Method for controlling the thickness of a sprayed coating on an aluminum veneer

Abstract

This invention discloses a method for controlling the uniformity of coating thickness in aluminum single-panel spraying, relating to the field of aluminum single-panel curved surface spraying technology. The method includes: constructing a set of historical triangular spraying regions based on historical three-dimensional data of the aluminum single-panel curved surface and calculating their surface geometric features; training a coating thickness analysis model by combining historical surface geometric features with spraying process parameters; acquiring real-time triangular regions and their surface geometric features during real-time spraying; inputting the real-time features and process parameters into the model to predict the real-time coating thickness; further predicting the coating thickness of the next area to be sprayed and calculating its deviation from the target thickness; generating process parameter adjustment instructions based on the deviation and the area's surface geometric features, and adjusting parameters such as spray gun speed and flow rate in real-time during spraying. This invention, by associating surface geometry and spraying process through a machine learning model, achieves accurate prediction and dynamic compensation control of coating thickness on complex curved surfaces, effectively improving coating uniformity.

Description

Technical Field

[0001] This invention relates to the field of aluminum single-panel curved surface spraying technology, specifically a method for controlling the uniformity of coating thickness in aluminum single-panel spraying. Background Technology

[0002] In aluminum panel curved surface spraying operations, the uniformity of coating thickness is a key indicator for measuring spraying quality. Currently, for aluminum panels with complex three-dimensional curved surfaces, coating thickness control mainly relies on the operator's experience or preset fixed spraying parameters. However, due to the differences in geometric features (such as local curvature and normal angle) at various points on the aluminum panel curved surface, using fixed spraying process parameters (such as spray gun movement speed, paint flow rate, and spray gun axis direction) can easily lead to uneven coating thickness in different areas of the curved surface, affecting the product's anti-corrosion and decorative properties.

[0003] Existing technologies lack a control method capable of dynamically linking the real-time geometric features of aluminum single-panel curved surfaces with spraying process parameters, and accordingly making accurate predictions and online adjustments. Therefore, how to accurately predict the coating thickness of any spraying area on a complex curved surface, and how to adjust the spraying process parameters in real-time and adaptively based on the prediction results to ensure the uniformity of the coating thickness across the entire aluminum single-panel curved surface, is a technical problem that needs to be solved in this field. Summary of the Invention

[0004] The purpose of this invention is to provide a method for controlling the uniformity of coating thickness in aluminum single-panel spraying, so as to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for uniformly controlling the thickness of aluminum single-panel spray coating. The methods include: Step S1: Collect historical spraying data of the aluminum single-panel curved surface; the historical spraying data includes: historical three-dimensional topography data, historical spraying process parameters, and historical coating thickness; based on the historical three-dimensional topography data, calculate the historical surface geometric features of the historical sprayed area, the historical surface geometric features include historical local curvature and historical normal tilt angle; the historical spraying process parameters include the historical moving speed of the spray gun, historical paint flow rate, and historical attitude angle; Step S2: Combine the calculated historical surface geometric features and historical spraying process parameters into a historical input feature vector, and use the historical coating thickness as the target output to perform supervised training on the selected machine learning model; iteratively adjust the model parameters through optimization algorithms so that the model learns and establishes a mapping relationship from the historical input feature vector to the coating thickness, thereby obtaining the coating thickness analysis model; Step S3: During the real-time spraying process of the aluminum single panel curved surface, acquire the real-time three-dimensional topography data and real-time spraying process parameters of the current spraying area; based on the real-time three-dimensional topography data, calculate the real-time surface geometric features of the current spraying area, including the real-time local curvature and the real-time normal tilt angle. Step S4: Obtain the surface geometric features of the next area to be sprayed; combine the surface geometric features of the next area to be sprayed with the real-time spraying process parameters of the current area to be sprayed into a prediction input feature vector, and input it into the coating thickness analysis model to obtain the predicted coating thickness of the next area to be sprayed under the current spraying process parameters; calculate the predicted thickness deviation between the predicted coating thickness and the target coating thickness. Step S5: Based on the predicted thickness deviation of the next area to be sprayed, generate a real-time adjustment command for the spraying process parameters when applying the coating to the next area; according to the adjustment command, adjust the moving speed of the spray gun, the paint flow rate, and the spray gun attitude angle in real time.

[0006] Furthermore, step S1 includes: Step S1-1: Using the surface of the aluminum single-panel spraying workbench as the reference plane, select a fixed corner point of the surface as the origin O of the coordinate system, define the two edges of the surface that pass through the origin and are perpendicular to each other as the X-axis and Y-axis respectively, and define the direction that extends upward perpendicular to the surface as the Z-axis to establish a three-dimensional coordinate system. Step S1-2: In the three-dimensional coordinate system, acquire historical point cloud data of the aluminum single-panel surface; perform denoising and triangulation processing on the historical point cloud data, discretizing the historical spraying area into several triangular spraying areas, obtaining a set of historical triangular spraying areas F={F k |k=1,2,…,n};where, the k-th historical triangular spraying region F k Composed of three vertices V k,1 V k,2 V k,3 The three-dimensional coordinates are uniquely determined; Step S1-3: Obtain the spray gun's spray area F in the kth historical triangular spraying region. k The historical spraying process parameters include: historical moving speed v. hist (k), Historical paint flow rate Q hist (k) and the historical direction vector n of the spray gun axis in the three-dimensional coordinate system. gun (k); After the spraying operation is completed, obtain the k-th historical triangular spraying area F. k Historical coating thickness T at the center position hist (k); Step S1-4: Based on the set of historical triangular spraying regions F obtained in step S1-2, firstly, according to the k-th historical triangular spraying region F... k The three vertices V k,1 V k,2 V k,3 Calculate the three-dimensional coordinates of the k-th historical triangular spraying area F. k Historical unit normal vector n surf (k); secondly, combined with the kth historical triangular spraying area F of the spray gun obtained in steps S1-3. k Historical direction vector n gun (k), through formula θ hist (k)=arccos(|n surf (k)·n gun (k)|) Calculate the k-th historical triangular spraying area F k Historical normal tilt angle θ hist (k), where · represents the vector dot product operation; finally, calculate the k-th historical triangular spraying region F. k Historical curvature C hist (k): Based on the kth historical triangular spraying area F k The three vertices V k,1 V k,2 V k,3 The three-dimensional coordinates and their relationship with the k-th historical triangular spraying area F k The surface information of adjacent historical triangular spraying areas is used to calculate F for the k-th historical triangular spraying area. k The curvature of the three vertices is taken, and the arithmetic mean of the curvatures of the three vertices is taken as the k-th historical triangular spraying region F. k Historical local curvature characterization value C hist (k), the kth historical triangular spraying area F k Historical normal tilt angle θ hist (k) and the kth historical triangular spraying area F k Historical curvature characterization value C hist (k) is the kth historical triangular spraying area F k Historical surface geometric features.

[0007] Furthermore, step S2 includes: Step S2-1: For the k-th historical triangular spraying region F in the set of historical triangular spraying regions F k The k-th historical triangular spraying area F k The historical surface geometric features and historical spraying process parameters are combined in the following order to form a seven-dimensional column vector, which serves as the historical input feature vector X. hist (k): X hist(k)=[θ hist (k),C hist (k),v hist (k),Q hist (k),n gun_x (k),n gun_y (k),n gun_z (k)] T , where θ hist (k) represents the k-th historical triangular spraying area F. k Historical normal tilt angle, C hist (k) represents the k-th historical triangular spraying area F. k Historical curvature representation value, v hist (k) represents the k-th historical triangular spraying area F. k Historical movement speed, Q hist (k) represents the k-th historical triangular spraying area F. k Historical paint flow rate, n gun_x (k),n gun_y (k),n gun_z (k) represents the k-th historical triangular spraying area F of the spray gun. k Historical direction vector n gun (k) Components in the X, Y, and Z axes of the three-dimensional coordinate system; Step S2-2: Use the historical input feature vectors corresponding to all historical triangular spraying areas as input features and the historical coating thickness corresponding to all historical triangular spraying areas as target output to form a historical spraying training dataset. Step S2-3: Supervised training of the selected machine learning model is performed using the historical spraying training dataset. The training process aims to minimize the error between the estimated coating thickness calculated by the machine learning model based on the historical input feature vector and the corresponding historical coating thickness. The internal weight parameters of the machine learning model are iteratively adjusted through an optimization algorithm so that the model learns and establishes a nonlinear mapping relationship from the seven-dimensional historical input feature vector to the coating thickness value. After the training converges, the trained model is used as the coating thickness analysis model.

[0008] Furthermore, step S3 includes: Step S3-1: During real-time spraying, based on a three-dimensional coordinate system defined in the same way as historical spraying, acquire real-time point cloud data of the real-time spraying area; perform denoising and triangulation processing on the real-time point cloud data, and extract the set of real-time triangular spraying areas G={G p |p=1,2,…,m}, where the real-time triangular spraying area G p Composed of three vertices U p,1 U p,2 Up,3 The three-dimensional coordinates are uniquely determined; Step S3-2: Obtain the spray gun's position in the real-time triangular spraying area G p The corresponding real-time spraying process parameters include the real-time moving speed v. real (p), Real-time paint flow rate Q real (p) and the real-time direction vector n of the spray gun axis in the three-dimensional coordinate system. gun_real (p); Step S3-3: Based on the real-time triangular spraying area set G, firstly, according to the real-time triangular spraying area G... p Calculate the coordinates of the three vertices of the real-time triangular spraying area G. p The real-time unit normal vector n surf_real (p); Secondly, combined with the real-time triangular spraying area G p Real-time direction vector n gun_real (p), through formula θ real (p)=arccos(|n surf_real (p)·n gun_real (p)|) Calculate the real-time triangular spraying area G p Real-time normal tilt angle θ real (p), where · represents the vector dot product operation; finally, based on the real-time triangular spraying area G p The three-dimensional coordinates of the three vertices and their relationship with the real-time triangular spraying area G p Using the geometric information of adjacent curved surfaces, calculate the real-time triangular spraying region G. p The curvature of the three vertices is used, and the arithmetic mean of the three vertices is taken as the real-time triangular spraying area G. p Real-time curvature characterization value C real (p); The real-time normal tilt angle θ real (p) and the real-time curvature characterization value C real (p) Together serve as the real-time triangular spraying area G p Real-time surface geometry features.

[0009] Furthermore, step S4 includes: Step S4-1: Based on the preset spraying path and real-time spraying progress, determine the next triangular spraying area G to be sprayed in the real-time spraying area. q ; Obtain the next triangular area to be sprayed, G q Surface geometric features, including the normal tilt angle θ pre (q) and curvature characterization value C pre (q); Step S4-2: Following the same feature vector construction rules as in step S2-1, construct the next triangular spraying area G to be sprayed.q Surface geometry features and real-time triangular spraying area G p The real-time spraying process parameters are combined into a seven-dimensional column vector, which serves as the input feature vector X. pre (q): X pre (q)=[θ pre (q),C pre (q),v real (p),Q real (p),n gun_real_x (p),n gun_real_y (p),n gun_real_z (p)] T , where θ pre (q) represents the next triangular area to be sprayed, G. q normal tilt angle, C pre (q) represents the next triangular area to be sprayed, G. q The curvature characterization value, v real (p) represents the current spraying area G. p Real-time movement speed, Q real (p) represents the real-time triangular spraying area G. p Real-time paint flow rate, n gun_real_x (p),n gun_real_y (p),n gun_real_z (p) represent the spray gun's position in the real-time triangular spraying area G. p Real-time direction vector n gun_real (p) Components in the X, Y, and Z axes of the three-dimensional coordinate system; Step S4-3: Convert the input feature vector X pre (q) Input is given to the coating thickness analysis model, and the coating thickness analysis model is based on the input feature vector X. pre (q) performs the calculation and outputs a coating thickness value, which is used as the next area to be sprayed, G. q In the real-time triangular spraying area G p Coating thickness T under real-time spraying process parameters pre (q); Calculate the next triangular spraying area G to be sprayed. q Thickness deviation ΔT pre (q)=T target -T pre (q).

[0010] Furthermore, step S5 includes: Step S5-1: Based on the next triangular spraying area G to be sprayed q Thickness deviation ΔT pre (q), by adjusting the policy function f adjust Generate the next triangular spraying area G to be sprayed.q The spraying process parameter adjustment command; the adjustment strategy function f adjust The specific form is: Δv(q) = K v ·(1+α·|C pre (q)|)·ΔT pre (q); ΔQ(q) = K Q ·(1+β·|cos(θ pre (q))|)·ΔT pre (q); Δn gun (q)=K n ·(n surf_pre (q)-n gun_real (p))·ΔT pre (q); among them, Δv(q), ΔQ(q), Δn gun (q) represents the adjustment amount of the spray gun moving speed, the paint flow rate, and the spray gun axis direction vector, respectively; K v K Q K n These are the preset speed, flow rate, and attitude adjustment gain coefficients, respectively; α and β are the weighting coefficients for the influence of curvature and angle; C pre (q) represents the next triangular area to be sprayed, G. q The curvature characterization value; n surf_pre (q) represents the next triangular area to be sprayed, G. q The unit normal vector; n gun_real (p) represents the spray gun's position in the real-time triangular spraying area G. p Real-time direction vector; K v ·(1+α·|C pre (q)|) is the first coefficient, K Q ·(1+β·|cos(θ pre (q))|) is the second coefficient; Step S5-2: Based on the adjustment instructions generated in step S5-1, spray the next triangular spraying area G to be sprayed. q Meanwhile, the spraying equipment is adjusted in real time: the moving speed of the spray gun is updated to v. real (q)=v real (p)+Δv(q), update the paint flow rate to Q. real (q)=Q real (p)+ΔQ(q), and update the axis direction vector of the spray gun to n. gun_real (q)=n gun_real (p)+Δn gun (q); Step S5-3: After completing the spraying of the next triangular area G to be sprayed qAfter adjusting the spraying process parameters and executing the spraying, the next triangular spraying area G to be sprayed will be... q Update to the new real-time spraying area, and based on the updated spraying path, redetermine the next triangular spraying area G to be sprayed. q The triangular area to be sprayed is then returned to step S4-1. Based on the new real-time spraying process parameters, steps S4 to S5 are re-executed to complete the adjustment of the spraying process parameters.

[0011] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention establishes a nonlinear mapping relationship between the surface geometry of the spraying area, spraying process parameters, and the final coating thickness by constructing a historical dataset and training a machine learning model. In actual spraying, the model can accurately predict the coating thickness based on the real-time geometric features of the area to be sprayed and the current process parameters, providing a reliable decision-making basis for subsequent process adjustments.

[0012] 2. Based on the predicted thickness of the current area, this invention further predicts the thickness of the next area to be sprayed and calculates the deviation from the target thickness. Based on this deviation and the geometric characteristics of the area, the system can generate adjustment instructions for spraying process parameters (such as speed, flow rate, and direction) in advance, and execute the adjustments in real time when the spray gun reaches the area. This prediction-based feedforward control method can proactively and promptly compensate for thickness unevenness caused by changes in surface geometry.

[0013] 3. The entire control process, from surface geometric feature extraction and thickness prediction to process parameter decision-making and adjustment, forms a complete closed-loop automatic control system. This method reduces reliance on operator experience, can automatically adapt to the complex changes of different surfaces, and significantly improves the uniformity and reliability of the coating uniformity of aluminum single-panel curved surfaces. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating the method for controlling the uniformity of coating thickness in aluminum single-panel spraying according to the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Example: Figure 1 As shown, the present invention provides a technical solution: a method for controlling the uniformity of coating thickness in aluminum single-panel spraying. The methods include: Taking an aluminum single-panel spraying production line as an application scenario, the production line adopts a robotic spraying system, and the spraying object is an aluminum single-panel workpiece with a complex curved surface.

[0017] Step S1: Collect historical spraying data of the aluminum single panel curved surface; historical spraying data includes: historical three-dimensional topography data, historical spraying process parameters and historical coating thickness; based on the historical three-dimensional topography data, calculate the historical surface geometric features of the historical sprayed area, including historical local curvature and historical normal tilt angle; historical spraying process parameters include the historical moving speed of the spray gun, historical paint flow rate and historical attitude angle; Step S1-1: Establish a three-dimensional coordinate system with the lower left corner of the platform as the origin O, the long side as the X-axis, the short side as the Y-axis, and the vertical upward direction of the platform as the Z-axis. Step S1-2: Select an S-shaped curved aluminum panel that has been sprayed and meets quality standards as a sample. Obtain its surface point cloud data using a 3D laser scanner, totaling 6238 points. After statistical filtering to remove outlier noise points, use the Delaunay triangulation algorithm to mesh the surface, obtaining a discretized representation of the curved surface composed of 1205 triangular facets, i.e., the set of historical triangular sprayed regions F={F1,F2,…,F…}. 1205}; Take the k=25th historical triangular spraying area F in the set 25 The coordinates (unit: mm) of its three vertices in the coordinate system are measured as follows: V 25,1 =(356.2,245.7,32.8); V 25,2 =(357.9,247.3,32.5); V 25,3 =(357.0,244.5,32.6); Steps S1-3: Spray the 25th historical triangular spraying area F 25 The process parameters recorded at that time are: historical moving speed v hist (25) = 320 mm / s, historical paint flow rate Q hist (25) = 165 mL / min, spray gun historical direction vector n gun (25) = [0.087, 0.174, 0.981] (unit vector); After spraying is completed, the 25th historical triangular spraying area F is measured. 25 The historical coating thickness at the center is T. hist (25) = 28.5 μm; Steps S1-4: Calculate the 25th historical triangular spraying area F 25 Surface geometric features: Calculate the historical unit normal vector n surf (10): Vector A=V 25,2 -V 25,1 =(357.9-356.2,247.3-245.7,32.5-32.8)=(1.7,1.6,-0.3); Vector B=V 25,3 -V 25,1 =(357.0-356.2,244.5-245.7,32.6-32.8)=(0.8,-1.2,-0.2); Calculate the normal vector: n = A × B; n x =A y ×B z A z ×B y =1.6×(-0.2) (-0.3)×(-1.2)=-0.32 0.36 = -0.68; n y =A z ×B x A x ×B z =(-0.3)×0.8 1.7 × (-0.2) = -0.24 + 0.34 = 0.1; n z =A x ×B y A y ×B x =1.7×(-1.2) 1.6 × 0.8 = -2.04 1.28 = -3.32; Therefore, n = (-0.68, 0.10, -3.32); Length ; Historical unit normal vector n surf (25)=n / |n|≈(-0.2006,0.0295,-0.9793); Calculate the historical normal tilt angle θ hist (25): Calculate the absolute value of the dot product between the spray gun direction vector and the surface normal vector: |n surf (25)·n gun (25)|=|-0.01745+0.00513 0.96059|=0.97291; Calculate the included angle: θ hist (25)=arccos(0.97291)≈13.4°; Calculate the historical curvature characterization value C hist (25): Select the 25th historical triangular spraying area F 25 For adjacent historical triangular spraying areas, the Gaussian curvature of each vertex is calculated based on the vertex coordinates; V is then calculated. 25,1 V 25,2 V 25,3 The curvatures are respectively: 0.018mm -1 0.02mm -1 0.017mm -1 ; C hist (25)=(0.018+0.02+0.017) / 3=0.0183mm -1 .

[0018] Step S2: Combine the calculated historical surface geometric features and historical spraying process parameters into a historical input feature vector, and use the historical coating thickness as the target output to supervise the training of the selected machine learning model; iteratively adjust the model parameters through optimization algorithms so that the model learns and establishes a mapping relationship from the historical input feature vector to the coating thickness, thereby obtaining the coating thickness analysis model; Step S2-1: Construct the 25th historical triangular spraying area F 25 Historical input feature vectors: X hist (25)=[θ hist (25),C hist (25),v hist (25),Q hist (25),n gun_x (25),n gun_y (25),n gun_z (5)] T =[13.4,0.0183,320,165,0.087,0.174,0.981] T ; Step S2-2: For all 1205 faces in the historical triangular spraying area set F, repeat steps S1 and S2-1 to obtain 1205 sets of data pairs {X}. hist (k),T hist (k)}(k=1to1205) constitutes a complete historical spraying training dataset.

[0019] Step S2-3: Select the support vector regression model as the machine learning model, use all 1205 sets of data as the training set, set the model parameters, and the training objective is to minimize the mean square error between the model's predicted thickness value and the actual historical thickness value. Iteratively adjust the model parameters through the optimization algorithm so that the model learns the nonlinear mapping relationship from the historical input feature vector of "surface geometric features + process parameters" to "coating thickness". The training process continues until the loss function converges to below the stable threshold. At this time, the model is considered to have completed training and is saved as "coating thickness analysis model".

[0020] Step S3: During the real-time spraying process of the aluminum single panel curved surface, acquire the real-time three-dimensional topography data and real-time spraying process parameters of the current spraying area; based on the real-time three-dimensional topography data, calculate the real-time surface geometric features of the current spraying area, including the real-time local curvature and the real-time normal tilt angle. Step S3-1: Perform online laser scanning on the real-time sprayed aluminum panel to acquire real-time point cloud data. After processing through the same procedure, extract the real-time triangular sprayed region set G={G1,G2,…,G…}. 1380 There are a total of 1380 real-time triangular spraying areas. The spraying robot is currently spraying the p=100th real-time triangular spraying area G. 100 Its vertex coordinates (mm) are: U 100,1 =(512.4,328.6,29.7), U 100,2 =(514.1,330.2,29.3), U 100,3 =(513.2,327.4,29.5); Step S3-2: Obtain the real-time triangular spraying area G from the robot's real-time data bus. 100 Process parameters at that time: real-time moving speed v real (100) = 310 mm / s, real-time paint flow rate Q real (100) = 160 mL / min, spray gun real-time direction vector (unit vector) n gun_real (100) = [0.122, 0.122, 0.985]; Step S3-3: Calculate the real-time triangular spraying area G 100 Real-time surface geometry features: Calculate the real-time unit normal vector n surf_real (100): Vector A real =U 100,2 U 100,1 =(1.7,1.6,-0.4); Vector B real =U 100,3 U100,1 =(0.8,-1.2,-0.2); Normal vector n real =A real ×B real ; n real_x =1.6×(-0.2)-(-0.4)×(-1.2)=-0.32-0.48=-0.8; n real_y =(-0.4)×0.8-1.7×(-0.2)=-0.32+0.34=0.02; n real_z =1.7×(-1.2)-1.6×0.8=-2.04-1.28=-3.32; n real =(-0.8, 0.02, -3.32); Length ; n surf_real (100)=n real / |n real |≈(-0.2343,0.00586,-0.9721); Calculate the real-time normal tilt angle θ real (100): |n surf_real (100)·n gun_real (100)|=|-0.02858+0.000715-0.95752|=0.98539; θ real (100)=arccos(0.98539)≈9.8°; Calculate the real-time curvature characterization value C real (100): Calculate the vertex U 100,1 U 100,2 U 100,3 The real-time curvature is 0.015mm. -1 0.016mm -1 0.014mm -1 ; C real (100)=(0.015+0.016+0.014) / 3=0.015mm -1 .

[0021] Step S4: Obtain the surface geometric features of the next area to be sprayed; combine the surface geometric features of the next area to be sprayed with the real-time spraying process parameters of the current area to be sprayed into a prediction input feature vector, and input it into the coating thickness analysis model to obtain the predicted coating thickness of the next area to be sprayed under the current spraying process parameters; calculate the predicted thickness deviation between the predicted coating thickness and the target coating thickness. Step S4-1: According to the preset spraying trajectory, the next triangular area to be sprayed is G. 101 By preprocessing its mesh data, its surface geometric features are obtained: normal tilt angle θ. pre (101) = 16.2°; curvature characterization value C pre (101) = 0.022 mm -1 Unit normal vector n surf_pre (101) = [-0.258, 0, -0.966]; Step S4-2: Construct the next triangular region G to be sprayed 101 The input feature vector X pre (101), here the real-time triangular spraying area G is used 100 Real-time process parameters X pre (101)=[θ pre (101),C pre (101),v real (100),Q real (100),n gun_real_x (100),n gun_real_y (100),n gun_real_z (100)] T =[16.2,0.022,310,160,0.122,0.122,0.985] T ; Step S4-3: Place X pre (101) Input the coating thickness analysis model to obtain the next triangular region G to be sprayed. 101 Coating thickness under real-time process parameters: T pre (101) = 25.1 μm; and the target coating thickness T is set. target =26μm; Calculate the next triangular region to be sprayed, G. 101 Thickness deviation under real-time process parameters: ΔT pre (101)=T target T pre (101) = 26.0 25.1 = 0.9 μm.

[0022] Step S5: Based on the predicted thickness deviation of the next area to be sprayed, generate in real time an adjustment command for the spraying process parameters when applying the next area to be sprayed; according to the adjustment command, adjust the moving speed of the spray gun, the paint flow rate, and the spray gun attitude angle in real time. Step S5-1: Based on the prediction deviation, generate parameter adjustment instructions through the adjustment strategy function; set the adjustment strategy parameters: speed adjustment gain coefficient: K v =-0.8(mm / s) / μm, the negative sign indicates reverse adjustment: when the prediction is too thin, the speed needs to be reduced to increase the thickness; flow rate adjustment gain coefficient: K Q =0.5(mL / min) / μm, positive sign indicates forward adjustment: when the predicted thickness is too thin, the flow rate needs to be increased to thicken it; attitude adjustment gain coefficient: K n =0.15; Curvature influence weight: α=8; Angle influence weight: β=0.3; Calculate the adjustment amount of each process parameter: The spray gun movement speed adjustment amount Δv(101): Δv(101) = K v ·(1+α·|C pre (101)|)·ΔT pre (101)=(-0.8)×(1+8×0.022)×0.9≈-0.846mm / s; Paint flow rate adjustment ΔQ(101): ΔQ(101) = K Q ·(1+β·|cos(θ pre (101))|)·ΔT pre (101)=0.5×(1+0.3×|cos(16.2°)|)×0.9≈0.580mL / min; spray gun axis direction vector adjustment amount Δn gun (101): Calculate the direction vector difference: n surf_pre (101) n gun_real (100) = [-0.380, -0.122, -1.951] Δn gun (101)=K n ·(n surf_pre (101) n gun_real (100))·ΔT pre (101) = 0.15 × [-0.380, -0.122, -1.951] × 0.9 ≈ [-0.0513, -0.0165, -0.2634]; Step S5-2: Spray the next triangular area G to be sprayed. 101 When using this parameter, apply the following adjusted parameters: Update movement speed: v real (101)=v real (100)+Δv(101)=310+(-0.846)=309.154mm / s; Update paint flow rate: Q real (101)=Q real (100)+ΔQ(101)=160+0.580=160.58mL / min; Update and normalize the spray gun direction vector: Calculate the original update vector: n gun_real (101)=n gun_real (100)+Δn gun (101)=[0.122,0.122,0.985]+[-0.0513,-0.0165,-0.2634]=[0.0707,0.1055,0.7216]; Normalization: Module length: Unit vector: n gun_real (101)≈[0.0707 / 0.7327,0.1055 / 0.7327,0.7216 / 0.7327]≈[0.0965,0.144,0.9849]; Step S5-3: Complete the spraying of the next triangular area G. 101 After spraying, the system status is updated: a new real-time spraying triangle area G. 101 The process parameters are the adjusted values ​​(speed 309.154 mm / s, flow rate 160.58 mL / min, spray gun direction [0.0965, 0.144, 0.9849]); subsequently, the system locks the new next triangular area to be sprayed, G, according to the path planning. 102 It will then automatically jump to step S4-1 to begin painting the next triangular area G. 102 Thickness prediction and parameter pre-adjustment are performed.

[0023] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for controlling the uniformity of coating thickness in aluminum single-panel spraying, characterized in that: Includes the following steps: Historical 3D data of aluminum single-panel curved surfaces are obtained to construct a set of historical triangular spraying areas; historical spraying process parameters and historical coating thickness of each historical triangular spraying area are collected, and historical surface geometric features are calculated. The historical surface geometric features and the historical spraying process parameters are combined to construct a historical input feature vector, and a coating thickness analysis model is trained with the historical coating thickness as the target. During real-time spraying, the set of real-time triangular spraying areas and real-time spraying process parameters are acquired, and the real-time surface geometric features are calculated. Obtain the surface geometric features of the next triangular area to be sprayed, combine the surface geometric features with the real-time spraying process parameters to construct an input feature vector, input the coating thickness analysis model, obtain the coating thickness, and calculate the thickness deviation. Based on the thickness deviation and the surface geometry of the next triangular area to be sprayed, a spraying process parameter adjustment command is generated. According to the spraying process parameter adjustment command, the moving speed of the spray gun, the paint flow rate and the attitude angle are adjusted. After the spraying process parameters of the spray gun are adjusted, the next area to be sprayed is sprayed.

2. The method for controlling the uniformity of coating thickness in aluminum single-panel spraying according to claim 1, characterized in that: Obtaining historical surface geometry features includes: A three-dimensional coordinate system for the spraying operation is established. Historical point cloud data of the aluminum single-panel surface is acquired and processed to discretize the aluminum single-panel surface into a set of historical triangular spraying regions composed of triangular units. The historical spraying process parameters include the moving speed of the spray gun in each region, the paint flow rate, and the spray gun axis direction vector. Based on the vertex coordinates of each historical triangular spraying region, the historical unit normal vector and the historical normal tilt angle are calculated. Combined with the information of the adjacent regions of each historical triangular spraying region, the historical curvature characterization value of each historical triangular spraying region is calculated. The historical normal tilt angle and historical curvature characterization value are used as historical surface geometric features.

3. The method for controlling the uniformity of coating thickness in aluminum single-panel spraying according to claim 1, characterized in that: The training model for coating thickness analysis includes: The historical surface geometric features and historical spraying process parameters of each historical triangular spraying area are combined into a historical input feature vector; a historical spraying training dataset is established based on the historical input feature vectors of all historical triangular spraying areas and the corresponding historical coating thickness; a machine learning model is trained through supervised learning using the historical spraying training dataset to establish a mapping relationship from historical input feature vectors to coating thickness values.

4. The method for controlling the uniformity of coating thickness in aluminum single-panel spraying according to claim 1, characterized in that: Obtaining real-time surface geometric features includes: During the real-time spraying process, real-time point cloud data of the real-time spraying area is collected to form a set of real-time triangular spraying areas; real-time spraying process parameters of the spray gun in the real-time spraying area are obtained, and the real-time normal tilt angle and real-time curvature characterization value of the real-time spraying area are calculated based on the vertex coordinates of the real-time spraying area in the set of real-time triangular spraying areas, as real-time surface geometric features.

5. The method for controlling the uniformity of coating thickness in aluminum single-panel spraying according to claim 1, characterized in that: The steps for calculating coating thickness and tolerances include: The next triangular spraying area to be sprayed is determined according to the spraying path, and its surface geometric features are obtained; the surface geometric features of the next triangular spraying area to be sprayed are combined with the real-time spraying process parameters of the real-time spraying area to form a predicted input feature vector; the predicted input feature vector is input into the coating thickness analysis model to obtain the predicted coating thickness; the difference between the predicted coating thickness and the target coating thickness is calculated as the thickness deviation.

6. The method for controlling the uniformity of coating thickness in aluminum single-panel spraying according to claim 1, characterized in that: The steps for generating adjustment instructions and making real-time adjustments include: Based on the thickness deviation and the surface geometry of the next triangular spraying area to be coated, an adjustment strategy function is constructed with the thickness deviation as the basic input and the surface geometry as the dynamic correction coefficient. The adjustment amount of the spray gun moving speed, paint flow rate, and axial direction vector is calculated according to the adjustment strategy function. Based on the adjustment amount, the spray gun moving speed, paint flow rate, and axial direction vector are updated in real time when spraying the next triangular spraying area. After completing the spraying of the next triangular spraying area, the next triangular spraying area is updated as a new real-time triangular spraying area, and the process returns to the step of calculating the coating thickness and deviation to perform the prediction and spraying process parameter adjustment for the next cycle.

7. The method for controlling the uniformity of coating thickness in aluminum single-panel spraying according to claim 6, characterized in that: The adjustment amount of the spraying process parameters calculated by the adjustment strategy function includes: A first coefficient is calculated based on the thickness deviation and the curvature characterization value of the next triangular spraying area to be sprayed. The thickness deviation is multiplied by the first coefficient to obtain the spray gun movement speed adjustment amount. A second coefficient is calculated based on the thickness deviation and the normal tilt angle of the next triangular spraying area to be sprayed. The thickness deviation is multiplied by the second coefficient to obtain the paint flow rate adjustment amount. The spray gun axis direction vector adjustment amount is calculated based on the thickness deviation and the vector difference between the surface unit normal vector of the next triangular spraying area to be sprayed and the current direction vector of the spray gun.

Images