Characteristic marking method, system and device based on monitoring signal and medium

Abstract

The invention relates to the technical field of feature recognition, and discloses a feature marking method and system based on a monitoring signal, and the method comprises the steps: carrying out the collection of an electromagnetic signal of a monitoring target through a power line signal monitoring device; after the collected electromagnetic signals are isolated and filtered, analog-to-digital conversion is completed, and digital spectrum data containing time and amplitude are obtained and serve as monitoring signals; analyzing the monitoring signal to obtain a candidate area to be marked; and constructing a relationship of the monitoring signals by using a knowledge graph, and performing feature analysis on the candidate region to obtain a feature marking result. The method not only has a clear feature analysis and decision basis, but also enhances semantic expression and interpretability of a marking result by introducing feature template matching and probabilistic reasoning. The method has higher identification accuracy and intelligent degree, is especially suitable for leakage signal monitoring and spectrum analysis in a complex electromagnetic environment, and has good engineering suitability and practical application value.

Description

Technical Field

[0001] The present invention relates to the technical field of feature recognition, and in particular to a feature marking method, system, device and medium based on monitoring signals. Background Art

[0002] With the increasing popularity of electronic information devices, the problem of signal leakage caused by electromagnetic radiation has gradually become an important concern in the field of information security. Especially in scenarios involving sensitive data processing, the unexpected electromagnetic signals generated by various display devices, transmission interfaces and hardware circuits may constitute potential leakage channels. To effectively monitor and identify such leakage signals, spectrum analysis technology is widely used in the visual analysis of the electromagnetic environment and interference tracing. In traditional methods, spectrum data often presents in the form of static frequency distribution, lacking in-depth modeling of time-domain features and relying on manual experience for signal judgment, with limited efficiency and accuracy.

[0003] In addition, in the face of a complex and changing electromagnetic environment, how to identify the part with key information features from the vast monitoring signals has become the core difficulty restricting the improvement of the intelligent analysis level. Therefore, there is an urgent need to establish an automated analysis mechanism with signal semantic understanding ability to achieve accurate identification and marking of potential information regions in the time-frequency spectrum diagram to meet the technical requirements of high-level information protection. Summary of the Invention

[0004] In view of the above problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by the present invention are: the inaccurate marking of the existing feature marking method for monitoring signals, the lack of specific judgment basis, and how to enhance the interpretability of the marking, etc.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: A feature marking method based on monitoring signals, comprising:

[0007] Collect electromagnetic signals of a monitoring target through a power line signal monitoring device;

[0008] After isolating and filtering the collected electromagnetic signals, perform analog-to-digital conversion to obtain digital spectrum data including time and amplitude as the monitoring signal;

[0009] Analyze the monitoring signal to obtain candidate regions to be marked;

[0010] Use a knowledge graph to construct the relationship of the monitoring signal, perform feature analysis on the candidate regions, and obtain the result of feature marking;

[0011] The feature analysis includes setting each of the candidate regions as a two-dimensional quantum state, and determining the final state of the candidate region through analysis and verification.

[0012] As a preferred embodiment of the feature marking method based on monitoring signals according to the present invention, wherein: the power line signal monitoring device includes a device interface and a fuselage;

[0013] The fuselage internally integrates a coupling circuit to receive the transmission signal on the power line, and realizes the acquisition, storage, and analysis of the signal carried by the power network; an isolation module is integrated to isolate the instantaneous high-power signal in the power network, filter out the power frequency signal in the power grid, and remove the monitoring interference caused; by continuously monitoring the electromagnetic environment in the area range, an alarm is given for the signal marked as abnormal.

[0014] The protocol of the device interface supports security authentication, remote management, work order configuration, real-time data and status upload, alarm log and historical data query.

[0015] As a preferred embodiment of the feature marking method based on monitoring signals according to the present invention, wherein: the analog-to-digital conversion includes converting a continuous analog electromagnetic signal into a digital time-domain signal at the sampling frequency, performing a Fourier transform on the digital time-domain signal to obtain spectral data containing frequency and amplitude information, and at the same time, setting a timestamp at each sampling moment.

[0016] As a preferred embodiment of the feature marking method based on monitoring signals according to the present invention, wherein: the analysis of the monitoring signal includes dividing the entire spectrogram into multiple continuous regions as candidate regions for feature marking by performing region division on the two-dimensional time-frequency matrix formed by each spectral data of the monitoring signal and the timestamp of each spectral data at the sampling moment;

[0017] Among them, edge overlap is allowed between adjacent regions.

[0018] As a preferred embodiment of the feature marking method based on monitoring signals according to the present invention, wherein: the process of the region division includes: starting from an initial high-energy point based on the region growing algorithm, expanding to form a signal region with time and frequency continuity;

[0019] The region growing algorithm includes, in the digital spectrum, using a binary classification method to identify the part with local feature anomalies as the seed point; setting the seed node to grow forward and backward, and constructing a dynamic candidate region through the feature analysis;

[0020] During the growth process, when the feature markers of two seed points are the same, if the candidate regions of the two seed points are adjacent or overlapping, the candidate regions of the two adjacent seed points are merged and divided as the same candidate region.

[0021] As a preferred embodiment of the feature marking method based on monitoring signals according to the present invention, wherein: the knowledge graph includes presetting the content of each feature marker, taking each content as a node, and representing the relationship between every two nodes by a connecting line with an arrow, and each arrow represents the occurrence probability of an association relationship between two nodes;

[0022] Among them, A←B represents the probability that node B occurs before node A.

[0023] As a preferred embodiment of the feature marking method based on monitoring signals according to the present invention, wherein: the feature analysis further includes setting the dimensions of the quantum state as: the content of the feature marker and the growth width of the candidate region;

[0024] By determining the quantum state, the feature marker of each candidate region is obtained;

[0025] The process of determining the quantum state is specifically as follows:

[0026] Step 1: Randomly generate a growth width for the candidate region of each seed point;

[0027] The growth width of the candidate region is less than or equal to a preset width threshold; the seed point is located at the center position of the candidate region;

[0028] Step 2: Using the feature template, calculate the similarity between each candidate region and the feature template, and retain the feature markers corresponding to the feature templates whose similarity exceeds the preset value φ as the selectable feature markers for each candidate region; in each candidate region, randomly select from the selectable feature markers to obtain the feature marker of each candidate region;

[0029] The feature template includes calculating the similarity between every two segments of monitoring signals for the content of the same feature marker in historical data, and screening to obtain template C; when the similarity between template C and other monitoring signals all exceeds the preset value φ, the sum of the similarities between template C and other monitoring signals is the smallest;

[0030] Step 3: According to the knowledge graph, calculate the fitness of all seed points under the current feature marker, expressed as:

[0031] where n represents the index of the seed; N represents the number of seeds; M n represents the probability that the nth seed occurs before the (n + 1)th seed;

[0032] Step 4: Keep the width of each current candidate region growing, and in each candidate region, randomly iterate and replace the selectable feature markers to obtain new feature markers for each candidate region;

[0033] For any seed point, calculate the fitness difference between the new feature marker and the original feature marker. If the difference is less than N·μ, calculate the matching degree, and select the feature marker with the largest matching degree as the result of this iteration; if the difference is not less than N·μ, directly select the feature marker with the largest fitness as the result of this iteration; iterate in turn until the maximum number of iterations is reached or there are no selectable feature markers;

[0034] The matching degree includes the sum of the similarities between the feature markers of each seed point and the corresponding feature templates in the current candidate region;

[0035] Among them, μ represents the unit threshold of fitness, which is multiplied by the number of seeds to obtain the threshold of fitness;

[0036] Step 5: Retain the optimal selection in Step 4, iteratively replace the width of the candidate region growth in Step 1, and repeat Steps 1 to 4 until the maximum number of iterations is reached, or there are no selectable feature markers and no replaceable region widths, complete the iteration, and output the marking results for each candidate region and the width of the candidate region growth;

[0037] Step 6: Merge the candidate regions according to the width of the candidate region growth.

[0038] A feature marking system based on monitoring signals, applying a feature marking method based on monitoring signals as described in the present invention, includes:

[0039] An acquisition unit, which collects electromagnetic signals of a monitoring target through a power line signal monitoring device;

[0040] A conversion unit, which isolates and filters the collected electromagnetic signals, and then completes analog-to-digital conversion to obtain digital spectrum data including time and amplitude as monitoring signals;

[0041] An analysis unit, which analyzes the monitoring signals to obtain candidate regions to be marked;

[0042] A marking unit, which constructs the relationship of the monitoring signals by using a knowledge graph, analyzes the features of the candidate regions, and obtains the results of feature marking.

[0043] A computer device includes: a memory and a processor; the memory stores a computer program, wherein: when the processor executes the computer program, it implements the steps of the method described in any one of the present invention.

[0044] A computer-readable storage medium stores a computer program thereon, wherein: when the computer program is executed by a processor, the steps of any one of the methods described in the present invention are implemented.

[0045] Advantages of the present invention: The feature marking method based on monitoring signals provided by the present invention realizes accurate marking of potential information regions in monitoring signals by constructing a knowledge graph, dividing candidate regions, and introducing a quantum state modeling mechanism. This method not only has clear feature analysis and decision-making bases, but also enhances the semantic expression and interpretability of marking results by introducing feature template matching and probability reasoning. Compared with traditional methods that rely on manual experience or static thresholds, the present invention has higher recognition accuracy and intelligence, and is particularly suitable for leakage signal monitoring and mapping analysis in complex electromagnetic environments, and has good engineering adaptability and practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0047] Figure 1 It is the overall flowchart of a feature marking method based on monitoring signals provided by the first embodiment of the present invention;

[0048] Figure 2 It is the power line signal monitoring device diagram in a feature marking method based on monitoring signals provided by the second embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific embodiments of the present invention in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0050] Embodiment 1, referring to Figure 1 , is an embodiment of the present invention, which provides a feature marking method based on monitoring signals, including:

[0051] S1: Collect electromagnetic signals of a monitoring target through a power line signal monitoring device.

[0052] Further, the power line signal monitoring device includes a device interface and a fuselage.

[0053] The fuselage integrates a coupling circuit inside to receive the transmission signals on the power line, and realizes the acquisition, storage, and analysis of the signals carried by the power network; it integrates an isolation module to isolate the instantaneous high-power signals in the power network, filter out the power frequency signals in the power grid, and remove the monitoring interference caused; by continuously monitoring the electromagnetic environment in the area, it gives an alarm for the signals marked as abnormal.

[0054] The protocol of the device interface supports security authentication, remote management, work order configuration, real-time data and status upload, alarm log and historical data query.

[0055] In an alternative embodiment, the signal analysis process can be completed by a cloud platform. The power line signal monitoring device uploads the collected digital spectrum data to the cloud server through a network interface. The cloud platform performs feature marking analysis on the candidate area based on the knowledge graph, feature template, and quantum state modeling algorithm, and synchronously returns the marking result to the device side for display or further processing. This method can significantly improve the computing efficiency and achieve centralized intelligent recognition and multi-terminal shared analysis model.

[0056] S2: After isolating and filtering the collected electromagnetic signals, perform analog-to-digital conversion to obtain digital spectrum data containing time and amplitude as the monitoring signal.

[0057] The analog-to-digital conversion includes converting the continuous analog electromagnetic signal into a digital time-domain signal at the sampling frequency, performing Fourier transform on the digital time-domain signal to obtain spectrum data containing frequency and amplitude information, and setting a timestamp at each sampling moment.

[0058] Specifically, the continuous analog electromagnetic signal is a voltage signal that changes continuously with time, denoted as s(t), which cannot be directly processed by a computer. Use an analog-to-digital converter (ADC) to sample and quantize s(t) at fixed time intervals: convert the voltage value into a digital value with a limited number of bits (such as 12 bits, 16 bits). A discrete time series is obtained, which is called a digital time-domain signal.

[0059] Convert the discrete time series signal into a frequency-domain representation to analyze which frequency components it contains. Send the digital time-domain signal x[n] within a certain window (such as 1024 points) into the fast Fourier transform (FFT) algorithm; the output of the FFT is a complex frequency-domain sequence X[k], and each point represents a frequency component (complex amplitude + phase).

[0060] Extract the amplitude spectrum (or power spectrum), that is: A[k] = ∣X[k]∣.

[0061] Each A[k] corresponds to a frequency fk. A set of spectral data points is obtained, and these points form the "frequency - amplitude" pairs you need, that is, the spectral data we mentioned.

[0062] S3: Analyze the monitoring signal to obtain candidate regions to be marked.

[0063] Furthermore, analyzing the monitoring signal includes dividing the entire spectrogram into multiple continuous regions as candidate regions for feature marking by using a two - dimensional time - frequency matrix composed of each spectral data of the monitoring signal and the time stamp of each spectral data at the sampling moment. There is allowed to be edge overlap between adjacent regions (each data point may provide a basis for the judgment of the seed point). Dividing the two - dimensional time - frequency matrix composed of digitized spectral data and allowing edge overlap between adjacent regions aims to more accurately reflect the continuity and complexity of the monitoring signal in the two dimensions of time and frequency. Since electromagnetic signals in the real environment often exhibit characteristics such as fuzzy boundaries, overlapping frequency bands, or time - varying drifts, if forced hard - boundary division is carried out, information loss or misclassification may occur.

[0064] By allowing edge overlap, a data point can participate in the determination process of multiple candidate regions simultaneously, thus retaining its supporting value for the judgment of multiple seed points and improving the flexibility and inclusiveness of region division. At the same time, this design provides more complete context information for subsequent feature marking, making region recognition no longer rely on isolated points but have structural relevance and multi - source support, thereby enhancing the stability and robustness of the marking result.

[0065] Based on the region - growing algorithm of signal amplitude threshold and neighborhood correlation, starting from the initial high - energy point, expand to form a signal region with time and frequency continuity;

[0066] The region - growing algorithm includes, in the digitized spectrum, using a binary classification method to identify the part with local feature anomalies as the seed point; assume that the seed node grows forward and backward, and through the feature analysis, construct a dynamic candidate region. During the growth process, when the feature marks of two seed points (which may not be adjacent) are the same, if the candidate regions of the two seed points are adjacent or there is overlap, then merge the candidate regions of the two adjacent seed points as the same candidate region for division. It should be noted that each data point can be the basis for the judgment of multiple nodes, so it can be divided into different candidate regions. Therefore, in this method, by superimposing multiple states on each data point, it can support the feature marking of multiple seed points.

[0067] In a preferred embodiment, a binary classification model based on a random forest algorithm is used to judge each time-frequency point in the spectrogram to identify whether it has the structural characteristics of a region-growing seed point. The model inputs include multi-dimensional features such as spectral amplitude values, time and frequency direction gradients, mean and variance of local regions, etc. The training process is based on a labeled data set for supervised learning. The random forest classifier has good robustness and interpretability and can adapt to the intelligent screening task of seed points under complex backgrounds.

[0068] S4: Use a knowledge graph to construct the relationship of the monitoring signal, perform feature analysis on the candidate region, and obtain the result of feature marking. The feature analysis includes setting each candidate region as a two-dimensional quantum state, and determining the final state of the candidate region through analysis and verification.

[0069] The knowledge graph includes presetting the content of each feature marking (which can include combinations of multiple "single marking contents"), taking each content as a node, representing the relationship between every two nodes through a connected line with an arrow, and each arrow represents the occurrence probability of an association relationship between two nodes. Among them, A←B represents the probability that node B occurs before node A.

[0070] By presetting the content of each feature marking as a node in the graph and using a directed connected line to represent the semantic or temporal association relationship between nodes, it aims to provide a reasoning basis based on prior knowledge for the feature marking of candidate regions, thereby avoiding only relying on the surface matching of spectral features. Considering that some feature events in the actual monitoring signal may appear in the form of combinations or sequences (for example, multiple signals are spatially or temporally correlated), this design allows each node in the graph to represent not only a single marking content but also a combined state of multiple markings, thereby enhancing the expression ability and coverage breadth of the graph.

[0071] In addition, by introducing an arrow to represent the "occurrence probability of the association relationship", for example, B→A represents "the probability that feature A appears subsequently when feature B appears", the graph has the ability to judge temporal evolution or semantic sequence. This design helps the candidate region to infer a more reasonable and more confident final marking result based on the context structure in the graph when there are multiple potential marking selections, enhancing the interpretability of the system and the consistency of the reasoning logic.

[0072] Furthermore, the feature analysis also includes setting the dimension of the quantum state as: the content of the feature marking and the width of the candidate region growth (the state of each dimension is a superposition of multiple states, and the state is confirmed through an iterative process). By determining the quantum state, the feature marking of each candidate region is obtained.

[0073] The process of determining the quantum state is specifically:

[0074] Step 1: Randomly generate a width for growing a candidate region for each seed point.

[0075] The width for growing the candidate region is less than or equal to a preset width threshold; the seed point is located at the center position of the candidate region.

[0076] Step 2: Use a feature template to calculate the similarity between each candidate region and the feature template, and retain the feature markers corresponding to the feature templates whose similarity exceeds a preset value φ as the selectable feature markers for each candidate region; in each candidate region, randomly select from the selectable feature markers to obtain the feature marker of each candidate region.

[0077] The feature template includes calculating the similarity between every two segments of monitoring signals for the content with the same feature marker in historical data, and screening to obtain template C; when the similarity between template C and other monitoring signals all exceeds a preset value the sum of the similarities between template C and other monitoring signals is minimized.

[0078] In a preferred embodiment, the similarity calculation method adopts a multi-scale similarity calculation method based on structural spectrum decomposition. Specifically, it includes: performing wavelet or Gaussian multi-scale decomposition on the candidate signal region, extracting the principal component structure vector; calculating the cosine similarity of the structure vector to measure the consistency of the signal contour; at the same time, extracting the energy mean and variance to evaluate the consistency of the regional amplitude characteristics. Finally, the structure similarity and the energy similarity are fused in a weighted manner as the matching score between the feature template and the candidate region, enhancing the discriminability and interpretability of the similarity calculation.

[0079] Step 3: According to the knowledge graph, calculate the fitness of all seed points under the current feature marker, expressed as:

[0080] where n represents the index of the seed; N represents the number of seeds; M n represents the probability that the nth seed occurs before the (n + 1)th seed.

[0081] Step 4: Keep the width of growing each current candidate region, and in each candidate region, randomly iteratively replace the selectable feature markers to obtain new feature markers for each candidate region (the newly selected markers are different from those previously selected).

[0082] For any seed point, calculate the fitness difference |S1 - S2| between the new feature marker and the original feature marker. If the difference is less than N·μ, calculate the matching degree, and select the feature marker with the maximum matching degree as the result of this iteration; if the difference is not less than N·μ, directly select the feature marker with the maximum fitness as the result of this iteration; iterate in turn until the maximum number of iterations is reached or there are no selectable feature markers. Among them, S1 represents the fitness of the original feature marker; S2 represents the fitness of the new feature marker.

[0083] The matching degree includes the sum of the similarities between the feature markers of each seed point and the corresponding feature templates in the current candidate region.

[0084]

[0085] Among them, μ represents the unit threshold of fitness, which is multiplied by the number of seeds to obtain the threshold of fitness; j represents the index of the seed, J represents the number of seeds, PP represents the matching degree of the current result, and D j represents the similarity of the j-th seed.

[0086] Step Five: Retain the optimal selection in Step Four, iteratively replace the width of the candidate region growth in Step One (the width is actually selected within a preset interval), and repeat Steps One to Four until the maximum number of iterations is reached, or there are no selectable feature markers and no replaceable region widths, complete the iteration, and output the marking results for each candidate region and the width of the candidate region growth.

[0087] Step Six: Merge the candidate regions according to the width of the candidate region growth.

[0088] In the actual monitoring process, monitoring signals often have characteristics such as complex morphology, obvious time-frequency drift, and non-linear amplitude change. Traditional similarity calculation methods based on simple distance or point-by-point comparison are easily affected by local noise interference and are difficult to accurately reflect the similarity of the overall structure and semantic pattern between two candidate regions. The multi-scale similarity calculation method based on structural spectral decomposition is designed to extract the core morphological features of candidate regions from a global perspective, obtain a stable spectral expression through methods such as principal component analysis, and combine multi-scale decomposition to eliminate local mutation interference and enhance structural robustness. At the same time, energy features (such as mean and variance) are introduced to evaluate the amplitude level of the signal, forming a dual expression of "shape + value", and then the final similarity score is output in a fused form. The purpose of this method design is to improve the accuracy and semantic consistency of template screening and matching, and provide a more reliable judgment basis for subsequent quantum state judgment and feature marking.

[0089] On the other hand, this embodiment also provides a feature marking system based on monitoring signals, which includes:

[0090] The acquisition unit collects electromagnetic signals of the monitoring target through a power line signal monitoring device.

[0091] The conversion unit isolates and filters the collected electromagnetic signals, then completes analog-to-digital conversion to obtain digital spectrum data including time and amplitude as the monitoring signal.

[0092] The analysis unit analyzes the digital spectrum data to obtain candidate regions to be marked.

[0093] The marking unit constructs the relationship of the monitoring signal using a knowledge graph, analyzes the characteristics of the candidate regions, and obtains the result of feature marking.

[0094] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs, etc., which can store program codes.

[0095] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or used in combination with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device.

[0096] More specific examples (a non-exhaustive list) of computer-readable media include the following: electrical connections (electronic devices) having one or more wirings, portable computer disk cartridges (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber devices, and portable compact disc read-only memory (CDROM). Additionally, the computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then stored in a computer memory.

[0097] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc.

[0098] Example 2, referring to Figure 2 This is an embodiment of the present invention, which provides a feature marking method based on monitoring signals. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0099] Through Figure 2 the D101 device, which includes a fuselage and an interface, data is collected.

[0100] To verify the effectiveness and superiority of the proposed feature marking method based on monitoring signals, a controlled experimental scenario is constructed: three types of typical information device signal sources are selected, namely, type A computing terminals, type B confidential processing modules, and type C embedded graphics output devices. The above devices are connected to the power supply system through standard power supply lines under typical operating conditions. The experiment is carried out in an electromagnetic isolation environment to exclude environmental radiation interference and ensure the effectiveness and accuracy of the collected signals. The signal collection uses the power line signal monitoring device D101 described in this embodiment. The device is connected to the power supply line through a fuselage coupling interface to achieve non-intrusive collection of electromagnetic leakage signals conducted on the device power line. The device internally integrates a coupling circuit, an isolation module, and an analog-to-digital conversion unit, supports high-resolution monitoring of signals in the frequency band from 1 MHz to 1 GHz, sets the sampling period to 5 ms, and the resolution bandwidth to 100 kHz.

[0101] During the experiment, a power line signal monitoring device will be used to connect to the power supply line of the target device through an interface; the collected analog electromagnetic signals will be sent to a high-precision analog-to-digital conversion module after isolation and filtering; the analog-to-digital converter samples the analog signals at a set sampling period (such as 5 ms) and converts them into a time-amplitude sequence; the sampled data is framed in the time domain, and a two-dimensional time-frequency matrix with timestamps is constructed through Fourier transform as the input for subsequent signal analysis.

[0102] A binary classification algorithm based on the spectral feature map is used to identify the high eigenvalue points in the time-frequency matrix as "seed points"; the region growing process is started at each seed point, and its constraint conditions include: the maximum frequency width is 15 frequency points; the time span does not exceed 80 ms; candidate regions with time-frequency continuity and feature correlation are obtained.

[0103] For each candidate region, its corresponding spectral shape feature vector is extracted; principal component analysis (PCA) is used for feature dimensionality reduction; similarity comparison is made with the historical feature template library constructed, and the similarity score is calculated; the feature template library is extracted from historical data through the multi-scale spectral shape decomposition + energy evaluation mechanism described in the present invention to ensure representativeness; the template screening criteria are that the similarity with other segments is ≥ 0.75 and the total distance is the smallest.

[0104] A feature marking knowledge graph is constructed, where each node represents a type of feature marking; directed edges are used to represent the semantic occurrence sequence and probability relationship between nodes (such as A←B represents the occurrence probability that B precedes A); semantic reasoning of the knowledge graph is performed on the matching results of each candidate region, and the final marking result and credibility score are calculated by integrating the template similarity and the knowledge graph path weight.

[0105] All signal analysis processes are executed on the local platform and the cloud platform respectively; the consistency of the marking results is compared to verify the migration robustness of the system deployed on different platforms.

[0106] Among them, traditional methods usually adopt a fixed threshold determination and static feature extraction method, and its specific process includes: setting a signal amplitude threshold (such as -80 dBm), and directly marking the spectrum points exceeding this threshold as "suspicious signals". Judging whether there is "abnormality" in the signal based on local mutation or energy density. Simple clustering of adjacent high-energy points is performed to form candidate regions. The final marking result mostly depends on manual experience adjustment.

[0107] In the experimental comparison, three types of typical signal sources (type A computing terminal, type B confidential processing module, type C graphics output device) are tested, and the number of candidate regions obtained by using the method of the present invention is reduced by 46%, 47% and 59% respectively compared with the traditional method, significantly reducing the redundant analysis cost of non-target regions.

[0108] In terms of the number of recognized valid signals, the method of the present invention respectively recognized 46, 53, and 41 valid feature regions, which is an increase of 39%, 47%, and 76% compared to the traditional method. This indicates that by introducing a high feature point-driven region growth mechanism, the present invention can more precisely focus on regions with abnormal time-frequency behavior and prominent feature forms, and cooperate with feature templates for structured comparison, enhancing the target consistency of recognition.

[0109] In terms of mislabeling and misjudgment control, there were 22, 25, and 21 mislabeled regions respectively under the traditional method, while the misjudgment of the method of the present invention was controlled within 1 to 3 in all tests, and the misjudgment rate was reduced by more than 86%, verifying the effectiveness of the knowledge graph reasoning mechanism in complex feature judgment.

[0110] Although the method of the present invention has a slightly higher computational time consumption than the traditional method due to the introduction of feature evolution modeling and graph reasoning, this increase is still within the acceptable range of engineering, and a significant improvement in judgment accuracy is obtained by trading computing resources, realizing the strategy optimization of "computing for accuracy", and reflecting strong engineering practicability and deployment value.

[0111] In summary, the present invention is significantly superior to the prior art in terms of the screening accuracy of candidate regions, effective marking ability, misjudgment control level, and semantic modeling ability, providing a technical path with high accuracy and high robustness for the intelligent recognition and marking of electromagnetic leakage behavior.

[0112] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A feature marking method based on monitoring signals, characterized in that, Including: Collect electromagnetic signals of the monitoring target through a power line signal monitoring device; After isolating and filtering the collected electromagnetic signals, perform analog-to-digital conversion to obtain digital spectrum data including time and amplitude as monitoring signals; Analyze the monitoring signals to obtain candidate regions to be marked; Use a knowledge graph to construct the relationship of the monitoring signals, perform feature analysis on the candidate regions, and obtain the results of feature marking; The feature analysis includes setting each candidate region as a two-dimensional quantum state, and determining the final state of the candidate region through analysis and verification.

2. The feature marking method based on a monitoring signal according to claim 1, wherein: The power line signal monitoring device includes a device interface and a fuselage; A coupling circuit is integrated inside the fuselage to receive transmission signals on the power line and realize the acquisition, storage, and analysis of signals carried by the power network; An isolation module is integrated to isolate instantaneous high-power signals in the power network, filter out power frequency signals in the power grid, and remove monitoring interference caused; continuously monitor the electromagnetic environment in the area range, and give an alarm for signals marked as abnormal; The protocol of the device interface supports security authentication, remote management, work order configuration, real-time data and status upload, alarm log and historical data query.

3. The feature marking method based on the monitoring signal according to claim 2, wherein: The analog-to-digital conversion includes converting a continuous analog electromagnetic signal into a digital time-domain signal at a sampling frequency, performing a Fourier transform on the digital time-domain signal to obtain spectrum data including frequency and amplitude information, and setting a timestamp at each sampling moment.

4. The feature marking method based on a monitoring signal according to claim 3, wherein: Analyzing the monitoring signals includes dividing regions through a two-dimensional time-frequency matrix composed of each spectrum data of the monitoring signals and the timestamps of each spectrum data at the sampling moment, and dividing the entire spectrum diagram into multiple continuous regions as candidate regions for feature marking; Among them, edge overlap is allowed between adjacent regions.

5. The feature marking method based on a monitoring signal according to claim 4, wherein: The process of the region division includes: starting from an initial high-energy point based on a region growing algorithm, expanding to form a signal region with time and frequency continuity; The region growing algorithm includes using a binary classification method in the digital spectrum to identify parts with local feature anomalies as seed points; setting the seed nodes to grow forward and backward, and constructing dynamic candidate regions through the feature analysis; During the growth process, when the feature markings of two seed points are the same, if the candidate regions of the two seed points are adjacent or overlapping, the candidate regions of the two adjacent seed points are merged and divided as the same candidate region.

6. The feature marking method based on a monitoring signal according to claim 5, wherein: The knowledge graph includes presetting the content of each feature marking, taking each content as a node, representing the relationship between every two nodes through a connecting line with an arrow, and each arrow represents the occurrence probability of an association relationship between two nodes; Among them, A←B represents the probability that node B occurs before node A.

7. The feature marking method based on a monitoring signal according to claim 6, characterized in that: The feature analysis also includes setting the dimensions of the quantum state as: the content of the feature marking and the growth width of the candidate region; Determine the feature marking of each candidate region by determining the quantum state; The process of determining the quantum state is specifically: Step 1: Randomly generate a growth width of a candidate region for each seed point; The width of the candidate region growth is less than or equal to a preset width threshold; the seed point is located at the center position of the candidate region; Step 2: Using the feature template, calculate the similarity between each candidate region and the feature template, and retain the feature markers corresponding to the feature templates whose similarity exceeds the preset value φ as the selectable feature markers for each candidate region; In each candidate region, randomly select from the selectable feature markers to obtain the feature marker of each candidate region; The feature template includes calculating the similarity between every two segments of monitoring signals for the content of the same feature marker in historical data, and screening to obtain template C; when the similarity between template C and other monitoring signals all exceeds the preset value φ, the sum of the similarities between template C and other monitoring signals is the smallest; Step 3: According to the knowledge graph, calculate the fitness of all seed points under the current feature marker, expressed as: where n represents the index of the seed; N represents the number of seeds; M n represents the probability that the nth seed occurs before the (n + 1)th seed; Step 4: Keep the width of the growth of each current candidate region, and in each candidate region, randomly iteratively replace the selectable feature markers to obtain new feature markers for each candidate region; For any seed point, calculate the fitness difference between the new feature marker and the original feature marker. If the difference is less than N·μ, calculate the matching degree, and select the feature marker with the largest matching degree as the result of this iteration; if the difference is not less than N·μ, directly select the feature marker with the largest fitness as the result of this iteration; iterate in turn until the maximum number of iterations is reached or there are no selectable feature markers; The matching degree includes the sum of the similarities between the feature markers of each seed point and the corresponding feature template in the current candidate region; Among them, μ represents the unit threshold of fitness, which is multiplied by the number of seeds to obtain the threshold of fitness; Step 5: Retain the optimal selection in Step 4, iteratively replace the width of the candidate region growth in Step 1, and repeat Steps 1 to 4 until the maximum number of iterations is reached, or there are no selectable feature markers and no replaceable region widths, complete the iteration, and output the marking result for each candidate region and the width of the candidate region growth; Step 6: Merge the candidate regions according to the width of the candidate region growth.

8. A feature marking system based on monitoring signals, applying a feature marking method based on monitoring signals as described in any one of claims 1-7, characterized in that, Including: An acquisition unit that acquires electromagnetic signals of the monitoring target through a power line signal monitoring device; A conversion unit that isolates and filters the acquired electromagnetic signals and then completes analog-to-digital conversion to obtain digital spectrum data including time and amplitude as monitoring signals; An analysis unit that analyzes the monitoring signals to obtain candidate regions to be marked; A marking unit that constructs the relationship of the monitoring signals using a knowledge graph, performs feature analysis on the candidate regions, and obtains the result of feature marking.

9. A computer device, comprising: A memory and a processor; the memory stores a computer program, and it is characterized in that: when the processor executes the computer program, it implements the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1-7.

Images