AI-based electromagnetic interference intelligent detection method and system

Abstract

The application discloses an AI-based electromagnetic interference intelligent detection method and system, relates to the technical field of signal processing, and solves the technical problem that the prior art directly identifies interference sources according to electromagnetic signals, and a large amount of calculation data exists, and in a high-frequency electromagnetic signal scene, timeliness is poor; the application obtains original electromagnetic signals; the original electromagnetic signals are pretreated to obtain multi-dimensional feature data; the original electromagnetic signals are divided according to the multi-dimensional feature data to obtain signal types; the risk level of the corresponding interference electromagnetic signals is analyzed according to the signal types; the interference electromagnetic signals are suppressed based on the risk level; interference signals can be suppressed in time, and interference sources can be screened.

Description

Technical Field

[0001] This application belongs to the field of signal processing and relates to electromagnetic interference detection technology, specifically an AI-based intelligent electromagnetic interference detection method and system. Background Technology

[0002] Electromagnetic interference (EMI) refers to any electromagnetic phenomenon that may degrade the performance of electronic equipment, systems, or signal transmission. Essentially, it is the unintended interference caused by electromagnetic energy through conduction or radiation, leading to performance degradation. Intelligent detection of EMI using advanced technologies such as artificial intelligence, big data, and the Internet of Things can significantly improve detection efficiency, accuracy, and response capabilities. Intelligent detection systems can operate 24 / 7, replacing manual inspections and reducing labor costs and time consumption. Intelligent detection algorithms can identify interference patterns, trends, and potential risks, improving detection accuracy and foresight. Simultaneously, by continuously monitoring the electromagnetic status of equipment, intelligent systems can detect potential interference risks in advance, triggering early warning mechanisms to prevent equipment damage or performance degradation due to interference, and reducing downtime losses caused by sudden failures.

[0003] Existing AI-based intelligent electromagnetic interference detection technologies can automatically learn hidden features in complex electromagnetic signals and accurately identify different types of interference sources. However, existing technologies directly identify interference sources based on electromagnetic signals, which involves a large amount of computational data and poor timeliness in high-frequency electromagnetic signal scenarios. Summary of the Invention

[0004] This application aims to solve at least one of the technical problems existing in the prior art; to this end, this application proposes an AI-based intelligent electromagnetic interference detection method and system to solve the technical problems of existing technologies that directly identify interference sources based on electromagnetic signals, which have the problems of large amount of computational data and poor timeliness in high-frequency electromagnetic signal scenarios.

[0005] To achieve the above objectives, the first aspect of this application provides an AI-based intelligent electromagnetic interference detection method, comprising: Acquire the raw electromagnetic signal; preprocess the raw electromagnetic signal to obtain multidimensional feature data; The original electromagnetic signal is divided based on multidimensional feature data to obtain the signal type; Analyze the risk level of the corresponding interfering electromagnetic signal based on the signal type; Interfering electromagnetic signals are suppressed based on risk level.

[0006] Based on the above steps, by acquiring the original electromagnetic signal and preprocessing it to obtain multi-dimensional feature data, we can more comprehensively and accurately capture the various characteristics of the electromagnetic signal. Compared with analysis methods that rely solely on a single feature, multi-dimensional feature data provides a richer and more reliable basis for signal type classification, thereby greatly improving the accuracy of signal type judgment. Analyzing the risk level of the corresponding interfering electromagnetic signal based on the accurate signal type analysis allows for a more realistic assessment of the potential impact of interfering electromagnetic signals. Suppressing interfering electromagnetic signals based on risk levels enables the adoption of appropriate suppression strategies according to different risk levels, avoiding over-processing or under-processing. This approach minimizes the interference of interfering electromagnetic signals on normal signals, ensuring the stability of signal transmission and system operation, while also making rational use of resources and improving processing efficiency. The entire process, from signal acquisition to final suppression processing, forms a complete, closed-loop electromagnetic signal management system. Through effective identification, risk assessment, and suppression of interfering electromagnetic signals, the impact of electromagnetic interference on various electronic devices and communication systems can be significantly reduced, enhancing the stability and reliability of the entire system and ensuring its normal operation in complex electromagnetic environments.

[0007] Preferably, the preprocessing of the original electromagnetic signal to obtain multidimensional feature data includes: The original electromagnetic signal is retrieved; the original electromagnetic signal is amplified using a low-noise amplifier, and a tunable filter is used for preliminary frequency band selection. The original electromagnetic signal is sampled using a high-speed analog-to-digital converter to obtain the original time series; the original time series is then denoised, filtered, and normalized; and features are extracted from the processed original time series to obtain multidimensional feature data.

[0008] Preferably, the step of extracting features from the processed original time series to obtain multidimensional feature data includes: Retrieve the processed original time series; calculate the statistical characteristics of a time window; the statistical characteristics include: mean, variance, kurtosis, skewness, pulse count rate, average pulse width, and duty cycle; The original time series was processed using Fast Fourier Transform to obtain frequency domain feature data, which includes: power spectral density, spectral centroid, spectral variance, and the frequency, amplitude, and bandwidth of the peak values. Integrate statistical features and frequency domain features into multidimensional feature data.

[0009] Preferably, the step of dividing the original electromagnetic signal based on multidimensional feature data to obtain the signal type includes: Retrieve multidimensional feature data and a trained type analysis model; input the multidimensional feature data into the type analysis model to obtain the type label corresponding to the original electromagnetic signal; wherein, the type analysis model is built based on an artificial intelligence model; the type label is set to a positive integer; Obtain the interference classification database; match the type labels of the original electromagnetic signals with the interference classification database to obtain the signal types of the original electromagnetic signals; the signal types include: interference signals and non-interference signals.

[0010] Preferably, the type analysis model is built based on an artificial intelligence model, including: Select model frameworks and deep learning algorithms from the artificial intelligence library; construct the model based on the deep learning algorithm and model framework to obtain the constructed model; Obtain the standard dataset; the standard dataset includes standard input data consistent with the content attributes of the multidimensional feature data, and standard output data consistent with the content attributes of the type label. The standard dataset is divided into a training set, a validation set, and a test set according to a preset ratio; a model is built using the training set for training; the internal parameters of the trained model are adjusted using the validation set; and the trained model is tested using the test set to obtain test metrics. Obtain the indicator threshold; compare the test indicator with the indicator threshold; if all test indicators are greater than the indicator threshold, mark the constructed model as a type analysis model; otherwise, reconstruct and train the type analysis model.

[0011] It should be noted that the split ratio of the standard dataset is set by experts, generally 8:1:1; the test metrics include: accuracy, F1 score, recall, and stability, and the thresholds for these metrics are set by technical professionals based on experimental simulations; the standard data is an integration of historical data and simulated data; the model is retrained by changing the model framework, deep learning algorithm, or the split ratio of the standard dataset.

[0012] Preferably, the step of analyzing the risk level of the corresponding interfering electromagnetic signal based on the signal type includes: Retrieve the original electromagnetic signal, which is an interference signal; extract the interference characteristics of the original electromagnetic signal; the interference characteristics include: frequency, bandwidth, and modulation type; Obtain a knowledge graph; match the interference characteristics of the original electromagnetic signal with the knowledge graph to obtain several interference sources; analyze the location of several interference sources and calculate the location distance between the location and the collection point; when the location distance is greater than the set distance threshold, the corresponding interference source is removed; otherwise, analyze the time correlation. Obtain the operating logs of the devices corresponding to the interference sources and the acquisition time of the original electromagnetic signals; calculate the time difference between the running time and the acquisition time in the operating logs; when the time difference is greater than the corresponding difference threshold, the corresponding interference source is removed; calculate and analyze the confidence level of the remaining interference sources and select the final interference source; calculate and analyze the risk level of the final interference source.

[0013] Preferably, the calculation and analysis of the confidence levels of the remaining interference sources, and the screening of the final interference sources, includes: The remaining interference sources are integrated and marked as a candidate interference source set. ; Construct the confidence analysis function: Calculate the confidence score of the candidate interference sources based on the confidence analysis function; select the interference source with the highest confidence score as the final interference source. in, ; This represents the time sequence of the occurrence of interference signals; Represents the working time series of candidate interference sources; The distribution of the time difference sequence between the time of interference occurrence and the time of equipment operation, where Var represents the variance; , , A proportionality coefficient greater than 0; It is the Kullback-Leibler divergence, used to measure the difference in feature distributions; Characteristics of interference signals; Indicates candidate interference sources Features; For spectral sensitivity parameters; To estimate the direction of arrival of the interference signal, , indicating candidate interference sources The prediction has reached the desired direction; For electromagnetic wave propagation models, Candidate interference sources The actual location; This represents the antenna beamwidth.

[0014] Preferably, the calculation and analysis of the risk level of the final interference source includes: Obtain component scores for system components; these scores include: inherent vulnerability scores and robustness scores. Construct vulnerability scoring functions for system components: ;in, as well as An adjustment coefficient greater than 0; This represents the inherent vulnerability of the j-th system component; This represents the component robustness score of the j-th system component; Construct a risk coefficient analysis function: ;in, This represents the confidence score of the final interference source; This indicates the intensity of the final interference source's impact on system component j; This represents the vulnerability score of the j-th system component; The risk coefficient of the final interference source is calculated based on the risk coefficient analysis function; the risk coefficient is matched with the corresponding level mapping table to obtain the risk level of the final interference source; the risk levels include: low, medium and high.

[0015] Preferably, the suppression processing of interfering electromagnetic signals based on risk level includes: Retrieve the risk level of the final interference source; identify the equipment model of the final interference source; combine the equipment model and risk level of the final interference source into a processing sequence; Obtain the equipment processing database; match the processing sequence with the equipment processing database to obtain the corresponding suppression measures; send the suppression measures to the corresponding technical personnel to suppress the interference signal.

[0016] The second aspect of this application provides an AI-based intelligent electromagnetic interference detection system, including: a type analysis module, and a signal acquisition module and a risk assessment module connected thereto; The signal acquisition module is used to acquire the raw electromagnetic signal; The type analysis module is used to preprocess the original electromagnetic signal to obtain multi-dimensional feature data; and to classify the original electromagnetic signal according to the multi-dimensional feature data to obtain the signal type. The risk assessment module is used to analyze the risk level of the corresponding interfering electromagnetic signal according to the signal type; and to suppress the interfering electromagnetic signal based on the risk level.

[0017] A third aspect of this application provides a computer-readable storage medium storing a computer program for performing the steps of the method described in the first aspect above.

[0018] Compared with the prior art, the beneficial effects of this application are: 1. Amplification using a low-noise amplifier enhances signal strength while reducing noise introduction, ensuring signal integrity. A tunable filter performs initial frequency band selection, filtering out irrelevant frequency bands and reducing interference, providing a cleaner and more targeted signal for subsequent processing. 2. By calculating the statistical characteristics of a time window and using Fast Fourier Transform to obtain frequency domain feature data, and integrating this into multi-dimensional feature data, the characteristics of the electromagnetic signal are comprehensively described from both the time and frequency domains. This allows for a more accurate characterization of the signal's essential attributes, providing rich information for subsequent signal classification. 3. A type analysis model based on artificial intelligence is used to process the multi-dimensional feature data, obtaining the type label corresponding to the original electromagnetic signal, which is then matched with an interference classification database to determine the signal type. The artificial intelligence model has powerful learning and classification capabilities, automatically learning the complex mapping relationship between signal features and types from large amounts of data, thus classifying signals more accurately.

[0019] 2. First, interference sources are initially screened based on distance thresholds to eliminate those too far from the acquisition point, reducing interference from irrelevant sources. Then, by comparing the operating time of the interference source device's log with the original electromagnetic signal acquisition time, interference sources with time differences exceeding a threshold are eliminated, further narrowing down the range of interference sources. This dual screening mechanism based on distance and time effectively eliminates interference sources with little correlation to the current interference signal, improving the accuracy of interference source location. A complex confidence analysis function is constructed, comprehensively considering factors such as differences in characteristic distribution, differences in direction of arrival, and time correlation. By calculating the confidence scores of candidate interference sources and selecting the highest score as the final interference source, candidate interference sources can be evaluated from multiple perspectives, ensuring that the finally determined interference source is the most likely source of the current interference signal, greatly improving the accuracy of interference source location. The constructed system component vulnerability scoring function comprehensively considers both the inherent vulnerability of the component and the component robustness score, enabling a more accurate understanding of the vulnerability level of each system component.

[0020] 3. Based on the confidence score of the final interference source, the vulnerability score of system components, and the intensity of the interference source's impact on system components, a risk coefficient analysis function is constructed to calculate the risk coefficient of the final interference source. This considers multiple key factors such as the reliability of the interference source, the vulnerability of system components, and the range and intensity of the interference's impact, enabling a more scientific assessment of the risk posed by the interference signal to the system. The risk coefficient is matched with a level mapping table to obtain a clear risk level. The device model and risk level of the final interference source are combined into a processing sequence and matched with the device processing database to quickly and accurately obtain corresponding suppression and treatment measures. The device processing database stores standard treatment schemes for various device models under different risk levels. This matching method avoids technicians blindly exploring when facing interference, saving time in developing treatment measures and improving processing efficiency. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram illustrating the overall steps of the method described in this application; Figure 2 This is a schematic diagram illustrating the signal type classification steps in this application; Figure 3 This is a schematic diagram of the signal risk assessment and suppression steps in this application; Figure 4 This is a schematic diagram of the system structure connection in this application. Detailed Implementation

[0023] The technical solutions of this application will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0024] Please see Figure 1 The first aspect of this application provides an AI-based intelligent electromagnetic interference detection method, including: S101. Acquire the original electromagnetic signal; preprocess the original electromagnetic signal to obtain multidimensional feature data; S102. Divide the original electromagnetic signal according to the multidimensional feature data to obtain the signal type; S103. Analyze the risk level of the corresponding interfering electromagnetic signal based on the signal type; S104. Suppress interference electromagnetic signals based on risk level.

[0025] Based on the above steps, by acquiring the original electromagnetic signal and preprocessing it to obtain multi-dimensional feature data, we can more comprehensively and accurately capture the various characteristics of the electromagnetic signal. Compared with analysis methods that rely solely on a single feature, multi-dimensional feature data provides a richer and more reliable basis for signal type classification, thereby greatly improving the accuracy of signal type judgment. Analyzing the risk level of the corresponding interfering electromagnetic signal based on the accurate signal type analysis allows for a more realistic assessment of the potential impact of interfering electromagnetic signals. Suppressing interfering electromagnetic signals based on risk levels enables the adoption of appropriate suppression strategies according to different risk levels, avoiding over-processing or under-processing. This approach minimizes the interference of interfering electromagnetic signals on normal signals, ensuring the stability of signal transmission and system operation, while also making rational use of resources and improving processing efficiency. The entire process, from signal acquisition to final suppression processing, forms a complete, closed-loop electromagnetic signal management system. Through effective identification, risk assessment, and suppression of interfering electromagnetic signals, the impact of electromagnetic interference on various electronic devices and communication systems can be significantly reduced, enhancing the stability and reliability of the entire system and ensuring its normal operation in complex electromagnetic environments.

[0026] In one possible implementation of the embodiments of this application, combined with Figure 1 ,like Figure 2 As shown, the above S101-S102 can be specifically implemented through the following S201-S203, which are explained in detail below: S201. Retrieve the original electromagnetic signal; amplify the original electromagnetic signal using a low-noise amplifier and perform preliminary frequency band selection using a tunable filter; sample the original electromagnetic signal using a high-speed analog-to-digital converter to obtain the original time series; perform noise reduction, filtering, and normalization on the original time series.

[0027] Example: At a wireless communication base station, raw electromagnetic signals from the surrounding environment are continuously collected. These signals may include signals emitted by various communication devices as well as environmental noise; The weak raw electromagnetic signal is amplified using a low-noise amplifier to enhance the signal strength while minimizing the introduction of additional noise. Use a tunable filter to perform preliminary screening based on a preset frequency band range, for example, only retaining signals within the 800MHz-2500MHz frequency band and filtering out interference from other irrelevant frequency bands; The filtered signal is sampled at a sampling rate of 100M times per second by a high-speed analog-to-digital converter, and the analog signal is converted into a digital signal to obtain the original time series data. The original time series is denoised by using a wavelet denoising algorithm to remove high-frequency noise; filtering is performed to further smooth the signal; and normalization is performed to adjust the signal amplitude range to between [0,1].

[0028] S202. Retrieve the processed original time series; calculate the statistical characteristics of a time window; process the processed original time series using Fast Fourier Transform to obtain frequency domain feature data; integrate the statistical characteristics and frequency domain feature data into multidimensional feature data.

[0029] The statistical features include: mean, variance, kurtosis, skewness, pulse count rate, average pulse width, and duty cycle; the frequency domain features include: power spectral density, spectral centroid, spectral variance, and the frequency, amplitude, and bandwidth of the peak value.

[0030] Example: Retrieve the preprocessed raw time series data; calculate the statistical characteristics within a 1-second time window: Mean: the average signal amplitude within this time window, calculated to be 0.3; Variance: reflects the dispersion of the signal amplitude, calculated to be 0.02; Kurtosis: describes the sharpness of the signal amplitude distribution, calculated to be 3.5; Skewness: measures the asymmetry of the signal amplitude distribution, calculated to be 0.1; Pulse Count Rate: counts the number of pulses occurring within this time window, calculated to be 10; Average Pulse Width: calculates the average of all pulse widths, calculated to be 0.05 ms; Duty Cycle: the ratio of pulse duration to the entire time window, calculated to be 0.05. The original time series was processed using Fast Fourier Transform to obtain frequency domain characteristic data: Power spectral density: showing the power distribution of the signal at different frequencies; analysis showed that the main power was concentrated around 1800MHz; Spectral centroid: indicating the center position of the signal power distribution, calculated to be 1820MHz; Spectral variance: measuring the dispersion of the spectrum, 50MHz²; Peak frequency, amplitude, and bandwidth: peak frequency is 1805MHz, amplitude is 0.8, and bandwidth is 10MHz. The statistical features and frequency domain features mentioned above are integrated into multi-dimensional feature data to form a feature vector containing multiple dimensions, which is then used for subsequent signal classification.

[0031] S203. Retrieve multidimensional feature data and the trained type analysis model; input the multidimensional feature data into the type analysis model to obtain the type label corresponding to the original electromagnetic signal; obtain the interference classification database; match the type label of the original electromagnetic signal with the interference classification database to obtain the signal type of the original electromagnetic signal.

[0032] The type analysis model is built based on an artificial intelligence model; the type label is set to a positive integer; the signal types include: interference signals and non-interference signals.

[0033] In one possible implementation, the type analysis model is built upon an artificial intelligence model, including: Select model frameworks and deep learning algorithms from the artificial intelligence library; construct the model based on the deep learning algorithm and model framework to obtain the constructed model; Obtain the standard dataset; the standard dataset includes standard input data consistent with the content attributes of the multidimensional feature data, and standard output data consistent with the content attributes of the type label. The standard dataset is divided into a training set, a validation set, and a test set according to a preset ratio; a model is built using the training set for training; the internal parameters of the trained model are adjusted using the validation set; and the trained model is tested using the test set to obtain test metrics. Obtain the indicator threshold; compare the test indicator with the indicator threshold; if all test indicators are greater than the indicator threshold, mark the constructed model as a type analysis model; otherwise, reconstruct and train the type analysis model.

[0034] It should be noted that the split ratio of the standard dataset is set by experts, generally 8:1:1; the test metrics include: accuracy, F1 score, recall, and stability, and the thresholds for these metrics are set by technical professionals based on experimental simulations; the standard data is an integration of historical data and simulated data; the model is retrained by changing the model framework, deep learning algorithm, or the split ratio of the standard dataset.

[0035] Example: Input multidimensional feature data into the type analysis model. After forward propagation calculation, the type label corresponding to the original electromagnetic signal is obtained. The type labels of the three original electromagnetic signals are 3, 7, and 4. The type labels of the three original electromagnetic signals are matched with the interference classification database. The original electromagnetic signals with type labels 4 and 7 are non-interference signals, and the original electromagnetic signal with type label 3 is an interference signal.

[0036] Based on the above steps, amplification using a low-noise amplifier enhances signal strength while reducing noise introduction and ensuring signal integrity. A tunable filter performs initial frequency band selection, filtering out irrelevant frequency bands and reducing interference, providing a cleaner and more targeted signal for subsequent processing. Calculating the statistical characteristics of a time window and using Fast Fourier Transform to obtain frequency domain feature data, and integrating this into multidimensional feature data, comprehensively describes the characteristics of the electromagnetic signal from both the time and frequency domains. This more accurately characterizes the essential attributes of the signal, providing rich information for subsequent signal classification. A type analysis model based on artificial intelligence is used to process the multidimensional feature data, obtaining the type label corresponding to the original electromagnetic signal, which is then matched with an interference classification database to determine the signal type. The artificial intelligence model has powerful learning and classification capabilities, automatically learning the complex mapping relationship between signal features and types from large amounts of data, thus classifying signals more accurately.

[0037] In one possible implementation of the embodiments of this application, combined with Figure 1 ,like Figure 3 As shown, the above S103-S104 can be specifically implemented through the following S3201-S303, which are explained in detail below: S301. Retrieve the original electromagnetic signal whose signal type is interference signal; extract the interference features of the original electromagnetic signal; obtain the knowledge graph; match the interference features of the original electromagnetic signal with the knowledge graph to obtain several interference sources; analyze the setting positions of several interference sources and calculate the position distance between the setting position and the collection point; when the position distance is greater than the set distance threshold, the corresponding interference source is removed; otherwise, analyze the time correlation.

[0038] Interference characteristics include frequency, bandwidth, and modulation type.

[0039] Example: Retrieve the original electromagnetic signal whose signal type is interference. Extract the interference characteristics of this signal, including a frequency of 1805MHz, a bandwidth of 10MHz, and a modulation type of QPSK; The knowledge graph is obtained, which contains information on various devices that may cause interference and their corresponding interference characteristics. The interference characteristics of the original electromagnetic signal are matched with the knowledge graph to obtain several possible sources of interference: a nearby walkie-talkie, a wireless router and a microwave oven. The locations of these interference sources were analyzed. Using the location information of the base station and the approximate locations of the known interference sources, the distances between them and the data collection point were calculated. The distances were: walkie-talkie 500 meters from the data collection point, wireless router 200 meters, microwave oven 100 meters; a distance threshold of 300 meters was set. Since the distance of the walkie-talkie exceeded the set threshold, its corresponding interference source was eliminated.

[0040] S302. Obtain the working log of the device corresponding to the interference source and the acquisition time of the original electromagnetic signal; calculate the time difference between the running time and the acquisition time in the working log; when the time difference is greater than the corresponding difference threshold, the corresponding interference source is removed.

[0041] Example: The wireless router's work log shows its running time as [10:00-12:00], and the original electromagnetic signal acquisition time as 11:30; the microwave oven's work log shows its running time as [11:00-14:40]; since the acquisition time is included in both of their working times, both should be retained.

[0042] S303. Integrate and mark the remaining interference sources as a candidate interference source set. Construct the confidence analysis function: Calculate the confidence score of the candidate interference sources based on the confidence analysis function; select the interference source with the highest confidence score as the final interference source.

[0043] in, ; This represents the time sequence of the occurrence of interference signals; Represents the working time series of candidate interference sources; The distribution of the time difference sequence between the time of interference occurrence and the time of equipment operation, where Var represents the variance; , , A proportionality coefficient greater than 0; It is the Kullback-Leibler divergence, used to measure the difference in feature distributions; Characteristics of interference signals; Indicates candidate interference sources Features; For spectral sensitivity parameters; To estimate the direction of arrival of the interference signal, , indicating candidate interference sources The prediction has reached the desired direction; For electromagnetic wave propagation models, Candidate interference sources The actual location; This represents the antenna beamwidth.

[0044] Example: Mark the wireless router and microwave oven as candidate interference sources, and set... ; ; ; ; The divergence between the characteristics of the interference signal and those of the wireless router is 0.5; the divergence between the characteristics of the interference signal and those of the microwave oven is 1.2; the estimated direction of arrival of the interference signal is 45°. The location of the wireless router is predicted to have a direction of arrival of 50° using an electromagnetic wave propagation model; the location of the microwave oven is predicted to have a direction of arrival of 30°. The wireless router's operating time series and the interference signal's time series have three overlapping time points; the microwave oven's operating time series and the interference signal's time series have two overlapping time points. Therefore, the calculated... ; The confidence score of the wireless router was calculated to be 2.4776, and the confidence score of the microwave oven was 1.0205. Therefore, the wireless router was taken as the final source of interference.

[0045] S304. Obtain the component scores of system components; construct the vulnerability scoring function for system components: Construct a risk coefficient analysis function: ; Calculate the risk coefficient of the final interference source based on the risk coefficient analysis function; match the risk coefficient with the corresponding level mapping table to obtain the risk level of the final interference source.

[0046] The component rating includes: the component's inherent vulnerability rating and the component's robustness rating. as well as An adjustment coefficient greater than 0; This represents the inherent vulnerability of the j-th system component; This represents the component robustness score of the j-th system component; This represents the confidence score of the final interference source; This indicates the intensity of the final interference source's impact on system component j; This represents the vulnerability score of the j-th system component; the risk levels include: low, medium, and high.

[0047] Example: For the receiver component in a communication base station, its inherent vulnerability is 0.6, and its robustness score is 0.4; setting δ=0.6 and ε=0.4, the calculated vulnerability score of the component is 0.6; the final impact strength of the interference source on the receiver component is 0.8; the risk score calculated by the risk coefficient analysis function is 4.004, which is medium risk; (0,3] is low risk, (3,5] is medium risk, and (5,+∞) is high risk.

[0048] S305. Retrieve the risk level of the final interference source; identify the equipment model of the final interference source; combine the equipment model and risk level of the final interference source into a processing sequence; obtain the equipment processing database; match the processing sequence with the equipment processing database to obtain the corresponding suppression measures; send the suppression measures to the corresponding technical personnel to suppress the interference signal.

[0049] Example: Match the model and risk level of the final interference source wireless router with the corresponding device processing library to obtain the suppression measures: It is recommended to adjust the transmission frequency or power of the wireless router, or to add shielding measures; Send the suppression measures to the corresponding technical personnel, who will suppress the interference signal according to the recommendations.

[0050] Based on the above steps, interference sources are first initially screened according to distance thresholds to eliminate those too far from the acquisition point, reducing interference from irrelevant sources. Then, by comparing the operating time of the interference source device's log with the original electromagnetic signal acquisition time, interference sources with time differences exceeding a threshold are eliminated, further narrowing down the range of interference sources. This dual screening mechanism based on distance and time effectively eliminates interference sources with little correlation to the current interference signal, improving the accuracy of interference source location. A complex confidence analysis function is constructed, comprehensively considering factors such as differences in characteristic distribution, differences in direction of arrival, and time correlation. By calculating the confidence scores of candidate interference sources and selecting the highest score as the final interference source, candidate interference sources can be evaluated from multiple perspectives, ensuring that the finally determined interference source is the most likely source of the current interference signal, greatly improving the accuracy of interference source location. The constructed system component vulnerability scoring function comprehensively considers both the inherent vulnerability and robustness scores of the components, enabling a more accurate understanding of the vulnerability level of each system component. Based on the confidence score of the final interference source and the vulnerability assessment of the system components... This paper analyzes the impact of interference sources on system components and constructs a risk coefficient analysis function to calculate the risk coefficient of the final interference source. It considers multiple key factors such as the reliability of the interference source, the vulnerability of system components, and the range and intensity of the interference's impact, enabling a more scientific assessment of the risk posed by interference signals to the system. The risk coefficient is matched with a level mapping table to obtain a clear risk level. The final interference source's equipment model and risk level are combined into a processing sequence and matched with an equipment processing database to quickly and accurately obtain corresponding suppression and treatment measures. The equipment processing database stores standard treatment schemes for various equipment models under different risk levels. This matching method avoids technicians blindly searching when facing interference, saving time in developing treatment measures and improving processing efficiency.

[0051] Please see Figure 4The second aspect of this application provides an AI-based intelligent electromagnetic interference detection system, including: a type analysis module, and a signal acquisition module and a risk assessment module connected thereto; The signal acquisition module is used to acquire the raw electromagnetic signal; The type analysis module is used to preprocess the original electromagnetic signal to obtain multi-dimensional feature data; and to classify the original electromagnetic signal according to the multi-dimensional feature data to obtain the signal type. The risk assessment module is used to analyze the risk level of the corresponding interfering electromagnetic signal according to the signal type; and to suppress the interfering electromagnetic signal based on the risk level.

[0052] A third aspect of this application provides a computer-readable storage medium storing a computer program for performing the steps of the method described in the first aspect above.

[0053] Some of the data in the above formula are calculated by removing dimensions and taking their numerical values. The formula is the closest to the real situation obtained by software simulation of a large amount of collected data. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data.

[0054] The working principle of this application is as follows: This application acquires the original electromagnetic signal; preprocesses the original electromagnetic signal to obtain multi-dimensional feature data; classifies the original electromagnetic signal according to the multi-dimensional feature data to obtain signal types; analyzes the risk level of the corresponding interfering electromagnetic signal according to the signal type; and suppresses the interfering electromagnetic signal based on the risk level.

[0055] The above embodiments are only used to illustrate the technical methods of this application and are not intended to limit it. Although this application has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of this application without departing from the spirit and scope of the technical methods of this application.

Claims

1. An AI-based intelligent detection method for electromagnetic interference, characterized in that, include: Acquire the raw electromagnetic signal; preprocess the raw electromagnetic signal to obtain multidimensional feature data; The original electromagnetic signal is divided based on multidimensional feature data to obtain the signal type; Analyze the risk level of the corresponding interfering electromagnetic signal based on the signal type; Interfering electromagnetic signals are suppressed based on risk level.

2. The AI-based intelligent electromagnetic interference detection method according to claim 1, characterized in that, The preprocessing of the original electromagnetic signal to obtain multidimensional feature data includes: The original electromagnetic signal is retrieved; the original electromagnetic signal is amplified using a low-noise amplifier, and a tunable filter is used for preliminary frequency band selection. The original electromagnetic signal is sampled using a high-speed analog-to-digital converter to obtain the original time series; the original time series is then denoised, filtered, and normalized; and features are extracted from the processed original time series to obtain multidimensional feature data.

3. The AI-based intelligent electromagnetic interference detection method according to claim 1, characterized in that, The process involves extracting features from the processed original time series to obtain multidimensional feature data, including: Retrieve the processed original time series; calculate the statistical characteristics of a time window; the statistical characteristics include: mean, variance, kurtosis, skewness, pulse count rate, average pulse width, and duty cycle; The original time series was processed using Fast Fourier Transform to obtain frequency domain feature data, which includes: power spectral density, spectral centroid, spectral variance, and the frequency, amplitude, and bandwidth of the peak values. Integrate statistical features and frequency domain features into multidimensional feature data.

4. The AI-based intelligent electromagnetic interference detection method according to claim 1, characterized in that, The process of classifying the original electromagnetic signal based on multidimensional feature data to obtain signal types includes: Retrieve multidimensional feature data and a trained type analysis model; input the multidimensional feature data into the type analysis model to obtain the type label corresponding to the original electromagnetic signal; wherein, the type analysis model is built based on an artificial intelligence model; the type label is set to a positive integer; Obtain the interference classification database; match the type labels of the original electromagnetic signals with the interference classification database to obtain the signal types of the original electromagnetic signals; the signal types include: interference signals and non-interference signals.

5. The AI-based intelligent electromagnetic interference detection method according to claim 4, characterized in that, The type analysis model is built based on an artificial intelligence model and includes: Select model frameworks and deep learning algorithms from the artificial intelligence library; construct the model based on the deep learning algorithm and model framework to obtain the constructed model; Obtain the standard dataset; the standard dataset includes standard input data consistent with the content attributes of the multidimensional feature data, and standard output data consistent with the content attributes of the type label. The standard dataset is divided into a training set, a validation set, and a test set according to a preset ratio; a model is built using the training set for training; the internal parameters of the trained model are adjusted using the validation set; and the trained model is tested using the test set to obtain test metrics. Obtain the indicator threshold; compare the test indicator with the indicator threshold; if all test indicators are greater than the indicator threshold, mark the constructed model as a type analysis model; otherwise, reconstruct and train the type analysis model.

6. The AI-based intelligent electromagnetic interference detection method according to claim 1, characterized in that, The risk level analysis of the corresponding interference electromagnetic signal based on the signal type includes: Retrieve the original electromagnetic signal, which is an interference signal; extract the interference characteristics of the original electromagnetic signal; the interference characteristics include: frequency, bandwidth, and modulation type; Obtain a knowledge graph; match the interference characteristics of the original electromagnetic signal with the knowledge graph to obtain several interference sources; analyze the location of several interference sources and calculate the location distance between the location and the collection point; when the location distance is greater than the set distance threshold, the corresponding interference source is removed; otherwise, analyze the time correlation. Obtain the operating logs of the devices corresponding to the interference sources and the acquisition time of the original electromagnetic signals; calculate the time difference between the running time and the acquisition time in the operating logs; when the time difference is greater than the corresponding difference threshold, the corresponding interference source is removed; calculate and analyze the confidence level of the remaining interference sources and select the final interference source; calculate and analyze the risk level of the final interference source.

7. The AI-based intelligent electromagnetic interference detection method according to claim 6, characterized in that, The calculation and analysis of the confidence levels of the remaining interference sources, and the screening of the final interference sources, include: The remaining interference sources are integrated and marked as a candidate interference source set. ; Construct the confidence analysis function: Calculate the confidence score of the candidate interference sources based on the confidence analysis function; select the interference source with the highest confidence score as the final interference source. in, ; This represents the time sequence of the occurrence of interference signals; Represents the working time series of candidate interference sources; The distribution of the time difference sequence between the time of interference occurrence and the time of equipment operation, where Var represents the variance; , , A proportionality coefficient greater than 0; It is the Kullback-Leibler divergence, used to measure the difference in feature distributions; Characteristics of interference signals; Indicates candidate interference sources Features; For spectral sensitivity parameters; To estimate the direction of arrival of the interference signal, , indicating candidate interference sources The prediction has reached the desired direction; For electromagnetic wave propagation models, Candidate interference sources The actual location; This represents the antenna beamwidth.

8. The AI-based intelligent electromagnetic interference detection method according to claim 1, characterized in that, The component scores of the system components are obtained; wherein, the component scores include: component inherent vulnerability scores and component robustness scores; Construct a vulnerability scoring function for system components: ;in, as well as An adjustment coefficient greater than 0; This represents the inherent vulnerability of the j-th system component; This represents the component robustness score of the j-th system component; Construct a risk coefficient analysis function: ;in, This represents the confidence score of the final interference source; This indicates the intensity of the final interference source's impact on system component j; This represents the vulnerability score of the j-th system component; The risk coefficient of the final interference source is calculated based on the risk coefficient analysis function; the risk coefficient is matched with the corresponding level mapping table to obtain the risk level of the final interference source; the risk levels include: low, medium and high.

9. The AI-based intelligent electromagnetic interference detection method according to claim 1, characterized in that, The suppression processing of interfering electromagnetic signals based on risk level includes: Retrieve the risk level of the final interference source; identify the equipment model of the final interference source; combine the equipment model and risk level of the final interference source into a processing sequence; Obtain the equipment processing database; match the processing sequence with the equipment processing database to obtain the corresponding suppression measures; send the suppression measures to the corresponding technical personnel to suppress the interference signal.

10. An AI-based intelligent electromagnetic interference detection system, applied to the AI-based intelligent electromagnetic interference detection method according to any one of claims 1-9, characterized in that, include: The type analysis module, and the connected signal acquisition module and risk assessment module; The signal acquisition module is used to acquire the raw electromagnetic signal; The type analysis module is used to preprocess the original electromagnetic signal to obtain multidimensional feature data; The original electromagnetic signal is divided based on multidimensional feature data to obtain the signal type; The risk assessment module is used to analyze the risk level of the corresponding interfering electromagnetic signal based on the signal type. Interfering electromagnetic signals are suppressed based on risk level.

Images