Control system of high-voltage power transmission and distribution perimeter early warning device based on microwave radar and AI analysis

Abstract

The invention relates to the technical field of microwave radars, in particular to a control system of a microwave radar and AI analysis high-voltage power transmission and distribution perimeter early warning device, which comprises a data acquisition module, a data processing module, a behavior recognition alarm module and a cloud storage module, according to the control system, the early warning data and the operation parameters are synchronously acquired through the data acquisition module, the comprehensive perception capability is provided for the system, the false alarm rate and the missing report rate of traditional perimeter security and protection are remarkably reduced, the environmental adaptability and the decision intelligent level of the system under long-term operation are improved, and the safety of the system is improved. The fundamental transformation from passive alarm to active, intelligent and reliable early warning is realized.

Description

Technical Field

[0001] This invention relates to the field of microwave radar technology, and in particular to a control system for a high-voltage power transmission and distribution perimeter early warning device based on microwave radar and AI analysis. Background Technology

[0002] As a critical national infrastructure, the perimeter security of high-voltage power transmission and distribution facilities is of paramount importance. Traditional security methods, such as video surveillance or infrared beam scanning, are susceptible to interference from severe weather, changing lighting conditions, and complex environments, resulting in both high false alarm and false negative rates. While some existing intelligent solutions attempt to incorporate multi-sensor fusion and behavior recognition algorithms, they still face numerous challenges in practical applications: First, it is difficult to accurately align multi-source heterogeneous data in time and space, leading to poor information fusion results; second, the recognition model lacks adaptive adjustment and fault tolerance mechanisms when facing complex real-world conditions such as sensor performance degradation and data logic conflicts, making it difficult to guarantee system reliability.

[0003] For example, Chinese patent CN118818487A discloses a method and system for monitoring perimeter anomalies of high-voltage transmission lines based on millimeter-wave radar, belonging to the field of millimeter-wave radar technology. The method includes acquiring raw millimeter-wave radar signal data of the transmission line corridor scene; acquiring point cloud data of the observed target using the raw millimeter-wave radar signal data; separating the transmission line point cloud data from the non-transmission line point cloud data based on the actual observation scene; identifying the transmission line point cloud data; identifying the point cloud data of the transmission line anomaly target; and combining and analyzing the transmission line point cloud data and the point cloud data of the transmission line anomaly target to identify the anomaly target at the transmission line perimeter. Therefore, it is evident that this solution still suffers from low accuracy in identifying pedestrian behavior at the high-voltage transmission and distribution perimeter, difficulty in obtaining evidence, and low intelligence. Summary of the Invention

[0004] To address this, the present invention provides a control system for a high-voltage power transmission and distribution perimeter early warning device based on microwave radar and AI analysis, in order to overcome the problems of low recognition accuracy, difficulty in obtaining evidence, and low intelligence in the existing technology for pedestrian behavior identification at high-voltage power transmission and distribution perimeters.

[0005] To achieve the above objectives, the present invention provides a control system for a high-voltage power transmission and distribution perimeter early warning device based on microwave radar and AI analysis, comprising: The data acquisition module is used to collect early warning data and operating parameters; The data processing module is used to perform wavelet denoising on the early warning data according to the wavelet denoising method to obtain denoised early warning data; it is also used to perform time alignment on the denoised early warning data according to the time alignment method to obtain aligned early warning data; it is also used to perform spatial alignment on the aligned early warning data according to the spatial alignment method to obtain aligned early warning data; and it is also used to perform multimodal data fusion on the aligned early warning data according to the multimodal data processing method to obtain fused data. The behavior recognition alarm module is used to construct an LSTM model using the LSTM model construction method, analyze the confidence of the warning behavior based on the LSTM model and fused data, and issue a behavior alarm based on the confidence of the warning behavior to obtain alarm data. It is also used to adjust the behavior alarm process according to the sensor status, calculate the failure factor based on the operating parameters, and optimize the failure adjustment process based on the failure factor. The cloud storage module is used to store alarm data in the cloud.

[0006] Furthermore, in the data processing module, the wavelet denoising method specifically includes: selecting the DB4 wavelet basis function to perform N-level wavelet decomposition on the early warning data to obtain high-frequency detail coefficients and low-frequency approximation coefficients at different frequency resolutions; using an adaptive threshold function based on Stan unbiased risk estimation to perform threshold quantization on the high-frequency detail coefficients; and reconstructing the high-frequency detail coefficients and low-frequency approximation coefficients after threshold quantization by inverse wavelet transform to generate the denoised early warning data.

[0007] Furthermore, the time alignment processing method executed in the data processing module includes: A global reference clock deployed in the high-voltage power transmission and distribution perimeter early warning device network is designated as the time synchronization benchmark; the local timestamps carried by the noise-reduced early warning data from different acquisition nodes are identified; and a cubic spline interpolation algorithm is used to resample the noise-reduced early warning data with non-uniform local timestamps onto a unified time series with a fixed sampling interval based on the global reference clock.

[0008] Furthermore, the spatial alignment processing method executed in the data processing module includes: Define a three-dimensional Cartesian coordinate system covering the entire protection perimeter as a common reference system; A rigid body transformation matrix is ​​pre-set for each data acquisition node. This matrix encapsulates the translation and rotation parameters of the node relative to the common reference system. By applying the corresponding rigid body transformation matrix, the aligned early warning data of each node is mapped from its local sensor coordinate system to the common reference system, thereby completing the unification of spatial coordinates.

[0009] Furthermore, the multimodal data processing method executed in the data processing module is a multi-source information fusion method based on Dempster evidence theory, specifically including: constructing a basic probability allocation function for the aligned warning data of different modalities; calculating the confidence interval for each warning proposition under the same recognition framework based on the basic probability allocation function; and synthesizing evidence from the confidence intervals from multiple information sources according to Dempster's combination rule, and outputting the fused probability distribution as the fused data.

[0010] Furthermore, the LSTM model construction method includes: Step A1: Select a convolutional neural network model as the basic framework of the power distribution confidence model, and divide 70% of the historically acquired early warning behavior confidence learning set into a confidence training set, 15% of the power distribution confidence sample dataset into a confidence validation set, and 15% of the power distribution confidence sample dataset into a confidence test set. Step A2: Train the convolutional neural network model based on the confidence training set, update the weights of the convolutional neural network model through the backpropagation algorithm, and after each epoch, validate the convolutional neural network model using the confidence validation set to obtain the confidence validation loss value and the confidence validation accuracy. Step A3: When the confidence verification loss value and the confidence verification accuracy meet the preset verification conditions, the convolutional neural network model is tested according to the confidence test set to obtain the confidence test accuracy. When the confidence test accuracy reaches the preset accuracy, the convolutional neural network model is output as an LSTM model.

[0011] Furthermore, the behavior recognition alarm module inputs the fused data into the LSTM model, obtains the warning behavior confidence score Q output by the LSTM model, compares the warning behavior confidence score Q with the preset warning behavior confidence score Q0, and issues a behavior alarm based on the comparison result, wherein: When Q < Q0, the behavior recognition alarm module does not issue a behavior alarm; When Q≥Q0, the behavior recognition alarm module will issue a behavior alarm and a voice prompt, reminding personnel to stay away from the area covered by the microwave radar and AI-analyzed high-voltage power transmission and distribution perimeter early warning device.

[0012] Furthermore, the fault adjustment process of the behavior recognition alarm module includes: The health indicators of each sensor are monitored and evaluated in real time. These indicators include at least signal strength, data update rate, and self-test error codes. When the health indicators of a sensor are lower than a preset fault threshold, the confidence level of the preset warning behavior Q0 is adjusted to obtain the adjusted confidence level of the preset warning behavior Q0T. Q0T is set to Q0 × 0.9, and the value of the preset warning behavior confidence level Q0 is replaced with the value of the adjusted confidence level of the preset warning behavior Q0T.

[0013] Furthermore, the behavior recognition alarm module calculates the failure factor S based on the ambient temperature and humidity m1, cumulative equipment operating cycle m2, historical alarm frequency m3, first weighting coefficient w1, second weighting coefficient w2, and third weighting coefficient w3 in the operating parameters, setting S = m1 × w1 + m2 × w2 + m3 × w3. The failure factor S is compared with the preset failure factor S0, and the failure situation is judged based on the comparison result. Based on the judgment result, the fault adjustment process is optimized for failure, wherein: When S≤S0, the behavior recognition alarm module determines the failure condition as a minor failure and does not perform failure optimization in the fault adjustment process; When S > S0, the behavior recognition alarm module determines the failure condition as a serious failure, optimizes the fault adjustment process, cancels the adjustment of the preset warning behavior confidence level Q0, and optimizes the preset warning behavior confidence level Q0 to obtain the optimized preset warning behavior confidence level Q0Y. Q0Y is set to Q0 × 0.7, and the value of the preset warning behavior confidence level Q0 is replaced with the value of the optimized preset warning behavior confidence level Q0T.

[0014] Furthermore, when storing the alarm data in the cloud, the cloud storage module adopts the following data structure: a unique event identifier is generated for each alarm data; the data structure includes at least a high-precision timestamp field, a warning level code field, a geographic location information field, and an associated original data hash value field; the data is indexed using the above fields to achieve rapid retrieval and tracing based on time and event severity.

[0015] Compared with existing technologies, the beneficial effects of this invention are as follows: the system synchronously acquires early warning data and operating parameters through a data acquisition module, providing the system with comprehensive perception capabilities; the data processing module performs progressive processing on the data, from noise reduction, alignment, and fusion, generating high-quality, standardized fused data, laying a solid foundation for accurate identification; the behavior recognition alarm module not only utilizes an LSTM model to achieve high-confidence identification and alarm for complex temporal behaviors, but also, through its unique multi-layer optimization mechanism of "fault adjustment - failure optimization - logic correction," endows the system with adaptive and self-repair capabilities in the event of component failure or performance degradation. Simultaneously, this module can also reverse-calibrate the data processing process, forming an intelligent closed loop that continuously improves system performance; the system also uses a cloud storage module to efficiently manage and store alarm data, ensuring data traceability and analyzability. In summary, this system ultimately constitutes a full-link intelligent solution from perception, processing, decision-making to optimization and storage, significantly improving the accuracy, reliability, adaptability, and maintainability of high-voltage power transmission and distribution perimeter early warning. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the control system of the high-voltage power transmission and distribution perimeter early warning device based on microwave radar and AI analysis in this embodiment. Detailed Implementation

[0017] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0018] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0019] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0020] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0021] Please see Figure 1 As shown, this is a schematic diagram of the control system of the high-voltage power transmission and distribution perimeter early warning device based on microwave radar and AI analysis in this embodiment. The control system includes: The data acquisition module is used to collect early warning data and operating parameters; The data processing module is used to perform wavelet denoising on the early warning data according to the wavelet denoising method to obtain denoised early warning data; it is also used to perform time alignment on the denoised early warning data according to the time alignment method to obtain aligned early warning data; it is also used to perform spatial alignment on the aligned early warning data according to the spatial alignment method to obtain aligned early warning data; and it is also used to perform multimodal data fusion on the aligned early warning data according to the multimodal data processing method to obtain fused data. The behavior recognition alarm module is used to construct an LSTM model using the LSTM model construction method, analyze the confidence of the warning behavior based on the LSTM model and fused data, and issue a behavior alarm based on the confidence of the warning behavior to obtain alarm data. It is also used to adjust the behavior alarm process according to the sensor status, calculate the failure factor based on the operating parameters, and optimize the failure adjustment process based on the failure factor. The cloud storage module is used to store alarm data in the cloud.

[0022] Specifically, the control system of the microwave radar and AI-analyzed high-voltage power transmission and distribution perimeter early warning device is applied to the device. This control system synchronously acquires early warning data and operating parameters through a data acquisition module, providing comprehensive perception capabilities. The data processing module performs progressive processing on the data, from noise reduction, alignment, and fusion, generating high-quality, standardized fused data, laying a solid foundation for accurate identification. The behavior recognition alarm module not only utilizes an LSTM model to achieve high-confidence recognition and alarm for complex temporal behaviors, but also, through its unique "fault adjustment-failure optimization-logic correction" multi-layer optimization mechanism, endows the system with adaptive and self-repairing capabilities in the event of component failure or performance degradation. Simultaneously, this module can reverse-calibrate the data processing process, forming an intelligent closed loop that continuously improves system performance. The system also uses a cloud storage module to efficiently manage and store alarm data, ensuring data traceability and analyzability. In summary, this system ultimately constitutes a full-link intelligent solution from perception, processing, decision-making to optimization and storage, significantly improving the accuracy, reliability, adaptability and maintainability of high-voltage power transmission and distribution perimeter early warning.

[0023] Specifically, in the data processing module, the wavelet denoising method includes: selecting the DB4 wavelet basis function to perform N-level wavelet decomposition on the warning data to obtain high-frequency detail coefficients and low-frequency approximation coefficients at different frequency resolutions; using an adaptive threshold function based on Stan unbiased risk estimation to perform threshold quantization on the high-frequency detail coefficients; and reconstructing the high-frequency detail coefficients and low-frequency approximation coefficients after threshold quantization by inverse wavelet transform to generate the denoised warning data.

[0024] Specifically, the time alignment processing method executed in the data processing module includes: A global reference clock deployed in the high-voltage power transmission and distribution perimeter early warning device network is designated as the time synchronization benchmark; the local timestamps carried by the noise-reduced early warning data from different acquisition nodes are identified; and a cubic spline interpolation algorithm is used to resample the noise-reduced early warning data with non-uniform local timestamps onto a unified time series with a fixed sampling interval based on the global reference clock.

[0025] Specifically, the spatial alignment processing method executed in the data processing module includes: Define a three-dimensional Cartesian coordinate system covering the entire protection perimeter as a common reference system; A rigid body transformation matrix is ​​pre-set for each data acquisition node. This matrix encapsulates the translation and rotation parameters of the node relative to the common reference system. By applying the corresponding rigid body transformation matrix, the aligned early warning data of each node is mapped from its local sensor coordinate system to the common reference system, thereby completing the unification of spatial coordinates.

[0026] Specifically, the multimodal data processing method executed in the data processing module is a multi-source information fusion method based on Dempster evidence theory, which includes: constructing a basic probability allocation function for the aligned warning data of different modalities; calculating the confidence interval for each warning proposition under the same recognition framework based on the basic probability allocation function; and synthesizing evidence from the confidence intervals from multiple information sources according to Dempster's combination rule, and outputting the fused probability distribution as the fused data.

[0027] The topology modeling module constructs a distribution confidence model using a distribution confidence model construction method, which includes: Specifically, the LSTM model construction method includes: Step A1: Select a convolutional neural network model as the basic framework of the power distribution confidence model, and divide 70% of the historically acquired early warning behavior confidence learning set into a confidence training set, 15% of the power distribution confidence sample dataset into a confidence validation set, and 15% of the power distribution confidence sample dataset into a confidence test set. Step A2: Train the convolutional neural network model based on the confidence training set, update the weights of the convolutional neural network model through the backpropagation algorithm, and after each epoch, validate the convolutional neural network model using the confidence validation set to obtain the confidence validation loss value and the confidence validation accuracy. Step A3: When the confidence verification loss value and the confidence verification accuracy meet the preset verification conditions, the convolutional neural network model is tested according to the confidence test set to obtain the confidence test accuracy. When the confidence test accuracy reaches the preset accuracy, the convolutional neural network model is output as an LSTM model. Specifically, the behavior recognition alarm module inputs fused data into an LSTM model, obtains the warning behavior confidence level Q output by the LSTM model, compares the warning behavior confidence level Q with a preset warning behavior confidence level Q0, and issues a behavior alarm based on the comparison result, wherein: When Q < Q0, the behavior recognition alarm module does not issue a behavior alarm; When Q≥Q0, the behavior recognition alarm module will issue a behavior alarm and a voice prompt, reminding personnel to stay away from the area covered by the microwave radar and AI-analyzed high-voltage power transmission and distribution perimeter early warning device.

[0028] Specifically, the fault adjustment process of the behavior recognition alarm module includes: The health indicators of each sensor are monitored and evaluated in real time. These indicators include at least signal strength, data update rate, and self-test error codes. When the health indicators of a sensor are lower than a preset fault threshold, the confidence level of the preset warning behavior Q0 is adjusted to obtain the adjusted confidence level of the preset warning behavior Q0T. Q0T is set to Q0 × 0.9, and the value of the preset warning behavior confidence level Q0 is replaced with the value of the adjusted confidence level of the preset warning behavior Q0T.

[0029] Specifically, the behavior recognition alarm module calculates the failure factor S based on the ambient temperature and humidity m1, the cumulative operating cycle m2, the historical alarm frequency m3, the first weighting coefficient w1, the second weighting coefficient w2, and the third weighting coefficient w3 in the operating parameters, setting S = m1×w1 + m2×w2 + m3×w3. The failure factor S is compared with the preset failure factor S0, and the failure situation is judged based on the comparison result. Furthermore, the fault adjustment process is optimized based on the judgment result. When S≤S0, the behavior recognition alarm module determines the failure condition as a minor failure and does not perform failure optimization in the fault adjustment process; When S > S0, the behavior recognition alarm module determines the failure condition as a serious failure, optimizes the fault adjustment process, cancels the adjustment of the preset warning behavior confidence level Q0, and optimizes the preset warning behavior confidence level Q0 to obtain the optimized preset warning behavior confidence level Q0Y. Q0Y is set to Q0 × 0.7, and the value of the preset warning behavior confidence level Q0 is replaced with the value of the optimized preset warning behavior confidence level Q0T.

[0030] Specifically, when storing the alarm data in the cloud, the cloud storage module adopts the following data structure: a unique event identifier is generated for each alarm data; the data structure includes at least a high-precision timestamp field, a warning level code field, a geographic location information field, and an associated original data hash value field; the data is indexed using the above fields to achieve rapid retrieval and tracing based on time and event severity.

[0031] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A control system of a high-voltage transmission and distribution perimeter early warning device with microwave radar and AI analysis, characterized in that, The method comprises the following steps: a data acquisition module is used to collect early warning data and operating parameters; a data processing module is used to perform wavelet denoising on the early warning data according to a wavelet denoising method, to obtain denoised early warning data, to perform time alignment on the denoised early warning data according to a time alignment method, to obtain aligned early warning data, to perform spatial alignment on the aligned early warning data according to a spatial alignment method, to obtain aligned early warning data, and to perform multi-modal data fusion on the aligned early warning data according to a multi-modal data processing method, to obtain fusion data; a behavior recognition and alarm module is used to construct an LSTM model by an LSTM model construction method, to analyze early warning behavior confidence according to the LSTM model and the fusion data, to perform behavior alarm according to the early warning behavior confidence, to obtain alarm data, to perform fault adjustment on the behavior alarm process according to the sensor state, to calculate a failure factor according to the operating parameters, and to perform failure optimization on the fault adjustment process according to the failure factor; a cloud storage module is used to store the alarm data in the cloud.

2. The control system of the microwave radar and AI analysis high-voltage transmission and distribution perimeter early warning device according to claim 1, characterized in that, In the data processing module, the wavelet denoising method specifically includes: selecting a DB4 wavelet basis function to perform N-layer wavelet decomposition on the early warning data to obtain high-frequency detail coefficients and low-frequency approximation coefficients at different frequency resolutions; an adaptive threshold function based on Stein's unbiased risk estimate is used to perform threshold quantization processing on the high-frequency detail coefficients; and the high-frequency detail coefficients and the low-frequency approximation coefficients after threshold quantization processing are subjected to wavelet inverse transform reconstruction to generate the denoised early warning data.

3. The control system of the microwave radar and AI analysis-based high-voltage transmission and distribution perimeter early warning device according to claim 1, characterized in that, The time alignment method executed in the data processing module includes: a global reference clock deployed in the high-voltage power transmission and distribution perimeter early warning device network is specified as a time synchronization reference; local time stamps carried by denoised early warning data from different collection nodes are identified; and a cubic spline interpolation algorithm is used to resample the denoised early warning data with non-uniform local time stamps to a uniform time sequence with fixed sampling intervals based on the global reference clock as a reference.

4. The control system of the microwave radar and AI analysis-based high-voltage transmission and distribution perimeter early warning device according to claim 1, characterized in that, The spatial alignment method executed in the data processing module includes: a three-dimensional rectangular coordinate system covering the entire protection perimeter is defined as a common reference system; a rigid body transformation matrix is pre-set for each data collection node, which encapsulates the translation parameters and rotation parameters of the node relative to the common reference system; the aligned early warning data of each node is mapped from its local sensor coordinate system to the common reference system by applying the corresponding rigid body transformation matrix, and the spatial coordinates are unified.

5. The control system of the microwave radar and AI analysis-based high-voltage transmission and distribution perimeter early warning device according to claim 1, characterized in that, The multi-modal data processing method executed in the data processing module is a multi-source information fusion method based on D-S evidence theory, specifically including: constructing a basic probability assignment function for the aligned early warning data of different modalities; calculating the confidence interval for each early warning proposition under the same identification framework based on the basic probability assignment function; and synthesizing the confidence intervals from multiple information sources according to the Dempster combination rule to output the fused probability distribution as the fusion data.

6. The control system of the microwave radar and AI analysis-based high-voltage transmission and distribution perimeter warning device according to claim 1, characterized in that, The LSTM model construction method comprises: Step A1, selecting a convolutional neural network model as a basic framework of a power distribution confidence model, and dividing 70% of a historical acquired early warning behavior confidence learning set into a confidence training set, 15% of a power distribution confidence sample data set into a confidence validation set, and 15% of the power distribution confidence sample data set into a confidence test set; Step A2, training the convolutional neural network model according to the confidence training set, updating the weights of the convolutional neural network model through a back propagation algorithm, and verifying the convolutional neural network model using the confidence validation set after each epoch to obtain a confidence validation loss value and a confidence validation accuracy; Step A3, when the confidence validation loss value and the confidence validation accuracy meet a preset verification condition, testing the convolutional neural network model according to the confidence test set to obtain a confidence test accuracy, and outputting the convolutional neural network model as an LSTM model when the confidence test accuracy reaches a preset accuracy.

7. The control system of the microwave radar and AI analysis-based high-voltage transmission and distribution perimeter early warning device according to claim 6, characterized in that, The behavior recognition alarm module inputs the fusion data into the LSTM model, obtains the early warning behavior confidence Q output by the LSTM model, compares the early warning behavior confidence Q with a preset early warning behavior confidence Q0, and performs behavior alarm according to the comparison result, wherein: When Q < Q0, the behavior recognition alarm module does not perform behavior alarm; When Q ≥ Q0, the behavior recognition alarm module performs behavior alarm and issues a voice prompt to prompt personnel to move away from a place covered by the high-voltage power distribution perimeter early warning device of the microwave radar and AI analysis.

8. The control system of the microwave radar and AI analysis-based high-voltage transmission and distribution perimeter early warning device according to claim 1, characterized in that, The fault adjustment process of the behavior recognition alarm module comprises: Real-time monitoring and evaluation of the health degree index of each sensor, which at least includes signal strength, data update rate and self-check error code; when the health degree index of the sensor is lower than a preset fault threshold, adjusting the preset early warning behavior confidence Q0 to obtain an adjusted preset early warning behavior confidence Q0T, setting Q0T = Q0 × 0.9, and replacing the value of the preset early warning behavior confidence Q0 with the value of the adjusted preset early warning behavior confidence Q0T.

9. The control system of the microwave radar and AI analysis-based high-voltage transmission and distribution perimeter early warning device according to claim 8, characterized in that, The behavior recognition alarm module calculates the failure factor S according to the environmental temperature and humidity m1, the equipment cumulative running period m2, the historical alarm frequency m3, the first weight coefficient w1, the second weight coefficient w2 and the third weight coefficient w3 in the operating parameters, sets S = m1 × w1 + m2 × w2 + m3 × w3, compares the failure factor S with a preset failure factor S0, judges the failure condition according to the comparison result, and optimizes the failure optimization of the fault adjustment process according to the judgment result, wherein: When S ≤ S0, the behavior recognition alarm module determines that the failure condition is slight failure and does not optimize the failure optimization of the fault adjustment process; When S>S0, the behavior recognition alarm module determines that the failure condition is a serious failure, optimizes the failure for the fault adjustment process, cancels the adjustment of the preset early warning behavior confidence Q0, optimizes the preset early warning behavior confidence Q0, obtains the optimized preset early warning behavior confidence Q0Y, sets Q0Y=Q0×0.7, and replaces the value of the preset early warning behavior confidence Q0 with the value of the optimized preset early warning behavior confidence Q0T.

10. The control system of the microwave radar and AI analysis-based high-voltage transmission and distribution perimeter warning device according to claim 1, characterized in that, The cloud storage module adopts the following data structure when storing the alarm data in the cloud: a unique event identifier is generated for each piece of alarm data; the data structure at least includes a high-precision timestamp field, a warning level code field, a geographic location information field, and an associated original data hash value field; the data is indexed based on the above fields to realize fast retrieval and tracing based on time and event severity.

Images