Intrusion detection method based on low frequency and microwave composite induction

Abstract

The invention provides an intrusion detection method based on low-frequency and microwave composite induction, and relates to the technical field of intrusion detection.The method comprises the steps that the distribution distance of a vibration sensor is determined according to early-warning target type low-frequency signal intensity, installation is conducted, mode decomposition is conducted on low-frequency signals, and according to the mode energy ratio, the low-frequency signals are obtained; the method comprises the following steps of: identifying overground and underground modal components, forming a low-frequency feature set according to a vibration position of a received low-frequency signal and a low-frequency frequency domain feature, forming a microwave feature set according to a detection position of a microwave detection target and after decomposition and noise reduction are carried out on an echo signal, and establishing an intrusion target identification model; the early-warning target type is identified according to the low-frequency feature set and the microwave feature set, the modal energy ratio judgment threshold is adjusted according to the soil humidity and the surface accumulated water, the noise threshold is adjusted according to the rainfall and the air humidity, and the identification precision of the early-warning target type is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intrusion detection, and in particular to an intrusion detection method based on low-frequency and microwave composite induction. Background Art

[0002] In modern security and intrusion detection systems, traditional monitoring methods often rely on a single type of sensor, such as infrared sensors, ground vibration sensors, or microwave radar. However, these single sensors often have limitations. For example, infrared sensors are susceptible to interference from weather conditions, microwave radars can be affected by physical obstacles, and vibration sensors can respond differently to different soil types. These traditional methods often fail to provide sufficient accuracy and robustness in complex environments. Therefore, combining the strengths of multiple sensors to effectively detect intrusions, especially in the face of changing environmental factors such as soil type, moisture, and water accumulation, has become a pressing technical challenge.

[0003] Prior art publication CN 112687068 B discloses an intrusion detection method based on microwave and vibration sensor data. The main scheme includes steps 1: testing different intrusion behaviors to obtain different characteristic waveforms corresponding to different scenarios, representing different event types, as characteristic wave templates; step 2: obtaining the sensor signal output, counting the number of times the sensor voltage intensity exceeds a threshold per unit time, and obtaining a data characteristic wave reflecting the intensity change of the sensor output signal in the "number-time" relationship; step 3: matching the characteristic wave template with the data characteristic wave for similarity, and determining the event type based on the similarity; and step 4: outputting different alarm results based on the combination of event types. However, this scheme only detects intrusion behaviors based on changes in the voltage intensity of the sensor output and identifies the event type by matching the characteristic wave template, without considering the impact of environmental changes.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0005] The object of the present invention is to provide an intrusion detection method based on low-frequency and microwave composite induction to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions: An intrusion detection method based on low-frequency and microwave composite sensing, the specific steps include: Step 1: Obtain soil type, warning target type, and environmental data for the area to be alerted. Using the fixed variable method, obtain the low-frequency signal strength of the vibration source for each warning target type in each soil type under different environmental data. Determine the distribution distance of the vibration sensor based on the low-frequency signal strength and install it. Step 2: Obtain a real-time low-frequency signal, perform modal decomposition on the low-frequency signal using the modal decomposition method, identify the aboveground and underground modal components based on the modal energy ratio, reconstruct the aboveground and underground modal components, obtain the aboveground and underground low-frequency reconstructed signals, and perform time-frequency analysis on the reconstructed signals; Step 3: Determine the vibration position of the low-frequency signal source based on the time difference of receiving the low-frequency signal, and form a low-frequency feature set with the low-frequency frequency domain features. According to the detection position of the microwave detection target, the echo signal is decomposed and denoised to form a microwave feature set. Step 4: Establish an intrusion target recognition model. Use the low-frequency feature set and microwave feature set as the model input data, and the warning target type and coordinate position as the model output to train the model. Use the trained model based on real-time input data to identify the warning target type. Step 5: Adjust the modal energy ratio judgment threshold based on soil moisture and ground water to remove the offset effect of water accumulation on the low-frequency signal. Adjust the noise threshold based on rainfall and air humidity to remove the impact of the environment on the microwave echo characteristics.

[0007] Furthermore, the soil types include: hardened soil, sandy soil, loose soil; The types of warning targets include: ground intrusion: personnel, vehicles, animals; underground intrusion: drilling, cave-ins; The environmental data include soil moisture, soil humidity, water depth, rainfall and air humidity; The method for obtaining the low-frequency signals of the vibration sources of the warning target types in different environments and different soil types by the fixed variable method is as follows: At the vibration source and at a distance of A vibration sensor is installed at the position, and the low-frequency signal strength of the warning target type is detected by changing the data of one of the warning target type, environmental data and soil type in sequence, while the remaining two data types remain unchanged.

[0008] Furthermore, the method for determining the distribution distance of the vibration sensor based on the low-frequency signal strength is: By calculating the attenuation coefficient of the low-frequency signal strength of each set of data, the distribution distance is calculated based on the attenuation coefficient, the low-frequency signal of each warning target type and the detection threshold of the vibration sensor; ; ; in, is the vibration attenuation coefficient, The distance detected by the vibration sensor is the distance to the vibration source. The low-frequency signal strength at is the low-frequency signal intensity at the vibration source detected by the vibration sensor, is the distribution distance of the vibration sensor; The installation method is: install the sensors according to the distribution distance so that any position on the boundary of the intrusion detection area can be detected by at least three vibration sensors.

[0009] Furthermore, the specific steps of performing modal decomposition on the low-frequency signal by the modal decomposition method are: Set the number of modal components , bandwidth constraints , initial value of center frequency , for each modal component , =1,2,…, , and angular frequency Update, judge whether the low-frequency signal decomposition is completed through the convergence condition, and output the modal components; Iteratively update the modal components: ; in, For the The modal component in the frequency domain The frequency domain value of the iteration, is the number of iterations, is a low-frequency signal, is the angular frequency, for Modal classification in The center frequency of the iteration, For the The frequency domain value of the modal component in the frequency domain, is the frequency domain representation of the Lagrange multiplier, ; Update center frequency: ; in, for Modal classification in The center frequency of the iteration; Determine whether to end the update based on the convergence condition: ; in, is the discrimination accuracy.

[0010] Furthermore, the modal components above and below ground are identified based on the modal energy ratio. The modal energy ratio is calculated as follows: ; in, is the modal energy ratio, For the The frequency representation of the modal components is: is the frequency integral, are the division thresholds of the above-ground and underground modal component frequencies respectively; when When , it is determined that the modal component comes from the underground part: when When , it is determined that the modal component comes from the ground part: in, is the modal energy ratio judgment threshold.

[0011] Furthermore, the method of respectively recombining the above-ground and underground modal components to obtain the above-ground and underground low-frequency recombined signals is: ; ; in, , They are the recombined signals underground and above ground, is the number of modal components judged to be underground, is the number of modal components judged to be above ground, for The moment modal components.

[0012] Furthermore, the method for determining the vibration position of the low-frequency signal source based on the time difference of receiving the low-frequency signal is: According to the time difference when the vibration sensor receives the low-frequency signal, the vibration position of the low-frequency signal source is determined by the time difference and the propagation speed of the low-frequency signal: ; ; in, , , , They are vibration source, , , The coordinate position of each sensor, is the propagation speed of low-frequency signals, For the , The time difference between the two sensors detecting the low-frequency signal, For the , The time difference between the two sensors detecting the low-frequency signal.

[0013] Furthermore, after decomposing and denoising the echo signal, the specific steps of forming a microwave feature set are as follows: The detection position of the detection target is identified according to the echo time of the microwave detection target, and the echo signal is decomposed and denoised through wavelet transform, and the detail coefficient characteristics are obtained. The microwave feature set is formed according to the detection position and detail coefficient characteristics; Firstly, the echo signal is transformed by wavelet and decomposed into a series of approximate coefficients and detail coefficients of different frequencies; ; in, is the echo signal, is the approximate coefficient, is the detail coefficient, is the scaling function, is the wavelet function, is the acquisition time point of the communication signal, are scale parameter and displacement parameter respectively; By setting the noise threshold, the noise signal in the echo signal can be removed: ; in, is the detail coefficient after denoising, The noise threshold is set.

[0014] Furthermore, the environmental data includes soil moisture, water depth, rainfall and air humidity; The steps to adjust the modal energy ratio according to soil moisture and ground water are: Correct the dielectric constant of the soil based on soil moisture: ; in, is the corrected dielectric constant, To detect the basic dielectric constant of the soil in the area, is soil moisture, is the dielectric constant of moisture-saturated soil, is the dielectric constant of dry soil; Determine the water attenuation coefficient based on the accumulated water: ; in, is the water attenuation coefficient, is the basic ponding attenuation coefficient of the soil type, is the depth of water accumulation, is the center frequency of the low-frequency signal; The modal energy ratio judgment threshold is adaptively modified according to the corrected dielectric constant and water attenuation coefficient. The calculation formula is: ; in, is the corrected modal energy ratio judgment threshold, is the modal energy ratio judgment threshold; The specific steps for adjusting the noise threshold according to rainfall and air humidity are as follows: ; in, is the corrected noise threshold, To set the basic noise threshold, is the basic correction constant of the transmitted microwave frequency, is the rainfall, For air humidity.

[0015] Compared with the prior art, the present invention has the following beneficial effects: determining the distribution distance of vibration sensors according to the low-frequency signal strength of the warning target type and installing them, performing modal decomposition on the low-frequency signal, and identifying the modal components above and below the ground according to the modal energy ratio, forming a low-frequency feature set according to the vibration position of the received low-frequency signal and the low-frequency frequency domain features, forming a microwave feature set according to the detection position of the microwave detection target and after decomposing and denoising the echo signal, establishing an intrusion target recognition model, identifying the warning target type according to the low-frequency feature set and the microwave feature set, adjusting the modal energy ratio judgment threshold according to soil moisture and ground water, and adjusting the noise threshold according to rainfall and air humidity, thereby improving the recognition accuracy of the warning target type; This invention leverages the complementary nature of low-frequency vibration signals and microwave echo signals to provide more accurate intrusion detection in complex environments. First, modal decomposition and time-frequency analysis of low-frequency vibration signals are used to detect both above-ground and underground targets. Second, microwave echo signals are processed using wavelet transforms to effectively remove noise caused by environmental changes, further improving the accuracy and stability of microwave signals. This invention also incorporates the impact of environmental factors (such as soil moisture and water depth) on signal characteristics, employing a dynamic adjustment approach to further enhance the adaptability of low-frequency and microwave echo signals under varying conditions. By integrating multimodal features, this method enables higher-precision target recognition and intrusion detection, significantly improving the reliability and accuracy of intrusion detection, particularly in inclement weather and complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Schematic diagram of the overall method of the present invention. DETAILED DESCRIPTION

[0017] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to specific embodiments.

[0018] It should be noted that, unless otherwise defined, the technical or scientific terms used in the present invention should have the usual meanings understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative position relationships. When the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0019] Example: See also Figure 1 , the present invention provides a technical solution: An intrusion detection method based on low-frequency and microwave composite sensing, the specific steps include: Step 1: Obtain the soil type, warning target type and environmental data of the area to be alarmed. Using the fixed variable method, obtain the low-frequency signal strength of the vibration source of the warning target type in different soil types in different environmental data. Determine the distribution distance of the vibration sensor based on the low-frequency signal strength and install it.

[0020] In this embodiment, the soil types include: hardened soil, sandy soil, and loose soil; The types of warning targets include: ground intrusion: personnel, vehicles, animals; underground intrusion: drilling, cave-ins; The environmental data include soil moisture, soil humidity, water depth, rainfall and air humidity; The method for obtaining the low-frequency signals of the vibration sources of the warning target types in different environments and different soil types by the fixed variable method is as follows: At the vibration source and at a distance of A vibration sensor is installed at the position, and the low-frequency signal strength of the warning target type is detected by changing the data of one of the warning target type, environmental data and soil type in sequence, while the remaining two data types remain unchanged.

[0021] Environmental factors such as soil type, moisture, water depth, and rainfall significantly influence the propagation characteristics of vibration signals. For example, the vibration propagation characteristics of hardened soil differ from those of loose or sandy soil, potentially leading to changes in signal attenuation, propagation range, and sensitivity. Soil moisture and water depth also affect soil density and viscosity, thereby affecting the strength of low-frequency signals. Therefore, the fixed variable method eliminates interference from unstable environmental factors, allowing accurate calculation of the impact of each soil type and warning target on the low-frequency signal under different environmental conditions.

[0022] The fixed variable method allows for the precise determination of the low-frequency vibration signal strength caused by different warning target types (such as people, vehicles, animals, and drilling rigs) under known environmental conditions such as soil type, moisture, and water accumulation. This helps establish a stable baseline model for precise vibration sensor placement based on varying soil types and environmental conditions. Knowing the signal strength under each soil type and environmental factor allows for more rational sensor distribution, enabling the system to effectively detect target vibration signals in a variety of complex environments.

[0023] Soil type and environmental conditions significantly impact the performance and detection range of vibration sensors. Different soil types cause vibration signals to attenuate at different rates, resulting in varying sensor sensitivities at different locations. Therefore, the fixed variable method accurately calculates the signal strength of the vibration source under varying environmental conditions, thereby determining the optimal installation location and spacing for vibration sensors. This optimized sensor placement not only improves vibration signal acquisition but also enhances the overall system efficiency and accuracy, reducing unnecessary false positives and false negatives.

[0024] In this embodiment, the method for determining the distribution distance of the vibration sensor based on the low-frequency signal strength is: By calculating the attenuation coefficient of the low-frequency signal strength of each set of data, the distribution distance is calculated based on the attenuation coefficient, the low-frequency signal of each warning target type and the detection threshold of the vibration sensor; ; ; in, is the vibration attenuation coefficient, The distance detected by the vibration sensor is the distance to the vibration source. The low-frequency signal strength at is the low-frequency signal intensity at the vibration source detected by the vibration sensor, is the distribution distance of the vibration sensor; The installation method is: install the sensors according to the distribution distance so that any position on the boundary of the intrusion detection area can be detected by at least three vibration sensors.

[0025] Step 2: Obtain a real-time low-frequency signal, perform modal decomposition on the low-frequency signal using the modal decomposition method, identify the above-ground and underground modal components based on the modal energy ratio, reconstruct the above-ground and underground modal components separately, obtain the above-ground and underground low-frequency reconstructed signals, and perform time-frequency analysis on the reconstructed signals.

[0026] Modal decomposition is a signal processing method that extracts distinct features from a signal by breaking it down into modal components. Different modal components correspond to distinct vibration modes, each with a specific frequency, amplitude, and phase. Modal decomposition can separate vibration signals from different sources, both above and below ground, eliminating environmental noise and irrelevant signal components, and clarifying the useful information in low-frequency signals. This is because above-ground vibration signals (e.g., from pedestrians, vehicles, and mechanical equipment) are typically higher in frequency and faster in propagation, with the wave propagation path influenced by both the ground and air. Underground vibration signals are typically lower in frequency and primarily influenced by the soil, resulting in slower propagation and a different attenuation pattern than above-ground signals. Due to these physical differences, above-ground and underground signals behave differently in the frequency and time domains, with significant differences in frequency characteristics, energy distribution, and propagation patterns.

[0027] Modal decomposition can be used to separate signals into different modal components based on their frequency content. Signals from different sources have different energy distributions, and modal decomposition can effectively separate these signals, giving aboveground and underground signals distinct frequency domain characteristics, allowing them to be distinguished.

[0028] In this embodiment, the specific steps of performing modal decomposition on the low-frequency signal by the modal decomposition method are: Set the number of modal components , bandwidth constraints , initial value of center frequency , for each modal component , =1,2,…, , and angular frequency Update, judge whether the low-frequency signal decomposition is completed through the convergence condition, and output the modal components; Iteratively update the modal components: ; in, For the The modal component in the frequency domain The frequency domain value of the iteration, is the number of iterations, is a low-frequency signal, is the angular frequency, for Modal classification in The center frequency of the iteration, For the The frequency domain value of the modal component in the frequency domain, is the frequency domain representation of the Lagrange multiplier, ; Update center frequency: ; in, for Modal classification in The center frequency of the iteration; Determine whether to end the update based on the convergence condition: ; in, is the discrimination accuracy.

[0029] In this embodiment, the aboveground and underground modal components are identified based on the modal energy ratio. The modal energy ratio is calculated as follows: ; in, is the modal energy ratio, For the The frequency representation of the modal components is: is the frequency integral, are the division thresholds of the above-ground and underground modal component frequencies respectively; when When , it is determined that the modal component comes from the underground part: when When , it is determined that the modal component comes from the ground part: in, is the modal energy ratio judgment threshold.

[0030] By recombining the aboveground and underground modal components, we can obtain low-frequency signals associated with each intrusion target. This signal separation enables the system to more accurately identify different types of intrusion targets. Aboveground and underground vibration signals may overlap or interfere with each other in certain frequency bands. The recombining step effectively reduces interference between different signal sources, making the characteristics of each signal clearer and more independent, facilitating subsequent processing and identification.

[0031] Time-frequency analysis simultaneously analyzes signals in both time and frequency, typically using the short-time Fourier transform (SFT). Intrusion signals (such as those of people, vehicles, and drilling) are often non-stationary, meaning their frequency and amplitude vary over time. Time-frequency analysis captures the temporal spectral characteristics of signals, which is crucial for identifying and monitoring these non-stationary intrusion signals. It also provides the instantaneous frequency and temporal characteristics of signals, accurately reflecting frequency variations at different time points, further improving intrusion detection accuracy. For example, time-frequency analysis can clearly identify the start, duration, and end times of intrusion events, helping to make more accurate alarm decisions.

[0032] In this embodiment, the method of respectively recombining the above-ground and underground modal components to obtain the above-ground and underground low-frequency recombined signals is: ; ; in, , They are the recombined signals underground and above ground, is the number of modal components judged to be underground, is the number of modal components judged to be above ground, for The moment modal components.

[0033] The method for performing time-frequency analysis on the recombinant signal is: By sliding a window function on the recombined signal through short-time Fourier transform, the recombined signal is divided into shorter segments, and then Fourier transform is performed on each segment, so that the time-frequency characteristics of the recombined signal can be obtained simultaneously. The specific calculation formula is: ; in, For The time-frequency characteristics of the reconstructed signal at the moment, For the recombination signal, is the window function, The center position of the time window at time , is the frequency variable of the recombined signal, is the number of samples in the window, Is an imaginary unit.

[0034] Step 3: Determine the vibration position of the low-frequency signal source based on the time difference of receiving the low-frequency signal, and form a low-frequency feature set with the low-frequency frequency domain features. According to the detection position of the microwave detection target, the echo signal is decomposed and denoised to form a microwave feature set.

[0035] A significant advantage of time-difference positioning is its ability to accurately locate vibration sources. By using multiple sensors in concert, the position of the vibration source relative to these sensors can be calculated based on time differences. This is extremely helpful for accurately locating low-frequency signal sources, especially in complex environments (such as underground or obscured areas). It effectively eliminates interference and precisely identifies the vibration source. When low-frequency signals are received by different sensors, they may be distorted due to noise, interference, or different signal propagation paths. However, time-difference analysis can effectively filter out irrelevant noise sources. This is because genuine low-frequency signal sources exhibit consistent time-difference variations across multiple sensors, while noise or interference signals often lack this consistency. Therefore, time-difference-based positioning can enhance the system's noise suppression and improve the accuracy of low-frequency signal identification.

[0036] A plane coordinate system is established with the center of the intrusion detection area as the coordinate origin, the south direction as the positive direction of the vertical coordinate, the east direction as the positive direction of the horizontal coordinate, and vertically upward as the positive direction of the vertical coordinate.

[0037] Using a fixed, standardized coordinate system makes data processing more unified across the entire system. Whether it's multiple sensors, monitoring equipment, or other detection systems, as long as a unified coordinate origin and coordinate axis definition are used, all measurement results can be directly compared and combined, avoiding conversion issues between different coordinate systems. When using a fixed coordinate system and setting the directions of each axis, the target position can be described more concisely and standardized. For example, the position of a vibration source can be described by three coordinate values ​​(x, y, z), where x and y are directly related to the relative position of the target within the plane, and the z coordinate describes the displacement in the vertical direction. This approach can simplify the analysis and judgment during target positioning and improve the accuracy and operability of the positioning system.

[0038] In this embodiment, the method for determining the vibration position of the low-frequency signal source based on the time difference of receiving the low-frequency signal is: According to the time difference when the vibration sensor receives the low-frequency signal, the vibration position of the low-frequency signal source is determined by the time difference and the propagation speed of the low-frequency signal: ; ; in, , , , They are vibration source, , , The coordinate position of each sensor, is the propagation speed of low-frequency signals, For the , The time difference between the two sensors detecting the low-frequency signal, For the , The time difference between the two sensors detecting the low-frequency signal.

[0039] Microwave echo signals are often affected by various noise sources, such as environmental noise and interference from electronic devices. This noise can degrade signal quality and affect the accuracy of intrusion detection. Wavelet transforms can localize signals through multi-scale analysis, decomposing them into approximate and detail coefficients. By selecting appropriate wavelet functions and thresholding methods, wavelet transforms can decompose signals at different scales, suppressing high-frequency noise components while retaining useful signal information. This effectively separates noise components from the signal, resulting in efficient noise reduction.

[0040] The noise-reduced detail coefficients contain high-frequency information in the signal, reflecting the signal characteristics across different frequency ranges and the signal's changing characteristics. This is particularly important for capturing local changes in microwave signals. In microwave echo signals, detail coefficients are often closely correlated with changes in the target object's motion, structure, or morphology. Therefore, extracting these detail coefficients can yield valuable target information.

[0041] In this embodiment, the specific steps of forming a microwave feature set after decomposing and denoising the echo signal are as follows: The detection position of the detection target is identified according to the echo time of the microwave detection target, and the echo signal is decomposed and denoised through wavelet transform, and the detail coefficient characteristics are obtained. The microwave feature set is formed according to the detection position and detail coefficient characteristics; Firstly, the echo signal is transformed by wavelet and decomposed into a series of approximate coefficients and detail coefficients of different frequencies; ; in, is the echo signal, is the approximate coefficient, is the detail coefficient, is the scaling function, is the wavelet function, is the acquisition time point of the communication signal, are scale parameter and displacement parameter respectively; By setting the noise threshold, the noise signal in the echo signal can be removed: ; in, is the detail coefficient after denoising, The noise threshold is set.

[0042] Step 4: Establish an intrusion target recognition model. Use the low-frequency feature set and microwave feature set as the model input data, and the warning target type and coordinate position as the model output to train the model. Identify the warning target type through the trained model based on real-time input data.

[0043] BP neural networks have powerful nonlinear mapping capabilities. In intrusion target identification, signal characteristics are often complex and nonlinear. For example, the relationship between microwave and low-frequency signal characteristics and target type is not linear. BP neural networks, through their multi-layered neuron hierarchy, can learn the complex nonlinear mapping relationship between input data and warning target types.

[0044] At the same time, the nonlinear activation function in the hidden layer effectively captures the complex patterns between microwave echo signals and low-frequency features, thereby improving recognition accuracy. Furthermore, in intrusion detection, target changes are highly dynamic. Due to the excellent self-learning capabilities of BP neural networks, in actual intrusion target recognition systems, the model is updated as new data enters. Through incremental learning, BP neural networks continuously optimize and adjust their parameters as new data is added, improving the model's adaptability.

[0045] The established intrusion target recognition model is based on the BP neural network, which includes the input layer, hidden layer, and output layer: Input layer: ; in, Input data to the model, The amount of fire extinguishing agent used, is the fire assessment value, is the fire grade; Hidden layer: ; ; in, is the weighted input of the hidden layer, is the output of the hidden layer, is the weight matrix of the hidden layer, is the bias term of the hidden layer, is the activation function of the hidden layer; Output layer: ; ; in, is the weighted input of the output layer, is the output warning target type information, is the weight matrix of the output layer, is the bias term of the output layer, is the activation function of the output layer.

[0046] First, for ground intrusions, microwave and low-frequency feature sets are fused together within the intrusion target recognition model for identification. Microwave signals can sensitively capture the dynamic changes of ground targets, especially for fast-moving targets. The detail coefficient helps extract high-frequency information from the signal, reflecting the target's fine features. Combined with the detection position, microwave features can provide more accurate ground target identification, effectively distinguishing different types of ground targets, such as people and vehicles, especially in complex environments or under obstruction. The advantage of combining vibration position and low-frequency frequency domain features is that low-frequency signals can effectively describe the interaction between the target and the ground, especially the stability and persistence of ground targets. Low-frequency frequency domain features can reflect the long-term vibration patterns generated by the target, making them particularly effective for identifying stationary or slow-moving ground targets. Combining the low-frequency feature set with the microwave feature set provides more comprehensive information. By fusing and analyzing signals from different frequency bands, the accuracy and robustness of ground target identification are further improved, ensuring the stability and reliability of the early warning system in complex environments.

[0047] For underground intrusions, low-frequency signals can reflect the vibration or motion characteristics of underground targets, especially when the underground targets are not exposed to the ground. The low-frequency frequency domain characteristics can capture the weak vibrations caused by underground target activities or structural changes. These low-frequency characteristics are usually not affected by ground interference and can effectively distinguish underground targets from other background noise, ensuring that the early warning system can accurately monitor the activities of underground targets in complex environments. Another major advantage of low-frequency frequency domain characteristics in identifying underground targets is that they can identify the long-term stability of underground targets and their interaction with the surrounding environment. They are particularly suitable for areas with complex terrain or poor signal propagation conditions. By combining the analysis of vibration position and frequency domain characteristics, underground targets can be identified more efficiently and accurately, improving the performance and response speed of the early warning system.

[0048] Step 5: Adjust the modal energy ratio judgment threshold based on soil moisture and ground water to remove the offset effect of water accumulation on the low-frequency signal. Adjust the noise threshold based on rainfall and air humidity to remove the impact of the environment on the microwave echo characteristics.

[0049] Adjusting the modal energy ratio threshold based on soil moisture and surface water accumulation effectively eliminates the effects of water accumulation or wet soil on low-frequency signal offsets. When soil moisture is excessively high or water accumulation is deep, the presence of water can cause low-frequency signal attenuation or distortion, affecting the modal energy ratio determination and, consequently, the accuracy of detection results. Real-time monitoring and dynamic adjustment of these environmental factors ensures that the modal energy ratio more accurately reflects the vibration characteristics of underground targets, eliminating errors caused by water accumulation or wet soil and improving the reliability and accuracy of signal processing. Furthermore, changes in rainfall and humidity can affect the propagation environment of echo signals. In particular, in conditions of high humidity or heavy rainfall, the signal noise level may increase, hindering echo feature extraction and target recognition. By adjusting the noise threshold based on changes in rainfall and humidity, the interference of environmental noise on the echo signal can be dynamically removed, thereby enhancing the effectiveness of wavelet transform denoising, ensuring clearer target features in the microwave echo signal, reducing the impact of environmental factors, and improving signal processing precision and target recognition accuracy.

[0050] In this embodiment, the environmental data includes soil moisture, water depth, rainfall and air humidity; The steps to adjust the modal energy ratio according to soil moisture and ground water are: Correct the dielectric constant of the soil based on soil moisture: ; in, is the corrected dielectric constant, To detect the basic dielectric constant of the soil in the area, is soil moisture, is the dielectric constant of moisture-saturated soil, is the dielectric constant of dry soil; Determine the water attenuation coefficient based on the accumulated water:

[0051] in, is the water attenuation coefficient, is the basic ponding attenuation coefficient of the soil type, is the depth of water accumulation, is the center frequency of the low-frequency signal; The modal energy ratio judgment threshold is adaptively modified according to the corrected dielectric constant and water attenuation coefficient. The calculation formula is: ; in, is the corrected modal energy ratio judgment threshold, is the modal energy ratio judgment threshold; The specific steps for adjusting the noise threshold according to rainfall and air humidity are as follows: ; in, is the corrected noise threshold, To set the basic noise threshold, is the basic correction constant of the transmitted microwave frequency, is the rainfall, For air humidity.

[0052] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0053] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software depends on the specific application and design constraints of the technical solution.

[0054] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0055] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. An intrusion detection method based on low-frequency and microwave composite induction, characterized in that: The specific steps include: Step 1: Obtain soil type, warning target type, and environmental data for the area to be alerted. Using the fixed variable method, obtain the low-frequency signal strength of the vibration source for each warning target type in each soil type under different environmental data. Determine the distribution distance of the vibration sensor based on the low-frequency signal strength and install it. Step 2: Obtain a real-time low-frequency signal, perform modal decomposition on the low-frequency signal using the modal decomposition method, identify the aboveground and underground modal components based on the modal energy ratio, reconstruct the aboveground and underground modal components, obtain the aboveground and underground low-frequency reconstructed signals, and perform time-frequency analysis on the reconstructed signals; Step 3: Determine the vibration position of the low-frequency signal source based on the time difference of receiving the low-frequency signal, and form a low-frequency feature set with the low-frequency frequency domain features. According to the detection position of the microwave detection target, the echo signal is decomposed and denoised to form a microwave feature set. Step 4: Establish an intrusion target recognition model. Use the low-frequency feature set and microwave feature set as the model input data, and the warning target type and coordinate position as the model output to train the model. Use the trained model based on real-time input data to identify the warning target type. Step 5: Adjust the modal energy ratio judgment threshold based on soil moisture and ground water to remove the offset effect of water accumulation on the low-frequency signal. Adjust the noise threshold based on rainfall and air humidity to remove the impact of the environment on the microwave echo characteristics.

2. The intrusion detection method based on low-frequency and microwave composite induction according to claim 1, characterized in that: The soil types include: hardened soil, sandy soil, loose soil; The types of warning targets include: ground intrusion: personnel, vehicles, animals; underground intrusion: drilling, cave-ins; The environmental data include soil moisture, soil humidity, water depth, rainfall and air humidity; The method for obtaining the low-frequency signals of the vibration sources of the warning target types in different environments and different soil types by the fixed variable method is as follows: At the vibration source and at a distance of A vibration sensor is installed at the position, and the low-frequency signal strength of the warning target type is detected by changing the data of one of the warning target type, environmental data and soil type in sequence, while the remaining two data types remain unchanged.

3. The intrusion detection method based on low-frequency and microwave composite induction according to claim 1, characterized in that: The method for determining the distribution distance of the vibration sensor based on the low-frequency signal strength is: By calculating the attenuation coefficient of the low-frequency signal strength of each set of data, the distribution distance is calculated based on the attenuation coefficient, the low-frequency signal of each warning target type and the detection threshold of the vibration sensor; ； ； in, is the vibration attenuation coefficient, The distance detected by the vibration sensor is the distance to the vibration source. The low-frequency signal strength at is the low-frequency signal intensity at the vibration source detected by the vibration sensor, is the distribution distance of the vibration sensor; The installation method is: install the sensors according to the distribution distance so that any position on the boundary of the intrusion detection area can be detected by at least three vibration sensors.

4. The intrusion detection method based on low-frequency and microwave composite induction according to claim 1 is characterized in that: The specific steps of performing modal decomposition on the low-frequency signal by the modal decomposition method are: Set the number of modal components , bandwidth constraints , initial value of center frequency , for each modal component , =1,2,…, , and the angular frequency 𝜔 is updated, and the convergence condition is used to determine whether the low-frequency signal decomposition is completed, and the modal components are output; Iteratively update the modal components: ； in, For the The modal component in the frequency domain The frequency domain value of the iteration, is the number of iterations, is a low-frequency signal, is the angular frequency, for Modal classification in The center frequency of the iteration, For the The frequency domain value of the modal component in the frequency domain, is the frequency domain representation of the Lagrange multiplier, ; Update center frequency: ； in, for Modal classification in The center frequency of the iteration; Determine whether to end the update based on the convergence condition: ； in, is the discrimination accuracy.

5. The intrusion detection method based on low-frequency and microwave composite induction according to claim 1 is characterized in that: The modal components above and below ground are identified based on the modal energy ratio. The formula for calculating the modal energy ratio is: ； in, is the modal energy ratio, For the The frequency representation of the modal components is: is the frequency integral, are the division thresholds of the above-ground and underground modal component frequencies respectively; when When , it is determined that the modal component comes from the underground part: when When , it is determined that the modal component comes from the ground part: in, is the modal energy ratio judgment threshold.

6. The intrusion detection method based on low-frequency and microwave composite induction according to claim 1, characterized in that: The method of respectively recombining the above-ground and underground modal components to obtain the above-ground and underground low-frequency recombined signals is as follows: ； ； in, , They are the recombined signals underground and above ground, is the number of modal components judged to be underground, is the number of modal components judged to be above ground, for The moment modal components.

7. The intrusion detection method based on low-frequency and microwave composite induction according to claim 1, characterized in that: The method for determining the vibration position of the low-frequency signal source based on the time difference of receiving the low-frequency signal is as follows: According to the time difference when the vibration sensor receives the low-frequency signal, the vibration position of the low-frequency signal source is determined by the time difference and the propagation speed of the low-frequency signal: ； ； in, , , , They are vibration source, , , The coordinate position of each sensor, is the propagation speed of low-frequency signals, For the , The time difference between the two sensors detecting the low-frequency signal, For the , The time difference between the two sensors detecting the low-frequency signal.

8. The intrusion detection method based on low-frequency and microwave composite induction according to claim 1 is characterized in that: After decomposing and denoising the echo signal, the specific steps of forming a microwave feature set are as follows: The detection position of the detection target is identified according to the echo time of the microwave detection target, and the echo signal is decomposed and denoised through wavelet transform, and the detail coefficient characteristics are obtained. The microwave feature set is formed according to the detection position and detail coefficient characteristics; Firstly, the echo signal is transformed by wavelet and decomposed into a series of approximate coefficients and detail coefficients of different frequencies; ； in, is the echo signal, is the approximate coefficient, is the detail coefficient, is the scaling function, is the wavelet function, is the acquisition time point of the communication signal, are scale parameter and displacement parameter respectively; By setting the noise threshold, the noise signal in the echo signal can be removed: ； in, is the detail coefficient after denoising, The noise threshold is set.

9. The intrusion detection method based on low-frequency and microwave composite induction according to claim 1, characterized in that: The environmental data include soil moisture, water depth, rainfall and air humidity; The steps to adjust the modal energy ratio according to soil moisture and ground water are: Correct the dielectric constant of the soil based on soil moisture: ； in, is the corrected dielectric constant, To detect the basic dielectric constant of the soil in the area, is soil moisture, is the dielectric constant of moisture-saturated soil, is the dielectric constant of dry soil; Determine the water attenuation coefficient based on the accumulated water: ； in, is the water attenuation coefficient, is the basic ponding attenuation coefficient of the soil type, is the depth of water accumulation, is the center frequency of the low-frequency signal; The modal energy ratio judgment threshold is adaptively modified according to the corrected dielectric constant and water attenuation coefficient. The calculation formula is: ； in, is the corrected modal energy ratio judgment threshold, is the modal energy ratio judgment threshold; The specific steps for adjusting the noise threshold according to rainfall and air humidity are as follows: ； in, is the corrected noise threshold, To set the basic noise threshold, is the basic correction constant of the transmitted microwave frequency, is the rainfall, For air humidity.

Images