An adaptive closed-loop control system for spraying process parameters

Abstract

This invention relates to the field of industrial automation technology and discloses an adaptive closed-loop control system for spraying process parameters. The system includes a process parameter monitoring module, a process execution control module, an online quality inspection module, a central intelligent control module, and an intelligent early warning and knowledge management module. The process parameter monitoring module constructs a digital mirror of the spraying process in real time, and the central intelligent control module runs a digital twin model and optimization algorithm to dynamically generate optimal control commands. The process execution control module then coordinates the adjustment of various process parameters, achieving adaptive closed-loop optimization and control of the spraying process, improving process accuracy and response speed. The online quality inspection module performs non-contact online detection of wet and dry films, identifying various coating defects in real time. Combined with the process status early warning unit of the intelligent early warning and knowledge management module, real-time monitoring and protection of coating quality and the production process are achieved.

Description

Technical Field

[0001] This invention relates to the field of industrial automation technology, specifically to an adaptive closed-loop control system for spraying process parameters. Background Technology

[0002] Spraying is a coating method that uses a spray gun or disc atomizer to disperse the material into uniform and fine droplets with the help of pressure or centrifugal force and apply them to the surface of the object to be coated. It can be divided into air spraying, airless spraying, electrostatic spraying, and various derivative methods of basic spraying forms, such as high-flow-rate low-pressure atomization spraying, thermal spraying, automatic spraying, and multi-group spraying.

[0003] In the field of spraying technology, traditional methods mainly rely on preset fixed process parameters and human experience for open-loop control, making it difficult to perceive and coordinate dynamic parameters of multiple dimensions such as ambient temperature and humidity, paint flow state, and robot motion trajectory in real time.

[0004] Therefore, an adaptive closed-loop control system for spraying process parameters is proposed to solve the above problems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an adaptive closed-loop control system for spraying process parameters, which solves the problem mentioned in the background art of difficulty in real-time and comprehensive perception and coordinated adjustment of multi-dimensional dynamic parameters such as ambient temperature and humidity, paint flow state, and robot motion trajectory during the spraying process.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an adaptive closed-loop control system for spraying process parameters, the system comprising a process parameter monitoring module, a process execution control module, an online quality detection module, and a central intelligent control module; The process parameter monitoring module is used to collect multi-dimensional dynamic parameters in real time throughout the entire spraying process, including spraying environment parameters, paint flow parameters, actuator motion parameters, and workpiece state parameters, and to construct a real-time digital image reflecting the current spraying conditions through multi-source data fusion technology. The process execution control module is used to receive optimization instructions from the central intelligent control module and to coordinate the control of the spraying robot, spray gun, paint supply and mixing system, temperature and humidity control device and conveying mechanism, and dynamically adjust the spraying trajectory, atomization pressure, paint flow rate, spray width, fan-shaped air pressure, electrostatic voltage, ambient temperature and humidity and chain speed process parameters. The online quality inspection module is used to perform non-contact online inspection of the wet film and dry film on the workpiece surface during the spraying process and before the coating is cured. It can acquire real-time image data of paint film thickness, uniformity, gloss, color difference and surface defects, and identify defect types such as sagging, orange peel, particles, missed spraying and bubbles based on visual algorithms and spectral analysis technology. The central intelligent control module is used to integrate and process multi-source heterogeneous data from the monitoring and detection module, run the built-in digital twin simulation model and process optimization algorithm, and dynamically generate and issue the optimal process parameter control command to the process execution control module by comparing the deviation between the real-time process status and the target quality model. The system also integrates an intelligent early warning and knowledge management module, which issues multi-level early warnings when process parameters exceed limits, quality defects exceed tolerances, or control efficiency is insufficient. It also continuously stores and learns the entire process data to build an iteratively optimized spraying process knowledge base.

[0007] Preferably, the process parameter monitoring module includes an environmental sensing unit, a flow field monitoring unit, a motion sensing unit, and a workpiece sensing unit; The environmental sensing unit monitors the temperature, relative humidity, cleanliness level, and airflow organization in the spray booth in real time through distributed temperature and humidity sensors, cleanliness particle counters, and anemometers. The flow field monitoring unit monitors the paint supply pressure, instantaneous flow rate, dynamic viscosity, and atomization cone angle and particle size distribution at the spray gun outlet in real time through pressure sensors, flow meters, online viscosity analyzers, and high-speed cameras integrated into the paint supply pipeline and spray gun. The motion sensing unit acquires the spatial trajectory coordinates, movement speed, acceleration, and attitude angle and spraying distance of the spray gun end in real time through the spraying robot body encoder, external laser tracker and spray gun attitude sensor. The workpiece sensing unit uses a 3D scanner and an RFID reader / writer to identify the workpiece's identification code and obtain its 3D point cloud model before spraying, providing geometric input for subsequent trajectory planning and film thickness simulation.

[0008] Preferably, the environmental sensing unit further includes a VOC concentration monitoring component, used to monitor the concentration of volatile organic compounds in the spray booth in real time, and to work in conjunction with the ventilation system to achieve safe emission control; The flow field monitoring unit also includes a paint mist overspray monitoring component, which uses a settling film and optical sensors placed around the workpiece to quantitatively evaluate the paint utilization rate and the distribution of oversprayed paint mist. The motion sensing unit uses multi-sensor fusion positioning technology, combining encoder data and laser tracker data to compensate for trajectory errors caused by robot joint flexibility and temperature drift in real time. The workpiece sensing unit is equipped with an infrared thermal imager at key workstations to detect the uniformity of the temperature field on the workpiece surface after the preheating and flash-drying processes.

[0009] Preferably, the process execution control module includes a robot trajectory control unit, a spray gun parameter control unit, an environmental control unit, and a conveying speed control unit; The robot trajectory control unit dynamically adjusts the motion path, speed profile, and posture sequence of the spraying robot according to the optimized trajectory file issued by the central intelligent control module, so as to realize the contour spraying of complex curved workpieces and the control of variable speed and variable distance switch gun. The spray gun parameter control unit uses a high-response proportional valve and a servo driver to coordinately adjust the paint output, atomizing air pressure, fan-shaped air pressure, and voltage and current of the electrostatic generator. The environmental control unit adjusts the working status of the spray booth air supply and exhaust unit, heater, humidifier and dehumidifier through the programmable logic controller to maintain the temperature, humidity and cleanliness of the spray booth within the process setting window; The conveyor speed control unit is used to control the running speed of the overhead chain and the ground conveyor, and to achieve synchronous speed control and positioning with the robot's spraying action according to the workpiece type and spraying cycle requirements.

[0010] Preferably, the robot trajectory control unit has an online trajectory replanning function. When the online quality detection module detects insufficient local film thickness, it generates a local respray trajectory in real time and inserts it into the original operation program. The spray gun parameter control unit integrates an adaptive fuzzy PID controller, which automatically adjusts the control parameters based on the dynamic deviation between the paint flow rate setpoint and the actual feedback value. The environmental control unit adopts a combined feedforward and feedback control strategy, predicts the control amount based on the heat capacity of the workpiece material to be coated and the initial environmental conditions, and performs feedback fine-tuning in combination with real-time monitoring data. The conveyor speed control unit is equipped with a servo drive and a high-precision position encoder, which supports the start and stop of the conveyor line, speed change and fixed-point stop, and maintains bus communication with the robot control system.

[0011] Preferably, the online quality inspection module includes a wet film inspection unit, a dry film inspection unit, and a defect analysis unit; The wet film detection unit, after spraying and before leveling, uses a laser triangulation sensor and a confocal chromatograph to measure the thickness and surface contour waviness of the wet paint film online in a non-contact manner, providing early feedback on the coating rate and leveling status. After the coating has cured, the dry film detection unit uses a multi-angle spectrophotometer, a machine vision camera, and a laser thickness gauge to detect the thickness, gloss, color coordinates, and macroscopic defects of the dry paint film online. The defect analysis unit integrates a deep learning-based image recognition algorithm to automatically analyze the surface images acquired by the dry film detection unit, classify and identify various defect types such as orange peel, particles, shrinkage cavities, bubbles, and blooming, and count their size, quantity, and distribution density.

[0012] Preferably, the wet film detection unit adopts a multi-sensor array scanning method to cover the key areas of the workpiece in a line scanning manner; The dry film detection unit is equipped with an adjustable light source system and a multi-axis motion mechanism to adapt to the stable measurement requirements of workpieces with different colors, glosses, and curvatures, and to eliminate interference from measurement angles and stray light. The defect analysis unit has a built-in defect feature database and judgment rule base. It performs correlation analysis between the identified defect features and the historical process parameter records to initially trace the process stage and cause of the defect.

[0013] Preferably, the central intelligent control module includes a digital twin simulation unit, a process optimization decision-making unit, a collaborative scheduling unit, and a human-machine interaction unit; The digital twin simulation unit runs a high-fidelity physical model of the spraying process based on the three-dimensional model of the workpiece, the coating characteristic parameters, and the real-time collected environmental and motion parameters. It predicts the coating thickness distribution, paint utilization rate, and key quality indicators in real time, forming a virtual image that is synchronized with the physical spraying process. The process optimization decision unit compares and analyzes the online quality inspection results with the prediction results of the digital twin simulation unit, and uses the model predictive control algorithm to dynamically calculate the optimal set of control strategies to eliminate quality deviations. The collaborative scheduling unit is responsible for decomposing the control strategy generated by the optimization decision unit into specific, time-sequential control instructions, and synchronously sending them to each subunit of the process execution control module. The human-computer interaction unit provides a graphical, configurable integrated operation interface, supporting functions such as importing spraying tasks, managing process formulas, 3D simulation preview, real-time monitoring visualization, quality report generation, and historical data traceability.

[0014] Preferably, the digital twin simulation unit uses a computational fluid dynamics and coating growth coupled model to simulate the movement, deposition and film formation of paint mist in the air, and the simulation results serve as a reference for process optimization. The process optimization decision unit integrates a reinforcement learning agent. This agent takes paint film uniformity, material consumption rate, and production cycle as comprehensive optimization objectives, and spray gun trajectory, flow rate, pressure, and speed as the main decision variables. Through continuous interaction with the environment and reward feedback, it learns autonomously and outputs a combination of process parameters that approximates the global optimum. The collaborative scheduling unit adopts an event-triggered distributed control architecture, and each actuator controller has edge computing and autonomous fault tolerance capabilities based on receiving a unified clock synchronization. The human-computer interaction unit supports virtual reality and augmented reality interfaces, allowing operators to immerse themselves in viewing spraying path simulation, real-time quality heat maps, and receiving remote expert guidance through wearable devices.

[0015] Preferably, the intelligent early warning and knowledge management module includes a process status early warning unit and a process knowledge base unit; The process status early warning unit is set with dynamic safety thresholds for process parameters and statistical process control limits for quality indicators. When the monitoring and detection data continuously deviate from the set range and the process parameter control response fails, it triggers operation interface prompts, audible and visual alarms, and production line interlock shutdown signals in stages. The process knowledge base unit is used to store environmental data, equipment parameters, control instructions, quality results and control logs for each spraying task. Through data mining and machine learning technologies, it automatically summarizes successful process parameter packages for different workpieces, different coatings and different quality requirements.

[0016] Compared with the prior art, the present invention provides an adaptive closed-loop control system for spraying process parameters, which has the following beneficial effects: 1. In this invention, a digital image of the spraying process is constructed in real time through a process parameter monitoring module, and the digital twin model and optimization algorithm are run through a central intelligent control module to dynamically generate the optimal control instructions. The process execution control module coordinates the adjustment of various process parameters, thereby realizing adaptive closed-loop optimization and control of the spraying process and improving the accuracy and response speed of the process.

[0017] 2. In this invention, the wet film and dry film are non-contactly inspected online through the online quality inspection module, which identifies various coating defects in real time. Combined with the process status early warning unit of the intelligent early warning and knowledge management module, multi-level early warnings can be issued when quality defects exceed tolerances or parameters exceed limits, thereby realizing real-time monitoring and protection of coating quality and production process.

[0018] 3. In this invention, the process knowledge base unit in the intelligent early warning and knowledge management module continuously stores and learns the process data of the entire process, and automatically summarizes the successful process parameter package using data mining and machine learning technologies, thus constructing an iteratively optimizeable spraying process expert system, which provides decision support for process optimization and new product trial production. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the adaptive closed-loop control system for spraying process parameters according to the present invention. Figure 2This is a schematic diagram of the process parameter monitoring module in this invention; Figure 3 This is a flowchart illustrating the operation steps of an adaptive closed-loop control system for spraying process parameters according to the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] For specific implementation examples, please refer to: Figures 1-3 An adaptive closed-loop control system for spraying process parameters, comprising a process parameter monitoring module, a process execution control module, an online quality detection module, and a central intelligent control module; The process parameter monitoring module is used to collect multi-dimensional dynamic parameters in real time throughout the entire spraying process, including spraying environment parameters, paint flow parameters, actuator motion parameters, and workpiece status parameters. It also constructs a real-time digital image reflecting the current spraying conditions through multi-source data fusion technology. The process execution control module is used to receive optimization instructions from the central intelligent control module and coordinate the control of the spraying robot, spray gun, paint supply and mixing system, temperature and humidity control device and conveying mechanism to dynamically adjust the spraying trajectory, atomization pressure, paint flow rate, spray width, fan-shaped air pressure, electrostatic voltage, ambient temperature and humidity and chain speed process parameters. The online quality inspection module is used to perform non-contact online inspection of the wet and dry film on the workpiece surface during the spraying process and before the coating is cured. It can acquire real-time image data of paint film thickness, uniformity, gloss, color difference and surface defects, and identify defect types such as sagging, orange peel, particles, missed spraying and bubbles based on visual algorithms and spectral analysis technology. The central intelligent control module is used to integrate and process multi-source heterogeneous data from the monitoring and detection module, run the built-in digital twin simulation model and process optimization algorithm, and dynamically generate and issue the optimal process parameter control command to the process execution control module by comparing the deviation between the real-time process status and the target quality model, so as to realize the adaptive closed-loop optimization and control of the spraying process. The system also integrates an intelligent early warning and knowledge management module, which issues multi-level early warnings when process parameters exceed limits, quality defects exceed tolerances, or control efficiency is insufficient. It also continuously stores and learns the entire process data to build an iteratively optimized spraying process knowledge base.

[0022] The process parameter monitoring module includes an environmental sensing unit, a flow field monitoring unit, a motion sensing unit, and a workpiece sensing unit; The environmental sensing unit monitors the temperature, relative humidity, cleanliness level, and airflow organization in the spray booth in real time through distributed temperature and humidity sensors, cleanliness particle counters, and anemometers. The flow field monitoring unit monitors the paint supply pressure, instantaneous flow rate, dynamic viscosity, and atomization cone angle and particle size distribution at the spray gun outlet in real time through pressure sensors, flow meters, online viscosity analyzers, and high-speed cameras integrated into the paint supply pipeline and spray gun. The motion sensing unit uses the robot body encoder, external laser tracker and spray gun attitude sensor to obtain the spatial trajectory coordinates, movement speed, acceleration and attitude angle and spraying distance of the spray gun end in real time. The workpiece sensing unit uses a 3D scanner and an RFID reader / writer to identify the workpiece's identification code and obtain its 3D point cloud model before spraying, providing geometric input for subsequent trajectory planning and film thickness simulation.

[0023] The environmental sensing unit also includes a VOC concentration monitoring component, which is used to monitor the concentration of volatile organic compounds in the spray booth in real time and to work in conjunction with the ventilation system to achieve safe emission control. The flow field monitoring unit also includes a paint mist overspray monitoring component, which uses a settling film and optical sensors placed around the workpiece to quantitatively assess the paint utilization rate and the distribution of oversprayed paint mist. The motion sensing unit uses multi-sensor fusion positioning technology, combining encoder data and laser tracker data to compensate for trajectory errors caused by robot joint flexibility and temperature drift in real time. In practical implementation, multi-sensor fusion localization technology is achieved through a Kalman filter. First, the filter state is initialized: the pose estimate and its covariance given by the robot encoder are used as the initial state estimate and initial error covariance of the filter; the measurement noise covariance matrix of the laser tracker is defined as... ; Subsequently, the filter performs the following recursive calculation in each control cycle: Prediction steps: ; ; in Let k be the prior state estimate at time k. Here is the state transition matrix. For the posterior state estimation at time k-1, Let be the error covariance matrix of the prior estimate at time k. Let be the error covariance matrix of the posterior estimate at time k-1. The process noise covariance matrix; Update steps: ; ; ; in For Kalman gain, For the observation matrix, To measure the noise covariance matrix, The actual measured value at time k. Let k be the posterior state estimate at time k. It is the identity matrix; This process is repeated cyclically, utilizing the continuous high-frequency characteristics of the encoder and the absolute precision of the laser tracker to output the fused pose in real time, suppressing the cumulative error and temperature drift of the encoder, thus providing a foundation for spraying. The workpiece sensing unit is equipped with an infrared thermal imager at key workstations to detect the uniformity of the temperature field on the workpiece surface after the preheating and flash-drying processes.

[0024] The process execution control module includes a robot trajectory control unit, a spray gun parameter control unit, an environmental control unit, and a conveyor speed control unit; The robot trajectory control unit dynamically adjusts the motion path, speed profile, and posture sequence of the painting robot according to the optimized trajectory file issued by the central intelligent control module, so as to realize the contour painting of complex curved workpieces and the control of variable speed and variable distance switch gun. In practical implementation, for complex curved workpieces, the system first generates an initial spraying trajectory based on the workpiece's 3D CAD model using the equidistant offset method; assuming the workpiece surface is S(u,v), the trajectory point P(s) and attitude O(s) at the end of the spray gun can be parameterized as follows: ; ; Where P(s) is the trajectory point at the end of the spray gun, S(u,v) is the parametric equation of the working surface, s is the arc length parameter of the trajectory, u(s) and v(s) are the surface parameters corresponding to the arc length parameter s, d is the set spraying distance, n(u(s),v(s)) is the unit normal vector of the surface at the point (u(s),v(s)), O(s) is the attitude of the end of the spray gun, b(s) is the unit binormal vector, i.e., b(s) = n(s) × t(s), n(s) is the unit normal vector, and t(s) is the unit tangent vector; The velocity profile is dynamically planned based on curvature, reducing the velocity in areas of high curvature to ensure paint film uniformity. ; Where v(s) is the planned movement speed at arc length s. The maximum allowable moving speed is given by c, where c is the attenuation coefficient and q(s) is the curvature of the trajectory at arc length s. The on / off control of the spray gun is achieved by judging the relationship between the distance between trajectory points and the spray width. When the distance between trajectory points is greater than 1.2 times the spray width, the spray gun is turned off. The spray gun parameter control unit uses a high-response proportional valve and a servo driver to coordinately adjust the paint output, atomizing air pressure, fan-shaped air pressure, and voltage and current of the electrostatic generator. The environmental control unit regulates the working status of the spray booth's air supply and exhaust units, heaters, humidifiers, and dehumidifiers through a programmable logic controller to maintain the temperature, humidity, and cleanliness of the spray booth within the process setting window. The conveyor speed control unit is used to control the running speed of the overhead chain and ground conveyor, and to achieve synchronous speed control and positioning with the robot's spraying action according to the workpiece type and spraying cycle requirements. Synchronous control of the conveyor line is achieved through real-time Ethernet communication between the robot controller and the conveyor line servo driver. The transformation relationship between the robot coordinate system and the conveyor line movement coordinate system is as follows: ; in Let the pose matrix of the robot's end effector in the world coordinate system be given. This represents the pose matrix of the transport line coordinate system relative to the world coordinate system, which varies with time. This is a fixed pose matrix for the robot's end effector relative to the conveyor line coordinate system; the robot receives position and speed feedback from the conveyor line encoder in real time and uses this to correct its trajectory, achieving dynamic tracking spraying.

[0025] The robot trajectory control unit has an online trajectory replanning function. When the online quality detection module detects insufficient local film thickness, it generates a local respray trajectory in real time and inserts it into the original operation program. The online trajectory replanning function is triggered when local film thickness is insufficient. The system generates a local respray trajectory centered on the defect area. The generation of the respray trajectory can be regarded as an optimization problem. The objective function is to minimize the respray time and trajectory length, and the constraint is to avoid collision with the original trajectory and the workpiece. The spray gun parameter control unit integrates an adaptive fuzzy PID controller, which automatically adjusts the control parameters based on the dynamic deviation between the paint flow setpoint and the actual feedback value. The core algorithmic expression of the control law is: ; Where u(t) is the output of the controller at time t. This is the proportional gain coefficient. This is the integral gain coefficient. The differential gain coefficient, Let y(t) be the deviation value at time t, which is defined as the difference between the set value and the actual measured value, i.e., e(t) = r(t) - y(t). The environmental control unit adopts a combined feedforward and feedback control strategy. It predicts the control amount based on the heat capacity of the workpiece material and the initial environmental conditions, and performs feedback fine-tuning in combination with real-time monitoring data to achieve rapid stabilization of temperature and humidity and energy-saving operation. The implementation of the feedforward and feedback combined control strategy is as follows: the feedforward controller directly calculates the compensation amount based on the thermodynamic model of the controlled object; the feedback controller calculates the adjustment amount based on the deviation between the temperature setpoint and the actual value; the total control quantity is: ; Where u(t) is the total control output. This is the feedforward control quantity. The feedback control quantity is calculated by the PID controller based on the real-time deviation. The feedforward control quantity can be estimated based on the heat balance equation: ; in For the system feedforward gain, Set the temperature value. This is the measured ambient temperature value; The conveyor speed control unit is equipped with a servo drive and a high-precision position encoder, which supports the start and stop of the conveyor line, speed change and fixed-point stop, and maintains bus communication with the robot control system to ensure that the two move synchronously.

[0026] The online quality inspection module includes a wet film inspection unit, a dry film inspection unit, and a defect analysis unit; After spraying and before leveling, the wet film detection unit uses a laser triangulation sensor and a confocal chromatograph to measure the thickness and surface contour waviness of the wet paint film online in a non-contact manner, providing early feedback on the coating rate and leveling status. After the coating has cured, the dry film inspection unit uses a multi-angle spectrophotometer, machine vision camera, and laser thickness gauge to detect the thickness, gloss, color coordinates, and macroscopic defects of the dry paint film online. The defect analysis unit integrates a deep learning-based image recognition algorithm to automatically analyze the surface images acquired by the dry film detection unit, classify and identify various defect types such as orange peel, particles, shrinkage cavities, bubbles, and blooming, and count their size, quantity, and distribution density. The deep learning-based image recognition algorithm employs a convolutional neural network with an encoder and decoder structure. The ResNet-34 backbone network encoder extracts image features, while the decoder gradually recovers spatial resolution through upsampling and skip connections, ultimately outputting a defect category probability map for each pixel. The network loss function is a weighted sum of cross-entropy loss and Dice loss. ; ; ; Where L is the total loss function. These are the weighting coefficients for cross-entropy loss and Dice loss. The loss is the cross-entropy loss, where M is the total number of defect categories. For real labels, To predict the probability that pixel o belongs to class c for the model, For Dice coefficient loss, Let be the predicted value of the i-th pixel. Let be the actual value of the i-th pixel, and N be the total number of pixels.

[0027] The wet film detection unit uses a multi-sensor array scanning method to cover the key areas of the workpiece in a line scanning manner; The dry film inspection unit is equipped with an adjustable light source system and a multi-axis motion mechanism to meet the stable measurement requirements of workpieces with different colors, glosses, and curvatures, and to eliminate interference from measurement angles and stray light. The defect analysis unit has a built-in defect feature database and judgment rule base. It performs correlation analysis between the identified defect features and the historical process parameter records to initially trace the process stage and cause of the defect. The defect feature database records the geometric features of each defect sample, namely area, perimeter, and roundness; texture features, namely gray-level co-occurrence matrix energy and entropy; and co-occurrence process parameters. Association analysis employs the Apriori association rule mining algorithm. When a strong association rule in the form of {fan-shaped air pressure below threshold A, moving speed above threshold B} => {orange peel defect appears} is found, the support and confidence must exceed a set threshold.

[0028] The central intelligent control module includes a digital twin simulation unit, a process optimization decision-making unit, a collaborative scheduling unit, and a human-machine interaction unit; The digital twin simulation unit runs a high-fidelity physical model of the spraying process based on the three-dimensional model of the workpiece, the coating characteristic parameters, and the real-time collected environmental and motion parameters. It predicts the coating thickness distribution, paint utilization rate, and key quality indicators in real time, forming a virtual image that is synchronized with the physical spraying process. The high-fidelity spraying process physical model is a coupling of computational fluid dynamics and coating growth model, which is executed in two layers: First layer: Fluid dynamics atomization and transport simulation; a transient three-dimensional two-phase flow model is established near the spray gun outlet; the governing equations for the air phase are the Navier-Stokes equations, and the paint mist phase is treated as a discrete particle swarm, tracked by solving the particle motion equations. The motion of each particle is described by the following equation: ; ; ; ; in and These are the mass and velocity vectors of the paint mist particles, respectively. For airflow drag, The drag coefficient, air density, For the windward area of ​​the particle, For air speed, For gravity, It is electrostatic force. Let E be the particle charge and E be the electric field strength; this model simulates the entire process of particles from atomization and spatial transport to impacting the workpiece surface, and outputs the spatial deposition flux distribution. ; Second layer: Coating growth and film thickness prediction; deposition flux based on hydrodynamic model output. The process involves integrating the spray gun trajectory on the workpiece surface. This step employs a discretized point spraying model, creating a biangular mesh on the workpiece surface. For any triangular facet j on the surface, the deposition thickness increment at time k is: ; in Let k be the increment of the deposition thickness at time k. It is the position of the spray gun at time k, o(k), at the center point of the surface j. The deposition flux generated at that location was obtained by interpolation using a hydrodynamic model. This is a correction factor for deposition efficiency. For time step, The dry film density of the paint film is given. By integrating the time of all surfaces over the entire spraying stroke, the coating thickness distribution of the entire workpiece can be predicted in real time. The paint utilization rate is obtained by calculating the ratio of the total solid mass deposited on the workpiece to the total solid mass sprayed out. The process optimization decision unit compares and analyzes the online quality inspection results with the prediction results of the digital twin simulation unit, and uses the model predictive control algorithm to dynamically calculate the optimal set of control strategies to eliminate quality deviations. The model predictive control algorithm solves the following finite-time optimization problem in each control cycle: ; ; ; Where min represents the minimization operation. The control increment at time k, To predict the time domain, To control the time domain, This is the system output at time k+i predicted at time k. As the expected output reference value, For matrix Weighted output error norm squared, This is the weight matrix for the output error. Let be the squared norm of the control increment, weighted by matrix R, where R is the weight matrix for the control variable changes, y is the system output variable, x is the system state variable, and u is the control input variable. To control the lower bound vector of the input variables, To control the upper limit vector of input variables, This is a system prediction model, namely a physical model of the high-fidelity spraying process; The collaborative scheduling unit is responsible for decomposing the control strategies generated by the optimization decision unit into specific, time-sequential control instructions, and synchronously sending them to each subunit of the process execution control module. The human-computer interaction unit provides a graphical, configurable integrated operation interface, supporting functions such as importing spraying tasks, managing process formulas, 3D simulation preview, real-time monitoring visualization, quality report generation, and historical data traceability.

[0029] The digital twin simulation unit uses a coupled computational fluid dynamics and coating growth model to simulate the movement, deposition and film formation of paint mist in the air, and its simulation results serve as a reference for process optimization. The computational fluid dynamics-coating growth coupled model is a more tightly coupled form of the high-fidelity physical model of the spraying process. It adopts a one-way coupling strategy: First, in the offline stage, a high-resolution transient fluid dynamics simulation is run for typical spray gun parameter combinations to generate a flow field database containing spatial deposition flux distribution, particle size, and velocity. During online operation, the coating growth model quickly interpolates and queries the corresponding deposition flux field from the database based on the current spray gun parameters, without having to solve complex fluid dynamics equations in real time. Then, the coating growth model uses this flux field, combined with the dynamic spray gun trajectory, to predict the film thickness in real time. The process optimization decision unit integrates a reinforcement learning agent. This agent takes paint film uniformity, material consumption rate, and production cycle as comprehensive optimization objectives, and spray gun trajectory, flow rate, pressure, and speed as the main decision variables. Through continuous interaction with the environment and reward feedback, it learns autonomously and outputs a combination of process parameters that approximates the global optimum. The specific implementation of the reinforcement learning agent uses a deep deterministic policy gradient algorithm to construct the process optimization agent. State space: includes current workpiece type code, ambient temperature and humidity, key geometric features, currently used process parameters, and quality feedback from the previous cycle; Action space: defined as the amount of adjustment to key process parameters, such as speed changes, flow rate changes, and atomization pressure changes, all within a continuous range; Reward function: Designed to be negatively correlated with the overall cost function to encourage optimization. ; Where r is the reward function, MSE is the mean square error between the predicted film thickness and the target film thickness, PaintUsed is the paint consumption, and Time is the operation time. , , It refers to the weight; the higher the reward value, the better the overall performance of that set of parameters. Network and Training: An agent consists of a policy network. A value network is used to select actions based on the state. It is used to evaluate the value of state-action pairs; among which Let s be the policy function, and s be the state. for Internal adjustable parameter set; For value function, For action, for The set of internally adjustable parameters; During training, the agent continuously explores and learns through trial and error in a digital twin environment; the update objective of the value network is to minimize the temporal difference error, while the update of the policy network follows the deterministic policy gradient, aiming to achieve higher values. Value direction adjustment strategy parameters After sufficient training, the policy network can directly output a set of process parameter adjustment actions that approximate the global optimum based on the current state. The collaborative scheduling unit adopts an event-triggered distributed control architecture, and each actuator controller has edge computing and autonomous fault tolerance capabilities based on receiving a unified clock synchronization. Specific implementation steps of distributed control architecture: Central Intelligent Control Layer: Composed of an industrial server and high-performance industrial control units, running the central intelligent control module; as the brain of the system, it is responsible for running complex algorithms such as digital twin simulation and process optimization decision-making, generating high-level instructions, and publishing the instructions to the global data space as the publisher through industrial communication protocols such as OPC UA and DDS.

[0030] Network communication layer: Time-sensitive network with deterministic transmission capability is adopted; industrial Ethernet switches provide high-bandwidth, low-latency, time-synchronized communication backbone for each node; the network topology is usually star or ring to ensure reliability.

[0031] Edge execution control layer: It consists of multiple edge controllers with independent processing capabilities. As subscribers, they receive instructions from the data space and as executors, they are responsible for local closed-loop control. Key edge nodes include: painting robot controller, spray gun integrated controller, environmental controller, and conveyor line controller. Each edge controller embeds a high-performance microprocessor, namely a multi-core ARM Cortex-A or FPGA, running a lightweight real-time operating system. Its edge computing tasks include local closed-loop control and local data preprocessing. Each edge controller also embeds a state machine and watchdog logic to achieve three levels of autonomous fault tolerance. Level 1: Input validity check and substitution; Level 2: Local closed-loop health monitoring and downgrade; Level 3: Communication interruption and autonomous shutdown; The human-computer interaction unit supports virtual reality and augmented reality interfaces, allowing operators to immerse themselves in viewing spray path simulations, real-time quality heat maps, and receiving remote expert guidance through wearable devices.

[0032] The intelligent early warning and knowledge management module includes a process status early warning unit and a process knowledge base unit; The process status early warning unit is set with dynamic safety thresholds for process parameters and statistical process control limits for quality indicators. When the monitoring and detection data continuously deviate from the set range and the process parameter control response fails, it will trigger operation interface prompts, audible and visual alarms, and production line interlock shutdown signals in stages. The system alerts are divided into four levels: prompt, warning, critical, and fault. The triggering logic is based on statistical process control and a rule engine. Warning level: Triggered when the process parameter of a single monitoring point deviates briefly from the set value but does not exceed the limit, and the prediction model shows a slight trend of future deviation; Response: Displays a warning message on the human-machine interface without interrupting production; Warning level: Triggered when any of the following conditions are met: The measured values ​​of the key parameters consistently exceed the 2σ control limit but do not reach 3σ, i.e. SPC rule triggered, i.e., "7 consecutive increases"; Response: Audible and visual alarm, interface highlight, operator is advised to check; where d is the current measurement value. Let σ be the process mean, and σ be the process standard deviation, 2σ and 3σ; Severity level: Triggered when any of the following conditions are met: Key quality indicators exceeding the 3σ control limit, i.e. ; Critical equipment is in abnormal condition; Autonomous fault-tolerant mechanism degrades operation; Response: Trigger production line speed reduction, automatically call alternative process parameters, and prepare for intervention; Fault level: Triggered by equipment hard fault, safety interlock trigger, or complete network interruption; Response: Immediately interlock and shut down, lock the equipment, and send an emergency message to the maintenance personnel's mobile phone; The process knowledge base unit is used to store environmental data, equipment parameters, control instructions, quality results and control logs for each spraying task. Through data mining and machine learning technologies, it automatically summarizes successful process parameter packages for different workpieces, different coatings and different quality requirements, forming an iteratively optimized spraying process expert system, providing data-driven decision support for new product trial production and process optimization. Data mining periodically performs cluster analysis on historical success cases, using Gaussian mixture models to cluster the high-dimensional process parameter space, with the goal of optimizing the log-likelihood function. ; in For the model with respect to the parameter set The log-likelihood function, where N is the total number of historical success cases. Let W be the feature vector of the process parameters for the i-th historical case, and W be the number of preset Gaussian distributions. Let w be the mean vector of the w-th Gaussian distribution. Let be the covariance matrix of the w-th Gaussian distribution.

[0033] The operation steps of an adaptive closed-loop control system for spraying process parameters are as follows: Step 1: Comprehensive Perception and Digital Mirror Construction The process parameter monitoring module collects multi-dimensional dynamic parameters of the entire spraying process in real time, including spraying environment parameters, paint flow parameters, actuator motion parameters, and workpiece status parameters. It also uses multi-source data fusion technology to construct a real-time digital image reflecting the current spraying conditions.

[0034] Step 2: Real-time monitoring and early defect identification: The online quality inspection module performs non-contact online inspection of the wet and dry films on the workpiece surface during the spraying process and before the coating cures. It acquires real-time image data of paint film thickness, uniformity, gloss, color difference, and surface defects, and identifies defect types such as sagging, orange peel, particles, missed spraying, and bubbles based on visual algorithms and spectral analysis technology.

[0035] Step 3: Intelligent Decision Making and Optimization Strategy Generation: The central intelligent control module integrates and processes multi-source heterogeneous data from the aforementioned monitoring and detection modules, runs the built-in digital twin simulation model and process optimization algorithm, and dynamically generates the optimal process parameter control strategy by comparing the deviation between the real-time process status and the target quality model.

[0036] Step 4: Collaborative Execution and Process Parameter Control The process execution and control module receives optimization instructions from the central intelligent control module and performs coordinated control of the spraying robot, spray gun, paint supply and mixing system, temperature and humidity control device and conveying mechanism, dynamically adjusting process parameters such as spraying trajectory, atomization pressure, paint flow rate, spray width, fan-shaped air pressure, electrostatic voltage, ambient temperature and humidity and chain speed.

[0037] Step 5: Early Warning Learning and Closed-Loop Optimization Iteration The integrated intelligent early warning and knowledge management module issues multi-level warnings when process parameters exceed limits, quality defects exceed tolerances, and control efficiency is insufficient. It also continuously stores and learns process data from the entire process, building an iteratively optimized spraying process knowledge base to achieve adaptive closed-loop optimization and control of the spraying process.

[0038] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0039] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An adaptive closed-loop control system for spraying process parameters, characterized in that: The system includes a process parameter monitoring module, a process execution control module, an online quality detection module, and a central intelligent control module; The process parameter monitoring module is used to collect multi-dimensional dynamic parameters in real time throughout the entire spraying process, including spraying environment parameters, paint flow parameters, actuator motion parameters, and workpiece state parameters, and to construct a real-time digital image reflecting the current spraying conditions through multi-source data fusion technology. The process execution control module is used to receive optimization instructions from the central intelligent control module and to coordinate the control of the spraying robot, spray gun, paint supply and mixing system, temperature and humidity control device and conveying mechanism, and dynamically adjust the spraying trajectory, atomization pressure, paint flow rate, spray width, fan-shaped air pressure, electrostatic voltage, ambient temperature and humidity and chain speed process parameters. The online quality inspection module is used to perform non-contact online inspection of the wet film and dry film on the workpiece surface during the spraying process and before the coating is cured. It can acquire real-time image data of paint film thickness, uniformity, gloss, color difference and surface defects, and identify defect types such as sagging, orange peel, particles, missed spraying and bubbles based on visual algorithms and spectral analysis technology. The central intelligent control module is used to integrate and process multi-source heterogeneous data from the monitoring and detection module, run the built-in digital twin simulation model and process optimization algorithm, and dynamically generate and issue the optimal process parameter control command to the process execution control module by comparing the deviation between the real-time process status and the target quality model. The system also integrates an intelligent early warning and knowledge management module, which issues multi-level early warnings when process parameters exceed limits, quality defects exceed tolerances, or control efficiency is insufficient. It also continuously stores and learns the entire process data to build an iteratively optimized spraying process knowledge base.

2. The adaptive closed-loop control system for spraying process parameters according to claim 1, characterized in that: The process parameter monitoring module includes an environmental sensing unit, a flow field monitoring unit, a motion sensing unit, and a workpiece sensing unit. The environmental sensing unit monitors the temperature, relative humidity, cleanliness level, and airflow organization in the spray booth in real time through distributed temperature and humidity sensors, cleanliness particle counters, and anemometers. The flow field monitoring unit monitors the paint supply pressure, instantaneous flow rate, dynamic viscosity, and atomization cone angle and particle size distribution at the spray gun outlet in real time through pressure sensors, flow meters, online viscosity analyzers, and high-speed cameras integrated into the paint supply pipeline and spray gun. The motion sensing unit acquires the spatial trajectory coordinates, movement speed, acceleration, and attitude angle and spraying distance of the spray gun end in real time through the spraying robot body encoder, external laser tracker and spray gun attitude sensor. The workpiece sensing unit uses a 3D scanner and an RFID reader / writer to identify the workpiece's identification code and obtain its 3D point cloud model before spraying, providing geometric input for subsequent trajectory planning and film thickness simulation.

3. The adaptive closed-loop control system for spraying process parameters according to claim 2, characterized in that: The environmental sensing unit also includes a VOC concentration monitoring component, which is used to monitor the concentration of volatile organic compounds in the spray booth in real time and to work in conjunction with the ventilation system to achieve safe emission control. The flow field monitoring unit also includes a paint mist overspray monitoring component, which uses a settling film and optical sensors placed around the workpiece to quantitatively evaluate the paint utilization rate and the distribution of oversprayed paint mist. The motion sensing unit uses multi-sensor fusion positioning technology, combining encoder data and laser tracker data to compensate for trajectory errors caused by robot joint flexibility and temperature drift in real time. The workpiece sensing unit is equipped with an infrared thermal imager at key workstations to detect the uniformity of the temperature field on the workpiece surface after the preheating and flash-drying processes.

4. The adaptive closed-loop control system for spraying process parameters according to claim 1, characterized in that: The process execution control module includes a robot trajectory control unit, a spray gun parameter control unit, an environmental control unit, and a conveying speed control unit; The robot trajectory control unit dynamically adjusts the motion path, speed profile, and posture sequence of the spraying robot according to the optimized trajectory file issued by the central intelligent control module, so as to realize the contour spraying of complex curved workpieces and the control of variable speed and variable distance switch gun. The spray gun parameter control unit uses a high-response proportional valve and a servo driver to coordinately adjust the paint output, atomizing air pressure, fan-shaped air pressure, and voltage and current of the electrostatic generator. The environmental control unit adjusts the working status of the spray booth air supply and exhaust unit, heater, humidifier and dehumidifier through the programmable logic controller to maintain the temperature, humidity and cleanliness of the spray booth within the process setting window; The conveyor speed control unit is used to control the running speed of the overhead chain and the ground conveyor, and to achieve synchronous speed control and positioning with the robot's spraying action according to the workpiece type and spraying cycle requirements.

5. The adaptive closed-loop control system for spraying process parameters according to claim 4, characterized in that: The robot trajectory control unit has an online trajectory replanning function. When the online quality detection module detects insufficient local film thickness, it generates a local respray trajectory in real time and inserts it into the original operation program. The spray gun parameter control unit integrates an adaptive fuzzy PID controller, which automatically adjusts the control parameters based on the dynamic deviation between the paint flow rate setpoint and the actual feedback value. The environmental control unit adopts a combined feedforward and feedback control strategy, predicts the control amount based on the heat capacity of the workpiece material to be coated and the initial environmental conditions, and performs feedback fine-tuning in combination with real-time monitoring data. The conveyor speed control unit is equipped with a servo drive and a high-precision position encoder, which supports the start and stop of the conveyor line, speed change and fixed-point stop, and maintains bus communication with the robot control system.

6. The adaptive closed-loop control system for spraying process parameters according to claim 1, characterized in that: The online quality inspection module includes a wet film inspection unit, a dry film inspection unit, and a defect analysis unit; The wet film detection unit, after spraying and before leveling, uses a laser triangulation sensor and a confocal chromatograph to measure the thickness and surface contour waviness of the wet paint film online in a non-contact manner, providing early feedback on the coating rate and leveling status. After the coating has cured, the dry film detection unit uses a multi-angle spectrophotometer, a machine vision camera, and a laser thickness gauge to detect the thickness, gloss, color coordinates, and macroscopic defects of the dry paint film online. The defect analysis unit integrates a deep learning-based image recognition algorithm to automatically analyze the surface images acquired by the dry film detection unit, classify and identify various defect types such as orange peel, particles, shrinkage cavities, bubbles, and blooming, and count their size, quantity, and distribution density.

7. The adaptive closed-loop control system for spraying process parameters according to claim 6, characterized in that: The wet film detection unit uses a multi-sensor array scanning method to cover the key areas of the workpiece in a line scanning manner. The dry film detection unit is equipped with an adjustable light source system and a multi-axis motion mechanism to adapt to the stable measurement requirements of workpieces with different colors, glosses, and curvatures, and to eliminate interference from measurement angles and stray light. The defect analysis unit has a built-in defect feature database and judgment rule base. It performs correlation analysis between the identified defect features and the historical process parameter records to initially trace the process stage and cause of the defect.

8. The adaptive closed-loop control system for spraying process parameters according to claim 1, characterized in that: The central intelligent control module includes a digital twin simulation unit, a process optimization decision-making unit, a collaborative scheduling unit, and a human-computer interaction unit. The digital twin simulation unit runs a high-fidelity physical model of the spraying process based on the three-dimensional model of the workpiece, the coating characteristic parameters, and the real-time collected environmental and motion parameters. It predicts the coating thickness distribution, paint utilization rate, and key quality indicators in real time, forming a virtual image that is synchronized with the physical spraying process. The process optimization decision unit compares and analyzes the online quality inspection results with the prediction results of the digital twin simulation unit, and uses the model predictive control algorithm to dynamically calculate the optimal set of control strategies to eliminate quality deviations. The collaborative scheduling unit is responsible for decomposing the control strategy generated by the optimization decision unit into specific, time-sequential control instructions, and synchronously sending them to each subunit of the process execution control module. The human-computer interaction unit provides a graphical, configurable integrated operation interface, supporting functions such as importing spraying tasks, managing process formulas, 3D simulation preview, real-time monitoring visualization, quality report generation, and historical data traceability.

9. The adaptive closed-loop control system for spraying process parameters according to claim 8, characterized in that: The digital twin simulation unit uses a computational fluid dynamics and coating growth coupled model to simulate the movement, deposition and film formation of paint mist in the air, and its simulation results serve as a reference for process optimization. The process optimization decision unit integrates a reinforcement learning agent. This agent takes paint film uniformity, material consumption rate, and production cycle as comprehensive optimization objectives, and spray gun trajectory, flow rate, pressure, and speed as the main decision variables. Through continuous interaction with the environment and reward feedback, it learns autonomously and outputs a combination of process parameters that approximates the global optimum. The collaborative scheduling unit adopts an event-triggered distributed control architecture, and each actuator controller has edge computing and autonomous fault tolerance capabilities based on receiving a unified clock synchronization. The human-computer interaction unit supports virtual reality and augmented reality interfaces, allowing operators to immerse themselves in viewing spraying path simulation, real-time quality heat maps, and receiving remote expert guidance through wearable devices.

10. The adaptive closed-loop control system for spraying process parameters according to claim 1, characterized in that: The intelligent early warning and knowledge management module includes a process status early warning unit and a process knowledge base unit; The process status early warning unit is set with dynamic safety thresholds for process parameters and statistical process control limits for quality indicators. When the monitoring and detection data continuously deviate from the set range and the process parameter control response fails, it triggers operation interface prompts, audible and visual alarms, and production line interlock shutdown signals in stages. The process knowledge base unit is used to store environmental data, equipment parameters, control instructions, quality results and control logs for each spraying task. Through data mining and machine learning technologies, it automatically summarizes successful process parameter packages for different workpieces, different coatings and different quality requirements.

Images