Power construction risk prediction method and system based on space curvature change

Abstract

The embodiment of the invention discloses a power construction risk prediction method and system based on spatial curvature change, and the method comprises the steps: obtaining electric field intensity time series data of each measurement point in a construction region, carrying out the multilayer wavelet packet decomposition of the electric field intensity time series data, and obtaining a plurality of space-time coupling electric field sub-band signals of different scales; extracting multi-scale disturbance characteristics from the electric field sub-band signals; constructing an electric field space distribution surface based on the electric field intensity time sequence data, and calculating discrete curvature characteristics; performing curvature sudden change position detection, curvature time change trend and curvature discontinuity detection on the electric field spatial distribution surface to obtain spatial disturbance curvature characteristics; fusing the multi-scale disturbance characteristics, the discrete curvature characteristics and the spatial disturbance curvature characteristics to obtain a comprehensive disturbance index; and when the comprehensive disturbance index reaches a preset risk condition, determining that abnormal disturbance exists, and generating early warning information. According to the invention, the risk identification precision in power grid construction is obviously improved.

Description

Technical Field

[0001] This application relates to the field of power safety technology, and in particular to a method and system for predicting power construction risks based on spatial curvature changes. Background Technology

[0002] Currently, proximity risk warnings during live-line work on high-voltage transmission lines or large-scale machinery construction primarily rely on deploying electric field sensors in the work area. These sensors monitor in real-time whether the electric field strength exceeds a fixed threshold or whether the rate of change is abnormal, triggering an alarm. This method is widely used in equipment such as power construction safety helmets, crane anti-collision systems, and personal electric field alarms. Representative technologies include threshold judgment based on a single sensor, judgment based on the average value of multiple sensors, and a simple time-domain difference method.

[0003] However, existing technologies still have the following problems that urgently need to be solved: relying solely on the amplitude or simple rate of change of electric field strength cannot effectively distinguish between two typical risk scenarios: rapid approach of a metal tool tip and slow approach of a large human body, resulting in a single warning strategy; the spatial propagation characteristics of disturbances among multiple sensors are not fully considered, and the ability to capture distributed, propagating electric field distortions (such as disturbance waves generated by hook swing) is insufficient; there is a lack of adaptive mechanisms for slowly changing environmental factors such as wind sway, temperature, humidity, and slight icing, resulting in severe background drift and a high false alarm rate during long-term operation; most existing methods extract features only in the time or spatial domain in a single dimension, failing to achieve multi-level deep fusion of time multi-scale and spatial curvature, resulting in the warning accuracy and reliability in complex construction scenarios still not fully meeting the safety requirements of voltage levels of 110 kV and above.

[0004] Therefore, this invention proposes a new method for predicting risks in power construction, which effectively solves the above-mentioned defects in the existing technology. Summary of the Invention

[0005] This application provides a method and system for predicting power construction risks based on spatial curvature changes, which significantly improves the accuracy of risk identification in power grid construction.

[0006] This application provides the following solution:

[0007] According to the first aspect, a method for predicting power construction risks based on spatial curvature variation is provided. The method includes: acquiring time-series electric field intensity data at various measuring points in the construction area; performing multi-level wavelet packet decomposition on the electric field intensity time-series data; the multi-level wavelet packet decomposition treats the electric field intensity time-series data from multiple measuring points as a multi-dimensional signal tensor; and recursively decomposes the tensor into high-frequency sub-bands and low-frequency sub-bands, thereby obtaining multiple spatiotemporally coupled electric field sub-band signals at different scales; and extracting multi-scale perturbation features from the electric field sub-band signals, the multi-scale perturbation features including energy change rate, energy distribution ratio, and transient energy... The system analyzes the relationship between pulse characteristics and energy redistribution; constructs a spatial distribution surface of the electric field based on the time-series data of the electric field intensity, and calculates discrete curvature characteristics, including Gaussian curvature and mean curvature; obtains spatial disturbance curvature characteristics by detecting curvature abrupt change locations, curvature time change trends, and curvature discontinuities on the spatial distribution surface of the electric field; fuses the multi-scale disturbance characteristics, the discrete curvature characteristics, and the spatial disturbance curvature characteristics to obtain a comprehensive disturbance index reflecting the overall stability of the electric field; when the comprehensive disturbance index reaches a preset risk condition, it is determined that there is an abnormal disturbance, and an early warning message is generated.

[0008] According to one achievable method in the embodiments of this application, the multi-layer wavelet packet decomposition treats the electric field intensity time-series data of multiple measurement points as a multi-dimensional signal tensor, and recursively decomposes the tensor into high-frequency sub-bands and low-frequency sub-bands, including: arranging the electric field intensity time-series data of all measurement points according to the actual spatial coordinates of the sensor in the construction area to form a four-dimensional signal tensor, whose dimensions are the number of sensor channels, the number of time sampling points, the spatial horizontal index, and the spatial vertical index, respectively; during each layer of the wavelet packet decomposition, the same wavelet packet filter bank is applied sequentially to the low-frequency sub-band and the high-frequency sub-band along the spatial horizontal and spatial vertical directions, respectively, to achieve a joint triple recursive decomposition of the time dimension and the two spatial dimensions, thereby obtaining multiple spatiotemporally coupled high-frequency sub-bands and low-frequency sub-bands at different time and space scales.

[0009] According to one achievable method in the embodiments of this application, the multi-layer wavelet packet decomposition method further includes, at each layer of recursive decomposition: for the current high-frequency sub-band to be decomposed, calculating the spatial synchronization correlation coefficient between all measurement point signals in the sub-band and the energy ratio between the sub-band and the adjacent low-frequency sub-band; when the spatial synchronization correlation coefficient is greater than a preset correlation threshold and the energy ratio between the sub-band and the adjacent low-frequency sub-band is greater than a preset energy ratio threshold, determining that the high-frequency sub-band contains obvious disturbance propagation waves, and performing the next layer of recursive decomposition on the high-frequency sub-band.

[0010] According to one achievable method in the embodiments of this application, when extracting multi-scale perturbation features, the method further includes: calculating the energy transfer direction and transfer intensity between adjacent scale sub-bands, and using them as multi-scale energy flow features characterizing the approach direction of the perturbation source, to constitute the multi-scale perturbation features.

[0011] According to one achievable method in an embodiment of this application, the curvature abrupt change location detection includes: calculating the difference between the Gaussian curvature of the current electric field spatial distribution surface and the Gaussian curvature of the pre-stored background model to obtain a curvature difference; applying a difference operation to the curvature difference along three spatial directions to obtain a curvature gradient amplitude field; finding local maxima points in the curvature gradient amplitude field, and determining the local maxima points that satisfy a preset amplitude threshold and are consistent with the curvature gradient direction of adjacent measuring points as curvature abrupt change locations; and using the coordinates of all curvature abrupt change locations and their corresponding curvature gradient amplitudes as one of the final spatial perturbation curvature features.

[0012] According to one achievable method in an embodiment of this application, the curvature discontinuity detection includes: triangulating the electric field spatial distribution surface at the current moment to generate interconnected spatial grid patches; for any two adjacent triangular grid patches sharing a common edge, calculating the normal vectors of the two adjacent triangular grid patches respectively, and obtaining the included angle between the two normal vectors; when the included angle is greater than a preset first angle threshold and the included angle maintains an increasing trend within at least three consecutive sampling periods, marking the corresponding common edge as a true curvature discontinuity edge, while common edges that only meet the first angle threshold but do not show an increasing trend are determined to be discontinuities caused by noise or slowly varying background and are removed; merging connected components of all true curvature discontinuities to form one or more closed or approximately closed discontinuous contours, and extracting the total length, closure degree, and geometric center position of each contour as the final spatial perturbation curvature feature.

[0013] According to one of the embodiments of this application, the method further includes: when the comprehensive disturbance index is continuously lower than the safety threshold for a preset duration, updating the electric field spatial distribution surface constructed from the electric field intensity time series data at the current moment and the discrete curvature features calculated based on the distribution surface to a new background model to replace the original background model.

[0014] According to the second aspect, a power construction risk prediction system based on spatial curvature variation is provided. The system includes: a multi-layer wavelet packet decomposition module, used to acquire time-series electric field intensity data at various measuring points in the construction area, and perform multi-layer wavelet packet decomposition on the electric field intensity time-series data. The multi-layer wavelet packet decomposition treats the electric field intensity time-series data from multiple measuring points as a multi-dimensional signal tensor, and recursively decomposes the tensor into high-frequency sub-bands and low-frequency sub-bands, thereby obtaining multiple spatiotemporally coupled electric field sub-band signals at different scales; and a multi-scale disturbance feature extraction module, used to extract multi-scale disturbance features from the electric field sub-band signals. The multi-scale disturbance features include energy change rate, energy distribution ratio, and transient energy pulse characteristics. The system includes a feature and energy redistribution relationship module; a spatial curvature feature calculation module, used to construct an electric field spatial distribution surface based on the electric field intensity time series data, and calculate discrete curvature features, including Gaussian curvature and mean curvature; and to obtain spatial perturbation curvature features by detecting curvature abrupt change locations, curvature time change trends, and curvature discontinuities on the electric field spatial distribution surface; a feature fusion module, used to fuse the multi-scale perturbation features, the discrete curvature features, and the spatial perturbation curvature features to obtain a comprehensive perturbation index reflecting the overall stability of the electric field; and a risk judgment and early warning module, used to determine that there is an abnormal perturbation when the comprehensive perturbation index reaches a preset risk condition, and to generate early warning information.

[0015] According to a third aspect, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in any one of the first aspects above.

[0016] According to the fourth aspect, an electronic device is provided, comprising:

[0017] One or more processors; and

[0018] A memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the method described in any one of the first aspects above.

[0019] According to the specific embodiments provided in this application, the following technical effects are disclosed:

[0020] This application achieves, for the first time, accurate capture of disturbance waves simultaneously across multiple temporal scales and spatial propagation paths by constructing multi-dimensional tensors from multi-sensor electric field time-series data and performing spatiotemporally coupled multi-layer wavelet packet decomposition. It combines multi-level spatial disturbance characteristics such as discrete Gaussian curvature, average curvature, abrupt change locations, temporal variation trends, and discontinuities on the electric field spatial distribution surface with multi-scale temporal domain characteristics to form a comprehensive disturbance index, which significantly improves the accuracy of risk identification in power grid construction.

[0021] Of course, any product implementing this application does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a system architecture diagram applicable to the embodiments of this application;

[0024] Figure 2 A flowchart of a power construction risk prediction method based on spatial curvature variation provided in an embodiment of this application;

[0025] Figure 3 A structural block diagram of a power construction risk prediction system based on spatial curvature variation provided in an embodiment of this application;

[0026] Figure 4 A schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0028] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0029] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0030] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0031] To facilitate understanding of this application, the system architecture on which this application is based will be described first. Figure 1 An exemplary system architecture that can be applied to embodiments of this application is shown, such as Figure 1 As shown, the system architecture may include: user equipment and a power construction risk prediction system based on spatial curvature changes located on the server side.

[0032] Users can input electric field intensity time-series data through user equipment, which then sends this data to the server-side power construction risk prediction system based on spatial curvature changes. The power construction risk prediction system based on spatial curvature changes can use the method provided in this embodiment to perform risk prediction on the electric field intensity time-series data and obtain early warning information. The server can then send the early warning information to the user terminal.

[0033] User devices can include, but are not limited to, smart mobile terminals, wearable devices, and PCs (Personal Computers). Smart mobile devices can include mobile phones, tablets, laptops, PDAs (Personal Digital Assistants), and connected cars. Wearable devices can include smartwatches, smart glasses, virtual reality devices, augmented reality devices, and mixed reality devices (i.e., devices that support both virtual and augmented reality).

[0034] The power construction risk prediction system based on spatial curvature changes can be configured as a standalone server, a server cluster, or a cloud server. A cloud server, also known as a cloud computing server or cloud host, is a hosting product within the cloud computing service system, designed to address the management difficulties and weak service scalability inherent in traditional physical hosts and Virtual Private Servers (VPS) services. Besides... Figure 1 In addition to the architecture shown, the power construction risk prediction system based on spatial curvature variation can also be set up on a computer terminal with strong computing power.

[0035] It should be understood that Figure 1The user equipment and the power construction risk prediction system based on spatial curvature variation shown are merely illustrative. Depending on implementation needs, any number of user equipment and the power construction risk prediction system based on spatial curvature variation can be included.

[0036] Figure 2 This is a flowchart illustrating a power construction risk prediction method based on spatial curvature variation, provided in an embodiment of this application. This method can be... Figure 1 The power construction risk prediction system based on spatial curvature variation in the system shown is executed. For example... Figure 2 As shown, the method may include the following steps:

[0037] Step 201: Obtain the electric field intensity time series data of each measuring point in the construction area, and perform multi-level wavelet packet decomposition on the electric field intensity time series data. The multi-level wavelet packet decomposition regards the electric field intensity time series data of multiple measuring points as a multi-dimensional signal tensor, and recursively decomposes the tensor into high-frequency sub-bands and low-frequency sub-bands, thereby obtaining multiple spatiotemporally coupled electric field sub-band signals of different scales.

[0038] Step 202: Extract multi-scale perturbation features from the electric field sub-band signal. The multi-scale perturbation features include energy change rate, energy distribution ratio, transient energy pulse features, and energy redistribution relationship.

[0039] Step 203: Construct an electric field spatial distribution surface based on the electric field intensity time series data, and calculate discrete curvature features, including Gaussian curvature and average curvature; and obtain spatial perturbation curvature features by detecting curvature abrupt change locations, curvature time change trends, and curvature discontinuities on the electric field spatial distribution surface.

[0040] Step 204: The multi-scale perturbation features, the discrete curvature features, and the spatial perturbation curvature features are fused to obtain a comprehensive perturbation index that reflects the overall stability of the electric field.

[0041] Step 205: When the comprehensive disturbance index reaches the preset risk condition, it is determined that there is an abnormal disturbance and an early warning message is generated.

[0042] As can be seen from the above process, this application, by constructing multi-dimensional tensors from multi-sensor electric field time-series data and performing spatiotemporally coupled multi-layer wavelet packet decomposition, has for the first time achieved accurate capture of disturbance waves at multiple temporal scales and spatial propagation paths. Combining the discrete Gaussian curvature, average curvature and its abrupt change locations, temporal variation trends, discontinuities, and other multi-level spatial disturbance characteristics on the electric field spatial distribution surface, and deeply integrating them with multi-scale temporal domain characteristics to form a comprehensive disturbance index, it significantly improves the accuracy of risk identification in power grid construction.

[0043] The following describes in detail each step of the above process and the effects that can be further produced, with reference to the embodiments. First, with reference to the embodiments, step 201, namely, "acquiring the electric field intensity time series data of each measuring point in the construction area, and performing multi-level wavelet packet decomposition on the electric field intensity time series data, wherein the multi-level wavelet packet decomposition regards the electric field intensity time series data of multiple measuring points as a multi-dimensional signal tensor, and recursively decomposes the tensor into high-frequency sub-bands and low-frequency sub-bands, thereby obtaining multiple spatiotemporally coupled electric field sub-band signals of different scales", will be described in detail.

[0044] First, multiple electric field sensors (typically 16 to 64) are deployed within the construction area. Each sensor outputs a scalar value of electric field intensity (or a triaxial modulus value) at the same sampling time. After all sensors continuously acquire data along the time axis, a multi-channel sequence that varies over time is naturally formed. Traditional methods often perform wavelet packet decomposition on each channel individually and then simply average or vote on the results. This invention, however, arranges the time-series data of all channels according to the actual spatial location of the sensors, organizing it into a multi-dimensional signal tensor. For example, when the sensors are arranged in an approximately rectangular grid, a four-dimensional tensor can be formed, with its four dimensions being the sensor channel number, the time sampling point number, the horizontal spatial grid number, and the vertical spatial grid number, respectively. In this way, the original data is upgraded from a bunch of isolated time-series curves into a holistic tensor carrying a complete spatiotemporal structure.

[0045] Next, recursive wavelet packet decomposition is directly applied to this multidimensional tensor. Unlike ordinary wavelet packet decomposition which only occurs along the time axis, the decomposition process of this invention performs recursive splitting simultaneously in both the time and spatial dimensions. At each level of decomposition, the system first applies low-pass and high-pass filters to the tensor along the time dimension to obtain low-frequency and high-frequency subbands; then, for each of the separated subbands, the same low-pass and high-pass filters are applied again along the horizontal and vertical spatial dimensions, respectively. After repeating this process several times, each subband ultimately possesses both a specific time scale (low-frequency or high-frequency) and a specific spatial scale (coarse or fine), truly achieving coupled temporal and spatial decomposition.

[0046] As an implementable approach, the multi-layer wavelet packet decomposition treats the time-series data of electric field intensity from multiple measurement points as a multi-dimensional signal tensor, and recursively decomposes the tensor into high-frequency and low-frequency subbands. This includes: arranging the time-series data of electric field intensity from all measurement points according to the actual spatial coordinates of the sensors within the construction area to form a four-dimensional signal tensor, whose dimensions are the number of sensor channels, the number of time sampling points, the spatial horizontal index, and the spatial vertical index; performing one-dimensional wavelet packet filtering on the four-dimensional signal tensor along the time dimension; and simultaneously applying the same wavelet packet filter bank to the low-frequency and high-frequency subbands along the spatial horizontal and spatial vertical directions, respectively, at each level of the wavelet packet decomposition, thereby achieving a joint triple recursive decomposition of the time dimension and the two spatial dimensions to obtain multiple spatiotemporally coupled high-frequency and low-frequency subbands at different time and space scales.

[0047] Specifically, assuming 32 electric field sensors are deployed on-site at a sampling frequency of 1000 Hz, and the system analyzes the data from the most recent second, the raw data consists of 32 time series of length 1000. Based on the actual installation coordinates of each sensor (e.g., ground coordinates or relative tower coordinates), these 32 channels are rearranged to their corresponding positions on the spatial grid. If the sensors are approximately arranged in a 4×8 rectangular array, the data is organized into a four-dimensional tensor: the first dimension is the number of sensor channels (32 here), the second dimension is the number of time sampling points (1000), the third dimension is the horizontal spatial grid index (8), and the fourth dimension is the vertical spatial grid index (4). In this way, the raw data transforms from 32 isolated curves into a "data cube" with a clear spatial topology, where any local disturbance leaves a traceable mark on the spatial grid.

[0048] Next, the decomposition strictly follows a fixed order of time, horizontal space, and vertical space, performing a complete "triple split" at each level. Specifically, the system first applies a selected wavelet packet filter bank (commonly the Daubechies or Symlets series) to the entire four-dimensional tensor along the time dimension, simultaneously decomposing the signal on the time axis into low-frequency and high-frequency subbands. At this point, two four-dimensional tensors are still obtained, but the time dimension is halved. Then, keeping the channel and spatial structure unchanged, the system applies the same filter bank along the horizontal space direction to each of the newly generated subbands (whether low-frequency or high-frequency), splitting the horizontal space dimension in half again. Next, the same filter bank is applied again along the vertical space direction to the four already split subbands, splitting the vertical space dimension in half as well. After one complete "triple split," the original tensor becomes eight subbands, each with a specific time scale (low-frequency or high-frequency), a specific horizontal space scale, and a specific vertical space scale.

[0049] When entering the next level of decomposition, the system uses all eight sub-bands generated in the previous level as a new starting point, repeating the same time, horizontal, and vertical triple splitting for each, recursively performing this process for 2 to 4 levels. Ultimately, this results in dozens to hundreds of spatiotemporally coupled sub-bands, each precisely corresponding to a combination of a specific time scale and a specific spatial scale. Crucially, because the decomposition process maintains the spatial topology unchanged, the propagation of electric field disturbances from near to far within the construction area will exhibit a clear "wavefront movement" trajectory in these sub-bands. For example, when a metal hook rapidly approaches a conductor, a sudden energy surge first occurs in the high-frequency sub-band of the sensor closest to the conductor, subsequently propagating to outer sensors at the speed of light. This spatiotemporal evolution characteristic is completely destroyed in traditional single-channel decomposition, but is fully preserved in the joint triple recursive decomposition of this invention.

[0050] Preferably, the present invention further includes the following steps in the recursive decomposition of each layer of multi-layer wavelet packet decomposition: for the high-frequency sub-band to be decomposed, calculating the spatial synchronization correlation coefficient between all measurement point signals in the sub-band and the energy ratio between the sub-band and the adjacent low-frequency sub-band; when the spatial synchronization correlation coefficient is greater than a preset correlation threshold and the energy ratio between the sub-band and the adjacent low-frequency sub-band is greater than a preset energy ratio threshold, determining that the high-frequency sub-band contains obvious disturbance propagation waves, and performing the next layer of recursive decomposition on the high-frequency sub-band.

[0051] The first criterion is the "spatial synchronization correlation coefficient." The system extracts signal segments from all measurement points (i.e., all sensor channels) within the current high-frequency sub-band, calculates the Pearson correlation coefficient or cross-correlation peak values ​​between them in pairs, and takes the average or median as the spatial synchronization correlation coefficient. If multiple sensors produce similar waveforms almost simultaneously, it indicates that a disturbance wave is sweeping across the entire construction area with a consistent phase, resulting in a high correlation coefficient; conversely, if only individual sensors exhibit noise or isolated spikes, the correlation coefficient will be low. This indicator is specifically used to capture the spatial consistency of disturbances.

[0052] The second criterion is the "energy ratio of the high-frequency subband to the adjacent low-frequency subband." When a real disturbance occurs, the high-frequency subband often suddenly accumulates a large amount of energy, while the low-frequency subband, which mainly reflects the background power frequency field, has relatively stable energy. Therefore, the energy ratio of high-frequency to low-frequency will increase significantly. If it is just random noise, the high-frequency energy may also increase, but it usually will not far exceed the low-frequency background. Only when a real tool or human body approaches rapidly will an energy burst far exceeding the background be generated in the high-frequency band. This indicator is used to capture the "frequency band abruptness" of the disturbance.

[0053] The system considers that there is indeed a "meaningful perturbation wave" in the current high-frequency subband, and it is worthwhile to continue to invest computing power for a more detailed next-level decomposition, only when both of these conditions are met simultaneously; if either condition is not met, the high-frequency subband is immediately sealed as a leaf node and no longer splits.

[0054] Through the above method, weak disturbances that originally manifest as transient pulses on a single sensor will appear as "wavefronts" propagating in a specific direction across multiple adjacent sub-bands in this invention. For example, when a crane's metal hook rapidly approaches a conductor, the resulting electric field disturbance will propagate from near to far in the form of a wave. This propagation characteristic cannot be captured at all by ordinary single-channel decomposition, while the spatiotemporal coupling sub-bands of this invention can completely preserve its propagation path and velocity information, thereby significantly improving the system's ability to identify distributed, propagating disturbances.

[0055] The following describes in detail step 202, namely, "extracting multi-scale perturbation features from the electric field sub-band signal, wherein the multi-scale perturbation features include energy change rate, energy distribution ratio, transient energy pulse features, and energy redistribution relationship," with reference to the embodiments.

[0056] After the aforementioned multi-layer wavelet packet decomposition, the system has obtained dozens to hundreds of high-frequency and low-frequency sub-bands at different time-space scales, recording the perturbation's behavior within a specific frequency range and spatial resolution. This invention further extracts four statistical quantities from each sub-band that best characterize the dangerous approach process, collectively forming a complete multi-scale perturbation feature map.

[0057] The first type is the energy change rate. It directly calculates the ratio of the energy increment of each subband between adjacent time windows to the time interval. When a real risk occurs, especially when the tip of a metal tool approaches rapidly, the energy of the high-frequency subband often rises exponentially within tens of milliseconds, and the energy change rate suddenly reaches a maximum value; while the energy rise of ordinary noise or slowly varying background is gradual. This feature is most sensitive to capturing "suddenness" and often serves as the first warning clue.

[0058] The second category is energy distribution ratio. It calculates the proportion of energy allocated to each sub-band at each scale and spatial direction within the total energy of all sub-bands at the current moment. Normally, low-frequency sub-bands account for the vast majority of energy, while high-frequency sub-bands account for a very small percentage. When a large area of ​​a human body approaches, the energy proportion of the mid-to-low-frequency sub-bands slowly but steadily increases. When a metal tip approaches rapidly, the proportion of the high-frequency, fine-scale sub-bands spikes dramatically. By observing abnormal shifts in the energy distribution ratio, the system can predict the physical size and approach velocity of the disturbance source in advance.

[0059] The third category is transient energy pulse characteristics. It specifically searches for short, sharp energy spikes in the high-frequency subband, including parameters such as pulse peak value, pulse width, and pulse rise slope. These spikes typically correspond to precursors of partial discharge when a tool tip momentarily penetrates a strong field or when a human finger points at a wire. Because of spatiotemporal coupling decomposition, these pulses are no longer isolated events but exhibit clear spatial propagation trajectories, thus reducing the probability of misjudgment.

[0060] The fourth category is energy redistribution, also known as cross-scale energy flow. It examines the direction and intensity of energy flow between adjacent scale sub-bands and between adjacent spatially oriented sub-bands. For example, when a disturbance source approaches from afar, energy first accumulates in the coarse-scale low-frequency sub-band, then rapidly "infuses" into the fine-scale high-frequency sub-band, forming a significant energy cascade effect from low to high frequency. Conversely, energy flowing back from high to low frequency often indicates that the disturbance is moving away. By quantifying the direction and speed of this cross-scale and cross-spatial energy transfer, the system can even predict the movement trend of the disturbance source before it enters a dangerous distance, gaining valuable early warning time.

[0061] These four types of features, which are independent yet complementary, together constitute the multi-scale perturbation feature vector. They are no longer simple statistics from a single sensor in traditional methods, but rather a fusion of global information from multiple temporal scales, multiple spatial directions, and propagation dynamics, giving the subsequent fusion with spatial curvature features extremely high discriminative power and robustness.

[0062] Preferably, when extracting multi-scale perturbation features, this application further calculates the energy transfer direction and transfer intensity between adjacent scale sub-bands, and uses them as multi-scale energy flow features characterizing the approach direction of the perturbation source, which are used to jointly constitute the multi-scale perturbation features.

[0063] After completing the spatiotemporal coupled wavelet packet decomposition, a hierarchical relationship naturally exists between adjacent scale subbands: the coarse-scale subband corresponds to large-scale, low-frequency background changes, while the fine-scale subband corresponds to local, high-frequency transient disturbances. When a real danger arrives, the disturbance energy does not appear out of thin air, but is transmitted layer by layer strictly according to physical laws. For example, when a metal hook or a human body approaches a wire from a distance, the energy first accumulates slowly in the coarsest-scale low-frequency subband, then rapidly pours down like a waterfall to the finer-scale high-frequency subband, eventually forming a sharp pulse in the highest-frequency subband. Conversely, if the disturbance source is moving away, the energy flow direction is completely reversed, first attenuating from the fine-scale high-frequency subband, and then gradually flowing back to the coarse-scale low-frequency subband.

[0064] This invention capitalizes on the physical essence of this unidirectional waterfall-like propagation, calculating the energy transfer intensity and direction between each pair of adjacent scale sub-bands. Specifically, the difference between the energy of the fine-scale sub-band at the current moment and the energy of the coarse-scale sub-band at the previous moment is taken as the transfer intensity. If the difference is positive and continuously increasing, it is marked as a positive energy flow "from coarse to fine," indicating that the disturbance source is approaching. If the difference is negative, it is marked as a reverse energy flow, indicating that the disturbance source is moving away. Simultaneously, the system also statistically analyzes the components of this energy flow in the horizontal and vertical directions, thereby directly obtaining the approaching direction of the disturbance source (e.g., "rapidly approaching from the lower left" or "slowly approaching from directly above").

[0065] The addition of this multi-scale energy flow characteristic enables the system to have predictive capabilities. Even before the disturbance source enters the traditional safe distance threshold, the energy flow has already begun to flow unidirectionally from the coarse scale to the fine scale, allowing the system to issue an early warning 1 to 2 seconds in advance, giving workers valuable time to evacuate or correct their actions.

[0066] The following describes in detail step 203, namely, "constructing an electric field spatial distribution surface based on the electric field intensity time series data and calculating discrete curvature features, the discrete curvature features including Gaussian curvature and average curvature; and obtaining spatial perturbation curvature features by detecting curvature abrupt change locations, curvature time change trends and curvature discontinuities on the electric field spatial distribution surface".

[0067] The system takes the electric field intensity values ​​of all sensors every tens of milliseconds, which are like scattered "height points" in space. This invention first uses methods such as radial basis function interpolation or Kriging interpolation to weave these discrete points into a smooth continuous surface—the electric field spatial distribution surface.

[0068] Mathematically, the degree of local curvature of a surface is entirely determined by two fundamental quantities: Gaussian curvature and mean curvature. Gaussian curvature is the product of the two principal curvatures, reflecting whether the surface "bends inward" like a sphere or "one positive and one negative" like a saddle surface; mean curvature is the average of the two principal curvatures, reflecting the intensity of the overall bulge or depression of the surface. This invention calculates these two discrete curvature values ​​at each grid point of the surface, forming two spatially varying "curvature maps." Under normal power frequency electric fields, these two maps change extremely smoothly; when danger approaches, the maps will exhibit extreme values ​​or violent oscillations near the nearest point of the disturbance source.

[0069] Knowing only the curvature value is not enough. This invention further performs three depth detections on the curvature map to form spatial perturbation curvature features:

[0070] The first step is curvature abrupt change location detection. The system subtracts the current Gaussian curvature from the background curvature to obtain the difference surface. It then calculates the spatial gradient of this difference surface and identifies the points with the largest gradients. These points often correspond precisely to the tip of a metal tool or the closest point of a human finger, because the tip causes an extreme concentration of electric field lines, resulting in a rapid increase in curvature within a very small range. By adding gradient direction consistency constraints, false abrupt changes caused by noise can be eliminated, achieving sub-meter level precision positioning.

[0071] As an implementable method, the curvature abrupt change location detection includes: calculating the difference between the Gaussian curvature of the current electric field spatial distribution surface and the Gaussian curvature of the pre-stored background model to obtain the curvature difference; applying difference operations to the curvature difference along three spatial directions to obtain the curvature gradient amplitude field; finding local maxima points in the curvature gradient amplitude field, and determining the local maxima points that satisfy a preset amplitude threshold and are consistent with the curvature gradient direction of adjacent measuring points as curvature abrupt change locations; and using the coordinates of all curvature abrupt change locations and their corresponding curvature gradient amplitudes as one of the final spatial perturbation curvature features.

[0072] Specifically, the first step is background subtraction. The system has already saved a background Gaussian curvature map (i.e., the Gaussian curvature portion of the background model) under safe conditions. After recalculating the Gaussian curvature of the entire area at the current moment, it is immediately subtracted point-by-point from the background to obtain a pure curvature difference map that only reflects the "new distortion added by this disturbance." This completely eliminates inherent curvature interference caused by the line itself, tower structure, and terrain undulations; the remaining signal almost entirely comes from external intrusions. The second step is three-dimensional gradient calculation. The system performs subtraction operations on this curvature difference map along the horizontal X, horizontal Y, and vertical Z directions (commonly using Sobel or Scharr operators) to obtain the rate of curvature change in the three directions. Then, the square root of the sum of the squares of these three values ​​is taken to obtain a curvature gradient amplitude field. This amplitude field is like a "volcano thermal imaging map," where the closest point to the actual disturbance source forms an isolated and sharp "crater," with an amplitude much higher than the surrounding area. The third step is precise peak finding. The system performs local maxima search within the curvature gradient magnitude field, but goes beyond simple non-maximum suppression, incorporating a stricter directional consistency constraint: candidate peak points must simultaneously meet two conditions: first, their gradient magnitude exceeds a preset threshold; second, the angle between their gradient directions and those of their eight neighboring points is within 30 degrees. Only points meeting both conditions are ultimately identified as true curvature abrupt change locations. This constraint easily filters out diffuse gradient increases caused by wind swaying, sensor noise, or large, distant objects, leaving only extremely concentrated gradient bursts generated by sharp discharges or near-fingers. Finally, the system outputs the coordinates of all curvature abrupt change locations and their corresponding gradient magnitudes as an independent spatial perturbation curvature feature. These points are typically one or very few in number, and almost all fall on the sharpest, most dangerous parts of the intruding object.

[0073] The second indicator is the curvature temporal trend. The system stores the Gaussian curvature and mean curvature sequences of the most recent frames for each spatial grid point, and calculates their first and second time derivatives. If the curvature increases rapidly and monotonically within a short period, it indicates that the disturbance source is accelerating its approach; if it first increases and then decreases, it indicates that it has passed the nearest point and is moving away. This trend information allows the system to determine in advance whether the risk is "worsening" or "has been mitigated."

[0074] The third step is curvature discontinuity detection. The system first triangulates the electric field surface, forming numerous small triangular patches, and then checks the angle between the normal vectors of adjacent patches one by one. If the angle suddenly exceeds 35 degrees and continues to increase for several consecutive frames, it indicates the appearance of a true "edge" or "ridge" on the surface. This usually corresponds to the distorted boundary cut out by the human body contour or the edge of a crane in the electric field. The system connects these discontinuous edges into closed or nearly closed contours, extracts the contour length, degree of closure, and geometric center, and can directly determine whether it is a "slender and sharp metal tool" or a "wide and closed human body contour".

[0075] As an implementable method, the curvature discontinuity detection includes: triangulating the spatial distribution surface of the electric field at the current moment to generate interconnected spatial grid patches; for any two adjacent triangular grid patches sharing a common edge, calculating the normal vectors of the two adjacent triangular grid patches respectively, and obtaining the angle between the two normal vectors; when the angle is greater than a preset first angle threshold and the angle continues to increase over at least three consecutive sampling periods, the corresponding common edge is marked as a true curvature discontinuity edge, while common edges that only meet the angle threshold but do not show an increasing trend are determined to be discontinuities caused by noise or slowly varying background and are removed; all true curvature discontinuities are merged into connected components to form one or more closed or approximately closed discontinuous contours, and the total length, closure degree, and geometric center position of each contour are extracted as the final spatial perturbation curvature features.

[0076] Specifically, the first step involves Delaunay triangulation of the current electric field spatial distribution surface, dividing the entire surface into thousands of closely connected small triangular facets. After this processing, any sharp bending of the surface will manifest as a sudden increase in the tilt of certain adjacent triangular facets. The second step involves the system examining each pair of adjacent triangular facets sharing the same edge, calculating their external normal vectors, and then determining the angle between these two normal vectors. Normally, adjacent facets are almost parallel, with an angle close to 0 degrees; however, if an intruding object appears, the electric field lines are cut or compressed, resulting in noticeable creases between adjacent facets, and the angle suddenly jumps to 30 degrees or even over 60 degrees. The third step is the filtering process. The system does not assume that a large angle in a single frame necessarily indicates a real danger; instead, it requires that the angle simultaneously meet two conditions: first, it exceeds a preset first angle threshold (usually set to 30-40 degrees); second, the angle continuously increases for at least three consecutive sampling periods (approximately 100-300 milliseconds). Only common edges that simultaneously meet both conditions are marked as "true curvature discontinuities." This eliminates false edges caused by slowly varying or random factors such as wind-induced line swaying, temperature drift, and instantaneous sensor noise. In the fourth step, the system merges all marked true curvature discontinuities into connected components, much like connecting scattered ink smudges into a complete character, ultimately forming one or more closed or semi-closed contour lines. Then, three key indicators are extracted: total contour length, closure degree (i.e., the extent to which the contour forms a closed shape), and geometric center position. For example, when the contour is long and thin, open, and longer than 3 meters, the system immediately identifies it as a metal boom or hook tip; when the contour is wide, nearly closed, and has a closure degree greater than 0.8, the system immediately identifies it as a human intrusion.

[0077] The following describes in detail step 204, namely, "integrating multi-scale disturbance characteristics, discrete curvature characteristics and spatial disturbance curvature characteristics to obtain a comprehensive disturbance index that reflects the overall stability of the electric field," with reference to the embodiments.

[0078] This invention integrates two completely different types of physical dimensions of information: multi-scale perturbation features from the time axis and discrete curvature features and spatial perturbation curvature features from spatial surfaces, into a single, unique scalar perturbation index that comprehensively reflects the overall stability of the electric field, thereby achieving extremely high accuracy and extremely low false alarm rate.

[0079] In practical fusion, simple addition or averaging can be used, or a layered weighted approach with dynamic adjustment can be employed. Specific implementation methods vary. For example, a weighted summation method can be used: assigning a weight of 0.4 to multi-scale perturbation features (due to their high time sensitivity), a weight of 0.3 to discrete curvature features (basic spatial distortion), and a weight of 0.3 to spatial perturbation curvature features (advanced morphological recognition), then summing the elements to obtain the index value. Alternatively, an attention mechanism can be used, allowing the system to dynamically adjust weights based on the current scenario—automatically increasing the attention to multi-scale perturbation features when high-frequency energy flow is detected; prioritizing spatial perturbation curvature features when curvature contour closure is high. More advanced implementations can leverage shallow neural networks to concatenate all features into an input vector, learning the optimal fusion parameters through training to ensure a balanced sensitivity of the index to different risk scenarios.

[0080] The following describes in detail step 205, namely, "when the comprehensive disturbance index reaches the preset risk condition, it is determined that there is an abnormal disturbance and an early warning information is generated," with reference to the embodiments.

[0081] The comprehensive disturbance index is a normalized value between 0 and 1, normally close to 0, representing overall electric field stability and no external interference. Preset risk conditions are one or more threshold lines, such as a mild risk threshold of 0.5 and a severe risk threshold of 0.8. These thresholds are pre-set based on extensive field experimental data and can be dynamically adjusted according to voltage levels, weather conditions, or construction type. When the index value exceeds the threshold, the system immediately determines that an abnormal disturbance exists. Here, "abnormal disturbance" specifically refers to electric field distortion caused by the approach of tools, human bodies, or sudden environmental changes (such as lightning strikes or bird interference), rather than slowly varying background factors such as temperature drift.

[0082] The judgment process is fully automated, requiring no human intervention. The system monitors the indicator's change curve in real time, and once it exceeds a threshold, it triggers multi-level response logic: if it only slightly exceeds the threshold, it is judged as a potential abnormal disturbance, possibly corresponding to a distant object slowly approaching; if it significantly exceeds the threshold, it is judged as a highly abnormal disturbance, usually corresponding to a metal tip entering a high-voltage zone. The judgment is based not only on the current value, but also on the rate of change and duration of the indicator, avoiding misjudgments caused by instantaneous noise.

[0083] Generating early warning information is the direct output of the judgment, and it takes various forms and is progressively graded. For example, a mild warning can be issued through a voice announcement such as "Attention, electric field disturbance is increasing, please check your distance" and a flashing yellow light; a severe warning will escalate to a red alert, a blaring buzzer, or even automatically trigger a forced power cut-off to on-site equipment, a shutdown of the crane's hydraulic system, or a "evacuate immediately" command sent to the monitor's mobile phone via a wireless module. At the same time, the system records the entire process of the event, including the disturbance index curve, the location of curvature abrupt changes, and discontinuous contour images, which facilitates post-event analysis and threshold optimization.

[0084] Preferably, the present invention further includes: when the comprehensive disturbance index is continuously lower than the safety threshold for a preset duration, updating the electric field spatial distribution surface constructed from the electric field intensity time series data at the current moment, and the discrete curvature features calculated based on the distribution surface, to a new background model to replace the original background model.

[0085] Specifically, the system sets a preset duration, such as 60 seconds or 5 minutes, as an observation window. Within this window, if the comprehensive disturbance index remains below the safety threshold (usually set to 0.15 or lower, indicating a highly stable electric field with no significant intrusion), the system considers the current state to be the "new normal," suitable for updating the background model. At this point, the system immediately captures all time-series data of the electric field intensity at the current moment and uses this data to reconstruct a completely new spatial distribution surface of the electric field.

[0086] Next, the system recalculates the discrete curvature features, including Gaussian curvature and mean curvature, based on this new distribution surface, forming a complete "new baseline." Then, this new surface and new curvature features directly replace the original background model. From then on, all subsequent curvature differencing, gradient calculations, and discontinuity detection will use this new model as the zero-point reference.

[0087] The greatest value of this update lies in its "zero-intervention" nature and robustness. The update is only triggered when the indicator remains low, avoiding the danger of erroneous updates when there is risk; at the same time, it only updates the two core data points, the distribution surface and the discrete curvature characteristics, ensuring minimal computational overhead.

[0088] The methods provided in this application can be applied to various scenarios, including but not limited to: In live-line maintenance work at voltage levels of 220kV and above, when workers approach conductors at heights or using insulated boom trucks, the system can monitor electric field disturbances in real time, accurately distinguish between large-area human approach and tool tip approach, provide graded warnings, and forcibly interrupt work when necessary to prevent high-voltage arc injuries. Secondly, in large-scale mechanical construction such as crane hoisting of towers or guy wires, the system, deployed in the boom and ground sensor array, can capture spatiotemporal coupling disturbance waves caused by the hook tip 1-2 seconds in advance, automatically triggering a voice command to stop or cutting off the power supply to prevent the hook from accidentally touching live parts. Finally, in cross-construction within transmission line corridors (such as highway bridge construction crossing lines), the system can be deployed on temporary protective frames, automatically adapting to background drift caused by wind swaying and temperature and humidity changes, ensuring long-term reliable operation and significantly reducing the risk of electric shock in multi-trade collaborative operations.

[0089] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0090] According to another embodiment, a power construction risk prediction system based on spatial curvature variation is provided. Figure 3 A schematic block diagram of a power construction risk prediction system based on spatial curvature variation is shown according to one embodiment. Figure 3 As shown, the system 300 includes:

[0091] The multi-layer wavelet packet decomposition module 301 is used to acquire the electric field intensity time series data of each measuring point in the construction area, and to perform multi-layer wavelet packet decomposition on the electric field intensity time series data. The multi-layer wavelet packet decomposition treats the electric field intensity time series data of multiple measuring points as a multi-dimensional signal tensor, and recursively decomposes the tensor into high-frequency sub-bands and low-frequency sub-bands, thereby obtaining multiple spatiotemporally coupled electric field sub-band signals of different scales.

[0092] The multi-scale perturbation feature extraction module 302 is used to extract multi-scale perturbation features from the electric field sub-band signal. The multi-scale perturbation features include energy change rate, energy distribution ratio, transient energy pulse features, and energy redistribution relationship.

[0093] The spatial curvature feature calculation module 303 is used to construct an electric field spatial distribution surface based on the electric field intensity time series data, and calculate discrete curvature features, including Gaussian curvature and average curvature; and obtain spatial perturbation curvature features by detecting curvature abrupt change locations, curvature time change trends and curvature discontinuities on the electric field spatial distribution surface.

[0094] The feature fusion module 304 is used to fuse the multi-scale perturbation features, the discrete curvature features and the spatial perturbation curvature features to obtain a comprehensive perturbation index that reflects the overall stability of the electric field.

[0095] The risk assessment and early warning module 305 is used to determine that there is an abnormal disturbance when the comprehensive disturbance index reaches the preset risk conditions, and to generate early warning information.

[0096] As an implementable approach, the multi-layer wavelet packet decomposition module 301, when treating the electric field intensity time-series data of multiple measurement points as a multi-dimensional signal tensor and recursively decomposing the tensor into high-frequency and low-frequency sub-bands, can be configured as follows: The electric field intensity time-series data of all measurement points are arranged according to the actual spatial coordinates of the sensor within the construction area, forming a four-dimensional signal tensor, whose dimensions are the number of sensor channels, the number of time sampling points, the spatial horizontal index, and the spatial vertical index, respectively; One-dimensional wavelet packet filtering is performed on the four-dimensional signal tensor along the time dimension; During each layer of wavelet packet decomposition, the same wavelet packet filter banks are simultaneously applied to the low-frequency and high-frequency sub-bands along the spatial horizontal and spatial vertical directions, respectively, to achieve a joint triple recursive decomposition of the time dimension and the two spatial dimensions, obtaining multiple spatiotemporally coupled high-frequency and low-frequency sub-bands at different time and space scales.

[0097] As an implementable approach, the multi-layer wavelet packet decomposition module 301 can also be configured to: calculate the spatial synchronization correlation coefficient between all measurement point signals within the current high-frequency sub-band to be decomposed, as well as the energy ratio between the sub-band and the adjacent low-frequency sub-band, when the spatial synchronization correlation coefficient is greater than a preset correlation threshold and the energy ratio between the sub-band and the adjacent low-frequency sub-band is greater than a preset energy ratio threshold, determine that the high-frequency sub-band contains obvious disturbance propagation waves, and perform the next layer of recursive decomposition on the high-frequency sub-band.

[0098] As an implementable approach, the multi-scale perturbation feature extraction module 302 can also be configured to: calculate the energy transfer direction and transfer intensity between adjacent scale sub-bands, and use them as multi-scale energy flow features characterizing the approach direction of the perturbation source to constitute the multi-scale perturbation features.

[0099] As an implementable approach, the spatial curvature feature calculation module 303 can be configured to: calculate the difference between the Gaussian curvature of the current electric field spatial distribution surface and the Gaussian curvature of the pre-stored background model to obtain the curvature difference; apply differential operations to the curvature difference along three spatial directions to obtain the curvature gradient amplitude field; find local maxima points in the curvature gradient amplitude field, and determine the local maxima points that satisfy the preset amplitude threshold and are consistent with the curvature gradient direction of adjacent measuring points as curvature abrupt change locations; and use the coordinates of all curvature abrupt change locations and their corresponding curvature gradient amplitudes as one of the final spatial perturbation curvature features.

[0100] As an implementable approach, the spatial curvature feature calculation module 303 can be configured to: triangulate the electric field spatial distribution surface at the current moment to generate interconnected spatial grid patches; for any two adjacent triangular grid patches sharing a common edge, calculate the normal vectors of the two adjacent triangular grid patches respectively, and obtain the included angle between the two normal vectors; when the included angle is greater than a preset first angle threshold and the included angle maintains an increasing trend for at least three consecutive sampling periods, mark the corresponding common edge as a true curvature discontinuity edge, while common edges that only meet the included angle threshold but do not have an increasing trend are determined as discontinuity edges caused by noise or slowly varying background and are removed; merge all true curvature discontinuity edges with connected components to form one or more closed or approximately closed discontinuous contours, and extract the total length, closure degree, and geometric center position of each contour as the final spatial perturbation curvature feature.

[0101] As an implementable approach, system 300 may also include a background model update module, configured to: when the comprehensive disturbance index is continuously lower than the safety threshold for a preset duration, update the electric field spatial distribution surface constructed from the electric field intensity time series data at the current moment, and the discrete curvature features calculated based on the distribution surface, to a new background model to replace the original background model.

[0102] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. Components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0103] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0104] In addition, embodiments of this application also provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the method described in any of the foregoing method embodiments.

[0105] And an electronic device comprising: one or more processors; and a memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the method described in any of the foregoing method embodiments.

[0106] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described in any of the foregoing method embodiments.

[0107] in, Figure 4 An exemplary architecture of an electronic device is shown, which may include a processor 410, a video display adapter 411, a disk drive 412, an input / output interface 413, a network interface 414, and a memory 420. The processor 410, video display adapter 411, disk drive 412, input / output interface 413, network interface 414, and memory 420 can communicate with each other via a communication bus 430.

[0108] The processor 410 can be implemented using a general-purpose CPU, microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits to execute relevant programs and implement the technical solution provided in this application.

[0109] The memory 420 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 420 can store the operating system 421 for controlling the operation of the electronic device 400, and the basic input / output system (BIOS) 422 for controlling the low-level operations of the electronic device 400. Additionally, it can store a web browser 423, a data storage management system 424, and a power construction risk prediction system 425 based on spatial curvature changes, etc. The aforementioned power construction risk prediction system 425 based on spatial curvature changes can be the application program that specifically implements the aforementioned steps in this embodiment. In summary, when implementing the technical solution provided in this application through software or firmware, the relevant program code is stored in the memory 420 and executed by the processor 410.

[0110] Input / output interface 413 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touch screens, microphones, various sensors, etc., and output devices may include displays, speakers, vibrators, indicator lights, etc.

[0111] Network interface 414 is used to connect a communication module (not shown in the figure) to enable communication and interaction between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0112] Bus 430 includes a pathway for transmitting information between various components of the device, such as processor 410, video display adapter 411, disk drive 412, input / output interface 413, network interface 414, and memory 420.

[0113] It should be noted that although the above-described device only shows the processor 410, video display adapter 411, disk drive 412, input / output interface 413, network interface 414, memory 420, bus 430, etc., in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the solution of this application, and does not necessarily include all the components shown in the figures.

[0114] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer program product. This computer program product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0115] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for predicting power construction risks based on spatial curvature variation, characterized in that, The method includes: The electric field intensity time series data of each measuring point in the construction area are acquired, and the electric field intensity time series data are subjected to multi-level wavelet packet decomposition. The multi-level wavelet packet decomposition regards the electric field intensity time series data of multiple measuring points as a multi-dimensional signal tensor, and recursively decomposes the tensor into high-frequency sub-band and low-frequency sub-band, thereby obtaining multiple spatiotemporally coupled electric field sub-band signals of different scales. Multi-scale perturbation features are extracted from the electric field sub-band signal. These multi-scale perturbation features include energy change rate, energy distribution ratio, transient energy pulse characteristics, and energy redistribution relationship. Based on the electric field intensity time series data, an electric field spatial distribution surface is constructed, and discrete curvature features are calculated, including Gaussian curvature and average curvature. Spatial perturbation curvature features are obtained by detecting curvature abrupt change locations, curvature time change trends, and curvature discontinuities on the electric field spatial distribution surface. By fusing the multi-scale perturbation features, the discrete curvature features, and the spatial perturbation curvature features, a comprehensive perturbation index reflecting the overall stability of the electric field is obtained. When the comprehensive disturbance index reaches the preset risk condition, it is determined that there is an abnormal disturbance and an early warning message is generated.

2. The method according to claim 1, characterized in that, The multi-layer wavelet packet decomposition treats the electric field intensity time-series data from multiple measurement points as a multi-dimensional signal tensor, and recursively decomposes the tensor into high-frequency subbands and low-frequency subbands, including: The electric field intensity time series data of all measuring points are arranged according to the actual spatial coordinates of the sensors in the construction area to form a four-dimensional signal tensor, whose dimensions are the number of sensor channels, the number of time sampling points, the spatial horizontal index, and the spatial vertical index, respectively. During each level of wavelet packet decomposition, the same wavelet packet filter bank is applied sequentially to the low-frequency subband and the high-frequency subband along the horizontal and vertical spatial directions, respectively, to achieve joint triple recursive decomposition of the time dimension and the two spatial dimensions, thereby obtaining multiple spatiotemporally coupled high-frequency and low-frequency subbands at different time and space scales.

3. The method according to claim 2, characterized in that, At each level of the wavelet packet decomposition, the method further includes: For the high-frequency subband to be decomposed, calculate the spatial synchronization correlation coefficient between all measurement points in the subband and the energy ratio between the subband and the adjacent low-frequency subband; When the spatial synchronization correlation coefficient is greater than the preset correlation threshold and the energy ratio of the sub-band to the adjacent low-frequency sub-band is greater than the preset energy ratio threshold, it is determined that the high-frequency sub-band contains obvious disturbance propagation waves, and the high-frequency sub-band is recursively decomposed into the next layer.

4. The method according to claim 1, characterized in that, When extracting multi-scale perturbation features, the method further includes: Calculate the energy transfer direction and intensity between adjacent scale sub-bands, and use them as multi-scale energy flow characteristics to characterize the approach direction of the disturbance source, which are then used to construct the multi-scale disturbance characteristics.

5. The method according to claim 1, characterized in that, The detection of curvature abrupt change locations includes: Calculate the difference between the Gaussian curvature of the electric field spatial distribution surface at the current moment and the Gaussian curvature of the pre-stored background model to obtain the curvature difference; By applying difference operations to the curvature difference along three spatial directions, a curvature gradient magnitude field is obtained. Find local maxima in the curvature gradient amplitude field, and determine the local maxima that satisfy the preset amplitude threshold and are consistent with the curvature gradient direction of the adjacent measurement points as curvature abrupt change locations; The coordinates of all curvature abrupt change locations and their corresponding curvature gradient magnitudes are used as one of the final spatial perturbation curvature features.

6. The method according to claim 1, characterized in that, The curvature discontinuity detection includes: The electric field spatial distribution surface at the current moment is triangulated to generate interconnected spatial grid patches; For any two adjacent triangular mesh faces that share a common edge, calculate the normal vectors of the two adjacent triangular mesh faces respectively, and find the angle between the two normal vectors; When the included angle is greater than the preset first angle threshold and the included angle continues to increase for at least three consecutive sampling periods, the corresponding common edge is marked as a true curvature discontinuity edge, while common edges that only meet the first angle threshold but do not have an increasing trend are determined to be discontinuities caused by noise or slowly changing background and are removed. All the real curvature discontinuity edges are merged into connected components to form one or more closed or nearly closed discontinuous contours. The total length, closure degree, and geometric center position of each contour are extracted as the final spatial perturbation curvature features.

7. The method according to claim 5, characterized in that, The method further includes: when the comprehensive disturbance index is continuously lower than the safety threshold for a preset duration, updating the electric field spatial distribution surface constructed from the electric field intensity time series data at the current moment, and the discrete curvature features calculated based on the distribution surface, to a new background model to replace the original background model.

8. A power construction risk prediction system based on spatial curvature variation, characterized in that, The system includes: The multi-layer wavelet packet decomposition module is used to acquire the electric field intensity time series data of each measuring point in the construction area, and to perform multi-layer wavelet packet decomposition on the electric field intensity time series data. The multi-layer wavelet packet decomposition treats the electric field intensity time series data of multiple measuring points as a multi-dimensional signal tensor, and recursively decomposes the tensor into high-frequency sub-bands and low-frequency sub-bands, thereby obtaining multiple spatiotemporally coupled electric field sub-band signals of different scales. A multi-scale perturbation feature extraction module is used to extract multi-scale perturbation features from the electric field sub-band signal. The multi-scale perturbation features include energy change rate, energy distribution ratio, transient energy pulse features, and energy redistribution relationship. The spatial curvature feature calculation module is used to construct an electric field spatial distribution surface based on the electric field intensity time series data, and calculate discrete curvature features, including Gaussian curvature and average curvature; and obtain spatial perturbation curvature features by detecting curvature abrupt change locations, curvature time change trends and curvature discontinuities on the electric field spatial distribution surface. The feature fusion module is used to fuse the multi-scale perturbation features, the discrete curvature features, and the spatial perturbation curvature features to obtain a comprehensive perturbation index that reflects the overall stability of the electric field. The risk assessment and early warning module is used to determine the existence of abnormal disturbances and generate early warning information when the comprehensive disturbance index reaches a preset risk condition.

9. An electronic device, characterized in that, include: One or more processors; and a memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the method according to any one of claims 1 to 7.

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

Images