Short-range unmanned aerial vehicle anti-electromagnetic interference device and method based on deep learning

Abstract

The invention relates to a short-range unmanned aerial vehicle anti-electromagnetic interference device and method based on deep learning. The device comprises an electromagnetic pulse induction module, a power supply module, an intelligent electromagnetic pulse shielding module, a signal recovery and enhancement module, an intelligent decision and control module and a deep learning and model updating module. The device integrates a plurality of intelligent modules, can sense and analyze electromagnetic pulses in real time, automatically adjust electromagnetic shielding strength, enhance signal recovery capability, and ensure stable operation of equipment in a complex electromagnetic environment; automatic decision making and online prediction capability based on deep learning are realized, and an electromagnetic interference event can be early warned and a protection strategy can be optimized; a high-sensitivity sensor is adopted, missing detection of electromagnetic pulse signals is avoided, meanwhile, a false alarm correction mechanism is established in a deep learning system, output of multiple detection models is comprehensively judged by introducing an integrated learning technology, and if judgment results of the multiple models are consistent, the detection result is more reliable.

Description

Technical Field

[0001] The present application relates to the field of electromagnetic pulse monitoring and shielding technology, and in particular to a short-range UAV anti-electromagnetic interference device and method based on deep learning. Background Art

[0002] At present, the physical size of anti-electromagnetic interference equipment is relatively large, and there is generally a lack of effective electromagnetic shielding mechanisms. In addition, these devices have limitations in performing the task of accurately monitoring electromagnetic pulses, especially when distinguishing the superposition of multiple electromagnetic pulses, their resolution is insufficient. When detecting electromagnetic pulses of low intensity or specific frequency bands, the sensitivity of the monitoring equipment is defective, which may lead to missed detection or false alarms of electromagnetic pulse events. Given that electromagnetic pulses cover an extremely wide spectrum range, some monitoring equipment lacks spectrum coverage capabilities, affecting the comprehensiveness and accuracy of monitoring data. Summary of the invention

[0003] Based on this, it is necessary to provide a short-range UAV anti-electromagnetic interference device and method based on deep learning to address the above technical problems.

[0004] A short-range UAV anti-electromagnetic interference device based on deep learning, the device comprising: The electromagnetic pulse sensing module is used to detect electromagnetic pulse signals in the surrounding environment in real time from low frequency to ultra-high frequency, extract features of the electromagnetic pulse signals, and transmit the extracted features to the intelligent electromagnetic pulse shielding module, the intelligent decision-making and control module, and the signal recovery and enhancement module.

[0005] An intelligent electromagnetic pulse shielding module is used to automatically start the intelligent electromagnetic pulse shielding module when the intensity of the electromagnetic pulse signal exceeds a preset threshold. After the intelligent electromagnetic pulse shielding module is started, the electromagnetic pulse signal is shielded and isolated according to the electromagnetic shielding strategy; the electromagnetic shielding strategy is obtained by continuously adjusting and optimizing according to the received feedback signal; The signal recovery and enhancement module is used to recover the signal damaged by electromagnetic interference and obtain high-quality recovered signal; The intelligent decision-making and control module is used to perform real-time data analysis on the signals detected by the electromagnetic pulse induction module and the high-quality recovery signals, and then use intelligent algorithms to optimize the electromagnetic shielding strategy and signal recovery mode, and adjust the operating status of the device.

[0006] The deep learning and model updating module is used to optimize the anti-interference model and protection strategy through deep learning methods and adaptive updating mechanisms.

[0007] The power module is used to provide power for the short-range UAV anti-electromagnetic interference device and automatically adjusts the output voltage and current according to the power requirements of each module.

[0008] In one embodiment, the electromagnetic pulse induction module includes: a plurality of high-sensitivity electromagnetic induction sensors, a signal analysis and processing unit, and a filtering unit.

[0009] The electromagnetic induction sensor is used to detect electromagnetic interference signals in the surrounding environment in real time from low frequency to ultra-high frequency. If the electromagnetic interference signal is an electromagnetic pulse signal, the electromagnetic pulse signal is transmitted to the signal analysis and processing unit; otherwise, the electromagnetic interference signal is marked and discarded, and electromagnetic detection continues.

[0010] The signal analysis and processing unit is used to decompose the electromagnetic pulse signals detected by multiple electromagnetic induction sensors by using a spectrum analysis algorithm.

[0011] The filtering unit is used to filter the interference signal and transmit the filtered signal to the intelligent decision-making and control module; if the intensity of the electromagnetic pulse signal exceeds the preset threshold, the intelligent electromagnetic pulse shielding module is triggered; if the intensity of the electromagnetic pulse signal does not exceed the preset threshold, the electromagnetic detection continues.

[0012] In one of the embodiments, the intelligent electromagnetic pulse shielding module includes an interference signal transmitting unit and an electromagnetic shielding cover.

[0013] The interference signal transmitting unit is used to dynamically adjust the frequency, amplitude, and phase through pseudo-random frequency generation technology, and intelligently control it. It adopts a multi-wave interference method and uses broadband noise and pseudo-random signals to cover multiple frequency bands to interfere with the target device.

[0014] The electromagnetic shielding cover has a rectangular shape, and a flexible electromagnetic shielding material is used as a shielding layer inside the electromagnetic shielding cover.

[0015] In one embodiment, the signal recovery and enhancement module includes: a noise removal unit, a signal time domain analysis unit and a signal reconstruction unit.

[0016] A noise removal unit, for suppressing low-frequency or high-frequency noise in a signal transmitted by the electromagnetic pulse induction module by using a filter that should be selected in an adaptive manner based on deep learning; The signal time domain analysis unit is used to decompose the superimposed electromagnetic pulse signal after denoising by using time-frequency analysis, and to separate the superimposed electromagnetic pulse signal after denoising from the time domain and the frequency domain.

[0017] The signal reconstruction unit is used to recover the signal loss caused by electromagnetic interference by using a signal recovery algorithm to obtain a high-quality recovery signal, and transmit the high-quality recovery signal to the deep learning and model updating module.

[0018] In one embodiment, the working process of the intelligent decision-making and control module includes: The electric field strength, magnetic field strength, frequency spectrum, pulse width, and peak power of the received electromagnetic pulse signal are pre-processed for possible noise or invalid information.

[0019] Through spectrum analysis and time domain analysis, the electromagnetic pulse signal is analyzed and the key information in the electromagnetic pulse signal is extracted. Using deep learning and machine learning algorithms, effective features are extracted from the signal to determine the type, intensity and potential impact of electromagnetic interference.

[0020] According to the results of data analysis, the interference source is identified and its threat level is evaluated. According to the evaluation results of the interference source, the working mode of the electromagnetic pulse shielding device is adjusted in real time and the most suitable shielding strategy is selected.

[0021] After the evaluation and decision are completed, control instructions will be generated and passed to the intelligent electromagnetic pulse shielding module and the signal recovery and enhancement module.

[0022] Real-time intelligent electromagnetic pulse shielding module monitors shielding effect and signal recovery quality of signal recovery and enhancement module; regularly collects feedback data from electromagnetic pulse sensing unit and shielding device, evaluates the effectiveness of current strategy, and automatically adjusts optimization strategy based on feedback data.

[0023] By continuously accumulating electromagnetic interference data, deep learning algorithms are used to learn interference patterns, and based on newly acquired experience and data, the anti-interference model is optimized and the control strategy is updated.

[0024] In one embodiment, the specific steps of updating the control strategy include: If the electromagnetic interference intensity reaches the preset threshold, the intelligent electromagnetic pulse shielding module is enabled to perform reverse interference.

[0025] If the electromagnetic interference intensity exceeds the preset safety limit, the signal recovery and enhancement module will be automatically started to increase the anti-interference ability of the device.

[0026] Through machine learning algorithms, historical electromagnetic pulse interference situations are learned and the rule sets and decision-making processes are gradually optimized.

[0027] Adjust the output of the power module according to the electromagnetic interference intensity.

[0028] In one of the embodiments, the intelligent decision-making and control module also includes an APP interface; the APP interface is used for users to remotely view the current electromagnetic pulse interference situation, system status and the working status of each module, and is also used to display the intensity, frequency, and interference type of the electromagnetic pulse signal in real time, and allows users to make adjustments or choose manual control according to actual conditions; it is also used to receive feedback from the system based on rule sets and real-time decisions, and provide optimization suggestions and early warnings.

[0029] In one embodiment, the working process of the deep learning and model updating module includes: Collect high-quality restored signals output by the signal recovery and enhancement module during historical detection.

[0030] Build a deep learning model based on the convolutional neural network architecture; the deep learning model includes: convolution layer, ReLU function, pooling layer and fully connected layer.

[0031] The high-quality restored signal of the historical detection process is used to train the deep learning model to obtain a trained deep learning model.

[0032] The high-quality recovered signal output by the signal recovery and enhancement module in the real-time detection process is processed using a trained deep learning model to update the electromagnetic shielding strategy and data recovery mode.

[0033] In one of the embodiments, during the deep learning model training process, if different electromagnetic pulse interferences are classified, the loss function used in the training process is the cross entropy loss function; if the signal strength is predicted, the loss function used in the training process is the mean square error loss function.

[0034] A method for anti-electromagnetic interference of a short-range UAV based on deep learning, the method comprising: The electromagnetic pulse sensing module monitors the surrounding electromagnetic interference signals in real time, analyzes and extracts the intensity, frequency, and waveform of the electromagnetic pulse signals, and then transmits the signals to the intelligent decision-making and control module and the intelligent electromagnetic pulse shielding module.

[0035] Based on the signal transmitted by the electromagnetic pulse sensing module, the intelligent electromagnetic pulse shielding module automatically enables electromagnetic shielding and cooperates with the signal recovery and enhancement module to ensure the stability of the communication signal.

[0036] Collect data from the electromagnetic pulse sensing module and the signal recovery and enhancement module, and dynamically optimize the system configuration based on real-time data to make real-time adjustments and optimization decisions.

[0037] Based on the data output by the signal recovery and enhancement module, deep learning is used to predict future electromagnetic interference events. According to the prediction results, early warnings are issued and the intelligent decision-making and control modules are prompted to adjust their strategies.

[0038] Based on the collected electromagnetic pulse signal data, online learning and optimization are carried out to continuously update the electromagnetic shielding strategy and enhance the system's anti-interference capability.

[0039] The above-mentioned short-range UAV anti-electromagnetic interference device and method based on deep learning, the device includes: electromagnetic pulse sensing module, power module, intelligent electromagnetic pulse shielding module, signal recovery and enhancement module, intelligent decision-making and control module, deep learning and model update module. This device integrates multiple intelligent modules, can sense and analyze electromagnetic pulses in real time, automatically adjust the electromagnetic shielding strength, enhance signal recovery capability, and ensure the stable operation of the equipment in a complex electromagnetic environment; this device has automated decision-making and online prediction capabilities based on deep learning, can warn of electromagnetic interference events in advance and optimize protection strategies; it uses highly sensitive sensors to avoid missed detection of electromagnetic pulse signals, and at the same time establishes a false alarm correction mechanism in the deep learning system. By introducing integrated learning technology, the outputs of multiple detection models are comprehensively judged. If the judgment results of multiple models are consistent, the detection results are more reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A block diagram of a short-range UAV anti-electromagnetic interference device based on deep learning in one embodiment; Figure 2 This is a workflow diagram of a short-range UAV anti-electromagnetic interference device based on deep learning in one embodiment; Figure 3 A flowchart of the working principle and function of an electromagnetic pulse induction module in one embodiment; Figure 4 A sample diagram of the inner shielding layer material in another embodiment; Figure 5 The present invention is a flowchart of a method for short-range UAV anti-electromagnetic interference based on deep learning in one embodiment. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0042] In one embodiment, Figure 1 As shown, a short-range UAV anti-electromagnetic interference device based on deep learning is provided, which is suitable for anti-electromagnetic pulse interference equipment of a short-range UAV platform, and the device includes: The electromagnetic pulse sensing module 10 is used to detect electromagnetic pulse signals in the surrounding environment in real time from low frequency to ultra-high frequency, extract features of the electromagnetic pulse signals, and transmit the extracted features to the intelligent electromagnetic pulse shielding module 20, the intelligent decision and control module 40 and the signal recovery and enhancement module 30.

[0043] Specifically, the electromagnetic pulse sensing module can detect electromagnetic interference signals in the surrounding environment in real time, especially electromagnetic pulse signals, and analyze their intensity, frequency, waveform and other characteristics.

[0044] The electromagnetic pulse sensing module 10 is a key component for monitoring and responding to electromagnetic interference. Its main function is to sense and detect electromagnetic pulses in the surrounding environment in real time, and feed back the detected signals to the intelligent electromagnetic pulse shielding module and the intelligent decision-making and control module, thereby triggering corresponding protection or adjustment measures.

[0045] The signal collection preprocessing performed by the electromagnetic pulse induction module 10 includes the collection of interference signals and low and high frequencies.

[0046] The electromagnetic pulse sensing module 10 is not limited to sensing the electromagnetic pulse signal in the surrounding environment, but can also capture the electromagnetic waveform and identify its frequency, amplitude and other related characteristics. In addition, the electromagnetic pulse sensing module will also evaluate the intensity of the electromagnetic pulse and send feedback information in time. If the electromagnetic pulse exceeds the threshold, the intelligent electromagnetic pulse shielding module 20 will be automatically turned on.

[0047] The intelligent electromagnetic pulse shielding module 20 is used to automatically start the intelligent electromagnetic pulse shielding module 20 when the intensity of the electromagnetic pulse signal exceeds a preset threshold value. After the intelligent electromagnetic pulse shielding module is started, the electromagnetic pulse signal is shielded and isolated according to the electromagnetic shielding strategy; the electromagnetic shielding strategy is obtained by continuously adjusting and optimizing according to the received feedback signal; Specifically, the intelligent electromagnetic pulse shielding module 20 is mainly used to respond to and resist electromagnetic pulses or electromagnetic interference signals, and protect the equipment from strong electromagnetic interference to electronic components, communication equipment and system functions. The intelligent electromagnetic shielding module 20 realizes shielding and protection against electromagnetic pulse interference through intelligent algorithms and hardware devices. Among them, the intelligent algorithm refers to the model in the deep learning and model updating module 50.

[0048] The signal recovery and enhancement module 30 is used to recover the signal damaged by electromagnetic interference to obtain a high-quality recovered signal.

[0049] Specifically, signal recovery can extract useful signals from complex signals affected by noise or interference, and signal enhancement can further improve the quality and strength of signals based on recovery. The signal recovery and enhancement module 30 is used to recover signals damaged by electromagnetic interference and improve signal quality through enhancement algorithms to ensure that important data and communication signals are not lost or affected.

[0050] The intelligent decision-making and control module 40 is used to perform real-time data analysis on the signals detected by the electromagnetic pulse induction module and the high-quality recovery signals, and then use intelligent algorithms to optimize the electromagnetic shielding strategy and signal recovery mode, and adjust the operating status of the device.

[0051] Specifically, the intelligent decision and control module 40 is responsible for optimizing the working state of the entire system in real time, adjusting the electromagnetic shielding strategy, signal recovery mode, etc. based on real-time data analysis and intelligent algorithms according to the data provided by the electromagnetic interference sensing module and the signal recovery module.

[0052] The steps of real-time data analysis and intelligent algorithms include: 1) Obtain data from the electromagnetic pulse sensing module and the signal recovery and enhancement module.

[0053] 2) Combine intelligent algorithms to perform real-time analysis of electromagnetic interference signals, specifically to extract new key features of interference, evaluate the quality of interference, and identify the type of interference model.

[0054] 3) According to the frequency spectrum characteristics of the interference signal, automatically select the appropriate shielding frequency band and restore the signal quality.

[0055] Preferably, a common power optimization system integrates power optimization decisions in an intelligent decision-making and control module to achieve optimized power management, and a specific implementation method is, for example, a mobile phone battery optimization system.

[0056] The deep learning and model updating module 50 is used to optimize the anti-interference model and protection strategy through deep learning methods and adaptive updating mechanisms.

[0057] Specifically, the deep learning and model updating module 50 performs online learning by continuously accumulating electromagnetic interference data to optimize the anti-interference model and protection strategy.

[0058] The power supply module 60 is used to provide power for the short-range UAV anti-electromagnetic interference device, and automatically adjusts the output voltage and current according to the power requirements of the electromagnetic pulse sensing module 10, the intelligent electromagnetic pulse shielding module 20, the signal recovery and enhancement module 30, the intelligent decision-making and control module 40, and the deep learning and model updating module 50.

[0059] Specifically, the power module 60 can distribute the electric energy to each module through the internal power control unit, and automatically adjust the output voltage and current according to the power requirements of each module. At the same time, the power module 60 supports multiple power inputs, such as batteries, external power supplies, solar photovoltaics, etc., and the relevant batteries should meet the minimum voltage requirements.

[0060] The device also includes a signal transceiver module for transmitting signals.

[0061] In the above-mentioned short-range UAV anti-electromagnetic interference device based on deep learning, the device includes: an electromagnetic pulse sensing module, a power module, an intelligent electromagnetic pulse shielding module, a signal recovery and enhancement module, an intelligent decision-making and control module, and a deep learning and model update module. This device integrates multiple intelligent modules, can sense and analyze electromagnetic pulses in real time, automatically adjust the electromagnetic shielding strength, enhance the signal recovery capability, and ensure the stable operation of the equipment in a complex electromagnetic environment; this device has automated decision-making and online prediction capabilities based on deep learning, can warn of electromagnetic interference events in advance and optimize protection strategies; it uses highly sensitive sensors to avoid missed detection of electromagnetic pulse signals, and at the same time establishes a false alarm correction mechanism in the deep learning system. By introducing integrated learning technology, the outputs of multiple detection models are comprehensively judged. If the judgment results of multiple models are consistent, the detection results are more reliable.

[0062] In one embodiment, the electromagnetic pulse induction module includes: a plurality of high-sensitivity electromagnetic induction sensors, a signal analysis and processing unit, and a filtering unit.

[0063] The electromagnetic induction sensor is used to detect electromagnetic interference signals in the surrounding environment in real time from low frequency to ultra-high frequency. If the electromagnetic interference signal is an electromagnetic pulse signal, the electromagnetic pulse signal is transmitted to the signal analysis and processing unit; otherwise, the electromagnetic interference signal is marked and discarded, and electromagnetic detection continues.

[0064] The signal analysis and processing unit is used to decompose the electromagnetic pulse signals detected by multiple electromagnetic induction sensors by using a spectrum analysis algorithm.

[0065] The filtering unit is used to filter the interference signal and transmit the filtered signal to the intelligent decision-making and control module; if the intensity of the electromagnetic pulse signal exceeds the preset threshold, the intelligent electromagnetic pulse shielding module is triggered; if the intensity of the electromagnetic pulse signal does not exceed the preset threshold, the electromagnetic detection continues.

[0066] Specifically, the electromagnetic pulse sensing module is the core component of the short-range UAV anti-electromagnetic interference device based on deep learning, which is responsible for real-time monitoring of electromagnetic pulse (EMP) signals in the environment. The module can detect multiple key parameters of electromagnetic pulses, including electric field strength (unit: volts per meter, V / m), magnetic field strength (unit: amperes per meter, A / m), frequency spectrum distribution, pulse width (unit: milliseconds, ms) and peak power (unit: kilowatts, kW).

[0067] The electromagnetic pulse induction module includes: multiple high-sensitivity electromagnetic induction sensors. Through the high-sensitivity electromagnetic induction sensors, the induction unit can fully capture the electromagnetic signal and perform preliminary processing. When the intensity of the electromagnetic pulse signal exceeds the set threshold, the induction unit will trigger the command and transmit the monitored signal data to the intelligent decision-making and control module. The specific high-sensitivity electromagnetic induction sensor is a high-frequency electromagnetic radiation detector. The specific model can refer to HF. In addition, the electromagnetic pulse induction module also includes a signal processing and analysis unit and a filtering unit (the filtering unit includes: low-pass, high-pass filter, band-pass filter), etc. The signal processing and analysis unit uses a spectrum analysis algorithm to decompose the captured electromagnetic signal, and the filtering unit filters the interference signal. Finally, the sensed electromagnetic pulse will be captured, pre-processed and transmitted to the intelligent decision-making and control module through a high-speed data transmission interface. The working principle and function workflow of the electromagnetic pulse induction module are shown in the following figure. Figure 2 shown.

[0068] In response to the existing phenomenon of missed detection or false alarm in electromagnetic pulse signal detection, the present invention adopts a high-sensitivity sensor to avoid missed detection of electromagnetic pulse signals. At the same time, a false alarm correction mechanism is established in the deep learning system. By introducing integrated learning technology, the outputs of multiple detection models are comprehensively judged. If the judgment results of multiple models are consistent, the detection results are more reliable.

[0069] The false alarm correction mechanism refers to: using integrated learning technology to comprehensively judge the output results of multiple independent detection models. Specifically, the collected electromagnetic signals will first be analyzed through multiple detection models. If a signal is determined to be an electromagnetic pulse signal by multiple models, but it is not actually an interference signal, the deep learning algorithm will mark these signals as "false alarms" and correct the model through training data to prevent the same type of false alarms from happening again.

[0070] The specific process of the false alarm correction mechanism is as follows: First, use high-sensitivity electromagnetic sensors to detect electromagnetic signals in the environment, including full-band signals from low frequency to ultra-high frequency.

[0071] Secondly, by deploying multiple basic models (such as SVM + RF + GBDT, CNN + LSTM + Transformer), each model judges the signal according to its own characteristics.

[0072] The output results of multiple models are integrated through ensemble learning technology (Stacking method). If the model judgment results are consistent, the signal is considered reliable; if there is a disagreement, it enters the false alarm correction module.

[0073] Finally, based on the trained deep learning model, the signal features are secondary judged, and then the results of the integrated learning are corrected.

[0074] In one of the embodiments, the intelligent electromagnetic pulse shielding module includes an interference signal transmitting unit and an electromagnetic shielding cover.

[0075] The interference signal transmitting unit is used to dynamically adjust the frequency, amplitude, and phase through pseudo-random frequency generation technology, and intelligently control it. It adopts a multi-wave interference method and uses broadband noise and pseudo-random signals to cover multiple frequency bands to interfere with the target device.

[0076] The electromagnetic shielding cover has a rectangular shape, and a flexible electromagnetic shielding material is used as a shielding layer inside the electromagnetic shielding cover.

[0077] Specifically, the operation effect of the intelligent electromagnetic pulse shielding module is to respond to electromagnetic pulses of different intensities and frequencies by dynamically emitting interference signals, and at the same time combine flexible electromagnetic shielding materials to shield electromagnetic pulses or electromagnetic signal interference. This module needs to be combined with the above-mentioned electromagnetic pulse sensing module to achieve intelligent detection, adjustment, and shielding.

[0078] The outer structure of the electromagnetic shield is made of stainless steel, while the shielding layer is made of nanocomposite materials, specifically, woven graphene, nanocarbon tubes, nanometal tubes and resin matrix. Figure 4 shown.

[0079] Dynamically transmitting interference signals mainly uses pseudo-random frequency generation technology to dynamically adjust the frequency, amplitude, phase, etc. At the same time, intelligent control is used to adopt multi-wave interference, using broadband noise and random signals to cover multiple frequency bands to interfere with the target device. Interference signals can also be specific to the interference target through target pointing.

[0080] Preferably, the electromagnetic shielding cover has a rectangular parallelepiped structure with a length of 20 cm, a width of 20 cm, and a height of 3.5 cm.

[0081] In one embodiment, the signal recovery and enhancement module includes: a noise removal unit, a signal time domain analysis unit and a signal reconstruction unit.

[0082] The noise removal unit is used to suppress low-frequency or high-frequency noise in the signal transmitted by the electromagnetic pulse induction module by using a filter that should be selected in an adaptive manner based on deep learning.

[0083] The signal time domain analysis unit is used to decompose the superimposed electromagnetic pulse signal after denoising by using time-frequency analysis, and to separate the superimposed electromagnetic pulse signal after denoising from the time domain and the frequency domain.

[0084] The signal reconstruction unit is used to recover the signal loss caused by electromagnetic interference by using a signal recovery algorithm to obtain a high-quality recovery signal, and transmit the high-quality recovery signal to the deep learning and model updating module.

[0085] Specifically, the signal recovery and enhancement module uses time-frequency analysis to decompose the superimposed electromagnetic pulse signals, and separates the superimposed signals in the time domain and frequency domain to improve the signal resolution. Finally, the deep learning algorithm is combined to further improve the accuracy of superimposed signal separation, so that the system can extract multiple independent electromagnetic pulse sources from the mixed signal.

[0086] Signal recovery can extract useful signals from complex signals affected by noise or interference, while signal enhancement and recovery can further improve the quality and strength of the signal.

[0087] Signal recovery and enhancement both rely on deep learning technology. In this application, signal recovery and enhancement technology is mainly aimed at noise removal, signal time domain analysis and signal reconstruction. Specifically: low-frequency or high-frequency noise is suppressed by low-pass filters and high-pass filters. The adaptive selection of filters here mainly relies on intelligent judgment, that is, the adaptive function of deep learning, and the signal time domain analysis is to segment the signal, and then perform Fourier transform on each segment to obtain the change of the signal in time and frequency, and then extract the intrinsic mode function of the signal through the combination of empirical mode decomposition (EMD) and Hilbert transform, so as to restore the main characteristics of the signal. Signal reconstruction is mainly aimed at the missing signal. At this time, interpolation and extrapolation are used to restore the missing part of the signal. The above three parts can be completed autonomously using intelligent modules.

[0088] In one of the embodiments, the working process of the intelligent decision-making and control module includes: receiving the electric field strength, magnetic field strength, frequency spectrum, pulse width, and peak power of the electromagnetic pulse signal, and pre-processing the information that may contain noise or invalid information; parsing the electromagnetic pulse signal through spectrum analysis and time domain analysis, extracting key information from the electromagnetic pulse signal, and using deep learning and machine learning algorithms to extract effective features from the signal (effective features refer to the features output by the feature matrix, such as time window, mean, variance, peak value, pulse width, frequency spectrum peak value, main frequency component, bandwidth, energy density, etc., determined according to needs), judging the type, intensity and potential impact of electromagnetic interference; identifying the interference source based on the results of data analysis, and evaluating its threat level, and changing the shielding strategy of the intelligent electromagnetic pulse shielding module in real time based on the evaluation results of the interference source, and selecting Select the most suitable shielding strategy; after the evaluation and decision are completed, a control instruction will be generated and passed to the intelligent electromagnetic pulse shielding module and the signal recovery and enhancement module; real-time monitoring of the shielding effect of the intelligent electromagnetic pulse shielding module and the signal recovery quality of the signal recovery and enhancement module; regular collection of feedback data from the electromagnetic pulse sensing unit and the shielding device, evaluate the effectiveness of the current strategy, and automatically adjust and optimize the electromagnetic shielding strategy based on the feedback data; through the continuous accumulation of electromagnetic interference data, the interference pattern is learned by a deep learning algorithm, and the anti-interference model is optimized based on the newly acquired experience and data (the anti-interference model is the core technology used to resist, mitigate or eliminate the impact of electromagnetic interference on the normal operation of the system), and the control strategy is updated (the control strategy includes: shielding optimization, signal recovery, intelligent algorithm system parameter optimization, decision optimization, model optimization).

[0089] Specifically, the intelligent decision-making and control module will conduct a detailed analysis of the transmitted data, first perform the screening process of erroneous data and incomplete data, and encode and convert the valid data. The specific steps are as follows: 1) Data reception and processing: receiving the electric field strength, magnetic field strength, frequency spectrum, pulse width, peak power, etc. of the electromagnetic pulse, and pre-processing the information that may contain noise or invalid information.

[0090] 2) Through spectrum analysis and time domain analysis, the electromagnetic pulse signal is analyzed in detail to extract key information from the signal. Using deep learning and machine learning algorithms, effective features are extracted from the signal to determine the type, intensity and potential impact of electromagnetic interference.

[0091] 3) Based on the results of data analysis, the intelligent decision module will identify the interference source and evaluate its threat level. The system will determine whether it poses a threat to the drone by comprehensively analyzing the intensity and spectrum characteristics of the electromagnetic pulse. According to the evaluation results of the interference source, the decision module will adjust the working mode of the electromagnetic pulse shielding device in real time and select the most suitable shielding strategy. 4) After the evaluation and decision-making are completed, the intelligent decision-making and control module will generate control instructions and pass them to related equipment (such as electromagnetic pulse shielding devices, signal recovery modules, etc.). These instructions will guide the system to adjust the current operating mode to maximize the anti-interference ability.

[0092] 5) The intelligent decision-making module monitors the shielding effect and signal recovery quality in real time. The system regularly collects feedback data from the electromagnetic pulse sensing unit and the shielding device to evaluate the effectiveness of the current strategy. Based on the feedback data, the decision-making module automatically adjusts the optimization strategy.

[0093] 6) By continuously accumulating electromagnetic interference data, the intelligent decision-making module uses deep learning technology to learn interference patterns. The system will optimize the anti-interference model and update the control strategy based on the newly acquired experience and data.

[0094] The specific steps of preprocessing include: 1) Remove high and low frequency signals through bandpass filters, decompose the signal using wavelet transform, separate the noise and signal, and retain the highly correlated wavelet coefficients when reconstructing the signal.

[0095] 2) Eliminate outliers through machine learning (SVM-support vector machine), and then repair missing data through deep learning prediction (LSTM model).

[0096] 3) Split the signal into fixed time windows and process each time window independently. After splitting the signal, extract key features from each signal segment.

[0097] 4) Output the feature matrix containing the features.

[0098] The processed data will be transmitted to the intelligent electromagnetic pulse shielding module to activate the protection mechanism. After the intelligent electromagnetic pulse shielding module is started, it will shield and isolate the electromagnetic pulse signal through hardware and algorithms to ensure that the key electronic equipment and communication systems of the drone are not interfered with. However, the shielding effectiveness is not static. The intelligent electromagnetic pulse shielding module will continuously adjust and optimize the shielding strategy based on real-time feedback. The feedback information of the shielding effectiveness will be transmitted back to the intelligent decision-making and control module. The intelligent decision-making and control module will optimize the control strategy based on this feedback information and send an ultra-high frequency signal activation command to the signal transmitter, thereby achieving effective suppression of electromagnetic pulse interference and improving the stability and anti-interference ability of the drone in an electromagnetic interference environment.

[0099] In one of the embodiments, the specific steps of updating the control strategy include: if the electromagnetic interference intensity reaches a preset threshold, enabling the intelligent electromagnetic pulse shielding module for reverse interference; if the electromagnetic interference intensity exceeds the preset safety limit, automatically starting the signal recovery and enhancement module to increase the anti-interference capability of the device; using a machine learning algorithm to learn from historical electromagnetic pulse interference situations and gradually optimize the rule set and decision-making process; adjusting the output of the power module according to the electromagnetic interference intensity to ensure that the system can still operate effectively under high load conditions.

[0100] Specifically, the intelligent decision-making and control module makes real-time adjustments to different modules according to the collected electromagnetic pulse signals and based on a predetermined set of rules. The intelligent decision-making and control module is mainly based on the collected electromagnetic pulse signals, and analyzes and processes the signals, and adjusts and optimizes decisions in real time according to the signals.

[0101] In one of the embodiments, the intelligent decision-making and control module also includes an APP interface; the APP interface is used for users to remotely view the current electromagnetic pulse interference situation, system status and the working status of each module, and is also used to display the intensity, frequency, and interference type of the electromagnetic pulse signal in real time, and allows users to make adjustments or choose manual control according to actual conditions; it is also used to receive feedback from the system based on rule sets and real-time decisions, and provide optimization suggestions and early warnings.

[0102] Specifically, the rule set is a set of decision rules based on historical data, environmental parameters, user needs and best practices. Real-time decision making is based on data analysis of various modules in the device (such as electromagnetic pulse sensing module, signal recovery module, etc.).

[0103] In one of the embodiments, the working process of the deep learning and model update module includes: collecting high-quality recovery signals output by the signal recovery and enhancement module during historical detection processes; constructing a deep learning model based on a convolutional neural network architecture; the deep learning model includes: a convolutional layer, a ReLU function, a pooling layer, and a fully connected layer; using the high-quality recovery signals of the historical detection process to train the deep learning model to obtain a trained deep learning model; using the trained deep learning model to process the high-quality recovery signals output by the signal recovery and enhancement module during real-time detection processes, and updating the electromagnetic shielding strategy and data recovery mode.

[0104] Specifically, the deep learning and model update module continuously learns the characteristics and interference patterns of electromagnetic pulse (EMP) signals, and optimizes the model based on historical data and real-time feedback to achieve accurate electromagnetic protection. The specific components of this module are: data collection, deep learning model training, real-time model update, decision output and adjustment, adaptive learning and feedback. The specific workflow of the deep learning and model update module is as follows: 1) This module uses the convolutional neural network (CNN) architecture. The input layer receives the processed electromagnetic signal and each signal is represented as an N-dimensional matrix. Each element of the matrix corresponds to the characteristics of the signal, such as frequency and intensity. Considering that the subsequent data may be multi-channel data, the input layer is an N-dimensional matrix. The next is the convolution layer. The specific design is that the size, step size and number of the convolution kernel have characteristics to determine the signal characteristics. or or larger. The activation function of the convolution layer uses the ReLU function. The window size of the pooling layer uses or ,The approach in the present invention is to reduce the signal to a smaller representation through ,pooling operation to improve computational efficiency.,Finally, there is a fully connected layer that is responsible for combining the ,features extracted by the convolutional layer and the pooling layer to make the final ,decision.

[0105] 2) The loss functions used by deep learning models are different. The loss function used to classify different types of electromagnetic pulse interference is the cross entropy loss function, and if the signal strength is predicted, the mean square error (MSE) loss function is used. The algorithm is optimized by using the Adam optimizer or the gradient descent method to update the network weights to minimize the loss function.

[0106] 3) For model training, the data is divided into several small batches (Batch), and only one batch of data is used each time the network weight is updated. In addition, all samples in the entire training set will be used for training multiple times, and each iteration is called an epoch. During the training process, after each epoch, the model will evaluate the validation set to monitor overfitting and underfitting during the training process.

[0107] 4) By increasing the number of convolutional layers or adjusting the size of the convolution kernel, the learning ability of the model can be enhanced, but it may also lead to an increase in computational complexity. During the training process, a certain proportion of neurons can be randomly discarded to prevent the model from overfitting, and overfitting can also be prevented by penalizing neurons with excessive weights.

[0108] In one of the embodiments, during the deep learning model training process, if different electromagnetic pulse interferences are classified, the loss function used in the training process is the cross entropy loss function; if the signal strength is predicted, the loss function used in the training process is the mean square error loss function.

[0109] In one embodiment, in a short-range UAV anti-electromagnetic interference device based on deep learning, all modules are installed on the control board and autonomously controlled by the AI ​​chip, which can detect and quantify the electromagnetic interference intensity of the surrounding environment of the short-range UAV in real time. The power module of the device adopts a compact rechargeable CTP technology battery, and the battery unit is also arranged at the end of the mainboard for easy charging. The electromagnetic pulse sensing module is installed on the front face of the system. In addition, a power indicator light and a charging interface are provided on the right side panel of the device. The power indicator light is designed to provide real-time power status feedback to alert users to pay attention to the power consumption of the device. The charging interface is designed to provide power supply for the device.

[0110] In one embodiment, Figure 5 As shown, a method for anti-electromagnetic interference of a short-range UAV based on deep learning is provided, and the method includes: Step 500: Monitor the surrounding electromagnetic interference signals in real time through the electromagnetic pulse sensing module, analyze and extract the strength, frequency, and waveform of the electromagnetic pulse signal, and then transmit the signal to the intelligent decision and control module and the intelligent electromagnetic pulse shielding module.

[0111] Step 502: Based on the signal transmitted by the electromagnetic pulse sensing module, the intelligent electromagnetic pulse shielding module automatically enables electromagnetic shielding and cooperates with the signal recovery and enhancement module to ensure stable communication signals.

[0112] Specifically, the electromagnetic pulse sensing module reports the detection result to the intelligent decision and control module, and based on this result, the intelligent decision and control module decides whether to start the intelligent electromagnetic pulse shielding module. After the intelligent electromagnetic pulse shielding module is started, the electromagnetic signal is isolated, and the signal recovery and enhancement module automatically starts working.

[0113] Step 504: Collect data from the electromagnetic pulse sensing module and the signal recovery and enhancement module, and dynamically optimize the system configuration based on real-time data to make real-time adjustments and optimization decisions.

[0114] Specifically, the process of optimizing device performance is carried out by automatically analyzing, evaluating and adjusting device settings through algorithms and intelligent decision-making mechanisms using data collected in real time.

[0115] The specific steps for real-time adjustment and optimization decision-making based on real-time data dynamic optimization system configuration include: 1) The device continuously monitors and collects data from the electromagnetic interference environment through the electromagnetic pulse sensing module and the signal recovery and enhancement module. 2) Through deep learning algorithms, data is analyzed in real time and decisions are made based on the current electromagnetic environment and signal status.

[0116] 3) Based on continuous feedback and data collection, the device can automatically update and adjust the optimization strategy after long-term operation.

[0117] Step 506: Use deep learning to predict future electromagnetic interference events based on the data output by the signal recovery and enhancement module, issue a warning based on the prediction results, and prompt the intelligent decision-making and control module to adjust the strategy.

[0118] Specific steps for intelligent decision-making and control module adjustment strategy: 1) Predict future electromagnetic interference events based on current signal data. For example, predict that a certain frequency band will experience increased interference in a certain period of time in the future, or that a certain interference pattern may continue or increase.

[0119] 2) Once a possible electromagnetic interference event is predicted, the system will trigger an early warning. A warning will be displayed on the APP.

[0120] 3) The intelligent decision-making and control module makes adjustments based on the prediction results, such as increasing the shielding strength of a specific frequency band, starting appropriate signal enhancement and recovery modes, switching to alternative frequency bands, or taking other anti-interference measures.

[0121] 4) By continuously accumulating data on electromagnetic interference events, the prediction model can be continuously improved to make future predictions more accurate.

[0122] Step 508: Based on the collected electromagnetic pulse signal data, online learning and optimization are performed to continuously update the electromagnetic shielding strategy and enhance the anti-interference capability of the system.

[0123] Compared with the prior art, the beneficial effects of the above-mentioned short-range UAV anti-electromagnetic interference method based on deep learning are as follows: 1) Traditional electromagnetic interference protection devices mostly adopt static shielding design and rely on preset protection parameters. This application uses intelligent algorithms to dynamically adjust the shielding strength according to real-time changes in the interference environment, which can effectively respond to different types of electromagnetic interference and provide more accurate protection.

[0124] 2) This application uses a signal recovery and enhancement module and an advanced signal recovery algorithm to efficiently recover signal loss caused by electromagnetic interference and enhance signal quality to ensure data integrity and communication stability.

[0125] 3) This application can perform online learning and model optimization based on historical data and real-time data. The deep learning algorithm can continuously update the anti-interference model according to the changing rules of electromagnetic interference, thereby improving the accuracy and response speed of anti-electromagnetic interference.

[0126] 4) The anti-electromagnetic interference equipment and technology proposed in this application have the potential for application in the field of electromagnetic interference monitoring and electromagnetic interference shielding. The device can realize real-time monitoring and accurate identification of dynamic changes of electromagnetic fields within a specified monitoring range, thereby effectively ensuring the safety of short-range drone operations in the monitoring area and avoiding potential electromagnetic interference risks; It should be understood that although Figure 5 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 5 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0127] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0128] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A short-range UAV anti-electromagnetic interference device based on deep learning, characterized in that: The device comprises: The electromagnetic pulse sensing module is used to detect electromagnetic pulse signals in the surrounding environment in real time, ranging from low frequency to ultra-high frequency, extract features of the electromagnetic pulse signals, and transmit the extracted features to the intelligent electromagnetic pulse shielding module, the intelligent decision-making and control module, and the signal recovery and enhancement module; An intelligent electromagnetic pulse shielding module is used to automatically start the intelligent electromagnetic pulse shielding module when the intensity of the electromagnetic pulse signal exceeds a preset threshold value. After the intelligent electromagnetic pulse shielding module is started, the electromagnetic pulse signal is shielded and isolated according to the electromagnetic shielding strategy; the electromagnetic shielding strategy is obtained by continuously adjusting and optimizing according to the received feedback signal; The signal recovery and enhancement module is used to recover the signal damaged by electromagnetic interference and obtain high-quality recovered signal; An intelligent decision-making and control module, used for performing real-time data analysis on the signal detected by the electromagnetic pulse induction module and the high-quality recovery signal, and then using an intelligent algorithm to optimize the electromagnetic shielding strategy and signal recovery mode, and adjust the device operation state; Deep learning and model update module, used to optimize anti-interference models and protection strategies through deep learning methods and adaptive update mechanisms; The power supply module is used to provide power to the short-range UAV anti-electromagnetic interference device and automatically adjust the output voltage and current according to the power requirements of each module.

2. According to the deep learning-based short-range UAV anti-electromagnetic interference device of claim 1, it is characterized in that: The electromagnetic pulse induction module includes: a plurality of high-sensitivity electromagnetic induction sensors, a signal analysis and processing unit, and a filtering unit; The electromagnetic induction sensor is used to detect electromagnetic interference signals in the surrounding environment in real time from low frequency to ultra-high frequency. If the electromagnetic interference signal is an electromagnetic pulse signal, the electromagnetic pulse signal is transmitted to the signal analysis and processing unit; otherwise, the electromagnetic interference signal is marked and discarded, and electromagnetic detection is continued; The signal analysis and processing unit is used to decompose the electromagnetic pulse signals detected by the multiple electromagnetic induction sensors by using a spectrum analysis algorithm; The filtering unit is used to filter the interference signal and transmit the filtered signal to the intelligent decision-making and control module; if the intensity of the electromagnetic pulse signal exceeds the preset threshold, the intelligent electromagnetic pulse shielding module is triggered; if the intensity of the electromagnetic pulse signal does not exceed the preset threshold, the electromagnetic detection continues.

3. The short-range UAV anti-electromagnetic interference device based on deep learning according to claim 1 is characterized in that: The intelligent electromagnetic pulse shielding module includes an interference signal transmitting unit and an electromagnetic shielding cover; The interference signal transmitting unit is used to dynamically adjust the frequency, amplitude, and phase through pseudo-random frequency generation technology, and intelligently control, using a multi-wave interference method, using broadband noise and pseudo-random signals to cover multiple frequency bands to interfere with the target device; The electromagnetic shielding cover has a structural shape of a cuboid, and a flexible electromagnetic shielding material is used as a shielding layer inside the electromagnetic shielding cover.

4. The short-range UAV anti-electromagnetic interference device based on deep learning according to claim 1 is characterized in that: The signal recovery and enhancement module includes: a noise removal unit, a signal time domain analysis unit and a signal reconstruction unit; The noise removal unit is used to suppress low-frequency or high-frequency noise in the signal transmitted by the electromagnetic pulse induction module by using a filter that should be selected in an adaptive manner based on deep learning; The signal time domain analysis unit is used to decompose the superimposed electromagnetic pulse signal after denoising by using time-frequency analysis, and separate the superimposed electromagnetic pulse signal after denoising from the time domain and the frequency domain; The signal reconstruction unit is used to use a signal recovery algorithm to recover the signal loss caused by electromagnetic interference, obtain a high-quality recovery signal, and transmit the high-quality recovery signal to the deep learning and model updating module.

5. The short-range UAV anti-electromagnetic interference device based on deep learning according to claim 1 is characterized in that: The working process of the intelligent decision-making and control module includes: Receive the electric field strength, magnetic field strength, frequency spectrum, pulse width, peak power of the electromagnetic pulse signal, and pre-process the information that may contain noise or invalid information; Through spectrum analysis and time domain analysis, electromagnetic pulse signals are analyzed to extract key information from them. Deep learning and machine learning algorithms are used to extract effective features from the signals to determine the type, intensity, and potential impact of electromagnetic interference. Identify interference sources based on the results of data analysis and evaluate their threat levels. Based on the evaluation results of interference sources, adjust the working mode of the electromagnetic pulse shielding device in real time and select the most suitable shielding strategy; After the evaluation and decision are completed, control instructions will be generated and transmitted to the intelligent electromagnetic pulse shielding module and the signal recovery and enhancement module; Real-time intelligent monitoring of electromagnetic pulse shielding module shielding effect and signal recovery quality of signal recovery and enhancement module; regular collection of feedback data from electromagnetic pulse sensing unit and shielding device, evaluation of the effectiveness of current strategy, and automatic adjustment of optimization strategy based on feedback data; By continuously accumulating electromagnetic interference data, deep learning algorithms are used to learn interference patterns, and based on newly acquired experience and data, the anti-interference model is optimized and the control strategy is updated.

6. The short-range UAV anti-electromagnetic interference device based on deep learning according to claim 5 is characterized in that: The specific steps to update the control strategy include: If the electromagnetic interference intensity reaches a preset threshold, the intelligent electromagnetic pulse shielding module is enabled to perform reverse interference; If the electromagnetic interference intensity exceeds the preset safety limit, the signal recovery and enhancement module is automatically activated to increase the anti-interference capability of the device; Use machine learning algorithms to learn from historical electromagnetic pulse interference situations and gradually optimize the rule set and decision-making process; The output of the power module is adjusted according to the electromagnetic interference intensity.

7. The short-range UAV anti-electromagnetic interference device based on deep learning according to claim 1 is characterized in that: The intelligent decision-making and control module also includes an APP interface; the APP interface is used by users to remotely view the current electromagnetic pulse interference situation, system status and the working status of each module, and is also used to display the strength, frequency and interference type of the electromagnetic pulse signal in real time, and allows users to adjust or choose manual control according to actual conditions; it is also used to receive feedback from the system based on rule sets and real-time decisions, and provide optimization suggestions and early warnings.

8. The short-range UAV anti-electromagnetic interference device based on deep learning according to claim 1 is characterized in that: The working process of the deep learning and model updating module includes: Collect high-quality restored signals output by the signal recovery and enhancement module during historical detection; Construct a deep learning model based on a convolutional neural network architecture; the deep learning model includes: a convolutional layer, a ReLU function, a pooling layer, and a fully connected layer; The deep learning model is trained using the high-quality restored signal of the historical detection process to obtain a trained deep learning model; The high-quality recovered signal output by the signal recovery and enhancement module in the real-time detection process is processed using a trained deep learning model to update the electromagnetic shielding strategy and data recovery mode.

9. The short-range UAV anti-electromagnetic interference device based on deep learning according to claim 8 is characterized in that: During the deep learning model training process, if different electromagnetic pulse interferences are classified, the loss function used in the training process is the cross entropy loss function; if the signal strength is predicted, the loss function used in the training process is the mean square error loss function.

10. A method for short-range UAV anti-electromagnetic interference based on deep learning, characterized in that: The method comprises: The electromagnetic pulse sensing module monitors the surrounding electromagnetic interference signals in real time, analyzes and extracts the strength, frequency and waveform of the electromagnetic pulse signals, and then transmits the signals to the intelligent decision-making and control module and the intelligent electromagnetic pulse shielding module; Based on the signal transmitted by the electromagnetic pulse sensing module, the intelligent electromagnetic pulse shielding module automatically activates electromagnetic shielding and cooperates with the signal recovery and enhancement module to ensure the stability of the communication signal; Collect data from the electromagnetic pulse sensing module and the signal recovery and enhancement module, dynamically optimize system configuration based on real-time data, and make real-time adjustments and optimization decisions; Use deep learning to predict future electromagnetic interference events based on the data output by the signal recovery and enhancement module, issue warnings based on the prediction results, and prompt the intelligent decision-making and control module to adjust the strategy; Based on the collected electromagnetic pulse signal data, online learning and optimization are carried out to continuously update the electromagnetic shielding strategy and enhance the system's anti-interference capability.

Images