Narrow-body self-walking operation platform with hybrid power system

Abstract

The invention belongs to the technical field of electric overhead working trucks, and provides a narrow-body self-walking working platform with a hybrid power system, which comprises a walking platform, a working device and a control system. The walking platform adopts a narrow-body four-wheel-drive chassis and is provided with a hybrid power system of a fuel engine and an electric pump, and the control system integrates an environment sensing unit, a stability calculation unit, a track prediction unit, an early warning unit and a control decision unit. The environment sensing unit constructs a high-precision three-dimensional environment map; the stability calculation unit calculates the overall gravity center position and the dynamic stability boundary of the platform in real time; the track prediction unit establishes a dynamic hydraulic system response model; and the early warning unit adopts an artificial potential field method to calculate a safety avoidance path. And the control decision unit automatically adjusts the control instruction according to the risk type and degree. According to the invention, the organic combination of environment perception, dynamic stable control, prediction trajectory planning and electrified body safety protection is realized, and the operation safety and efficiency of the aerial work platform are obviously improved.

Description

Technical Field

[0001] This invention relates to the field of electric work vehicle technology, specifically to a narrow-body self-propelled work platform with a hybrid power system. Background Technology

[0002] Electric work vehicles are essential equipment for the power industry in carrying out maintenance, repair, and live-line work on transmission lines. With the rapid development of urban power grid construction and the continuous improvement of power system safety requirements, electric work vehicles are being used more and more widely in complex urban environments and live-line work scenarios.

[0003] Traditional aerial work platforms typically have a wide chassis to ensure stability and load-bearing capacity. However, this wide chassis design presents challenges in confined spaces such as narrow urban streets, factory workshops, and residential roads, making many work locations inaccessible due to narrow passageways, severely limiting the platform's application range and operational flexibility. How to achieve a narrower chassis design while maintaining platform stability, thereby improving the platform's mobility and accessibility in confined spaces, is a pressing technical challenge in aerial work platform design. Existing aerial work platforms primarily rely on operators' visual observation to determine the location and distance of obstacles in the surrounding environment. Human visual judgment has inherent limitations, including low accuracy, susceptibility to lighting and line-of-sight obstruction, and the inability to quantify distances. Especially at upward angles and long distances, distance estimation errors can be significant, making it difficult for operators to accurately determine the actual distance between the work platform and obstacles. While some advanced aerial work platforms are equipped with single-type sensors for obstacle detection, these sensors suffer from blind spots, limited accuracy, and inability to provide comprehensive coverage, making it difficult to construct accurate and complete three-dimensional environmental models. Insufficient environmental awareness forces operators to adopt overly conservative strategies when working in complex environments, keeping operating distances far beyond safety requirements, resulting in low workspace utilization and reduced work efficiency. Existing technologies typically employ fixed-threshold stability monitoring methods, setting a fixed tilt angle threshold and issuing an alarm when the platform's tilt angle exceeds this threshold. This simple threshold method cannot consider real-time changes in the platform's center of gravity during dynamic movement, nor can it dynamically adjust the stability judgment criteria based on different boom postures and load conditions. When the platform performs actions such as boom lifting, extension, or rotation, the center of gravity changes rapidly. Fixed-threshold monitoring methods cannot predict the potential stability deterioration caused by these movements, only issuing passive alarms when the platform is already close to instability. This passive response mode forces operators to frequently stop operations and wait for the platform to stabilize, repeatedly interrupting the work process and severely impacting work continuity and efficiency. The lack of real-time dynamic stability calculation capabilities based on multibody dynamics models prevents existing platforms from fully utilizing their operational capabilities while ensuring safety. Existing aerial work platform control systems typically employ a simple direct-mapping control method, where the position of the control lever directly corresponds to the opening of the hydraulic valve. Operator commands are executed directly without any risk assessment or intelligent adjustment. This control mode lacks the ability to predict future movement trajectories; the system cannot assess in advance the potential trajectory and risks that current commands may lead to, and can only passively issue alarms when the work platform actually approaches an obstacle or enters a hazardous area. This passive response control strategy requires operators to immediately adjust their operating direction to avoid risks upon receiving an alarm. Each adjustment interrupts the work process and prolongs operation time, resulting in poor operational smoothness and low work efficiency.The lack of proactive planning capabilities based on predicted trajectories prevents existing platforms from automatically calculating and executing optimal job paths in complex environments.

[0004] Furthermore, high-voltage live conductors pose a serious safety threat to workers and equipment, necessitating strict control of the safe distance between working equipment and live conductors. Current technology primarily relies on operators judging the location of live conductors based on conductor location diagrams provided by the power company and visual observation. This method cannot achieve precise positioning of live conductors, nor can it monitor changes in their position in real time. Because the three-dimensional coordinates of the live conductor cannot be accurately known, operators are forced to adopt extremely conservative operating strategies to ensure safety, maintaining the actual distance far exceeding the prescribed safe distance. This results in many work locations that could be safely reached being inaccessible due to excessive conservatism. While some platforms are equipped with live conductor detection devices, they lack integration with trajectory prediction and avoidance path planning. They can only passively alarm when the work bucket actually approaches the live conductor, unable to plan safe working paths to avoid it in advance.

[0005] To address the technical problems existing in the above-mentioned technologies, there is an urgent need to develop an aerial work platform system that integrates hybrid power drive, intelligent environmental perception, dynamic stability calculation, predictive trajectory planning, and accurate detection of charged parts. This system would improve the platform's operational safety, efficiency, and intelligence in complex power environments, and meet the increasingly stringent technical requirements of the power industry for aerial work equipment. Summary of the Invention

[0006] To address the problems existing in the background art, the present invention provides a narrow-body self-propelled work platform with a hybrid power system, including the following control method: S1: High-precision 3D environmental modeling. LiDAR sensors, a binocular vision system, and an ultrasonic sensor array are used to synchronously acquire environmental data. Multi-source data undergoes time synchronization and coordinate system transformation. Point cloud registration algorithms and data fusion processing are used to construct a real-time 3D environmental map containing obstacle geometry, position coordinates, and ground slope information. S2: Dynamic stability envelope calculation. Real-time reading of platform attitude angle data from IMU sensors, bucket load data from load sensors, boom attitude data from encoders at each joint, and ground tilt angle data are used. This data is input into a multibody dynamics model to calculate the overall center of gravity coordinates of the platform and the support reaction forces at each support point. The dynamic stability boundary is calculated based on the overturning moment balance condition. S3: Predictive trajectory planning and risk assessment. Control commands from the operator are received and converted into target motion parameters for each actuator. Based on the platform's dynamic state equations and the hydraulic system's response characteristics, a numerical integration method is used to calculate the position coordinate sequence of the bucket's center point within the next 1-3 seconds. The predicted trajectory is then compared with the environmental data. The diagram compares and verifies the location of obstacles, charged bodies detected by electric field sensors, and dynamic stability boundaries to identify and mark trajectory segments with risks of collision, approaching charged bodies, or instability; S4: Charged body approach warning, reads spatial electric field intensity distribution data measured by electric field sensor array to infer the location of charged bodies, calculates the safe distance between each point on the predicted trajectory and the charged body, and determines whether to enter the warning area, danger area, or prohibited area. When entering the warning or danger area, the audible and visual alarm device is activated and the artificial potential field method is used to calculate the suggested avoidance path; S5: Cooperative collision avoidance and dynamic stability control, determines the intervention level according to the risk type and risk level. When in the first-level warning, a prompt is issued to the operator through the display device and tactile feedback device. When in the second-level intervention, the hydraulic valve control current is automatically reduced to slow down the boom movement speed. When in the third-level avoidance, the control command is automatically corrected and the deflection angle is superimposed on the original movement direction to guide the boom to move along the avoidance path. When in the emergency braking, the hydraulic valve control signal in the dangerous direction is immediately cut off and the chassis automatic leveling function is activated.

[0007] In a preferred embodiment, step S1 specifically includes: S11: Activating the LiDAR sensor to perform a surround scan, setting the scan frequency to 10-20Hz, the sensor emitting laser pulses and receiving reflected signals, calculating the distance value based on the time of light flight, acquiring three-dimensional point cloud data including azimuth angle, pitch angle, and distance value, and storing the point cloud data in a buffer; S12: Activating the left and right cameras of the binocular vision system to simultaneously capture images, performing grayscale processing on the left and right images, and using the SGBM stereo matching algorithm to calculate the disparity value of the same feature point in the left and right images. Calculate the depth distance of the point based on the parallax value. ;in For depth distance; The baseline distance of the binocular cameras; The focal length of the camera; The parallax value is used; the depth data is mapped to the RGB image to generate a color depth map; S13: each sensor in the ultrasonic sensor array is activated to emit ultrasonic pulses in sequence, receive the ultrasonic signals reflected back from the obstacle, and measure the round-trip time of the ultrasonic waves. Calculate the distance to the obstacle based on the speed of sound. ;in Distance to the obstacle; The speed of sound is taken as 340 m / s; 1 represents the echo time; 2 represents the round-trip coefficient; acquire the location data of nearby obstacles within a range of 0.2-5 meters; S14: align the LiDAR point cloud data, visual depth data, and ultrasonic distance data with timestamps, remove abnormal data exceeding the detection range, transform the data from each sensor coordinate system to the platform center coordinate system using a coordinate transformation matrix, and perform data fusion using an extended Kalman filter algorithm. The state update equation is: ;in This is the fused state estimation vector; This is a predicted value based on the state at the previous time step; The Kalman gain matrix; This represents the vector of sensor observations at the current moment. The observation matrix is ​​used; the fused environmental point cloud data is obtained through iterative calculation; S15: the fused point cloud data is processed by voxel rasterization, the space is divided into cubic grids with a side length of 5-10cm, the number of point clouds in each grid is counted, the grids containing point clouds are marked as occupied, the grids with a number of point clouds exceeding the threshold are identified as obstacle grids, the boundary patches of the obstacle grids are extracted, the ICP point cloud registration algorithm is used to match and align the current frame point cloud with the historical frame point cloud, the 3D environment map is updated frame by frame, and a real-time environment map containing obstacle geometry, 3D coordinate position and ground slope angle information is generated.

[0008] In the preferred embodiment, step S2 specifically includes: S21: Read the triaxial acceleration data output by the IMU sensor. , , Three-axis angular velocity data , , and triaxial magnetic field data , , The platform's pitch angle is calculated by fusing acceleration and angular velocity data using a complementary filtering algorithm. and roll angle The formula for calculating the pitch angle is: ;in The pitch angle is the current angle; α is the filter coefficient, ranging from 0.95 to 0.98. The pitch angle at the previous moment; y-axis angular velocity; dt is the sampling time interval; It is the arctangent function in the four quadrants; The acceleration is along the x-axis; The acceleration is along the y-axis; Z-axis acceleration; sqrt is the square root operator; roll angle. The platform's real-time attitude angle data was obtained using the same method; S22: Read the voltage signal output by the load sensor at the bottom of the work bucket. The voltage signal is converted into load weight through a calibration curve. ;in This refers to the load weight; These are calibration coefficients; The sensor output voltage; Zero-point offset; obtain real-time load data within the work bucket; S23: read the number of pulses output by the angle encoder at the connection between the main boom and the turntable. Calculate the main boom angle based on the encoder resolution. ;in The angle of elevation of the main arm relative to the horizontal plane; For encoder pulse counting; 360° represents the angle corresponding to one revolution. This refers to the total number of pulses per revolution of the encoder; it also reads the number of pulses output from the angle encoder at the connection point between the auxiliary arm and the main arm. Calculate the arm angle Read the number of pulses output by the displacement encoder of the telescopic mechanism. Calculate the telescopic length Read the number of pulses output by the angle encoder of the rotary mechanism. Calculate the rotation angle S24: Obtain real-time attitude data of each joint of the boom; S25: Treat each component of the platform as a rigid body, establish a multibody dynamics coordinate system, and define the chassis mass with the platform center O as the origin. and its centroid coordinates Define the quality of the turntable and its centroid coordinates Define the mass of the main boom and its centroid coordinates Define the mass of the auxiliary arm and its centroid coordinates Define the quality of the work bucket and its centroid coordinates Define load mass and its centroid coordinates Based on the boom's attitude angle and telescopic length, the position of the center of gravity of each component in the platform coordinate system is calculated step by step using a homogeneous coordinate transformation matrix, thus calculating the overall center of gravity coordinates of the platform. , ;in The x-coordinate of the centroid; The y-coordinate of the centroid; Σ is the summation operator; Let the mass of the i-th component be denoted as . Let X be the centroid coordinate of the i-th component; Let the Y-coordinate of the center of gravity of the i-th component be denoted; obtain the overall center of gravity position of the platform; S25: construct a support polygon with the four support points of the platform as vertices, determine whether the projection point of the center of gravity is inside the support polygon, calculate the distance from the projection point of the center of gravity to each side of the support polygon, and take the minimum distance as the stability margin. Calculate the overturning moment at each support point. ;in The overturning moment; The total gravity of the platform; The horizontal distance from the center of gravity to the overturning axis; calculate the overturning moment. ;in To resist overturning moment; For lateral gravity; The distance from the lateral center of gravity to the overturning axis; Determine To determine if the condition is valid, draw the dynamic stability boundary polygon under the current working condition, delineate the safe working area with the centroid projection point as the center and the stability margin as the radius, and update and output the dynamic stability envelope data in real time.

[0009] In the preferred embodiment, step S3 specifically includes: S31: Receives control command signals from the remote controller and parses the main boom lifting lever position values ​​contained in the commands. Telescopic operating lever position value Rotary control lever position value and travel control lever position value The lever position value is converted into the target velocity of the corresponding actuator through linear mapping. ;in For target speed; This is the maximum speed of the actuator; S32: Establish the platform dynamics state equations and define the state vector. (Note: The original text contains some inconsistencies and inconsistencies. A more accurate translation would require the full context.) ; in Main arm angle, For the auxiliary arm angle, For the rotation angle, For telescopic length, Indicates angular velocity. Indicates linear velocity. Indicates transpose; defines the control input vector. ,in Represent the control current of the hydraulic valve; establish the state transition equation. ;in The prediction time step is set to 0.1s; S33: Based on the characteristics of the hydraulic system, establish a model relating the actuator speed and the valve control current. (This is a nonlinear state transition function.) ;in For actuator speed; This refers to the flow coefficient of the proportional valve. For valve control current; This is the square root operator; This is the difference between the hydraulic pump output pressure and the return oil pressure. Let the hydraulic oil density be denoted by ; considering the first-order inertial delay characteristics of the hydraulic system, a transfer function is used. Describe the response process, where The time constant is set to 0.1-0.3 s. For the Laplace operator; S34: with the current state As initial conditions, the target velocity command is substituted into the state equation, and numerical integration is performed using the fourth-order Runge-Kutta method to calculate step by step. , ... The state value at time t, where For prediction durations of 1-3 seconds, extract the position coordinates of the center point of the working bucket at each moment. To form a sequence of predicted trajectory coordinates ; S35: Extracting boundary point cloud sets of obstacles from a 3D environment map Extracting the position coordinates of charged bodies from electric field sensor data For each point on the predicted trajectory The Euclidean distances between it and all obstacle points are calculated as follows: ;in Let be the distance between the i-th trajectory point and the j-th obstacle point; take the minimum distance among all obstacle points. ,in This is the minimum value operator; it calculates the distance between the trajectory point and the charged body: Extracting stability margin from dynamic stability boundary data Determine whether the condition is met. , , ,in , , These are the obstacle safety distance, the charged body safety distance, and the stability safety margin, respectively. Trajectory points that do not meet any of the safety conditions are marked as risk points. Continuous risk points are combined into risk trajectory segments. The risk type and risk level are recorded, and the trajectory risk assessment results are output.

[0010] In the preferred method of inspection, step S4 specifically includes: S41: Read the voltage signal output from the electric field sensor array installed on the boom and bucket. It is converted into electric field strength through the calibration formula. ;in Let be the electric field strength measured by the i-th sensor; These are calibration coefficients; S42: Based on the relationship between electric field strength and distance, the position of the charged body is inferred using a point charge model, and the electric field strength equation is established. (This is the output voltage of the sensor; data on the electric field intensity distribution in the space surrounding the platform is acquired.) ;in To measure the electric field strength at the point; Let the electrostatic constant be taken as N·m 2 / C 2 ; This is the equivalent charge of a charged body; To measure the distance from the point to the charged body; 2 The square operator is used; a system of equations is formed from measurement data from multiple sensors, and the position coordinates of the charged body are solved using the least squares method. and voltage level S43: Defines three levels of safe distance thresholds and warning distances. m corresponds to a low-risk area, and the danger distance is... m corresponds to a medium-risk area, and the prohibited distance is... m corresponds to a high-risk area; S44: Calculate the points on the predicted trajectory. With charged body Distance between ;in The distance from the trajectory point to the charged body; determine The distance range in which it is located, when The audible and visual alarm device is activated at certain times, and the alarm frequency is set. Hz, when Increase alarm frequency to Hz, while marking the location of charged objects and danger zones in red on the AR display; S45: Calculate the avoidance path using the artificial potential field method, defining the target attraction potential field. ;in The strength of the attractive potential field; The attraction gain coefficient is set to 1-5; The distance from the current position to the target position; 0.5 is a coefficient; the repulsive potential field of the charged body is defined. when , when ;in The strength of the repulsive potential field; The repulsion gain coefficient is set to 10-50; The distance to the charged body; For early warning distance; calculate the composite potential field. ;in The total potential field strength; This is the addition operator; it calculates the direction of the potential field gradient. ;in The gradient vector; It is a partial derivative operator; , , It is a unit vector along the three coordinate axes; It is a vector dot product operator; it generates a sequence of avoidance path points along the negative gradient direction, and outputs the avoidance path to the AR display in the form of navigation arrows and numerical displays to guide the operator to adjust the operation to avoid the charged body.

[0011] In a preferred embodiment, step S5 specifically includes: S51: Based on the risk types and risk levels identified in S3 and S4, establish risk level assessment criteria, when the minimum safe distance... satisfy It was determined to be a Level 1 warning at that time. When it was determined to be a level two intervention, When it is determined to be a level three avoidance, when The time was determined to be emergency braking; among which S52: When at Level 1 warning, a display command is sent to the AR display, marking the risk location in yellow on the display screen and drawing an arrow pointing from the current work bucket position to the suggested safe direction. Simultaneously, a control signal is sent to the haptic feedback device in the remote control, driving the vibration motor to generate a warning vibration at a frequency of 1Hz and an amplitude of 50%, alerting the operator to potential risks. S53: When at Level 2 intervention, the current hydraulic valve control current is read. Calculate the adjustment coefficient ;in For adjustment coefficients; Minimum safe distance; The safety distance threshold is 0.3, 0.4, and 0.5, which are constant coefficients. The adjusted control current is then calculated. ;in For the new control current; As the control command output of the hydraulic valve, the boom movement speed is reduced to 30-70% of its original speed, extending the operator's reaction time; S54: When in the third-level avoidance state, the coordinates of the next target point are extracted from the avoidance path calculated by S4. Calculate the current working bucket position Direction vector to the target point ;in Given the target direction vector; calculate the motion direction vector corresponding to the original control command. Calculate the required deflection angle ;in This refers to the deflection angle; It is the inverse cosine function; This is the vector dot product operator; This is a vector magnitude operator; it decomposes the deflection angle into individual actuators and modifies the control instructions for each actuator. ;in This is the corrected control current for the j-th actuator; This is the original control current; To correct the current increment; output the corrected control command to make the boom move automatically along the avoidance path while maintaining the operator's control over the overall direction of movement; S55: When in emergency braking mode, immediately identify the direction of movement that causes the risk and send a zero current command to the hydraulic valve corresponding to that direction. mA cuts off the hydraulic oil flow to the actuator, bringing the movement in that direction to a complete stop within 0.5 seconds; it then determines whether other movement directions are safe, maintaining the original control command for safe directions or allowing the operator to continue control; simultaneously, it sends a leveling command to the chassis leveling system and reads the height sensor data from the four support points. , , , Calculation platform tilt angle ;in The tilt angle; It is the arctangent function; This is the maximum value operator; This is the minimum value operator; To adjust the spacing between support points; control the leveling hydraulic cylinder to adjust the height of each support point, so that... This enables the platform to quickly return to a horizontal state, eliminating the risk of overturning.

[0012] The present invention provides a narrow-body self-propelled work platform with a hybrid power system, comprising a walking platform, a work device, and a control system; The traveling platform includes a wheeled four-wheel drive chassis, on which are mounted a fuel engine, a fuel tank, an electric pump, and a battery pack. The outputs of the fuel engine and the electric pump are connected to a hydraulic distribution valve group via hydraulic lines to provide hydraulic power to the traveling platform and the working device. The working device includes a turntable, which is mounted on the wheeled four-wheel drive chassis via a slewing bearing. The turntable is movably connected to the main boom via a first hinge shaft. A hydraulic mechanism I is provided between the turntable and the main boom. The end of the main boom away from the turntable is movably connected to the auxiliary boom via a second hinge shaft. A hydraulic mechanism II is provided between the main boom and the auxiliary boom. The end of the auxiliary boom away from the main boom is movably connected to the work bucket via a slewing mechanism. The control system includes a control box and a remote controller. The control box contains an environmental perception unit, a stability calculation unit, a trajectory prediction unit, an early warning unit, and a control decision unit.

[0013] Furthermore, the environmental perception unit includes a LiDAR sensor, a binocular vision system, an ultrasonic sensor array, and an environmental data processor. The LiDAR sensor is mounted on the top of the turntable, the binocular vision system is mounted on the front face of the work bucket, and multiple ultrasonic sensors of the ultrasonic sensor array are respectively mounted on the four sides of the four-wheel drive chassis. The environmental data processor is connected to the LiDAR sensor, the binocular vision system, and the ultrasonic sensor array, and is also connected to the trajectory prediction unit for transmitting 3D environmental map data. The stability calculation unit includes an IMU sensor, a load sensor, an angle encoder, and a stability calculation processor. The IMU sensor is mounted at the geometric center of the four-wheel drive chassis, the load sensor is mounted below the bottom plate of the work bucket, and the angle encoders include a main boom angle encoder mounted on the first hinge shaft, a secondary boom angle encoder mounted on the second hinge shaft, a slewing angle encoder mounted on the slewing bearing, and a telescopic length encoder mounted on the main boom telescopic mechanism. The stability calculation processor is connected to the IMU sensor, the load sensor, and each angle encoder, and is also connected to the trajectory prediction unit for transmitting dynamic stability boundary data. The trajectory prediction unit includes a remote control signal receiving module, a dynamic model calculation module, and a trajectory generation module. The remote control signal receiving module... The module connects to the remote controller via wireless communication to receive control commands. The dynamic model calculation module connects to the remote controller signal receiving module, the environmental data processor, and the stability calculation processor to receive the current state data of the platform. The trajectory generation module connects to the dynamic model calculation module and to the early warning unit to transmit the predicted trajectory coordinate sequence. The early warning unit includes an electric field sensor array, an electric field data processor, and an avoidance path planning module. Multiple electric field sensors of the electric field sensor array are respectively installed on the outer surface of the auxiliary arm and around the working bucket. The electric field data processor is connected to the electric field sensor array signal. The avoidance path planning module connects to the electric field data processor and the trajectory generation module and to the control decision unit to transmit risk assessment results and suggested avoidance paths. The control decision unit includes a risk level determination module, an intervention strategy selection module, and a control command correction module. The risk level determination module connects to the avoidance path planning module to receive risk assessment results. The intervention strategy selection module connects to the risk level determination module. The control command correction module connects to the intervention strategy selection module and the remote controller signal receiving module to receive the original control commands and intervention strategies. The output of the control command correction module connects to the control terminals of each hydraulic valve in hydraulic mechanism I, hydraulic mechanism II, and the slewing mechanism to output the corrected control commands.

[0014] Furthermore, the front and rear axles of the wheeled four-wheel drive chassis are each equipped with an independent steering mechanism. Each steering mechanism includes a steering hydraulic cylinder and a steering control valve. The control terminal of the steering control valve is electrically connected to the control decision unit within the control box. The inlet and outlet ports of the steering hydraulic cylinder are connected to the output port of the steering control valve via hydraulic lines. The wheeled four-wheel drive chassis also includes a horizontal tilt sensor, the signal output terminal of which is connected to the stability calculation unit within the control box. The signal output terminal of the LiDAR sensor is connected to the environmental data processor via a CAN bus. The binocular vision system includes a left camera and a right camera, the signal output terminals of which are connected via a U... The SB interface is connected to the environmental data processor; each ultrasonic sensor in the ultrasonic sensor array is connected to the environmental data processor via a serial communication interface; the environmental data processor is connected to the trajectory prediction unit via a data interface; the IMU sensor is connected to the stability calculation processor via a digital interface; the load sensor is connected to the analog-to-digital converter input of the stability calculation processor via an analog signal line; the main boom angle encoder, auxiliary boom angle encoder, slewing angle encoder, and telescopic length encoder are all connected to the stability calculation processor via incremental encoder interfaces; the stability calculation processor exchanges data with the trajectory prediction unit via a shared data channel. The remote controller communicates bidirectionally with the remote controller signal receiving module via a wireless communication frequency band. The remote controller's joystick sends control commands to the remote controller signal receiving module through a signal processing circuit. The dynamic model calculation module simultaneously receives output data from the remote controller signal receiving module, the environmental data processor, and the stability calculation processor via an internal data bus. The output of the trajectory generation module is connected to the early warning unit via a data channel. Each electric field sensor in the electric field sensor array is connected to the multi-channel analog-to-digital converter input of the electric field data processor via analog signal lines. The output of the electric field data processor is connected to the avoidance path planning module via a data bus. The avoidance path planning module... The output terminal is simultaneously connected to the display devices of the control decision unit and the remote controller via a parallel data channel; the risk level determination module outputs a risk level signal to the intervention strategy selection module based on the input risk assessment data; the intervention strategy selection module selects the corresponding intervention strategy based on the risk level and outputs the strategy code to the control command correction module; the control command correction module processes the original control command and the intervention strategy to generate a corrected control command, which is then output to the hydraulic valve control terminals of hydraulic mechanism I, hydraulic mechanism II, and slewing mechanism respectively; when the risk level reaches the emergency braking level, the risk level determination module directly sends an emergency start signal to the leveling hydraulic cylinder of the four-wheel drive chassis.

[0015] Furthermore, the output of the control command correction module is connected to the electromagnet of the hydraulic valve in hydraulic mechanism I via a PWM signal line to control the opening degree of the hydraulic valve; the output of the control command correction module is connected to the electromagnet of the hydraulic valve in hydraulic mechanism II via a PWM signal line to control the opening degree of the hydraulic valve; the output of the control command correction module is connected to the electromagnet of the hydraulic valve in the slewing mechanism via a PWM signal line to control the opening degree of the hydraulic valve; an angular displacement sensor is provided on the working bucket, and the angular displacement sensor is mounted on the output shaft of the slewing mechanism. The signal output of the angular displacement sensor is connected to the input of the stability calculation processor to monitor the slewing angle of the working bucket; the auxiliary boom is made of FRP insulation. Made of insulating material, the outer surface of the auxiliary arm is connected to the main arm via an insulating sleeve. The remote control contains a display device and a tactile feedback device. The input terminal of the display device is wirelessly connected to the output terminal of the intervention strategy selection module to receive display data. The display data includes risk location coordinates, risk level indicators, and suggested operation direction vectors. The display device converts the data into a graphical interface display through a graphics processing unit. The tactile feedback device includes a vibration motor and a drive circuit. The input terminal of the drive circuit is wirelessly connected to the output terminal of the intervention strategy selection module to receive vibration control signals. The output terminal of the drive circuit is connected to the power supply terminal of the vibration motor to drive the vibration motor to generate tactile feedback.

[0016] The beneficial effects achieved by this invention are as follows: First, this invention designs a hybrid power system and multi-sensor environmental perception technology, enabling the aerial work platform to operate safely in complex electrical environments. The platform adopts a hybrid power configuration of a fuel engine and an electric pump, with the two power sources automatically switching or operating in parallel according to working conditions. The narrow-body self-propelled chassis design allows the platform to smoothly pass through narrow passages and confined spaces, significantly expanding the applicable range of working scenarios, especially suitable for space-constrained environments such as urban streets and factory workshops. The environmental perception unit integrates three sensors with different ranging principles: a LiDAR sensor, a binocular vision system, and an ultrasonic sensor array. Through multi-source data fusion using an extended Kalman filter algorithm, the constructed high-precision 3D environmental map accurately reflects the geometry, location coordinates, and ground slope information of obstacles, providing a reliable environmental cognitive foundation for trajectory planning and risk assessment.

[0017] Secondly, this invention designs a dynamic stability envelope calculation technology. By establishing a multibody dynamics model, it calculates the overall center of gravity position of the platform and the support reactions at each support point in real time, dynamically generating a stability boundary based on the overturning moment balance condition. The system integrates platform attitude data measured by IMU sensors, bucket load data acquired by load sensors, and boom attitude data output by encoders at each joint into the dynamic model. A homogeneous coordinate transformation matrix is ​​used to calculate the position of the center of gravity of each component in the platform coordinate system step by step, thereby accurately obtaining the overall center of gravity coordinates of the platform. The system determines the stability margin by constructing support polygons and calculating the distances from the center of gravity projection point to each side of the polygons. Simultaneously, it evaluates the stability state by comparing the magnitude of the overturning moment and the anti-overturning moment. Compared to the traditional fixed threshold judgment method, this real-time dynamic calculation stability assessment method can more accurately reflect the true stability state of the platform under various working conditions, allowing the system to fully utilize the platform's operational capabilities while ensuring safety, avoiding reduced operational efficiency due to excessive conservatism.

[0018] Third, this invention establishes a dynamic state equation including state variables such as the main boom angle, auxiliary boom angle, slewing angle, and telescopic length. It uses a fourth-order Runge-Kutta numerical integration method to calculate the position coordinate sequence of the work bucket center point within the next one to three seconds, forming a predicted trajectory. This predicted trajectory is comprehensively compared and verified with obstacle boundary point clouds in the environmental map, the positions of charged bodies detected by electric field sensors, and dynamic stability boundary data, identifying and marking trajectory segments with collision risks, risks of approaching charged bodies, or risks of instability. This proactive trajectory prediction technology transforms the control system from passive response to active planning, assessing the possible motion trajectory and potential risks of the control command the instant the operator issues it, thereby achieving early warning and proactive avoidance, significantly reducing the number of operational adjustments, and improving operational smoothness and continuity.

[0019] Fourth, this invention addresses the specific safety requirements of power operation environments by designing a live-body proximity warning system based on an electric field sensor array. The system deploys multiple non-contact electric field sensors on the outer surface of the auxiliary boom and around the work platform to measure the electric field intensity distribution in the surrounding space. It establishes a set of electric field intensity equations using a point charge model and employs the least squares method to deduce the three-dimensional coordinates and voltage level of the live-body. The system establishes a three-tiered safety distance threshold system: warning distance, danger distance, and prohibited distance, corresponding to graded protection strategies for low-risk, medium-risk, and high-risk areas. When the predicted trajectory point enters the warning or danger zone, the system uses an artificial potential field method to calculate the avoidance path, defining the target's attractive potential field and the live-body's repulsive potential field. The optimal avoidance path is generated by calculating the gradient direction of the synthesized potential field. This technical solution, combining electric field detection with potential field-based avoidance path planning, enables the platform to accurately locate the live-body and automatically plan a safe path, maximizing the use of the allowable working space while meeting safety distance requirements, achieving an optimal balance between safety and operational efficiency.

[0020] Fifth, this invention designs a graded intervention collaborative collision avoidance and dynamic stability control strategy, implementing differentiated control interventions based on risk type and degree. The system divides risk levels into four levels: Level 1 warning, Level 2 intervention, Level 3 avoidance, and emergency braking, and takes corresponding control measures for each level. During Level 1 warning, a prompt is issued to the operator via an augmented reality display and haptic feedback device, maintaining the operator's complete control over the platform. During Level 2 intervention, the system automatically adjusts the hydraulic valve control current, reducing the boom's movement speed to extend the operator's reaction time. During Level 3 avoidance, the system extracts the target point coordinates from the avoidance path, calculates the required deflection angle, and corrects the control commands of each actuator, causing the boom to automatically move along the avoidance path while maintaining the operator's control over the overall movement direction. During emergency braking, the system immediately sends a zero-current command to the hydraulic valve in the dangerous direction, cutting off the hydraulic oil flow to that actuator, and simultaneously activates the chassis leveling system to quickly restore the platform to a level state. This graded intervention strategy implements different levels of control intervention at different risk levels, ensuring both safety and operational flexibility, significantly improving the system's intelligence and practicality. Attached Figure Description

[0021] Figure 1 This is a front view structural diagram of a narrow-body self-propelled work platform with a hybrid power system according to the present invention; Figure 2 This is a front view structural diagram of a narrow-body self-propelled work platform with a hybrid power system according to the present invention; Figure 3 This is a top view of a narrow-body self-propelled work platform with a hybrid power system according to the present invention. Figure 4This is a structural diagram of a narrow-body self-propelled work platform with a hybrid power system according to the present invention. Figure 5 This is a comprehensive comparison chart of the operational efficiency of the embodiments and various comparative examples, including completion time, number of adjustments, and response time; Figure 6 Examples and comparative examples show the control accuracy of safe distances, including a comparison chart of distance to charged objects, control accuracy, distance to obstacles, and number of collisions. Figure 7 The stability performance of the embodiment and each comparative platform, including tilt angle, stability margin, and alarm count comparison chart; Figure 8 The environmental perception capabilities of the embodiments and comparative examples are compared, including sensor configuration, update frequency, and detection accuracy.

[0022] Numbering on the map: 1. Walking platform; 11. Wheeled four-wheel drive chassis; 111. Steering mechanism; 12. Combustion engine; 13. Fuel tank; 14. Electric pump; 15. Battery pack; 2. Working device; 21. Turntable; 22. Main boom; 23. Hydraulic mechanism I; 24. Auxiliary boom; 25. Hydraulic mechanism II; 26. Slewing mechanism; 27. Working bucket; 28. Angular displacement sensor; 3. Control system; 31. Control box; 32. Remote control. Detailed Implementation

[0023] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. In addition, the forms of the various structures described in the following embodiments are merely illustrative. The present invention is not limited to the structures described in the following embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Example 1, this example refers to Figures 1-4 A narrow-body self-propelled work platform with a hybrid power system was designed, comprising three main parts: a walking platform 1, a work device 2, and a control system 3. The three parts work together to realize the intelligent and safe operation function of the aerial work platform.

[0025] The walking platform 1 uses a wheeled four-wheel drive chassis 11 as its mobile carrier. The width of this chassis is controlled within 1.2 meters to meet the requirements of narrow-body operations. The front and rear axles of the chassis are equipped with independent steering mechanisms 111. Each steering mechanism 111 includes a steering hydraulic cylinder and a steering control valve. The steering hydraulic cylinder is preferably a double-acting hydraulic cylinder with a cylinder diameter of 50 mm and a stroke of 200 mm. The steering control valve is preferably an electro-hydraulic proportional directional valve with a rated flow rate of 40 liters / minute. The control terminal of the steering control valve is electrically connected to the control decision unit in the control box 31 via a CAN bus. The inlet and outlet ports of the steering hydraulic cylinder are connected to the output port of the steering control valve via high-pressure hoses, realizing independent steering control of the front and rear wheels and significantly improving the platform's mobility in narrow spaces. A fuel engine 12 is mounted on top of the wheeled four-wheel drive chassis 11. It is preferably a four-cylinder turbocharged fuel engine with a power of 55 kW and a speed range of 1500-3000 rpm. The fuel engine 12 is connected to the hydraulic pump via a flexible coupling, providing the main power source for the entire system. The fuel tank 13, with a capacity of 60 liters, is installed at the rear of the chassis and is made of anti-static material with an internal anti-sway structure. The electric pump 14, installed next to the fuel engine 12, is preferably a gear pump driven by a permanent magnet synchronous motor, with a power of 15 kW and a flow rate of 80 liters / minute. The input of the electric pump 14 is connected to the battery pack 15 via a power cable. The battery pack 15 uses lithium iron phosphate batteries with a total capacity of 20 kWh and a nominal voltage of 96 volts. It is installed in a sealed battery compartment in the middle of the chassis and has a built-in battery management system (BMS) to monitor battery temperature, voltage, current, and other parameters in real time. The outputs of both the fuel engine 12 and the electric pump 14 are connected to a hydraulic distribution valve assembly via hydraulic lines. The hydraulic distribution valve assembly includes a main relief valve, a flow divider valve, and a multi-way directional valve, which can automatically switch power sources or achieve parallel operation of two power sources according to working conditions, providing stable hydraulic power to the traveling platform 1 and the working device 2. This hybrid power design ensures both long-term operating capability and energy conservation and environmental protection.

[0026] The four-wheel drive chassis 11 is also equipped with a horizontal tilt sensor, preferably a dual-axis tilt sensor with a measurement range of ±30 degrees and an accuracy of 0.01 degrees. This sensor is installed at the geometric center of the chassis, and its signal output is connected to the stability calculation unit in the control box 31 via an RS485 interface, enabling real-time monitoring of the chassis's tilt status. Leveling hydraulic cylinders are located at each of the four corners of the chassis, with a cylinder diameter of 80 mm and a stroke of 300 mm. Each leveling hydraulic cylinder is equipped with a displacement sensor and a pressure sensor. By independently controlling the extension and retraction of each leveling hydraulic cylinder, the platform's automatic leveling function is achieved, ensuring stability during operation.

[0027] The core of the working device 2 is the turntable 21. The turntable 21 is mounted at the center of the wheeled four-wheel drive chassis 11 via a slewing bearing. The slewing bearing is preferably a single-row four-point contact ball bearing with a diameter of 1200 mm and a load capacity of 50 tons. The inner ring of the slewing bearing is fixed to the chassis, and the outer ring is connected to the turntable 21. It is driven by a slewing hydraulic motor to achieve 360-degree continuous rotation. The main boom 22 is movably connected to the turntable 21 via a first hinge shaft. The first hinge shaft is made of high-strength alloy steel with a diameter of 100 mm and self-lubricating bearings at both ends. The main boom 22 adopts a box-beam structure welded from high-strength steel plates, with a length of 12 meters, a cross-sectional dimension of 400×300 mm, a wall thickness of 8 mm, and internal reinforcing ribs to improve structural strength. A hydraulic mechanism I23 is provided between the turntable 21 and the main boom 22, including two parallel luffing hydraulic cylinders with a cylinder diameter of 125 mm, a stroke of 1500 mm, and a working pressure of 25 MPa. The cylinder barrel end of the hydraulic cylinder is hinged to the turntable 21, and the piston rod end is hinged to the middle of the main boom 22. The lifting and lowering action of the main boom 22 is achieved by the extension and retraction of the hydraulic cylinder.

[0028] The main boom 22 has an internal telescopic mechanism with a multi-stage sleeve structure, including a basic boom and two telescopic boom sections. The telescopic booms are guided by sliders and driven by built-in telescopic hydraulic cylinders, with a maximum telescopic length of 8 meters. The telescopic hydraulic cylinders are equipped with telescopic length encoders, preferably wire-type displacement sensors with a range of 10 meters and an accuracy of 0.1 mm. The end of the main boom 22 furthest from the turntable 21 is movably connected to the auxiliary boom 24 via a second hinge shaft. The structure of the second hinge shaft is similar to that of the first hinge shaft. The auxiliary boom 24 is 6 meters long and made of FRP insulating material, which has excellent electrical insulation performance with an insulation resistance greater than 500 megohms, effectively preventing electric shock accidents. The outer surface of the auxiliary boom 24 is connected to the main boom 22 by an insulating isolation sleeve made of epoxy resin material with a thickness of 20 mm. A hydraulic mechanism II 25 is provided between the main boom 22 and the auxiliary boom 24, including a leveling hydraulic cylinder with a cylinder diameter of 80 mm and a stroke of 800 mm. This hydraulic cylinder can automatically adjust the angle of the auxiliary boom 24 to keep the working bucket 27 always in a horizontal state.

[0029] The end of the auxiliary boom 24 furthest from the main boom 22 is movably connected to the work bucket 27 via a slewing mechanism 26. The slewing mechanism 26 includes a small slewing bearing and a slewing hydraulic motor, with a slewing range of ±90 degrees. The work bucket 27 is welded from aluminum alloy profiles, measuring 2000×800×1100 mm, with a rated load of 230 kg. It has an anti-slip patterned plate at the bottom and is surrounded by guardrails 1100 mm high, with a foldable extension platform on top of the guardrails. An angular displacement sensor 28, preferably a magnetic encoder with 12-bit resolution, is installed on the output shaft of the slewing mechanism 26. Its signal output is connected to the input of a stability calculation processor via a shielded cable, used to accurately monitor the slewing angle of the work bucket and ensure operational positioning accuracy.

[0030] The main body of the control system 3 is the control box 31, a sealed enclosure with IP65 protection rating, measuring 600×500×300 mm. It is installed on the side of the turntable 21 and features a modular design, comprising five functional modules: an environmental perception unit, a stability calculation unit, a trajectory prediction unit, an early warning unit, and a control decision unit. The environmental perception unit includes a LiDAR sensor, a binocular vision system, an ultrasonic sensor array, and an environmental data processor. The LiDAR sensor preferably uses a 16-line lidar with a detection range of 100 meters, a vertical field of view of 30 degrees, a horizontal field of view of 360 degrees, and a scanning frequency of 10 Hz. It is installed at the top center of the turntable 21 and communicates with the environmental data processor via a CAN bus. The binocular vision system includes two industrial cameras with a resolution of 1920×1080 pixels, a frame rate of 30 frames per second, and a baseline distance of 300 mm. They are installed on both sides of the front face of the workbench 27, and the signal outputs of the left and right cameras are connected to the environmental data processor via USB 3.0 interfaces. The ultrasonic sensor array comprises eight ultrasonic sensors with a detection range of 0.2-5 meters, a beam angle of 15 degrees, and an operating frequency of 40 kHz. Two sensors are mounted on each of the four sides of the four-wheel drive chassis 11. Each ultrasonic sensor is connected to the environmental data processor via an RS485 serial communication interface. The environmental data processor is an embedded industrial computer with a quad-core ARM Cortex-A72 processor, a clock speed of 2.0 GHz, and 8 GB of memory. It features multiple communication interfaces and connects to the trajectory prediction unit via a gigabit Ethernet interface, enabling real-time processing of multi-source sensor data and the construction of a 3D environmental map.

[0031] The stability calculation unit includes an IMU sensor, a load sensor, an angle encoder assembly, and a stability calculation processor. The IMU sensor preferably employs a nine-axis inertial measurement unit, integrating a three-axis accelerometer, a three-axis gyroscope, and a three-axis magnetometer, with a sampling frequency of 100 Hz. It is mounted on a shock absorber bracket at the geometric center of the four-wheel drive chassis 11 and connected to the stability calculation processor via an SPI digital interface. The load sensor is a strain gauge load cell with a range of 500 kg and an accuracy of 0.1%FS. It is installed at the four corners below the bottom plate of the work bucket 27, outputting an analog signal via a Wheatstone bridge circuit, which is then connected to the stability calculation processor via a 24-bit high-precision analog-to-digital converter AD7799. The angle encoder group includes a main boom angle encoder, a secondary boom angle encoder, a slewing angle encoder, and a telescopic length encoder, all of which are incremental photoelectric encoders with a resolution of 4096 pulses / revolution. The main boom angle encoder is mounted on the first hinge shaft, the secondary boom angle encoder is mounted on the second hinge shaft, and the slewing angle encoder is mounted on the slewing bearing. Each encoder is connected to the encoder interface card of the stability calculation processor via differential signal transmission. The stability calculation processor uses a TMS320F28335 DSP digital signal processor with a main frequency of 150MHz and a floating-point arithmetic unit. It exchanges data at high speed with the trajectory prediction unit through dual-port RAM and can calculate the dynamic stability boundary of the platform in real time.

[0032] The trajectory prediction unit includes a remote control signal receiving module, a dynamics model calculation module, and a trajectory generation module. The remote control signal receiving module uses a 2.4GHz wireless communication module, supports frequency hopping spread spectrum technology, has an effective communication distance of 100 meters, and communicates bidirectionally with the remote control 32 via an SMA antenna. The dynamics model calculation module is implemented based on the Xilinx Zynq-7020 FPGA chip, and incorporates the platform's multibody dynamics model and hydraulic system model. It simultaneously receives output data from the remote control signal receiving module, the environmental data processor, and the stability calculation processor via an AXI bus, with a calculation frequency of 100 Hz. The trajectory generation module uses a dedicated trajectory planning processor with a built-in RRT path planning algorithm. Its output is connected to the early warning unit via a high-speed serial bus, and it can generate a 1-3 second predicted trajectory coordinate sequence.

[0033] The early warning unit includes an electric field sensor array, an electric field data processor, and an avoidance path planning module. The electric field sensor array comprises 12 non-contact electric field sensors with a measurement range of 0-100 kV / m and a response time of less than 1 millisecond. Six sensors are evenly distributed on the outer surface of the auxiliary arm 24, and the other six are mounted on the top guardrail of the work bucket 27. Each electric field sensor is connected to the 16-channel synchronous sampling ADC of the electric field data processor via a shielded coaxial cable. The electric field data processor uses an STM32F407 microcontroller with a main frequency of 168 MHz and a built-in digital filtering algorithm. Its output is connected to the avoidance path planning module via a CAN bus. The avoidance path planning module is based on the artificial potential field method and uses a dedicated motion planning processor. Its output is simultaneously connected to the control decision unit and the display device of the remote controller 32 via a parallel data bus, enabling real-time calculation of safe avoidance paths.

[0034] The control decision unit includes a risk level determination module, an intervention strategy selection module, and a control command correction module. The risk level determination module is implemented using a fuzzy logic controller with a built-in risk assessment rule base, outputting four risk level signals based on the input risk assessment data. The intervention strategy selection module is based on a finite state machine design, selecting the corresponding intervention strategy according to the risk level and outputting the strategy code to the control command correction module. The control command correction module uses a 32-bit ARM processor STM32F767 with six PWM output channels, an output frequency of 1kHz, and a duty cycle resolution of 12 bits. These PWM signals are connected to the electromagnets of the hydraulic valves in hydraulic mechanisms I 23, II 25, and the rotary mechanism 26 via PWM signal lines. The PWM signals drive the proportional solenoid valves through a power amplifier circuit, controlling the current range of 0-800 mA, thereby precisely controlling the opening degree of the hydraulic valves and achieving precise control of the speed and direction of the actuators. When the risk level reaches the emergency braking level, the risk level determination module sends an emergency start signal directly to the leveling hydraulic cylinder of the four-wheel drive chassis 11 via an emergency stop relay, bypassing the normal control circuit to ensure rapid response in extreme situations.

[0035] The remote controller 32 features an ergonomic design, measuring 200×150×50 mm and weighing 500 grams. Its casing is made of non-slip rubber, with an IP54 protection rating. It includes a 7-inch color LCD display with a resolution of 1024×600 pixels, readable even in sunlight. The display's input connects to the output of the intervention strategy selection module via a 2.4GHz wireless link, receiving display data including risk location coordinates, risk level indicators, and suggested operation direction vectors. A built-in graphics processing unit converts this data into an intuitive graphical interface. The remote controller 32 also integrates haptic feedback, including an eccentric vibration motor and an H-bridge drive circuit. The vibration motor has a power of 3 watts and an adjustable vibration frequency range of 0-10 Hz. The drive circuit's input receives vibration control signals wirelessly, enabling it to deliver risk warnings to the operator through different vibration modes. The remote controller 32 is equipped with four dual-axis joysticks with a travel range of ±30 degrees. Built-in Hall sensors detect the joystick positions with a resolution of 12 bits. These joysticks control the main boom lifting, auxiliary boom leveling, rotation, and walking functions. The joystick signals are processed by a signal conditioning circuit and sampled by a 12-bit ADC before being transmitted to the remote controller signal receiving module via a 2.4GHz wireless module, enabling remote and precise control of the work platform.

[0036] Through the organic combination and synergistic operation of the above-mentioned components, the narrow-body self-propelled work platform of the present invention realizes advanced functions such as hybrid power drive, intelligent environmental perception, dynamic stability control, live conductor safety protection and intelligent auxiliary decision-making, which significantly improves the safety and efficiency of high-altitude operations. It is particularly suitable for live-line working environments in the power industry and provides reliable technical support for high-altitude operations under complex working conditions.

[0037] This invention also designs a control method for a narrow-body self-propelled work platform with a hybrid power system. Through multi-sensor fusion sensing, real-time dynamic stability calculation, predictive trajectory planning, electric field detection and early warning, and intelligent control decision-making, the method enables safe and intelligent operation of the aerial work platform in complex power environments. The control method includes the following detailed steps: Step S1: High-precision 3D environmental modeling; Step S1 constructs a real-time 3D environmental map using multi-sensor data fusion technology, providing a precise environmental information foundation for subsequent trajectory planning and risk assessment. This step first activates LiDAR sensors, a binocular vision system, and an ultrasonic sensor array to synchronously acquire environmental data. Then, it performs time synchronization alignment and coordinate system transformation on the multi-source data. Finally, through point cloud registration algorithms and data fusion processing, a real-time 3D environmental map containing obstacle geometry, location coordinates, and ground slope information is constructed.

[0038] Step S11: LiDAR Environment Scan; Step S11 specifically implements the environment scanning function of the LiDAR sensor. LiDAR is an abbreviation for Light Detection and Ranging, which acquires distance information by emitting laser pulses and measuring the reflected light signals. In this step, the LiDAR sensor is activated to perform a surround scan, with the scanning frequency set to 10-20Hz. This frequency range ensures data accuracy while meeting real-time requirements. The sensor emits laser pulses and receives reflected signals, calculates distance values ​​based on the time-of-flight of light, and acquires three-dimensional point cloud data including azimuth, pitch, and distance values. The point cloud data is stored in a buffer. This time-of-flight-based ranging method has the advantages of high accuracy and strong anti-interference capability, and can effectively perceive the three-dimensional environmental structure around the work platform.

[0039] Step S12: Binocular Vision Depth Calculation; Step S12 implements the depth calculation function of the binocular vision system. The binocular vision system simulates the stereoscopic vision principle of the human eye, capturing the same scene from different angles using two cameras and calculating the depth distance of objects using disparity information. This step initiates simultaneous capture by the left and right cameras of the binocular vision system, performs grayscale processing on the left and right images, and uses the SGBM stereo matching algorithm to calculate the disparity value of the same feature point in the left and right images. SGBM is an abbreviation for Semi-Global Block Matching, a semi-global block matching algorithm that achieves relatively stable and accurate disparity results through dynamic programming optimization in multiple directions.

[0040] Calculate the depth distance of this point based on the disparity value: ;in: Depth distance (mm) Baseline distance of the binocular cameras (mm) For camera focal length (pixels) The disparity value (pixel) is used to map depth data to RGB images, generating a color depth map that provides rich texture and depth information for environmental perception.

[0041] Step S13: Ultrasonic Near-Range Detection; Step S13 completes the near-range obstacle detection of the ultrasonic sensor array. Ultrasonic sensors operate based on the principle of sound wave reflection, making them particularly suitable for detecting near-range obstacles. This step activates each sensor in the ultrasonic sensor array to sequentially emit ultrasonic pulses, receive the ultrasonic signals reflected back from the obstacle, and measure the round-trip time of the ultrasonic waves.

[0042] Calculate the distance to the obstacle based on the speed of sound: ;in: Distance to the obstacle (m) The speed of sound is taken as 340 m / s. Echo time (s); This method acquires near-field obstacle location data within a range of 0.2-5 meters, effectively supplementing the shortcomings of LiDAR and vision systems in near-field detection.

[0043] Step S14: Multi-sensor data fusion; Step S14 achieves time synchronization and coordinate system unification of multi-sensor data. Since different sensors have different sampling frequencies and processing times, it is necessary to align the acquired data with timestamps to ensure that data from the same moment can be correctly matched. This step aligns the timestamps of LiDAR point cloud data, visual depth data, and ultrasonic distance data, removes abnormal data exceeding the detection range, and transforms the data from each sensor coordinate system to the platform's central coordinate system using a coordinate transformation matrix. Subsequently, an extended Kalman filter algorithm is used for data fusion, and the state update equation is: ;in: The fused state estimation vector The predicted value based on the state at the previous time step. Kalman gain matrix The vector of sensor observations at the current moment The observation matrix is ​​used; the fused environmental point cloud data is obtained through iterative calculation, which significantly improves the accuracy and reliability of environmental perception.

[0044] Step S15: 3D Environment Map Construction; Step S15 completes the construction and updating of the 3D environment map. This step performs voxel rasterization on the fused point cloud data. A voxel is the smallest unit in 3D space, similar to a pixel in a 2D image. The space is divided into cubic grids with sides of 5-10 cm. This resolution ensures both the accuracy of the environment representation and controls computational complexity. The number of point clouds in each grid is counted, and grids containing point clouds are marked as occupied. Grids with more than a threshold number of point clouds are identified as obstacle grids. The boundary patches of the obstacle grids are extracted to form the geometric representation of the obstacles. The ICP point cloud registration algorithm is used to match and align the current frame point cloud with the historical frame point clouds. ICP is an abbreviation for Iterative Closest Point, which continuously optimizes the correspondence between two point clouds to achieve accurate point cloud registration. The 3D environment map is updated frame by frame, generating a real-time environment map containing obstacle geometry, 3D coordinate positions, and ground slope angle information, providing an accurate environmental cognition basis for the operation platform.

[0045] Step S2: Dynamic stability envelope calculation; Step S2 provides stability constraints for safe operation by calculating the dynamic stability boundary of the platform in real time. This step reads the platform attitude angle data output by the IMU sensor, the bucket load data output by the load sensor, the boom attitude data output by each joint encoder, and the ground tilt angle data in real time. The above data is input into the multibody dynamics model to calculate the overall center of gravity coordinates of the platform and the support reaction forces at each support point, and the dynamic stability boundary is calculated based on the overturning moment balance condition.

[0046] Step S21: Platform Attitude Angle Measurement; Step S21 achieves accurate measurement of the platform's attitude angle. IMU is an abbreviation for Inertial Measurement Unit, which integrates an accelerometer, gyroscope, and magnetometer, and can measure the acceleration, angular velocity, and magnetic field information of an object. This step reads the three-axis acceleration data, three-axis angular velocity data, and three-axis magnetic field data output by the IMU sensors.

[0047] The pitch angle is calculated using the formula for fusing acceleration and angular velocity data using a complementary filtering algorithm: ; in: The current pitch angle (rad). These are the filter coefficients, with values ​​ranging from 0.95 to 0.98. The pitch angle (rad) at the previous moment. y-axis angular velocity (rad / s) The sampling time interval (s) The arctangent function in the four quadrants x-axis acceleration (m / s²) 2 ) y-axis acceleration (m / s²) 2 ) Z-axis acceleration (m / s²) 2 The roll angle was calculated using the same method to obtain real-time attitude angle data of the platform, providing basic parameters for stability analysis.

[0048] Step S22: Working Bucket Load Monitoring; Step S22 completes the real-time monitoring of the working bucket load. The load sensor typically uses a strain gauge sensor, which obtains weight information by measuring the deformation of the elastic body under load. This step reads the voltage signal output by the load sensor at the bottom of the working bucket and converts the voltage signal into load weight through a calibration curve. ;in: Load weight (kg) Calibration factor (kg / mV) The sensor output voltage (mV) The zero-point offset (kg) is used to monitor load changes in real time, allowing for timely adjustment of stability calculation parameters to ensure operational safety.

[0049] Step S23: Boom joint posture measurement; Step S23 achieves precise measurement of the posture of each joint of the boom. An encoder is a sensor that converts angular or linear displacement into electrical signals, measuring the displacement by counting pulse signals. This step reads the number of pulses output from the angle encoder at the connection between the main boom and the turntable, and calculates the main boom angle based on the encoder resolution. ;in: The elevation angle (°) of the main arm relative to the horizontal plane. Encoder pulse counting One circle corresponds to the angle This represents the total number of pulses per encoder revolution. Similarly, the encoder data for the boom, telescopic mechanism, and slewing mechanism are read to obtain real-time attitude data for each joint of the boom, providing accurate geometric parameters for dynamic modeling.

[0050] Step S24: Multibody Dynamics Modeling; Step S24 establishes a multibody dynamics model and calculates the overall center of gravity of the platform. Multibody dynamics studies the motion laws of a system composed of multiple rigid or flexible bodies connected by constraints. In this step, each component of the platform is treated as a rigid body, and a multibody dynamics coordinate system is established, with the platform center O as the origin. The mass and center of gravity coordinates of each component are defined. Based on the boom attitude angle and extension length, the position of the center of gravity of each component in the platform coordinate system is calculated step by step using a homogeneous coordinate transformation matrix.

[0051] Overall center of gravity coordinates of the computing platform: ;in: The x-coordinate of the centroid (m) The Y-coordinate of the centroid (m) Summation operator Let i be the mass (kg) of the i-th component. Let x be the centroid coordinate (m) of the i-th component. Let Y be the center of gravity (m) of the i-th component; by accurately calculating the position of the overall center of gravity, key parameters are provided for stability analysis.

[0052] Step S25: Calculation of Dynamic Stable Boundary; Step S25 completes the calculation and update of the dynamic stable boundary. This step constructs a supporting polygon with the four support points of the platform as vertices, determines whether the centroid projection point is inside the supporting polygon, calculates the distance from the centroid projection point to each side of the supporting polygon, and takes the minimum distance as the stability margin.

[0053] Calculate the overturning moment at each support point: ;in: The overturning moment (N·m) The total weight of the platform (N) The horizontal distance (m) from the center of gravity to the overturning axis; calculate the anti-overturning moment: ;in: To resist overturning moment (N·m) For the opposing gravity (N) The distance (m) from the lateral center of gravity to the overturning axis; determine whether the condition is met. The system determines the stability conditions and updates and outputs dynamically stable envelope data in real time.

[0054] Step S3: Predictive Trajectory Planning and Risk Assessment; Step S3 predicts the trajectory of the work bucket in the future time period by establishing a platform dynamics model and hydraulic system response characteristics, and assesses various risks on the trajectory. This step receives control commands issued by the operator and converts them into target motion parameters for each actuator. Based on the platform dynamics state equation and hydraulic system response characteristics, a numerical integration method is used to calculate the position coordinate sequence of the work bucket center point in the next 1-3 seconds. The predicted trajectory is compared and verified with obstacles in the environmental map, the positions of charged bodies detected by electric field sensors, and dynamic stability boundaries. Trajectory segments with collision, proximity to charged bodies, or instability risks are identified and marked.

[0055] Step S31: Control Command Parsing; Step S31 completes the reception and parsing of control commands. This step receives control command signals from the remote control and parses the position values ​​of each joystick contained in the command. The joystick position values ​​are then converted into the target speed of the corresponding actuator through linear mapping. ;in: Target velocity (m / s or rad / s) This represents the maximum speed of the actuator (m / s or rad / s). The value represents the lever position (%); 100 is the percentage conversion factor. Step S32: Establishing the dynamic state equations; Step S32 establishes the platform's dynamic state equations. Define the state vector: ; in: Main arm angle (rad) The angle of the secondary arm (rad) Rotation angle (rad) The extension / retraction length (m) Represents angular velocity (rad / s) Expressing linear velocity (m / s) Indicate matrix transpose; establish the state transition equation: ; in: To predict the time step, we take 0.1s; The nonlinear state transition function is used; Step S33: Hydraulic system model establishment; Step S33 establishes the hydraulic system response characteristic model. Based on the characteristics of the hydraulic system, establish a model showing the relationship between actuator speed and valve control current: ;in: The actuator speed (m / s); This refers to the flow coefficient of the proportional valve. Valve control current (mA); The difference between the hydraulic pump output pressure and the return oil pressure (MPa); Let the hydraulic oil density be denoted by ; considering the first-order inertial delay characteristics of the hydraulic system, the transfer function is adopted as follows: ;in: The time constant is 0.1-0.3 s; For the Laplace operator; Step S34: Numerical integration calculation of the trajectory; Step S34 performs numerical integration calculation of the trajectory. Using the current state as the initial condition, the target velocity command is substituted into the state equation, and numerical integration is performed using the fourth-order Runge-Kutta method. The state values ​​at each future moment are calculated step by step, with a prediction duration of 1-3 seconds, corresponding to 10-30 calculation steps. The position coordinates of the center point of the working bucket at each moment are extracted to form a sequence of predicted trajectory coordinates.

[0056] Step S35: Trajectory Risk Assessment; Step S35 completes the trajectory risk assessment. Extract the boundary point cloud set of obstacles from the 3D environment map, and extract the position coordinates of charged bodies from the electric field sensor data. For each point on the predicted trajectory, calculate its Euclidean distance to all obstacle points: ; in: Let be the distance (m) between the i-th trajectory point and the j-th obstacle point; calculate the distance between the trajectory point and the charged body: Determine whether all safety conditions are met, mark trajectory points that do not meet any safety conditions as risk points, and output the trajectory risk assessment results.

[0057] Step S4: Live conductor proximity warning; Step S4 is specifically designed to address the safety issues of live conductors in the power operation environment. It uses an electric field sensor array to detect the location of live conductors, establishes a graded warning mechanism, and provides intelligent avoidance paths.

[0058] Step S41: Electric Field Strength Measurement; Step S41 completes the measurement and conversion of the electric field strength. The voltage signals output by the electric field sensor arrays mounted on the auxiliary boom and the work bucket are read and converted into electric field strength using the calibration formula: ;in: The electric field strength (V / m) measured by the i-th sensor; For calibration coefficients (V / m / mV); The sensor output voltage (mV); Step S42: Reverse Calculation of Charged Body Position; Step S42 performs the reverse calculation of the charged body position. The position of the charged body is calculated using a point charge model, and the electric field strength equation is established: ;in: To measure the electric field strength (V / m) at the point. Let be the electrostatic constant, and take... N·m 2 / C 2 The equivalent charge (C) of the charged body. The distance (m) from the measurement point to the charged body is determined by solving a system of equations based on the measurement data from multiple sensors, and then using the least squares method to solve for the position coordinates and voltage level of the charged body.

[0059] Step S43: Define safe distance thresholds; Step S43 establishes a three-level safe distance threshold system. Define warning distance, danger distance, and prohibited distance, corresponding to low-risk areas, medium-risk areas, and high-risk areas.

[0060] Step S44: Distance Calculation and Classification Judgment; Step S44 calculates and classifies the distances between trajectory points and charged bodies. It calculates the distances between each point on the predicted trajectory and the charged body: It determines the distance range and activates an audible and visual alarm when entering a dangerous area, marking the location of the live conductor and the dangerous area on the augmented reality display.

[0061] Step S45: Calculation of the avoidance path; Step S45 uses the artificial potential field method to calculate the avoidance path. Define the target attractive potential field: ;in: The strength of the attractive potential field; The attraction gain coefficient is set to 1-5. The distance (m) from the current position to the target position is defined; the repulsive potential field of the charged body is defined, the composite potential field is calculated, and the sequence of avoidance path points is generated along the negative gradient direction.

[0062] Step S5: Collision Avoidance and Dynamic Stability Control; Step S5 is the execution stage of the entire control method. Based on the risk type and degree identified in the preceding steps, a graded intervention strategy is adopted to achieve intelligent safety control. Step S51: Risk Level Assessment; Step S51 establishes risk level assessment criteria. Based on the relationship between the minimum safe distance and the safety threshold, it is determined as Level 1 warning, Level 2 intervention, Level 3 avoidance, or emergency braking. Step S52: Level 1 Warning Prompt; Step S52 implements human-machine interaction prompts for Level 1 warnings. A display command is sent to the augmented reality display to mark the risk location in yellow, and a control signal is sent to the haptic feedback device of the remote control to generate a prompt vibration. Step S53: Level 2 Intervention Speed ​​Adjustment; Step S53 implements automatic speed adjustment for Level 2 intervention. The current hydraulic valve control current is read, the adjustment coefficient is calculated, and the adjusted control current is output to automatically reduce the boom movement speed. Step S54: Level 3 Avoidance Path Correction; Step S54 implements automatic path correction for Level 3 avoidance. The coordinates of the next target point are extracted from the avoidance path, and the required deflection angle is calculated. ;in: The deflection angle is in rad. It is an inverse cosine function; the deflection angle is decomposed into each actuator, and the control commands of each actuator are corrected so that the boom moves automatically along the avoidance path.

[0063] Step S55: Emergency Braking Response; Step S55 enables rapid emergency braking response. Immediately identify the direction of movement causing the risk, send a zero-current command to the hydraulic valve corresponding to that direction, cutting off the hydraulic oil flow to the actuator. Simultaneously, send a leveling command to the chassis leveling system, controlling the leveling hydraulic cylinders to adjust the height of each support point, allowing the platform to quickly return to a level state.

[0064] Through the coordinated operation of the five main steps described above, the control method of this invention organically combines multi-sensor environmental perception, real-time dynamic stability monitoring, predicted trajectory risk assessment, live conductor proximity warning, and intelligent safety control. This control method boasts significant advantages such as high perception accuracy, strong prediction accuracy, fast response speed, and high safety and reliability, effectively improving the operational safety and intelligence level of aerial work platforms in complex power environments, and providing an advanced technical solution for aerial work in the power industry.

[0065] Example 2: This example uses live-line work on a 110kV transmission line of a power company as an application scenario. A narrow-body self-propelled working platform with a hybrid power system, based on this invention, is used to verify its practical application effect. The work site is located in a narrow urban alley with an 8-degree slope. The objective is to replace insulators on a conductor 18 meters above the ground. The conductor voltage is 110kV. Obstacles such as trees and buildings exist in the work environment, with the nearest obstacle approximately 4.5 meters from the work location.

[0066] Before the operation began, the operator moved the narrow-body self-propelled work platform to the work position. The platform chassis is 1.15 meters wide, allowing it to easily traverse the 1.5-meter-wide community roads. After activating the control system, the system automatically executed the high-precision 3D environmental modeling steps. The LiDAR sensor scanned the environment at a frequency of 15Hz, completing the acquisition of point cloud data within a 50-meter radius within 3 seconds, obtaining approximately 4.5 million point cloud data points. The binocular vision system simultaneously acquired color depth images within a 15-meter radius in front, with an image resolution of 1920×1080 pixels, a baseline distance of 300 mm, and a focal length of 1000 pixels. For an object 10 meters away, the parallax value was approximately 30 pixels, and the depth distance was calculated to be 10,000 mm according to the depth calculation formula. The ultrasonic sensor array detected a tree trunk obstacle 1.2 meters to the left of the chassis and a building wall 2.8 meters to the right.

[0067] The system employs an extended Kalman filter algorithm to fuse multi-source sensor data. The Kalman gain matrix converges after 10 iterations, achieving an accuracy of 0.05 meters for the fused environmental point cloud data. Through voxel rasterization, the space is divided into 8-centimeter cube grids, generating approximately 1.25 million grid cells, of which about 85,000 are marked as obstacles. The ICP point cloud registration algorithm is used for frame-by-frame updates, with registration errors controlled within 0.03 meters. The final constructed 3D environmental map accurately reflects the spatial structure of the work site.

[0068] In the dynamic stability envelope calculation step, the IMU sensor outputs platform attitude data in real time at a sampling frequency of 100Hz. Since the ground slope is 8 degrees, the initial pitch angle measured by the IMU sensor is 8.2 degrees, and the initial roll angle is 0.5 degrees. A complementary filtering algorithm is used for attitude calculation, with the filter coefficient set to 0.96. After a 0.5-second filtering convergence process, the attitude angle data is output stably. The load sensor at the bottom of the work bucket detects a total weight of 185 kg for the workers and tools, and the sensor output voltage is 370 mV. Based on the calibration coefficient of 0.5 kg / mV and the zero-point offset of 0 kg, the load weight is calculated to be 185 kg.

[0069] The main boom angle encoder outputs 2048 pulses, with an encoder resolution of 4096 pulses / revolution, calculating the main boom elevation angle to be 45 degrees. The auxiliary boom angle encoder outputs 1024 pulses, calculating the auxiliary boom angle relative to the main boom to be 22.5 degrees. The telescopic mechanism's displacement encoder outputs 3276 pulses, corresponding to a telescopic length of 4 meters. The slewing mechanism's angle encoder outputs 819 pulses, calculating the slewing angle to be 72 degrees. Based on the above attitude data, the multibody dynamics model calculates the platform's overall center of gravity coordinates to be 2.35 meters in the X direction, 0.42 meters in the Y direction, and 8.65 meters in the Z direction.

[0070] The system constructs a supporting polygon using four support points on the platform. The coordinates of these four support points are A (-0.6, -1.0, 0), B (0.6, -1.0, 0), C (0.6, 1.0, 0), and D (-0.6, 1.0, 0). The coordinates of the center of gravity projection point are (2.35, 0.42), which is located outside the supporting polygon, indicating a risk of overturning. The distances from the center of gravity projection point to each side of the supporting polygon are calculated, with a minimum distance of 1.75 meters, indicating a stability margin of 1.75 meters. Due to the large stability margin, the system determines that the current state is within a safe range. The overturning moment is calculated. The total weight of the platform is 18,500 Newtons, the horizontal distance from the center of gravity to the nearest overturning axis is 1.75 meters, and the overturning moment is 32,375 N·m. The anti-overturning moment was calculated. With a lateral gravity of 12,000 Newtons and a lateral center of gravity distance of 3.2 meters from the overturning axis, the anti-overturning moment was 38,400 N·m. Since the anti-overturning moment was greater than the overturning moment, the system confirmed that the platform was in a stable state.

[0071] In the trajectory planning and risk assessment step, the operator issues a control command via remote control to continue raising the main boom, with the main boom lifting lever position set to 60%. Based on the maximum lifting angular velocity of the main boom being 3 degrees / second, the target angular velocity is calculated to be 1.8 degrees / second. The system establishes a dynamic state equation, and the current state vector is: [45°, 22.5°, 72°, 4m, 0° / s, 0° / s, 0° / s, 0m / s]. Based on the hydraulic system model, the hydraulic pump output pressure is 20MPa, the return oil pressure is 0.5MPa, the pressure difference is 19.5MPa, the hydraulic oil density is 850kg / m³, the proportional valve flow coefficient is 0.8, and the valve control current is 480mA. The calculated actuator speed is 0.031m / s, corresponding to a boom angular velocity of 1.8 degrees / second.

[0072] Considering the first-order inertial delay characteristic of the hydraulic system, the time constant is set to 0.2 seconds. The system uses the fourth-order Runge-Kutta method for numerical integration, with a prediction duration of 2 seconds and a time step of 0.1 seconds, calculating a total of 20 prediction points. After numerical integration, the predicted trajectory coordinate sequence of the center point of the working bucket within the next 2 seconds is obtained, with the trajectory starting coordinates at (15.8, 8.2, 18.0) meters and the ending coordinates at (16.5, 8.9, 19.2) meters. The system compares the predicted trajectory with obstacles in the environmental map, calculates the minimum distance between each trajectory point and the obstacle, and finds that the distance between the 12th trajectory point and the tree branch is 0.85 meters, which is less than the safe distance of 1.0 meter from the obstacle. The system marks this trajectory point as a risk point.

[0073] In the charged body proximity warning step, the electric field sensor array monitors the electric field intensity distribution in the surrounding space in real time. The output voltage signals of the six electric field sensors on the auxiliary arm are 125mV, 98mV, 156mV, 142mV, 110mV, and 88mV, respectively. Based on the calibration coefficient of 800V / m / mV, the calculated electric field intensities are 100kV / m, 78.4kV / m, 124.8kV / m, 113.6kV / m, 88kV / m, and 70.4kV / m, respectively. The system uses a point charge model to inversely determine the location of the charged body, with an electrostatic constant of 9×10⁻⁶. 9 N·m 2 / C 2 By solving the system of equations using the least squares method, the coordinates of the charged body are obtained as (17.2, 9.5, 21.5) meters, and the equivalent voltage level is 108kV.

[0074] The system defines three levels of safety distance thresholds: a warning distance of 3.0 meters, a danger distance of 1.5 meters, and a prohibited distance of 0.7 meters. The system calculates the distance between each point on the predicted trajectory and the live conductor: the first trajectory point is 3.8 meters away, the tenth is 2.9 meters away, the fifteenth is 2.3 meters away, and the twentieth is 1.8 meters away. The system determines that the fifteenth trajectory point has entered the warning zone and the twentieth is entering the danger zone. The system activates the audible and visual alarm device with an alarm frequency set to 2Hz, and simultaneously marks the location of the live conductor in red on the AR display.

[0075] The system uses an artificial potential field method to calculate the avoidance path. With an attraction gain coefficient of 3 and a distance of 1.2 meters from the current working bucket position to the target position, the calculated attraction potential field strength is 2.16. With a repulsion gain coefficient of 30 and a distance of 2.3 meters from the current position to the charged body, the calculated repulsion potential field strength is 0.83. The combined potential field strength is 2.99. An avoidance path point sequence is generated along the negative gradient direction. The avoidance path shifts the working bucket 0.6 meters to the left, maintaining a safe distance of 3.2 meters from the charged body.

[0076] In the collaborative collision avoidance and dynamic stability control steps, the system intervenes in stages based on the identified risk type and severity. For the obstacle risk at the 12th trajectory point, the minimum safe distance is 0.85 meters, the safe distance threshold is 1.0 meter, and the calculated distance ratio is 0.85, which the system classifies as a level two intervention. The system reads the current hydraulic valve control current as 480mA, calculates the adjustment coefficient, and uses the formula α=0.3+0.4×(0.85-0.5) / (0.3)=0.77 to calculate the adjusted control current as 370mA. The system outputs the adjusted control command, and the main boom lifting speed automatically decreases to 77% of the original speed, giving the operator more reaction time.

[0077] For the risk of a live conductor at the 20th trajectory point, the minimum safe distance is 1.8 meters, the safe distance threshold is 3.0 meters, and the calculated distance ratio is 0.6, which the system classifies as a level two intervention. The system also reduces its movement speed and generates a warning vibration at a frequency of 2Hz and an amplitude of 70% via the tactile feedback device on the remote control. Upon sensing the warning, the operator adjusts the direction of the control lever, causing the work bucket to move along the avoidance path, ultimately successfully reaching the work position, maintaining a safe distance of 3.2 meters from the live conductor and 1.3 meters from tree obstacles.

[0078] Throughout the operation, the system updated the environmental map in real time at a frequency of 10Hz, performed dynamic stability calculations at 50Hz, updated trajectory predictions at 10Hz, sampled electric field detection at 1kHz, and maintained a control command correction delay of less than 50ms. The operator completed the insulator replacement within the work bucket in approximately 25 minutes. During this time, the platform's stability margin remained above 1.5 meters, and the safe distance from live parts remained above 3.0 meters, with no safety incidents occurring. After the operation was completed, the system controlled the boom to return to its initial position along the planned path. The entire operation was safe and efficient.

[0079] Comparative Example 1 employs a traditional aerial work platform control method. This method lacks environmental perception and trajectory prediction capabilities, relying solely on the operator's visual judgment and experience. The platform is equipped with basic angle and load sensors but lacks environmental perception devices such as LiDAR, binocular vision, and ultrasonic sensors, as well as an electric field sensor array for detecting charged bodies. The control system uses a simple proportional control strategy, directly mapping the lever position to the hydraulic valve opening, lacking risk assessment and intelligent intervention functions. A comparative experiment was conducted using the same operating scenario as Example 2. During operation, the operator visually judged the distance between the work platform and obstacles and charged bodies. Due to the inherent errors in visual judgment and the inability to accurately measure distances, the operator tended to adopt a conservative operating strategy, maintaining an actual distance of over 5.5 meters between the work platform and the charged wires, far exceeding the prescribed 3.0-meter safety distance. Lacking environmental perception capabilities, the operator failed to promptly detect tree branches obstructing the view when approaching tree obstacles, resulting in a slight collision between the work platform and the branches. Although no equipment damage occurred, this impacted operational efficiency. Regarding stability control, traditional methods rely solely on fixed stability monitoring indicators, issuing an alarm when the platform tilt angle exceeds 3 degrees, but they cannot predict stability changes during dynamic movement. During the experiment, when the main boom was rapidly raised, the platform's center of gravity shifted significantly, and the tilt angle instantaneously reached 2.8 degrees. Although this did not exceed the alarm threshold, it was close to the safety boundary. Based on experience, the operator proactively reduced the operating speed, extending the operation time to 38 minutes, a 52% increase compared to Example 2. In the experiment, the closest distance between the work bucket and the live wire was 5.2 meters, far exceeding the 3.0-meter safety distance requirement. While this ensured safety, it significantly limited the working space, affecting operational efficiency. The entire operation failed to fully utilize the permitted working space, and operators reported inconvenience in operation.

[0080] Comparative Example 2 employs a partially intelligent aerial work platform control method. This method is equipped with LiDAR and tilt sensors, possessing basic environmental perception and stability monitoring functions, but lacks trajectory prediction, electric field detection, and intelligent intervention capabilities. The control system can construct a 3D environmental map in real time and issue an alarm when an obstacle is detected within 1.0 meter, but it cannot predict the trajectory in advance and can only passively issue an alarm when a collision is imminent. A comparative experiment was conducted using the same operating scenario as Example 2. During the operation, the LiDAR sensor collected environmental point cloud data at a frequency of 10Hz, and the 3D environmental map constructed by the system accurately reflected the obstacle's location. When the work platform approached a tree obstacle, the system issued a warning at a distance of 1.2 meters. The operator adjusted the operating direction based on the alarm, avoiding a collision. Compared to Comparative Example 1, Comparative Example 2 showed significantly improved environmental perception capabilities, and the operation time was shortened to 32 minutes. During the experiment, the system issued 12 obstacle approach alarms, and the operator made 12 operational adjustments, while Example 2 only made 3 active avoidance adjustments. Regarding the safety control of live parts, Comparative Example 2 lacks an electric field sensor, making it impossible to accurately locate the live parts; it still relies on visual observation and experience. To ensure safety, operators maintain a distance of over 4.5 meters between the work bucket and the live wire, which, while lower than the 5.2 meters in Comparative Example 1, is still greater than the 3.0-meter safety distance requirement. In terms of stability control, the system can monitor the platform's tilt angle in real time, but it lacks dynamic stability boundary calculation capabilities and cannot predict stability changes during dynamic movement.

[0081] Comparative Example 3 employs a different intelligent aerial work platform control method. This method is equipped with environmental perception sensors and electric field sensors, possessing environmental modeling and charged body detection capabilities, but lacking trajectory prediction and intelligent intervention functions. The control system can detect the position of the charged body in real time and issue an alarm when the distance is less than 3.0 meters, but it cannot predict the trajectory in advance or proactively avoid risks. A comparative experiment was conducted using the same operating scenario as Example 2. During the operation, the electric field sensor array detected the electric field intensity distribution in the surrounding space in real time, and the system accurately located the position of the charged conductor. When the work bucket was 3.2 meters away from the charged conductor, the system issued a warning, and the operator adjusted the operating direction according to the warning. Compared with Comparative Examples 1 and 2, Comparative Example 3 has a significant advantage in the safety control of charged bodies, and can accurately control the safe distance from charged bodies. However, due to the lack of trajectory prediction function, the system cannot assess in advance the risk of approaching charged bodies that may result from operating commands, and can only issue an alarm when the work bucket actually approaches the charged body. During the experiment, the system issued a total of 8 charged body approach alarms, and the operator made 8 operational adjustments. Example 2, through trajectory prediction, proactively assesses risks and adjusts control commands in advance, requiring only two avoidance adjustments, resulting in smoother operation. Regarding stability control, Comparative Example 3 possesses basic stability monitoring capabilities but lacks dynamic stability boundary calculation functionality. In the experiment, when the main boom was raised to a significant angle, the system detected a platform tilt angle of 2.5 degrees, approaching the preset 3-degree alarm threshold. However, the system could not predict the potential overturning risk from continued raising, forcing operators to stop operation and wait for the platform to stabilize before resuming work, extending the operation time to 30 minutes.

[0082] To objectively evaluate the performance of Example 2 and the comparative examples, a comparative experiment was designed according to the relevant requirements of the national standard GB / T9465-2008 "Aerial Work Platforms" and the power industry standard DL / T5285-2018 "Hydraulic Crimping Process Specification for Overhead Conductors and Ground Wires in Power Transmission and Transformation Engineering". The experimental site was set up in a standard test area, simulating a typical live-line working environment. A 110kV simulated live conductor was arranged in the site, with a conductor height of 18 meters and a ground slope of 8 degrees. Obstacles such as trees and buildings were set up in the site, with the nearest obstacle 4.5 meters from the working position. The experiment used the same model of narrow-body self-propelled working platform, configured with the control systems of Example 2 and the comparative examples respectively. The experimental task was to complete a simulated insulator replacement operation, requiring the working bucket to reach a working position 3.0 meters from the live conductor, maintaining a safe distance of more than 1.0 meter from obstacles, and the platform tilt angle not exceeding 3 degrees. Each control method was repeated 10 times, and the performance indicators were recorded. The average values ​​were taken for comparative analysis. The experimental indicators include operation completion time, minimum distance to a live conductor, minimum distance to an obstacle, maximum platform tilt angle, number of operation adjustments, control command response time, trajectory prediction accuracy, and number of safety accidents. The measurement methods for each indicator are as follows: Operation completion time is recorded using a stopwatch, showing the total time from platform startup to completion of the task. Minimum distance to a live conductor is measured in real-time using a laser rangefinder, measuring the distance from the center point of the work bucket to the live conductor, and the minimum value is recorded during the experiment. Minimum distance to an obstacle is measured in real-time using a laser rangefinder, measuring the distance from the outer surface of the work bucket to the nearest obstacle, and the minimum value is recorded during the experiment. Maximum platform tilt angle is measured using a high-precision digital inclinometer, measuring the tilt angle of the platform chassis, and the maximum value is recorded during the experiment. Number of operation adjustments is recorded by the experimenters, noting the number of times the operator actively changed the operating direction. Control command response time is recorded using a high-speed data acquisition card, showing the time delay from the movement of the control lever to the start of hydraulic valve action. Trajectory prediction accuracy is measured by comparing the root mean square error of the predicted trajectory coordinates with the actual trajectory coordinates. Number of safety accidents is recorded by noting the number of collisions, overturning incidents, or violations of safety distance requirements during the experiment.

[0083] Experimental results are as follows Figure 5-8 As shown, Figure 5 This is a comprehensive comparison chart of the work efficiency of Example 2 and Comparative Examples 1-3. Figure 5The following key information can be gleaned from the three sub-graphs: Regarding average task completion time, Example 2 completes the task in just 26 minutes, while Comparative Example 1 requires 39 minutes, Comparative Example 2 requires 33 minutes, and Comparative Example 3 requires 31 minutes. Example 2 shortens the time by 13 minutes compared to Comparative Example 1, achieving an efficiency improvement of 33.3%. In terms of the number of operation adjustments, Example 2 requires only 3.2 directional adjustments throughout the entire task, while Comparative Example 1 requires 15.8 adjustments, Comparative Example 2 requires 12.5 adjustments, and Comparative Example 3 requires 8.7 adjustments. The number of operation adjustments in Example 2 is less than one-quarter of that in Comparative Example 1. Regarding control command response time, the system response time of Example 2 is 45 milliseconds, significantly faster than Comparative Example 1's 120 milliseconds, Comparative Example 2's 85 milliseconds, and Comparative Example 3's 95 milliseconds, representing a response speed improvement of 62.5% to 166.7%. Figure 5 As can be seen, Example 2 significantly outperforms all comparative examples in terms of work efficiency, operational smoothness, and system response speed, demonstrating superior overall work efficiency performance. Example 2 can complete the task in the shortest time, while minimizing the number of operational adjustments and maintaining the fastest system response speed. The synergistic optimization of these three indicators shows that Example 2 achieves truly intelligent and efficient operation.

[0084] This invention designs a trajectory prediction unit that, by establishing the platform's dynamic state equations and the hydraulic system's response characteristic model, can predict the working bucket's trajectory 1 to 3 seconds in advance. The trajectory prediction works by taking the current platform state and operating commands as inputs, and calculating the future position coordinate sequence using state transition equations and the fourth-order Runge-Kutta numerical integration method. This predictive control technology transforms the system from passive response to active planning, enabling the assessment of the possible trajectory and potential risks of a control command the instant it is issued. Comparative Examples 1 to 3 lack complete trajectory prediction functionality; they can only employ a passive response control strategy, issuing an alarm only when the working bucket actually approaches an obstacle or live conductor. This passive response mode requires operators to frequently adjust their operating direction to avoid risks, disrupting the work process and extending operation time with each adjustment. Example 2, by combining trajectory prediction technology with an avoidance path planning module, allows the system to automatically calculate the optimal movement path from the current position to the target position, achieving the shortest path operation while ensuring safety. Therefore, the operation completion time is significantly shortened, and the number of operational adjustments is greatly reduced. Furthermore, Example 2 employs a high-performance FPGA processor and a DSP digital signal processor to construct the control system. The FPGA chip, Xilinx Zynq-7020, has parallel processing capabilities, enabling it to simultaneously process multiple sensor data streams and control algorithm calculations. The DSP processor, TMS320F28335, has a dedicated floating-point unit, capable of efficiently performing complex dynamic equation solving and trajectory integration calculations. This hardware architecture optimization reduces the system's control command response time to 45 milliseconds, significantly faster than comparable systems using ordinary embedded processors. This rapid response capability further enhances the system's real-time control performance and operational smoothness.

[0085] Figure 6 This is a comparison of the safety distance control accuracy between Example 2 and Comparative Examples 1-3, from... Figure 6As can be seen from the four sub-figures, in terms of controlling the minimum distance to live parts, Example 2 precisely controls the minimum distance between the work bucket and the 110 kV live conductor at 3.15 meters, very close to the safety distance requirement of 3.0 meters. In contrast, Comparative Example 1's minimum distance is 5.35 meters, Comparative Example 2's is 4.58 meters, and Comparative Example 3's is 3.82 meters; all three comparative examples adopted relatively conservative safety strategies. Regarding distance control accuracy, Example 2 achieves 95.2% accuracy, while Comparative Example 1 achieves only 56.1%, Comparative Example 2 65.5%, and Comparative Example 3 78.5%, with Example 2 showing an accuracy improvement of 17% to 39%. In terms of controlling the minimum distance to obstacles, Example 2 maintains a safety distance of 1.08 meters, slightly higher than the safety requirement of 1.0 meter. Comparative Examples 2 and 3 maintain 1.15 meters and 1.22 meters respectively, while Comparative Example 1's minimum distance is only 0.75 meters and it experienced three minor collisions. Regarding safety, no collisions occurred in Example 2, Comparative Example 2, and Comparative Example 3, while Comparative Example 1 suffered from safety hazards due to a lack of environmental awareness. Example 2 significantly outperformed all comparative examples in terms of safe distance control accuracy, maximizing the use of permissible workspace while strictly ensuring safety, achieving precise distance control and safety protection. Example 2 not only avoided the collision risk caused by insufficient conservatism in Comparative Example 1 but also avoided the wasted workspace caused by excessive conservatism in Comparative Examples 2 and 3, achieving an optimal balance between safety and work efficiency.

[0086] This invention integrates an electric field sensor array and an electric field data processor to construct a precise positioning system for charged bodies. In Example 2, 12 non-contact electric field sensors are arranged on the outer surface of the auxiliary arm and around the work bucket. These sensors measure the electric field intensity distribution in space, with a measurement range of 0 to 100 kV / m and a response time of less than 1 millisecond. After the electric field data processor collects the voltage signals from each sensor, it converts them into electric field intensity values ​​using a calibration formula. Then, a set of electric field intensity equations is established using a point charge model. The data from multiple measurement points are solved simultaneously, and the three-dimensional coordinate position and voltage level of the charged body are deduced using the least squares method. Compared with traditional visual observation and experience-based judgment, this non-contact electric field detection technology has significant advantages such as high accuracy, fast response, and immunity to the effects of lighting and line-of-sight obstruction. Comparative Examples 1 and 2 do not have electric field sensors; operators can only rely on visual observation to determine the position of the charged conductor. Human visual judgment has significant distance estimation errors, especially at upward angles and at long distances, where the error can reach tens of centimeters or even more than one meter. Because the exact location of the charged body cannot be known, operators must adopt a very conservative strategy to ensure safety, keeping the actual distance above 5 meters. This results in low utilization of the workspace, making many accessible work locations inaccessible. Although Comparative Example 3 is equipped with an electric field sensor that can accurately locate the charged body, it lacks trajectory prediction capabilities. The system cannot assess the potential risks of approaching the charged body in advance and can only issue a passive alarm when the work bucket actually approaches the charged body. Operators still need to maintain a large safety margin to deal with emergencies, thus keeping the actual distance at 3.82 meters. Example 2 organically combines electric field detection technology with trajectory prediction technology and avoidance path planning technology. The system first accurately obtains the three-dimensional coordinates of the charged body through the electric field sensor. Then, the trajectory prediction unit calculates the predicted trajectory for the next 1 to 3 seconds based on the current operation command and calculates the distance between each point on the trajectory and the charged body. The avoidance path planning module uses an artificial potential field method to construct a potential field function, setting the target location as an attractive potential field and the charged body location as a repulsive potential field. The optimal avoidance path is generated by calculating the gradient direction of the synthesized potential field. This combination of three technologies enables the system to accurately identify risk locations, predict movement trajectories in advance, and proactively plan avoidance paths. Therefore, while meeting the 3.0-meter safety distance requirement, the actual distance can be precisely controlled at 3.15 meters, which is only 0.15 meters more than the safety requirement. This makes full use of the allowable working space and significantly improves work efficiency.In terms of obstacle collision protection, Example 2 constructs a high-precision three-dimensional environmental map through multi-sensor fusion. The LiDAR sensor provides long-range environmental scanning, the binocular vision system provides mid-range depth information and texture recognition, and the ultrasonic sensor array provides near-range blind spot detection. After the data from the three sensors are fused by the extended Kalman filter algorithm, the accuracy of the environmental map reaches 0.05 meters. This high-precision environmental perception capability, combined with trajectory prediction technology, enables active collision avoidance, and therefore no collision accidents occur.

[0087] Figure 7 This is a comparison chart of the platform stability performance of Example 2 and Comparative Examples 1-3. Figure 7 As shown in the two sub-figures, regarding the control of the maximum tilt angle of the platform, Example 2 achieved a maximum tilt angle of 2.15 degrees throughout the entire operation, significantly lower than Comparative Example 1's 2.82 degrees, Comparative Example 2's 2.68 degrees, and Comparative Example 3's 2.55 degrees. All methods did not exceed the alarm threshold of 3.0 degrees. In terms of stability margin, Example 2 maintained an average stability margin of 1.68 meters, significantly higher than Comparative Example 1's 1.24 meters, Comparative Example 2's 1.35 meters, and Comparative Example 3's 1.42 meters, representing an improvement of 18% to 35%. Regarding the number of stability alarms, Example 2 triggered zero stability alarms throughout the experiment, while Comparative Example 1 triggered two alarms, and Comparative Examples 2 and 3 each triggered one alarm. Although no actual overturning event occurred with any of the methods, frequent stability alarms would disrupt the work process and affect the operator's confidence. Example 2 demonstrates excellent performance in platform stability control, effectively limiting the platform tilt angle and maintaining a large stability margin, ensuring the safety and reliability of the operation. Example 2 utilizes intelligent stability control technology to ensure that the platform remains in optimal stability during high-altitude operations, providing operators with a safer and more reliable working environment.

[0088] Figure 8 This is a comparison chart of the environmental perception capabilities of Example 2 and Comparative Examples 1-3. Figure 8As shown in the three sub-figures, in terms of sensor configuration, Example 2 is equipped with a complete environmental perception sensor system, including three sensors based on different principles: a LiDAR laser radar sensor, a binocular vision system, and an ultrasonic sensor array. Comparative Example 1 lacks any environmental perception sensors, Comparative Example 2 is only equipped with a LiDAR sensor, and Comparative Example 3 is equipped with a LiDAR sensor and an ultrasonic sensor array but lacks a binocular vision system. Regarding the environmental map update frequency, Example 2, Comparative Example 2, and Comparative Example 3 can all update the 3D environmental map at a frequency of 10 Hz, equivalent to updating once every 100 milliseconds. Comparative Example 1, lacking environmental perception capabilities, cannot construct an environmental map. In terms of obstacle detection accuracy, Example 2 achieves a detection accuracy of 0.05 meters, Comparative Example 2 achieves 0.12 meters, and Comparative Example 3 achieves 0.08 meters. The detection accuracy of Comparative Example 1 cannot be quantified. The detection accuracy of Example 2 is 140% higher than Comparative Example 2 and 60% higher than Comparative Example 3. Example 2 achieves the highest environmental perception capability and obstacle detection accuracy through multi-sensor fusion technology, providing the most reliable environmental information foundation for safe operation. The environmental perception system in Example 2 not only has the most comprehensive range of sensor types, but also achieves the highest detection accuracy through data fusion technology.

[0089] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A narrow-body self-propelled work platform with a hybrid power system, characterized in that, It is equipped with the following control methods: S1: High-precision 3D environmental modeling. Start LiDAR sensor, binocular vision system and ultrasonic sensor array to collect environmental data synchronously. Perform time synchronization alignment and coordinate system transformation on multi-source data. Construct a real-time 3D environmental map containing obstacle geometry, position coordinates and ground slope information through point cloud registration algorithm and data fusion processing. S2: Dynamic stability envelope calculation, real-time reading of platform attitude angle data output by IMU sensor, working bucket load data output by load sensor, boom attitude data output by encoder of each joint, and ground tilt angle data, input of the above data into multibody dynamics model to calculate the overall center of gravity position coordinates of the platform and the support reaction force of each support point, and calculate dynamic stability boundary according to overturning moment balance condition. S3: Predictive trajectory planning and risk assessment. It receives control commands from operators and converts them into target motion parameters for each actuator. Based on the platform dynamic state equation and hydraulic system response characteristics, it uses numerical integration to calculate the position coordinate sequence of the center point of the working bucket within the next 1-3 seconds. It compares and verifies the predicted trajectory with obstacles in the environmental map, the positions of charged bodies detected by electric field sensors, and dynamic stability boundaries. It identifies and marks trajectory segments with risks of collision, approach to charged bodies, or instability. S4: Charged body approach warning. Read the spatial electric field intensity distribution data measured by the electric field sensor array to infer the position of the charged body, calculate the safe distance between each point on the predicted trajectory and the charged body, and determine whether to enter the warning area, danger area or restricted area. When entering the warning or danger area, activate the audible and visual alarm device and use the artificial potential field method to calculate the suggested avoidance path. S5: Collision avoidance and dynamic stability control. The intervention level is determined according to the risk type and risk level. When it is in the first level of warning, the operator is prompted through the display device and tactile feedback device. When it is in the second level of intervention, the hydraulic valve control current is automatically reduced to slow down the boom movement speed. When it is in the third level of avoidance, the control command is automatically corrected and the deflection angle is superimposed on the original movement direction to guide the boom to move along the avoidance path. When it is in the emergency braking, the hydraulic valve control signal in the dangerous direction is immediately cut off and the chassis leveling is started.

2. The platform according to claim 5, characterized in that, Step S1 specifically includes: S11: Start the LiDAR sensor to perform a surround scan, set the scan frequency to 10-20Hz, the sensor emits laser pulses and receives reflected signals, calculates the distance value based on the time of light flight, obtains three-dimensional point cloud data including azimuth, pitch angle and distance value, and stores the point cloud data in the buffer. S12: Simultaneously capture images using the left and right cameras of the binocular vision system, perform grayscale processing on the left and right images, and calculate the disparity value of the same feature point in the left and right images using the SGBM stereo matching algorithm. Calculate the depth distance of the point based on the parallax value. ;in For depth distance; The baseline distance of the binocular cameras; The focal length of the camera; The disparity value is used to map the depth data to the RGB image, generating a color depth map. S13: Activate each sensor in the ultrasonic sensor array to sequentially emit ultrasonic pulses, receive the ultrasonic signals reflected back from the obstacle, and measure the round-trip time of the ultrasonic waves. Calculate the distance to the obstacle based on the speed of sound. ;in Distance to the obstacle; The speed of sound is taken as 340 m / s; 2 represents the echo time; 2 represents the round-trip coefficient; This data is used to acquire the location of nearby obstacles within a range of 0.2-5 meters. S14: Timestamp-align LiDAR point cloud data, visual depth data, and ultrasonic distance data; remove outlier data outside the detection range; transform the data from each sensor coordinate system to the platform's central coordinate system using a coordinate transformation matrix; and perform data fusion using an extended Kalman filter algorithm. The state update equation is: ;in This is the fused state estimation vector; This is a predicted value based on the state at the previous time step; The Kalman gain matrix; This represents the vector of sensor observations at the current moment. The observation matrix is ​​used; the fused environmental point cloud data is obtained through iterative calculation. S15: The fused point cloud data is processed into voxel rasterization, dividing the space into cubic grids with a side length of 5-10cm. The number of point clouds in each grid is counted, and grids containing point clouds are marked as occupied. Grids with a number of point clouds exceeding a threshold are identified as obstacle grids. The boundary patches of the obstacle grids are extracted, and the ICP point cloud registration algorithm is used to match and align the point clouds of the current frame with those of the historical frames. The 3D environment map is updated frame by frame to generate a real-time environment map containing obstacle geometry, 3D coordinate position and ground slope angle information.

3. The platform according to claim 1, characterized in that, Step S2 specifically includes: S21: Read the triaxial acceleration data output by the IMU sensor. , , Three-axis angular velocity data , , and triaxial magnetic field data , , The platform's pitch angle is calculated by fusing acceleration and angular velocity data using a complementary filtering algorithm. and roll angle The formula for calculating the pitch angle is: ;in The pitch angle is the current angle; α is the filter coefficient, ranging from 0.95 to 0.

98. The pitch angle at the previous moment; y-axis angular velocity; dt is the sampling time interval; It is the arctangent function in the four quadrants; The acceleration is along the x-axis; The acceleration is along the y-axis; Z-axis acceleration; sqrt is the square root operator; roll angle. The platform's real-time attitude angle data were obtained by calculating using the same method. S22: Read the voltage signal output by the load sensor at the bottom of the work hopper. The voltage signal is converted into load weight through a calibration curve. ;in This refers to the load weight; These are calibration coefficients; The sensor output voltage; This is the zero-point offset; it obtains real-time load data within the working bucket. S23: Read the number of pulses output from the angle encoder at the connection between the main boom and the turntable. Calculate the main boom angle based on the encoder resolution. ;in The angle of elevation of the main arm relative to the horizontal plane; For encoder pulse counting; 360° represents the angle corresponding to one revolution. This refers to the total number of pulses per revolution of the encoder; it also reads the number of pulses output from the angle encoder at the connection point between the auxiliary arm and the main arm. Calculate the arm angle Read the number of pulses output by the displacement encoder of the telescopic mechanism. Calculate the telescopic length Read the number of pulses output by the angle encoder of the rotary mechanism. Calculate the rotation angle To obtain real-time attitude data of each joint of the boom; S24: Treat each component of the platform as a rigid body, establish a multibody dynamics coordinate system, and define the chassis mass with the platform center O as the origin. and its centroid coordinates Define the quality of the turntable and its centroid coordinates Define the mass of the main boom and its centroid coordinates Define the mass of the auxiliary arm and its centroid coordinates Define the quality of the work bucket and its centroid coordinates Define load mass and its centroid coordinates Based on the boom's attitude angle and telescopic length, the position of the center of gravity of each component in the platform coordinate system is calculated step by step using a homogeneous coordinate transformation matrix, thus calculating the overall center of gravity coordinates of the platform. , ;in The x-coordinate of the centroid; The y-coordinate of the centroid; Σ is the summation operator; Let the mass of the i-th component be denoted as . Let X be the centroid coordinate of the i-th component; Let Y be the center of gravity of the i-th component; obtain the overall center of gravity position of the platform; S25: Construct a supporting polygon using the four support points of the platform as vertices. Determine whether the centroid projection point is inside the supporting polygon. Calculate the distance from the centroid projection point to each side of the supporting polygon and take the minimum distance as the stability margin. Calculate the overturning moment at each support point. ;in The overturning moment; The total gravity of the platform; The horizontal distance from the center of gravity to the overturning axis; calculate the overturning moment. ;in To resist overturning moment; For lateral gravity; The distance from the lateral center of gravity to the overturning axis; Determine To determine if the condition is valid, draw the dynamic stability boundary polygon under the current working condition, delineate the safe working area with the centroid projection point as the center and the stability margin as the radius, and update and output the dynamic stability envelope data in real time.

4. The platform according to claim 1, characterized in that, Step S3 specifically includes: S31: Receives control command signals from the remote controller and parses the main boom lifting lever position values ​​contained in the commands. Telescopic operating lever position value Rotary control lever position value and travel control lever position value The lever position value is converted into the target velocity of the corresponding actuator through linear mapping. ;in For target speed; This is the maximum speed of the actuator; The value represents the lever position; 100 is the percentage conversion factor; this provides the target motion speed command for each actuator. S32: Establish the platform dynamic state equations and define the state vectors: ； in Main arm angle, For the auxiliary arm angle, For the rotation angle, For telescopic length, Indicates angular velocity. Indicates linear velocity. Indicates transpose; defines the control input vector. ,in Represent the control current of the hydraulic valve; establish the state transition equation. ;in The prediction time step is set to 0.1s; It is a nonlinear state transition function; S33: Based on the characteristics of the hydraulic system, establish a model relating actuator speed to valve control current. ;in For actuator speed; This refers to the flow coefficient of the proportional valve. For valve control current; This is the square root operator; This is the difference between the hydraulic pump output pressure and the return oil pressure. Let the hydraulic oil density be denoted by ; considering the first-order inertial delay characteristics of the hydraulic system, a transfer function is used. Describe the response process, where The time constant is set to 0.1-0.3 s. For the Laplace operator; S34: In the current state As initial conditions, the target velocity command is substituted into the state equation, and numerical integration is performed using the fourth-order Runge-Kutta method to calculate step by step. , ... The state value at time t, where For prediction durations of 1-3 seconds, extract the position coordinates of the center point of the working bucket at each moment. To form a sequence of predicted trajectory coordinates ; S35: Extracting boundary point cloud sets of obstacles from a 3D environment map Extracting the position coordinates of charged bodies from electric field sensor data For each point on the predicted trajectory The Euclidean distances between it and all obstacle points are calculated as follows: ;in Let be the distance between the i-th trajectory point and the j-th obstacle point; take the minimum distance among all obstacle points. ,in This is the minimum value operator; it calculates the distance between the trajectory point and the charged body: Extracting stability margin from dynamic stability boundary data Determine whether the condition is met. , , ,in , , These are the obstacle safety distance, the charged body safety distance, and the stability safety margin, respectively. Trajectory points that do not meet any of the safety conditions are marked as risk points. Continuous risk points are combined into risk trajectory segments. The risk type and risk level are recorded, and the trajectory risk assessment results are output.

5. The platform according to claim 1, characterized in that, Step S4 specifically includes: S41: Read the voltage signal output from the electric field sensor array installed on the boom and bucket. It is converted into electric field strength through the calibration formula. ;in Let be the electric field strength measured by the i-th sensor; These are calibration coefficients; The sensor outputs voltage; data on the electric field intensity distribution in the space surrounding the platform is acquired. S42: Based on the relationship between electric field strength and distance, the position of the charged body is deduced using a point charge model, and the electric field strength equation is established. ;in To measure the electric field strength at the point; Let the electrostatic constant be taken as N·m 2 / C 2 ; This is the equivalent charge of a charged body; To measure the distance from the point to the charged body; 2 The square operator is used; a system of equations is formed from measurement data from multiple sensors, and the position coordinates of the charged body are solved using the least squares method. and voltage level ; S43: Defines three levels of safety distance thresholds and warning distances. m corresponds to a low-risk area, and the danger distance is... m corresponds to a medium-risk area, and the prohibited distance is... m corresponds to a high-risk area; S44: Calculate the points on the predicted trajectory With charged body Distance between ;in The distance from the trajectory point to the charged body; determine The distance range in which it is located, when The audible and visual alarm device is activated at certain times, and the alarm frequency is set. Hz, when Increase alarm frequency to Hz, while marking the location of charged objects and danger zones in red on the AR display; S45: Calculate the avoidance path using the artificial potential field method, and define the target attraction potential field. ;in The strength of the attractive potential field; The attraction gain coefficient is set to 1-5; The distance from the current position to the target position; 0.5 is a coefficient; the repulsive potential field of the charged body is defined. when , when ;in The strength of the repulsive potential field; The repulsion gain coefficient is set to 10-50; The distance to the charged body; For early warning distance; calculate the composite potential field. ;in The total potential field strength; This is the addition operator; it calculates the direction of the potential field gradient. ;in The gradient vector; It is a partial derivative operator; , , It is a unit vector along the three coordinate axes; It is a vector dot product operator; it generates a sequence of avoidance path points along the negative gradient direction, and outputs the avoidance path to the AR display in the form of navigation arrows and numerical displays to guide the operator to adjust the operation to avoid the charged body.

6. The platform according to claim 1, characterized in that, Step S5 specifically includes: S51: Based on the risk types and risk levels identified in S3 and S4, establish risk level assessment criteria, when the minimum safe distance... satisfy It was determined to be a Level 1 warning at that time. When it was determined to be a level two intervention, When it is determined to be a level three avoidance, when The time was determined to be emergency braking; among which This refers to the safe distance threshold corresponding to this risk type; S52: When the warning level is at level 1, a display command is sent to the AR display to mark the risk location in yellow on the display screen and draw an arrow pointing from the current work bucket position to the recommended safe direction. At the same time, a control signal is sent to the haptic feedback device in the remote control to drive the vibration motor to generate a warning vibration at a frequency of 1Hz and an amplitude of 50% to remind the operator to pay attention to potential risks. S53: When at the secondary intervention level, read the current hydraulic valve control current. Calculate the adjustment coefficient ;in For adjustment coefficients; Minimum safe distance; The safety distance threshold is 0.3, 0.4, and 0.5, which are constant coefficients. The adjusted control current is then calculated. ;in For the new control current; As a control command output for the hydraulic valve, the boom movement speed is reduced to 30-70% of its original speed, extending the operator's reaction time; S54: When the evasion level is three, extract the coordinates of the next target point from the evasion path calculated by S4. Calculate the current working bucket position Direction vector to the target point ;in Given the target direction vector; calculate the motion direction vector corresponding to the original control command. Calculate the required deflection angle ;in This refers to the deflection angle; It is the inverse cosine function; This is the vector dot product operator; This is a vector magnitude operator; it decomposes the deflection angle into individual actuators and modifies the control instructions for each actuator. ;in This is the corrected control current for the j-th actuator; This is the original control current; To correct the current increment, the corrected control command is output to make the boom move automatically along the avoidance path while maintaining the operator's control over the overall movement direction. S55: When in emergency braking mode, immediately identify the direction of motion that poses a risk and send a zero-current command to the hydraulic valve corresponding to that direction. mA cuts off the hydraulic oil flow to the actuator, bringing the movement in that direction to a complete stop within 0.5 seconds; it then determines whether other movement directions are safe, maintaining the original control command for safe directions or allowing the operator to continue control; simultaneously, it sends a leveling command to the chassis leveling system and reads the height sensor data from the four support points. , , , Calculation platform tilt angle ;in The tilt angle; It is the arctangent function; This is the maximum value operator; This is the minimum value operator; To adjust the spacing between support points; control the leveling hydraulic cylinder to adjust the height of each support point, so that... This enables the platform to quickly return to a horizontal state, eliminating the risk of overturning.

7. The platform according to claim 1, characterized in that, It includes: Walking platform, working device and control system; The traveling platform includes a wheeled four-wheel drive chassis, on which a fuel engine, a fuel tank, an electric pump and a battery pack are respectively installed. The output ends of the fuel engine and the electric pump are connected to the hydraulic distribution valve group through hydraulic pipelines to provide hydraulic power to the traveling platform and the working device. The working device includes a turntable, which is mounted on a wheeled four-wheel drive chassis via a slewing bearing. The turntable is movably connected to the main boom via a first hinge shaft. A hydraulic mechanism I is provided between the turntable and the main boom. The end of the main boom away from the turntable is movably connected to the auxiliary boom via a second hinge shaft. A hydraulic mechanism II is provided between the main boom and the auxiliary boom. The end of the auxiliary boom away from the main boom is movably connected to the working bucket via a slewing mechanism. The control system includes a control box and a remote controller. The control box contains an environmental perception unit, a stability calculation unit, a trajectory prediction unit, an early warning unit, and a control decision unit.

8. The platform according to claim 7, characterized in that: The environmental perception unit includes a LiDAR sensor, a binocular vision system, an ultrasonic sensor array, and an environmental data processor. The LiDAR sensor is mounted on the top of the turntable, the binocular vision system is mounted on the front face of the work bucket, and multiple ultrasonic sensors of the ultrasonic sensor array are mounted on the four sides of the wheeled four-wheel drive chassis. The environmental data processor is connected to the LiDAR sensor, the binocular vision system, and the ultrasonic sensor array, and is also connected to the trajectory prediction unit for transmitting three-dimensional environmental map data. The stability calculation unit includes an IMU sensor, a load sensor, an angle encoder, and a stability calculation processor. The IMU sensor is installed at the geometric center of the four-wheel drive chassis, the load sensor is installed under the bottom plate of the work bucket, and the angle encoders include a main boom angle encoder installed on the first hinge shaft, a secondary boom angle encoder installed on the second hinge shaft, a slewing angle encoder installed on the slewing bearing, and a telescopic length encoder installed on the main boom telescopic mechanism. The stability calculation processor is connected to the IMU sensor, the load sensor, and each angle encoder, and is also connected to the trajectory prediction unit to transmit dynamic stability boundary data. The trajectory prediction unit includes a remote control signal receiving module, a dynamic model calculation module, and a trajectory generation module. The remote control signal receiving module is connected to the remote control via wireless communication to receive control commands. The dynamic model calculation module is connected to the remote control signal receiving module, the environmental data processor, and the stability calculation processor to receive the current state data of the platform. The trajectory generation module is connected to the dynamic model calculation module and to the early warning unit for transmitting the predicted trajectory coordinate sequence. The early warning unit includes an electric field sensor array, an electric field data processor, and an avoidance path planning module. Multiple electric field sensors in the electric field sensor array are respectively installed on the outer surface of the auxiliary arm and around the working bucket. The electric field data processor is connected to the electric field sensor array. The avoidance path planning module is connected to the electric field data processor and the trajectory generation module, and also to the control decision unit to transmit risk assessment results and suggested avoidance paths. The control decision unit includes a risk level determination module, an intervention strategy selection module, and a control command correction module. The risk level determination module is connected to the avoidance path planning module to receive risk assessment results. The intervention strategy selection module is connected to the risk level determination module. The control command correction module is connected to the intervention strategy selection module and the remote control signal receiving module to receive original control commands and intervention strategies. The output of the control command correction module is connected to the control terminals of each hydraulic valve in hydraulic mechanism I, hydraulic mechanism II, and the slewing mechanism to output corrected control commands.

9. The platform according to claim 7, characterized in that: The front and rear axles of the wheeled four-wheel drive chassis are each equipped with an independent steering mechanism. Each steering mechanism includes a steering hydraulic cylinder and a steering control valve. The control terminal of the steering control valve is electrically connected to the control decision unit within the control box. The inlet and outlet ports of the steering hydraulic cylinder are connected to the output port of the steering control valve via hydraulic lines. The wheeled four-wheel drive chassis also includes a horizontal tilt sensor. The signal output terminal of the horizontal tilt sensor is connected to the stability calculation unit within the control box. The signal output terminal of the LiDAR sensor is connected to the environmental data processor via a CAN bus. The binocular vision system includes a left camera and a right camera. The signal output terminals of the left and right cameras are connected to the environmental data processor via USB interfaces. Each ultrasonic sensor in the ultrasonic sensor array is connected to the environmental data processor via a serial communication interface. The environmental data processor is connected to the trajectory prediction unit via a data interface. The IMU sensor is connected to the stability calculation processor via a digital interface; the load sensor is connected to the analog-to-digital converter input of the stability calculation processor via an analog signal line; the main arm angle encoder, auxiliary arm angle encoder, slewing angle encoder, and telescopic length encoder are all connected to the stability calculation processor via incremental encoder interfaces; the stability calculation processor exchanges data with the trajectory prediction unit via a shared data channel; the remote controller communicates bidirectionally with the remote controller signal receiving module via a wireless communication frequency band, and the remote controller's joystick sends control commands to the remote controller signal receiving module via a signal processing circuit; the dynamic model calculation module simultaneously receives output data from the remote controller signal receiving module, the environmental data processor, and the stability calculation processor via an internal data bus; the output of the trajectory generation module is connected to the early warning unit via a data channel. Each electric field sensor in the electric field sensor array is connected to the multi-channel analog-to-digital converter input terminal of the electric field data processor via an analog signal line. The output of the electric field data processor is connected to the avoidance path planning module via a data bus; the output of the avoidance path planning module is simultaneously connected to the display device of the control decision unit and the remote controller via a parallel data channel. The risk level determination module outputs a risk level signal to the intervention strategy selection module based on the input risk assessment data. The intervention strategy selection module selects the corresponding intervention strategy based on the risk level and outputs the strategy code to the control command correction module. The control command correction module processes the original control command and the intervention strategy to generate a corrected control command, which is then output to the hydraulic valve control terminals of hydraulic mechanism I, hydraulic mechanism II, and slewing mechanism, respectively. When the risk level reaches the emergency braking level, the risk level determination module directly sends an emergency start signal to the leveling hydraulic cylinder of the four-wheel drive chassis.

10. The platform according to claim 7, characterized in that: The output of the control command correction module is connected to the electromagnet of the hydraulic valve in the hydraulic mechanism I via a PWM signal line to control the opening degree of the hydraulic valve. The output of the control command correction module is connected to the electromagnet of the hydraulic valve in the hydraulic mechanism II via a PWM signal line to control the opening degree of the hydraulic valve. The output of the control command correction module is connected to the electromagnet of the hydraulic valve in the rotary mechanism via a PWM signal line to control the opening degree of the hydraulic valve. The working bucket is equipped with an angular displacement sensor, which is mounted on the output shaft of the rotary mechanism. The signal output terminal of the angular displacement sensor is connected to the input terminal of the stability calculation processor to monitor the rotation angle of the working bucket. The auxiliary arm is made of FRP insulating material, and the outer surface of the auxiliary arm is connected to the main arm through an insulating sleeve. The remote control includes a display device and a tactile feedback device. The input terminal of the display device is wirelessly connected to the output terminal of the intervention strategy selection module to receive display data. The display data includes risk location coordinates, risk level indicators, and suggested operation direction vectors. The display device converts the data into a graphical interface display through a graphics processing unit. The tactile feedback device includes a vibration motor and a drive circuit. The input terminal of the drive circuit is wirelessly connected to the output terminal of the intervention strategy selection module to receive vibration control signals. The output terminal of the drive circuit is connected to the power supply terminal of the vibration motor to drive the vibration motor to generate tactile feedback.

Images