Aircraft spectrum detection and identification method and device based on deep convolutional network

Abstract

The application discloses a kind of aircraft spectrum detection identification methods based on deep convolution network, comprising: collecting the antenna rotation control signal component and data feedback component of aircraft placed on ground, and as correction data;Acquisition detection signal in the process of aircraft flight;The detection signal includes horizontal polarization and longitudinal polarization data signal;The detection signal is preprocessed, and is divided into training data set and verification data set;Deep convolution network is built, and preset activation function and iteration threshold value;Training data set and correction data are input into deep convolution network respectively, training correction is carried out, and the network parameter of deep convolution network is optimized, to obtain the optimal deep convolution network;Optimal deep convolution network is used to verify verification data set, and output verification result, and realize spectrum detection identification.Through the above scheme, the application has the advantages of simple logic, accurate and reliable and the like.

Description

Technical Field

[0001] This invention relates to the field of aircraft spectrum detection technology, and in particular to an aircraft spectrum detection and identification method and apparatus based on deep convolutional networks. Background Technology

[0002] Spectrum monitoring is a portable signal analyzer used for radio monitoring and interference localization from 9kHz to 6GHz. It is particularly suitable for outdoor radio and spectrum monitoring, and for locating interference, bursts, and illegal signals. When using aircraft for low-altitude spectrum monitoring, multiple antennas are often required to work together, and both horizontal and vertical polarization modes need to be supported. For example, the Chinese invention patent with publication number "CN107367294A" and titled "Dual-Polarization Detection System Based on Resonant Surface Acoustic Wave Wireless Passive Sensor" consists of a dual-polarization sensing antenna capable of exciting two electromagnetic waves with mutually perpendicular polarization directions, a dual-polarization interrogation antenna capable of exciting two electromagnetic waves with mutually perpendicular polarization directions, phase matching and impedance matching circuits, polarization direction switching switches, transceiver switching switches, radio frequency signal transceiver processing circuits, duty cycle error detection circuits, a display unit, and a microprocessor unit.

[0003] For example, Chinese invention patent publication number "CN111381111A" entitled "Antenna Polarization Test Circuit and Test Device" describes an antenna polarization test circuit that includes a linearly polarized antenna module, a circularly polarized antenna module, a matching module, a multiplexer RF switch, and a controller. The linearly polarized and circularly polarized antenna modules are electrically connected to the input terminals of the multiplexer RF switch via the matching module. The control terminal of the multiplexer RF switch is electrically connected to the controller. The output terminal of the multiplexer RF switch is used to electrically connect to an RF detector. The matching module provides impedance matching between the multiplexer RF switch and the linearly polarized and circularly polarized antenna modules, respectively. The controller controls the multiplexer RF switch to select either the linearly polarized or circularly polarized antenna module to receive the wireless signal from the antenna under test.

[0004] The aforementioned technology provides the collaborative use of multiple antennas, such as dual-polarized sensing antennas or linearly polarized antenna modules and circularly polarized antenna modules. However, it does not provide a method for collaborative detection and identification using dual-polarized sensing antennas or multiple antennas. Furthermore, when using aircraft-borne spectrum detection, the detection signal contains electromagnetic interference signals from aircraft control and feedback, and these interference signals are not part of sudden or illegal signals.

[0005] Therefore, there is an urgent need to propose a simple, accurate, and reliable method and device for aircraft spectrum detection and identification based on deep convolutional networks. Summary of the Invention

[0006] To address the aforementioned problems, the present invention aims to provide a method and apparatus for aircraft spectrum detection and identification based on deep convolutional networks. The technical solution adopted by the present invention is as follows:

[0007] The first part of this technology provides a method for detecting and identifying the spectrum of an aircraft based on a deep convolutional network, which uses an antenna for spectrum monitoring mounted on the aircraft. The antenna is mounted on the aircraft support, and an antenna rotation drive mechanism is provided between the antenna and the aircraft support. The antenna rotation drive mechanism drives the antenna to rotate and switch between the horizontal and vertical directions.

[0008] The aircraft spectrum detection and identification method includes the following steps:

[0009] The antenna rotation control signal components and data feedback components of the aircraft placed on the ground are collected and used as correction data;

[0010] The system collects detection signals during the flight of the aircraft; these detection signals include horizontally polarized and longitudinally polarized data signals.

[0011] The detected signals are preprocessed and divided into training and validation datasets;

[0012] Build a deep convolutional network and preset the activation function and iteration threshold;

[0013] The training dataset and calibration data are respectively input into the deep convolutional network for training and calibration, and the network parameters of the deep convolutional network are optimized to obtain the optimal deep convolutional network.

[0014] The optimal deep convolutional network is used to validate the validation dataset, output the validation results, and realize spectrum detection and recognition.

[0015] The second part of this technology provides an aircraft spectrum detection and identification device based on deep convolutional networks, which includes:

[0016] A spectrum monitor is installed on the aircraft to monitor the spectrum.

[0017] An antenna, mounted on the aircraft's support frame, is connected to a spectrum monitor. The antenna collects the antenna rotation control signal components and data feedback components when the aircraft is positioned on the ground, and uses these as correction data. The antenna also collects detection signals during the aircraft's flight. These detection signals include horizontally polarized and longitudinally polarized data signals.

[0018] An antenna rotation drive mechanism is located between the antenna and the aircraft support, and drives the antenna to rotate and switch between the horizontal and vertical directions;

[0019] The preprocessing module, connected to the spectrum monitor and antenna, preprocesses the probed signals and divides them into training and validation datasets.

[0020] The network model building module builds a deep convolutional network and presets the activation function and iteration threshold.

[0021] The training module, connected to the preprocessing module, the network model building module, and the spectrum monitor, inputs the training dataset and correction data into the deep convolutional network for training and correction, and optimizes the network parameters of the deep convolutional network to obtain the optimal deep convolutional network.

[0022] The verification module, connected to the training and preprocessing modules, uses an optimal deep convolutional network to verify the verification dataset, outputs the verification results, and implements spectrum detection and recognition.

[0023] Thirdly, this technology provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements an aircraft spectrum detection and identification method based on a deep convolutional network.

[0024] Part Four: This technology provides a computer-readable storage medium storing a computer program, characterized in that, when the computer program is executed by a processor, it implements the steps of an aircraft spectrum detection and identification method based on a deep convolutional network.

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] This invention incorporates antenna rotation control signal components and data feedback components as correction data, and uses the inverse error value to correct the network parameters of the neural network. This eliminates interference signals other than non-polarized data, ensuring reliable identification and location of interference, bursts, and illegal signals. In summary, this invention has advantages such as simple logic and high accuracy and reliability, and has high practical and promotional value in the field of aircraft spectrum detection technology. Attached Figure Description

[0027] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope of protection. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This is a schematic diagram of the hardware structure of the present invention.

[0029] Figure 2This is a logic flowchart of the present invention.

[0030] In the above figures, the component names corresponding to the reference numerals are as follows:

[0031] 100. Aircraft; 101. Aircraft support frame; 200. Antenna; 300. Antenna rotation drive mechanism; 400. Spectrum monitor. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of this application clearer, the present invention will be further described below with reference to the accompanying drawings and embodiments. The embodiments of the present invention include, but are not limited to, the following embodiments. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort are within the scope of protection of this application.

[0033] In this embodiment, the term "and / or" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0034] The terms "first" and "second," etc., used in the specification and claims of this embodiment are used to distinguish different objects, not to describe a specific order of objects. For example, "first target object" and "second target object," etc., are used to distinguish different target objects, not to describe a specific order of target objects.

[0035] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0036] In the description of the embodiments in this application, unless otherwise stated, "multiple" means two or more. For example, multiple processing units means two or more processing units; multiple systems means two or more systems.

[0037] like Figures 1 to 2As shown, this embodiment provides a method for aircraft spectrum detection and identification based on deep convolutional networks. The method employs an aircraft 100 equipped with a spectrum monitoring antenna 200. The antenna 200 is mounted on an aircraft support 101, and an antenna rotation drive mechanism 300 is provided between the antenna 200 and the aircraft support 101. The antenna rotation drive mechanism 300 drives the antenna to rotate in both horizontal and vertical directions. It should be noted that the antenna rotation drive mechanism 300 in this embodiment uses existing mature technology, which can drive the antenna 200 to rotate 90° to achieve rotation switching between horizontal and vertical directions. Alternatively, the antenna rotation drive mechanism 300 is located at one end of the rotation axis of the antenna 200 and fixed to the aircraft support 101. The other end of the rotation axis of the antenna 200 is connected to the aircraft support 101 via a rotating shaft. A stepper motor for rotation drive is installed within the antenna rotation drive mechanism 300, thus driving the antenna 200 to rotate along the rotation axis.

[0038] In this embodiment, the aircraft spectrum detection and identification method specifically includes the following steps:

[0039] The first step involves collecting the antenna rotation control signal components and data feedback components from the aircraft's ground-based location, which will be used as calibration data. At this stage, detection and acquisition are not initiated; instead, non-detection and acquisition signals are collected for calibration purposes.

[0040] The second step is to collect detection signals during the aircraft's flight; these signals include horizontally polarized and longitudinally polarized data signals. In actual data acquisition, antenna rotation control signal components and data feedback components are inevitably present.

[0041] The third step involves preprocessing the probe signals, dividing them into training and validation datasets. This includes filtering, normalizing, and sliding window segmentation of the probe signals.

[0042] The fourth step is to build a deep convolutional network and preset the activation function and iteration threshold. A deep convolutional neural network mainly consists of an input layer, convolutional layers, activation functions, pooling layers, fully connected layers, and an output layer. The input layer allows the deep convolutional network to directly use images as input, extracting features through training. The convolutional layer, through convolution operations, essentially represents the input in another way. If we consider the convolutional layer as a black box, then we can view the output as another representation of the input, and the entire network training is essentially training the intermediate parameters needed to achieve this representation. The pooling layer is a special data processing operation in a convolutional neural network. It reduces the size of image features through pooling, effectively eliminating the computational burden caused by using the results of the previous layer as input. The activation function is a linear operation in the network where both convolution and pooling operations are linear.

[0043] The fifth step involves inputting the training dataset and the calibration data into the deep convolutional network for training and calibration, and optimizing the network parameters of the deep convolutional network to obtain the optimal deep convolutional network.

[0044] Specifically, when using a horizontally oriented antenna to detect signals, the following steps are included:

[0045] (11) Drive the antenna to rotate to the horizontal direction and collect the detection signal polarized in the horizontal direction.

[0046] (12) Preprocess the detection signal that is polarized in the horizontal direction.

[0047] (13) The preprocessed horizontally polarized detection signal and correction data are sequentially input into the deep convolutional network for training and correction.

[0048] In addition, when using an antenna positioned vertically to detect signals, the following steps are included:

[0049] (21) Drive the antenna to rotate to the vertical direction and collect the detection signal polarized in the longitudinal direction.

[0050] (22) Preprocess the detection signal that is polarized in the longitudinal direction.

[0051] (23) The preprocessed longitudinally polarized detection signal and correction data are sequentially input into the deep convolutional network for training and correction.

[0052] Furthermore, training calibration and network parameter optimization include the following steps:

[0053] (31) Input the training dataset into the deep convolutional network for training and output the first recognition result.

[0054] (32) Input the correction data into a deep convolutional network for verification and output the second recognition result.

[0055] (33) The error value between training and correction is obtained. In this embodiment, the error value δ satisfies the following formula:

[0056] δ=1-δ1

[0057] Wherein, δ1 represents the error between the first recognition result and the second recognition result, and its value is between (0,1).

[0058] (34) The error value is back-transmitted into the deep convolutional network, and the network parameters are optimized; the network parameters include weights;

[0059] (35) When the error value is less than the preset iteration threshold, training and correction shall be stopped.

[0060] The sixth step involves using an optimal deep convolutional network to validate the validation dataset, outputting the validation results, and implementing spectrum detection and recognition.

[0061] The above embodiments are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any changes made based on the design principles of the present invention, or any non-creative modifications made thereon, shall fall within the scope of protection of the present invention.

Claims

1. A method for detecting and identifying aircraft spectrum based on deep convolutional networks, characterized in that, An antenna for spectrum monitoring is mounted on an aircraft. The antenna is mounted on the aircraft's support frame, and an antenna rotation drive mechanism is provided between the antenna and the support frame. The antenna rotation drive mechanism drives the antenna to rotate and switch between the horizontal and vertical directions. The aircraft spectrum detection and identification method includes the following steps: The antenna rotation control signal components and data feedback components of the aircraft placed on the ground are collected and used as correction data; The system collects detection signals during the flight of the aircraft; these detection signals include horizontally polarized and longitudinally polarized data signals. The detected signals are preprocessed and divided into training and validation datasets; Build a deep convolutional network and preset the activation function and iteration threshold; The training dataset and calibration data are respectively input into the deep convolutional network for training and calibration, and the network parameters of the deep convolutional network are optimized to obtain the optimal deep convolutional network. The optimal deep convolutional network is used to validate the validation dataset, output the validation results, and realize spectrum detection and recognition.

2. The aircraft spectrum detection and identification method based on deep convolutional networks according to claim 1, characterized in that, The detection signal is preprocessed, including filtering, normalization, and sliding window segmentation.

3. The aircraft spectrum detection and identification method based on deep convolutional networks according to claim 1, characterized in that, Also includes: The antenna is driven to rotate to a horizontal position and the detection signal polarized in the horizontal direction is acquired. Preprocessing of horizontally polarized detection signals; The preprocessed horizontally polarized detection signal and correction data are sequentially input into a deep convolutional network for training and correction.

4. The aircraft spectrum detection and identification method based on deep convolutional networks according to claim 1, characterized in that, Also includes: The antenna is driven to rotate to the vertical direction and the detection signal polarized in the longitudinal direction is acquired; Preprocessing of longitudinally polarized detection signals; The preprocessed longitudinally polarized detection signal and correction data are sequentially input into a deep convolutional network for training and correction.

5. The aircraft spectrum detection and identification method based on deep convolutional networks according to claim 1, 3, or 4, characterized in that, The training dataset and calibration data are respectively input into the deep convolutional network for training and calibration, and the network parameters of the deep convolutional network are optimized to obtain the optimal deep convolutional network. This includes the following steps: The training dataset is input into a deep convolutional network for training, and the first recognition result is output. The correction data is input into a deep convolutional network for verification, and a second recognition result is output. Calculate the error value between the first recognition result and the second recognition result; The error value is then fed back into the deep convolutional network for network parameter optimization; the network parameters include weights. Training and correction will stop when the error value is less than the preset iteration threshold.

6. The aircraft spectrum detection and identification method based on deep convolutional networks according to claim 5, characterized in that, The error value Satisfy the following formula: ； in, This represents the error between the first and second recognition results, and its value is between (0,1).

7. An aircraft spectrum detection and identification device based on deep convolutional networks, characterized in that, include: A spectrum monitor is installed on the aircraft to monitor the spectrum. An antenna, mounted on the aircraft's support frame, is connected to a spectrum monitor. The antenna collects the antenna rotation control signal components and data feedback components when the aircraft is positioned on the ground, and uses these as correction data. The antenna also collects detection signals during the aircraft's flight. These detection signals include horizontally polarized and longitudinally polarized data signals. An antenna rotation drive mechanism is located between the antenna and the aircraft support, and drives the antenna to rotate and switch between the horizontal and vertical directions; The preprocessing module, connected to the spectrum monitor and antenna, preprocesses the probed signals and divides them into training and validation datasets. The network model building module builds a deep convolutional network and presets the activation function and iteration threshold. The training module, connected to the preprocessing module, the network model building module, and the spectrum monitor, inputs the training dataset and correction data into the deep convolutional network for training and correction, and optimizes the network parameters of the deep convolutional network to obtain the optimal deep convolutional network. The verification module, connected to the training and preprocessing modules, uses an optimal deep convolutional network to verify the verification dataset, outputs the verification results, and implements spectrum detection and recognition.

8. The aircraft spectrum detection and identification device based on deep convolutional networks according to claim 7, characterized in that, Also includes: The backpropagation module, located within the training module, inputs the training dataset into the deep convolutional network for training and outputs the first recognition result. The correction data is input into a deep convolutional network for verification, and a second recognition result is output. The error value between training and correction is calculated. The error value is then transmitted back into the deep convolutional network for network parameter optimization.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the aircraft spectrum detection and identification method based on deep convolutional networks as described in any one of claims 1 to 6.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the aircraft spectrum detection and identification method based on deep convolutional networks as described in any one of claims 1 to 6.

Images