Intelligent imaging generation system and method

Abstract

The application discloses an intelligent imaging generation system and method, which realizes closed-loop control of spraying quality by real-time linkage adjustment of spraying mechanical arm and imaging equipment parameters through a dynamic parameter adjustment module, and combines a defect prediction and optimization decision module and a multi-modal process knowledge base. The system integrates a high-resolution visual sensor, a capacitive thickness detector and a temperature and humidity sensor, fuses multi-source data by using a Kalman filtering algorithm, and dynamically adjusts the speed of the mechanical arm, the nozzle pressure and the imaging gain coefficient. The defect prediction module simulates coating defect morphology based on a physically constrained generative adversarial network and generates a process optimization instruction set through reinforcement learning. The multi-modal process knowledge base stores an associated model of material characteristics, environmental data and defect patterns, supports incremental learning and historical data decay update. Through dynamic parameter collaborative optimization, defect prediction prepositioning and multi-modal data fusion, the spraying quality detection efficiency and coating consistency are improved, and the rework rate is reduced.

Description

A smart image generation system and method Technical Field

[0001] This invention relates to the field of spray coating production technology, and in particular to an intelligent imaging generation system and method that combines dynamic parameter adjustment, multimodal data fusion and spray coating process optimization to achieve real-time monitoring and closed-loop control of industrial spray coating quality, and is applied to spray coating production lines. Background Technology

[0002] In the industrial manufacturing sector, the spraying production line, as a crucial link in surface treatment, directly impacts the product's appearance, durability, and market competitiveness through its spraying quality. With the continuous development of intelligent manufacturing technology, spraying imaging systems, as an important component of the spraying production line, undertake the vital task of real-time monitoring and feedback control of spraying quality. However, existing spraying imaging systems still have several shortcomings in practical applications, specifically as follows:

[0003] 1. Fixed spraying imaging parameters cannot be dynamically adjusted according to the characteristics of the spraying material, leading to missed defects or overexposure. 2. Defect assessment requires manual visual inspection or offline analysis, and cannot provide real-time feedback to the spraying equipment for parameter adjustment. 3. Spraying process parameters, material data, and imaging data are processed independently, lacking cross-modal optimization capabilities. 4. When adjusting spraying process parameters, the existing system may over-invest resources in pursuit of high quality. For example, continuously adjusting spraying parameters to reduce the defect rate may not consider the increased costs associated with these adjustments, such as higher paint costs and extended production time. This can lead to a significant increase in costs while improving quality, reducing the company's economic benefits and market competitiveness. These problems result in low efficiency and insufficient accuracy in spraying quality inspection, an inability to balance quality and cost, and hinder real-time optimization and control of the spraying process.

[0004] Therefore, overcoming the aforementioned shortcomings has become an important issue that urgently needs to be addressed by those skilled in the art. Summary of the Invention

[0005] This invention overcomes the shortcomings of the above-mentioned technologies and provides an intelligent imaging generation system and method.

[0006] The purpose of this application is to provide an intelligent imaging generation system and method, as well as a spraying imaging method, which has the advantages of real-time dynamic adjustment of spraying imaging parameters, realization of multi-modal data collaborative optimization, and improvement of spraying quality detection efficiency and accuracy.

[0007] The first aspect of this application provides an intelligent image generation system and method, including:

[0008] The dynamic parameter adjustment module 11 acquires the real-time state of the sprayed surface through multi-sensor fusion and adjusts the operating parameters of the imaging device and the spraying robot arm in conjunction.

[0009] The defect prediction and optimization decision module 12 simulates the coating defect morphology under different process parameters based on a generative adversarial network model with physical constraints. It generates a process optimization instruction set through reinforcement learning. The process optimization instruction set includes: collaborative optimization instructions for coating viscosity adjustment value, spraying speed threshold, and spray gun distance parameter.

[0010] The multimodal process knowledge base 13 stores the correlation between the material properties data, environmental sensing data, historical process parameter data, and defect modes of the sprayed device.

[0011] Preferably, the dynamic parameter adjustment module 11 includes:

[0012] Multiple data acquisition units 111, including:

[0013] A high-resolution vision sensor 1111 is used to capture surface texture features and generate a first data stream;

[0014] The 1112 capacitive thickness gauge monitors the coating thickness distribution in real time and generates a second data stream.

[0015] Temperature and humidity sensor 1113 collects environmental parameters and generates a third data stream;

[0016] The data fusion unit 112 uses the Kalman filter algorithm to perform spatiotemporal alignment and noise suppression on the first, second and third data streams.

[0017] Preferably, the dynamic parameter adjustment module 11 further includes:

[0018] Intelligent sensing and decision-making unit 113, which includes:

[0019] Robotic arm moving speed adjustment unit 1131: used to extract the surface roughness feature value Ra of the sprayed surface in real time by combining the first data stream, and dynamically adjust the robotic arm moving speed v according to the preset robotic arm moving speed formula; wherein, the robotic arm moving speed formula is: v=k1*Ra+b1; k1 and b1 are roughness-speed response coefficients obtained by experimental calibration; Ra ranges from 0.1-10μm;

[0020] Nozzle pressure optimization and adjustment unit 1132: used to combine the coating thickness distribution data monitored by the second data stream and the adhesion level F obtained by the mechanical adhesion tester, and optimize the nozzle pressure P using a preset nozzle pressure optimization formula; wherein, the nozzle pressure optimization formula is: P=k2*F+b2; k2 and b2 are experimentally calibrated pressure compensation coefficients; the adhesion level F is divided into 1-5 levels;

[0021] Imaging device gain coefficient adjustment unit 1133: used to combine a third data source and use the imaging device gain coefficient compensation formula to compensate the imaging device gain coefficient G; the imaging device gain coefficient compensation formula is: G=G0*(1+α*RH%+β*T℃); G0 is the reference gain coefficient; RH is the ambient relative humidity; T is the ambient temperature; α and β are temperature and humidity compensation coefficients, α is in 1 / %, β is in 1 / ℃.

[0022] Preferably, the dynamic adjustment process of each adjustment unit of the dynamic parameter adjustment module 11 satisfies:

[0023] When the surface roughness Ra > 5μm, the workpiece surface is judged to be relatively rough, and the spiral spraying trajectory is automatically switched.

[0024] When the surface roughness Ra < 1 μm, the workpiece surface is judged to be relatively smooth, and the normal high-speed spraying trajectory is automatically switched.

[0025] When the adhesion rating F < 1, the adhesion is judged to be too low, and the nozzle pressure is automatically increased to enhance the adhesion between the coating and the substrate.

[0026] When the adhesion rating F > 5, the adhesion is judged to be too high, and the nozzle pressure is automatically reduced to avoid coating embrittlement.

[0027] When the temperature T>35℃ or the humidity RH>75%, the working environment is determined to be too hot or too humid. The gain is dynamically adjusted using the gain coefficient compensation formula of the imaging equipment to correct environmental interference.

[0028] Preferably, the dynamic parameter adjustment module 11 further includes an exception handling mechanism:

[0029] When the instantaneous wind speed is detected to be greater than a preset threshold, execute:

[0030] Pause spraying and initiate 3D laser scanning to reconstruct the surface morphology;

[0031] Generate local spraying paths based on point cloud data;

[0032] When the coating thickness non-uniformity is greater than 1%, the adaptive mesh spraying mode is triggered.

[0033] Preferably, the defect prediction and optimization decision module 12 includes:

[0034] The defect simulation unit 121 generates coating defect prediction maps under different process parameters based on a physical constraint adversarial network model; the adversarial network model is a GAN model.

[0035] The generator of the GAN model takes into account a set of process parameters, material property data, environmental data, and substrate state parameters, and outputs a coating defect prediction map under different process parameters through its discriminator. The discriminator of the GAN model integrates physical equation constraints for coating defects, including:

[0036] Flow length Where ρ is the coating density, g is the gravitational acceleration, h is the coating thickness, η is the coating viscosity, v is the spraying speed, "C" is an empirical coefficient including the dimension of length, and σ is the surface tension of the coating.

[0037] The reinforcement learning optimization unit 122 uses real-time spraying quality indicators as the state space and process parameter adjustment amounts as the action space, and optimizes process parameters through a reward function.

[0038] The closed-loop control unit 123 sends the optimized process parameters to the spraying robot arm 301 and the imaging device 302 in real time, and receives new sensor data for iterative optimization.

[0039] The reward function is defined as follows:

[0040] R = w1·(1-(∣D_{actual}-D_{target}∣ / D_{target}))+w2·C_{saved} / C_{total}; D_{actual} is the actual defect rate, D_{target} is the target defect rate, C_{saved} is the actual saved paint cost, C_{total} is the total paint cost budget, and w1 and w2 are weighting coefficients that satisfy w1+w2=1.

[0041] Preferred options also include:

[0042] The mode selection module (14) is used to dynamically select the quality priority mode and the cost priority mode according to the workpiece production type;

[0043] When the system detects a critical workpiece in production, it automatically selects the quality-first mode and limits w_1 to w_2; the system will call the preset process optimization instruction set to reduce defects.

[0044] When the system detects a non-critical workpiece, it automatically selects the cost-priority mode and limits w_2 to w_1; the system will call the preset process optimization instruction set to ensure cost.

[0045] Preferably, the multimodal process knowledge base 13 includes:

[0046] Material property storage unit 131 is used to store material property data of the device being coated. The material property data includes at least: substrate type and its physical property parameters; coating properties;

[0047] The environmental data storage unit 132 is used to store environmental data, which includes at least: temperature, humidity parameters, and wind speed parameters.

[0048] The process parameter storage unit 133 is used to store process parameters, which include at least: nozzle pressure and spraying speed parameters.

[0049] Image data storage unit 134 is used to store optical inspection images of defective areas, including at least micrographs of orange peel and dripping.

[0050] The text recording storage unit 135 is used to store operation logs, maintenance reports, and quality inspection notes during the production process.

[0051] The defect case storage unit 136 is used to store the operating condition feature data package of historical defect cases. The operating condition feature data includes process parameter combinations and defect information. The defect information includes at least the defect type, defect severity, and defect repair measures.

[0052] Defect mode association storage unit 137 is used to store the association between material properties, environmental data, and defect modes, including:

[0053] a) The mapping relationship between material type and environmental parameters on defect type;

[0054] b) A quantitative correlation model between environmental parameters and defect morphology;

[0055] c) A correlation model between substrate surface properties and the spatial distribution of defects.

[0056] Preferably, the multimodal process knowledge base 13 is updated in the following ways:

[0057] When the prediction bias of a new defect case is greater than 15%, incremental learning is triggered to update the association graph weights.

[0058] Apply a time decay factor to historical data exceeding 90 days;

[0059] The relationship between the porosity distribution of the storage coating and process parameters.

[0060] The second aspect of this application discloses an intelligent image generation method, characterized in that it is applied to the spraying imaging system described in the first aspect, and includes the following steps:

[0061] S1. Real-time data acquisition: Simultaneously acquires data on the texture features of the sprayed surface, the coating thickness distribution, and the ambient temperature and humidity through multiple sensors;

[0062] S2. Dynamic parameter coordinated adjustment: The robotic arm moving speed v is dynamically calculated based on the surface roughness Ra, the nozzle pressure P is optimized based on the adhesion level F, and the gain coefficient G of the imaging device is adjusted in combination with the temperature and humidity compensation formula.

[0063] S3. Defect Simulation and Prediction: Input the set of process parameters into the physical constraint GAN model to generate a coating defect morphology prediction map and identify the risk area of ​​sagging defects.

[0064] S4. Process optimization decision: Select weight coefficients based on the quality / cost model, and generate a collaborative optimization instruction set for paint viscosity, spraying speed, and gun distance through reinforcement learning iteratively;

[0065] S5. Closed-loop execution and feedback: Optimization instructions are sent to the spraying robot arm and imaging equipment, the reward function is updated based on the newly collected coating quality data, and incremental learning is triggered to optimize the knowledge base association model.

[0066] Furthermore, the method of this application also includes: S6. Anomaly adaptive processing: when the wind speed is detected to be >2m / s or the thickness non-uniformity is detected to be >1%, the spraying is paused and the local respraying path planning is initiated, and the exposure parameters of the imaging equipment are corrected simultaneously.

[0067] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0068] 1. This case achieves dynamic optimization of spraying imaging parameters and real-time defect prediction by using a dynamic parameter adjustment module to link equipment parameters in real time, a defect prediction module to generate process optimization instruction sets, and a multimodal knowledge base to establish a cross-modal correlation model. This has the advantages of improving detection efficiency and accuracy.

[0069] 2. This application establishes a mathematical model for multi-source data fusion, achieving cross-modal correlation between surface condition, coating performance, and environmental parameters, thus resolving the issues of parameter adjustment lag and single-dimensional optimization. Furthermore, the robotic arm speed is dynamically adjusted based on surface roughness, avoiding defects caused by paint buildup or insufficient coverage; nozzle pressure is optimized based on adhesion level, improving coating bonding strength and durability; and the imaging gain coefficient is compensated for by temperature and humidity, ensuring detection accuracy under different environments. This achieves real-time collaborative optimization of spraying and imaging parameters, effectively solving the defect omission problem caused by parameter rigidity in traditional systems. Simultaneously, through multi-dimensional parameter linkage optimization, it improves spraying quality and detection efficiency.

[0070] 3. This case establishes a dynamic correlation between surface features, environmental parameters, and equipment control parameters by setting multi-dimensional threshold conditions, enabling adaptive parameter adjustment and improving system response speed. Through intelligent matching of spray trajectory and surface roughness, this case avoids the problems of paint accumulation on rough surfaces or insufficient coverage on smooth surfaces; it establishes a dynamic balance mechanism between adhesion level and nozzle pressure, ensuring coating bonding strength while preventing embrittlement defects; and it automatically corrects imaging parameters through temperature and humidity compensation formulas, ensuring the stability of the detection system under different environments.

[0071] 4. This project addresses the issue of unforeseen environmental interference and the inability to automatically correct coating defects during the spraying process through a specific anomaly handling mechanism. Real-time scanning and dynamic path planning enable localized repairs without system downtime, effectively reducing paint waste and improving spraying quality consistency. Furthermore, the grid-based spraying mode allows for precise adjustments to thickness distribution differences, avoiding overspraying or underspraying caused by traditional uniform spraying. Attached Figure Description

[0072] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0073] Figure 1 is a schematic diagram of the overall system architecture of Embodiment 1 of the present invention.

[0074] Figure 2 is a schematic diagram of the operation process of Embodiment 1 of the present invention. Detailed Implementation

[0075] 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.

[0076] It should be noted that the terms "first," "second," "third," "fourth," etc., used in the specification and claims of this invention are used to distinguish different objects, not to describe a specific order. The terms "comprising" and "having," and any variations thereof, in the embodiments of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or devices.

[0077] Example 1

[0078] As shown in Figures 1 and 2, this application proposes an intelligent imaging generation system and method, including a dynamic parameter adjustment module 11, which acquires the state of the sprayed surface in real time through multi-sensor fusion and adjusts the operating parameters of the imaging device and the spraying robot arm in linkage; a defect prediction and optimization decision module 12, which simulates the coating defect morphology under different process parameters based on a generative adversarial network model with physical constraints, and generates a process optimization instruction set through reinforcement learning; and a multimodal process knowledge base 13, which stores the correlation between the material property data, environmental sensing data, historical process parameter data and defect patterns of the sprayed device.

[0079] The dynamic parameter adjustment module refers to a system component that integrates vision, thickness, and environmental sensors to achieve data acquisition and fusion. Specifically, it can be implemented using a combination of a high-resolution industrial camera, a capacitive thickness gauge, and a temperature and humidity sensor to eliminate data bias from a single sensor. The defect prediction and optimization decision module is a defect simulation engine built on a generative adversarial network (GAN). Specifically, it can employ an adversarial neural network structure with physical constraints, outputting a defect prediction map based on a set of input process parameters. The multimodal process knowledge base is a database system that stores the correlation between material properties, environmental parameters, and defects. Specifically, it can adopt a hybrid architecture of distributed time-series database and graph database to achieve efficient retrieval and correlation analysis of cross-modal data.

[0080] Specifically, during the spraying process, a vision sensor captures surface texture features in real time, a thickness gauge monitors the coating distribution simultaneously, and environmental sensors collect temperature and humidity data. This data, after spatiotemporal alignment, is input into a parameter adjustment model to drive the coordinated optimization of the robotic arm speed and nozzle pressure. Simultaneously, a generative adversarial network simulates potential defect morphologies based on current process conditions, and a reinforcement learning algorithm dynamically adjusts the paint viscosity and spray gun distance based on quality indicators. A multimodal knowledge base continuously accumulates correlation data between materials, environment, and defects, providing historical case studies to support process optimization.

[0081] Compared to existing technologies, traditional systems using fixed imaging parameters result in an imbalance between detection sensitivity and coating quality. This solution achieves dynamic parameter matching through multi-source data fusion. Existing technologies rely on manual experience to adjust process parameters; this solution utilizes adversarial networks with physical constraints to achieve pre-defect prediction. Existing systems process process data and inspection data in isolation; this solution constructs a cross-domain data association model through a multimodal knowledge base.

[0082] Through the above technical solutions, this application achieves real-time collaborative optimization of spraying parameters and imaging parameters, significantly reducing coating rework rate. The defect prediction model identifies sagging risk areas in advance, guiding timely correction of process parameters. Multimodal data correlation analysis effectively improves the consistency of spraying quality and reduces material waste and equipment idle time.

[0083] This application further proposes a dynamic parameter adjustment module comprising a multi-data acquisition unit and a data fusion unit. The multi-data acquisition unit includes a high-resolution visual sensor, a capacitive thickness gauge, and a temperature and humidity sensor. The high-resolution visual sensor captures surface texture features and generates a first data stream; the capacitive thickness gauge monitors the coating thickness distribution in real time and generates a second data stream; the temperature and humidity sensor acquires environmental parameters and generates a third data stream. The data fusion unit employs a Kalman filter algorithm to perform spatiotemporal alignment and noise suppression on the first, second, and third data streams.

[0084] The core of this application lies in constructing a technical architecture for dynamic parameter adjustment, defect prediction, and knowledge base closed-loop optimization, which specifically includes the following modules:

[0085] I. Dynamic Parameter Adjustment Module 11

[0086] Multi-data acquisition unit 111: integrates a high-resolution vision sensor 1111, a capacitive thickness gauge 1112, and a temperature and humidity sensor 1113 to collect surface roughness Ra, coating thickness μm, and ambient temperature and humidity T / RH in real time; so as to obtain multi-dimensional information of the sprayed surface through the synchronous acquisition of different physical quantities.

[0087] Data fusion unit 112: Uses Kalman filtering algorithm to align multi-source data;

[0088] Among them, high-resolution vision sensors refer to image acquisition devices with micron-level resolution, specifically industrial linear or area scan cameras, used to capture the microscopic texture features of the sprayed surface. Capacitive thickness gauges are instruments that measure the thickness of non-conductive coatings through changes in electric field, specifically multi-plate array sensors, used to acquire real-time spatial distribution data of coating thickness. Temperature and humidity sensors are devices that measure ambient temperature and relative humidity, specifically digital temperature and humidity composite sensors, used to collect environmental parameters of the spraying area. The data fusion unit is a module that integrates and processes multi-source data, specifically implemented using an embedded processor equipped with a Kalman filter algorithm, improving data reliability through spatiotemporal coordinate registration and noise suppression.

[0089] In practical implementation, a high-resolution vision sensor captures images of the workpiece surface at a fixed frame rate, extracts surface roughness features using an edge detection algorithm, and generates a first data stream containing positional information. A capacitive thickness gauge continuously scans the coating thickness during the robotic arm's movement, converting capacitance changes into thickness distribution data, generating a second data stream. A temperature and humidity sensor collects environmental temperature and humidity parameters of the spraying area at a preset sampling frequency, generating a third data stream. After receiving the three data streams, the data fusion unit uses a Kalman filter algorithm to eliminate timestamp differences between the sensors and suppress measurement noise, ultimately outputting a spatiotemporally synchronized multidimensional dataset. This dataset is transmitted to the subsequent decision-making unit for real-time adjustment of the robotic arm's operating parameters and the imaging device's detection parameters. Thus, this solution, through the collaborative work of multiple sensors, synchronously collects three key data types: surface texture, coating thickness, and environmental parameters. Combined with the Kalman filter algorithm to eliminate time delays and measurement errors from different sensors, it provides more complete and accurate input data for subsequent parameter adjustments.

[0090] Intelligent Sensing and Decision-Making Unit 113: Dynamically adjusts the robotic arm speed v, nozzle pressure P, and imaging gain G using preset formulas. This unit includes a robotic arm movement speed adjustment unit 1131, a nozzle pressure optimization adjustment unit 1132, and an imaging device gain coefficient adjustment unit 1133. The robotic arm movement speed adjustment unit extracts the surface roughness feature value Ra of the sprayed surface in real time based on the first data stream, and dynamically adjusts the robotic arm movement speed v according to the preset robotic arm movement speed formula; the robotic arm movement speed formula is v = k1Ra + b1, where k1 and b1 are the roughness-speed response coefficients obtained through experimental calibration. The nozzle pressure optimization adjustment unit combines the coating thickness distribution data monitored by the second data stream and the adhesion level F obtained by the mechanical adhesion tester, and optimizes the nozzle pressure P using the preset nozzle pressure optimization formula; the nozzle pressure optimization formula is P = k2F + b2, where k2 and b2 are the pressure compensation coefficients determined through experimental calibration. The imaging device gain coefficient adjustment unit combines the third data stream and uses the imaging device gain coefficient compensation formula to compensate the imaging device gain coefficient G. The imaging device gain coefficient compensation formula is G = G0(1 + αRH% + β*T℃), where G0 is the reference gain coefficient, RH is the ambient relative humidity, T is the ambient temperature, and α and β are temperature and humidity compensation coefficients. The surface roughness characteristic value Ra refers to the arithmetic mean deviation of the micro-profile of the sprayed surface, which can be measured by three-dimensional laser scanning or white light interferometer, and is used to quantitatively characterize the flatness of the workpiece surface. The roughness-speed response coefficients k1 and b1 are linear relationship parameters calibrated through spraying experiments, for example, obtained by spraying samples with different roughness and measuring the optimal moving speed for fitting, used to establish a quantitative mapping between surface state and robotic arm movement. The adhesion grade F refers to the grading index of the bonding strength between the coating and the substrate, which can be achieved by cross-cut test or pull-out test, and is used to reflect the mechanical bonding performance of the coating. Pressure compensation coefficients k2 and b2 are parameters determined through adhesion testing and nozzle pressure optimization experiments. For example, they can be calibrated by adjusting the pressure and measuring the coating quality at different adhesion levels to dynamically match the spraying pressure with the coating bonding requirements. Temperature and humidity compensation coefficients α and β are quantitative parameters based on the influence of ambient temperature and humidity on the gain of imaging equipment. For example, they can be calibrated by simulating different environmental conditions in a temperature and humidity chamber and testing the imaging quality to eliminate the interference of environmental factors on imaging detection.

[0091] In practice, the robotic arm's movement speed adjustment unit acquires the texture features of the sprayed surface in real time through a high-resolution vision sensor, extracts the Ra value through image processing algorithms, and automatically adjusts the movement speed according to a preset linear formula. When the surface roughness increases, the system increases the movement speed to reduce localized paint accumulation; when the surface is smooth, the speed is reduced to ensure uniform coating coverage. The nozzle pressure optimization adjustment unit dynamically calculates the optimal pressure value based on the coating distribution data obtained from the capacitive thickness gauge and the adhesion test results. When the adhesion level is low, the system increases the pressure to enhance paint penetration; when the adhesion is too high, the pressure is reduced to prevent coating cracking. The imaging equipment gain coefficient adjustment unit acquires environmental parameters through temperature and humidity sensors and uses a compensation formula to correct the imaging equipment's gain setting in real time. For example, in a high-temperature and high-humidity environment, the system automatically increases the gain coefficient to compensate for changes in light refractive index, ensuring image clarity.

[0092] As mentioned above, traditional spraying parameter adjustments typically rely on fixed empirical values ​​or single sensor feedback, making multi-parameter collaborative optimization impossible. Therefore, this solution establishes a mathematical model for multi-source data fusion, achieving cross-modal correlation between surface state, coating performance, and environmental parameters, thus resolving the issues of parameter adjustment lag and single-dimensional optimization. Furthermore, the robotic arm speed is dynamically adjusted based on surface roughness, avoiding defects caused by paint buildup or insufficient coverage; nozzle pressure is optimized based on adhesion level, improving coating bonding strength and durability; and the imaging gain coefficient is compensated for by temperature and humidity, ensuring detection accuracy under different environments. This achieves real-time collaborative optimization of spraying and imaging parameters, effectively solving the defect omission problem caused by parameter rigidity in traditional systems. Simultaneously, through multi-dimensional parameter linkage optimization, it improves spraying quality and detection efficiency.

[0093] As a preferred embodiment, this application further proposes that the dynamic adjustment process of each adjustment unit of the dynamic parameter adjustment module satisfies the following:

[0094] When the surface roughness Ra > 5μm, the workpiece surface is judged to be relatively rough, and the system automatically switches to a spiral spraying trajectory. When the surface roughness Ra = 1μm, the workpiece surface is judged to be relatively smooth, and the system automatically switches to a normal high-speed spraying trajectory. When the adhesion level F = 1, the adhesion is judged to be too low, and the nozzle pressure is automatically increased to enhance the adhesion between the coating and the substrate. When the adhesion level F = 5, the adhesion is judged to be too high, and the nozzle pressure is automatically reduced to avoid coating embrittlement. When the temperature T = 35℃ or the humidity RH = 75%, the working environment is judged to be too hot or too humid, and the gain is dynamically adjusted through the gain coefficient compensation formula of the imaging equipment to correct environmental interference. The spiral spraying trajectory refers to a spraying method that covers the workpiece surface in a spiral path, which can be implemented using a robotic arm motion control system to form a uniform coating on rough surfaces. Nozzle pressure adjustment refers to dynamically adjusting the spraying pressure based on the adhesion level, which can be implemented using a proportional valve or a servo motor-driven pressure pump to maintain the coating adhesion within a reasonable range. Gain compensation formula refers to a mathematical model that dynamically corrects the gain of the imaging device based on temperature and humidity parameters. Specifically, it can be implemented using a linear regression model to eliminate the impact of environmental interference on image quality. Specifically, when the surface roughness Ra exceeds a threshold, the system automatically switches the spraying trajectory according to preset conditions. For example, a spiral trajectory on rough surfaces improves the uniformity of paint coverage, while a high-speed trajectory on smooth surfaces increases work efficiency. When the adhesion level is below the preset lower limit, the nozzle pressure is increased to enhance paint penetration; when the adhesion level exceeds the upper limit, the pressure is reduced to prevent coating cracking. When temperature and humidity exceed the safe range, the imaging parameters are adjusted using the gain coefficient compensation formula to reduce the interference of environmental factors on detection accuracy. This adjustment process triggers corresponding control commands by comparing real-time sensor data with preset thresholds, forming a closed-loop feedback mechanism.

[0095] As mentioned above, traditional systems use fixed spraying trajectories and pressure parameters, which cannot adapt to different surface conditions and environmental changes, leading to fluctuations in coating quality. This invention establishes a dynamic correlation between surface features, environmental parameters, and equipment control parameters by setting multi-dimensional threshold conditions, achieving adaptive parameter adjustment and improving system response speed. This invention avoids the problems of paint accumulation on rough surfaces or insufficient coverage on smooth surfaces by intelligently matching the spraying trajectory with surface roughness; it establishes a dynamic balance mechanism between adhesion level and nozzle pressure, ensuring both coating bonding strength and preventing embrittlement defects; and it automatically corrects imaging parameters through temperature and humidity compensation formulas, ensuring the stability of the detection system under different environments.

[0096] As a preferred embodiment, this application further proposes that the dynamic parameter adjustment module also includes an anomaly handling mechanism: when the instantaneous wind speed is detected to be greater than a preset threshold, such as wind speed > 2 m / s, spraying is paused and a three-dimensional laser scan is initiated to reconstruct the surface morphology; a local respraying path is generated based on point cloud data; when the coating thickness non-uniformity exceeds a set threshold, such as > 1%, an adaptive mesh spraying mode is triggered. The adaptive mesh spraying mode refers to dividing the spraying area into multiple sub-regions and adjusting parameters independently. For example, dynamic mesh division technology can be used to automatically adjust the spraying speed and spray gun angle within each mesh according to the thickness distribution differences. Instantaneous wind speed detection refers to real-time monitoring of airflow disturbance in the spraying area through a wind pressure sensor. For example, a digital anemometer can be used to collect data at a sampling frequency of at least twice per second. This detection provides a trigger condition for anomaly handling, avoiding airflow interference that could cause coating scattering. Three-dimensional laser scan reconstruction of the surface morphology refers to acquiring three-dimensional coordinate data of the workpiece surface through a line laser scanner. For example, a blue laser combined with a high-speed CMOS sensor can be used to generate a millimeter-precision point cloud model for accurately identifying areas requiring respraying. Localized touch-up spraying paths refer to spraying trajectories generated based on surface topography features. For example, path planning algorithms can generate spiral or sawtooth trajectories in the touch-up spraying area to ensure coating integrity. In specific implementation, when a wind speed greater than 2 m / s is detected, the system immediately suspends spraying to prevent paint scattering. Subsequently, a 3D laser scanning device is activated, for example, using a rotating line laser scanning head, to reconstruct the current surface topography at a resolution of 0.1 mm. Based on the point cloud data obtained from the scan, for example, point cloud density analysis algorithms are used to identify coating defect areas, generate localized touch-up spraying paths, and control a robotic arm to perform precise touch-up spraying. When a coating thickness difference exceeding 1% is detected, the system divides the spraying area into multiple grid units, for example, using an octree spatial segmentation algorithm to dynamically generate the grid, and independently adjusts the spraying parameters based on the measured thickness value within each grid. For example, the paint flow rate is increased in areas with insufficient thickness, and the spray gun movement speed is reduced in areas with excessive thickness.

[0097] As mentioned above, traditional spraying systems can only issue alarms or shut down for repair when encountering abnormal wind speeds or uneven thickness. This solution, however, achieves autonomous repair through 3D scanning reconstruction and dynamic path planning, avoiding production line shutdowns caused by manual intervention. Furthermore, this solution generates adaptive paths based on real-time point cloud data, accurately covering defective areas. In addition, the design of a specific anomaly handling mechanism solves the problem of untimely correction of sudden environmental interference and coating defects during spraying. Through real-time scanning and dynamic path planning, local repairs are completed without shutting down the system, effectively reducing paint waste and improving spraying quality consistency. Simultaneously, the grid-based spraying mode can finely adjust for differences in thickness distribution, avoiding over- or under-spraying of material caused by traditional uniform spraying.

[0098] As a preferred embodiment, this application further proposes a defect prediction and optimization decision module 12, including a defect simulation unit 121, a reinforcement learning optimization unit 122, and a closed-loop control unit 123. The defect simulation unit 121 generates coating defect prediction maps under different process parameters based on a physically constrained adversarial network model, where the adversarial network model is a GAN model. The generator input of the GAN model includes a set of process parameters (e.g., spraying speed v, nozzle pressure P), material property data (e.g., surface energy γ = 35 mN / m, coating density ρ, coating viscosity η, coating surface tension σ, coating thickness h), environmental data (e.g., ambient temperature T, humidity RH), and substrate state parameters (e.g., roughness Ra). The discriminator outputs coating defect prediction maps under different process parameters.

[0099] The discriminator of the GAN model integrates physical equation constraints for coating defects, including:

[0100] Length of sagging defects Where ρ is the density of the coating and g is the acceleration due to gravity, which is a constant, approximately 9.81 m / s². 2Here, h represents the coating thickness, η represents the coating viscosity, v represents the spraying speed, and C is an empirical coefficient used to adjust the accuracy of the formula; σ represents the surface tension of the coating, indicating the attractive force between molecules on the coating surface. In practice, the discriminator is responsible for judging the difference between the predicted map output by the generator and the actual defect map. Through continuous adversarial training, the accuracy of the generator's predicted map is improved. During the training of the GAN model, physical constraints (i.e., the aforementioned physical equations for coating defects) are introduced to ensure that the generated predicted coating defect maps conform to the physical laws of coating defect formation. Then, the generator and discriminator continuously optimize through adversarial training. For example, the generator strives to generate more realistic predicted maps to deceive the discriminator, while the discriminator strives to improve its judgment accuracy to distinguish between real and generated maps. After adversarial training, the generator of the GAN model can generate accurate predicted coating defect maps based on the input multi-source data. These maps intuitively demonstrate the types of defects that may occur in the coating under different process parameters and their distribution, providing an important basis for optimizing the spraying process. For example, when the system detects a decrease in paint viscosity, the model can predict potential sagging defects and provide the possible length of the sagging. These predictions can guide the system to adjust process parameters in a timely manner, such as reducing the spraying speed or increasing the paint viscosity, thereby preventing defects from occurring. Sagging defects are formed when paint flows downwards due to gravity during the coating drying process, and their formation mechanism can be explained by fluid dynamics equations. The sagging defect length formula describes the quantitative relationship between sagging defect length and paint viscosity (η), spraying speed (v), paint density (ρ), gravitational acceleration (g), and coating thickness (h). For example, high-viscosity paint (η↑) or high spraying speed (v↑) increases intermolecular cohesion, inhibiting sagging (L↓); while thin coatings (h↓) or low-density paints (ρ↓) are more prone to sagging due to increased gravity. By embedding this physical equation into the GAN model discriminator, the generator output can be constrained to conform to the laws of fluid dynamics, ensuring that the prediction results match the actual physical process.

[0101] The reinforcement learning optimization unit (122) uses real-time spraying quality indicators as the state space and process parameter adjustment amounts as the action space, optimizing process parameters through a reward function. In specific implementation, the state space is the set of all environmental state parameters that the system needs to observe during the decision-making process, used to characterize the real-time quality state of the spraying process. The state space includes the following real-time spraying quality indicators: thickness uniformity, adhesion level, and surface gloss. The action space consists of process parameter adjustment amounts (speed change Δv, pressure change ΔP); where the speed change Δv is the adjustment range of the spraying robot arm's moving speed (e.g., ±0.2 m / s); and the pressure change ΔP is the adjustment range of the nozzle pressure (e.g., ±5 kPa). The reward function is defined as R = w1·(1-(∣D_actual-D_target∣ / D_target))+w2·C_saved / C_total, where D_actual is the actual defect rate, D_target is the target defect rate, C_saved is the actual saved paint cost, C_total is the total paint cost budget, and w1 and w2 are weighting coefficients, satisfying w1+w2=1. The state space of the reinforcement learning optimization unit refers to quantifying the spraying quality indicators into a multi-dimensional vector, including the coating thickness standard deviation, adhesion grade, and surface roughness. Specifically, normalized sensor data can be stitched together to form the state vector, thereby achieving a comprehensive evaluation of complex quality indicators. The action space refers to the set of discretized operation instructions for adjusting process parameters, specifically using the adjustment step size of spraying speed, nozzle pressure, and gun distance parameters as the action dimension. The weighting coefficient in the reward function refers to the balance factor between quality priority and cost priority. Specifically, the ratio of w1 and w2 can be dynamically adjusted. When the system detects a critical workpiece, the value of w1 is automatically increased, thereby achieving multi-objective optimization decision-making.

[0102] The closed-loop control unit 123 sends the optimized process parameters to the spraying robot arm and imaging device in real time, and receives new sensor data for iterative optimization. Specifically, the defect simulation unit receives a set of process parameters and environmental data from multiple sensors, and inputs them along with material property data into the generator of the GAN model. Under physical constraints, the generator outputs a microscopic defect distribution map of the coating surface, and the discriminator identifies the difference between the predicted results and the actual defects through adversarial training. The reinforcement learning optimization unit constructs a state vector based on the current spraying quality indicators, selects process parameter adjustment instructions in the action space, executes them, obtains reward values, and updates the system. The closed-loop control unit sends the optimized spraying speed and nozzle pressure parameters to the robot arm controller, and simultaneously adjusts the gain coefficient of the imaging device to adapt to environmental changes. After a preset period, the system collects new coating thickness and surface morphology data, recalculates the reward function value, and triggers model parameter updates, forming an iterative optimization loop.

[0103] As mentioned above, traditional spraying systems can only adjust process parameters according to fixed rules, failing to simulate the nonlinear relationship between complex defect morphologies and process parameters. Existing defect detection methods rely on offline analysis and lack real-time feedback mechanisms, leading to lag in parameter adjustment. This project achieves accurate prediction of defect morphologies through a physically constrained GAN model, optimizes multi-parameter collaborative adjustment strategies by combining the dynamic exploration capabilities of reinforcement learning, and utilizes closed-loop control to achieve real-time linkage between process parameters and imaging parameters, effectively solving the problems of parameter solidification and feedback delay. Thus, the defect prediction and optimization decision module 12 in this project can accurately predict the distribution of coating defects under different process conditions through the settings of its various units, dynamically generate optimization instructions to balance quality and cost objectives, and achieve adaptive control of the spraying process. Moreover, by continuously correcting the prediction model through real-time data closed-loop feedback, the collaborative efficiency of defect identification and process adjustment is improved, reducing the need for manual intervention.

[0104] As a preferred embodiment, this application further proposes a mode selection module 14 for dynamically selecting a quality-priority mode and a cost-priority mode based on the workpiece production type.

[0105] When the system detects a critical workpiece in production, it automatically selects the quality-first mode and limits w_1 to w_2; the system will call the preset process optimization instruction set to reduce defects.

[0106] When the system detects a non-critical workpiece, it automatically selects the cost-first mode and limits w_2>

[0107] w_1; The system will invoke a preset process optimization instruction set to ensure cost control. The quality-first mode prioritizes coating quality by adjusting weight coefficients, which can be achieved by setting preset weight thresholds, such as a quality weight coefficient of 0.7 and a cost weight coefficient of 0.3. This can be implemented using reward function weight adjustment in the reinforcement learning unit to prioritize different optimization objectives within the objective function. The process optimization instruction set is a set of collaborative control instructions including coating viscosity adjustment values, spraying speed thresholds, and spray gun distance parameters, which can be achieved by using a generative adversarial network model with physical constraints to output multi-dimensional parameter combinations. In practice, the mode selection module uses the workpiece type identification unit to determine the attribute labels of the currently produced workpiece. When a critical workpiece label is identified, the quality-first mode is activated, forcibly setting the quality weight coefficient higher than the cost weight coefficient. At this time, when the reinforcement learning optimization unit generates the coating viscosity adjustment value, it prioritizes reducing the defect rate, and the spraying speed threshold is set to the lower limit within the allowable range to extend the effective spraying time. When a non-critical workpiece label is identified, the cost priority mode is activated, and the cost weight coefficient is forced to be higher than the quality weight coefficient. At this time, the nozzle pressure parameter will be calculated based on the paint consumption optimization formula, and the spraying speed threshold will be set to the upper limit within the allowable range to shorten the production cycle.

[0108] As described above, this case achieves dynamic weight allocation through a mode selection mechanism. In critical workpiece scenarios, the quality weight coefficient is increased to enhance defect suppression capabilities; in non-critical workpiece scenarios, the cost weight coefficient is increased to achieve resource conservation. This solves the imbalance between detection accuracy and cost control caused by parameter rigidity, enabling automatic matching of the optimal optimization strategy based on workpiece attributes. This ensures the quality of critical workpieces while reducing the production costs of non-critical workpieces. When producing high-value workpieces, a quality-first mode reduces quality risks caused by coating defects; when producing ordinary workpieces, a cost-first mode reduces paint consumption and production time, achieving dynamic balance control between quality and cost.

[0109] As a preferred embodiment, the multimodal process knowledge base 13 includes:

[0110] The material property storage unit 131 is used to store the material property data of the device to be coated. The material property data includes at least: the substrate type (metal / plastic / composite material) and its physical property parameters (density, thermal conductivity, surface energy); the coating properties (viscosity, solid content, surface tension coefficient, activation energy of curing reaction); the substrate physical property parameters are obtained by material testing equipment (such as surface tension meter, thermal conductivity meter), and the coating properties are determined by rheometer, gas chromatograph.

[0111] The environmental data storage unit 132 is used to store environmental data, which includes at least: temperature, humidity parameters, and wind speed parameters.

[0112] The process parameter storage unit 133 is used to store process parameters, which include at least: nozzle pressure and spraying speed parameters.

[0113] Image data storage unit 134 is used to store optical inspection images of defective areas, including at least micrographs of orange peel and dripping.

[0114] The text recording storage unit 135 is used to store operation logs, maintenance reports, and quality inspection notes during the production process.

[0115] The defect case storage unit 136 is used to store the operating condition feature data package of historical defect cases. The operating condition feature data includes process parameter combinations and defect information. The defect information includes at least the defect type, defect severity, and defect repair measures.

[0116] Defect mode association storage unit 137 is used to store the association between material properties, environmental data, and defect modes, including:

[0117] a) The mapping relationship between material type and environmental parameters on defect type;

[0118] b) A quantitative correlation model between environmental parameters and defect morphology;

[0119] c) A correlation model between substrate surface properties and the spatial distribution of defects.

[0120] As described above, this invention integrates key material characteristic data such as substrate type and its physical properties, and coating properties through the material characteristic storage unit 131, providing comprehensive basic material information for optimizing the spraying process. This data is acquired through professional testing equipment, ensuring its accuracy and reliability. The environmental data storage unit 132 records environmental parameters during the spraying process, such as temperature, humidity, and wind speed. These parameters significantly impact spraying quality and help analyze the effects of environmental factors on the spraying effect. The process parameter storage unit 133 stores key process parameters during the spraying process, such as nozzle pressure and spraying speed. These parameters are direct targets for spraying process optimization. The image data storage unit 134 stores optical inspection images of defective areas, such as micrographs of orange peel and runs. These images visually demonstrate defects occurring during the spraying process, helping technicians quickly identify and analyze problems. The text record storage unit 135 records operation logs, maintenance reports, and quality inspection notes during the production process. This text information provides detailed background information for defect analysis and resolution. The defect case storage unit 136 stores data packages of historical defect cases, including process parameter combinations and defect information. These case studies provide valuable experience for technicians, helping them quickly locate and resolve similar problems. The defect pattern association storage unit 137 stores the correlation between material properties, environmental data, and defect patterns, including the mapping relationship between material type and environmental parameters to defect type, a quantitative correlation model between environmental parameters and defect morphology, and a correlation model between substrate surface properties and defect spatial distribution. These correlations help to deeply understand the formation mechanism of defects during the spraying process, providing a scientific basis for process optimization. This achieves comprehensive data integration, enabling the system to automatically adjust spraying process parameters based on real-time data and historical experience, achieving continuous optimization of spraying quality.

[0121] As a preferred implementation, this application further proposes that the multimodal process knowledge base be updated in the following ways: when the prediction deviation of a new defect case is greater than 15%, incremental learning is triggered to update the association graph weights; a time decay factor is applied to historical data exceeding 90 days; and the relationship between coating porosity distribution and process parameters is stored.

[0122] Incremental learning refers to updating the weight parameters of the correlation graph through incremental model training. Specifically, this can be achieved by using online learning algorithms combined with historical data to fine-tune model parameters, dynamically adapting to changes in new defect patterns and addressing the prediction bias of older models for new operating conditions. The time decay factor is a coefficient that exponentially decays the importance of historical data. Specifically, it can be calculated using the formula lambda = 0.98^{t / 30} (where t is the number of days), used to reduce the interference of outdated data on the current model and maintain the timeliness of the knowledge base. The correlation between coating porosity distribution and process parameters refers to the mapping relationship between microstructure data obtained through high-resolution μCT scanning and spraying parameters. Specifically, 3D image processing technology can be used to extract porosity distribution feature values ​​and establish a regression model to optimize spraying parameters and reduce porosity defects. In practice, when the prediction bias of a new defect case exceeds a threshold, the system automatically initiates the incremental learning process, for example, using an online gradient descent algorithm to adjust the correlation weights between material properties and defect types, and synchronously updating the quantified correlation model in the knowledge base. For historical data stored for more than three months, its contribution to model training is reduced by a time decay factor. For example, the weight of coating thickness data from 90 days ago is reduced to less than 50% of the initial value, thus prioritizing the learning of recent operating conditions. In addition, by storing the correspondence between high-resolution μCT scan data and spraying parameters, such as establishing a correlation between areas with porosity exceeding 5% and process parameters with spray gun movement speeds below 0.8 m / s, a physical basis is provided for subsequent optimization of spraying parameters.

[0123] As mentioned above, traditional knowledge bases typically employ a static storage model, making it impossible to dynamically adjust the association model based on new defect data. Furthermore, the fixed weight allocation of historical data limits the model's generalization ability. This application achieves continuous evolution of the knowledge base through an incremental learning mechanism, eliminates interference from stale data by combining it with a time decay factor, and introduces microstructure data to enhance the physical interpretability of process parameter optimization, significantly improving the collaborative optimization capability of defect prediction and parameter adjustment. In addition, this application achieves dynamic optimization of the knowledge base and cross-scale data fusion, solving the problem of inaccurate process parameter matching caused by data update lag in traditional systems.

[0124] Example 2

[0125] This case proposes an intelligent imaging generation method, including the following steps: S1. Real-time data acquisition: synchronously acquire the texture features of the sprayed surface, the coating thickness distribution, and the ambient temperature and humidity data through multiple sensors;

[0126] S2. Dynamic parameter coordinated adjustment: The robotic arm moving speed v is dynamically calculated based on the surface roughness Ra, the nozzle pressure P is optimized based on the adhesion level F, and the gain coefficient G of the imaging device is adjusted in combination with the temperature and humidity compensation formula.

[0127] S3. Defect Simulation and Prediction: Input the set of process parameters into the physical constraint GAN model to generate a coating defect morphology prediction map and identify the risk area of ​​sagging defects.

[0128] S4. Process optimization decision: Select weight coefficients based on the quality / cost model, and generate a collaborative optimization instruction set for paint viscosity, spraying speed, and gun distance through reinforcement learning iteratively;

[0129] S5. Closed-loop execution and feedback: The optimization instructions are sent to the spraying robot arm and imaging equipment, the reward function is updated based on the newly collected coating quality data, and incremental learning is triggered to optimize the knowledge base association model;

[0130] S6. Anomaly Adaptive Handling: When a wind speed > 2 m / s or a thickness non-uniformity > 1% is detected, spraying is paused and local respraying path planning is initiated, simultaneously correcting the exposure parameters of the imaging equipment. Anomaly adaptive handling refers to triggering an adaptive control strategy by real-time monitoring of environmental parameters and coating status. Specifically, 3D laser scanning combined with point cloud data processing algorithms can be used to generate local respraying paths, such as calculating the optimal respraying trajectory based on a dynamic path planning model. The correction of the imaging equipment's exposure parameters can be achieved through environmental interference compensation algorithms, such as establishing the correlation between wind speed and exposure time using a multivariate regression model. Local respraying path planning includes two stages: surface morphology reconstruction and spraying parameter re-optimization. Specifically, a gridded spraying parameter adjustment module can be used to achieve differentiated spraying in different areas. Synchronous correction refers to the coordinated issuance of imaging equipment parameter adjustments and spraying robot arm control commands, specifically achieved through real-time communication between devices via an industrial bus protocol.

[0131] Specifically, the process begins by capturing surface texture images using a visual sensor, while a capacitive thickness gauge acquires coating distribution data. Temperature and humidity sensors collect environmental parameters and transmit them to the central processing unit. Surface roughness features are input into a preset speed formula to calculate the robotic arm's movement speed, and nozzle pressure parameters are adjusted based on real-time adhesion test results. The imaging device's gain coefficient automatically corrects for environmental interference using a compensation formula, such as automatically increasing the gain value to compensate for thermal noise in high-temperature environments. A set of process parameters is input into a generative adversarial network to predict defect morphology, and the prediction results trigger a reinforcement learning module to generate optimization instructions. The closed-loop control module synchronously sends the optimized spraying speed and coating viscosity parameters to the actuator, while simultaneously feeding newly acquired coating quality data back to the knowledge base to update the associated model. Furthermore, an emergency response process can be initiated when sudden environmental interference is detected; for example, spraying can be paused and a 3D scan can be started to reconstruct the surface morphology when wind speed exceeds the limit, generating a local respray path based on point cloud data. The imaging device's exposure parameters can be dynamically adjusted according to environmental changes; for example, the optical filtering mode can be automatically switched to eliminate water mist interference when humidity changes abruptly. Knowledge base updates can employ time decay mechanisms to process historical data, such as applying exponential decay weights to process data that has exceeded its storage period.

[0132] As described above, the method in this case overcomes the technical shortcomings of manual inspection due to its lag through real-time data acquisition and closed-loop feedback mechanisms. The established multimodal correlation model effectively integrates process parameters and material property data, improving the prediction accuracy of coating defects and the response speed of process adjustments. The introduction of an anomaly handling mechanism enhances the system's robustness under complex working conditions, ensuring the stability of quality control during continuous production. Through anomaly adaptive handling steps, a closed-loop control mechanism of anomaly response and parameter correction is established, effectively solving the problem of uncontrollable coating quality under abnormal spraying conditions. Local respraying path planning reduces material waste and improves repair efficiency. Dynamic correction of imaging equipment exposure parameters avoids image distortion caused by environmental interference, providing reliable data support for online quality inspection. The synergistic optimization of process parameters and imaging parameters enhances the system's adaptability to complex working conditions, achieving full-process quality control in the spraying process.

[0133] Specifically, Example 1 uses automotive body painting as an application. This system applies topcoat to car doors. The substrate is galvanized steel sheet (surface energy γ = 35 mN / m), and the coating is water-based polyurethane (coating viscosity η = 250 mPa·s, coating surface tension σ = 30 mN / m, coating density ρ = 1.2 g / cm³). 3 The ambient temperature is T = 28℃, the ambient humidity is RH = 65%, and the instantaneous wind speed is < 1m / s. The baseline environment is T = 25℃ and RH = 50%. The system needs to achieve real-time collaborative optimization of spraying parameters and imaging parameters to ensure consistent coating quality and reduce the defect rate.

[0134] The system operation process is as follows:

[0135] 1. System initialization and parameter configuration

[0136] Workpiece type identification

[0137] The system identifies the current workpiece as a "car door" by scanning with RFID or QR codes, and automatically loads the material property data of the galvanized steel substrate and the water-based polyurethane coating.

[0138] The mode selection module (14) determines the quality priority mode (because the door is a key component), and sets the reward function weights as follows: w1 = 0.7 (quality priority), w2 = 0.3 (cost secondary).

[0139] 2. Data Acquisition and Fusion:

[0140] A high-resolution visual sensor captures the surface roughness Ra = 2.5 μm of the galvanized steel sheet, generating the first data stream; a capacitive thickness gauge measures the average thickness h = 85 μm, generating the second data stream; a temperature and humidity sensor collects environmental data, T = 28 ± 0.5℃, RH = 65 ± 3%, generating the third data stream. The data fusion unit (112) uses a Kalman filter algorithm to perform timestamp alignment and noise suppression on the three data streams, outputting a spatiotemporally synchronized multidimensional dataset:

[0141] 3. Dynamic parameter adjustment:

[0142] Using the formula for the robotic arm's movement speed, v = k1 * Ra + b1, where k1 is 0.2 and b1 is 0.5, the robotic arm speed is calculated to be v = 0.2 * 2.5 + 0.5 = 1.0 m / s. A spiral spraying trajectory is then adopted (because Ra = 2.5 μm, which is between 1 and 5 μm).

[0143] The adhesion level F = 3 obtained by the mechanical adhesion tester is P = k2 * F + b2; k2 is 5, b2 is 20, P = 5 * 3 + 20 = 35 kPa.

[0144] The gain coefficient compensation formula for imaging equipment is: G=G0*(1+α*RH%+β*T℃); G0 is 1.0, α is 0.5, and β is 0.3, where G0 is the reference gain coefficient; RH is the ambient relative humidity, and T is the ambient temperature; α and β are the temperature and humidity compensation coefficients. The physical meaning of parameters α=0.5( / %) and β=0.3( / ℃) is: every 1% change in humidity results in a 0.5% change in gain, and every 1℃ change in temperature results in a 0.3% change in gain. G=G0*(1+0.5*65%*0.01+0.3*28℃*0.01)=1.409G0=1.409.

[0145] 4. Defect simulation prediction and process optimization

[0146] 4.1 Defect Simulation Unit (121)

[0147] The process parameters (v = 1.0 m / s, P = 35 kPa), material properties (surface roughness Ra = 2.5 μm, coating viscosity η = 250 mPa·s, coating surface tension σ = 30 mN / m, coating density ρ = 1.2 g / cm³) were used. 3 The coating thickness h = 85 μm and environmental data (T = 28 °C, RH = 65%) were input into the physical constraint GAN model.

[0148] The generator outputs a coating defect prediction map, and the discriminator identifies potential defects.

[0149] Predict the risk area of ​​sag defects (assuming C=1, in practical applications, the value of C needs to be determined through experiments); (Length of sagging defect L = 3.4 mm)

[0150] The GAN model predicts that the probability of overhang under the current parameters is 12%.

[0151] After adjustment, the actual defect rate dropped to 0.4%, resulting in a 18% saving in paint.

[0152] 4.2 Reinforcement Learning Optimization Unit (122)

[0153] State space configuration: thickness uniformity = 95%, adhesion level = 3.

[0154] Select optimization instructions in the action space:

[0155] Reduce the spraying speed to 0.9 m / s (Δv = -0.1 m / s)

[0156] Increase the coating viscosity to 280 mPa·s (by adding a thickener).

[0157] 4.3 Closed-loop control unit (123)

[0158] The optimization instructions are sent to the painting robot arm and imaging equipment, and the knowledge base and associated models are updated simultaneously.

[0159] 5. Adaptive Exception Handling

[0160] 5.1 Detection of abnormal wind speed

[0161] The real-time wind speed is 1.2 m / s (not exceeding the threshold of 2 m / s), and the system is operating normally.

[0162] 5.2 Thickness Non-uniformity Detection

[0163] Thickness non-uniformity = 4.0% (<1% threshold), no need to trigger mesh spraying mode.

[0164] 6. Closed-loop feedback and knowledge base updates

[0165] 6.1 New Data Acquisition and Reward Function Update

[0166] New coating quality: thickness uniformity = 96%, adhesion grade = 4, defect rate = 0.35%.

[0167] Reward function value:

[0168] R=0.7*(1-(∣0.35-0.3∣ / 0.3))+0.3*(5 / 10)=0.7*0.833+0.3*0.05=0.583+0.015=0.598

[0169] 6.2 Incremental Learning of Knowledge Base

[0170] Prediction bias = |0.35-0.3| / 0.3 = 16.7% > 15%, triggering incremental learning and updating the material-defect association weights.

[0171] 7. Spraying Completion and Quality Verification

[0172] 7.1 Evaluation of Spraying Effect

[0173] Final coating thickness: 85±3μm (improved uniformity)

[0174] Defect rate: 0.2% (below the target value of 0.5%)

[0175] 7.2 Data Archiving

[0176] Storage process parameters, environmental data, and defect cases are stored in a multimodal process knowledge base (13).

[0177] Parameter calibration process description:

[0178] I. The process of obtaining k1 and b1 in the robotic arm speed formula:

[0179] In the robotic arm speed formula, k1 and b1 are roughness-speed response coefficients obtained through experimental calibration. These coefficients were determined by selecting galvanized steel sheets with different surface roughness (Ra values) as the coating substrate in the experiment. Galvanized steel sheets are widely used in industries such as automotive and construction, and their surface characteristics directly affect the coating quality (such as coating adhesion and uniformity). Multiple samples with roughnesses ranging from 1 μm to 1 μm were prepared through machining or surface treatment to cover common situations in actual production. The optimal moving speed was then measured, and a linear relationship was fitted using the least squares method.

[0180] Table 1: Experimental Data for Calibrating the Parameters of the Robotic Arm Speed ​​Formula

[0181]

[0182] As shown in the table above, the optimal robotic arm speed was experimentally determined by spraying galvanized steel plates with different roughnesses. The least squares method was used to fit the linear formula, and the goodness of fit R0 was [value missing]. 2 =0.97, with the error controlled within ±3%. R 2 =0.97 indicates that the model can explain 97% of the data variation, demonstrating that the linear relationship between roughness and speed is highly reliable and can effectively guide parameter adjustments in actual production.

[0183] 2. Adjust the nozzle pressure under different adhesion grades (F = 1 to 5), and determine the optimal pressure value by testing the coating bonding strength using the cross-cut test.

[0184] Table 2: Experimental Data for Nozzle Pressure Formula Parameter Calibration

[0185] The error in the fitting formula for the optimal pressure P obtained from the adhesion grade experiment is 124.825 -0.2230.230 +0.2334.935 -0.1440.340 +0.3544.745 -0.3 surface

[0186] As shown in the table above, the maximum error between the measured pressure value and the theoretical value is ±0.3 kPa, which meets the industrial standard (±0.5 kPa). The coating adhesion grade under different pressures was tested using the cross-cut adhesion test to determine the formula parameters (k2 = 5, b2 = 20).

[0187] The above provides a detailed description of an intelligent imaging generation system and method disclosed in the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. An intelligent imaging generation system, characterized in that, include: The dynamic parameter adjustment module (11) acquires the sprayed surface state in real time through multi-sensor fusion and adjusts the operating parameters of the imaging device and the spraying robot arm in linkage; the defect prediction and optimization decision module (12) simulates the coating defect morphology under different process parameters based on the physical constraint generative adversarial network model, and generates a process optimization instruction set through reinforcement learning. The process optimization instruction set includes: the coating viscosity adjustment value, the spraying speed threshold, and the collaborative optimization instruction of the spray gun distance parameter; the multimodal process knowledge base (13) stores the material property data, environmental sensing data, historical data of process parameters and the correlation between defect modes of the sprayed device; wherein, the defect prediction and optimization decision module (12) includes: a defect simulation unit (121), which generates a coating defect prediction map under different process parameters based on the physical constraint adversarial network model; the adversarial network model is a GAN model; the generator input of the GAN model includes a set of process parameters, material property data, environmental data and substrate state parameters, so as to output a coating defect prediction map under different process parameters through its discriminator; the discriminator of the GAN model integrates the physical equation constraint of coating defects, including: the length of the drip. Where ρ is the coating density, g is the gravitational acceleration, h is the coating thickness, η is the coating viscosity, v is the spraying speed, C is an empirical coefficient, and σ is the surface tension of the coating; the reinforcement learning optimization unit (122) uses the real-time spraying quality index as the state space and the process parameter adjustment amount as the action space, and optimizes the process parameters through the reward function; the closed-loop control unit (123) sends the optimized process parameters to the spraying robot arm (301) and the imaging device (302) in real time, and receives new sensor data for iterative optimization; where the reward function is defined as: R=w1⋅(1-(∣D_{actual} -D_{target}∣ / D_{target}))+w2⋅C_{saved} / C_{total} ; D_{actual} is the actual defect rate, D_{target} is the target defect rate, C_{saved} is the actual saved coating cost, and C_{total} is the actual defect rate. This is the total paint cost budget, where w1 and w2 are weighting coefficients, satisfying w1+w2=1.

2. The system according to claim 1, characterized in that, The dynamic parameter adjustment module (11) includes: a multi-data acquisition unit (111), which includes: a high-resolution vision sensor (1111) for capturing surface texture features and generating a first data stream; a capacitive thickness detector (1112) for real-time monitoring of coating thickness distribution and generating a second data stream; a temperature and humidity sensor (1113) for acquiring environmental parameters and generating a third data stream; and a data fusion unit (112) for using a Kalman filter algorithm to perform spatiotemporal alignment and noise suppression on the first, second, and third data streams.

3. The system according to claim 2, characterized in that, The dynamic parameter adjustment module (11) further includes: an intelligent sensing and decision-making unit (113), which includes: a robotic arm moving speed adjustment unit (1131): used to extract the surface roughness feature value Ra of the sprayed surface in real time by combining the first data stream, and dynamically adjust the robotic arm moving speed v according to the preset robotic arm moving speed formula; wherein, the robotic arm moving speed formula is: v = k1*Ra + b1; k1 and b1 are roughness-velocity response coefficients obtained through experimental calibration; Ra ranges from 0.1 to 10 μm; Nozzle pressure optimization adjustment unit (1132): used to combine the coating thickness distribution data monitored by the second data stream and the adhesion level F obtained by the mechanical adhesion tester, and optimize the nozzle pressure P using the preset nozzle pressure optimization formula; where, the nozzle pressure optimization formula is: P=k2*F+b2; k2 and b2 are experimentally calibrated pressure compensation coefficients; the adhesion level F is divided into 1-5 levels; Imaging device gain coefficient adjustment unit (1133): used to combine the third data source and use the imaging device gain coefficient compensation formula to compensate the imaging device gain coefficient G; the imaging device gain coefficient compensation formula is: G=G0*(1+α*RH%+β*T℃); G0 is the reference gain coefficient; RH is the ambient relative humidity; T is the ambient temperature; α and β are temperature and humidity compensation coefficients, α is in units of 1 / %, and β is in units of 1 / ℃.

4. The system according to claim 3, characterized in that, The dynamic adjustment process of each adjustment unit in the dynamic parameter adjustment module (11) satisfies the following: when the surface roughness Ra > 5 μm, it is judged that the workpiece surface is relatively rough, and the spiral spraying trajectory is automatically switched; when the surface roughness Ra < 1 μm, it is judged that the workpiece surface is relatively smooth, and the normal high-speed spraying trajectory is automatically switched; when the adhesion level F < 1, it is judged that the adhesion is too low, and the nozzle pressure is automatically increased to enhance the adhesion between the coating and the substrate; when the adhesion level F > 5, it is judged that the adhesion is too high, and the nozzle pressure is automatically reduced to avoid coating embrittlement; when the temperature T > 35℃ or the humidity RH > 75%, it is judged that the working environment is too hot or too humid, and the gain is dynamically adjusted through the gain coefficient compensation formula of the imaging equipment to correct environmental interference.

5. The system according to claim 3, characterized in that, The dynamic parameter adjustment module (11) also includes an anomaly handling mechanism: when the instantaneous wind speed is detected to be greater than the preset threshold, the following actions are taken: pause spraying and start three-dimensional laser scanning to reconstruct the surface morphology; generate a local respray path based on point cloud data; when the coating thickness non-uniformity is greater than 1%, trigger the adaptive mesh spraying mode.

6. The system according to claim 1, characterized in that, Also includes: The mode selection module (14) is used to dynamically select the quality priority mode and the cost priority mode according to the production type of the workpiece; when the system detects the production of a critical workpiece, it automatically selects the quality priority mode and limits w_1 > w_2; the system will call the preset process optimization instruction set to reduce defects; When the system detects a non-critical workpiece, it automatically selects the cost-priority mode and limits w_2 to w_1; the system will call the preset process optimization instruction set to ensure cost.

7. The system according to claim 1, characterized in that, The multimodal process knowledge base (13) includes: a material property storage unit (131) for storing material property data of the device being coated, the material property data including at least: substrate type and its physical property parameters, coating properties; an environmental data storage unit (132) for storing environmental data, the environmental data including at least: temperature, humidity parameters, wind speed parameters; a process parameter storage unit (133) for storing process parameters, the process parameters including at least: nozzle pressure, spraying speed parameters; an image data storage unit (134) for storing optical inspection images of defective parts, the optical inspection images including at least: micrographs of dripping parts; a text record storage unit (135) for storing operation logs, maintenance reports, and quality inspection notes during the production process; a defect case storage unit (136) for storing working condition feature data packages of historical defect cases, the working condition feature data including process parameter combinations and defect information; the defect information including at least: defect type, defect severity, and defect repair measures; and a defect pattern association storage unit (137) for storing the association between material properties, environmental data, and defect patterns, including: a) the mapping relationship between material type and environmental parameters to defect type; b) a) Quantitative correlation model between environmental parameters and defect morphology; b) Correlation model between substrate surface properties and spatial distribution of defects.

8. The system according to claim 7, characterized in that, The multimodal process knowledge base (13) is updated in the following ways: when the prediction deviation of a new defect case is greater than 15%, incremental learning is triggered to update the association map weights; a time decay factor is applied to historical data of more than 90 days; and the relationship between coating porosity distribution and process parameters is stored.

9. A method for generating intelligent images, characterized in that, The system applied to any one of claims 1-8 comprises the following steps: S1. Real-time data acquisition: synchronously acquiring surface texture features, coating thickness distribution, and ambient temperature and humidity data through multiple sensors; S2. Dynamic parameter collaborative adjustment: dynamically calculating the robotic arm's moving speed v based on the surface roughness Ra, optimizing the nozzle pressure P based on the adhesion level F, and adjusting the imaging device gain coefficient G in conjunction with the temperature and humidity compensation formula; S3. Defect simulation prediction: inputting the set of process parameters into a physical constraint GAN model to generate a coating defect morphology prediction map and identify risk areas for sagging defects; S4. Process optimization decision: selecting weight coefficients based on the quality / cost model, and generating a collaborative optimization instruction set for coating viscosity, spraying speed, and gun distance through reinforcement learning iteration; S5. Closed-loop execution and feedback: issuing optimization instructions to the spraying robotic arm and imaging device, updating the reward function based on newly acquired coating quality data, and triggering incremental learning to optimize the knowledge base association model.

Images