Substation safety operation area dynamic modeling method based on electric field-point cloud coupling

Abstract

The application relates to the technical field of substation modeling, and discloses a substation safety operation area dynamic modeling method based on electric field-point cloud coupling. The application solves the low-temperature operation problem through two mechanisms: firstly, a temperature-phase-time three-dimensional compensation model is constructed, temperature change rate is used for segmented adaptive compensation and residual self-optimization, coupling amplification of phase distortion and spatial mapping deviation is eliminated, the corrected phase is used as an extended Kalman filter prior constraint, closed-loop collaborative suppression of two types of errors is realized, and distance measurement accuracy reaches the centimeter level; secondly, an environment temperature and phase compensation residual double-factor adaptive weighted fusion is adopted, residual trend prediction is used for weight pre-adjustment, working condition partition gradual smooth transition is adopted, low-quality electric field data pollution of the fusion result is avoided, and position reconstruction standard deviation is reduced by about 60%. The two mechanisms jointly guarantee the accuracy and stability of the low-temperature safety area dynamic modeling.

Description

Technical Field

[0001] This invention relates to the field of substation modeling technology, and in particular to a dynamic modeling method for safe operation areas in substations based on electric field-point cloud coupling. Background Technology

[0002] In substation live-line working scenarios, existing safety area monitoring methods typically rely on single point cloud positioning or electric field strength judgment, with sensor weights often fixed. However, during winter operations in cold northern regions, broadband electric field sensors are affected by the low-temperature environment, causing changes in the response speed of their internal electronic components and resulting in a systematic phase shift in the output signal. Simultaneously, the mechanical deformation of the boom of special vehicles and the increased delay in positioning signal transmission at low temperatures cause the spatial mapping deviation to increase with decreasing temperature. These two errors are not independent—phase distortion leads to angular deviation of the positioning reference, which is further amplified into distance error during spatial mapping, creating a superimposed amplification effect. Existing static phase compensation methods cannot respond to sudden temperature changes, and fixed-weight fusion strategies struggle to balance the confidence levels of point cloud and electric field data under harsh low-temperature conditions, leading to inaccurate real-time distance judgment between the boom end and the live conductor, failing to meet the accuracy and real-time requirements of dynamic safety operation area modeling. Therefore, this invention aims to solve the following two technical problems:

[0003] First, how to decouple and synergistically suppress the coupling amplification effect between the signal phase distortion of the broadband electric field sensor and the deviation distance of the boom space mapping under low temperature conditions, so as to achieve joint correction of the two types of errors.

[0004] Second, how to dynamically adjust the fusion weights of point cloud positioning data and spatial electric field gradient positioning data based on changes in ambient temperature and phase compensation residuals, so as to avoid a decrease in positioning accuracy under low temperature and adverse chemical conditions. Summary of the Invention

[0005] The technical problem to be solved by this invention is that the existing technology has the disadvantages of decoupling the coupling of low temperature phase distortion and mapping deviation and dynamically adjusting the weight to maintain the accuracy of low temperature positioning. To this end, we propose a dynamic modeling method for the safe operation area of ​​substations based on electric field-point cloud coupling.

[0006] To achieve the above objectives, this application adopts the following technical solution: a dynamic modeling method for safe operation areas in substations based on electric field-point cloud coupling, comprising the following steps:

[0007] Construct a spatiotemporal synchronous acquisition and error pre-compensation system for multi-source heterogeneous data: Integrate and install lidar, broadband electric field sensor, inertial measurement unit and temperature compensation module at the end of the boom of special vehicle, trigger synchronous acquisition based on GPS disciplined clock source, and obtain raw three-dimensional point cloud data, spatial electric field gradient data, boom attitude data and ambient temperature data;

[0008] The collected raw data is preprocessed and features are extracted to establish a two-way coupling relationship between 3D point cloud data and spatial electric field gradient data.

[0009] Electric field-point cloud coupled modeling and dynamic safety zone construction: The bidirectionally calibrated 3D point cloud-spatial electric field gradient correlation data is input into the joint error compensation module. This module constructs a closed-loop collaborative suppression architecture with temperature-phase compensation as the prior constraint of extended Kalman filter, realizing the decoupling and iterative correction of the phase distortion of broadband electric field sensor signal and the distance of boom spatial mapping deviation, and outputs the boom spatial position signal and spatial electric field gradient information after joint correction.

[0010] Based on the ambient temperature threshold and phase compensation residual, the fusion weights of the three-dimensional point cloud positioning data and the spatial electric field gradient positioning data are dynamically allocated, and the two are weighted and fused to reconstruct the real-time spatial position of the boom end; based on the reconstructed boom spatial position and the spatial electric field gradient intensity distribution, three-layer working areas of core restricted area, warning area and safe area are dynamically generated.

[0011] Based on the constructed dynamic security zone model, intelligent monitoring, visual early warning, and decision support are provided.

[0012] Furthermore, the system for constructing a spatiotemporal synchronous acquisition and error pre-compensation system for multi-source heterogeneous data also includes:

[0013] A three-dimensional rotation matrix is ​​constructed using the real-time pitch, roll, and yaw angles output by the inertial measurement unit to transform the spatial electric field gradient direction acquired by the broadband electric field sensor to align with the three-dimensional point cloud coordinate system. The ambient temperature data is acquired in real time through the temperature compensation module, and the pre-stored temperature-phase response characteristic curve of the broadband electric field sensor is called. The corresponding phase correction coefficient is matched according to the current ambient temperature to perform phase shift compensation on the original output signal of the broadband electric field sensor.

[0014] The temperature-phase response characteristic curve is obtained through laboratory calibration. The calibration method is as follows: place the broadband electric field sensor in a temperature control chamber, set multiple calibration points within a preset temperature range, measure the phase shift of the sensor output signal at each calibration point, fit the correspondence between temperature and phase shift into a continuous curve and store it.

[0015] Furthermore, the joint error compensation module implements decoupling and iterative correction, specifically including:

[0016] Dynamic compensation for signal phase distortion based on temperature-phase characteristic curve: An adaptive filtering algorithm is used to track the signal phase deviation of a broadband electric field sensor in real time. The temperature-phase compensation coefficient is used as the initial constraint of the adaptive filter to construct a three-dimensional compensation model of temperature-phase-time and output phase information that has been adaptively corrected by ambient temperature.

[0017] Extended Kalman Filter Iterative Correction with Phase Correction as State Update Constraint: The corrected phase information is used as the prior constraint for the state update of the extended Kalman filter. A multibody kinematic model of the vehicle boom is constructed. In the filtering update stage, the phase correction is introduced into the observation equation in the form of a time-varying bias term to perform constraint correction on the recursive result and output the boom spatial position signal after joint correction.

[0018] Furthermore, the dynamic compensation for signal phase distortion based on the temperature-phase characteristic curve also includes:

[0019] Segmented adaptive compensation parameter adjustment based on temperature change rate: In the temperature-phase-time three-dimensional compensation model, the ambient temperature change rate for two consecutive sampling periods is calculated; when the absolute value of the temperature change rate exceeds the preset change rate threshold, it is determined that the ambient temperature is in a state of drastic fluctuation, and the system automatically switches to the fast response compensation mode.

[0020] In the fast response compensation mode, the current temperature change rate is used as input, and the pre-stored temperature change rate-phase response correction table is called to perform a secondary dynamic adjustment on the phase correction coefficient.

[0021] Furthermore, the dynamic compensation for signal phase distortion based on the temperature-phase characteristic curve also includes:

[0022] Self-evaluation of compensation effect and iterative optimization of correction coefficient based on phase compensation residual: After each phase compensation is completed, the residual value between the compensated signal phase and the theoretical undistorted phase is calculated;

[0023] When the residual value exceeds the preset residual tolerance threshold for multiple consecutive sampling periods, online iterative optimization of the correction coefficient is triggered. The iterative optimization adopts the gradient descent algorithm, which uses the partial derivative of the current residual value with respect to each feature point of the temperature-phase response characteristic curve as the gradient direction to fine-tune the feature point parameters and generate an updated characteristic curve.

[0024] Furthermore, the dynamic allocation of fusion weights for 3D point cloud positioning data and spatial electric field gradient positioning data specifically includes: using ambient temperature and phase compensation residual as dual-factor judgment criteria; when the ambient temperature is lower than a preset threshold and the phase compensation residual exceeds a set range, the system determines that it has entered a low-temperature signal degradation condition, automatically reduces the weight ratio of the broadband electric field sensor in the fusion calculation, and at the same time increases the prediction weight based on the inertial measurement unit and kinematic model; when the ambient temperature rises and the phase compensation residual returns to the normal range, the fusion weight of the broadband electric field sensor is gradually restored.

[0025] Furthermore, the dynamic allocation of fusion weights also includes a weight pre-adjustment mechanism based on residual trend prediction: a residual trend prediction submodule is constructed, which uses a linear regression method to fit a residual change trend line based on the phase compensation residual values ​​of the most recent sampling periods, and predicts the residual change direction of future sampling periods based on the slope of the trend line; when the prediction result shows that the residual will exceed the residual tolerance threshold in the next period, the system reduces the fusion weight of the broadband electric field sensor one sampling period in advance, and allocates the weight adjustment amount proportionally according to the prediction confidence.

[0026] Furthermore, the dynamic allocation of fusion weights also includes a weight smoothing transition strategy based on operating condition partitions: the two-dimensional plane formed by ambient temperature and phase compensation residuals is divided into four operating condition partitions: low temperature high residual zone, low temperature low residual zone, normal temperature high residual zone, and normal temperature low residual zone; each partition corresponds to a set of preset weight change rate parameters, including the rising rate when the weight increases and the falling rate when the weight decreases; when the system detects that the current operating condition point crosses the boundary from one partition to another, it performs a gradual adjustment according to the change rate corresponding to the target partition, and the weight value within the gradual transition period is transitioned by linear interpolation.

[0027] Furthermore, the intelligent monitoring, visual early warning, and decision support include:

[0028] Construction of a safe distance prediction model for boom motion trend: Using the boom kinematic model, combined with the boom's historical motion trajectory and current motion state, a trajectory prediction algorithm based on extended Kalman filter is used to predict the position trajectory of the boom end within the future time window;

[0029] Spatial collision detection is performed between the predicted trajectory and the 3D point cloud model of the substation's energized equipment; the dynamic safe distance between the boom end and the live conductor within the predicted time window is calculated in real time; tiered early warning and risk alerts are provided.

[0030] When the predicted safe distance is less than the first preset threshold, a level one warning is triggered, prompting the operator to slow down or adjust the boom posture; when the predicted safe distance is less than the second preset threshold, a level two warning is triggered, triggering an audible and visual alarm and automatically restricting boom movement.

[0031] Furthermore, the step of feeding back the warning signals triggered by the tiered early warning and risk alert to the step of dynamically allocating fusion weights specifically involves:

[0032] The early warning signal is fed back to the weight pre-adjustment mechanism based on residual trend prediction and the weight smooth transition strategy based on working condition partitioning.

[0033] Under high-risk operating conditions, the residual tolerance threshold and operating condition boundary parameters are temporarily adjusted to form a dynamic closed loop of early warning response and data collection and fusion strategy adjustment.

[0034] The technical effects and advantages of this invention are as follows:

[0035] This invention solves the problem of coupled amplification of phase distortion and spatial mapping deviation in broadband electric field sensor signals under low-temperature environments by constructing a three-dimensional compensation model based on temperature, phase, and time, and introducing a piecewise adaptive compensation mechanism based on the rate of temperature change. Specifically, the essence of phase distortion is that temperature changes cause changes in the response speed of the electronic components inside the sensor, resulting in a dynamic phase shift related to the rate of temperature change. This invention calculates the rate of temperature change and quickly adjusts the compensation parameters at the initial stage of temperature abrupt change, advancing the peak occurrence time of the phase error to the forefront of temperature change, thus eliminating the response blind zone of traditional hysteresis compensation. By continuously monitoring the compensation residual and iteratively optimizing the characteristic curve online, the compensation model can adaptively track the aging drift of the sensor. These two mechanisms work together to reduce the amplitude of the phase error by about 40%, and significantly reduce the variance of the phase compensation residual. Furthermore, the corrected phase information is used as a priori constraint for the state update of the extended Kalman filter, so that the phase correction and spatial position correction form a closed-loop cooperative suppression at the state vector level—the phase correction immediately affects the position estimation, while the position residual, in turn, corrects the subsequent phase compensation parameters. As a result, the two types of errors are transformed from mutual amplification to mutual suppression, and the distance measurement accuracy between the boom end and the live wire is improved to the centimeter level in an environment of minus 25 degrees Celsius, effectively solving the problem of inaccurate distance judgment caused by low temperature coupling error in the existing technology.

[0036] This invention employs a two-factor adaptive weighted fusion algorithm based on ambient temperature and phase compensation residuals to address the issue of decreased positioning accuracy caused by fixed-weight fusion under low-temperature and adverse working conditions. The logical chain is as follows: low temperatures reduce the signal-to-noise ratio of the electric field sensor, decreasing the reliability of its positioning contribution; if fixed weights are still used, low-quality electric field data will contaminate the fusion results. By introducing two independent indicators—ambient temperature and phase compensation residuals—the reliability of the electric field data is evaluated in real time. When the temperature is low and the residuals are large, the electric field weight is automatically reduced while the kinematic prediction weight is increased. Furthermore, residual trend prediction is used to achieve weight pre-adjustment, advancing the weight response time by one sampling period, avoiding positioning jitter caused by feedback delay; and a gradual change strategy based on working conditions eliminates abrupt changes in position reconstruction caused by weight jumps. The synergy of these three factors results in a significantly better position reconstruction continuity and accuracy in low-temperature conditions compared to the fixed-weight scheme, reducing the standard deviation of the boom end coordinate output by approximately 60%. Attached Figure Description

[0037] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts:

[0038] Figure 1 This is a flowchart illustrating the overall method of the present invention;

[0039] Figure 2 This is the logic diagram of the joint error compensation and extended Kalman filter iterative correction of the present invention;

[0040] Figure 3 This is the logic diagram for dynamic weight allocation and smooth transition in this invention;

[0041] Figure 4 This is a flowchart of the closed-loop process for safe distance prediction and graded early warning in this invention;

[0042] Figure 5 This is a rendering of the spatial situational awareness of the present invention. Detailed Implementation

[0043] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0044] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1This is a flowchart illustrating a dynamic modeling method for substation safe operation areas based on electric field-point cloud coupling, provided by an embodiment of the present invention. The following is a detailed description of this dynamic modeling method for substation safe operation areas based on electric field-point cloud coupling.

[0045] Step S110: Construct a spatiotemporal synchronous acquisition and error pre-compensation system for multi-source heterogeneous data.

[0046] In this embodiment, we will take the example of a winter live-line maintenance operation using a special vehicle, such as an aerial work platform, at a 500kV substation in northern China. First, an integrated data acquisition unit needs to be constructed at the end of the special vehicle's boom. This integrated unit rigidly mounts a lidar, binocular depth camera, broadband electric field sensor, voltage monitoring unit, inertial measurement unit, and temperature compensation module onto the same mechanical support, ensuring that the relative spatial positions of each sensor are fixed. Among them, lidar refers to a sensor that measures the distance and spatial coordinates of a target by emitting a laser beam and receiving the reflected signal, used to acquire high-precision three-dimensional point cloud data; a binocular depth camera refers to a camera that calculates the depth of a target by simulating the parallax principle of the human eye through two cameras, used to assist in acquiring depth information to compensate for the data loss of a single sensor; a broadband electric field sensor refers to a sensing device that can continuously measure the intensity and gradient of the spatial electric field within a certain frequency range, used to monitor the spatial electric field gradient distribution in real time; a voltage monitoring unit refers to a device used to acquire the voltage amplitude and phase state of charged equipment in real time, used to obtain the real-time voltage state of the equipment; an inertial measurement unit refers to a sensor that integrates a three-axis accelerometer and a three-axis gyroscope and can measure the attitude angle of the carrier, used to acquire the pitch angle, roll angle and yaw angle data of the boom; and a temperature compensation module refers to a compensation circuit that has a built-in temperature sensing element and stores the sensor's temperature characteristic curve, used to acquire the ambient temperature in real time and generate corresponding signal correction coefficients.

[0047] Synchronous acquisition is triggered by a GPS-disciplined clock source to obtain raw 3D point cloud data, auxiliary depth information, spatial electric field gradient data, equipment voltage status, boom attitude data, and ambient temperature data, ensuring precise alignment of all data in the time dimension. A GPS-disciplined clock source refers to a time synchronization device that uses the high-precision time signal of the Global Positioning System to calibrate the local clock, achieving microsecond-level synchronization accuracy and ensuring alignment of data collected by multiple sensors under the same time reference. In this embodiment, the lidar acquires 3D point cloud data at a frequency of 20Hz, the broadband electric field sensor acquires spatial electric field gradient data at a sampling frequency of 1kHz, the inertial measurement unit outputs boom attitude data at a frequency of 200Hz, and the temperature compensation module acquires ambient temperature data at a frequency of 10Hz.

[0048] Step S111: Attitude alignment compensation of the inertial measurement unit

[0049] A three-dimensional rotation matrix is ​​constructed using the real-time pitch, roll, and yaw angles output by the inertial measurement unit (IMU) to transform the spatial electric field gradient direction acquired by the broadband electric field sensor to align with the three-dimensional point cloud coordinate system. Specifically, the three-dimensional rotation matrix is ​​constructed as follows: using the pitch, roll, and yaw angles output by the IMU as inputs, three basic rotation matrices are constructed around the X, Y, and Z axes respectively. These three basic rotation matrices are then multiplied to obtain the final three-dimensional rotation matrix. Multiplying this rotation matrix by the spatial electric field gradient vector output by the broadband electric field sensor on the left achieves the transformation of the electric field gradient direction from the sensor coordinate system to the three-dimensional point cloud coordinate system. This step eliminates measurement errors in the electric field direction caused by boom movement or equipment installation deviations, providing a directional reference for the subsequent accurate fusion of electric field data and spatial geometry. In this embodiment, the angle between the transformed electric field gradient direction and the point cloud normal vector must not exceed 5 degrees; otherwise, an attitude calibration alarm is triggered.

[0050] Step S112: Pre-correction of the phase of the broadband electric field sensor signal by the temperature compensation module.

[0051] Ambient temperature data is collected in real time by a temperature compensation module, and the pre-stored temperature-phase response characteristic curve of a broadband electric field sensor is retrieved at the signal processing front end. This characteristic curve is obtained through laboratory calibration. Specifically, the broadband electric field sensor is placed in a temperature-controlled chamber, and calibration points are set every 5 degrees Celsius within a temperature range of -40°C to 60°C. At each calibration point, the phase shift of the sensor's output signal is measured, and the relationship between temperature and phase shift is fitted into a continuous curve, which is stored in the non-volatile memory of the temperature compensation module. A phase correction coefficient is matched to the current ambient temperature to compensate for the phase shift of the original output signal of the broadband electric field sensor. For example, when the ambient temperature drops to -25°C, the corresponding phase correction coefficient is retrieved from the temperature-phase response characteristic curve, and the phase of the original output signal is compressed and corrected. This step eliminates the systematic influence of the low-temperature environment on the phase of the electric field signal in advance, providing a stable input source for subsequent signal processing. In this embodiment, the calculation period for phase compensation does not exceed 10 milliseconds, and the compensation delay error is less than one-thousandth.

[0052] Step S113: Synchronous acquisition of auxiliary depth information

[0053] Auxiliary depth information is simultaneously acquired using a binocular depth camera and then fused with the raw 3D point cloud data acquired by a lidar system. The binocular depth camera works by simultaneously capturing images of the same scene from two cameras. Due to the baseline distance between the two cameras, there is a parallax in the position of the same target on the imaging planes of the left and right cameras. The depth information of the target can be obtained by calculating the parallax using triangulation principles. When lidar data is missing in complex occluded scenarios, such as when there is cross-occlusion between the boom and equipment, the auxiliary depth information acquired by the binocular depth camera can fill in these missing areas, improving the completeness and density of the 3D point cloud data. In this embodiment, the baseline distance of the binocular depth camera is set to 20 cm, the depth measurement range is 0.5 m to 30 m, and the depth measurement accuracy reaches the centimeter level. The fused 3D point cloud data density reaches 500 points per square meter, providing richer geometric information for subsequent 3D spatial reconstruction.

[0054] Step S120: Preprocessing and feature extraction of multi-source data

[0055] The raw data acquired by the acquisition layer is cleaned, calculated, and feature extracted to establish a two-way coupling relationship between 3D point cloud data and spatial electric field gradient data, providing a high-quality data foundation for subsequent modeling and fusion.

[0056] Step S121: Comprehensive preprocessing of 3D point cloud data

[0057] The original 3D point cloud data is sequentially subjected to denoising filtering, multi-view registration and stitching, equipment region segmentation and identification, and 3D spatial reconstruction to form a complete 3D point cloud model of the substation scene. Denoising filtering employs a statistical filtering algorithm, specifically: calculating the average distance between each point and its several neighboring points; assuming the average distance follows a Gaussian distribution; and identifying and removing points whose average distance exceeds the mean plus or minus a certain number of standard deviations. Multi-view registration and stitching uses an iterative nearest-point algorithm, continuously optimizing the rotation matrix and translation vector between two point clouds to align overlapping areas. Equipment region segmentation and identification uses a Euclidean clustering-based segmentation algorithm to separate point clouds of different equipment such as towers, conductors, and insulators based on the spatial distance between points. 3D spatial reconstruction uses a Poisson surface reconstruction algorithm to generate a complete 3D geometric structure based on the point cloud and its normal vectors. Geometric features, including curvature, flatness, and linearity, are extracted, and surface normal directions are calculated to provide a spatial geometric basis for safety boundary construction.

[0058] Step S122: Comprehensive processing and risk assessment of spatial electric field gradient data

[0059] Based on real-time spatial electric field gradient data acquired by a broadband electric field sensor and combined with equipment voltage status obtained by a voltage monitoring unit, the spatial electric field distribution is calculated. The spatial electric field distribution calculation employs the finite element method (FEM), dividing the substation operating area into a finite number of grid cells. The Poisson equation is solved on each grid cell to obtain the potential distribution of the entire space, and then the electric field intensity distribution is obtained by calculating the gradient of the potential. Risk assessment of the operating area is performed based on preset safety thresholds: when the electric field intensity exceeds 10 kV / m, it is marked as a core restricted area; when the electric field intensity is between 5 kV / m and 10 kV / m, it is marked as a warning zone; and when the electric field intensity is below 5 kV / m, it is marked as a safe zone. Real-time tracking of dynamic field changes is performed, and multi-parameter data fusion processing is completed to output an electric field intensity distribution map and risk level assessment results.

[0060] Step S123: Two-way coupling of 3D point cloud data and spatial electric field gradient data

[0061] A data association and feedback calibration mechanism is established between the 3D point cloud data processing module and the spatial electric field gradient data processing module. 3D point cloud data provides precise spatial geometric constraints for calculating the spatial electric field distribution. Specifically, during finite element mesh generation, the equipment surface in the 3D point cloud model is used as the mesh boundary, ensuring that the electric field calculation accurately reflects the influence of the actual spatial location of the equipment on the electric field distribution. Spatial electric field gradient data provides physical field feature assistance for 3D point cloud segmentation and recognition. Specifically, during point cloud clustering and segmentation, the spatial electric field gradient intensity is used as a weighting factor, assigning higher weights to regions with high electric field intensity during clustering, helping to distinguish equipment regions with different electric field characteristics. The two modules form a collaborative processing closed loop, outputting bidirectionally calibrated 3D point cloud-spatial electric field gradient association data.

[0062] Step S130: Electric field-point cloud coupled modeling and dynamic safe region construction

[0063] This step is the core innovation of the present invention. Through the synergistic mechanism of joint error compensation and adaptive weighted fusion, the phase distortion and spatial mapping deviation of the broadband electric field sensor signal are decoupled and suppressed. On this basis, the electric field distribution and spatial geometry are deeply integrated to construct a dynamically adaptive safe working area.

[0064] Step S131: Joint error compensation and dynamic correction of mapping deviation

[0065] The bidirectionally calibrated 3D point cloud-spatial electric field gradient correlation data output from the data processing layer is input to the joint error compensation module. This module constructs a closed-loop cooperative suppression architecture with temperature-phase compensation as the prior constraint of the extended Kalman filter, realizing the decoupling and iterative correction of the two types of errors, and outputting the boom spatial position signal and spatial electric field gradient information after joint correction.

[0066] Step S1311: Dynamic compensation for signal phase distortion based on temperature-phase characteristic curve

[0067] An adaptive filtering algorithm is employed to track the signal phase deviation of a broadband electric field sensor in real time. The temperature-phase compensation coefficient output in step S112 is used as the initial constraint for the adaptive filter, constructing a three-dimensional compensation model based on temperature, phase, and time. Specifically, the adaptive filtering algorithm uses the minimum mean square error criterion, iteratively adjusting the filter's weight coefficients to minimize the mean square error between the filter's output and the desired output. The core of the temperature-phase-time three-dimensional compensation model is a three-dimensional lookup table, with its three dimensions being ambient temperature, phase compensation residual, and timestamp. The lookup table stores the optimal phase compensation parameters under different operating conditions. This model can dynamically adjust the compensation parameters based on changes in ambient temperature and the historical trend of phase error, outputting phase information adaptively corrected by ambient temperature. Since the phase shift caused by low temperature is not a fixed value but changes with temperature and exhibits a memory effect, a single static compensation coefficient cannot track this dynamic change. This step, by constructing a three-dimensional compensation model, incorporates the rate of temperature change into the adjustment basis of the compensation parameters, enabling the compensation coefficient to respond in advance according to the direction and amplitude of temperature fluctuations. Simultaneously, the phase compensation residual is used as a feedback signal to evaluate the compensation effect in real time, thereby controlling the phase error within a preset tolerance range. As a result, the angular interference of phase distortion on the positioning reference is significantly suppressed, providing high-precision observation input for subsequent extended Kalman filtering and reducing the coupling strength of the two types of errors from the source.

[0068] Step S13111: Piecewise adaptive compensation parameter adjustment based on temperature change rate

[0069] In the temperature-phase-time three-dimensional compensation model, a temperature change rate calculation unit is added. This unit uses the ratio of the ambient temperature difference between two consecutive sampling periods to the sampling interval as the temperature change rate. For example, if the temperature in the current sampling period is -20 degrees Celsius, the temperature in the previous sampling period was -22 degrees Celsius, and the sampling interval is 0.1 seconds, then the temperature change rate is 20 degrees Celsius per second. When the absolute value of the temperature change rate exceeds the preset change rate threshold, it is determined that the ambient temperature is in a state of drastic fluctuation, and the model automatically switches to the fast response compensation mode. In the fast response compensation mode, using the sign and amplitude of the current temperature change rate as input, the pre-stored temperature change rate-phase response correction table is called to perform a secondary dynamic adjustment on the phase correction coefficient output in step S112, so that the compensation parameters can respond to the sudden temperature trend in advance rather than tracking it with lag. The phase correction coefficient after secondary adjustment is simultaneously fed back to step S13131 as a reference input for predicting the phase compensation residual. In winter operations, the temperature may drop by more than 10 degrees Celsius in a short period of time, and the traditional lag compensation method will generate an error window of several seconds due to the response delay. This step calculates the rate of temperature change and sets up a piecewise adaptive mechanism to quickly adjust the compensation parameters at the initial stage of a temperature surge, bringing the peak of the phase error to the forefront of the temperature change, thereby eliminating the positioning blind spot caused by the hysteresis response. This improvement reduces the maximum deviation of phase distortion by about 40% under conditions of rapid temperature change, providing a more stable observation sequence for subsequent extended Kalman filtering.

[0070] Step S13112: Self-evaluation of compensation effect and iterative optimization of correction coefficients based on phase compensation residuals

[0071] After each phase compensation, the residual value between the compensated signal phase and the theoretical distortion-free phase is calculated, and this residual value is used as a self-evaluation index of the compensation effect. The theoretical distortion-free phase refers to the ideal phase value that the electric field sensor should output when there is no temperature influence. This value is obtained by statistical analysis of historical distortion-free period data. When the residual value exceeds the preset residual tolerance threshold for multiple consecutive sampling periods, online iterative optimization of the correction coefficient is triggered. The iterative optimization adopts the gradient descent algorithm. The specific process is as follows: using the partial derivative of the current residual value with respect to each feature point of the temperature-phase response characteristic curve as the gradient direction, the feature point parameters are finely adjusted along the gradient descent direction to generate an updated characteristic curve. The updated characteristic curve is then stored in the non-volatile memory of the temperature compensation module for subsequent compensation cycles. The residual sequence and its statistical characteristics generated during this iterative optimization process, including the residual mean and residual variance, are output to step S13311 as the dynamic benchmark for "phase compensation residual" in the two-factor weight allocation. As the sensor is used for longer periods and subjected to low-temperature cycling, its temperature-phase characteristic curve may slowly drift. This step involves continuously monitoring the compensation residuals. When the residuals exceed limits, online iterative optimization is initiated. The gradient descent method is used to fine-tune key points of the characteristic curve, enabling the compensation model to adaptively track sensor aging or performance changes. This closed-loop self-optimization mechanism ensures the continuous stability of phase compensation accuracy during long-term use, avoids the accumulation of phase errors caused by model mismatch, and thus maintains the fundamental effectiveness of decoupling and suppressing the two types of errors.

[0072] See Figure 2 As shown, step S1312: Extended Kalman filter iterative correction with phase correction as the state update constraint.

[0073] The corrected phase information output in step S1311 is used as the prior constraint for the state update of the extended Kalman filter to construct a multibody kinematics model of the vehicle boom. The multibody kinematics model decomposes the boom into multiple rigid bodies, including the base, lower boom, upper boom, and end effector. Each rigid body is connected by a rotary joint, and the spatial position of the end effector can be calculated using joint angles and boom length parameters. The extended Kalman filter uses the boom end position, velocity, and attitude as state variables, totaling 11 state variables, including three-dimensional position components, three-dimensional velocity components, quaternion attitude components, and receiver clock bias. The phase correction is introduced as a time-varying bias term in the observation equation.

[0074] In the filtering prediction phase, the state is recursively predicted using the boom kinematic model, i.e., the current state is predicted based on the state estimate from the previous moment and the kinematic model. In the filtering update phase, a phase correction is used to constrain the recursive result. Specifically, the Kalman gain is calculated, and the state estimate is updated based on the difference between the corrected observation and the predicted value, ensuring that the updated state estimate satisfies both the constraints of the kinematic model and the constraints of the observation. Phase distortion and spatial mapping bias are not independent—phase distortion first causes an angular information bias in the observation, which is amplified into a distance error by geometric relationships when mapped to three-dimensional space through the kinematic model. This step uses the corrected phase information as a priori constraint for the extended Kalman filter's state update, and the phase correction is introduced into the observation equation as a time-varying bias term in the filtering update phase. Specifically, the corrected phase value directly participates in calculating the Kalman gain and updating the state covariance matrix, making the filter's state estimate subject to dual constraints from the phase correction and the kinematic model. In this way, the phase correction result and the spatial position correction result form a closed-loop synergistic suppression at the state vector level: the phase error, once corrected in time, immediately affects the position estimation, and the residual of the position estimation, in turn, corrects the subsequent phase compensation parameters. Thus, the coupled amplification effect of the two types of errors is transformed into a synergistic mechanism of mutual suppression, which significantly improves the convergence speed and steady-state accuracy of the boom end position estimation.

[0075] Step S132: Spatial mapping and temporal synchronization of 3D point cloud-spatial electric field gradient coupling

[0076] The spatial mapping and association module accurately maps the boom's spatial position and spatial electric field gradient information, corrected in step S1312, into a 3D point cloud coordinate system. The core of this module is a coordinate transformation matrix, calculated based on the 3D rotation matrix and boom kinematic model from step S111. This matrix converts the boom end position in the local coordinate system to its position in the global 3D point cloud coordinate system, aligning the spatial electric field gradient data with the 3D spatial geometry data in the spatial dimension. The temporal dynamic synchronization module uses timestamp interpolation to unify data from different sampling frequencies onto the same time axis. Specifically, using the time axis of the electric field data with the highest sampling frequency as a reference, linear interpolation is performed on the low-frequency point cloud data and inertial measurement unit data to ensure consistency between the spatial electric field gradient data and the 3D point cloud data in the temporal dimension. This achieves a deep dual fusion of electric field distribution and spatial geometry, outputting spatiotemporally synchronized 3D point cloud-spatial electric field gradient fused data.

[0077] Step S133: Dynamic Adaptive Safe Region Modeling

[0078] Based on the fused 3D point cloud-spatial electric field gradient fusion data, an adaptive weighted fusion algorithm based on ambient temperature threshold and phase compensation residual is used to fuse and reconstruct the 3D point cloud positioning data and spatial electric field gradient positioning data, construct a dynamic safe region, and output a safe region model with dynamic boundaries.

[0079] Step S1331: Dynamic allocation of sensor fusion weights based on ambient temperature threshold and phase compensation residual.

[0080] A dynamic allocation mechanism for the positioning weights of 3D point cloud and spatial electric field gradient is established. Ambient temperature and the phase compensation residual output in step S1311 are used as the two-factor judgment criteria. The specific judgment logic is as follows: when the ambient temperature is below a preset threshold and the phase compensation residual exceeds a set range, the system determines that it has entered a low-temperature signal degradation condition, automatically reducing the weight ratio of the broadband electric field sensor in the fusion calculation, while increasing the prediction weight based on the inertial measurement unit and kinematic model; when the ambient temperature rises and the phase compensation residual returns to the normal range, the fusion weight of the broadband electric field sensor is gradually restored, achieving a smooth transition of the fusion strategy. In low-temperature environments, the signal-to-noise ratio of the broadband electric field sensor decreases, reducing the reliability of its positioning contribution; while point cloud data is less affected by temperature. If a fixed weight is still used, low-quality electric field data will contaminate the fusion results, leading to increased boom position reconstruction errors. This step introduces two independent indicators—ambient temperature and phase compensation residual—to evaluate the reliability of the electric field data in real time: when the temperature is low and the residual is large, the electric field data is deemed unreliable, its fusion weight is automatically reduced, and the kinematic prediction weight is increased; conversely, the weight is restored. This dynamic allocation allows the fusion system to automatically select data sources with higher confidence under different operating conditions, avoiding interference from low-quality data under low-temperature and adverse operating conditions, thus ensuring the continuous accuracy of boom end position reconstruction.

[0081] Step S13311: Weight pre-adjustment mechanism based on residual trend prediction

[0082] In the two-factor judgment criterion, in addition to using the absolute value of the phase compensation residual at the current moment, the statistical characteristics of the residual sequence output in step S13112 are further introduced to construct a residual trend prediction submodule. This submodule uses a linear regression method to fit a residual change trend line based on the phase compensation residual values ​​of the most recent sampling periods, and predicts the direction of residual change in the next two sampling periods based on the slope of the trend line. When the prediction result shows that the residual will exceed the residual tolerance threshold in the next period, the system reduces the fusion weight of the broadband electric field sensor one sampling period in advance, and allocates the weight adjustment amount proportionally according to the prediction confidence, i.e., the goodness of fit of the trend line. At the same time, the pre-adjusted weight value is fed back to the iterative optimization stage of step S13112 as an external reference for the self-evaluation of the compensation effect, so that the dynamic adjustment of the residual threshold can take into account the impact of changes in the fusion weight. Traditional weight feedback adjustment has an inherent delay—the weight change is only triggered when the residual has exceeded the limit, which will lead to a brief decrease in fusion accuracy during the rapid rise of the residual. This step uses linear regression to predict the residual change trend and reduces the electric field weight in advance before the residual exceeds the limit, enabling the fusion system to have feedforward control capabilities. Simultaneously, the pre-adjustment magnitude is linked to the prediction confidence level, avoiding over-adjustment. This pre-adjustment mechanism advances the weight response time by one sampling period, effectively suppressing positioning jitter caused by weight switching lag, ensuring a smooth transition in boom position reconstruction even when phase errors deteriorate rapidly.

[0083] See Figure 3 As shown, step S13312: Weighted smooth transition strategy based on working condition partitioning

[0084] Based on the two-factor judgment criteria, the two-dimensional plane formed by ambient temperature and phase compensation residuals is divided into four operating condition zones: low temperature high residual zone, low temperature low residual zone, normal temperature high residual zone, and normal temperature low residual zone. Each zone corresponds to a set of preset weight change rate parameters, including the rate of increase when the weight increases and the rate of decrease when the weight decreases. When the system detects that the current operating point crosses the boundary from one zone to another, it does not directly jump the weight, but gradually adjusts it according to the change rate corresponding to the target zone. The weight value within the gradual change period is transitioned by linear interpolation. The intermediate weight value generated during this gradual change process is output to step S1332 in real time for smooth spatial position reconstruction. If the weight changes abruptly, the fused position coordinates will produce a step change, which manifests as a "jump" in the boom position in the monitoring screen. This not only affects visual continuity but may also trigger false alarms. This step divides the operating condition plane into four zones and sets different weight change rates for each zone. When crossing zones, linear interpolation is used to gradually change the weight value, making the weight value change continuously. For example, when switching from a low-temperature, high-residual region to a normal-temperature, low-residual region, the weight values ​​gradually increase over multiple sampling periods, rather than jumping instantaneously. This gradual approach limits the rate of change of the fusion position, fundamentally eliminating reconstruction jitter caused by sudden weight changes and ensuring the smoothness of the boom's motion trajectory and the stability of distance judgment in the 3D visualization scene.

[0085] Step S1332: Spatial location reconstruction by weighted fusion

[0086] The 3D point cloud positioning weights, after weight pre-adjustment in step S13311 and smoothing transition processing in step S13312, are combined with the spatial electric field gradient positioning weights to perform a weighted fusion of the 3D point cloud positioning results and the spatial electric field gradient positioning results. The weighted fusion calculation formula is: the fused position equals the point cloud positioning result multiplied by the point cloud weight plus the electric field positioning result multiplied by the electric field weight. Through weighted fusion, the real-time position coordinates of the vehicle boom end in 3D space are reconstructed, ensuring the continuity and accuracy of spatial position reconstruction under low-temperature conditions, and outputting a high-precision real-time spatial position of the boom end. These position coordinates will serve as the target input for the safety distance prediction model in step S141.

[0087] Step S1333: Dynamic generation of safe zones and identification of hazardous zones

[0088] Based on preset safety thresholds, and combined with the reconstructed boom spatial position and spatial electric field gradient intensity distribution, a three-tiered working area—core restricted area, warning zone, and safe zone—is dynamically generated. The specific generation rules are as follows: using the surface of the energized equipment as a reference, expanding outwards along the surface normal direction, different area boundaries are defined according to the spatial electric field gradient intensity. The boundary of the core restricted area is defined as an isosurface with a spatial electric field gradient intensity of 10 kV / m; the boundary of the warning zone is defined as an isosurface with a spatial electric field gradient intensity of 5 kV / m; and the safe zone is the area outside the warning zone. The core restricted area is the area with the highest spatial electric field gradient intensity and the greatest safety risk; the warning zone is the area requiring close attention and approaching the critical value; and the safe zone is the area that meets the requirements for safe operation. Areas with excessive spatial electric field gradient intensity are automatically marked as hazardous areas, enabling real-time identification and dynamic updating of hazardous areas, and outputting a dynamic safe zone model with hierarchical markings.

[0089] Step S140: Intelligent monitoring, visual early warning and decision support

[0090] Based on the dynamic safety zone model output by the modeling fusion layer, it provides users with 3D visualization, real-time early warning and intelligent decision support, and outputs visualized safety operation guidelines.

[0091] Step S141: Construction of a safe distance prediction model for boom movement trends

[0092] Using the boom kinematic model established in step S1312, and combining the boom's historical trajectory with its current motion state, the short-term motion trend of the boom tip is predicted. The prediction method employs a trajectory prediction algorithm based on extended Kalman filtering. Specifically, the process involves using the historical position sequence of the boom tip within the past second as input, fitting the motion model parameters (including velocity and acceleration) through extended Kalman filtering, and then predicting the position trajectory for the next two seconds. Spatial collision detection is performed between the predicted trajectory and the 3D point cloud model of the substation's energized equipment, and the dynamic safe distance between the boom tip and the energized conductor within the prediction time window is calculated in real time. The safe distance is calculated by determining the minimum Euclidean distance between each point on the predicted trajectory and the surface of the energized equipment, and taking the minimum of these minimum distances as the dynamic safe distance. The predicted safe distance value in the future time domain is output, and this predicted value will serve as the triggering basis for the graded early warning in step S143.

[0093] Step S142: 3D visualization and spatial electric field gradient superposition

[0094] See Figure 5As shown, the constructed dynamic safety zone is rendered in real-time within a 3D point cloud scene. Simultaneously, real-time spatial electric field gradient intensity data is overlaid and displayed in 3D space as a color cloud map. The color mapping rules for the color cloud map are as follows: areas with electric field strength below 2kV / m are displayed in green, areas between 2kV / m and 5kV / m in yellow, areas between 5kV / m and 8kV / m in orange, areas between 8kV / m and 10kV / m in red, and areas above 10kV / m in dark red. This intuitively presents the distribution of hazardous areas and changes in spatial electric field gradient intensity, providing operators with immersive spatial situational awareness and outputting a visual monitoring screen. The visual monitoring screen supports multi-view switching, including overhead, side, and boom-following views, and supports zoom functionality, allowing operators to observe the spatial relationship between the boom and live equipment from different angles.

[0095] See Figure 4 As shown, step S143: Tiered early warning and risk alert

[0096] When the predicted safe distance output in step S141 enters the preset threshold range, the system triggers a graded warning based on the risk level. The preset threshold range is set as follows: a Level 1 warning is triggered when the predicted safe distance is less than 1.5 times the minimum safe distance; a Level 2 warning is triggered when the predicted safe distance is less than the minimum safe distance. The Level 1 warning prompts the operator to slow down or adjust the boom posture, with specific prompts including voice broadcast and a yellow flashing prompt on the cab display. The Level 2 warning triggers an audible and visual alarm, including a buzzer sound and a red strobe light flashing, and automatically restricts boom movement, specifically by reducing boom movement speed and prohibiting further movement towards live equipment. Simultaneously, the safety boundary is highlighted in the 3D point cloud scene, and the boundary of the danger zone is marked with a red semi-transparent sphere or cube, providing the operator with an intuitive spatial location prompt and outputting a graded warning signal. This warning signal will be fed back to the residual trend prediction submodule in step S13311 and the working condition zoning judgment stage in step S13312, temporarily adjusting the residual tolerance threshold and zoning boundary parameters under high-risk working conditions, forming a dynamic closed loop between warning response and data acquisition.

[0097] Step S144: Intelligent Decision Support

[0098] Based on a dynamic safety zone model and operational task objectives, the system automatically recommends safe operational paths. The safe operational path generation method employs the A-Star search algorithm, starting from the current boom position, ending at the target operational position, using the safety zone and warning zone in the safety zone model as passable areas, and the core restricted area as obstacles, to search for the shortest path that satisfies safety distance constraints. Furthermore, it performs safety procedure compliance checks on key nodes during the operation process, specifically checking whether the distance between each point on the path and live equipment meets the safety distance requirements stipulated in the electrical industry safety work regulations; whether there are any risk points on the path that could lead to collisions between the boom and equipment; and whether the operational posture complies with the operating procedures. This ensures compliance and safety throughout the entire operation process, outputting operational path planning and safety compliance recommendations. These recommendations are optimized by incorporating historical warning records from step S143, achieving an intelligent decision-making upgrade from passive warning to proactive avoidance.

[0099] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. A dynamic modeling method for safe operation areas in substations based on electric field-point cloud coupling, characterized in that, Includes the following steps: Construct a spatiotemporal synchronous acquisition and error pre-compensation system for multi-source heterogeneous data: Integrate and install lidar, broadband electric field sensor, inertial measurement unit and temperature compensation module at the end of the boom of special vehicle, trigger synchronous acquisition based on GPS disciplined clock source, and obtain raw three-dimensional point cloud data, spatial electric field gradient data, boom attitude data and ambient temperature data; The collected raw data is preprocessed and features are extracted to establish a two-way coupling relationship between 3D point cloud data and spatial electric field gradient data. Electric field-point cloud coupled modeling and dynamic safety zone construction: The bidirectionally calibrated 3D point cloud-spatial electric field gradient correlation data is input into the joint error compensation module. This module constructs a closed-loop collaborative suppression architecture with temperature-phase compensation as the prior constraint of extended Kalman filter, realizing the decoupling and iterative correction of the phase distortion of broadband electric field sensor signal and the distance of boom spatial mapping deviation, and outputs the boom spatial position signal and spatial electric field gradient information after joint correction. Based on the ambient temperature threshold and phase compensation residual, the fusion weights of the three-dimensional point cloud positioning data and the spatial electric field gradient positioning data are dynamically allocated, and the two are weighted and fused to reconstruct the real-time spatial position of the boom end; based on the reconstructed boom spatial position and the spatial electric field gradient intensity distribution, three-layer working areas of core restricted area, warning area and safe area are dynamically generated. Based on the constructed dynamic security zone model, intelligent monitoring, visual early warning, and decision support are provided.

2. The method according to claim 1, characterized in that, The system for constructing a spatiotemporal synchronous acquisition and error pre-compensation system for multi-source heterogeneous data also includes: A three-dimensional rotation matrix is ​​constructed using the real-time pitch, roll, and yaw angles output by the inertial measurement unit to transform the spatial electric field gradient direction acquired by the broadband electric field sensor to align with the three-dimensional point cloud coordinate system. The ambient temperature data is acquired in real time through the temperature compensation module, and the pre-stored temperature-phase response characteristic curve of the broadband electric field sensor is called. The corresponding phase correction coefficient is matched according to the current ambient temperature to perform phase shift compensation on the original output signal of the broadband electric field sensor. The temperature-phase response characteristic curve is obtained through laboratory calibration. The calibration method is as follows: place the broadband electric field sensor in a temperature control chamber, set multiple calibration points within a preset temperature range, measure the phase shift of the sensor output signal at each calibration point, fit the correspondence between temperature and phase shift into a continuous curve and store it.

3. The method according to claim 2, characterized in that, The joint error compensation module implements decoupling and iterative correction, specifically including: Dynamic compensation for signal phase distortion based on temperature-phase characteristic curve: An adaptive filtering algorithm is used to track the signal phase deviation of a broadband electric field sensor in real time. The temperature-phase compensation coefficient is used as the initial constraint of the adaptive filter to construct a three-dimensional compensation model of temperature-phase-time and output phase information that has been adaptively corrected by ambient temperature. Extended Kalman Filter Iterative Correction with Phase Correction as State Update Constraint: The corrected phase information is used as the prior constraint for the state update of the extended Kalman filter. A multibody kinematic model of the vehicle boom is constructed. In the filtering update stage, the phase correction is introduced into the observation equation in the form of a time-varying bias term to perform constraint correction on the recursive result and output the boom spatial position signal after joint correction.

4. The method according to claim 3, characterized in that, The dynamic compensation for signal phase distortion based on the temperature-phase characteristic curve also includes: Segmented adaptive compensation parameter adjustment based on temperature change rate: In the temperature-phase-time three-dimensional compensation model, the ambient temperature change rate for two consecutive sampling periods is calculated; when the absolute value of the temperature change rate exceeds the preset change rate threshold, it is determined that the ambient temperature is in a state of drastic fluctuation, and the system automatically switches to the fast response compensation mode. In the fast response compensation mode, the current temperature change rate is used as input, and the pre-stored temperature change rate-phase response correction table is called to perform a secondary dynamic adjustment on the phase correction coefficient.

5. The method according to claim 3, characterized in that, The dynamic compensation for signal phase distortion based on the temperature-phase characteristic curve also includes: Self-evaluation of compensation effect and iterative optimization of correction coefficient based on phase compensation residual: After each phase compensation is completed, the residual value between the compensated signal phase and the theoretical undistorted phase is calculated; When the residual value exceeds the preset residual tolerance threshold for multiple consecutive sampling periods, online iterative optimization of the correction coefficient is triggered. The iterative optimization adopts the gradient descent algorithm, which uses the partial derivative of the current residual value with respect to each feature point of the temperature-phase response characteristic curve as the gradient direction to fine-tune the feature point parameters and generate an updated characteristic curve.

6. The method according to claim 1, characterized in that, The dynamic allocation of fusion weights for 3D point cloud positioning data and spatial electric field gradient positioning data specifically includes: using ambient temperature and phase compensation residual as dual-factor judgment criteria; when the ambient temperature is lower than a preset threshold and the phase compensation residual exceeds a set range, the system determines that it has entered a low-temperature signal degradation condition, automatically reduces the weight ratio of the broadband electric field sensor in the fusion calculation, and at the same time increases the prediction weight based on the inertial measurement unit and kinematic model; when the ambient temperature rises and the phase compensation residual returns to the normal range, the fusion weight of the broadband electric field sensor is gradually restored.

7. The method according to claim 6, characterized in that, The dynamic allocation of fusion weights also includes a weight pre-adjustment mechanism based on residual trend prediction: a residual trend prediction submodule is constructed. This submodule uses a linear regression method to fit a residual change trend line based on the phase compensation residual values ​​of the most recent sampling periods, and predicts the residual change direction of future sampling periods based on the slope of the trend line. When the prediction result shows that the residual will exceed the residual tolerance threshold in the next period, the system reduces the fusion weight of the broadband electric field sensor one sampling period in advance, and allocates the weight adjustment amount proportionally according to the prediction confidence.

8. The method according to claim 6, characterized in that, The dynamic allocation of fusion weights also includes a weight smoothing transition strategy based on operating condition partitions: the two-dimensional plane formed by ambient temperature and phase compensation residuals is divided into four operating condition partitions: low temperature high residual zone, low temperature low residual zone, normal temperature high residual zone, and normal temperature low residual zone; each partition corresponds to a set of preset weight change rate parameters, including the rate of increase when weights increase and the rate of decrease when weights decrease. When the system detects that the current operating point crosses the boundary from one partition to another, it makes a gradual adjustment according to the change rate of the target partition, and the weight value within the gradual adjustment period is transitioned by linear interpolation.

9. The method according to claim 1, characterized in that, The intelligent monitoring, visual early warning, and decision support include: Construction of a safe distance prediction model for boom motion trend: Using the boom kinematic model, combined with the boom's historical motion trajectory and current motion state, a trajectory prediction algorithm based on extended Kalman filter is used to predict the position trajectory of the boom end within the future time window; Spatial collision detection is performed between the predicted trajectory and the 3D point cloud model of the substation's energized equipment; the dynamic safe distance between the boom end and the live conductor within the predicted time window is calculated in real time; tiered early warning and risk alerts are provided. When the predicted safe distance is less than the first preset threshold, a level one warning is triggered, prompting the operator to slow down or adjust the boom posture; when the predicted safe distance is less than the second preset threshold, a level two warning is triggered, triggering an audible and visual alarm and automatically restricting boom movement.

10. The method according to claim 9, characterized in that, The warning signals triggered by the tiered early warning and risk alert are fed back to the step of dynamically allocating fusion weights, specifically as follows: The early warning signal is fed back to the weight pre-adjustment mechanism based on residual trend prediction and the weight smooth transition strategy based on working condition partitioning. Under high-risk operating conditions, the residual tolerance threshold and operating condition boundary parameters are temporarily adjusted to form a dynamic closed loop of early warning response and data collection and fusion strategy adjustment.

Images