A green intelligent coating equipment pilot test verification platform process optimization simulation system

Abstract

The application discloses a kind of green intelligent coating equipment pilot test verification platform process optimization simulation system, it is related to coating test platform simulation technical field, the system includes the coating process multiple physical field coupling simulation module, paint film quality and energy consumption collaborative prediction module, VOC emission dynamic monitoring and tracing module, process parameter intelligent optimization decision module and pilot test verification digitization twin module that are sequentially communicated with;The application is by building multiple physical field coupling simulation model including spraying flow field, temperature field, humidity field and curing reaction field, realize the high-precision numerical simulation to the whole process of paint film formation, replace traditional physical prototype repeatedly debugging, shorten pilot test verification period, reduce material and energy consumption, by paint film quality and energy consumption collaborative prediction module combines multi-objective optimization algorithm, realize the multi-objective collaborative optimization of paint film quality, production energy consumption and VOC emission, avoid the limitation of single factor optimization.

Description

Technical Field

[0001] This invention relates to the field of coating test platform simulation technology, specifically a process optimization simulation system for a green and intelligent coating equipment test platform. Background Technology

[0002] Green intelligent coating equipment is a new generation of surface treatment equipment that integrates automated control, digital twins, environmental protection and emission reduction technologies, and intelligent sensing. Through full-process digital management and green process integration, it significantly reduces energy consumption and volatile organic compound emissions during production while ensuring consistent coating quality. It is one of the core equipment for achieving green transformation and high-quality development in the manufacturing industry, and is widely used in the surface protection and decoration treatment of parts in the automotive, construction machinery, and aerospace industries. The pilot-scale verification platform for green intelligent coating equipment is a crucial link connecting laboratory R&D and large-scale production. It is used to verify the feasibility and stability of new coating processes, equipment, and materials under near-real-world production conditions. Its process optimization simulation capability directly determines the verification efficiency and the speed of results transformation in the pilot-scale stage. Process optimization simulation uses digital means to predict quality defects, energy consumption levels, and emission characteristics during the coating process in advance, effectively reducing the number of physical pilot-scale tests, shortening the R&D cycle, and lowering R&D costs.

[0003] However, in existing technologies, pilot-scale verification of coating still mainly relies on repeated debugging and destructive testing of physical prototypes. This not only results in long verification cycles and high material and energy consumption, but also makes it difficult to traverse process parameters under all operating conditions. At the same time, existing process optimization methods mostly adopt single-factor variable methods, which cannot simultaneously take into account the multi-objective synergistic optimization of paint film quality, production energy consumption, and VOC emissions. Furthermore, there is a lack of dynamic monitoring and precise source tracing methods for the entire process of VOC generation, migration, and enrichment during coating. As a result, process optimization often suffers from the problem of neglecting one aspect while focusing on another, making it difficult to meet the comprehensive requirements of high efficiency, low carbon emissions, and environmental protection for green and intelligent coating equipment. Therefore, it is of great significance to develop a pilot-scale verification platform process optimization simulation system for green and intelligent coating equipment. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a green and intelligent coating equipment pilot-scale verification platform process optimization simulation system. It can overcome the problems of existing technologies, such as coating pilot-scale verification relying on physical tests, the inability to coordinate multiple objectives in process optimization, and insufficient VOC emission control capabilities. This system integrates digital twin verification, multi-physics simulation, and intelligent process optimization, and can realize digital verification and parameter optimization of the entire coating process. While ensuring coating quality, it reduces energy consumption and VOC emissions in the pilot-scale stage.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a green intelligent coating equipment pilot-scale verification platform process optimization simulation system, which includes a coating process multi-physics field coupling simulation module, a paint film quality and energy consumption collaborative prediction module, a VOC emission dynamic monitoring and source tracing module, a process parameter intelligent optimization decision-making module, and a pilot-scale verification digital twin module connected in sequence.

[0006] The multi-physics coupling simulation module for the coating process is used to construct a multi-physics coupling simulation model and generate a sample dataset. The sample dataset generated by this module, which matches the process parameters and simulation results, will be transmitted in real time unidirectionally to the paint film quality and energy consumption collaborative prediction module, providing standardized basic data for feature extraction and model training of the prediction model.

[0007] The paint film quality and energy consumption collaborative prediction module is used to build a collaborative prediction model based on the sample dataset. The nonlinear mapping relationship between process parameters, paint film quality and equipment energy consumption mined by this module will be pushed to the intelligent optimization decision module of process parameters as the core calculation basis to support subsequent multi-objective optimization calculation.

[0008] The VOC emission dynamic monitoring and source tracing module is used to collect VOC concentration data and calculate emission flux. The VOC emission flux, emission source location results and other environmental protection data obtained by this module are simultaneously connected to the process parameter intelligent optimization decision module, so that parameter optimization can take into account three core indicators: coating quality, equipment energy consumption and exhaust gas emissions.

[0009] The intelligent optimization decision-making module for process parameters is used to perform multi-objective global optimization and generate process parameter combinations. The multiple optimal process parameter combinations obtained by the module through algorithm screening are uniformly distributed to the pilot-scale digital twin module to complete the transition and connection from digital process scheme to physical pilot-scale platform control instructions.

[0010] The pilot-scale digital twin module is used to realize two-way data interaction between the physical platform and the digital twin. The on-site measured data collected by this module will be transmitted back to the paint film quality and energy consumption collaborative prediction module and the VOC emission dynamic monitoring and traceability module, driving the continuous iterative updates of each simulation and prediction model, and gradually improving the computational accuracy and process adaptability of the entire system.

[0011] Furthermore, the multiphysics coupling simulation module for the coating process performs the following operations when constructing the multiphysics coupling simulation model:

[0012] Import the 3D model of the workpiece to be coated and the rheological properties parameters of the coating, and initialize the simulation boundary conditions;

[0013] A coupled computational domain comprising the spraying flow field, temperature field, humidity field, and paint film curing reaction field is constructed, and discretization is performed using a numerical calculation method combining the finite element method and computational fluid dynamics.

[0014] By introducing substrate surface morphology parameters, the processes of coating atomization, droplet deposition, leveling, and film crosslinking are simulated, generating a sample dataset containing process parameters and corresponding simulation results.

[0015] Furthermore, the coupling calculation domain of the multiphysics coupling simulation module for the coating process adopts a meshing method that combines structured and unstructured meshes. The spraying area and drying area are divided using structured meshes, while the complex curved surface area of ​​the workpiece is divided using unstructured meshes. After the meshing is completed, mesh independence is verified. Once the verification is successful, it is used for subsequent multiphysics coupling calculations. During the coupling calculation process, a loosely coupled iterative method is used to realize data transfer and synchronous solution between various physical fields.

[0016] Furthermore, the input parameters of the paint film quality and energy consumption collaborative prediction module include spraying pressure, paint viscosity, gun travel speed, drying temperature curve, spraying path, and exhaust system airflow. The output parameters of the module include paint film thickness uniformity, leveling, adhesion, curing degree, spraying unit energy consumption, and drying unit energy consumption. The module performs feature extraction and normalization on the sample dataset, extracts the nonlinear mapping relationship between process parameters and output parameters in the sample dataset, and constructs a deep learning-based collaborative prediction model. This model is trained using a multi-task joint loss function, the expression of which is: ,in The total loss function of the collaborative prediction model is... The mean squared error loss for the paint film quality prediction subtask. The mean square error loss for the energy consumption prediction subtask. , The weight coefficients for each subtask were determined using the analytic hierarchy process (AHP) combined with historical production data accumulated from the pilot-scale verification platform. and The sum of their values ​​is 1.

[0017] Furthermore, the VOC emission dynamic monitoring and source tracing module performs the following operations when calculating VOC emission flux:

[0018] A workshop airflow organization simulation model was constructed based on the three-dimensional layout of the pilot-scale verification platform. Boundary conditions for air inlets and outlets were set to simulate the airflow distribution under different ventilation conditions.

[0019] Real-time VOC concentration data is collected by distributed gas sensors placed in the spraying area, drying area, and exhaust duct.

[0020] By combining the airflow field distribution results with the mass conservation inversion algorithm, the VOC emission flux of different process stages is calculated, and the spatial location of VOC emission sources is located.

[0021] Furthermore, the distributed gas sensor of the VOC emission dynamic monitoring and source tracing module adopts a composite detection method combining electrochemical sensors and photoionization sensors. The distributed gas sensors are arranged according to a preset three-dimensional spatial grid layout, and the sampling frequency of each sensor is consistent with the data transmission frequency. The VOC emission dynamic monitoring and source tracing module performs moving average filtering and outlier removal on the collected concentration data. The processed data is used for subsequent emission flux calculation and emission source location.

[0022] Furthermore, the intelligent optimization decision-making module for process parameters performs the following operations when executing multi-objective global optimization:

[0023] Receive the parameter mapping relationship output by the paint film quality and energy consumption co-prediction module, and set the preset threshold of the paint film quality index as a constraint condition.

[0024] Using production energy consumption and VOC emissions as optimization objectives, a multi-objective optimization algorithm is invoked for global search to calculate the comprehensive fitness value of each individual. The expression for the comprehensive fitness value is as follows: ,in The first in the population The overall fitness value of an individual For the first Pareto dominance level of an individual For the first The crowding degree of an individual within its dominance level, in order to eliminate differences in units and value ranges, needs to be determined by... and Normalization is performed so that both are mapped to the interval [0, 1]. , The fitness weight coefficients were determined by orthogonal experimentation combined with typical working condition data from the pilot-scale verification platform, generating a Pareto optimal solution set.

[0025] A multi-attribute decision method is used to sort the Pareto optimal solution set and output multiple sets of process parameter combinations that meet different priority requirements.

[0026] Furthermore, the intelligent optimization decision-making module for process parameters employs a multi-objective optimization algorithm that introduces adaptive crossover and mutation operators based on the traditional non-dominated sorting genetic algorithm. The adaptive crossover operator dynamically adjusts the crossover probability according to the number of generations of the population. The expression for the crossover probability is: ,in This represents the crossover probability of the current generation. The preset maximum crossover probability, The minimum crossover probability is preset. For the current generation, The maximum number of generations is preset. The adaptive mutation operator dynamically adjusts the mutation probability based on the fitness value of the individual and the average fitness of the population. The intelligent optimization decision-making module of process parameters verifies the feasibility of the equipment operation boundary of the generated process parameter combination and outputs it to the pilot-scale verification digital twin module.

[0027] Furthermore, the pilot-scale digital twin module performs the following operations when realizing two-way data interaction:

[0028] The system receives the process parameter combination output by the intelligent optimization decision-making module and converts it into a standard control command format that can be recognized by the physical testing and verification platform.

[0029] Control commands are sent to the painting robot, drying oven, and exhaust system for execution.

[0030] Data on the operating status of various sensors deployed on the physical testing platform, as well as data on paint film detection and VOC emissions, are collected and fed back to the digital twin.

[0031] Furthermore, the pilot-scale verification digital twin module constructs a digital twin that corresponds one-to-one with the physical pilot-scale verification platform. The digital twin includes the equipment geometric model, kinematic model, and control logic model of the physical platform. The pilot-scale verification digital twin module supports three-dimensional visualization of the pilot-scale verification process and historical data backtracking. Historical data is stored hierarchically and indexed quickly according to timestamps. The pilot-scale verification digital twin module synchronizes feedback data to the paint film quality and energy consumption collaborative prediction module and the process parameter intelligent optimization decision-making module, triggering the iterative update process of the model.

[0032] Compared with existing technologies, the process optimization simulation system of this green intelligent coating equipment pilot verification platform has the following beneficial effects:

[0033] This invention constructs a multi-physics coupled simulation model that includes the spraying flow field, temperature field, humidity field, and curing reaction field, achieving high-precision numerical simulation of the entire paint film formation process. This replaces the repeated debugging of traditional physical prototypes, shortens the pilot-scale verification cycle, and reduces material and energy consumption. Through a paint film quality and energy consumption collaborative prediction module combined with a multi-objective optimization algorithm, it achieves multi-objective collaborative optimization of paint film quality, production energy consumption, and VOC emissions, avoiding the limitations of single-factor optimization. Through a VOC emission dynamic monitoring and source tracing module, it achieves precise location of emission sources and dynamic calculation of emission flux, providing data support for exhaust system optimization. Through bidirectional mapping and iterative optimization of the pilot-scale verification digital twin module, it further improves the accuracy and practicality of process optimization results.

[0034] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0036] Figure 1 A schematic diagram of the process optimization simulation system of a pilot verification platform for a green and intelligent coating equipment;

[0037] Figure 2 A flowchart of the process optimization simulation system for a pilot verification platform of a green and intelligent coating equipment;

[0038] Figure 3 This is a flowchart of the multiphysics coupling simulation module for the painting process during the construction of the multiphysics coupling simulation model. Detailed Implementation

[0039] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0040] This invention provides a green and intelligent coating equipment pilot-scale verification platform process optimization simulation system, targeting pilot-scale verification scenarios for coating equipment in the automotive, construction machinery, and aerospace industries. It addresses industry pain points such as traditional coating pilot-scale testing relying on repeated debugging of physical prototypes, long verification cycles, high material and energy consumption, process optimization considering only a single indicator, inability to simultaneously consider paint film quality, production energy consumption and VOC emissions, and the lack of dynamic monitoring and precise traceability of VOC emissions. It establishes a multi-module collaborative digital simulation and intelligent optimization closed-loop system.

[0041] The entire system consists of a multi-physics field coupled simulation module for the coating process, a collaborative prediction module for paint film quality and energy consumption, a dynamic monitoring and tracing module for VOC emissions, an intelligent optimization and decision-making module for process parameters, and a pilot-scale digital twin module that are connected in sequence. Each module works together according to the data flow order and forms an iterative closed loop.

[0042] The multi-physics coupled simulation module for the coating process integrates the finite element method and computational fluid dynamics to build coupled models of four types of physical fields: spray flow field, temperature field, humidity field, and paint film curing reaction field. It simulates the entire process of paint atomization, deposition, leveling, and paint film cross-linking, generating a standardized sample dataset. The paint film quality and energy consumption collaborative prediction module relies on deep learning algorithms and multi-task joint loss functions to uncover the nonlinear correlation between process parameters, paint film quality, and equipment energy consumption. The VOC emission dynamic monitoring and source tracing module uses composite sensors to collect concentration data and combines airflow simulation and mass conservation inversion algorithms to calculate emission flux and locate emission sources. The intelligent optimization decision-making module for process parameters uses a non-dominated sorting genetic algorithm to perform multi-objective global optimization, outputting multiple optimal combinations of process parameters based on constraints. The pilot-scale verification digital twin module constructs a digital twin equivalent to the physical platform, completing process command issuance, on-site data acquisition, and reverse transmission, driving continuous iterative updates of the front-end models. The entire system replaces numerous physical experiments with digital simulation, achieving multi-objective synergistic optimization of three major indicators: paint film quality, production energy consumption, and VOC emissions. Simultaneously, it completes virtual-real linkage verification of the entire coating process, comprehensively improving the efficiency and greening level of coating pilot production. A detailed description follows with specific embodiments.

[0043] This embodiment applies the system to a green intelligent coating equipment testing and verification platform for automotive chassis components. The platform is equipped with an automated painting robot, a segmented constant-temperature drying oven, a distributed ventilation system, and a complete set of sensing sensors. It primarily performs testing and verification work on the coating process of irregularly shaped metal components such as automotive steering knuckles and suspension brackets. The entire system is designed according to the following... Figure 1 The modular architecture shown completes hardware deployment and software debugging. The five functional modules maintain real-time communication and complete data transmission, computation and analysis, parameter optimization, and virtual-physical interaction according to a fixed process. The overall workflow follows the following... Figure 2The operational logic shown forms a complete closed-loop operation system, from simulation modeling to pilot-scale verification and data feedback iteration.

[0044] The multiphysics coupling simulation module for the painting process is the data source of the entire system, responsible for providing sufficient and accurate simulation samples for subsequent prediction models, and is also the core unit for realizing full-process digital simulation. For example... Figure 3 As shown, this module completes the construction of multiphysics coupling models, numerical calculations, and generation of sample datasets according to a fixed process.

[0045] The first step is to import the basic simulation data, import the three-dimensional solid model of the automotive steering knuckle component into the simulation software, and at the same time enter the rheological properties parameters of the industrial coating used, including the viscosity variation range, surface tension, curing reaction rate and other basic parameters. Combined with the actual operating environment of the pilot platform, the simulation boundary conditions are initialized, and the constraints such as the pressure range of the spraying area, the basic values ​​of temperature and humidity in the workshop environment, and the upper and lower limits of temperature in the drying area are defined.

[0046] The second step involves constructing a multi-physics coupled computational domain and performing numerical discretization. This simulation integrates four major physical fields: the spraying flow field, temperature field, humidity field, and paint film curing reaction field. A hybrid numerical calculation method combining the finite element method and computational fluid dynamics is used to discretize the entire computational domain. In the mesh generation stage, a combination of structured and unstructured meshes is employed. Regular spraying and drying areas within the platform are uniformly divided using structured meshes to ensure computational efficiency and stability in regular regions. For automotive steering knuckles, which have numerous curved surfaces and sharp edges, unstructured meshes are used for subdivision of complex curved surface areas to ensure simulation accuracy in these areas. After all mesh generation is completed, mesh independence verification is performed. The mesh density is gradually adjusted and the simulation results are compared. When the mesh size continues to decrease but the fluctuation range of the simulation output results is below the industry allowable error value, the mesh generation is deemed qualified, and the coupled computation is officially initiated. In the multiphysics coupling solution stage, a loosely coupled iterative method is used throughout the process. Data is transferred and solved synchronously between the flow field, temperature field, humidity field, and solidification reaction field according to a fixed iteration frequency, ensuring that the operating states of each physical field remain linked and matched.

[0047] The third step involves introducing substrate surface morphology parameters, incorporating parameters such as surface roughness and microtexture of the metal workpiece into the simulation system. This fully simulates the entire coating process, sequentially reproducing a series of physical and chemical reaction processes, including high-pressure atomization of the coating, deposition of tiny droplets onto the workpiece surface, natural leveling of the coating, and thermal cross-linking and curing of the paint film. The entire simulation process runs continuously, constantly adjusting process parameters such as spraying pressure, gun speed, and drying temperature, recording the full-process simulation results for each set of process parameters. Finally, all data is integrated to generate a standardized sample dataset containing process parameters and corresponding simulation results. This sample dataset is transmitted in real-time unidirectionally to the paint film quality and energy consumption collaborative prediction module via a communication interface, providing solid basic data support for model training.

[0048] The paint film quality and energy consumption co-prediction module receives sample datasets from the multiphysics coupled simulation module. Its core task is to explore the intrinsic relationship between process parameters and paint film quality and equipment energy consumption, and to build a stable and reliable co-prediction model. The input parameters set for this module include six categories of process parameters: spraying pressure, paint viscosity, gun speed, drying temperature profile, spraying path, and exhaust system airflow. The output parameters correspond to four paint film quality indicators: paint film thickness uniformity, leveling, adhesion, and curing degree, as well as two energy consumption indicators: spraying unit energy consumption and drying unit energy consumption.

[0049] First, data preprocessing is performed on the received sample dataset, including data cleaning, feature extraction, and global normalization to eliminate the influence of different parameter units and strengthen the feature correlation between process parameters and output indicators. After preprocessing, a deep learning-based multi-task collaborative prediction model is built, which simultaneously undertakes two sub-tasks: paint film quality prediction and energy consumption prediction. In the specific implementation of this embodiment, the collaborative prediction model uses a multi-task joint loss function for model training, and the loss function expression is as follows: The parameters in the formula are defined as follows: This represents the total loss function during the training process of the collaborative prediction model. The mean squared error loss corresponds to the paint film quality prediction subtask and is used to measure the deviation between the predicted values ​​and the actual simulated values ​​of various paint film quality indicators. The mean squared error loss corresponding to the energy consumption prediction subtask is used to characterize the error level of the energy consumption prediction results. and These are the weighting coefficients for the paint film quality subtask and the energy consumption subtask, respectively, and the sum of the two coefficients is fixed at 1.

[0050] This embodiment employs the analytic hierarchy process (AHP) to quantify and determine the two weight coefficients. The specific implementation process involves collecting twelve consecutive months of historical operating data from the automotive parts coating pilot platform, encompassing tens of thousands of process parameters, paint film quality inspection reports, and equipment energy consumption statistics. A hierarchical analysis model is built using paint film quality and production energy consumption as the two main evaluation dimensions. A pairwise comparison judgment matrix is ​​constructed, and consistency checks are performed on the matrix. After eliminating invalid data interference, the matrix eigenvectors are calculated, and then normalized to obtain the initial weight values. Combining the quality-first production principle of the automotive parts coating industry, the final weight values ​​are determined based on the comprehensive calculation results. The value is 0.6. The value is set to 0.4. After determining the weight parameters, iterative model training is initiated, continuously optimizing the network weights until the model loss function converges. The trained model can accurately output the predicted results of paint film quality and energy consumption corresponding to different combinations of process parameters, and simultaneously extract the nonlinear mapping relationship between process parameters, paint film quality, and equipment energy consumption. The module uses this mapping relationship as the core calculation basis and pushes it synchronously to the intelligent optimization decision module for process parameters, providing core support for subsequent multi-objective optimization calculations.

[0051] The VOC emission dynamic monitoring and source tracing module focuses on monitoring, flux calculation, and emission source location of volatile organic compounds throughout the entire coating process. The acquired VOC-related data will be used as an important optimization indicator in the intelligent optimization process. First, based on the overall three-dimensional layout of the automotive parts coating pilot platform, a global airflow organization simulation model of the workshop is built. Simulation boundary conditions are set according to the actual locations of the workshop's air inlets, outlets, and ventilation ducts to simulate the airflow patterns inside the workshop under different exhaust volumes and ventilation modes, clarifying basic airflow characteristics such as airflow direction and wind speed distribution.

[0052] Next, the deployment and debugging of distributed gas sensors were completed. In this embodiment, a composite detection device combining electrochemical and photoionization sensors was selected. The two types of sensors complement each other and are suitable for the detection needs of low-concentration and high-concentration VOC scenarios, respectively. All sensors were installed according to a preset three-dimensional spatial grid layout. The grid spacing in the spraying area was set to two meters, and the grid spacing in the drying area was set to three meters. A set of composite sensors was arranged every five meters inside the exhaust duct to achieve full-area monitoring of the coating area without blind spots. The sampling frequency and data transmission frequency of all sensors were uniformly set to ensure the temporal consistency of the collected data. After the sensors were started, they continuously collected real-time raw VOC concentration data in the spraying area, drying area, and exhaust duct. The module background performed noise reduction processing on the raw data, used a moving average filtering algorithm to smooth data fluctuations, and set reasonable thresholds to remove outliers and invalid data caused by equipment interference and instantaneous environmental fluctuations.

[0053] After data preprocessing, the module combines the previously obtained workshop airflow distribution results with a mass conservation inversion algorithm to calculate the VOC emission flux for different process stages, including spraying, drying, and waste gas transport. Simultaneously, it precisely locates the spatial position of VOC emission sources based on concentration distribution and airflow direction. The module ultimately outputs VOC emission flux data and emission source tracing results, simultaneously transmitting these two types of environmental indicator data to the intelligent optimization decision-making module for process parameters. This allows subsequent process optimization to simultaneously consider the three core objectives of paint film quality, production energy consumption, and VOC emissions.

[0054] The intelligent optimization decision-making module for process parameters is the core of the entire system's process optimization. It integrates all data from the front-end prediction module and the VOC monitoring module, conducting multi-objective global optimization of energy consumption and VOC emissions while ensuring paint film quality. The module first receives the process quality-energy consumption mapping relationship and VOC emission-related data. Combining this with the coating technology standards for automotive chassis components, it sets fixed thresholds for key quality indicators such as paint film thickness uniformity, adhesion, and curing degree. These thresholds serve as hard constraints for the algorithm's operation, ensuring that all optimization results meet the paint film quality standards. Based on this, it sets reducing overall production energy consumption and decreasing total VOC emissions as the two major optimization objectives, and then uses a non-dominated sorting genetic algorithm to perform a global search operation.

[0055] In the specific implementation of this embodiment, the algorithm needs to calculate the overall fitness value of each individual in the population, and the corresponding expression is: In the formula Represents the first in the population The overall fitness value corresponding to each individual For the first The Pareto dominance level corresponds to each individual, and the level value can reflect the degree of excellence or inferiority of the individual in multi-objective optimization. This represents the crowding level of an individual within its dominance hierarchy, used to measure the density of the population's individual distribution. Because the two indicators have significantly different dimensions and ranges, it is necessary to perform calculations beforehand. and Normalization is performed to map the values ​​of the two indicators to a range of 0 to 1. and To determine the fitness weight coefficients for the two indicators, this embodiment employs an orthogonal experimental design combined with typical operating data from the pilot-scale platform. Ten mainstream coating conditions commonly encountered during platform operation are selected, and a four-factor, three-level orthogonal experimental scheme is designed. Multiple weight combinations are configured to test the algorithm, comparing the optimization speed, solution set diversity, and process parameter rationality under different weights. The final coefficients are determined based on a comprehensive analysis of all experimental results. The value is 0.7. The value is 0.3.

[0056] To further improve the optimization performance of traditional genetic algorithms, this algorithm introduces adaptive crossover and adaptive mutation operators to avoid problems such as premature convergence and local optima. In the specific implementation of this embodiment, the adaptive crossover operator dynamically adjusts the crossover probability based on the real-time generation number of the population; the corresponding calculation formula is as follows: , in the formula This represents the crossover probability corresponding to the current generation. The maximum crossover probability preset for the algorithm is set to 0.9, taking into account the characteristics of this painting condition. The preset minimum crossover probability is set to 0.3. This represents the current generation of the algorithm. The maximum number of generations preset for the algorithm is uniformly set to 200 in this embodiment. In the early stages of the algorithm's operation, the number of generations is relatively small, and the crossover probability remains at a high level, strengthening gene exchange between individuals in the population and expanding the search range. In the later stages of operation, the number of generations gradually approaches the maximum value, and the crossover probability continuously decreases, ensuring the stability of high-quality individuals. The adaptive mutation operator dynamically adjusts the mutation probability based on the difference between the fitness value of an individual and the average fitness of the population. Individuals with low fitness have their mutation probability increased, increasing their evolutionary potential, while individuals with high fitness have their mutation probability decreased, protecting high-quality solutions.

[0057] After multiple rounds of iterative computation, the algorithm generates a Pareto optimal solution set. Then, a multi-attribute decision-making method is used to prioritize the parameter combinations within the solution set. Based on different application requirements such as quality priority, energy saving priority, and emission reduction priority, multiple sets of differentiated optimal process parameter combinations are output. After the parameter combinations are generated, the module verifies whether each process parameter is within the rated operating boundaries of the spraying robot, drying oven, and exhaust system. Invalid parameters exceeding the equipment's operating range are eliminated, and the verified optimal process parameter combinations are uniformly distributed to the pilot-scale verification digital twin module.

[0058] The pilot-scale digital twin module plays a crucial role in virtual-real interaction, on-site execution, and data feedback. It serves as the hub connecting the digital simulation system and the physical pilot-scale platform, and the closed-loop iterative capability of the entire system also relies on this module. The module first completes the construction of the digital twin. Referring to the physical equipment of the automotive parts coating pilot-scale platform, it constructs a one-to-one corresponding digital twin model. The model fully reproduces the geometric shape, kinematic laws, and internal control logic of equipment such as the painting robot, drying oven, and ventilation system. At the same time, it builds an overall workshop scene model to achieve full-process 3D visualization.

[0059] After receiving the optimal combination of process parameters from the intelligent optimization module, the module first converts the parameter format, transforming the digital process parameters into standard control commands that can be recognized by various devices on the physical pilot platform. Then, it sends these control commands to the on-site painting robot, segmented drying oven, and exhaust system via the industrial communication bus, driving the entire physical equipment to perform coating operations according to the optimized process parameters. Throughout the workpiece coating process, temperature sensors, thickness sensors, gas sensors, and power monitoring sensors deployed throughout the physical platform continuously collect equipment operating status data, measured paint film quality data, and on-site VOC emission data. All measured data is transmitted back to the digital twin in real time, achieving real-time synchronization between the physical operating conditions and the digital model.

[0060] The module is equipped with a data storage unit that stores all historical operational and testing data in layers according to timestamps. It also features a rapid indexing mechanism, allowing staff to retrieve pilot-scale process data from any time period at any time for retrospective analysis of the process. After each batch of parts is coated in a pilot-scale test, the module automatically synchronizes the measured paint film quality data, energy consumption data, and VOC emission data from the field back to the front-end paint film quality and energy consumption collaborative prediction module and the VOC emission dynamic monitoring and traceability module, triggering iterative updates to the models within these two modules. The system continuously corrects parameter deviations in the simulation and prediction models using real-world data, constantly improving the simulation accuracy and optimization effect of the entire system, creating a continuously self-improving closed loop for the entire process optimization system.

[0061] In summary, this embodiment, by applying the process optimization simulation system to the pilot-scale verification platform for automotive chassis component coating, comprehensively solves various problems existing in the traditional pilot-scale coating mode and achieves significant application results. Relying on the multi-physics coupled simulation module to complete the digital simulation of the entire coating process, it significantly reduces the number of physical prototype debugging and physical test sprays, while reducing the consumption of materials and energy such as paint and electricity, thus lowering the operating costs of the pilot-scale stage. The deep learning collaborative prediction model, combined with multi-objective optimization algorithms, breaks through the limitations of traditional single-factor process optimization, achieving synergistic optimization of three indicators: paint film quality, production energy consumption, and VOC emissions. The optimized process parameters fully meet the quality standards for automotive component coating, while simultaneously reducing production energy consumption and VOC emissions. The VOC emission dynamic monitoring and source tracing module can monitor the exhaust gas emission status of each process in real time, accurately locate the emission source, and provide accurate data support for exhaust system control and exhaust gas treatment scheme optimization. The pilot-scale verification digital twin module achieves bidirectional linkage between the physical platform and the digital model, continuously iterating and updating the entire system model with the help of on-site measured data, allowing the system's simulation accuracy and process optimization capabilities to steadily improve over the long term.

[0062] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A process optimization simulation system for a pilot-scale verification platform of green intelligent coating equipment, characterized in that, The system includes a multi-physics coupled simulation module for the coating process, a collaborative prediction module for paint film quality and energy consumption, a dynamic monitoring and tracing module for VOC emissions, an intelligent optimization decision-making module for process parameters, and a digital twin module for pilot-scale verification, all connected in sequence. The multi-physics coupling simulation module for the coating process is used to construct a multi-physics coupling simulation model and generate a sample dataset. The paint film quality and energy consumption collaborative prediction module is used to build a collaborative prediction model based on the sample dataset; The VOC emission dynamic monitoring and source tracing module is used to collect VOC concentration data and calculate emission flux. The intelligent optimization decision-making module for process parameters is used to perform multi-objective global optimization and generate combinations of process parameters. The pilot-scale digital twin module is used to enable two-way data interaction between the physical platform and the digital twin.

2. The green intelligent coating equipment pilot verification platform process optimization simulation system according to claim 1, characterized in that, The multiphysics coupling simulation module for the coating process performs the following operations when constructing the multiphysics coupling simulation model: Import the 3D model of the workpiece to be coated and the rheological properties parameters of the coating, and initialize the simulation boundary conditions; A coupled computational domain comprising the spraying flow field, temperature field, humidity field, and paint film curing reaction field is constructed, and discretization is performed using a numerical calculation method combining the finite element method and computational fluid dynamics. By introducing substrate surface morphology parameters, the processes of coating atomization, droplet deposition, leveling, and film crosslinking are simulated, generating a sample dataset containing process parameters and corresponding simulation results.

3. The green intelligent coating equipment pilot-scale verification platform process optimization simulation system according to claim 2, characterized in that, The coupling calculation domain of the multiphysics coupling simulation module for the coating process adopts a meshing method that combines structured and unstructured meshes. The spraying area and drying area are divided using structured meshes, while the complex curved surface area of ​​the workpiece is divided using unstructured meshes. After the mesh division is completed, the mesh independence is verified. After the verification is passed, it is used for subsequent multiphysics coupling calculations. During the coupling calculation process, a loosely coupled iterative method is adopted to realize data transfer and synchronous solution between various physical fields.

4. The green intelligent coating equipment pilot verification platform process optimization simulation system according to claim 1, characterized in that, The input parameters of the paint film quality and energy consumption co-prediction module include spraying pressure, paint viscosity, gun speed, drying temperature curve, spraying path, and exhaust system airflow. The output parameters include paint film thickness uniformity, leveling, adhesion, curing degree, spraying unit energy consumption, and drying unit energy consumption. The module performs feature extraction and normalization on the sample dataset, extracts the nonlinear mapping relationship between process parameters and output parameters in the sample dataset, and constructs a deep learning-based co-prediction model. The co-prediction model is trained using a multi-task joint loss function, the expression of which is: ,in The total loss function of the collaborative prediction model is... The mean squared error loss for the paint film quality prediction subtask. The mean square error loss for the energy consumption prediction subtask. , These are the weighting coefficients for each subtask.

5. The green intelligent coating equipment pilot verification platform process optimization simulation system according to claim 1, characterized in that, The VOC emission dynamic monitoring and source tracing module performs the following operations when calculating VOC emission flux: A workshop airflow organization simulation model was constructed based on the three-dimensional layout of the pilot-scale verification platform. Boundary conditions for air inlets and outlets were set to simulate the airflow distribution under different ventilation conditions. Real-time VOC concentration data is collected by distributed gas sensors placed in the spraying area, drying area, and exhaust duct. By combining the airflow field distribution results with the mass conservation inversion algorithm, the VOC emission flux of different process stages is calculated, and the spatial location of VOC emission sources is located.

6. The green intelligent coating equipment pilot verification platform process optimization simulation system according to claim 5, characterized in that, The distributed gas sensor of the VOC emission dynamic monitoring and source tracing module adopts a composite detection method combining electrochemical sensors and photoionization sensors. The distributed gas sensors are arranged according to a preset three-dimensional spatial grid layout. The sampling frequency of each sensor is consistent with the data transmission frequency. The VOC emission dynamic monitoring and source tracing module performs moving average filtering and outlier removal on the collected concentration data. The processed data is used for subsequent emission flux calculation and emission source location.

7. The green intelligent coating equipment pilot verification platform process optimization simulation system according to claim 1, characterized in that, The intelligent optimization decision-making module for process parameters performs the following operations when executing multi-objective global optimization: Receive the parameter mapping relationship output by the paint film quality and energy consumption co-prediction module, and set the preset threshold of the paint film quality index as a constraint condition. Using production energy consumption and VOC emissions as optimization objectives, a multi-objective optimization algorithm is invoked for global search to calculate the comprehensive fitness value of each individual. The expression for the comprehensive fitness value is as follows: ,in The first in the population The overall fitness value of an individual For the first Pareto dominance level of an individual For the first The degree of crowding of an individual within its dominance hierarchy , Generate Pareto optimal solution sets using fitness weight coefficients; A multi-attribute decision method is used to sort the Pareto optimal solution set and output multiple sets of process parameter combinations that meet different priority requirements.

8. The green intelligent coating equipment pilot verification platform process optimization simulation system according to claim 7, characterized in that, The intelligent optimization decision-making module for process parameters employs a multi-objective optimization algorithm that introduces adaptive crossover and mutation operators based on the traditional non-dominated sorting genetic algorithm. The adaptive crossover operator dynamically adjusts the crossover probability according to the number of generations of the population. The expression for the crossover probability is: ,in This represents the crossover probability of the current generation. The preset maximum crossover probability, The minimum crossover probability is preset. For the current generation, The maximum number of generations is preset. The adaptive mutation operator dynamically adjusts the mutation probability based on the fitness value of the individual and the average fitness of the population. The intelligent optimization decision-making module of process parameters verifies the feasibility of the equipment operation boundary of the generated process parameter combination and outputs it to the pilot-scale verification digital twin module.

9. The green intelligent coating equipment pilot verification platform process optimization simulation system according to claim 1, characterized in that, The pilot-scale digital twin module performs the following operations when realizing two-way data interaction: The system receives the process parameter combination output by the intelligent optimization decision-making module and converts it into a standard control command format that can be recognized by the physical testing and verification platform. Control commands are sent to the painting robot, drying oven, and exhaust system for execution. Data on the operating status of various sensors deployed on the physical testing platform, as well as data on paint film detection and VOC emissions, are collected and fed back to the digital twin.

10. The green intelligent coating equipment pilot verification platform process optimization simulation system according to claim 1, characterized in that, The pilot-scale verification digital twin module constructs a digital twin that corresponds one-to-one with the physical pilot-scale verification platform. The digital twin includes the equipment geometric model, kinematic model, and control logic model of the physical platform. The pilot-scale verification digital twin module supports three-dimensional visualization of the pilot-scale verification process and historical data backtracking. Historical data is stored hierarchically and indexed quickly according to timestamps. The pilot-scale verification digital twin module synchronizes feedback data to the paint film quality and energy consumption collaborative prediction module and the process parameter intelligent optimization decision module, triggering the iterative update process of the model.

Images