Spraying process self-adaptive adjustment method and system based on temperature measurement

Abstract

The invention relates to the technical field of spraying control, and discloses a spraying process self-adaptive adjustment method and system based on temperature measurement, and the method comprises the steps: presetting multi-sensor parameters, calculating a dynamic weight, calibrating sensors, and solving the problems of real-time interference and cross-process sensitivity difference; acquiring environmental data of a target spraying process, generating a global temperature field thermodynamic diagram, associating the film thickness with an environmental interference label, and providing a three-dimensional reference; coating workpiece heat flux and empty workpiece heat flux are collected, net heat flux and real-time reaction enthalpy change are calculated, a three-dimensional curing degree model is constructed, and the real-time curing degree is calculated to reflect the coating film curing progress; based on real-time curing degree fusion temperature spatial-temporal characteristics and phase change characteristics, core characteristics are extracted, a digital twin model is established to preview a process adjustment effect, an optimal control vector is output by constructing a spray gun three-dimensional optimization function and an improved quantum genetic algorithm, a targeted actuator is triggered for regulation and control, and a coating thermodynamic parameter library is constructed. And updating model parameters to optimize the overall situation.

Description

Technical Field

[0001] This invention relates to the field of spraying control technology, and more specifically to an adaptive adjustment method and system for spraying processes based on temperature measurement. Background Technology

[0002] Existing technologies often employ fixed-weight fusion of sensor data, failing to consider real-time changes in signal-to-noise ratio and differences in sensor sensitivity across process scenarios. For example, infrared signal attenuation caused by spray droplets can easily introduce measurement errors. Traditional methods often rely on a single sensor, using only infrared to measure surface temperature without integrating key factors such as substrate temperature, film thickness, and temperature and humidity, making it difficult to reflect the overall temperature distribution. Curing models often ignore the impact of film thickness on reaction rate, leading to misjudgments of insufficient curing of thick films or over-curing of thin films. There is a lack of process pre-simulation mechanisms based on digital twins, with multi-spray gun parameter adjustments relying on experience or simple rules, failing to consider coupling relationships such as airflow interference between spray guns, easily resulting in uneven coatings and inconsistent curing. Control of process anomalies is mostly done by adjusting a single parameter, lacking specificity. Model parameter updates rely on local data, failing to achieve global optimization in distributed scenarios, and long-term use can easily accumulate errors. There is a lack of a unified coating thermodynamic parameter library, requiring repeated experiments for new coating adaptation, and the models have not undergone extensive process compatibility verification, making it difficult to quickly migrate to different production scenarios. Therefore, there is a need to provide an adaptive adjustment method and system for the spraying process based on temperature measurement. Summary of the Invention

[0003] The purpose of this invention is to provide an adaptive adjustment method and system for spraying processes based on temperature measurement. To solve the aforementioned problems in the prior art, this invention achieves this through the following technical solution: In a first aspect, the present invention provides an adaptive adjustment method for a spraying process based on temperature measurement, which specifically includes the following steps: Step 1: Preset multi-sensor parameters, calculate dynamic weights through signal-to-noise ratio and process adaptation coefficient, and solve the problems of real-time interference and cross-process sensitivity differences by fusing reference temperature to calibrate sensors; Step 2: Obtain environmental data of the target spraying process through calibrated sensors, generate a global temperature field thermogram by calculating and fusing the temperature field, and provide a three-dimensional reference by associating film thickness with environmental interference annotations; Step 3: Collect the heat flux of the coated workpiece and the empty workpiece using a heat flux sensor, calculate the net heat flux and real-time reaction enthalpy change, and construct a three-dimensional curing degree model by combining the autocatalytic reaction kinetic model and film thickness correction term to calculate the real-time curing degree, reflecting the coating curing progress. Step 4: Based on the real-time curing degree fusion temperature spatiotemporal characteristics and phase transition characteristics, extract the core features, combine FEA and CFD to establish a digital twin model to pre-simulate the process adjustment effect, and output the optimal control vector by constructing a three-dimensional optimization function for the spray gun and improving the quantum genetic algorithm. Step 5: Based on the optimal control vector, trigger targeted actuator regulation, construct a coating thermodynamic parameter library, and update model parameters through Bayesian deep neural network and federated learning to optimize the global model.

[0004] The method for calibrating the sensor by fusing a reference temperature is as follows: Calculate the signal-to-noise ratio of the infrared sensor Signal-to-noise ratio compared to capacitive sensors The weights are then assigned based on the squared ratio of the signal-to-noise ratios of the infrared sensor and the capacitive sensor to obtain dynamic weights. Dynamic weights Includes: dynamic weights of infrared sensors Dynamic weights of capacitive sensors And the dynamic weight of the infrared sensor Dynamic weights of capacitive sensors The sum of is 1; Real-time calibration and verification are performed based on the obtained dynamic weights. The dynamic weights of the infrared sensor are then fused and calculated using a weighted formula. Calibration temperature with infrared sensor Multiply to obtain the infrared sensor reference temperature, and then apply the dynamic weights of the capacitive sensor. Temperature measurement with capacitive sensor Multiplying these gives the reference temperature of the capacitive sensor; The reference temperature of the infrared sensor and the reference temperature of the capacitive sensor are summed to obtain the fused reference temperature. If the calibration temperature difference is greater than the preset temperature difference threshold, the sensor parameter fine-tuning is triggered. The calibration temperature difference is multiplied by the sensor's preset parameter fine-tuning coefficient and then added to the correction base value to obtain the temperature correction coefficient. Furthermore, to acquire environmental data for the target coating process, a capacitive probe is placed below the workpiece to directly measure the substrate temperature through the coating; an acoustic emission sensor is installed on the side of the workpiece to collect the acoustic spectrum signal of droplets impacting the workpiece. It also filters low-frequency noise from the airflow using a bandpass filter; a temperature and humidity sensor collects ambient temperature data. and relative humidity Laser thickness gauge collects real-time film thickness. Replace the fixed film thickness d value; Based on environmental data acquired by sensors, an adaptive fusion formula is constructed, which utilizes dynamic weights from infrared sensors. Two-dimensional temperature field corresponding to infrared sensor Combined with temperature correction factor Opposite number, real-time film thickness Infrared transmittance compensation coefficient for temperature and humidity The product of these is the product of exponential functions of exponents, and the fused temperature field is obtained through the environmental adaptive fusion formula. ; Set the coordinates of each pixel Corresponding fusion temperature field By splicing the data according to their spatial location, a temperature field thermogram of the entire workpiece is generated. Furthermore, the method for calculating the net heat flux is as follows: A heat flux sensor is attached to the back of the workpiece, directly opposite the coating area, to simultaneously collect two types of heat flux data: the heat flux of the coated workpiece and the heat flux of the uncoated workpiece. The heat flux of the coated workpiece includes the heat released during coating curing and the heat dissipation from the environment. A laser thickness gauge transmits the real-time film thickness to the enthalpy change calculation unit, and the heat flux sensor simultaneously records the heat flux data corresponding to the real-time film thickness. The heat flux of the coated workpiece at time t is then calculated. Heat flux of empty workpiece at time t The difference is used to calculate the net heat flux of the spraying process using the net heat flux formula. ; The net heat flux is integrated over time, and the real-time enthalpy change is calculated by combining the sensor coverage area and the coating quality. Furthermore, the method for obtaining the real-time reaction enthalpy change value is as follows: By integrating the net heat flux over time and combining the sensor coverage area with the coating mass, the real-time enthalpy change is calculated. This real-time enthalpy change directly reflects the progress of the curing reaction. The coating mass at time t is then used to determine the enthalpy change. The reciprocal of the time Net heat flux The real-time enthalpy change is obtained by integrating the product of the heat flux sensor coverage area S. ; Furthermore, the method for constructing the three-dimensional curing degree model is as follows: The method for constructing the three-dimensional solidification model is as follows: By incorporating a film thickness correction term and utilizing the kinetic differential equation, multiplying the pre-exponential factor A by an exponential term, the rate of change of curing degree over time can be obtained through the kinetic differential equation. ; The real-time curing degree integral is calculated by numerical integration, and the real-time curing degree is obtained by subtracting the exponent term from 1. ; Furthermore, the method for adjusting the effect of the pre-simulation process is as follows: By integrating the spatiotemporal characteristics of temperature and phase transition characteristics, core features are extracted. Based on the obtained core features, the temperature field, degree of curing, and film thickness after process parameter adjustment are simulated. The specific model is as follows: Interferometric correction thermal field simulation, using gradient operator thermal conductivity of coating With temperature gradient The product is used for divergence calculation, plus the heat flux released during solidification. Subtract air density Specific heat capacity of air Interference speed of airflow from adjacent spray guns With temperature gradient The product of these two factors equals the coating density. Specific heat capacity of coating With temperature The product of the partial derivatives with respect to time; where, and Index the spray guns by number; Interference-corrected film thickness simulation, through coating flow rate Coating density Spraying time The product of these three factors, divided by the spray gun's movement speed Area of ​​the sprayed area The product of these two factors, multiplied by the correction term, yields the adjusted coating thickness. ; The pre-simulation results of multiple spray guns are obtained by running the twin model of M spray guns simultaneously through NPU, and the pre-simulation temperature, pre-simulation degree of curing and pre-simulation coating thickness of each spray gun are output. Furthermore, the method for outputting the optimal control vector is as follows: Output the pre-test temperature of each spray gun Pre-curing degree and pre-coating thickness Minimize the pre-simulation temperature Pre-curing degree and pre-coating thickness With the corresponding target temperature Target curing degree and target film thickness The norm of the difference is summed with the sum of the difference in the maximum moving speed of the spray gun and the absolute value of the adjustment of the process parameter of the i-th spray gun, to construct a three-dimensional optimization function for the spray gun; an improved quantum genetic algorithm (QGA) is used for iteration to output the optimal control vector for multiple spray guns. ; Furthermore, the method for updating the model parameters is as follows: Using real-time collected input and output data, update the model parameters, changing the i-th model parameter before the update. plus the preset learning rate Same expectation operator Acting on the i-th measured output data Regarding the i-th input data and model parameters Likelihood function and output layer weights The product of the regularization coefficients and The sum of the products of divergences with respect to the model parameters The product of gradients yields the updated model parameters. .

[0005] Secondly, the present invention provides an adaptive adjustment system for spraying processes based on temperature measurement, which specifically includes the following modules: Fusion calibration module: presets multiple sensor parameters, calculates dynamic weights through signal-to-noise ratio and process adaptation coefficient, and calibrates sensors by fusing reference temperature to solve the problems of real-time interference and cross-process sensitivity differences; Temperature field construction module: It acquires environmental data of the target spraying process through calibrated sensors, generates a global temperature field thermogram by calculating and fusing the temperature field, and provides a three-dimensional reference by associating film thickness with environmental interference annotations; Thermodynamic analysis module: It collects the heat flux of the coated workpiece and the empty workpiece through heat flow sensors, calculates the net heat flux and real-time reaction enthalpy change, and constructs a three-dimensional curing degree model by combining the autocatalytic reaction kinetic model and film thickness correction term to calculate the real-time curing degree and reflect the curing progress of the coating film. Modeling and optimization module: Based on real-time curing degree fusion temperature spatiotemporal features and phase transition features, core features are extracted, and a digital twin model is established by combining FEA and CFD to pre-simulate the process adjustment effect. The optimal control vector is output by constructing a three-dimensional optimization function for the spray gun and improving the quantum genetic algorithm. Adaptive update module: Based on the optimal control vector, targeted actuator regulation is triggered, a coating thermodynamic parameter library is constructed, and the model parameters are updated and optimized globally through Bayesian deep neural network and federated learning.

[0006] The beneficial effects of this invention are: The sensor fusion results are dynamically calibrated based on the ratio of the square of the sensor signal-to-noise ratio and the process adaptation coefficient, while simultaneously addressing real-time interference and cross-process sensitivity differences. Data such as infrared two-dimensional temperature field, capacitor substrate temperature, acoustic spectrum signal, temperature and humidity, and film thickness are integrated to construct a multi-factor coupled fusion temperature field formula, achieving accurate characterization of the global temperature field. A film thickness correction term is added to the autocatalytic kinetic model, and the integration time is dynamically adjusted through net heat flux integration and real-time enthalpy change calculation to solve the problem of misjudgment of thick / thin film curing progress. A digital twin model of the spraying process is constructed by combining FPGA real-time preprocessing, FEA / CFD simulation, and NPU parallel computing. Global optimization of multi-spray gun parameters is achieved through a multi-gun three-dimensional optimization function and an improved quantum genetic algorithm. Precise control strategies are designed for different process anomalies, a dynamic parameter library is constructed, and distributed updates and global optimization of model parameters are achieved through Bayesian deep neural networks and federated learning. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0008] Figure 1 This is a flowchart of the steps of an adaptive adjustment method for spraying process based on temperature measurement provided in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the structure of an adaptive adjustment system for spraying process based on temperature measurement provided in Embodiment 2 of the present invention. Detailed Implementation

[0009] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0010] Example 1: As Figure 1 As shown in the figure, the adaptive adjustment method for spraying process based on temperature measurement provided by the present invention specifically includes the following steps: Step 1: Preset multi-sensor parameters, calculate dynamic weights through signal-to-noise ratio and process adaptation coefficient, and solve the problems of real-time interference and cross-process sensitivity differences by fusing reference temperature to calibrate sensors; In a specific embodiment, the target spraying process is selected through the human-computer interaction interface, and the system automatically retrieves the sensor preset parameters corresponding to the target spraying process from the distributed database; Specifically, the infrared thermal imager's temperature measurement range is preset to [80℃, 300℃], and the epoxy resin coating's transmission attenuation coefficient is preset to 500. The temperature measurement range of the capacitive temperature probe is preset to [25℃, 200℃], and the anti-static interference threshold is preset to 100kV; the acoustic emission sensor: the standard acoustic spectrum characteristic frequency reference for droplet impact on metal substrate is preset to 100kHz. Calculate the signal-to-noise ratio (SNR) of the infrared sensor and the capacitive sensor, and assign weights according to the square of the SNR to ensure that the sensor with high SNR has a high proportion in the fusion. Specifically, calculate the signal-to-noise ratio of the infrared sensor. Signal-to-noise ratio compared to capacitive sensors The weights are allocated based on the square ratio of the signal-to-noise ratios of the infrared sensor and the capacitive sensor, by using the real-time signal-to-noise ratio of the sensor at time t. The square of the value, divided by the real-time signal-to-noise ratio of the infrared sensor at time t. The square of the real-time signal-to-noise ratio of the capacitance sensor at time t The sum of squares is used to obtain dynamic weights through an adaptation fusion formula. ; in, For time steps, Sensor types include: infrared sensors and capacitive sensors, with dynamic weighting. Includes: dynamic weights of infrared sensors Dynamic weights of capacitive sensors And the dynamic weight of the infrared sensor Dynamic weights of capacitive sensors The sum of is 1; Real-time calibration and verification are performed based on the obtained dynamic weights to avoid calibration drift of the dynamic weights. The dynamic weights of the infrared sensor are fused and calculated using a weighted formula. Calibration temperature with infrared sensor Multiply to obtain the infrared sensor reference temperature, and then apply the dynamic weights of the capacitive sensor. Temperature measurement with capacitive sensor Multiplying these gives the reference temperature of the capacitive sensor; The reference temperature of the infrared sensor and the reference temperature of the capacitive sensor are summed to obtain the fused reference temperature. ; If the calibration temperature difference (the absolute value of the difference between the fused reference temperature and the standard reference temperature) is greater than the preset temperature difference threshold, sensor parameter fine-tuning is triggered. The calibration temperature difference is multiplied by the sensor's preset parameter fine-tuning coefficient, and then added to the correction baseline value to obtain the temperature correction coefficient. The base value is determined based on the sensor principle; Dynamic weighting addresses the signal-to-noise ratio (SNR) drop caused by real-time interference, while process adaptation coefficients resolve the sensitivity differences between sensors across different processes. Step 2: Obtain environmental data of the target spraying process through calibrated sensors, generate a global temperature field thermogram by calculating and fusing the temperature field, and provide a three-dimensional reference by associating film thickness with environmental interference annotations; Environmental data for the target coating process is acquired, including: two-dimensional temperature field, substrate temperature, acoustic spectrum signal, ambient temperature, relative humidity, and real-time film thickness. An infrared thermal imager is mounted at a 30° angle above the spray gun to avoid obstruction by spray droplets, and the two-dimensional temperature field of the workpiece surface is collected. ,in Here are the pixel coordinates of the workpiece surface. For collection timestamps; A capacitive probe is placed under the workpiece to directly measure the temperature of the substrate by penetrating the coating; An acoustic emission sensor is installed on the side of the workpiece to collect the acoustic spectrum signal of droplets impacting the workpiece. And low-frequency noise from the airflow is filtered out by a bandpass filter; Temperature and humidity sensor collects ambient temperature and relative humidity ; Laser thickness gauge collects real-time film thickness. Replace the fixed film thickness d value; Based on environmental data acquired by sensors, an adaptive fusion formula is constructed, which utilizes dynamic weights from infrared sensors. Two-dimensional temperature field corresponding to infrared sensor Combined with temperature correction factor Opposite number, real-time film thickness Infrared transmittance compensation coefficient for temperature and humidity The product of these is the product of exponential functions of exponents, and the fused temperature field is obtained through the environmental adaptive fusion formula. ; Each pixel coordinate Corresponding fusion temperature field By splicing the data according to spatial location, a temperature field thermal map of the entire workpiece is generated. The fused temperature field thermal map is updated according to a preset period, and environmental interference labels are added. For example, if the relative humidity... When the humidity is greater than 65%, the edge of the heat map is marked as a high humidity interference area. At the same time, the film thickness data is associated, and the area with the film thickness greater than the preset thickness threshold is marked as a thick film that needs to be monitored for curing, providing a three-dimensional reference of environment, film thickness and temperature. Step 3: Collect the heat flux of the coated workpiece and the empty workpiece using a heat flux sensor, calculate the net heat flux and real-time reaction enthalpy change, and construct a three-dimensional curing degree model by combining the autocatalytic reaction kinetic model and film thickness correction term to calculate the real-time curing degree, reflecting the coating curing progress. In a specific embodiment, a heat flux sensor is attached to the back of the workpiece, directly opposite the spraying area, to simultaneously collect two types of heat flux data: the heat flux of the coated workpiece and the heat flux of the uncoated workpiece. The heat flux of the coated workpiece includes the heat released during coating curing and the heat dissipation from the environment. A laser thickness gauge transmits the real-time film thickness to the enthalpy change calculation unit, and the heat flux sensor simultaneously records the heat flux data corresponding to the real-time film thickness. The heat flux of the coated workpiece at time t is then used to calculate the heat flux. Heat flux of empty workpiece at time t The difference is used to calculate the net heat flux of the spraying process using the net heat flux formula. ; By integrating the net heat flux over time and combining the sensor coverage area with the coating mass, the real-time enthalpy change is calculated. This real-time enthalpy change directly reflects the progress of the curing reaction. The coating mass at time t is then used to determine the enthalpy change. The reciprocal of the time Net heat flux The real-time enthalpy change is obtained by integrating the product of the heat flux sensor coverage area S. ; A self-catalytic reaction kinetic model was used to establish the relationship between the degree of curing and real-time temperature and time t. A three-dimensional curing degree model was constructed by adding a film thickness correction term to the kinetic differential equation. Specifically, by adding a film thickness correction term and using the kinetic differential equation, the pre-exponential factor A is multiplied by an exponential term, and the rate of change of curing degree with time is obtained through the kinetic differential equation. ; The real-time curing degree integral is calculated by numerical integration, and the real-time curing degree is obtained by subtracting the exponent term from 1. ; When the real-time film thickness is greater than the preset thickness threshold, the reaction rate increases, the heat dissipation inside the thick film is slow, the curing reaction is accelerated, and the integration time is extended, avoiding misjudgment of insufficient curing due to slow accumulation of enthalpy change inside the thick film; when the real-time film thickness is less than the preset thickness threshold, the reaction rate decreases and the integration time is shortened, avoiding missed judgment of over-curing due to rapid heat dissipation of the thin film; when the real-time film thickness is equal to the preset thickness threshold, the preset integration time remains unchanged. Step 4: Based on the real-time curing degree fusion temperature spatiotemporal characteristics and phase transition characteristics, extract the core features, combine FEA and CFD to establish a digital twin model to pre-simulate the process adjustment effect, and output the optimal control vector by constructing a three-dimensional optimization function for the spray gun and improving the quantum genetic algorithm. In a specific embodiment, temperature spatiotemporal features and phase transition features are integrated to extract core features, which include: global temperature standard deviation, edge-center temperature difference, temperature change rate, enthalpy change rate, curing degree deviation, and enthalpy change completion rate. Based on the obtained core features, a field-programmable gate array (FPGA) is used for real-time data preprocessing. A digital twin model of the spraying process is established based on finite element analysis (FEA) and computational fluid dynamics (CFD) to simulate the temperature field, degree of curing, and film thickness after process parameter adjustments. The specific model is as follows: Interferometric correction thermal field simulation, using gradient operator thermal conductivity of coating With temperature gradient The product is used for divergence calculation, plus the heat flux released during solidification. Subtract air density Specific heat capacity of air Interference speed of airflow from adjacent spray guns With temperature gradient The product of these two factors equals the coating density. Specific heat capacity of coating With temperature The product of the partial derivatives with respect to time; where, and Index the spray guns by number; Interference-corrected film thickness simulation, through coating flow rate Coating density Spraying time The product of these three factors, divided by the spray gun's movement speed Area of ​​the sprayed area The product of these two factors, multiplied by the correction term, yields the adjusted coating thickness. ; The pre-simulation results of multiple spray guns are obtained by running a twin model of M spray guns simultaneously using an NPU, and the pre-simulation temperature of each spray gun is output. Pre-curing degree and pre-coating thickness ; Based on the multi-spray gun pre-simulation results, minimize the pre-simulation temperature. Pre-curing degree and pre-coating thickness With the corresponding target temperature Target curing degree and target film thickness The norm of the difference is summed with the sum of the difference in the maximum moving speed of the spray gun and the absolute value of the adjustment of the process parameter of the i-th spray gun, and a three-dimensional optimization function for the spray gun is constructed. An improved quantum genetic algorithm (QGA) is used for iterative processing to output the optimal control vector for multiple guns. ; Step 5: Based on the optimal control vector, trigger targeted actuator regulation, construct a coating thermodynamic parameter library, and update model parameters through Bayesian deep neural network and federated learning to optimize the global model. In a specific embodiment, based on the obtained optimal control vector for the multi-gun, targeted actuator regulation is triggered. For example, the corresponding relationship is shown in Table 1 below: Table 1. Correspondence between actuator control methods Based on the completion of regulation, a thermodynamic parameter library for coatings is constructed. This library stores key thermodynamic parameters of commonly used spray coatings in a distributed database, categorized by coating type and applicable process. The key thermodynamic parameters include: Curing kinetic parameters: pre-indicator factor, activation energy, and reaction order; Thermophysical parameters: thermal conductivity, density, and specific heat capacity; Phase transition parameters: enthalpy change upon complete curing and coating transmission attenuation coefficient; The parameter library supports two update methods: the system automatically imports industry standard parameters; users add new coatings and fit the parameters through three verification experiments before adding them to the library to ensure parameter accuracy. Each spraying equipment calculates the prediction error of the local model in real time. If the prediction error is greater than or equal to the preset error threshold, the key thermodynamic parameters of the corresponding spraying equipment are marked as abnormal data, and the local model parameters are marked as parameters to be attenuated. The system utilizes real-time collected input and output data. The input data includes: coating temperature, degree of curing, coating thickness, ambient temperature, and relative humidity; the output data is the coating pass rate. Update the model parameters using a Bayesian deep neural network, changing the i-th model parameter before the update. plus the preset learning rate Same expectation operator Acting on the i-th measured output data Regarding the i-th input data and model parameters Likelihood function and output layer weights The product of the regularization coefficients and The sum of the products of divergences with respect to the model parameters The product of gradients yields the updated model parameters. ; After receiving the updated model parameters and anomaly markers from each client, the federated server performs weighted aggregation based on both data quality and data volume to generate global model parameters. The server then verifies the process compatibility of the global model parameters by substituting typical parameters from the coating parameter library and verifying that the prediction error meets the preset error range. Once the verification is successful, the parameters are sent to each client. The client updates its local model using the global model parameters and retains local anomaly data records for subsequent fault diagnosis.

[0011] Example 2: As Figure 2 As shown in the figure, the adaptive adjustment system for spraying process based on temperature measurement provided in this embodiment of the invention specifically includes the following modules: Fusion calibration module: presets multiple sensor parameters, calculates dynamic weights through signal-to-noise ratio and process adaptation coefficient, and calibrates sensors by fusing reference temperature to solve the problems of real-time interference and cross-process sensitivity differences; Temperature field construction module: It acquires environmental data of the target spraying process through calibrated sensors, generates a global temperature field thermogram by calculating and fusing the temperature field, and provides a three-dimensional reference by associating film thickness with environmental interference annotations; Thermodynamic analysis module: It collects the heat flux of the coated workpiece and the empty workpiece through heat flow sensors, calculates the net heat flux and real-time reaction enthalpy change, and constructs a three-dimensional curing degree model by combining the autocatalytic reaction kinetic model and film thickness correction term to calculate the real-time curing degree and reflect the curing progress of the coating film. Modeling and optimization module: Based on real-time curing degree fusion temperature spatiotemporal features and phase transition features, core features are extracted, and a digital twin model is established by combining FEA and CFD to pre-simulate the process adjustment effect. The optimal control vector is output by constructing a three-dimensional optimization function for the spray gun and improving the quantum genetic algorithm. Adaptive update module: Based on the optimal control vector, targeted actuator regulation is triggered, a coating thermodynamic parameter library is constructed, and the model parameters are updated and optimized globally through Bayesian deep neural network and federated learning.

[0012] The above provides a detailed description of one embodiment of the present invention, but the content described is only a preferred embodiment of the present invention and should not be considered as limiting the scope of the present invention. The above formulas are all dimensionless numerical calculations, and the formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world situation. The preset parameters in the formulas are set by those skilled in the art based on actual conditions and historical experience, and can be adjusted according to actual conditions. The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. All equivalent changes and improvements made in accordance with the scope of the present invention should still fall within the patent coverage of the present invention.

Claims

1. A method for adaptive adjustment of a spray process based on temperature measurements, characterized in that, The method comprises the following steps: presetting multi-sensor parameters, calculating dynamic weights through signal-to-noise ratio and process adaptation coefficients, calibrating sensors through fused reference temperature, solving real-time interference and cross-process sensitivity difference problems; acquiring environmental data of a target spraying process through the calibrated sensors, generating a global temperature field thermodynamic map through calculation and fusion of temperature fields, and providing a three-dimensional reference by associating film thickness and environmental interference labels; acquiring coating workpiece heat flux and empty workpiece heat flux through heat flow sensors, calculating net heat flux and real-time reaction enthalpy, combining a self-catalytic reaction kinetics model and a film thickness correction term to construct a three-dimensional solidification degree model to calculate real-time solidification degree and reflect coating film solidification progress; based on real-time solidification degree, fusing temperature spatiotemporal characteristics and phase change characteristics, extracting core features, establishing a digital twinborn model to simulate process adjustment effects, outputting an optimal control vector through a three-dimensional optimization function of a spray gun and an improved quantum genetic algorithm; based on the optimal control vector, triggering targeted actuator control, constructing a coating thermodynamic parameter library, and updating model parameters through a Bayesian deep neural network and federated learning to optimize the global.

2. The method of claim 1, wherein, The method for calibrating the sensors through the fused reference temperature is: Computing signal-to-noise ratio of infrared sensor Computing signal-to-noise ratio of capacitive sensor And according to the square ratio of the signal-to-noise ratio of the infrared sensor and the capacitive sensor, a dynamic weight is obtained Dynamic weight The infrared sensor dynamic weight The capacitive sensor dynamic weight And the sum of the infrared sensor dynamic weight And the capacitive sensor dynamic weight Is 1 The dynamic weight is obtained, real-time calibration verification is carried out based on the dynamic weight, and the infrared sensor dynamic weight is fused and calculated through a weighting formula The calibration temperature of the infrared sensor is multiplied by the reference temperature of the infrared sensor to obtain the reference temperature of the infrared sensor The dynamic weight of the capacitive sensor is multiplied by the measured temperature of the capacitive sensor The reference temperature of the capacitive sensor is obtained The reference temperature of the capacitive sensor is obtained summing the infrared sensor reference temperature with the capacitive sensor reference temperature to obtain a fused reference temperature ; if the calibration temperature difference is greater than a preset temperature difference threshold, a sensor parameter fine adjustment is triggered, the calibration temperature difference is multiplied by a preset parameter fine adjustment coefficient of the sensor, and then added to a correction base value to obtain a temperature correction coefficient.

3. The method of claim 1, wherein, The method for generating the global temperature field thermodynamic map is: To acquire environmental data for the target coating process, a capacitive probe is placed beneath the workpiece to directly measure the substrate temperature through the coating; an acoustic emission sensor is installed on the side of the workpiece to collect the acoustic spectrum signal of droplets impacting the workpiece. It also filters low-frequency noise from the airflow using a bandpass filter; a temperature and humidity sensor collects ambient temperature data. and relative humidity Laser thickness gauge collects real-time film thickness. Replace the fixed film thickness d value; Based on environmental data acquired by sensors, an adaptive fusion formula is constructed, which utilizes dynamic weights from infrared sensors. Two-dimensional temperature field corresponding to infrared sensor Combined with temperature correction factor Opposite number, real-time film thickness Infrared transmittance compensation coefficient for temperature and humidity The product of these factors is the product of exponential functions of exponents, and the fused temperature field is obtained through the environmental adaptive fusion formula. ; Set the coordinates of each pixel Corresponding fusion temperature field By splicing the data according to their spatial location, a temperature field thermogram of the entire workpiece is generated.

4. The method of adaptive adjustment of a spray process based on temperature measurements according to claim 1, characterized in that, The method for calculating the net heat flux is: The heat flow sensor is attached to the back of the workpiece and opposite the spraying area, and two types of heat flux data are synchronously collected, including: coated workpiece heat flux and empty workpiece heat flux, the coated workpiece heat flux includes paint curing heat release and environmental heat dissipation; the laser thickness gauge transmits real-time film thickness to the enthalpy calculation unit, and the heat flow sensor synchronously records the heat flux data under the corresponding real-time film thickness, and the net heat flux of the spraying process is obtained by the difference between the coated workpiece heat flux at time t and the empty workpiece heat flux at time t . ; time integration is performed on the net heat flux, and real-time reaction enthalpy values are calculated in combination with sensor coverage area and coating mass.

5. A method of adaptive adjustment of a spray process based on temperature measurements according to claim 4, characterized in that, The method for acquiring the real-time reaction enthalpy values is: By integrating the net heat flux over time and combining the sensor coverage area with the coating mass, the real-time enthalpy change is calculated. This real-time enthalpy change directly reflects the progress of the curing reaction. The coating mass at time t is then used to determine the enthalpy change. The reciprocal of the time Net heat flux The real-time enthalpy change is obtained by integrating the product of the heat flux sensor coverage area S. .

6. The method of adaptive adjustment of a spray process based on temperature measurements according to claim 1, characterized in that, The method for constructing the three-dimensional solidification degree model is: By adding a film thickness correction term to the kinetic differential equation, the pre-exponential factor A is multiplied by an exponential term, which gives the rate of change of the degree of cure with time by the kinetic differential equation ; The real-time degree of cure integral is calculated by numerical integration, and the real-time degree of cure is obtained by subtracting the exponential term from 1 .

7. The method of adaptive adjustment of a spray process based on temperature measurements according to claim 1, characterized in that, The method for simulating the process adjustment effects is: fusing temperature spatiotemporal characteristics and phase change characteristics, extracting core features, based on the obtained core features, simulating the temperature field, solidification degree and film thickness after adjustment of the process parameters, and the specific model is as follows: interference correction thermal field simulation by gradient operator on paint thermal conductivity with temperature gradient divergence, plus solidification exothermic flux minus air density , air specific heat , adjacent torch gas flow interference velocity with temperature gradient divergence, equals paint density , paint specific heat times temperature partial derivative with respect to time; wherein, and are torch number indices; Interference-corrected film thickness simulation, through coating flow rate Coating density Spraying time The product of these three factors, divided by the spray gun's movement speed Area of ​​the sprayed area The product of these two factors, multiplied by the correction term, yields the adjusted coating thickness. ; the twinborn model of M spray guns is simultaneously run through an NPU to obtain multi-spray gun simulation results, and the simulation temperature, simulation solidification degree and simulation coating thickness of each spray gun are output.

8. The method of adaptive adjustment of a spray process based on temperature measurements according to claim 1, characterized in that, The method for outputting the optimal control vector is: outputting a pre-play temperature of each spray gun , a pre-play curing degree , and a pre-play coating thickness , minimizing the difference between the pre-play temperature , the pre-play curing degree , and the pre-play coating thickness and the corresponding target temperature , target curing degree , and target film thickness , and the maximum movement speed difference of the spray gun .

9. The method of adaptive adjustment of a spray process based on temperature measurements according to claim 1, characterized in that, The method for updating the model parameters is: The model parameters are updated using the input data and the output data collected in real time, and the product of the likelihood function of the ith measured output data and the output layer weight , plus a preset learning rate , and the expectation operator , acting on the ith measured output data , with respect to the ith input data , and the model parameters , and the product of the output layer weight , and the regularization coefficient , and the product of the divergence , with respect to the model parameters , and the product of the gradient, to obtain the updated model parameters .

10. A temperature measurement based spray process adaptive adjustment system for performing the adjustment method of any of the preceding claims 1 to 9, characterized in that, It comprises: a calibration module: presetting multi-sensor parameters, calculating dynamic weights through signal-to-noise ratio and process adaptation coefficients, calibrating sensors through fused reference temperature, solving real-time interference and cross-process sensitivity difference problems; a temperature field construction module: acquiring environmental data of a target spraying process through the calibrated sensors, generating a global temperature field thermodynamic map through calculation and fusion of temperature fields, and providing a three-dimensional reference by associating film thickness and environmental interference labels; a thermodynamic analysis module: acquiring coating workpiece heat flux and empty workpiece heat flux through heat flow sensors, calculating net heat flux and real-time reaction enthalpy, combining a self-catalytic reaction kinetics model and a film thickness correction term to construct a three-dimensional solidification degree model to calculate real-time solidification degree and reflect coating film solidification progress; Modeling optimization module: based on real-time solidification degree, integrate the space-time characteristics and phase change characteristics, extract the core features, combine FEA and CFD to establish digital twin model to simulate the process adjustment effect, and output the optimal control vector through building three-dimensional optimization function of spray gun and improved quantum genetic algorithm; Adaptive update module: based on the optimal control vector, trigger targeted actuator control, build a thermodynamic parameter library of coating, update model parameters through Bayesian deep neural network and federated learning to optimize the global.

Images