Method, device and medium for early warning of safety risk of construction machinery of transformer substation based on beidou positioning and three-dimensional data

Abstract

This invention discloses a method, equipment, and medium for early warning of safety risks of substation construction machinery based on BeiDou positioning and 3D data. It adaptively corrects the observation covariance based on the observation integrity coefficient and generates a pose uncertainty description object and a positioning status identifier. The nominal geometric occupant and the uncertainty ellipsoid are expanded and combined to form a mechanically reliable occupant. A dynamic risk distance field for energized equipment is generated. The minimum predicted safety gap between the predicted reliable occupant and the dynamic risk distance field is obtained. Risk warnings are given based on the minimum predicted safety gap, the gap change trend, the positioning status, and the status of the energized equipment. This invention avoids the problem of missed warnings caused by existing solutions relying on a single nominal position to determine the safety distance when positioning quality deteriorates. It provides a reliable basis for accurate safety determination under different equipment and operating conditions, realizing a shift from "post-event alarm" to "predictive constraints, hierarchical intervention, and safety closed-loop control."

Description

Technical Field

[0001] This invention belongs to the field of safety early warning technology, specifically relating to a method, equipment and medium for early warning of safety risks of substation construction machinery based on Beidou positioning and three-dimensional data. Background Technology

[0002] In the construction, expansion and routine maintenance of substations, the safety of large construction machinery (such as cranes and excavators) operating in energized environments has always been a core concern for the power industry.

[0003] Traditional substation operation safety management mainly relies on manual on-site monitoring. Monitors visually observe the relative position between the crane boom and the energized equipment, and use rangefinders or telescopic insulating rods to intermittently measure spatial distances. In this mode, the movement of construction machinery is entirely judged by humans based on experience and procedures. At the same time, static protective facilities such as ground fences, safety warning signs, and infrared beam alarms are usually set up on site to define the physical boundary between the construction area and the energized operation area.

[0004] With the development of satellite navigation technology, positioning technology based on the Global Navigation Satellite System (GNSS) has been introduced into substation construction management. In existing technologies, a single-point positioning GNSS receiver is typically installed on top of construction machinery to obtain the latitude and longitude coordinates of the equipment in real time. The location information of the vehicle is obtained using satellite signals and displayed on a two-dimensional map background, thereby achieving basic trajectory tracking of the vehicle's position on site. In some solutions, a loosely coupled mode is used to simply fuse GNSS data with inertial sensors to perform short-term estimation and positioning when satellite signals are weak.

[0005] To obtain a digital representation of substations, 3D laser scanning technology (LiDAR) is widely used in scene modeling. Typically, a station-mounted laser scanner is used to perform a full scan of the substation's towers, structures, conductors, and transformers, acquiring raw 3D point cloud data of the scene. Subsequently, through point cloud denoising, inter-station registration, and manual feature extraction, a static 3D model of the substation is established, such as a Building Information Modeling (BIM) model or a refined surface mesh model. This model reflects the geometry and spatial arrangement of the power facilities within the substation.

[0006] Regarding safety warning algorithms, existing solutions are mostly based on simple geometric bounding boxes for collision detection. This involves creating simplified cylindrical or cuboid bounding box models for construction machinery and live equipment, and calculating the Euclidean distance between the two geometric objects. The warning logic typically presets a fixed safety distance threshold (e.g., 5 meters or 8 meters). Once the calculated geometric distance is less than this threshold, the system triggers an alarm signal, such as through on-site loudspeaker broadcasts, flashing warning lights in the driver's cab, or pushing alarm information to the monitoring backend via the network.

[0007] Some existing technologies have incorporated deep learning point cloud algorithms to semantically annotate the collected scene point clouds. Based on the geometric features of the point clouds (such as curvature and normal vectors), they are classified into different categories of power semantic tags, such as towers, conductors, and insulators. Meanwhile, some systems are attempting to use database mapping methods to logically associate substation ledger information, voltage levels, and other attribute data with corresponding spatial geometric coordinates, thus initially realizing a "space + attribute" digital management logic.

[0008] To address the complex and ever-changing operating environment of substations, some existing solutions have introduced auxiliary environmental monitoring methods. For example, weather stations or monocular video surveillance cameras are installed in the construction area to acquire real-time video footage of the temperature, humidity, visibility, and cranes. These technologies use manual identification of video streams or simple background modeling algorithms to help determine whether crane operations exceed limits. At the communication level, existing technologies mostly rely on wireless communication networks such as 4G / 5G or ZigBee to aggregate data from various sensing terminals on-site to a unified monitoring screen or management platform for centralized display.

[0009] Most existing technologies focus on the static reconstruction of substation scenarios, with modeling typically completed before or during construction. These static models (such as offline BIM or static point clouds) cannot reflect the dynamic flow of objects and space occupancy at the construction site in real time. When large machinery like cranes enter the site, the lack of a real-time dynamic mapping mechanism makes it difficult for the system to promptly identify newly appearing temporary cables, scaffolding, or terrain changes caused by excavation. This "time lag" between the static foundation and dynamic operations results in significant blind spots for safety warning systems when facing dynamic obstacles, failing to provide complete spatial references for the operating machinery.

[0010] In the unique electromagnetic environment of substations, traditional GNSS positioning technology faces severe challenges. Due to the dense network of metal towers, gantry cranes, and complex lightning rod systems within substations, satellite signals are highly susceptible to multipath effects during propagation, leading to positioning drifts on the order of several meters. Existing technologies, such as single-point positioning or loosely coupled GNSS / IMU modes, lack underlying constraints for absolute pose. In scenarios with partial signal obstruction or loss of lock (e.g., when a crane is operating beneath equipment structures), positioning errors accumulate rapidly. This fluctuation in accuracy directly impacts the accuracy of determining the distance between the crane boom tip and live parts, failing to meet the stringent centimeter-level monitoring requirements of the power industry for near-electrical operations.

[0011] Existing voxel modeling techniques mostly employ a uniform partitioning method with a fixed resolution. When dealing with intricate electrical components such as conductors, insulators, or fittings in substations, reducing voxel sizes to ensure modeling accuracy leads to an exponential increase in the number of voxel units, resulting in massive data redundancy and consuming significant system computing resources. Conversely, using larger voxel sizes to improve rendering efficiency results in blurred edge features of critical electrical components, making it impossible to accurately describe the spatial contours of complex fittings. This makes it difficult for existing solutions to provide sufficient fine-grained local details to support safe distance calculations while ensuring smooth operation in large-scale scenes.

[0012] Most current safety early warning systems employ warning logic based on fixed physical distances, failing to adequately consider the complex operating conditions of substations. Existing technologies typically set a static safety threshold (e.g., 5 meters), often designed based on ideal air insulation conditions. However, in actual construction, the breakdown voltage of air varies significantly with factors such as ambient humidity, smog concentration, and rainfall intensity. Traditional solutions, lacking coupling between environmental perception factors and distance-based early warning algorithms, cannot automatically adjust the safety threshold under severe weather conditions, while potentially generating unnecessary and frequent false alarms in favorable environments, thus lacking intelligent and flexible decision-making capabilities.

[0013] Existing technologies exhibit a significant "information silo" phenomenon in data organization. While they can acquire geometric spatial data of a scene, this data is often an isolated set of coordinates, lacking deep correlation with power infrastructure management operations such as risk lists, project schedules, and quality responsibility areas. Because voxel units or point cloud clusters lack corresponding business tags, when the system detects a risk, managers struggle to immediately obtain the corresponding voltage level, construction personnel qualifications, or historical inspection records for that area. This disconnect between spatial data and management attributes limits the early warning system to a basic alert function, failing to support a fully traceable, intelligent digital twin management closed loop.

[0014] In mobile measurement and dynamic monitoring, existing technologies often overlook the timestamp alignment issue between sensors. Due to the inconsistent sampling frequencies of LiDAR, GNSS, and IMU hardware, and the lack of a precise hardware pulse triggering mechanism, the acquired multi-source data exhibits slight time-axis offsets. When a crane boom rotates rapidly or a mobile measurement platform moves at high speed, this time lag can cause "ghosting" or stretching distortion in the mapped 3D spatial point cloud, leading to false spatial occupancy judgments. This underlying spatiotemporal heterogeneity problem results in errors in the geometric topology of the early warning system in dynamic scenes, reducing the reliability of collision detection. Summary of the Invention

[0015] The purpose of this invention is to provide a method, equipment, and medium for early warning of safety risks of substation construction machinery based on BeiDou positioning and three-dimensional data, so as to solve the above-mentioned technical problems in the monitoring of safety risks of substation construction machinery.

[0016] This invention is achieved through the following technical solution: A safety risk early warning method for substation construction machinery based on BeiDou positioning and 3D data includes the following steps: S1. Acquire observation data from BeiDou RTK, lidar, inertial measurement unit, and mechanical attitude sensor to determine the posterior fusion state estimate of the construction machinery state vector; obtain the observation integrity coefficient corresponding to each observation source based on the observation residual, effective observation ratio, continuous anomaly duration, and cross-source consistency deviation; adaptively correct the observation covariance based on the observation integrity coefficient, and generate the pose uncertainty description object and positioning status identifier. S2. Based on the nominal pose of the mechanical reference and the structural parameters and real-time motion parameters of the construction machinery, establish the mechanical kinematic chain of the construction machinery and the nominal geometric occupancy of the mechanical components; propagate the pose uncertainty description object along the mechanical kinematic chain to each mechanical component to generate the corresponding spatial uncertainty ellipsoid; expand and combine the nominal geometric occupancy and the uncertainty ellipsoid to form a reliable mechanical occupancy. S3. Establish a basic safety distance field based on the actual geometric surface of the energized equipment, and generate a dynamic risk distance field for the energized equipment based on the environmental correction, directional correction and state correction corresponding to the energized equipment. S4. Obtain the minimum predicted safety gap between the predicted credible occupant and the dynamic risk distance field, and conduct risk warning based on the minimum predicted safety gap, gap change trend, positioning status and energized equipment status.

[0017] In some embodiments, the positioning status is divided into normal status, degraded status and untrusted status according to the integrity coefficient of each observation source and the status identifier. When it is a degraded status, step S2 is triggered to expand the mechanically trustworthy occupying body. When it is an untrusted status, step S2 is triggered to generate a conservatively reachable occupying body.

[0018] In some embodiments, when the mechanical positioning state is an untrusted positioning state, the trusted occupant at the last normal positioning moment is used. Based on this, combined with the maximum permissible speed of the machinery For, maximum permissible acceleration and duration of abnormality Generate a conservatively accessible occupied body , represented as: ; in, This represents a space dilation combination operation. This represents the safety margin corresponding to mechanical structure error and control response error. This represents a spatial expansion volume generated by the expansion distance.

[0019] In some embodiments, in step S2, the key point coordinates of the mechanical components are calculated based on the coordinate transformation matrix corresponding to the mechanical components in the mechanical kinematic chain, and the nominal geometric occupancy of each mechanical component is established based on the key point coordinates and the shape parameters of the corresponding components.

[0020] In some embodiments, in step S2, a kinematic Jacobian matrix is ​​constructed relative to the coordinates of key points of the mechanical component with respect to the mechanical reference position, mechanical reference posture, joint angle, and extension length. The spatial position covariance matrix of the mechanical component is obtained based on the mechanical reference pose covariance matrix and the mechanical state covariance matrix corresponding to the mechanical joint angle, extension length and the attitude parameters of the suspended object. The spatial uncertainty ellipsoid of the mechanical component is then constructed based on the spatial position covariance matrix.

[0021] In some embodiments, in step S3, one or more environmental monitoring data from humidity, rainfall, visibility, temperature, wind speed, and wind direction are acquired, and an environmental correction amount is determined using a duration confirmation rule. , ; Approaching speed at - The projection of the wind speed vector in the direction that brings the suspended conductor, flexible lead wire, or suspended object closer together is used as the mechanical proximity component, and the projection of the wind speed vector in the direction that brings the suspended conductor, flexible lead wire, or suspended object closer together is used as the wind-induced proximity component to generate the directional correction amount. ; To point to a safe area away from electrical equipment; No. Dynamic safety distance of a live device at surface position x and time t Represented as: ; in, , For the first The basic safety distance for each device , This is the equipment status correction amount caused by temporary power supply, special operating mode, unconfirmed status, or communication failure relative to the normal energized reference state; The dynamic risk distance field of the j-th charged device Represented as: ; For the first A geometric set of hazards for a single piece of electrical equipment; The minimum risk distance field is taken for all currently energized or equipment that needs to be treated as energized, to generate a global dynamic risk distance field. .

[0022] In some embodiments, in step S4, based on the posterior fusion state estimate, the mechanical motion state, and the control commands, the construction machinery is predicted to operate within the prediction time window. Action parameters at each discrete prediction time; Based on the mechanical kinematics chain, the predicted motion parameters and pose uncertainty description objects are propagated to the mechanical components to obtain the predicted reliable occupant of the mechanical components at the prediction time. When in a normal location state, the predicted credible occupant is used as the risk assessment object; when in a degraded location state, the uncertainty range corresponding to the predicted credible occupant is increased and then used as the risk assessment object; when in an unreliable location state, the predicted conservatively reachable occupant is used as the risk assessment object.

[0023] In some embodiments, the method further includes the following after step S4: Obtain the control sensitivity of the mechanical control component to the minimum predicted safety clearance. ,like If <0, mark the current k-th control component as a dangerous direction action. If the value is >0, mark it as a hazard avoidance action; When the risk level is safe, maintain the original control commands and continue monitoring; when the risk level is warning, output the nearest equipment, the nearest mechanical component, the remaining safety clearance, the estimated intrusion time and the suggested avoidance direction to the driver's cab, on-site audio-visual devices and the back-end terminal, and increase the prediction update frequency; when the risk level is high-risk, project the control commands to the permissible control set that meets the predicted safety constraints to obtain constraint control commands. And / or, when the current trusted occupying body or conservatively reachable occupying body has entered the dynamic safety boundary, stop the dangerous direction action and lock the relevant actuators; when there is a safe avoidance direction, perform deceleration or reverse avoidance action; when the safe avoidance direction cannot be determined, perform emergency shutdown.

[0024] On the other hand, the present invention also provides an electronic device, comprising: Processor; and Memory for storing the executable instructions of the processor; The processor is configured to execute the substation construction machinery safety risk early warning method based on BeiDou positioning and three-dimensional data by executing the executable instructions.

[0025] On the other hand, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the aforementioned method for early warning of safety risks of substation construction machinery based on BeiDou positioning and three-dimensional data.

[0026] Compared with the prior art, the present invention has the following advantages and beneficial effects: This invention performs integrity assessments on BeiDou RTK, LiDAR, IMU, and mechanical attitude observations, converting observation anomalies, signal obstruction, registration degradation, and time asynchrony into pose uncertainty and positioning status, rather than simply outputting a single position coordinate. This allows changes in positioning quality to directly affect subsequent mechanical safety space calculations, reducing the risk of making proximity judgments based on ideal positioning status when observation quality deteriorates. It also avoids the problem of missed reports caused by existing schemes that still rely on a single nominal position to determine safety distances when positioning quality deteriorates.

[0027] This invention propagates positional uncertainty, joint errors, structural tolerances, and the state of the suspended load along the mechanical kinematic chain to the boom, hook, suspended load, and other hazardous components, forming a credible or conservatively reachable occupant. Compared to methods that rely solely on vehicle positioning points, basket centers, or nominal trajectories to determine safe distances, this invention reflects the spatial range that the machinery may reach during periods of amplified errors at the end of long-arm components and positioning anomalies. The safety assessment object is no longer a single BeiDou positioning point, machinery center point, or nominal three-dimensional trajectory of the construction machinery, but rather the overall credible occupant space including positioning errors and mechanical attitude errors. Furthermore, it generates a conservatively reachable occupant when positioning is downgraded or becomes unreliable, thus avoiding safety blind spots during periods of positioning anomalies.

[0028] This invention is based on the actual geometric surface of the live equipment and combines the equipment status, the continuous environmental status, the mechanical approach direction and the wind-induced approach direction to construct a dynamic risk distance field. This allows the safety boundary to be updated with the local geometry of the live equipment, the operating conditions and the environmental status, providing a reliable basis for accurate safety judgment under different equipment and different operating conditions. When the safety boundary version changes, the system re-executes the prediction judgment to avoid using outdated boundaries to output risk conclusions.

[0029] This invention enables the system to identify future risk trends when the boundary is not currently intruded upon by calculating the minimum safe gap between a trusted occupant and a dynamic risk boundary within a prediction time window. Different control strategies are implemented for different risk sources, and the control commands are not directly triggered by a single distance threshold, but are determined based on the mechanically trusted occupant, the dynamic safety boundary, the predicted intrusion trend, and the location status. This achieves a shift from "post-event alarm" to "predictive constraints, tiered intervention, and closed-loop safety control." Attached Figure Description

[0030] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 This is a flowchart of a safety risk early warning method for substation construction machinery according to an embodiment of the present invention. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of this application clearer, specific embodiments of this application will be described in further detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely for explaining this application and not for limiting it. It should also be noted that, for ease of description, only the parts relevant to this application are shown in the drawings, not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe operations (or steps) as sequential processes, many of these operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but may also have additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0033] The technical terms and explanations involved in the embodiments of this invention are as follows: BeiDou RTK: Real-time dynamic carrier phase differential positioning technology based on the BeiDou satellite navigation system, used to obtain high-precision absolute position and positioning solution status of construction machinery.

[0034] Multi-source observation integrity: A parameter used to characterize the reliability of observation results from BeiDou RTK, LiDAR, IMU, and mechanical attitude sensors at the current moment, which is at least related to observation residuals, observation duration, and number of effective observations.

[0035] Position uncertainty description object: Error range data object corresponding to the nominal position of the machine, including at least position uncertainty, attitude uncertainty and their related parameters, used to describe the spatial range in which the actual position of the machine may deviate from the nominal position.

[0036] Mechanical kinematic chain: A coordinate transformation relationship established according to the connection relationship of various components of construction machinery. It is used to calculate the position and attitude of the boom, hook and the suspended object based on parameters such as the mechanical reference posture, rotation angle, amplitude angle and telescopic length.

[0037] Mechanically reliable occupant: A conservative spatial occupancy range formed by the combination of the nominal geometric occupant of the machine, the positional uncertainty, the mechanical attitude error, the structural tolerance, and the swing error of the suspended object. It is used to replace a single mechanical positioning point for safety determination.

[0038] Conservatively reachable occupied body: The mechanically reachable spatial range generated based on the last reliable pose, maximum mechanical velocity, maximum acceleration, and abnormal duration when the location is unreliable.

[0039] Dynamic safety boundary: Based on the actual geometric surface of the energized equipment, the near-electric risk boundary is dynamically generated according to the equipment type, voltage level, energized state, environmental conditions and spatial orientation.

[0040] Dynamic risk distance field: a data field used to represent the remaining safety clearance from any spatial location in the substation to the dynamic safety boundary of energized equipment; when its value is less than or equal to zero, it indicates that the location has entered the dynamic safety boundary.

[0041] Predicted safety gap: The minimum spatial gap between the mechanically predicted trusted occupant and the dynamic safety boundary within a preset prediction time window.

[0042] Directional motion constraints: Based on the impact of mechanical motion on the predicted safety clearance, speed limits, prohibitions, or lockouts are imposed on dangerous directional motions such as slewing, luffing, telescoping, hoisting, or descent, while retaining control methods for hazard avoidance directional motions.

[0043] The embodiments of the present invention relate to the near-electric safety risk early warning management of mechanical equipment such as cranes, aerial work platforms, and excavators during the construction, expansion, maintenance, and live-line work of substations.

[0044] Research has revealed that existing technologies typically involve directly mapping positioning results, mechanical trajectories, or video footage onto a 3D model, and then issuing alarms based on fixed areas or distances. This method fails to distinguish between the "nominal estimated value" of the mechanical pose and the "actually possible space occupied" by each mechanical component. Under conditions such as BeiDou signal obstruction, multipath propagation of metal structures, point cloud registration degradation, mechanical joint errors, or swaying of suspended objects, it is prone to underestimating safety clearances or providing unstable warning results.

[0045] While existing multi-source positioning and mapping schemes can improve trajectory accuracy through GNSS, IMU, and LiDAR, they typically aim to improve map or positioning results, failing to convert the integrity, residuals, and continuous anomalies of each observation source into pose uncertainties that can be directly used in near-electric safety determination. When positioning quality deteriorates, the system may still make decisions based on a single nominal position or static safety boundary, making it difficult to avoid missing spatial risks associated with mechanical end effectors, hooks, and suspended loads.

[0046] Meanwhile, existing dynamic safety distances are typically adjusted based on voltage levels, weather, or personnel status, without fully considering the actual geometry of energized equipment, the direction of mechanical approach, the wind-induced swaying direction of conductors or suspended loads, and changes in equipment condition. In situations where machinery rapidly approaches energized components, or where safety boundaries are updated during the assessment period, static zones or uniform radius expansions are insufficient to provide safety boundaries that match actual operating conditions.

[0047] Furthermore, most existing warnings only appear after the machinery is approaching or entering a danger zone, lacking a mechanism for identifying and controlling the direction of movement based on predicted safety clearances. Indiscriminately stopping machinery can also cause instability of the load or loss of available safety maneuvers. Therefore, it is necessary to identify the specific actuators and directions of movement that reduce the safety clearance, while ensuring safety.

[0048] In some embodiments of the present invention, the method for early warning of safety risks of substation construction machinery based on BeiDou positioning and three-dimensional data adopts the following steps: Acquire data from BeiDou RTK, LiDAR, IMU, mechanical attitude, electrical equipment status, and environmental observations; Completeness assessment is performed on the residuals, effective proportions, continuous anomaly times and cross-source consistency of each observation source, and the mechanical nominal pose, state covariance, pose uncertainty description object and positioning status are output. Based on the mechanical kinematics chain, the pose uncertainty, joint error, structural tolerance and the state of the suspended object are propagated to each mechanical component to generate a reliable occupying body or a conservatively reachable occupying body when the positioning is unreliable; A dynamic risk distance field for the energized equipment is generated based on its actual geometric surface, voltage level, energized state, environmental conditions, mechanical approach direction, and wind-induced approach direction. Calculate the minimum safety gap between the predicted credible occupant and the dynamic risk distance field within the prediction time window; Based on the sensitivity of each control component to the minimum predicted safety clearance, the dangerous direction and the avoidance direction are determined, and speed limit, directional constraint, locking, shutdown or avoidance control commands are output. Receive feedback from the actuator and recalculate the predicted safety gap to form a safety closed loop.

[0049] In some embodiments, a system for executing the above-described method for early warning of safety risks of substation construction machinery based on BeiDou positioning and three-dimensional data includes: a multi-source observation integrity assessment unit, a mechanical credible occupant generation unit, a dynamic risk boundary unit for energized equipment, a predicted safety gap determination unit, a control sensitivity and permissible control set generation unit, an execution feedback verification unit, and a safety event tracing unit.

[0050] The system achieves early warning and proactive constraint of safety risks in substation construction machinery operations by continuously processing the machine position, the space occupied by each machine component, the boundary of the energized equipment, and the actions of the actuator.

[0051] Reference Figure 1 In this embodiment of the invention, BeiDou RTK, LiDAR, IMU, mechanical attitude, equipment status, and environmental monitoring data are used as inputs. First, a multi-source observation integrity assessment is performed to obtain the nominal mechanical pose, pose uncertainty, and positioning status. Then, the uncertainty is propagated along the mechanical kinematic chain to generate a credible occupant for the boom, hook, and load. Simultaneously, a dynamic safety boundary is generated based on the geometry, voltage level, energized state, and environmental conditions of the energized equipment. Within the prediction time window, the intrusion risk of the credible occupant is determined, and warnings, speed limits, action constraints, locking, or shutdown commands are output according to the risk level, forming a safety closed loop.

[0052] A 3D LiDAR, a BeiDou RTK receiver, and an IMU were deployed at the substation site for data acquisition. The LiDAR was used to acquire 3D point cloud data of the construction machinery and environment within the substation; the BeiDou RTK receiver provided global absolute positioning coordinates; and the IMU acquired the attitude angles, angular velocities, and accelerations of the construction machinery during its movement.

[0053] The Beidou RTK mobile station is installed near the top of the construction machinery to obtain the global position of the machinery. An IMU is installed on the slewing platform to acquire changes in the machinery's attitude. A slewing angle sensor is installed at the slewing mechanism to acquire the rotation angle of the construction machinery's upper body relative to the chassis; a pitch angle sensor is installed at the boom hinge to acquire the boom pitch angle; a boom length or extension sensor is installed at the telescopic boom to acquire the current boom length; a hook height sensor acquires the height of the hook or working end; and vehicle tilt angle and outrigger status sensors determine the machinery's stability. Through this deployment, real-time data on the machinery's chassis position, boom attitude, working radius, hook height, and overall machine status are obtained.

[0054] The specific operating steps are as follows: Step S1: Determine the pose uncertainty associated with the integrity assessment and safety of multi-source observations. This step is used to obtain the nominal mechanical pose of the construction machinery under a unified spatiotemporal reference, and simultaneously generate a pose uncertainty description object that can directly drive the expansion of the credible occupant volume in step S2.

[0055] Collect observation data from BeiDou RTK mobile station, lidar, inertial measurement unit and mechanical attitude sensor.

[0056] The BeiDou RTK mobile station should at least output the location of the construction machinery, the number of effective satellites, the carrier phase residual, and the pseudorange residual; the lidar should at least output the point cloud frame, the proportion of effective matching points, and the registration residual; the inertial measurement unit should at least output the three-axis acceleration, the three-axis angular velocity, the attitude increment, and the zero bias estimate; and the mechanical attitude sensor should at least output the rotation angle, amplitude angle, telescopic length, hook height, and the attitude parameters of the suspended object.

[0057] Multi-source synchronous observation frames were established using BeiDou RTK time as the reference time axis, and the BeiDou RTK coordinates, local lidar coordinates, and mechanical body coordinates were transformed to a unified station area coordinate system. For inertial observations and mechanical attitude observations that were not completely synchronized with the target time, interpolation was performed based on adjacent observation values.

[0058] Let the state vector of the construction machinery at time t be... : (1-1); in, Indicates the position of the mechanical positioning reference point. Indicates mechanical speed. Indicates mechanical attitude parameters. and These represent the zero bias of acceleration and the zero bias of angular velocity of the inertial measurement unit, respectively.

[0059] The state vector in this step Describe the overall integration state of the construction machinery, among which and The pose of the mechanical reference coordinate system relative to the station area coordinate system is jointly determined; the poses of components such as the boom, hook, and load are not directly incorporated. In step S2, calculations are performed step by step based on the mechanical reference pose, mechanical structure parameters, and real-time motion parameters.

[0060] The mechanical positioning reference point is the phase center of the Beidou antenna pre-calibrated on the mechanical body or a reference point with a fixed transformation relationship with it.

[0061] Posterior fusion state estimate obtained from multi-source synchronous observation set Furthermore, observation integrity assessments were performed on the observation data from BeiDou RTK, lidar, and inertial measurement units respectively.

[0062] Posterior fusion state estimation The process of obtaining it is as follows: Using the posterior fusion state estimation from the previous time step By combining the IMU measurements with state prediction, a prior state estimate for the current time step can be obtained. and prior state covariance ; Then, the BeiDou RTK position / velocity, LiDAR registration pose, and IMU pre-integration quantities within the same synchronous observation frame are respectively written into observation equations. ;in, For the first The observation function corresponding to the class observation, To observe the noise, its covariance is ; Equivalent joint estimation can be achieved using extended Kalman filtering, iterative weighted least squares, or factor graph optimization. (1-2); In equation (1-2), For the first Class observation at time Innovation relative to the prediction of observations from prior states (observation residuals). For observation function In the prior state The first-order Jacobian matrix at a given location is used to map small state perturbations to the observation space.

[0063] After completing the integrity assessment, the corrected observation covariance is used. Participating in joint updates yields posterior fusion state estimates. Among them, the mechanical reference position extracted from the posterior fusion state estimation. and mechanical reference attitude Together they constitute the nominal pose of the mechanical reference. .

[0064] Therefore, posterior fusion state estimation It is the posterior state vector obtained by multi-source synchronous observations within a unified station area coordinate system through prediction, residual verification, and weighted fusion; the nominal pose of the mechanical reference. It consists only of the position and attitude components in the posterior fusion state estimate.

[0065] For any observation source Based on the observation residuals of this observation source and residual covariance matrix Calculate the normalized observation residuals: (1-3).

[0066] The observation source j referred to in this step refers to an independent information source participating in the joint estimation of the mechanical reference pose. The set of observation sources J includes at least BeiDou RTK observation sources, LiDAR inter-frame or map registration observation sources, and IMU inertial estimation observation sources. When a visual odometry or other independent positioning sensor is configured, it can be added as a new observation source and processed according to the same rules. The mechanical rotation angle, amplitude angle, and extension length are used as kinematic inputs for step S2, and their measurement uncertainty is included in the mechanical state covariance. .

[0067] Further obtain the effective observation ratio of observation source j Duration of continuous abnormality and cross-source consistency deviation Cross-source consistency deviation The position and attitude differences between the independent pose of observation source j and the joint pose of other effective observation sources are used to distinguish between single observation source anomalies and the common degradation of multiple observation sources.

[0068] The above three integrity parameters , , In length of The time window is determined within the sliding time window.

[0069] set up This represents the number of observations that should be reached within the window. The effective observation ratio is the number of observations that pass the checks for timestamp, range, format, and residual threshold. Calculate using the following formula: (1-4); set up For abnormal indication quantities; when When the chi-square threshold corresponding to its degrees of freedom is exceeded, the time difference exceeds the synchronization tolerance, or the observation is invalid. ,otherwise ; Then the duration of continuous abnormality according to The process is repeated until observations return to normal, at which point the value is reset to zero.

[0070] The position is obtained by solving for the observation source j separately. and attitude rotation matrix After removing this observation source, the solution is obtained by jointly solving the problem using the remaining valid observation sources. , Then the cross-source consistency deviation is: (1-5); in, This means mapping the attitude difference rotation matrix to a rotation vector. , These are the position consistency scale and attitude consistency scale obtained from sensor calibration data, respectively. and The joint estimation is obtained by leaving one observation source unchanged. That is, the state prediction model, coordinate system and solution algorithm are kept unchanged, only the observation source j is removed from the current synchronous observation frame, and then the joint pose estimation is re-executed using the remaining observation sources that have passed the validity check.

[0071] thus, and All can be directly calculated from synchronous observation frames, residual test results, and independent / joint pose solutions.

[0072] Based on sensor calibration data, a completeness mapping function corresponding to different observation sources is pre-established. The observation integrity coefficient is determined according to the following formula. and the corrected observation covariance An implementable integrity mapping function for: (1-6); in, The smaller the value of , the lower the credible contribution of the observation source to the current pose; and These are the original observation covariance and the corrected observation covariance, respectively. is the preset minimum integrity coefficient for the j-th type of observation source.

[0073] When only one observation source experiences an increase in residuals, a decrease in the proportion of effective observations, or an anomaly in cross-source consistency, the system only expands the observation covariance of that observation source. When multiple observation sources are simultaneously abnormal, the system expands the corresponding observation covariance according to their respective integrity coefficients, avoiding the application of a single uniform reduction coefficient to all observation sources.

[0074] In some embodiments, the following may be adopted: (1-7); in, ; To normalize the residual threshold, To minimize the effective observation ratio, To allow for continuous abnormal duration, As a cross-source consistency threshold, , , , The sensitivity coefficient is non-negative. This is the preset minimum integrity coefficient.

[0075] The aforementioned thresholds and sensitivity coefficients were determined through sensor calibration tests and statistical analysis of normal, occluded, registration degradation, and communication delay samples, ensuring that the normal samples... When the value is close to 1 and the degree of abnormality increases Monotonically decreasing. This expression is only a preferred implementation; other piecewise linear functions or lookup tables that satisfy the same monotonic relationship can also be applied.

[0076] Preset minimum integrity coefficient Based on the maximum permissible covariance expansion factor Determined, that is ; Determined through verification tests at the mechanically permissible speed, control response time, and minimum safety confidence level.

[0077] when achieve And the abnormality lasted for more than At that time, instead of continuing to use infinitely expanding covariance to replace fault handling, the corresponding observation source is judged as failed, and the downgrade or untrusted state is determined based on the remaining observation sources.

[0078] Based on the corrected covariance of each observation, the system obtains the posterior fusion state estimate. The corresponding state covariance matrix And generate a pose uncertainty description object. .

[0079] It should include at least the position uncertainty, attitude uncertainty, correlation between position and attitude of the mechanical positioning reference point in three spatial directions, and integrity identifiers of each observation source. This object is not used for map accuracy evaluation, but for step S2 to calculate the spatial covariance of each mechanical component and the expansion range of the credible occupant.

[0080] When using the information form of extended Kalman filtering, the state covariance matrix It can be obtained by the following formula: (1-8); in, For a moment The set of observation sources that have passed format and time validity checks; Let be the prior state error covariance at time t, and its inverse matrix is ​​the prior information matrix.

[0081] When using factor graph optimization, The result is obtained by inverting the Hessian matrix or its approximate matrix at the optimal state of the objective function. Both reflect the influence of the corrected observation covariance on the joint pose uncertainty.

[0082] Position uncertainty describes the object It can be represented as: (1-9); in, , The selection matrix for choosing the mechanical reference position and attitude components from the posterior fusion state estimate; , respectively by The position-attitude cross-correlation matrix is ​​obtained by taking the square root of the diagonal elements of the position and attitude sub-blocks; it is defined as follows: ,in For position-attitude cross covariance blocks, and These are diagonal matrices composed of the standard deviations of position and attitude, respectively. This indicates whether the observation source is normal, abnormal, or malfunctioning.

[0083] To avoid directly comparing the covariance matrix with the scalar threshold, let's denote... , They are respectively The position and attitude sub-blocks are defined, and the maximum position uncertainty is defined. Maximum attitude uncertainty ,in This represents the largest eigenvalue of the matrix.

[0084] Based on the integrity coefficient of each observation source Status indicators , and The location status is divided into normal, degraded, and untrusted states: 1) All key observation sources meet the integrity condition, and , When this occurs, it is considered a normal state; 2) At least one key observation source is anomalous, but the remaining valid observations still maintain continuous pose, and , When this happens, it is determined to be in a downgraded state; 3) Multiple key observation sources continue to show abnormalities or fail, or , If at least one of the following conditions is met, the state is determined to be untrustworthy.

[0085] Degraded state trigger step S2 expands the mechanically trusted occupying body. Untrusted state trigger step S2 generates a conservatively reachable occupying body.

[0086] Conventional threshold , and safety limit , The sensor calibration, permissible mechanical speed, control response time, and safety margin for near-electric work are predetermined, and , .

[0087] This step outputs the posterior fusion state estimation sequence. Mechanical reference nominal pose sequence Object of pose uncertainty description Integrity coefficient of each observation source The system includes a positioning status identifier and a unified time index. The nominal pose of the mechanical reference and the object describing the pose uncertainty serve as inputs for the propagation uncertainty along the mechanical kinematic chain in step S2. The velocity component in the posterior fusion state estimation serves as an auxiliary input for the approach direction verification in step S3 and the short-time prediction in step S4. The positioning status identifier is used in step S2 to select a credible or conservatively reachable occupier and in step S5 to perform positioning degradation protection.

[0088] Step S2: Generation of a mechanically reliable occupant based on pose uncertainty propagation This step uses the nominal pose sequence of the mechanical reference output in step S1. The system takes the pose uncertainty description object U(t), the positioning status identifier, and the unified time index as inputs, and reads the structural parameters and real-time motion parameters of the construction machinery to generate the nominal geometric occupants, reliable occupants, and conservatively reachable occupants when the positioning is unreliable at each time point.

[0089] The structural parameters of the construction machinery include at least: the fixed offset of the machinery's rotation center relative to the machinery's positioning reference point, the nominal length of each section of the boom, the connection relationship of each joint, the connection relationship of the hook relative to the end of the boom, the external dimensions of the load, and the preset mechanical structural errors. These structural parameters can be pre-stored in a machinery parameter library and retrieved according to the machinery model or the task at hand.

[0090] The real-time motion parameters of the construction machinery include at least the slewing angle, amplitude angle, telescopic length, hook height, and the attitude parameters of the suspended object.

[0091] First, based on the nominal pose of the mechanical reference and the real-time motion parameters, the mechanical kinematics chain of the construction machinery at time t is established.

[0092] The kinematic chain is a transformation relationship formed by sequentially multiplying the rigid body transformations from the mechanical reference coordinate system to the coordinate systems of each component, according to the actual connection sequence of the mechanical components. Taking a truck crane as an example, the mechanical reference coordinate system sequentially connects the slewing base, luffing hinge point, each section of the telescopic boom, the boom end, the hook, and the load. The slewing angle corresponds to rotation about the vertical axis, the luffing angle corresponds to rotation about the luffing hinge axis, the telescopic length corresponds to translation along the boom axis, the hook height corresponds to translation along the wire rope direction, and the load attitude parameters correspond to the attitude transformation from the hook coordinate system to the load coordinate system. Excavators or aerial work platforms can establish similar chains according to the connection sequence of the slewing platform, boom, telescopic boom, and end effector.

[0093] Using the mechanical positioning reference point as the origin of the mechanical reference coordinate system, the nominal spatial position and nominal attitude of the slewing base, each section of the boom, the end of the boom, the hook, and the suspended load are calculated in sequence.

[0094] Let the coordinate transformation matrix corresponding to the i-th mechanical component in the mechanical kinematic chain be... ,but: (2-1); in, The coordinate transformation matrix corresponding to the mechanical reference pose output in step S1. Indicates the first Real-time angle parameters of each joint. This represents the real-time length parameter of the r-th section of the robotic arm; This indicates that, according to the connection sequence of the mechanical kinematic chain, from the first joint to the second joint... The coordinate transformation matrices of each level between the mechanical components are multiplied sequentially to obtain the mechanical reference coordinate system up to the nth level. Comprehensive transformation of the coordinate systems of individual mechanical components.

[0095] The overall meaning of formula (2-1) is: First, transform the coordinates in the mechanical reference coordinate system to the station area coordinate system, then sequentially superimpose the rotation and translation transformations from the mechanical reference part to the i-th mechanical component, finally obtaining the i-th... The pose transformation matrix of each mechanical component in the station area coordinate system Since matrix multiplication does not satisfy the commutative law, the transformation matrices must be arranged according to the actual connection order of the mechanical components.

[0096] The product order in equation (2-1) is consistent with the connection order of the mechanical components. For the aforementioned end of the crane boom, it can be specifically expressed as: The change of hook and load in This is then multiplied by the translational changes of the wire rope length and the changes in the attitude of the suspended object. Each fixed transformation is obtained from mechanical dimension calibration, while each real-time rotational and translational transformation is obtained from the measurement values ​​of the corresponding motion sensors.

[0097] Based on the coordinate transformation matrix, the coordinates of key points of each mechanical component are calculated, and the nominal geometric volume of each mechanical component is established based on the key point coordinates and the shape parameters of the corresponding components. .

[0098] Let the first The preset geometric template of a mechanical component in its local coordinate system is: Then the nominal geometric volume can be uniformly represented as: (2-2); in, The dimensions are established from the component outlines in the mechanical model parameter library. Taking the boom as an example, if the endpoint of the local centerline is... , Circumradius is Then capsules can be used. For suspended objects, the measured or recorded length, width, and height can be used to form a cuboid template, or a point cloud convex hull can be used to form a polyhedral template.

[0099] Set the preset geometry template through Transform to the station area coordinate system to obtain the nominal geometrical volume occupied by the component at time t.

[0100] The nominal geometrical occupying volume can be represented by a swept-segment volume, a capsule, a cuboid, a cylinder, a polyhedron, or a combination thereof. Specifically, for the boom, a swept-segment volume formed along the boom's centerline is used; for the hook and the suspended load, an envelope volume matching their actual shape is used; and for the rotating parts of the machinery, a rotational envelope volume corresponding to their rotation radius is used. This yields the set of nominal geometrical occupying volumes of the construction machinery under ideal positioning conditions: (2-3).

[0101] The nominal geometrical occupancy volume only reflects the spatial range calculated by the construction machinery according to its nominal pose and nominal motion parameters. It cannot reflect the actual spatial deviation caused by BeiDou observation anomalies, point cloud registration degradation, inertial cumulative errors, and mechanical attitude sensor errors. Therefore, this step describes the pose uncertainty output from step S1 as an object. It propagates along the mechanical kinematic chain to each mechanical component.

[0102] Specifically, regarding the first For each mechanical component, construct the kinematic Jacobian matrix of the key point coordinates of the mechanical component relative to the mechanical reference position, mechanical reference attitude, joint angles, and extension length. The kinematic Jacobian matrix is ​​used to characterize the moment when the mechanical reference pose and mechanical motion parameters undergo minute changes. The relationship between the coordinate changes of key points of a mechanical component.

[0103] Let the error variable consisting of the mechanical reference pose and motion parameters be... , No. The coordinates of the key points of each mechanical component are: Then the kinematic Jacobian matrix The expression is: (2-4); in, For mechanical reference position error, For small angular errors in the mechanical reference posture, This is the joint angle error. For the error of the extension length, This refers to the error in the posture or swing of the suspended object; It is a third-order identity matrix. It is an antisymmetric matrix composed of the positions of key points relative to the mechanical reference. , and These represent the partial derivatives of the key point coordinates with respect to the joint angle, extension length, and suspended object attitude parameters, respectively. This represents an operator that converts a three-dimensional vector into an antisymmetric matrix.

[0104] Regarding the first For joints independent of any component, their corresponding Jacobian columns are set to zero. Let the mechanical reference pose covariance matrix output in step S1 be... The mechanical state covariance matrix corresponding to the mechanical joint angle, telescopic length, and suspended object posture parameters is: Then the first Spatial position covariance matrix of each mechanical component Determined according to the following formula: (2-5); In the formula That is, the state covariance matrix in step S1 The position-attitude submatrix satisfies .

[0105] Let the mechanical motion parameter vector Then its covariance matrix Determine using the following formula: (2-6); in, The values ​​are determined by the calibration accuracy of the rotation angle, amplitude angle, telescopic length, and the attitude sensor of the suspended object, as well as the measurement residual of the sliding window. Based on the structural dimensional tolerances, the variance is converted to the corresponding parameter variance: when the given bilateral symmetric tolerance is ±a, the variance is taken according to a uniform distribution. When the total width of the given tolerance interval is Δ, take the variance. ; The determination is based on the residual of the suspended object's swing angle measurement, wind speed, and the rate of change of motion parameters.

[0106] thus, This always represents the covariance resulting from the combined effects of motion parameter measurement error, structural tolerance, and load sway. It is determined based on the accuracy of the mechanical attitude sensor, mechanical structural tolerances, load sway state, and the rate of change of motion parameters.

[0107] When the mechanical reference pose error and motion parameter error are approximately independent, the first... The spatial location covariance of each component is calculated in the block diagonal form of equation (2-4); if the calibration results indicate that the two are correlated, the block diagonal matrix is ​​replaced with one containing cross covariance. The complete covariance matrix of the matrix and its transpose.

[0108] The complete form of covariance propagation is .in, This represents the cross-covariance between the mechanical reference pose error and the motion parameter error; when the two are approximately independent, it is expressed as... That is, it degenerates into the block diagonal form of equation (2-5).

[0109] Based on the spatial location covariance matrix , construct the first The spatial uncertainty ellipsoid of a mechanical component : (2-7); in, Let be the nominal coordinates of the i-th key point of the mechanical component. These are the preset parameters corresponding to the security confidence level.

[0110] The spatial uncertainty ellipsoid This is used to represent the possible spatial offset range of the i-th mechanical component under the current observation and mechanical action conditions.

[0111] nominal geometric occupying volume With spatial uncertainty ellipsoid Perform an expansion combination to obtain the first... The trusted occupant of a mechanical component : (2-8); in, This indicates a space expansion combination operation.

[0112] The credible occupant does not only represent the theoretical geometric shape of the mechanical component, but also includes the conservative space occupancy range after taking into account the nominal geometric range of the machine, positioning error, attitude error, mechanical structure error and the swing error of the suspended object.

[0113] Perform a union operation on the trusted occupants of all mechanical components to generate the overall trusted occupant of the construction machinery at time t: (2-9).

[0114] When the mechanical positioning state is normal, the system generates an overall trusted occupying body according to the conventional security confidence level.

[0115] When the mechanical positioning state is a degraded positioning state, the system increases the parameters. Or increase the mechanical state covariance matrix This generates an expanded overall trusted occupying body.

[0116] When the mechanical positioning state is unreliable, the system no longer generates the occupied volume based solely on the current nominal pose, but instead uses the last normal positioning time. Based on the trusted occupying body, and combined with the maximum permissible speed of the machine, the maximum permissible acceleration, and the duration of the anomaly, a conservatively reachable occupying body is generated: (2-10); in, , representing the abnormal duration. The maximum permissible speed of the machine. This is the maximum permissible acceleration of the machine. This represents the safety margin corresponding to mechanical structure error and control response error. This represents a spatial expansion volume generated by the expansion distance.

[0117] The conservative reachable occupant is used to avoid false alarms caused by using the last nominal position for safety assessment during a location failure.

[0118] To improve the efficiency of subsequent spatial determination, the overall trusted occupancy volume can be... Or conservatively reachable occupying body The dynamic occupancy unit set is discretized, and each dynamic occupancy unit is assigned a time index, mechanical component identifier, positioning status identifier, and uncertainty level. The dynamic occupancy unit set can be represented using voxels, octree nodes, or bounding box index units.

[0119] This step outputs the nominal geometric occupying volume of the mechanical component. Component Trusted Occupation Overall trustworthy occupancy Conservatively accessible occupying body The dynamic occupancy unit set, the temporal location information of the nominal key points or material points of each component, and the corresponding time index and positioning status identifier.

[0120] in, The timing and position information is input in step S3 to calculate the actual approach direction of the machine; The dynamic occupied unit set is input into step S4 for intrusion prediction; When the location is untrusted, enter step S4 as the object of conservative risk assessment, and enter step S5 for degradation control, directional constraints and shutdown control.

[0121] Step S3: Generation of dynamic risk distance field of electrically coupled equipment in mechanical proximity direction This step utilizes the substation's 3D engineering model, the attributes and real-time operating status of energized equipment, on-site environmental monitoring data, a unified time index, and the nominal geometric occupancy of mechanical components output from step S2. The temporal location information of nominal key points or material points and component identifiers are used as inputs to generate a dynamic risk distance field related to the actual geometric surface of the energized equipment, the actual mechanical approach direction, and the continuous state of the environment.

[0122] Trustworthy Occupants It is only used to identify the closest dangerous mechanical component and subsequent intrusion determination, and is not directly used for differential calculation of mechanical movement speed, so as to avoid the change in positioning uncertainty being mistaken for the actual movement of the machine.

[0123] From the 3D engineering model of the substation, live components such as busbars, conductors, disconnectors, circuit breaker bushings, transformer bushings, and high-voltage leads are extracted to establish the first... Geometric set of hazards for each electrical device It also binds the equipment type, voltage level, energized status, phase, operating mode, and spatial orientation.

[0124] The aforementioned binding attributes are used for the subsequent generation of dynamic risk distance fields or the screening of effective hazard sources: equipment type, voltage level, and exposure method are used to select the basic safety distance. , The energized state, temporary energized state, phase, and operating mode are used to determine whether the equipment has entered the current effective hazard source set Ω(t) and to determine the equipment state correction amount. Spatial orientation, together with the local topology of the device geometry, is used to determine the surface normal vector pointing towards the safe area. .

[0125] The basic safety distance should be determined based on the equipment voltage level, equipment type, and the way live parts are exposed. , The basic risk distance field is established based on the actual geometric surface of the energized equipment rather than its center point. : (3-1).

[0126] Regarding the first Given a local surface position x of a live device, determine the surface normal vector pointing towards the safe area. And determine the approach velocity vector of the closest mechanical component from the position change of the same material point in the nominal geometrically occupied body at adjacent time points in step S2. .

[0127] Let the mechanical component at time t that is closest to the local surface position x of the energized equipment be... Mechanical parts By each trusted occupying entity The minimum distance to x is used to determine the distance.

[0128] Subsequently, the nominal geometric occupant of the mechanical component Determine the local material point closest to x. and use the same local material point at adjacent time points. Instead of reselecting the nearest point on the trusted occupying body boundary.

[0129] make Then the approximate velocity vector and its normal component toward the device for: (3-2); in, To point towards a safe area away from energized equipment, therefore only when the machinery moves toward the surface of the equipment. .

[0130] Using the same material point for difference calculation can avoid spurious velocities caused by the expansion and contraction of the credible occupied volume due to changes in covariance or confidence parameters; the velocity of the material point can also be directly calculated using the kinematic chain in step S2.

[0131] Acquire one or more environmental monitoring data points, including humidity, rainfall, visibility, temperature, wind speed, and wind direction, and determine environmental corrections using duration verification rules. , If environmental parameters exceed the corresponding threshold and remain so for a certain period of time, the correction amount is increased. If the parameters return to a safe range and remain so for a certain period of time of time, the correction amount is decreased to avoid repeated boundary switching caused by instantaneous fluctuations.

[0132] Approaching speed at - The projection of the wind speed vector in the direction that brings the suspended conductor, flexible lead wire, or suspended object closer together is used as the mechanical proximity component, and the projection of the wind speed vector in the direction that brings the suspended conductor, flexible lead wire, or suspended object closer together is used as the wind-induced proximity component. Based on this, a directional correction amount is generated. The preferred expression for the directional correction is: (3-3); in, The wind speed is the non-negative projection of the direction in which the conductor or suspended object moves closer to each other. and The values ​​are determined based on the mechanical control response time and the wind-induced oscillation calibration data, respectively.

[0133] When the machine moves parallel to or away from the local surface, no directional expansion caused by the machine's approach is applied; when the machine moves toward the local surface, the directional correction increases with the approach component. Dynamic safety distance of a live device at position x and time t Determined according to the following formula: (3-4); in, Defined as the first Under normal energized conditions, with confirmed operating mode and normal communication, the basic safety distance is determined based on voltage level, equipment type and exposure method, so no further correction is added to the general energized state. , This refers to the equipment status correction amount caused by temporary power supply, special operating mode, unconfirmed status, or communication failure relative to the normal energized reference state.

[0134] The equipment status correction amount can be determined by the equipment status indicator according to the following formula: (3-5); in, and , These are indicators for temporary power supply, special operating modes, unconfirmed status, and communication failure, respectively; corresponding to... and These are non-negative additional safety margins. Each margin is pre-written into the equipment status parameter table based on applicable safety regulations and on-site risk assessments; when multiple abnormal states occur simultaneously, the maximum value of the corresponding additional margin is preferred, or a combination of values ​​that has been verified by safety is used.

[0135] When device status information fails, it is handled according to the worst-case scenario of unconfirmed status and communication failure.

[0136] When the device status information fails, a safety boundary is generated according to the conservative condition of being energized and having an unconfirmed status. This yields the dynamic risk distance field of the j-th energized device. : (3-6); when When ≤0, it indicates position. Entering the The dynamic safety boundary of a live device; when When the value is greater than 0, it represents the remaining safety clearance.

[0137] The minimum risk distance field is taken for all currently energized or equipment that needs to be treated as energized, and a global dynamic risk distance field is generated: (3-7); A new safety boundary version number is generated for each update of device status, environmental persistence status, or directional correction, and only the local distance field within the affected device and its neighborhood is updated.

[0138] When performing the predicted intrusion determination in step S4, the data must be read simultaneously. And the corresponding security boundary version number; if the security boundary version number changes during the judgment period, the original judgment result becomes invalid and is recalculated based on the updated dynamic risk distance field.

[0139] This step outputs the geometric set of energized hazard sources, the local dynamic safety boundary, and the global dynamic risk distance field. Equipment status indicators, environmental status indicators, and safety boundary version numbers. Among these, The remaining safety clearances at each spatial location within the mechanically predicted credible occupant are used in step S4; the directional correction results and safety boundary version numbers are used in step S5 to explain the source of risk and verify the validity of control instructions.

[0140] Step S4: Predictive Intrusion Determination of Trusted Occupants of Mechanical Components This step uses the posterior fusion state estimate output from step S1. Object of pose uncertainty description And the positioning status identifier, the trusted occupant of the mechanical component output in step S2 Overall trustworthy occupancy Conservatively accessible occupying body The mechanical kinematic chain and component state information, as well as the global dynamic risk distance field output in step S3. Device status identifiers and security boundary version numbers Using the input as the basis, the system predicts and determines the near-electricity risk within the current time and the prediction time window.

[0141] Specifically, it reads the current slewing angle, amplitude angle, telescopic length, hook height, hoisting posture, joint angular velocity, joint angular acceleration, and mechanical control commands of the construction machinery.

[0142] Based on the current posterior fusion state estimation By integrating the state of construction machinery, its motion status, and control commands, we can predict the construction machinery's performance within the predicted time window. Action parameters at each discrete prediction time.

[0143] Prediction Time Window Pre-set based on machine type, maximum operating speed, control system response time, and safety level. Let the predicted time be... .

[0144] Using the mechanical kinematics chain established in step S2, the predicted motion parameters and U(t) output in step S1 are propagated to each mechanical component; when it is necessary to consider the increase in uncertainty with the prediction time, process noise is further added. The predicted reliable occupant is obtained according to the occupant generation method in step S2: (4-1); in, This represents the credible occupant of the i-th mechanical component at the predicted time. The predicted credible occupant also includes the space range that the boom, hook, load, and other hazardous mechanical components may occupy at the predicted time.

[0145] When the machine is in its normal positioning state, to predict the credible occupant. As a risk assessment entity When the machine is in a degraded positioning state, the system increases the uncertainty range corresponding to the predicted credible occupant and uses the expanded predicted credible occupant as... When the machine is in an unreliable positioning state, the current conservatively reachable occupying body is output in step S2. Based on the predicted duration The maximum permissible speed and maximum permissible acceleration of the machine continue to expand the reachability range, generating a predicted conservative reachable occupied body. and take it as This allows for coverage of the spatial range that may be reached at future predicted times during the location anomaly period.

[0146] Risk assessment of the occupied body The global dynamic risk distance field generated in step S3 for querying various spatial locations within the region is denoted as the safety boundary version used at the start of this prediction. .

[0147] If no version update is triggered within a short prediction time window, then define... That is, the current version of the risk distance field remains unchanged within the time window; if any state triggers a new version ν′, the original prediction result is immediately discarded, and the minimum safe gap of the predicted trusted occupant is regenerated based on the new version of the risk distance field.

[0148] This yields the minimum safety gap for the predicted time: (4-2); Where m(t+τ) represents the risk assessment occupying the volume at the prediction time t+τ. The minimum safe gap between all spatial locations within the area and the global dynamic risk distance field Fall.

[0149] When m(t+τ)>0, the risk-determined occupying body is located outside the dynamic safety boundary; when m(t+τ)≤0, the risk-determined occupying body has entered the dynamic safety boundary of at least one energized device.

[0150] The system calculates the minimum predicted safety gap within the prediction time window: (4-3); The It is the minimum safety gap for all discrete prediction moments within the prediction time window.

[0151] Recording makes m(t+τ) reach The most dangerous predicted moment The system calculates the nearest live equipment number, the nearest mechanical component number, and the minimum clearance location; and calculates the safety clearance change trend based on the safety clearance change at adjacent prediction times. When the current safety clearance is greater than zero but continues to decrease within the prediction time window and is expected to enter the dynamic safety boundary, it is determined as a predicted intrusion risk; when the minimum safety clearance at the current or prediction time is less than or equal to zero, it is determined as an actual intrusion risk.

[0152] Risk levels are determined based on the minimum predicted safety clearance, clearance variation trends, positioning status, and equipment status. A risk level must include at least: 1) Safety status: The minimum safety gap within the prediction time window is greater than the preset warning margin, and the safety gap does not show a continuous decreasing trend.

[0153] 2) Warning status: The minimum safety gap is less than or equal to the preset warning margin, but has not yet entered the dynamic safety boundary or there is an intrusion trend within the predicted time window.

[0154] 3) High-risk state: It is predicted that the trusted occupier will enter the dynamic security boundary within the prediction time window.

[0155] 4) Intrusion status: The current trusted occupying entity or conservatively reachable occupying entity has entered the dynamic security boundary.

[0156] 5) Location is unreliable: The location is unreliable and cannot generate a valid prediction result that meets the preset security requirements.

[0157] To ensure consistency between the risk assessment result and the status of the energized equipment, the security boundary version number used in this assessment is compared with the current security boundary version number in step S3. If they are inconsistent, the current assessment result is discarded, and the predicted intrusion assessment is re-executed based on the updated dynamic risk distance field.

[0158] This step outputs the minimum predicted safety gap. The most dangerous predicted moment, the nearest energized equipment identifier, the nearest mechanical component identifier, the risk level, the risk source identifier, and the safety boundary version number are all included. Among these, the risk level, the risk source identifier, and the most dangerous predicted moment serve as inputs for graded warnings, speed limits, directional constraints, and locking or shutdown controls in step S5.

[0159] Step 5: Directional active constraint and closed-loop control based on safety clearance control sensitivity This step uses the minimum predicted safety gap output from step S4. The system takes the most dangerous predicted moment, the nearest energized equipment identifier, the nearest mechanical component identifier, the risk level, the risk source identifier, and the safety boundary version number as inputs to determine the specific actuator and action direction that will lead to a reduction in the predicted safety clearance, and generates active constraint commands while preserving avoidance actions as much as possible.

[0160] The mechanical control quantity is expressed as: (5-1); in, For rotation speed command, For variable speed command, For telescopic speed command, This is the command for the lifting speed of the hook.

[0161] For the k-th control component, while keeping other control components unchanged, a preset small disturbance ε is applied, and the short-time prediction in step S4 is re-executed. The control sensitivity of this control component to the minimum predicted safety gap is calculated according to the following formula: (5-2); in, This is the unit vector corresponding to the k-th control component.

[0162] like <0 indicates that the current k-th control component reduces the predicted safety gap, and it is marked as a dangerous directional action. A value >0 indicates that the control component increases the predicted safety clearance, and it is marked as a hazard avoidance directional action.

[0163] Therefore, the sources of risk have been refined from the general concept of "mechanical proximity to live equipment" to specific actuators, directions of motion, and their impact on safety clearances.

[0164] When the risk level is safe, the system maintains the original control commands and continues monitoring. When in a warning state, it outputs the nearest equipment, the nearest mechanical component, remaining safety clearance, estimated intrusion time, and suggested avoidance direction to the driver's cab, on-site audio-visual devices, and the back-end terminal, while increasing the prediction update frequency. When in a high-risk state, the operator control commands are... Projecting onto the set of permissible controls that satisfy the predicted safety constraints yields the constraint control instructions: (5-3); in, This provides a preset warning margin. Through this projection processing, the system prioritizes reducing the operating speed that minimizes the safety clearance, prohibits actions in the corresponding dangerous direction when necessary, and retains the operating components that can keep the machinery away from energized equipment, thus avoiding indiscriminate shutdowns that could cause instability of the suspended load or sudden changes in the machinery's attitude.

[0165] In equation (5-3) Only applied to control sensitivity Furthermore, the dangerous direction control component that would reduce the minimum safety clearance if movement continued in that direction; The risk aversion component is retained or limited.

[0166] When there is no satisfying When a feasible control command is issued, the allowed control set will be determined to be empty, and all dangerous directional actions will be prohibited; if step S4 still cannot verify a safe avoidance direction, the relevant actuators will be locked or the emergency stop will be executed directly.

[0167] When the current trusted or conservatively reachable occupier has entered the dynamic safety boundary, stop dangerous directional actions and lock the relevant actuators; when there is a safe avoidance direction verified by step S4, allow limited deceleration or reverse avoidance actions, and execute an emergency shutdown when the safe avoidance direction cannot be determined.

[0168] When the positioning status is unreliable, control verification is performed using a conservatively reachable occupying body. Actions toward energized equipment are prohibited, and the speed of other actions is limited until the positioning status continuously meets the recovery conditions and the recalculated minimum predicted safety gap is greater than the recovery margin.

[0169] This step outputs on-site early warning information, background push information, permitted control sets, directional action constraint commands, lockout or shutdown commands, and safety event records. Safety event records must include at least control sensitivity, restricted actuators and directions, minimum predicted safety clearance before and after control, safety boundary version number, and execution feedback, thus forming a safety closed loop of "prediction judgment - command constraint - execution feedback - re-judgment."

[0170] Corresponding to the above method, the system of this embodiment includes: a multi-source synchronous acquisition and integrity assessment unit, a mechanically trusted occupant generation unit, a dynamic risk distance field generation unit for energized equipment, a predicted intrusion determination unit, a directional active constraint and execution feedback unit, and a security event recording unit. Each unit executes steps S1 to S5 respectively, and in accordance with... Figure 1 The data stream shown transmits a unified time index, location status identifier, occupied volume data, security boundary version number, risk level, and control feedback.

[0171] In some embodiments of the present invention, for pose acquisition, in addition to the combination of BeiDou RTK, LiDAR, and IMU, UWB base stations and tags, visual identification codes, visual inertial odometry, or millimeter-wave radar can be used to acquire mechanical reference pose. Regardless of the positioning method used, the pose estimation result and the corresponding observation quality parameters or uncertainty parameters must be output for subsequent reliable occupant generation.

[0172] In some embodiments of the present invention, in addition to using the expansion combination of the spatial uncertainty ellipsoid and the nominal geometric occupant, Monte Carlo sampling, unscented transformation, interval propagation or polyhedral envelope methods can also be used to calculate the spatial range that the mechanical component may occupy in order to generate the mechanically reliable occupant. The spatial index unit can be represented by voxels, octree nodes, hierarchical bounding boxes or signed distance field grids.

[0173] In some embodiments of the present invention, for the generation of dynamic safety boundaries, energized equipment can be represented by a triangular mesh model, a parametric BIM model, a point cloud model, or a bounding volume model; environmental conditions can be obtained from meteorological sensors, on-site image recognition, weather forecast interfaces, or insulation condition detection devices. Dynamic safety boundaries can be implemented using distance fields, isosurfaces, bounding box expansion, or layered electronic fences, and are all updated according to the equipment status and environmental conditions.

[0174] In some embodiments of the present invention, for predictive intrusion determination and control, short-term prediction based on current speed and acceleration, kinematic simulation based on mechanical control commands, or predictive models based on historical operating trajectories can be used to generate a future reliable occupant. Control methods, in addition to speed limiting, directional constraints, locking, and stopping, can also be implemented using cab prompts, remote authorization confirmation, or automatic avoidance trajectory planning.

[0175] On the other hand, the present invention also provides an electronic device, comprising: Processor; and Memory for storing the executable instructions of the processor; The processor is configured to execute the substation construction machinery safety risk early warning method based on BeiDou positioning and three-dimensional data by executing the executable instructions.

[0176] On the other hand, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the aforementioned method for early warning of safety risks of substation construction machinery based on BeiDou positioning and three-dimensional data.

[0177] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications or equivalent changes made to the above embodiments based on the technical essence of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for early warning of safety risks of substation construction machinery based on BeiDou positioning and 3D data, characterized in that, Includes the following steps: S1. Acquire observation data from BeiDou RTK, lidar, inertial measurement unit, and mechanical attitude sensor to determine the posterior fusion state estimate of the construction machinery state vector; obtain the observation integrity coefficient corresponding to each observation source based on the observation residual, effective observation ratio, continuous anomaly duration, and cross-source consistency deviation of each observation source. The observation covariance is adaptively corrected based on the observation integrity coefficient, and a pose uncertainty description object and a positioning status identifier are generated. S2. Based on the nominal pose of the mechanical reference and the structural parameters and real-time motion parameters of the construction machinery, establish the mechanical kinematic chain of the construction machinery and the nominal geometric occupancy of the mechanical components; propagate the pose uncertainty description object along the mechanical kinematic chain to each mechanical component to generate the corresponding spatial uncertainty ellipsoid; expand and combine the nominal geometric occupancy and the uncertainty ellipsoid to form a reliable mechanical occupancy. S3. Establish a basic safety distance field based on the actual geometric surface of the energized equipment, and generate a dynamic risk distance field for the energized equipment based on the environmental correction, directional correction and state correction corresponding to the energized equipment. S4. Obtain the minimum predicted safety gap between the predicted credible occupant and the dynamic risk distance field, and conduct risk warning based on the minimum predicted safety gap, gap change trend, positioning status and energized equipment status.

2. The method for early warning of safety risks of substation construction machinery based on BeiDou positioning and three-dimensional data according to claim 1, characterized in that, Based on the integrity coefficient and status identifier of each observation source, the positioning status is divided into normal status, degraded status and untrusted status. When it is in a degraded status, step S2 is triggered to expand the mechanically trustworthy occupying body. When it is in an untrusted status, step S2 is triggered to generate a conservatively reachable occupying body.

3. The substation construction machinery safety risk early warning method based on BeiDou positioning and three-dimensional data according to claim 2, characterized in that, When the mechanical positioning state is an untrusted positioning state, the trusted occupant at the last normal positioning moment is used. Based on this, combined with the maximum permissible speed of the machinery For, maximum permissible acceleration and duration of abnormality Generate a conservatively accessible occupied body , is represented as: ； in, This represents a space expansion combination operation. This represents the safety margin corresponding to mechanical structure error and control response error. This represents a spatial expansion volume generated by the expansion distance.

4. The method for early warning of safety risks of substation construction machinery based on BeiDou positioning and three-dimensional data according to claim 1, characterized in that, In step S2, the coordinates of the key points of the mechanical components are calculated based on the coordinate transformation matrix corresponding to the mechanical components in the mechanical kinematic chain, and the nominal geometric occupancy of each mechanical component is established based on the coordinates of the key points and the shape parameters of the corresponding components.

5. The substation construction machinery safety risk early warning method based on BeiDou positioning and three-dimensional data according to claim 4, characterized in that, In step S2, the kinematic Jacobian matrix of the key point coordinates of the mechanical component relative to the mechanical reference position, mechanical reference posture, joint angle and extension length is constructed. The spatial position covariance matrix of the mechanical component is obtained based on the mechanical reference pose covariance matrix and the mechanical state covariance matrix corresponding to the mechanical joint angle, extension length and the attitude parameters of the suspended object. The spatial uncertainty ellipsoid of the mechanical component is then constructed based on the spatial position covariance matrix.

6. The method for early warning of safety risks of substation construction machinery based on BeiDou positioning and three-dimensional data according to claim 2, characterized in that, In step S3, one or more environmental monitoring data points are acquired, including humidity, rainfall, visibility, temperature, wind speed, and wind direction, and the environmental correction amount is determined using duration confirmation rules. , ; Approaching speed at - The projection of the wind speed vector in the direction that brings the suspended conductor, flexible lead wire, or suspended object closer together is used as the mechanical proximity component, and the projection of the wind speed vector in the direction that brings the suspended conductor, flexible lead wire, or suspended object closer together is used as the wind-induced proximity component to generate the directional correction amount. ; To point to a safe area away from electrical equipment; No. Dynamic safety distance of a live device at surface position x and time t Represented as: ； in, , For the first The basic safety distance for each device , This is the equipment status correction amount caused by temporary power supply, special operating mode, unconfirmed status, or communication failure relative to the normal energized reference state; The dynamic risk distance field of the j-th charged device Represented as: ; For the first A geometric set of hazards for a single piece of electrical equipment; The minimum risk distance field is taken for all currently energized or equipment that needs to be treated as energized, to generate a global dynamic risk distance field. .

7. The method for early warning of safety risks of substation construction machinery based on BeiDou positioning and three-dimensional data according to claim 6, characterized in that, In step S4, based on the posterior fusion state estimate, mechanical motion state, and control commands, the construction machinery is predicted to operate within the prediction time window. Action parameters at each discrete prediction time; Based on the mechanical kinematics chain, the predicted motion parameters and pose uncertainty description objects are propagated to the mechanical components to obtain the predicted reliable occupant of the mechanical components at the prediction time. When in a normal positioning state, the predicted trusted occupant is used as the object of risk assessment; When in a downgraded positioning state, the uncertainty range corresponding to the predicted credible occupant is increased and then used as the risk assessment object; When in an unreliable location state, the predicted conservatively reachable occupied body is used as the object of risk assessment.

8. The method for early warning of safety risks of substation construction machinery based on BeiDou positioning and three-dimensional data according to claim 7, characterized in that, The process after step S4 also includes: Obtain the control sensitivity of the mechanical control component to the minimum predicted safety clearance. ,like If <0, mark the current k-th control component as a dangerous direction action. If the value is >0, mark it as a hazard avoidance action; When the risk level is safe, maintain the original control commands and continue monitoring; when the risk level is warning, output the nearest equipment, the nearest mechanical component, the remaining safety clearance, the estimated intrusion time and the suggested avoidance direction to the driver's cab, on-site audio-visual devices and the back-end terminal, and increase the prediction update frequency; when the risk level is high-risk, project the control commands to the permissible control set that meets the predicted safety constraints to obtain constraint control commands. And / or, when the current trusted occupying body or conservatively reachable occupying body has entered the dynamic safety boundary, stop the dangerous direction action and lock the relevant actuators; when there is a safe avoidance direction, perform deceleration or reverse avoidance action; when the safe avoidance direction cannot be determined, perform emergency shutdown.

9. An electronic device, characterized in that, include: processor; as well as Memory for storing the executable instructions of the processor; The processor is configured to execute the substation construction machinery safety risk early warning method based on BeiDou positioning and three-dimensional data as described in any one of claims 1-8 by executing the executable instructions.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the substation construction machinery safety risk early warning method based on Beidou positioning and three-dimensional data as described in any one of claims 1-8.

Images