency signal, the second attenuation time-frequency signal, and the third attenuation time-frequency signal, align the features in the frequency domain of the first attenuation time-frequency signal, the second attenuation time-frequency signal, and the third attenuation time-frequency signal, which is convenient for comparing the features in the frequency domain within the first attenuation time-frequency signal, the second attenuation time-frequency signal, and the third attenuation time-frequency signal, and is also convenient for comparing the features in the frequency domain with the sample time-frequency signal, and amplifies the small features with fluctuations in the frequency domain.

[0102] Step 1064: Input the first attenuation time-frequency signal, the second attenuation time-frequency signal, and the third attenuation time-frequency signal into the third interaction module to jointly extract the interactive time-frequency features in the time domain and the frequency domain.

[0103] The third interaction module is responsible for extracting features from the first attenuation time-frequency signal, the second attenuation time-frequency signal, and the third attenuation time-frequency signal in the time domain and the frequency domain. Then, during self-checking, the first attenuation time-frequency signal, the second attenuation time-frequency signal, and the third attenuation time-frequency signal can be input into the third interaction module to jointly extract features from the first attenuation time-frequency signal, the second attenuation time-frequency signal, and the third attenuation time-frequency signal in the time domain and the frequency domain, denoted as the interactive time-frequency features.

[0104] In one design, as Figure 2 shown, the third interaction module includes the second time-frequency convolutional layer ConvTP_2, the third time-frequency convolutional layer ConvTP_3, the fourth time-frequency convolutional layer ConvTP_4, and the fifth convolutional layer Conv_5.

[0105] Among them, the convolution kernel in the second time-frequency convolutional layer ConvTP_2 has a dimension of n2 in the time domain direction and a dimension of m2 in the frequency domain direction, the convolution kernel in the third time-frequency convolutional layer ConvTP_3 has a dimension of n3 in the time domain direction and a dimension of m3 in the frequency domain direction, and the convolution kernel in the fourth time-frequency convolutional layer ConvTP_4 has a dimension of n4 in the time domain direction and a dimension of m4 in the frequency domain direction.

[0106] In this design, input the first attenuation time-frequency signal E12 into the second time-frequency convolutional layer ConvTP_2 to extract the first attenuation time-frequency features in the time domain and the frequency domain.

[0107] Input the second attenuated time-frequency signal E22 into the third time-frequency convolutional layer ConvTP_3 to perform a convolutional operation, and extract the second attenuated time-frequency features in the time domain and the frequency domain.

[0108] Input the third attenuated time-frequency signal E32 into the fourth time-frequency convolutional layer ConvTP_4 to perform a convolutional operation, and extract the third attenuated time-frequency features in the time domain and the frequency domain.

[0109] Concatenate the first attenuated time-frequency features, the second attenuated time-frequency features, and the third attenuated frequency domain features Concat into the fourth attenuated time-frequency features.

[0110] Input the fourth attenuated time-frequency features into the fifth convolutional layer Conv_5 to perform a convolutional operation, and extract the interactive time-frequency features.

[0111] In this embodiment, when decomposing and mining the features in the time domain and the frequency domain of the first attenuated time-frequency signal, the second attenuated time-frequency signal, and the third attenuated time-frequency signal, align the features of the first attenuated time-frequency signal, the second attenuated time-frequency signal, and the third attenuated time-frequency signal in the time domain and the frequency domain, which is convenient for comparing the features in the time domain and the frequency domain within the first attenuated time-frequency signal, the second attenuated time-frequency signal, and the third attenuated time-frequency signal, and is also convenient for comparing the features in the time domain and the frequency domain with the features of the sample time-frequency signal, and amplifying the small features with fluctuations in the time domain and the frequency domain.

[0112] Step 1065: Concatenate the interactive time domain features, the interactive frequency domain features, and the interactive time-frequency features into the attenuated composite features.

[0113] In this embodiment, as Figure 2 shown, the interactive time domain features, the interactive frequency domain features, and the interactive time-frequency features can be Concat concatenated into the attenuated composite features.

[0114] Step 1066: Input the attenuated composite features into the second convolutional layer to extract the attenuated signal features.

[0115] In this embodiment, as Figure 2 shown, input the attenuated composite features into the second convolutional layer Conv_2 to perform a convolutional operation, and extract the attenuated signal features.

[0116] Step 107: Input the sample signal features and the attenuated signal features into the classification network to identify the probability of abnormal attenuator status.

[0117] The antenna attenuation detection model includes a classification network, and the classification network is responsible for performing a binary classification task. Then, input the sample signal features and the attenuated signal features into the classification network to perform binary classification and identify the probability of abnormal attenuator status.

[0118] In an embodiment of the present invention, as Figure 2As shown, the classification network includes a sixth convolutional layer Conv_6, a self-attention layer Self-Attention, and a fully connected layer FC.

[0119] In this embodiment, the sample signal feature and the attenuation signal feature are concatenated Concat into a first signal pair feature.

[0120] The first signal pair feature is input into the sixth convolutional layer Conv_6 to perform a convolutional operation to extract a second signal pair feature.

[0121] The second signal pair feature is input into the self-attention layer Self-Attention to extract a third signal pair feature.

[0122] Since the sample signal feature and the attenuation signal feature come from different structures, therefore, a sixth convolutional layer Conv_6 and a self-attention layer Self-Attention are provided in the classification network to integrate the sample signal feature and the attenuation signal feature, and then the fully connected layer FC is used for feature dimension conversion.

[0123] Among them, the self-attention layer Self-Attention implements a self-attention mechanism, which can establish a global dependence relationship for the sample time-frequency signal, the first attenuation time-frequency signal, the second attenuation time-frequency signal, and the third attenuation time-frequency signal, expand the receptive field in the feature map, and obtain more context information in the comparison of time domain, frequency domain, time domain and frequency domain.

[0124] The third signal pair feature is input into the fully connected layer FC to perform a fully connected operation and mapped into a fourth signal pair feature.

[0125] An activation function such as Sigmoid (S-shaped function) is used to perform an activation operation on the fourth signal pair feature to obtain the probability of the attenuator state being abnormal.

[0126] Step 108: Detect whether the state of the attenuator is abnormal according to the probability.

[0127] In this embodiment, methods such as the threshold method and the trend method can be used to detect whether the state of the attenuator is abnormal or normal according to the probability.

[0128] In specific implementation, multiple different sample test signals can be used to perform self-check on the attenuator. At this time, multiple probabilities can be accumulated, the average value of the multiple probabilities can be calculated to obtain a confidence level, so as to smooth the multiple probabilities and reduce the influence of errors.

[0129] The confidence level is compared with a preset second threshold. Since the self-check of the attenuator is a preliminary maintenance, it can prevent sudden major failures of the attenuator. Therefore, the second threshold can use empirical values with low sensitivity, such as 0.9, 0.95, etc.

[0130] If the confidence level is less than a preset second threshold, determine that the state of the attenuator is normal.

[0131] If the confidence level is greater than or equal to the preset second threshold, determine that the state of the attenuator is abnormal.

[0132] When it is detected that the state of the attenuator in the marine antenna is abnormal, it indicates that there are performance hazards in the attenuator of the marine antenna. At this time, an alarm operation can be performed on the attenuator in the marine antenna, and a prompt message can be sent to the accounts of relevant users (such as the owner or lessee of the ship, etc.), prompting them to pay attention to the attenuator in the marine antenna and contact the manufacturer that produces or sells the marine antenna in a timely manner to arrange for professionals to repair the attenuator in the marine antenna.

[0133] In this embodiment, communication operation data is collected for the marine antenna on the ship; there is an attenuator in the marine antenna; maintenance nodes are set for the attenuator according to the communication operation data; when the maintenance node is reached, a sample test signal is input into the attenuator, and a first attenuation test signal is output at the first attenuation position, a second attenuation test signal is output at the second attenuation position, and a third attenuation test signal is output at the third attenuation position in sequence; an antenna attenuation detection model is determined; the antenna attenuation detection model includes a first feature network, a second feature network, and a classification network; the sample test signal is input into the first feature network, and sample signal features are extracted in the time domain and / or frequency domain; the first attenuation test signal, the second attenuation test signal, and the third attenuation test signal are input into the second feature network, and attenuation signal features are jointly extracted in the time domain and / or frequency domain; the sample signal features and the attenuation signal features are input into the classification network to identify the probability of abnormal state of the attenuator; whether the state of the attenuator is abnormal is detected according to the probability. In this embodiment, a comparison set is constructed for the attenuation of the signal by the attenuator in the marine antenna at different attenuation positions, and the features are decomposed in the time domain and / or frequency domain, which can amplify the small fluctuations of the signal in the time domain and frequency domain, ensure the accuracy of detecting the state of the attenuator, realize the self-check of the attenuator, and the self-check process has a high degree of automation, convenient operation, and can effectively improve the efficiency of detecting the state of the attenuator.

[0134] Moreover, the self-check nodes are adaptively set according to the communication operation conditions of the marine antenna, the maintenance frequency is reasonably increased, and the performance hazards of the attenuator are effectively eliminated while minimizing the impact on the marine antenna as much as possible, ensuring the safety of navigation.

[0135] Embodiment 2

[0136] See Figure 3 , which shows a schematic structural diagram of an abnormal detection device for a marine antenna attenuator provided in Embodiment 2 of the present invention. As Figure 3 shown, the device includes:

[0137] The communication operation data acquisition module 301 is used to collect communication operation data from the maritime antenna on the ship; an attenuator is provided in the maritime antenna;

[0138] The maintenance node setting module 302 is used to set maintenance nodes for the attenuator according to the communication operation data;

[0139] The test signal attenuation module 303 is used to input a sample test signal into the attenuator when reaching the maintenance node, and sequentially output a first attenuation test signal at a first attenuation position, a second attenuation test signal at a second attenuation position, and a third attenuation test signal at a third attenuation position;

[0140] The model determination module 304 is used to determine an antenna attenuation detection model; the antenna attenuation detection model includes a first feature network, a second feature network, and a classification network;

[0141] The sample signal feature extraction module 305 is used to input the sample test signal into the first feature network and extract sample signal features in the time domain and / or frequency domain;

[0142] The attenuation signal feature extraction module 306 is used to input the first attenuation test signal, the second attenuation test signal, and the third attenuation test signal into the second feature network and jointly extract attenuation signal features in the time domain and / or frequency domain;

[0143] The abnormal probability identification module 307 is used to input the sample signal features and the attenuation signal features into the classification network to identify the probability of abnormal state of the attenuator;

[0144] The state detection module 308 is used to detect whether the state of the attenuator is abnormal according to the probability.

[0145] In an embodiment of the present invention, the maintenance node setting module 302 includes:

[0146] The intensity query module is used to query the intensity of the received signal during each communication in the communication operation data;

[0147] The communication frequency accumulation module is used to accumulate the communication frequency by 1 if the intensity is greater than a preset first threshold;

[0148] The adjustment coefficient generation module is used to generate an adjustment coefficient according to the communication frequency; the adjustment coefficient is negatively correlated with the communication frequency;

[0149] The time adjustment module is used to multiply the adjustment coefficient by a preset unit time to obtain the maintenance node of the attenuator.

[0150] In an embodiment of the present invention, the adjustment coefficient is:

[0151] H = min(2 / e 2x , α);

[0152] Wherein, H is the adjustment coefficient, x is the frequency of the communication, α is a preset upper limit value, and min is a function for taking the minimum value.

[0153] In an embodiment of the present invention, the first feature network includes a first time-domain convolutional layer, a first frequency-domain convolutional layer, a first time-frequency convolutional layer, and a first convolutional layer;

[0154] The sample signal feature extraction module 305 is further configured to:

[0155] Perform continuous wavelet transform on the sample test signal to obtain a sample time-frequency signal;

[0156] Input the sample time-frequency signal into the first time-domain convolutional layer to extract sample time-domain features in the time domain;

[0157] Input the sample time-frequency signal into the first frequency-domain convolutional layer to extract sample frequency-domain features in the frequency domain;

[0158] Input the sample time-frequency signal into the first time-frequency convolutional layer to extract sample time-frequency features in the time domain and frequency domain;

[0159] Concatenate the sample time-domain features, the sample frequency-domain features, and the sample time-frequency features into a sample composite feature;

[0160] Input the sample composite feature into the first convolutional layer to extract sample signal features.

[0161] In an embodiment of the present invention, the second feature network includes a first interaction module, a second interaction module, a third interaction module, and a second convolutional layer;

[0162] The attenuation signal feature extraction module 306 includes:

[0163] A time-frequency conversion module, configured to perform continuous wavelet transform on the first attenuation test signal, the second attenuation test signal, and the third attenuation test signal respectively to obtain a first attenuation time-frequency signal, a second attenuation time-frequency signal, and a third attenuation time-frequency signal;

[0164] An interactive time-domain feature extraction module, configured to input the first attenuation time-frequency signal, the second attenuation time-frequency signal, and the third attenuation time-frequency signal into the first interaction module to jointly extract interactive time-domain features in the time domain;

[0165] The interactive frequency-domain feature extraction module is used to input the first attenuated time-frequency signal, the second attenuated time-frequency signal, and the third attenuated time-frequency signal into the second interactive module to jointly extract interactive frequency-domain features in the frequency domain;

[0166] The interactive time-frequency feature extraction module is used to input the first attenuated time-frequency signal, the second attenuated time-frequency signal, and the third attenuated time-frequency signal into the third interactive module to jointly extract interactive time-frequency features in the time domain and the frequency domain;

[0167] The attenuation composite feature splicing module is used to splice the interactive time-domain feature, the interactive frequency-domain feature, and the interactive time-frequency feature into an attenuation composite feature;

[0168] The attenuation signal feature generation module is used to input the attenuation composite feature into the second convolutional layer to extract attenuation signal features.

[0169] In an embodiment of the present invention, the first interactive module includes a second time-domain convolutional layer, a third time-domain convolutional layer, a fourth time-domain convolutional layer, and a third convolutional layer; the interactive time-domain feature extraction module is further used for:

[0170] Input the first attenuated time-frequency signal into the second time-domain convolutional layer to extract the first attenuated time-domain feature in the time domain;

[0171] Input the second attenuated time-frequency signal into the third time-domain convolutional layer to extract the second attenuated time-domain feature in the time domain;

[0172] Input the third attenuated time-frequency signal into the fourth time-domain convolutional layer to extract the third attenuated time-domain feature in the time domain;

[0173] Splice the first attenuated time-domain feature, the second attenuated time-domain feature, and the third attenuated time-domain feature into a fourth attenuated time-domain feature;

[0174] Input the fourth attenuated time-domain feature into the second convolutional layer to extract the interactive time-domain feature;

[0175] The second interactive module includes a second frequency-domain convolutional layer, a third frequency-domain convolutional layer, a fourth frequency-domain convolutional layer, and a fourth convolutional layer; the interactive frequency-domain feature extraction module is further used for:

[0176] Input the first attenuated time-frequency signal into the second frequency-domain convolutional layer to extract the first attenuated frequency-domain feature in the frequency domain;

[0177] Input the second attenuated time-frequency signal into the third frequency-domain convolutional layer to extract the second attenuated frequency-domain feature in the frequency domain;

[0178] Input the third attenuated time-frequency signal into the fourth frequency-domain convolutional layer to extract third attenuated frequency-domain features in the frequency domain;

[0179] Concatenate the first attenuated frequency-domain feature, the second attenuated frequency-domain feature, and the third attenuated frequency-domain feature into a fourth attenuated frequency-domain feature;

[0180] Input the fourth attenuated frequency-domain feature into the fourth convolutional layer to extract interactive frequency-domain features;

[0181] The third interaction module includes a second time-frequency convolutional layer, a third time-frequency convolutional layer, a fourth time-frequency convolutional layer, and a fifth convolutional layer; The interactive time-frequency feature extraction module is further configured to:

[0182] Input the first attenuated time-frequency signal into the second time-frequency convolutional layer to extract first attenuated time-frequency features in the time domain and the frequency domain;

[0183] Input the second attenuated time-frequency signal into the third time-frequency convolutional layer to extract second attenuated time-frequency features in the time domain and the frequency domain;

[0184] Input the third attenuated time-frequency signal into the fourth time-frequency convolutional layer to extract third attenuated time-frequency features in the time domain and the frequency domain;

[0185] Concatenate the first attenuated time-frequency feature, the second attenuated time-frequency feature, and the third attenuated frequency-domain feature into a fourth attenuated time-frequency feature;

[0186] Input the fourth attenuated time-frequency feature into the fifth convolutional layer to extract interactive time-frequency features.

[0187] In an embodiment of the present invention, the classification network includes a sixth convolutional layer, a self-attention layer, and a fully-connected layer;

[0188] The abnormal probability recognition module 307 is further configured to:

[0189] Concatenate the sample signal feature and the attenuated signal feature into a first signal pair feature;

[0190] Input the first signal pair feature into the sixth convolutional layer to extract a second signal pair feature;

[0191] Input the second signal pair feature into the self-attention layer to extract a third signal pair feature;

[0192] Input the third signal pair feature into the fully-connected layer to map it into a fourth signal pair feature;

[0193] Perform an activation operation on the fourth signal pair feature to obtain the probability of the abnormal state of the attenuator.

[0194] In an embodiment of the present invention, the state detection module 308 includes:

[0195] A confidence calculation module, configured to calculate the average value of the multiple probabilities to obtain a confidence level;

[0196] A normal determination module, configured to determine that the state of the attenuator is normal if the confidence level is less than a preset second threshold;

[0197] An abnormal determination module, configured to determine that the state of the attenuator is abnormal if the confidence level is greater than or equal to the preset second threshold.

[0198] The abnormal detection device for a marine antenna attenuator provided by an embodiment of the present invention can execute the abnormal detection method for a marine antenna attenuator provided by any embodiment of the present invention, and has corresponding functional modules and beneficial effects for executing the abnormal detection method for a marine antenna attenuator.

[0199] Embodiment III

[0200] See Figure 4 , which shows a schematic structural diagram of an electronic device provided by an embodiment of the present invention. The electronic device is intended to represent various forms of digital computers, such as, for example, a laptop computer, a desktop computer, a workbench, a personal digital assistant, a blade server, a mainframe computer, and other suitable computers. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present invention described herein and / or claimed.

[0201] As Figure 4 shown, the electronic device 10 includes at least one processor 11, and a memory communicatively connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc. Among them, the memory stores a computer program executable by the at least one processor. The processor 11 can execute various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. The input / output (I / O) interface 15 is also connected to the bus 14.

[0202] Multiple components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disc, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0203] The processor 11 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include but are not limited to a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the anomaly detection method for the maritime antenna attenuator.

[0204] In some embodiments, the anomaly detection method for the maritime antenna attenuator can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the anomaly detection method for the maritime antenna attenuator described above can be executed. Alternatively, in other embodiments, the processor 11 can be configured to execute the anomaly detection method for the maritime antenna attenuator by any other suitable means (e.g., by means of firmware).

[0205] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor, and can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0206] A computer program for implementing the method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, a special purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowchart and / or block diagram to be implemented. The computer programs can be executed entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine, or entirely on a remote machine or server.

[0207] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0208] In order to provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0209] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected with each other by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), blockchain network, and the Internet.

[0210] A computing system can include a client and a server. The client and the server are generally far apart from each other and typically interact through a communication network. The client-server relationship is created by computer programs running on respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, and solves the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services.

[0211] Embodiment 4

[0212] The embodiment of the present invention also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the abnormal detection method of the marine antenna attenuator provided in any embodiment of the present invention.

[0213] In the process of implementing the computer program product, computer program code for performing the operations of the present invention can be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network - including a local area network (LAN) or a wide area network (WAN) - or can be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0214] It should be understood that the various forms of processes shown above can be used, with steps reordered, added or deleted. For example, the steps recited in the present invention can be executed in parallel, sequentially or in a different order, as long as the desired results of the technical solution of the present invention can be achieved, and no limitation is imposed herein.

[0215] The above specific embodiments do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub - combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for detecting abnormality of a maritime antenna attenuator, characterized in that: include: Collect communication operation data from maritime antennas on ships; The maritime antenna has an attenuator; Setting a maintenance node for the attenuator according to the communication operation data; When reaching the maintenance node, the sample test signal is input into the attenuator, and a first attenuated test signal is output at a first attenuation position, a second attenuated test signal is output at a second attenuation position, and a third attenuated test signal is output at a third attenuation position in sequence; Determine an antenna attenuation detection model; the antenna attenuation detection model includes a first feature network, a second feature network and a classification network; Inputting the sample test signal into the first feature network to extract sample signal features in the time domain and / or frequency domain; Inputting the first attenuated test signal, the second attenuated test signal and the third attenuated test signal into the second feature network, and jointly extracting attenuated signal features in the time domain and / or frequency domain; Inputting the sample signal feature and the attenuation signal feature into the classification network to identify the probability of abnormal state of the attenuator; Whether the state of the attenuator is abnormal is detected according to the probability.

2. The method according to claim 1, characterized in that The step of setting a maintenance node for the attenuator according to the communication operation data comprises: querying the received signal strength during each communication in the communication operation data; If the intensity is greater than a preset first threshold, the frequency of communication is accumulated by 1; generating an adjustment coefficient according to the frequency of the communication; the adjustment coefficient is negatively correlated with the frequency of the communication; The adjustment coefficient is multiplied by a preset unit time to obtain the maintenance node of the attenuator.

3. The method according to claim 2, characterized in that The adjustment coefficient is: H=min(2 / e 2x ,a); Among them, H is the adjustment coefficient, x is the frequency of the communication, α is the preset upper limit value, and min is the minimum value function.

4. The method according to any one of claims 1 to 3, characterized in that The first feature network includes a first time domain convolution layer, a first frequency domain convolution layer, a first time-frequency convolution layer and a first convolution layer; The step of inputting the sample test signal into the first feature network and extracting sample signal features in the time domain and / or frequency domain includes: Performing continuous wavelet transform on the sample test signal to obtain a sample time-frequency signal; Inputting the sample time-frequency signal into the first time-domain convolutional layer to extract sample time-domain features in the time domain; Inputting the sample time-frequency signal into the first frequency domain convolution layer to extract sample frequency domain features in the frequency domain; Inputting the sample time-frequency signal into the first time-frequency convolution layer, and extracting sample time-frequency features in the time domain and the frequency domain; splicing the sample time domain feature, the sample frequency domain feature and the sample time-frequency feature into a sample composite feature; The sample composite features are input into the first convolutional layer to extract sample signal features.

5. The method according to claim 4, characterized in that The second feature network includes a first interaction module, a second interaction module, a third interaction module and a second convolutional layer; The step of inputting the first attenuated test signal, the second attenuated test signal, and the third attenuated test signal into the second feature network, and jointly extracting attenuated signal features in the time domain and / or frequency domain, comprises: Performing continuous wavelet transform on the first attenuated test signal, the second attenuated test signal and the third attenuated test signal respectively to obtain a first attenuated time-frequency signal, a second attenuated time-frequency signal and a third attenuated time-frequency signal; Inputting the first attenuated time-frequency signal, the second attenuated time-frequency signal and the third attenuated time-frequency signal into the first interaction module, and jointly extracting interaction time-domain features in the time domain; Inputting the first attenuated time-frequency signal, the second attenuated time-frequency signal and the third attenuated time-frequency signal into the second interaction module, and jointly extracting interactive frequency domain features in the frequency domain; Inputting the first attenuated time-frequency signal, the second attenuated time-frequency signal and the third attenuated time-frequency signal into the third interaction module, and jointly extracting interactive time-frequency features in the time domain and the frequency domain; splicing the interactive time domain feature, the interactive frequency domain feature and the interactive time-frequency feature into an attenuated composite feature; The attenuated composite feature is input into the second convolutional layer to extract the attenuated signal feature.

6. The method according to claim 5, characterized in that The first interaction module includes a second time domain convolution layer, a third time domain convolution layer, a fourth time domain convolution layer and a third convolution layer; the first attenuated time-frequency signal, the second attenuated time-frequency signal and the third attenuated time-frequency signal are input into the first interaction module, and the interactive time domain features are jointly extracted in the time domain, including: Inputting the first attenuated time-frequency signal into the second time-domain convolutional layer to extract first attenuated time-domain features in the time domain; Inputting the second attenuated time-frequency signal into the third time-domain convolutional layer to extract second attenuated time-domain features in the time domain; Inputting the third attenuated time-frequency signal into the fourth time-domain convolutional layer to extract the third attenuated time-domain feature in the time domain; splicing the first attenuation time domain feature, the second attenuation time domain feature and the third attenuation time domain feature into a fourth attenuation time domain feature; Inputting the fourth attenuated time domain feature into the second convolutional layer to extract the interactive time domain feature; The second interaction module includes a second frequency domain convolution layer, a third frequency domain convolution layer, a fourth frequency domain convolution layer and a fourth convolution layer; the first attenuated time-frequency signal, the second attenuated time-frequency signal and the third attenuated time-frequency signal are input into the second interaction module, and the interactive frequency domain features are jointly extracted in the frequency domain, including: Inputting the first attenuated time-frequency signal into the second frequency domain convolution layer to extract first attenuated frequency domain features in the frequency domain; Inputting the second attenuated time-frequency signal into the third frequency domain convolution layer to extract second attenuated frequency domain features in the frequency domain; Inputting the third attenuated time-frequency signal into the fourth frequency domain convolution layer to extract third attenuated frequency domain features in the frequency domain; splicing the first attenuation frequency domain feature, the second attenuation frequency domain feature and the third attenuation frequency domain feature into a fourth attenuation frequency domain feature; Inputting the fourth attenuated frequency domain feature into the fourth convolutional layer to extract the interactive frequency domain feature; The third interaction module includes a second time-frequency convolution layer, a third time-frequency convolution layer, a fourth time-frequency convolution layer and a fifth convolution layer; the first attenuated time-frequency signal, the second attenuated time-frequency signal and the third attenuated time-frequency signal are input into the third interaction module, and the interactive time-frequency features are jointly extracted in the time domain and the frequency domain, including: Inputting the first attenuated time-frequency signal into the second time-frequency convolution layer, and extracting first attenuated time-frequency features in the time domain and the frequency domain; Inputting the second attenuated time-frequency signal into the third time-frequency convolution layer, and extracting second attenuated time-frequency features in the time domain and the frequency domain; Inputting the third attenuated time-frequency signal into the fourth time-frequency convolution layer, and extracting the third attenuated time-frequency features in the time domain and the frequency domain; splicing the first attenuation time-frequency feature, the second attenuation time-frequency feature and the third attenuation frequency domain feature into a fourth attenuation time-frequency feature; The fourth attenuated time-frequency feature is input into the fifth convolutional layer to extract the interactive time-frequency feature.

7. The method according to claim 6, characterized in that The classification network includes a sixth convolutional layer, a self-attention layer and a fully connected layer; The step of inputting the sample signal feature and the attenuation signal feature into the classification network to identify the probability of the attenuator state being abnormal includes: splicing the sample signal feature and the attenuated signal feature into a first signal pair feature; Inputting the first signal pair feature into the sixth convolutional layer to extract the second signal pair feature; Inputting the second signal pair feature into the self-attention layer to extract the third signal pair feature; Inputting the third signal pair feature into the fully connected layer and mapping it into a fourth signal pair feature; An activation operation is performed on the fourth signal pair feature to obtain the probability that the attenuator state is abnormal.

8. The method according to any one of claims 1-3, 5-7, characterized in that: The detecting whether the state of the attenuator is abnormal according to the probability includes: Calculate an average value of the multiple probabilities to obtain a confidence level; If the confidence level is less than a preset second threshold, determining that the state of the attenuator is normal; If the confidence level is greater than or equal to a preset second threshold, it is determined that the state of the attenuator is abnormal.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the abnormality detection method for a maritime antenna attenuator according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the abnormality detection method for a maritime antenna attenuator according to any one of claims 1 to 8 is implemented.

Images