A positive crane wire active anti-collision early warning control method

Abstract

The application relates to the technical field of safety control of engineering machinery, and discloses a positive front-lifting crane electric wire active anti-collision early warning control method, which comprises the following steps: collecting image data, point cloud data and real-time attitude parameters of the positive front-lifting crane of the electric wire through a visual sensor, a radar sensor and an attitude sensor respectively; performing time synchronization and space coordinate unification on the collected multi-source data, and performing deep fusion on the image data and the point cloud data, so as to output three-dimensional space position information and an approaching speed of the electric wire. The application aims to solve the technical problems that in the existing positive front-lifting crane electric wire anti-collision scheme, passive alarm is only relied on the current static distance, dynamic prediction on the lifting arm movement trend is lacked, early warning is seriously lagged and active intervention is impossible.

Description

Technical Field

[0001] This invention relates to the field of safety control technology for engineering machinery, specifically to an active anti-collision early warning control method for overhead power lines. Background Technology

[0002] Attractive cranes (reach stackers) are widely used in container yards such as ports and railway freight yards, where overhead power lines often exist above or around their operating areas. Because the boom length of an attractive crane can reach over 20 meters, the lifting height can exceed 15 meters, and the operator's cab has limited visibility, it is difficult for operators to accurately judge the real-time distance between the boom, spreader, and power lines during operation. To address this, various collision avoidance solutions have emerged in existing technologies. These include using a single camera for image recognition and triggering an audible and visual alarm when an obstacle is detected, or installing mechanical stops or limit switches at specific extreme positions on the boom to restrict its range of motion.

[0003] However, the aforementioned existing technologies can only provide passive alarms based on the static distance at the current moment, and cannot predict the movement trend of the boom. When the boom approaches the line at high speed, the warning signal is severely delayed, and the driver often does not have time to take braking measures, making it difficult to fundamentally avoid collision accidents. Summary of the Invention

[0004] The purpose of this invention is to solve the technical problems of existing front-mounted overhead wire collision prevention schemes, which rely solely on the current static distance for passive alarms, lack dynamic prediction of the boom's movement trend, resulting in severe delays in early warning and an inability to actively intervene. Therefore, this invention proposes an active collision prevention and early warning control method for front-mounted overhead wires.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: A method for active anti-collision warning and control of overhead power lines includes the following steps: Image data, point cloud data, and real-time attitude parameters of the front-end suspension are collected by visual sensors, radar sensors, and attitude sensors, respectively. The collected multi-source data is synchronized in time and unified in spatial coordinates. The image data and point cloud data are deeply fused to output the three-dimensional spatial position information and approach speed of the power line. A kinematic model of the boom end is established based on the real-time attitude parameters. A filtering algorithm is used to predict the trajectory of the boom end within a predetermined time period in the future. The minimum spatial distance and its rate of change between the predicted trajectory and the three-dimensional spatial position information are calculated. Based on the line voltage level, the approach speed, and the ambient humidity, a three-level safety threshold, including warning distance, deceleration distance, and locking distance, is dynamically calculated. The minimum spatial distance is compared with the three-level safety threshold, and control commands for audible and visual alarms, speed limits, or directional locking in dangerous directions are executed according to the comparison results. When the minimum spatial distance is greater than the locking distance and the duration exceeds the preset threshold, the speed limit and directional locking commands are automatically released.

[0006] Based on the above technical solution, the present invention can be further improved as follows.

[0007] Furthermore, the deep fusion adopts a fusion strategy of radar-prioritized ranging and visual fine positioning: the point cloud data of the radar sensor provides the distance and approach speed of the wire, and the image data of the visual sensor provides the geometric direction and type of the wire after instance segmentation, and finally outputs a three-dimensional bounding box of the line with confidence.

[0008] Furthermore, the warning distance in the three-level safety threshold... Deceleration distance and locking distance Calculated dynamically according to the following formula:

[0009] in, This refers to the basic safety distance corresponding to the line voltage level. For speed coefficient, Humidity influence coefficient This represents the ambient humidity value.

[0010] Furthermore, the real-time attitude parameters include the boom pitch angle. boom length Rotation angle and lifting height The kinematic model is represented as a position function of the boom end in the vehicle coordinate system. The velocity at the end of the boom is obtained by differentiating with respect to time. and acceleration .

[0011] Furthermore, the filtering algorithm employs an extended Kalman filter to construct a state vector. The trajectory of the boom end effector is recursively predicted within the next 500ms to 1500ms.

[0012] Furthermore, when the minimum spatial distance is greater than the locking distance and less than or equal to the deceleration distance, a speed limit command is sent to the vehicle controller of the front-end crane via the CAN bus; The vehicle controller limits the hydraulic pump's displacement according to the speed limit command, thereby reducing the maximum operating speed in the dangerous direction by 50%. The dangerous direction is determined based on the relative kinematic decoupling result between the predicted trajectory at the end of the boom and the three-dimensional bounding box of the line, and includes at least one of the boom pitching and descent direction, the boom extension direction, or the boom rotation direction.

[0013] Furthermore, when the minimum spatial distance is less than or equal to the locking distance, a zero-speed locking command is sent to the corresponding hydraulic proportional valve controller via the CAN bus to lock only dangerous directional actions that may cause collision risks, while retaining the operating authority in the safe direction.

[0014] Furthermore, the preset threshold is 1 second, and a hysteresis interval Δd is set; when the minimum spatial distance is greater than the sum of the locking distance and the hysteresis interval and the duration reaches 1 second, the speed limit and directional locking command are automatically released.

[0015] Furthermore, it also includes: monitoring the controller's operating status through a watchdog timer, and judging the validity of data from each sensor in real time through a data health check module; When the data interruption or invalid time of the radar sensor or attitude sensor exceeds 100ms, the warning distance, deceleration distance and locking distance are automatically amplified to the preset most conservative value, and sensor abnormality prompt information is output.

[0016] Furthermore, the time synchronization adopts the PTP protocol to provide timing for all sensors through a high-precision clock source, ensuring that the timestamp error of image data, point cloud data and real-time attitude parameters is less than 1ms; The spatial coordinates are unified by converting all sensor data into a vehicle coordinate system with the front crane's rotation center as the origin.

[0017] Compared with the prior art, the technical solution of this application has the following beneficial technical effects: This invention, by fusing visual and radar sensors and synchronizing attitude parameters, can acquire the three-dimensional spatial position and relative approach speed of a line in real time, providing a precise data foundation for dynamic judgment and overcoming the problem of inaccurate distance measurement by a single sensor under complex working conditions. Based on kinematic models and filtering algorithms, it predicts the future trajectory of the boom and calculates the minimum spatial distance and its rate of change, enabling the system to proactively identify risks before a collision occurs, completely changing the passive "post-event alarm" approach of traditional solutions. Based on dynamically calculated three-level safety thresholds, it performs graded interventions, proactively limiting speed when danger approaches, directionally locking dangerous movements in emergency situations, while retaining safe-direction work permissions, and automatically resetting after the risk is eliminated. Thus, this invention achieves a leap from "passive response" to "proactive prediction and intervention," effectively solving the problem of delayed early warning and significantly improving the safety and continuity of line collision avoidance. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the overall method steps of the present invention; Figure 2 This is a schematic diagram illustrating the multi-source data acquisition and deep fusion of the present invention; Figure 3 This is a schematic diagram of the kinematic modeling and trajectory prediction of the boom end of the present invention; Figure 4 This is a schematic diagram illustrating the dynamic three-level safety threshold calculation and hierarchical control of the present invention; Figure 5 This is a schematic diagram illustrating the danger direction determination and execution control of the present invention; Figure 6 This is a schematic diagram illustrating the state recovery and abnormal safety redundancy of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] The present invention provides an active anti-collision warning control method for front-mounted overhead power lines, comprising the following steps: Image data, point cloud data, and real-time attitude parameters of the front-end suspension are collected by visual sensors, radar sensors, and attitude sensors, respectively. During data acquisition, an industrial camera with high dynamic range capability is mounted on the front top of the cab, with the lens facing the boom's working area, to continuously acquire image frames of the power lines. A 77GHz millimeter-wave radar is installed at the boom root or on the cab top, its beam covering the space in front of the boom, outputting point cloud data containing distance, speed, and azimuth information. A magnetoresistive angle sensor is installed at the boom root hinge point, an coded displacement sensor is installed at the telescopic cylinder, an absolute encoder is installed at the slewing bearing center, and a rotary encoder is installed at the hoisting drum shaft end, to measure the boom pitch angle, telescopic length, slewing angle, and spreader height in real time.

[0021] The collected multi-source data is synchronized in time and unified in spatial coordinates. The image data and point cloud data are deeply fused to output the three-dimensional spatial position information and approach speed of the power line. like Figure 2As shown, after receiving the multi-source data, the controller first performs time synchronization and spatial unification. The controller has a built-in PTP master clock to provide a unified timing reference for all sensors, grouping data frames with acquisition times differing by less than 1 millisecond into the same processing cycle. Through offline calibrated rotation matrices and translation vectors, all sensor data are transformed into a vehicle coordinate system with the front crane's rotation center as the origin.

[0022] A kinematic model of the boom end is established based on real-time attitude parameters. A filtering algorithm is used to predict the trajectory of the boom end within a predetermined time period in the future. The minimum spatial distance and its rate of change between the predicted trajectory and the three-dimensional spatial position information are calculated. Then, deep data fusion is performed. Image frames are input into a pre-trained instance segmentation network to extract pixel-level contours of the power lines. Ground filtering and density clustering are performed on the radar point cloud to filter out point cloud clusters with linear distribution. Using coordinate mapping relationships, the visual contours are associated and matched with the point cloud clusters. A strategy of "radar ranging as the main method and visual positioning as the auxiliary method" is adopted to output the three-dimensional bounding box of the power line and its approach speed relative to the end of the boom.

[0023] Based on line voltage level, approach speed and ambient humidity, a three-level safety threshold including warning distance, deceleration distance and locking distance is dynamically calculated; Next, dynamic trajectory prediction is performed. The controller reads the boom pitch angle, extension length, slewing angle, and spreader height in real time, substitutes them into the boom end-effector kinematic equations, calculates the current position of the boom end-effector in the vehicle coordinate system, and obtains the end-effector velocity and acceleration by differentiating with respect to time. Using the extended Kalman filter algorithm, a state vector is constructed based on the aforementioned position, velocity, and acceleration to recursively predict the boom end-effector trajectory at multiple future time points. The minimum spatial distance and its rate of change between the predicted trajectory and the 3D bounding box of the line are calculated.

[0024] The minimum spatial distance is compared with the three-level safety threshold, and control commands such as audible and visual alarm, speed limit, or directional locking are executed according to the comparison result. When the minimum spatial distance is greater than the locking distance and the duration exceeds the preset threshold, the speed limit and directional locking commands are automatically released.

[0025] like Figure 1 As shown, the three-level safety thresholds are then dynamically calculated. The basic safety distance is determined based on the line voltage level. Combined with the current approach speed and ambient humidity, the warning distance, deceleration distance, and locking distance are calculated respectively. The warning distance increases with increasing speed and humidity, while the deceleration distance and locking distance decrease progressively at fixed ratios.

[0026] Subsequently, tiered intervention control is implemented. The minimum spatial distance is compared with the three-level safety threshold: if it is between the warning distance and the deceleration distance, an audible and visual alarm is triggered in the driver's cab; if it is between the deceleration distance and the locking distance, a speed limit command is sent to the vehicle controller via the CAN bus to reduce the maximum operating speed in the dangerous direction; if it is less than or equal to the locking distance, a zero-speed locking command is sent to the hydraulic proportional valve controller to lock only the dangerous direction that poses a collision risk, while retaining operating authority in the safe direction.

[0027] Finally, an automatic reset is performed. The minimum spatial distance is continuously monitored. When it exceeds the locking distance and the duration exceeds a preset threshold (e.g., 1 second), the controller automatically sends a release command to cancel all speed limits and directional locks, and the equipment returns to normal operation.

[0028] In a preferred embodiment, the present invention can be further configured as follows: the deep fusion adopts a fusion strategy of radar-prioritized ranging and visual fine-tuning: the point cloud data from the radar sensor provides the distance and approach velocity of the wire, while the image data from the visual sensor, after instance segmentation, provides the geometric orientation and type of the wire, ultimately outputting a 3D bounding box of the line with confidence. Specifically, the deep fusion process adopts a fusion strategy of "radar-prioritized ranging and visual fine-tuning." The point cloud data from the radar sensor provides the distance d and approach velocity v of the wire.

[0029] The controller first performs ground filtering and DBSCAN clustering on the radar point cloud to identify continuous point cloud clusters belonging to the power line, and calculates the distance and radial velocity of the centroid of the cluster. The image data from the vision sensor is processed by a trained improved YOLOv5s-seg instance segmentation model to output a pixel-level mask of the power line, thereby extracting the geometric orientation (horizontal, inclined, or vertical) and type (single, double, or split conductor) of the power line.

[0030] Using a pre-calibrated radar-camera extrinsic parameter matrix, the radar point cloud is mapped onto the image plane and matched with a visual mask. For successfully matched regions, the radar ranging value is used as the final distance, while visual information is used to confirm the target's category and spatial pose. For regions with sparse or missing radar point clouds, the distance information is supplemented using the bounding box of the visual mask combined with a monocular depth estimation algorithm. The final output is a linear 3D bounding box with confidence.

[0031] In a preferred embodiment, the present invention may be further configured as follows: Figure 4 As shown, the warning distance in the three-level safety threshold... Deceleration distance and locking distance Calculated dynamically according to the following formula:

[0032] in, This refers to the basic safety distance corresponding to the line voltage level. For speed coefficient, Humidity influence coefficient This represents the ambient humidity value. According to the minimum safety distance settings for different voltage levels in the "State Grid Corporation of China Power Safety Work Regulations," the specific values ​​are: 0.7m for 10kV, 1.0m for 35kV, 1.5m for 110kV, and 3.0m for 220kV. (Speed ​​coefficient) Based on the maximum permissible speed calibration of the boom tip in the dangerous direction, this application specifically obtains the speed through field testing. =0.2 s -1 Humidity Influence Coefficient Piecewise linear mapping based on relative humidity: when humidity ≤ 30% =0.5, humidity ≥90% =2.0, intermediate humidity linear interpolation.

[0033] In a preferred embodiment, the present invention can be further configured such that the real-time attitude parameters include the boom pitch angle. boom length Rotation angle and lifting height The kinematic model is represented as the position function of the boom end in the vehicle coordinate system. The velocity at the end of the boom is obtained by differentiating with respect to time. and acceleration ,like Figure 3 As shown, the boom pitch angle in the real-time attitude parameters The angle is directly measured by a magnetoresistive angle sensor installed at the base of the boom shaft, with a measurement range of 0° to 90° and a resolution of 0.03°. Boom length. The displacement is measured by a wire-type or magnetostrictive displacement sensor mounted on the telescopic cylinder. In this application, an encoder is used in conjunction with a measuring wheel, with a measurement accuracy of ±1 cm. Rotation angle. Measured by an absolute encoder mounted at the center of the slewing bearing, with a resolution of 0.1°. Spreader height. Indirect measurement is performed using a rotary encoder mounted on the hoisting wire rope drum. The specific expression for the kinematic model is:

[0034] in Let be the height of the center of rotation above the ground. Differentiating with respect to time yields the velocity components, which are then synthesized into the final velocity. and acceleration In this application, the backward difference method is specifically used for numerical differentiation, with a sampling interval of 20ms.

[0035] In a preferred embodiment, the present invention can be further configured such that: the filtering algorithm employs an extended Kalman filter to construct a state vector. The system recursively predicts the trajectory of the boom end effector within the next 500ms to 1500ms. Specifically, the filtering algorithm uses an extended Kalman filter (EKF), whose state vector... The model has 9 dimensions. The prediction model is based on the assumption of constant acceleration motion: X(t+Δt)=F·X(t)+w, where F is the state transition matrix and w is the process noise. The observation model establishes a nonlinear mapping relationship between the measured boom pitch angle θ, boom length L, slewing angle φ, and spreader height h and the position components in the state vector, and uses the linearization method of EKF (first-order Taylor expansion) to calculate the Jacobian matrix.

[0036] In this application, the prediction time Δt is specifically set to 500ms, 1000ms, and 1500ms, respectively, and the predicted positions at the three time points are output to form the predicted trajectory. 500ms was chosen as the minimum prediction time because the minimum reaction time required for emergency braking of the front-end crane is approximately 300–500ms; 1500ms was chosen as the maximum prediction time because beyond this time, the uncertainty of the boom movement increases significantly, and the prediction error exceeds the acceptable range.

[0037] In a preferred embodiment, the present invention can be further configured such that when the minimum spatial distance is greater than the locking distance and less than or equal to the deceleration distance, a speed limit command is sent to the vehicle controller of the front-end crane via the CAN bus; The vehicle controller limits the hydraulic pump's displacement based on this speed limit command, thereby reducing the maximum operating speed in the dangerous direction by 50%. The danger direction is determined based on the relative kinematic decoupling results between the predicted trajectory at the boom end and the three-dimensional bounding box of the line, including at least one of the boom pitch and descent direction, boom extension direction, or boom rotation direction, such as... Figure 5 As shown, when the minimum space distance enters the deceleration zone, the controller sends a speed limit command to the vehicle controller VCU of the front crane via the CAN bus. After receiving the command, the controller adjusts the duty cycle of the PWM signal of the proportional displacement valve of the hydraulic pump according to the dangerous direction specified in the command, limiting the output flow of the hydraulic pump to 50% of the normal flow, thereby reducing the maximum operating speed in the dangerous direction by 50%.

[0038] Specifically, the rate of change of the boom's pitch, extension, and slewing speeds are calculated separately, and the direction with the largest rate of change is taken as the dangerous direction. In the specific implementation of this application, when the boom's pitch descent causes a collision risk, only the pitch descent speed is limited, while pitch ascent, extension, slewing, and travel are unaffected.

[0039] In a preferred embodiment, the present invention can be further configured as follows: when the minimum spatial distance is less than or equal to the locking distance, a zero-speed locking command is sent to the corresponding hydraulic proportional valve controller via the CAN bus, locking only actions in the dangerous direction that pose a collision risk, while retaining the operating permission in the safe direction. When the minimum spatial distance enters the locking zone, the controller immediately sends a zero-speed locking command to the corresponding hydraulic proportional valve controller via the CAN bus. This command is the highest priority safety frame, requiring the solenoid coil of the target proportional valve to be de-energized or kept at zero position, thereby cutting off the pressure oil circuit in that direction. The locking target is limited to actions in the dangerous direction that pose a collision risk.

[0040] For example, if kinematic decoupling determines that boom extension is the primary hazard, the zero-speed locking command only applies to the extension direction of the boom, while the retraction direction retains normal operating permissions, allowing the operator to move the boom away from the danger zone. Operating permissions in safe directions (such as boom pitching, slewing away from the line, etc.) are fully preserved.

[0041] In other embodiments of this application, the controller may also output an audible and visual alarm simultaneously, and graphically indicate the specific direction of the locked action on the driver's cab display screen, prompting the driver to perform the opposite action to escape danger.

[0042] In a preferred embodiment, the present invention can be further configured as follows: a preset threshold of 1 second is set, and a hysteresis interval Δd is set; when the minimum spatial distance is greater than the sum of the locking distance and the hysteresis interval and the duration reaches 1 second, the speed limit and directional locking command are automatically released. like Figure 6 As shown, the automatic reset logic is implemented independently as follows. The preset reset delay time is 1 second, and a hysteresis distance value is set, which is 0.1 meters in this embodiment. The controller calculates the current minimum spatial distance in each control cycle (e.g., 20 milliseconds) and compares it with the locking distance. When the current minimum spatial distance is detected to be greater than the sum of the locking distance and the hysteresis distance, the controller starts a timer; If this condition is maintained for more than 1 second, it is considered that the front crane boom has moved away from the power line and the risk of collision has been completely eliminated. At this time, the controller automatically sends a release command via the CAN bus to cancel the previously issued speed limit command and directional lock command, allowing the equipment to return to normal operation.

[0043] If the current minimum spatial distance falls below the locking distance again within 1 second, the timer is reset to zero. The purpose of the hysteresis distance is to prevent the system from frequently entering and exiting the locking state when the minimum spatial distance fluctuates around the locking distance, thereby preventing oscillations in the hydraulic system.

[0044] In this application, the hysteresis distance of 0.1 meters is calibrated based on the typical hydraulic valve response time and boom inertia. This value effectively suppresses oscillations without causing reset lag due to excessive size. In other embodiments of this application, the hysteresis distance can be dynamically adjusted according to the boom's movement speed; the higher the speed, the greater the hysteresis distance.

[0045] In a preferred embodiment, the present invention may be further configured to include: monitoring the operating status of the controller through a watchdog timer, and judging the validity of the data from each sensor in real time through a data health check module; When the data interruption or invalidation time of the radar sensor or attitude sensor exceeds 100ms, the warning distance, deceleration distance, and lock-up distance are automatically increased to the preset most conservative values, and a sensor abnormality prompt message is output. The safety fault tolerance steps are implemented independently as follows: The controller integrates a watchdog timer, which runs with an independent clock source. Every 100ms, the main control program performs a watchdog feeding operation. If the main program fails to feed the watchdog in time due to a software infinite loop or hardware failure, the watchdog timer will overflow, automatically triggering a hardware reset and setting all output instructions to a safe state, i.e., executing zero-speed locking of all movement directions, and simultaneously issuing an audible and visual alarm. The data health check module monitors the communication status of the radar sensor and attitude sensor, the data frame cyclic redundancy check results, and data validity in real time.

[0046] When a radar sensor data interruption is detected for more than 100 milliseconds, or when the attitude sensor output exceeds a reasonable physical range (e.g., the pitch angle measurement value is greater than 90 degrees), the sensor is deemed to be faulty.

[0047] At this point, the controller automatically increases the warning distance, deceleration distance, and locking distance to the preset most conservative values: specifically, the basic safety distance is forcibly set to the safety distance corresponding to the highest voltage level (e.g., 3.0 meters), while the speed influence coefficient and humidity influence coefficient are both set to the maximum allowable value, thereby increasing the safety thresholds at each level to 2 to 3 times the normal value.

[0048] Simultaneously, the controller outputs a text message "Sensor malfunction, entering safety protection mode" to the cab display via the CAN bus, alerting the operator. In other embodiments of this application, when a critical sensor fails, it can automatically switch to a backup sensor to continue operating, or request manual intervention.

[0049] In a preferred embodiment, the present invention can be further configured as follows: time synchronization adopts the PTP protocol to provide timing for all sensors through a high-precision clock source, ensuring that the timestamp error of image data, point cloud data and real-time attitude parameters is less than 1ms; Spatial coordinates are unified by converting all sensor data to a vehicle coordinate system with the center of rotation of the front-end crane as the origin. Time synchronization is implemented independently of spatial coordinate unification as follows: Time synchronization uses the PTP protocol, with the controller acting as the PTP master clock, providing time synchronization to the vision, radar, and attitude sensors via Ethernet or a dedicated synchronization signal line. Each sensor supports hardware timestamp capture, with each data frame including a precise timestamp of its acquisition time, achieving nanosecond-level accuracy. When receiving data, the controller aligns data frames from different sensors based on the timestamps, allowing a maximum time difference of 1 millisecond.

[0050] Spatial coordinate unification is performed according to the following steps: First, the rotation matrix and translation vector from the radar sensor coordinate system to the visual sensor camera coordinate system are obtained through offline calibration. Using these two transformation parameters, the coordinates of the point cloud data acquired by the radar sensor are transformed to the camera coordinate system.

[0051] Then, using another set of pre-calibrated rotation matrices and translation vectors, all sensor data (including visual data, radar data, and attitude data) are uniformly transformed into a vehicle coordinate system with the center of rotation of the front-end crane as the origin. The origin of the vehicle coordinate system is located at the center of rotation of the front-end crane, with the positive X-axis pointing towards the front of the vehicle, the positive Y-axis pointing to the driver's left, and the positive Z-axis pointing vertically upward.

[0052] The vehicle coordinate system calibration is performed using a laser tracker or total station, with the rotation center as the reference point, while the vehicle is stationary. In other embodiments of this application, the radar coordinates can be directly transformed to the vehicle coordinate system without going through the camera coordinate system, thereby reducing the cumulative error caused by the coordinate transformation process.

[0053] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0054] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for active anti-collision early warning control of front-mounted overhead power lines, characterized in that, Includes the following steps: Image data, point cloud data, and real-time attitude parameters of the front-end suspension are collected by visual sensors, radar sensors, and attitude sensors, respectively. The collected multi-source data is synchronized in time and unified in spatial coordinates. The image data and point cloud data are deeply fused to output the three-dimensional spatial position information and approach speed of the power line. A kinematic model of the boom end is established based on the real-time attitude parameters. A filtering algorithm is used to predict the trajectory of the boom end within a predetermined time period in the future. The minimum spatial distance and its rate of change between the predicted trajectory and the three-dimensional spatial position information are calculated. Based on the line voltage level, the approach speed, and the ambient humidity, a three-level safety threshold, including warning distance, deceleration distance, and locking distance, is dynamically calculated. The minimum spatial distance is compared with the three-level safety threshold, and control commands for audible and visual alarms, speed limits, or directional locking in dangerous directions are executed according to the comparison results. When the minimum spatial distance is greater than the locking distance and the duration exceeds the preset threshold, the speed limit and directional locking commands are automatically released.

2. The active anti-collision early warning control method for front-mounted overhead power lines according to claim 1, characterized in that, The deep fusion adopts a fusion strategy of radar-prioritized ranging and visual fine positioning: the point cloud data of the radar sensor provides the distance and approach speed of the wire, and the image data of the visual sensor provides the geometric direction and type of the wire after instance segmentation. Finally, a 3D bounding box of the line with confidence is output.

3. The active anti-collision early warning control method for front-mounted overhead power lines according to claim 1, characterized in that, The warning distance in the three-level safety threshold Deceleration distance and locking distance Calculated dynamically according to the following formula: ； in, This refers to the basic safety distance corresponding to the line voltage level. For speed coefficient, Humidity influence coefficient This represents the ambient humidity value.

4. The active anti-collision early warning control method for front-mounted overhead power lines according to claim 3, characterized in that, The real-time attitude parameters include the boom pitch angle. boom length Rotation angle and lifting height The kinematic model is represented as a position function of the boom end in the vehicle coordinate system. The velocity at the end of the boom is obtained by differentiating with respect to time. and acceleration .

5. The active anti-collision early warning control method for front-mounted overhead power lines according to claim 4, characterized in that, The filtering algorithm employs an extended Kalman filter to construct a state vector. The trajectory of the boom end effector is recursively predicted within the next 500ms to 1500ms.

6. The active anti-collision early warning control method for front-mounted overhead power lines according to claim 5, characterized in that, When the minimum spatial distance is greater than the locking distance and less than or equal to the deceleration distance, a speed limit command is sent to the vehicle controller of the front-end crane via the CAN bus. The vehicle controller limits the hydraulic pump's displacement according to the speed limit command, thereby reducing the maximum operating speed in the dangerous direction by 50%. The dangerous direction is determined based on the relative kinematic decoupling result between the predicted trajectory at the end of the boom and the three-dimensional bounding box of the line, and includes at least one of the boom pitching and descent direction, the boom extension direction, or the boom rotation direction.

7. The active anti-collision early warning control method for front-mounted overhead power lines according to claim 6, characterized in that, When the minimum spatial distance is less than or equal to the locking distance, a zero-speed locking command is sent to the corresponding hydraulic proportional valve controller via the CAN bus to lock only dangerous directional actions that may cause collision risks, while retaining the right to work in safe directions.

8. The active anti-collision early warning control method for front-mounted overhead power lines according to claim 7, characterized in that, The preset threshold is 1 second, and a hysteresis interval Δd is set; when the minimum spatial distance is greater than the sum of the locking distance and the hysteresis interval and the duration reaches 1 second, the speed limit and directional locking command are automatically released.

9. The active anti-collision early warning control method for front-mounted overhead power lines according to claim 8, characterized in that, Also includes: The watchdog timer monitors the controller's operating status, and the data health check module determines the validity of data from each sensor in real time. When the data interruption or invalid time of the radar sensor or attitude sensor exceeds 100ms, the warning distance, deceleration distance and locking distance are automatically amplified to the preset most conservative value, and sensor abnormality prompt information is output.

10. The active anti-collision early warning control method for front-mounted overhead power lines according to claim 9, characterized in that, The time synchronization adopts the PTP protocol to provide timing for all sensors through a high-precision clock source, ensuring that the timestamp error of image data, point cloud data and real-time attitude parameters is less than 1ms. The spatial coordinates are unified by converting all sensor data into a vehicle coordinate system with the front crane's rotation center as the origin.

Images