Intelligent spraying control method and system suitable for large bridge end face

Abstract

The invention relates to the technical field of intelligent spraying, in particular to an intelligent spraying control method and system suitable for the end face of a large bridge. The method comprises the following steps: positioning the initial operation position of the bridge end surface spraying equipment based on machine vision and a multi-sensor calibration method; constructing a bridge end face spraying model, and obtaining an optimal spraying path based on a path generation algorithm; intelligent adaptive control is conducted on a spraying mechanism based on multi-sensor data fusion, and the position posture of an executing mechanism is adjusted through a dynamic angle compensation algorithm; a multi-degree-of-freedom stable operation control algorithm is used for controlling the speed and the action period of an executing mechanism; and a multi-mechanism joint control algorithm is used for controlling linkage control of the executing mechanism and the spray head. Aiming at the core problems of low manual operation efficiency, large quality fluctuation, heavy and limited traditional equipment, prominent potential safety hazards and the like in railway bridge end face waterproof construction, systematic technical breakthrough is realized through integration of electromechanical integration innovation and an intelligent algorithm.

Description

Technical Field

[0001] The present invention relates to the field of intelligent spraying technology, and in particular to an intelligent spraying control method and system suitable for the end face of a large bridge. Background Art

[0002] In the manufacture of railway prefabricated box girders, after demolding, concrete pouring requires tensioning, grouting, sealing anchor holes, and applying a waterproof layer to the beam ends. The quality of the waterproof layer at the beam ends is crucial to the safety and durability of the structure. Poor construction quality can allow rainwater to seep into the anchor holes, causing corrosion to critical structural components such as anchors and steel strands.

[0003] The manual painting method involves first preparing a two-component polyurethane waterproof coating by hand, then applying it repeatedly with rollers, brushes, scrapers, and other hand tools. Each end face of a high-speed railway box girder requires 7.7 square meters of polyurethane waterproof coating. Standards require a coating thickness of at least 2 mm, and approximately eight coats are typically required. Therefore, a total of 16 coats of polyurethane waterproof coating are required at both ends of each box girder, covering an area of ​​123 square meters. This is a significant workload, and manual control of coating thickness and uniformity results in poor consistency and low construction efficiency. Furthermore, high-temperature construction, volatilization of cleaning agents, and working at height pose health risks to personnel.

[0004] Existing bridge end-face spray coating equipment faces challenges such as bulky equipment, limited mobility within confined spaces, and inconvenient turnover. This bulky equipment is limited during transportation and on-site operation, increasing transportation difficulties and operational risks. Within confined spaces, the spray equipment's mobility is reduced, impacting spray quality and efficiency while also increasing safety risks. This inconvenient turnover compromises operational continuity, increases labor and material costs, and can extend project duration. These issues collectively impact the overall performance and cost-effectiveness of spray coating operations. Summary of the Invention

[0005] In order to solve the above-mentioned problems, the present invention provides an intelligent spraying control method and system suitable for the end face of a large bridge.

[0006] In the first aspect, the present invention provides an intelligent spraying control method for the end surface of a large bridge, which adopts the following technical solutions: An intelligent spraying control method for large bridge end faces, comprising: Obtain bridge end face dimension data and various sensor data; Positioning the starting position of bridge end surface spraying equipment based on machine vision and multi-sensor calibration methods; Construct a bridge end face spraying model and obtain the optimal spraying path based on the path generation algorithm; The spraying mechanism is intelligently adaptively controlled based on multi-sensor data fusion, wherein the position and posture of the actuator are adjusted using a dynamic angle compensation algorithm; the speed and motion cycle of the actuator are controlled using a multi-degree-of-freedom smooth operation control algorithm; and the linkage control of the actuator and the nozzle is controlled using a multi-mechanism joint control algorithm.

[0007] Furthermore, the method based on machine vision and multi-sensor calibration is used to locate the starting position of the bridge end face spraying equipment, including using manual remote control guidance or automatic following methods, using machine vision and multi-sensor calibration to move the equipment to the corresponding working position, and based on the self-correction algorithm and automatic deviation correction control algorithm to enable the equipment to complete posture self-correction, combined with the spatial three-dimensional coordinate fusion algorithm, to achieve accurate positioning of the starting position of the bridge end face intelligent spraying equipment operation, wherein, through the machine vision algorithm, automatic following and guiding personnel are achieved, moving to the corresponding position, extracting the current scene image features, performing marker position matching and identification, and completing the calibration of the camera coordinate system, pixel coordinate system, and bridge end face coordinate system.

[0008] Furthermore, the self-correction algorithm and automatic deviation correction control algorithm enable the equipment to complete posture self-correction, including calculating the deviation between the equipment and the bridge end face through the data fed back by the GPS-RTK and laser ranging sensors at the front and rear ends of the equipment, forming the input of the automatic deviation correction control algorithm, and adopting an improved PID adjustment algorithm. The output needs to be adjusted to the angle. With the center of the equipment as the reference point, the in-situ rotation control method is adopted to convert the required adjustment amount into the position information of the steering motor, driving the chassis to complete the steering. During the process, the distance between the front and rear ends and the bridge end face is continuously monitored and fed back to the automatic deviation correction process control to ensure that the equipment remains parallel to the bridge end face.

[0009] Furthermore, the combination of the three-dimensional spatial coordinate fusion algorithm can realize the precise positioning of the starting position of the bridge end face intelligent spraying equipment operation, including constructing a bridge end face coordinate system based on the bridge end face size data and various sensor data, and utilizing the equipment action execution mechanism and the coordinate system space coordinate fusion algorithm. According to the principle of spatial coordinate transformation, the rotation matrix and translation vector are used to realize the mutual conversion between the equipment action execution mechanism coordinate system and the bridge end face coordinate system, and the actual coordinates of the bridge section are used as the target to realize the precise transmission of the action execution mechanism.

[0010] Furthermore, the bridge end face spraying model is constructed, and the optimal spraying path is derived based on a path generation algorithm. This involves optimizing the bridge end face dimensions, combining them with actual measurement information, and then dividing the spraying area after optimization. Bridge ends are bilaterally symmetrical structures, and are normally divided into three spraying areas: left, right, and center. After the spraying areas are divided, each area is subdivided, and irregular areas are filled with regular shapes. The principle is that sub-areas on the same horizontal line have regular shapes and consistent sizes, while sub-areas at different heights have different sizes. The sub-area division of the three areas is completed in sequence, and the area number, layer number, sub-area number, and the length, width, slope, and start time data of each sub-area number are automatically generated.

[0011] Furthermore, the construction of the bridge end face spraying model and the acquisition of the optimal spraying path based on the path generation algorithm also include planning the spraying path according to the geometric shape of the bridge end face spraying model, from top to bottom, with the lower spray covering half of the upper spray, and spraying in sequence from the left, middle and right spraying areas; by adopting a layered spraying method, the designed spraying thickness is achieved through multiple spraying; when spraying each layer, the height and moving speed of the nozzle are adjusted so that the paint evenly covers the bridge end face, thereby realizing the automatic generation of the optimal spraying path.

[0012] Furthermore, the use of a dynamic angle compensation algorithm to adjust the position and posture of the actuator includes, in response to the uneven ground at the construction site, obtaining the three-axis angle offset of the actuator in real time through a pre-installed high-precision inclination sensor, and constructing an error model to automatically output the compensation amount of the action actuator, and accurately adjust the position and posture of the actuator to ensure that it performs the task according to the predetermined trajectory and accuracy.

[0013] Furthermore, the multi-degree-of-freedom smooth operation control algorithm is used to control the speed and action cycle of the actuator, including real-time output of the parameters of the motion control of each drive motor according to the planned path, including acceleration and deceleration, process smooth operation speed, and the time period of each action execution, thereby realizing fast, efficient, safe and stable operation of each action actuator.

[0014] Furthermore, the multi-mechanism linkage control algorithm is used to control the actuator and the nozzle linkage control, including a multi-mechanism linkage control algorithm based on PID algorithm, kinematics forward solution, and kinematics inverse solution, combined with position feedback, to accurately control the acceleration section, uniform speed section, and deceleration section of the movement of each mechanical mechanism, formulate a linkage control strategy, and realize orderly execution and linkage control of multiple mechanisms.

[0015] In the second aspect, an intelligent spraying control system for the end face of a large bridge is provided, comprising: An omnidirectional four-wheel drive chassis, a multi-stage horizontal slide structure, a nozzle lifting and rotating structure, and a coating mechanism. The omnidirectional four-wheel drive chassis is used to drive the device forward, backward, turn left, turn right, and rotate on the spot; the multi-stage horizontal slide structure is used to output torque to achieve multi-stage linkage control; the nozzle lifting and rotating structure is used to achieve three-dimensional spatial positioning of the nozzle, and cover the entire working surface of the bridge bottom plate and web; the coating mechanism is used to realize the ratio control and stirring function of the bridge end face spraying material, and the control system is connected through the reserved communication interface to complete the pressure, flow, amplitude and angle control of the spraying process.

[0016] In summary, the present invention has the following beneficial technical effects: This invention addresses the core challenges of low manual operation efficiency, high quality fluctuations, bulky and limited traditional equipment, and significant safety hazards in railway bridge end-face waterproofing construction. By integrating mechatronics innovation with intelligent algorithms, it achieves a systematic technological breakthrough. Compared to conventional processes, this solution replaces the traditional robotic arm structure with a modular, omnidirectional four-wheel drive chassis. Combined with a multi-stage horizontal slide, a 4m-stroke lifting mechanism, and a 360° rotating nozzle, the equipment is reduced in size by over 40%, achieving precise three-dimensional coverage within an 8m x 3.5m work surface. Assembly and disassembly efficiency is tripled, effectively resolving the problem of equipment becoming stuck in confined spaces.

[0017] Through the layered spraying algorithm and dynamic parameter compensation technology, the 8-pass manual painting is optimized to 5-pass automated spraying, and the single-end surface operation time is compressed from 4 hours to 1.5 hours, the coating uniformity is improved by 70% and the amount is saved by 15%. At the same time, the GPS-RTK and laser ranging dual-mode positioning system are integrated, combined with a 0.5-second high-frequency attitude correction algorithm to achieve a trajectory deviation of <2cm under a 5° slope condition, eliminating 90% of the risk of high-altitude calibration operations. In terms of environmental adaptability, the closed coating system and remote control technology reduce the exposure to harmful substances by 80%, and the fault breakpoint resume spraying function makes the abnormal recovery time ≤5 minutes. The present invention provides a quantifiable and replicable technology upgrade path for the waterproofing construction of railway projects through the collaborative innovation of lightweight equipment, intelligent paths, precise control, and safe operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a schematic diagram of an intelligent spraying control method for the end face of a large bridge according to embodiment 1 of the present invention. DETAILED DESCRIPTION

[0019] The present invention will be further described in detail below with reference to the accompanying drawings.

[0020] Example 1 Reference Figure 1 The present embodiment provides an intelligent spraying control method for large bridge end faces, comprising: Obtain bridge end face dimension data and various sensor data; Positioning the starting position of bridge end surface spraying equipment based on machine vision and multi-sensor calibration methods; Construct a bridge end face spraying model and obtain the optimal spraying path based on the path generation algorithm; The spraying mechanism is intelligently adaptively controlled based on multi-sensor data fusion, wherein the position and posture of the actuator are adjusted using a dynamic angle compensation algorithm; the speed and motion cycle of the actuator are controlled using a multi-degree-of-freedom smooth operation control algorithm; and the linkage control of the actuator and the nozzle is controlled using a multi-mechanism joint control algorithm.

[0021] Specifically, the following steps are included: 1. Accurate positioning of spray equipment in unknown scenarios Through manual remote control guidance, or the equipment's automatic following function, using machine vision, multi-sensor calibration (GPS-RTK, laser ranging, position marking) and other technologies, the equipment can be moved to the corresponding working position and complete posture self-correction without the need for global map scanning. Combined with the spatial three-dimensional coordinate fusion algorithm, the precise positioning of the starting position of the bridge end face intelligent spraying equipment operation can be achieved.

[0022] Among them, (1) a deep learning-based visual positioning and multi-sensor fusion algorithm is used. During the data processing process, the YOLOv8 neural network is used to identify the guide personnel and the bridge end face features, and the detection results with a target coordinate confidence level ≥ 0.8 are output. The IMU inertial measurement unit (sampling frequency 200Hz) and the lidar point cloud data (point cloud density 10 points / cm²) are integrated and data fusion is achieved through Kalman filtering.

[0023] Deep learning object detection: b=(x,y,w,h,c) Among them, (x, y) is the coordinate of the target center, w, h is the width and height of the bounding box, and c is the confidence level.

[0024] Kalman filter state update: State vector , including position and attitude, Fk is the state transfer matrix, Hk is the observation matrix, and Kk is the Kalman gain.

[0025] The YOLOv8 model weights are pre-trained using a bridge scene dataset, and the Kalman filter process noise covariance Q and observation noise covariance R are set according to the sensor accuracy (position noise ±5 cm, angle noise ±0.5°).

[0026] (2) Automatic following, equipment posture self-correction algorithm: Through the machine vision algorithm, it can automatically follow the guide, move to the corresponding position, extract the current scene image features, perform marker position matching and recognition, and complete the calibration of the camera coordinate system, pixel coordinate system, and bridge end face coordinate system. Through the data feedback from the GPS-RTK and laser ranging sensors at the front and rear ends of the equipment, the deviation between the equipment and the bridge end face (front and back, left and right) is calculated to form the input of the automatic correction control algorithm. The improved PID adjustment algorithm is used to adjust the output to the required angle. With the center of the equipment as the reference point, the in-situ rotation control method is used to convert the required adjustment amount into the position information of the steering motor, driving the chassis to complete the steering. During the process, the distance between the front and rear ends and the bridge end face is continuously monitored and fed back to the automatic correction process control to ensure that the equipment remains parallel to the bridge end face.

[0027] The system uses a posture correction algorithm based on model predictive control (MPC) to obtain front and rear-end LiDAR point clouds (distance measurement accuracy ±2mm), fit the plane equation of the bridge end face, establish a kinematic model for the equipment, predict attitude changes over the next 500ms, and optimize the control variables.

[0028] Plane fitting: ax+by+cz+d=0 The RANSAC algorithm is used to remove outliers, and the threshold is set to 5 cm.

[0029] MPC optimization problem: Constraints: Among them, Np is the prediction time domain (take 10 steps), Q is the error weight matrix, and R is the control amount weight matrix.

[0030] The prediction time domain Np=10 corresponds to a prediction length of 500ms, the control time domain Nc=5, the equipment maximum steering angular velocity u{max}=10° / s, the distance deviation weight Q{dist}=10, and the angle deviation weight Q{ang}=50.

[0031] (3) Spatial three-dimensional coordinate fusion algorithm: A spatial coordinate fusion algorithm for the bridge end face coordinate system and the equipment action actuator coordinate system is studied. The principle of spatial coordinate transformation is adopted, and the rotation matrix and translation vector are used to realize the mutual conversion between the equipment action actuator coordinate system and the bridge end face coordinate system. The actual coordinates of the bridge section are used as the target to realize the precise transmission of the action actuator.

[0032] Among them, an octree-based three-dimensional coordinate fast fusion algorithm is used to construct an octree structure for the bridge end face point cloud (leaf node voxel size 10cm×10cm×10cm), and the coordinates of the actuator end are calculated through the UR5 robotic arm kinematic model (DH parameter error ±0.1mm), and the octree index is used to realize fast coordinate query.

[0033] About octree construction: in, m is the maximum dimension of the bounding box, and l is the tree depth (8 layers).

[0034] 2. Coordinate transformation: Homogeneous transformation matrix cascade, To transform the equipment chassis to the world coordinate system, Transform the end of the robot arm to the chassis.

[0035] The voxel size of the octagonal leaf node is 10 cm, which meets the centimeter-level accuracy requirements of the spraying operation; the connecting rod length αai, torsion angle αi and other parameters of the robot arm DH are obtained through factory calibration, and the error compensation matrix is ​​updated every 10 minutes.

[0036] 2. Build a bridge end face spraying model and automatically complete the spraying path planning Construct a bridge end face spray model. After inputting the bridge end face dimension data, the spray area is automatically divided and the special-shaped areas are filled. A spray model is generated with the area number, layer number, sub-area number as index, and the length, width, slope, start time, etc. of each sub-area number as data; combined with the spray construction process requirements, the spray path is planned, and the formatted output of the spray model data is realized to form the action instruction set of the spray mechanism. Combined with the control system, one-button automatic spraying is realized to ensure the completion of the operation surface and meet the construction process requirements.

[0037] in, (1) Construction method of bridge end face spraying model: Based on the bridge end face dimension information and combined with the actual measurement information, the spraying area is divided after optimization. The bridge end faces are all bilaterally symmetrical structures, and are normally divided into three spraying areas: left, right, and middle. After the spraying area is divided, each area is sub-divided into sub-areas, and the irregular areas are filled with regular shapes. The principle is that the sub-areas on the same horizontal line have regular shapes and consistent sizes, and the sub-areas at different heights can have different sizes. In sequence, the sub-area division of the three areas is completed, and the area number, layer number, sub-area number, and the length, width, slope, start time and other data of each sub-area number are automatically generated.

[0038] in, (1) An adaptive region partitioning algorithm based on a genetic algorithm was used to read the BIM model end face contour and extract the feature point cloud. The region partitioning scheme was optimized using a genetic algorithm with the sub-region regularity (rectangularity ≥ 0.8) and spraying efficiency (empty stroke ≤ 15%) as the goals.

[0039] About rectangular calculation: Among them, area is the area of ​​the sub-region, is the minimum enclosing rectangle area.

[0040] About the genetic algorithm fitness function: Weights w1=0.6, w2=0.4.

[0041] The genetic algorithm population size was set to 50, the crossover probability was 0.8, the mutation probability was 0.1, and convergence was achieved after 50 iterations. The sub-area width was determined according to the nozzle coverage diameter (30 cm), and the height did not exceed 2 m to accommodate the working range of the robotic arm.

[0042] (2) Automatic generation algorithm of the optimal spraying path: The spraying path is planned based on the geometric shape of the bridge end spraying model. The principle is to spray from top to bottom, with the lower spray covering half of the upper spray, and spray from the left, middle, and right spraying areas in sequence. By adopting a layered spraying method, the designed spraying thickness is achieved through multiple spraying. When spraying each layer, the height and movement speed of the nozzle are adjusted to ensure that the paint evenly covers the bridge end, thereby realizing the automatic generation of the optimal spraying path.

[0043] The spiral spray path algorithm, based on time window optimization, introduces a time window constraint (spray time deviation per layer ≤ 5%) based on zigzag scanning. This algorithm uses a spiral filling path to reduce acceleration spikes around corners. The paint drying time constant (τ = 15 minutes at 23°C) is calculated to determine the waiting time between layers.

[0044] Regarding time window constraints: Among them, ti is the spraying time of the i-th layer, is the average time, =0.05.

[0045] 2. Spiral path generation: The radius r(t) decreases linearly, and the angular velocity θ(t) increases uniformly.

[0046] Time window deviation threshold =5% to ensure uniform coating thickness; the starting radius of the spiral path is 1 / 2 of the sub-area width, the angular velocity is 0.1rad / s, and the paint drying time is corrected based on real-time data from the temperature and humidity sensor (accuracy ±1°C, ±5%RH).

[0047] 3) Formatted Data Synthesis Method: After the spraying path plan is automatically generated, the path plan data needs to be formatted and converted into action instructions that can be executed by the actuator. This mainly completes the data mapping task, including the execution order of sub-areas and the start and end positions of each sub-area relative to the reference point, to form formatted data. Once automatic spraying is enabled, the equipment will execute the formatted instructions to complete the spraying process, and the current execution instructions will be recorded during the process to facilitate abnormal recovery.

[0048] 4. Precision spraying adaptive control technology based on multi-sensor data fusion Various sensor devices are pre-installed and combined with the position feedback of the action mechanism, real-time perception of all elements of the equipment operation process is achieved. It has functions such as automatic equipment correction, dynamic angle compensation, adaptive adjustment of nozzle angle, linkage control of multi-degree-of-freedom action actuators, and smooth control of equipment operation process. At the same time, environmental data, operation time and other information are fed back to the control system as control parameters to ensure that the construction environment and operation cycle meet the requirements of spray material construction and realize refined control.

[0049] Among them, (1) dynamic angle compensation algorithm: in view of the uneven ground conditions at the construction site, a high-precision inclination sensor is pre-installed to obtain the three-axis angle offset of the actuator in real time, and an error model is constructed to automatically output the compensation amount of the action actuator, accurately adjust the position and posture of the actuator, and ensure that it performs the task according to the predetermined trajectory and accuracy.

[0050] Among them, the dynamic angle compensation algorithm based on neural network prediction uses an LSTM neural network (12-dimensional input layer, 64-dimensional hidden layer, and 6-dimensional output layer) to learn the mapping relationship between terrain undulation and angle deviation. The tilt sensor (accuracy 0.005°) data is input into the model after sliding average filtering (window size 10).

[0051] About LSTM unit calculation: in, is the forget gate, ot is the output gate, is the Sigmoid function.

[0052] Compensation calculation: Scaling factor λ=0.95.

[0053] The LSTM model is trained using 1000 sets of field terrain data, with a sliding average filter window size of 10 corresponding to 100ms smoothing. The compensation scaling factor \lambda avoids over-adjustment to ensure system stability.

[0054] (2) Smooth operation control algorithm for multi-DOF motion actuators: The intelligent spraying equipment for bridge end faces contains multiple motion actuators, and the working surface covers an area of ​​8m*3.5m, which is relatively large. A smooth operation control algorithm for multi-DOF motion actuators was developed. Based on the planned path, the parameters of each drive motor motion control are output in real time, including acceleration and deceleration, process smooth operation speed, and the time period for each action execution, so as to achieve fast, efficient, safe and stable operation of each motion actuator.

[0055] A multi-degree-of-freedom smooth trajectory algorithm based on fuzzy control collects path curvature (threshold 0.1 / m) and velocity deviation (±10%). A fuzzy controller (with two input variables and three output variables) dynamically adjusts jerk. A motor load-current mapping table (with a resolution of 0.1A) is established to prevent overloads.

[0056] The fuzzy rules are expressed as: 2. Acceleration calculation: Proportional coefficient .

[0057] The fuzzy controller has 7 levels of input variable quantization and 5 levels of output, and uses the center of gravity method for defuzzification; the motor current threshold is set to 120% of the rated current, and the speed is automatically reduced by 30% when the overload protection is triggered.

[0058] 3) Real-time sensing active protection system: Based on the construction process requirements of polyurethane waterproof coatings on bridge end faces, a real-time sensing active protection system was developed to monitor ambient temperature and wind speed to prevent the equipment from operating in unsuitable outdoor environments. Environmental information is used as a compensation parameter and fed back into the spraying process control. Automated control of the spraying material stirring process and dynamic control of the spraying material temperature are achieved to ensure that the spraying material is in optimal performance. Spraying time control is achieved to ensure that the spraying material is used up within the set time. Based on the feedback of the actuator position information, active speed reduction control is performed when approaching the limit, realizing a limit soft protection mechanism.

[0059] 4) Multi-mechanism linkage control algorithm: To achieve precise control of the nozzle movement, a multi-mechanism linkage control algorithm based on PID algorithm, kinematics forward solution, and kinematics inverse solution is developed. Combined with position feedback, the acceleration, constant speed, and deceleration sections of each mechanical mechanism’s movement are precisely controlled, and a linkage control strategy is formulated to achieve orderly execution and linkage control of multiple mechanisms.

[0060] A real-time multi-mechanism linkage algorithm based on a twin model was used to construct a digital twin of the spray equipment (including the robotic arm, printhead, and chassis dynamics models). The linkage errors were predicted through simulation using a physics engine (NVIDIA PhysX). A model predictive control (MPC) solver (with a solution time of ≤10ms) was used to generate linkage control variables.

[0061] The twin model dynamics is expressed as: Among them, M is the inertia matrix, C is the Coriolis force matrix, G is the gravity vector, is the joint torque.

[0062] Linkage optimization: The deviation between the twin model and the actual system state is minimized.

[0063] The simulation step of the physics engine is 1ms, the MPC solution time domain N=20 (corresponding to 20ms), the robot arm joint torque limit τmax=50N·m, and the nozzle opening and closing delay compensation time is 50ms.

[0064] 5) Precise control method for spraying quality: The built-in laser distance sensor monitors the distance between the nozzle and the end surface in real time, and automatically adjusts the nozzle's front and rear position, direction, pressure, and flow rate according to the planned path to ensure the stability of the spraying operation and improve the quality of the nozzle's finished surface.

[0065] Example 2 This embodiment provides an intelligent spray control system for large bridge end surfaces, including: An omnidirectional four-wheel drive chassis, a multi-stage horizontal slide structure, a nozzle lifting and rotating structure, and a coating mechanism. The omnidirectional four-wheel drive chassis is used to drive the device forward, backward, turn left, turn right, and rotate on the spot; the multi-stage horizontal slide structure is used to output torque to achieve multi-stage linkage control; the nozzle lifting and rotating structure is used to achieve three-dimensional spatial positioning of the nozzle, and cover the entire working surface of the bridge bottom plate and web; the coating mechanism is used to realize the ratio control and stirring function of the bridge end face spraying material, and the control system is connected through the reserved communication interface to complete the pressure, flow, amplitude and angle control of the spraying process.

[0066] The above are all preferred embodiments of the present invention, and are not intended to limit the scope of protection of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for controlling intelligent spraying of large bridge end faces, characterized in that: include: Obtain bridge end face dimension data and various sensor data; Positioning the starting position of bridge end surface spraying equipment based on machine vision and multi-sensor calibration methods; Construct a bridge end face spraying model and obtain the optimal spraying path based on the path generation algorithm; Intelligent adaptive control of the spray mechanism is performed based on multi-sensor data fusion, in which the position and posture of the actuator are adjusted using a dynamic angle compensation algorithm; The multi-degree-of-freedom smooth operation control algorithm is used to control the speed and action cycle of the actuator; the multi-mechanism joint control algorithm is used to control the linkage control of the actuator and the nozzle.

2. The intelligent spraying control method for large bridge end faces according to claim 1 is characterized in that: The method based on machine vision and multi-sensor calibration is used to locate the starting position of the bridge end face spraying equipment, including using manual remote control guidance or automatic following methods, using machine vision and multi-sensor calibration to move the equipment to the corresponding working position, and based on the self-correction algorithm and automatic deviation correction control algorithm to enable the equipment to complete posture self-correction, combined with the spatial three-dimensional coordinate fusion algorithm, to achieve accurate positioning of the starting position of the bridge end face intelligent spraying equipment operation. Among them, through the machine vision algorithm, automatic following and guiding personnel are achieved, moving to the corresponding position, extracting the current scene image features, performing mark position matching and identification, and completing the calibration of the camera coordinate system, pixel coordinate system, and bridge end face coordinate system.

3. The intelligent spraying control method for large bridge end faces according to claim 2 is characterized in that: The self-correction algorithm and automatic deviation correction control algorithm are used to enable the equipment to complete posture self-correction, including calculating the deviation between the equipment and the bridge end face through data fed back by the GPS-RTK and laser ranging sensors at the front and rear ends of the equipment, forming the input of the automatic deviation correction control algorithm, and adopting an improved PID adjustment algorithm. The output needs to be adjusted to the angle. With the center of the equipment as the reference point, an in-situ rotation control method is adopted to convert the required adjustment amount into position information of the steering motor, driving the chassis to complete the steering. During the process, the distance between the front and rear ends and the bridge end face is continuously monitored and fed back to the automatic deviation correction process control to ensure that the equipment remains parallel to the bridge end face.

4. The intelligent spraying control method for large bridge end faces according to claim 3 is characterized in that: The combined spatial three-dimensional coordinate fusion algorithm is used to achieve precise positioning of the starting position of the bridge end face intelligent spraying equipment operation, including constructing a bridge end face coordinate system based on the bridge end face size data and various sensor data, utilizing the equipment action actuator and the coordinate system spatial coordinate fusion algorithm, and according to the principle of spatial coordinate transformation, utilizing the rotation matrix and translation vector to realize the mutual conversion between the equipment action actuator coordinate system and the bridge end face coordinate system, with the actual coordinates of the bridge section as the target, realizing precise transmission of the action actuator.

5. The intelligent spraying control method for large bridge end faces according to claim 4 is characterized in that: The bridge end face spraying model is constructed, and the optimal spraying path is obtained based on the path generation algorithm, including the spraying area division after optimization based on the bridge end face dimension information and actual measurement information. The bridge end face has a bilaterally symmetrical structure and is normally divided into three spraying areas, namely the left, right and middle areas. After the spraying area is divided, each area is subdivided into subareas, and irregular areas are filled with regular shapes. The principle is that subareas on the same horizontal line have regular shapes and consistent sizes, while subareas at different heights have different sizes. The subarea division of the three areas is completed in sequence, and the area number, layer number, subarea number, and length, width, slope, and start time data of each subarea number are automatically generated.

6. The intelligent spraying control method for large bridge end faces according to claim 5 is characterized in that: The method of constructing a bridge end face spraying model and obtaining an optimal spraying path based on a path generation algorithm also includes planning a spraying path according to the geometric shape of the bridge end face spraying model, spraying from top to bottom, with the lower spray covering half of the upper spray, and spraying from the left, middle, and right spraying areas in sequence; using a layered spraying method, achieving a designed spraying thickness through multiple sprayings; and adjusting the height and movement speed of the nozzle during each layer spraying so that the paint evenly covers the bridge end face, thereby realizing automatic generation of the optimal spraying path.

7. The intelligent spraying control method for large bridge end faces according to claim 6 is characterized in that: The dynamic angle compensation algorithm is used to adjust the position and posture of the actuator. This includes obtaining the three-axis angle offset of the actuator in real time through a pre-installed inclination sensor in response to the uneven ground at the construction site, building an error model, automatically outputting the compensation amount of the action actuator, and adjusting the position and posture of the actuator. The LSTM neural network is used to learn the mapping relationship between terrain undulation and angle deviation. The inclination sensor data is input into the model after sliding average filtering. The LSTM model is trained with 1,000 sets of on-site terrain data, and the sliding average filter window size is 10 corresponding to 100ms smoothing processing. The compensation amount scaling coefficient is used to avoid over-adjustment and ensure system stability.

8. The intelligent spraying control method for large bridge end faces according to claim 7 is characterized in that: The multi-degree-of-freedom smooth operation control algorithm is used to control the speed and action cycle of the actuator, including real-time output of the parameters of the drive motor motion control according to the planned path, including acceleration and deceleration, process operation smooth speed, and the time period of each action execution, thereby realizing the operation of each action actuator. Among them, the multi-degree-of-freedom smooth trajectory algorithm based on fuzzy control collects path curvature and speed deviation, dynamically adjusts the acceleration through the fuzzy controller, and establishes a motor load-current mapping table to prevent overload.

9. The intelligent spraying control method for large bridge end faces according to claim 8 is characterized in that: The multi-mechanism linkage control algorithm is used to control the actuator and the nozzle linkage control, including a multi-mechanism linkage control algorithm based on PID algorithm, kinematics forward solution, and kinematics inverse solution. The acceleration section, uniform speed section, and deceleration section of the movement of each mechanical mechanism are controlled by position feedback, and a linkage control strategy is formulated to achieve orderly execution and linkage control of multiple mechanisms. Among them, a digital twin of the spraying equipment is constructed, the mechanism linkage error is predicted through physical engine simulation, and the model predictive control MPC solver is used to generate the linkage control quantity.

10. An intelligent spray control system for large bridge end surfaces, executing the method according to any one of claims 1 to 9, comprising: An omnidirectional four-wheel drive chassis, a multi-stage horizontal slide structure, a nozzle lifting and rotating structure, and a coating mechanism. The omnidirectional four-wheel drive chassis is used to drive the device forward, backward, turn left, turn right, and rotate on the spot; the multi-stage horizontal slide structure is used to output torque to achieve multi-stage linkage control; the nozzle lifting and rotating structure is used to achieve three-dimensional spatial positioning of the nozzle, and cover the entire working surface of the bridge bottom plate and web; the coating mechanism is used to realize the ratio control and stirring function of the bridge end face spraying material, and the control system is connected through the reserved communication interface to complete the pressure, flow, amplitude and angle control of the spraying process.

Images