Intelligent spraying method and system for electronic product parts

Abstract

The application relates to the technical field of mechanical arm control, in particular to an intelligent electronic product part spraying method and system, which comprises the following steps: acquiring curvature characteristics of each position point on a surface of a part to be sprayed, and planning a spraying track containing multiple control points; performing local optimization on multiple current control points in a current optimization window; converting a global optimal solution of the current optimization window into a control instruction to drive a mechanical arm to perform a spraying action; sliding the current optimization window along the spraying track and repeating the local optimization until all the control points in the spraying track are traversed. Through the technical scheme, spraying control can be adjusted with high precision by using local geometric information in the optimization window, and the problems of low spraying quality and low control precision in spraying of a complex curved surface part are effectively solved.

Description

Technical Field

[0001] This application relates to the field of robotic arm control technology, and in particular to an intelligent spraying method and system for electronic product parts. Background Technology

[0002] With the rapid development of the consumer electronics industry, the appearance design of electronic products such as mobile phones, tablets, and wearable devices is increasingly pursuing thinness and streamlined design, and their shell components often contain complex geometric features such as large curvature R-angles and hyperboloids. Surface coating, as a key process in electronic product manufacturing, not only determines the appearance and texture of the product, but also plays a protective role such as corrosion resistance and wear resistance.

[0003] Currently, the painting of electronic product components is mainly accomplished by industrial robotic arms. However, due to the complex and varied geometric features of the surfaces of electronic product components, and the significant differences in curvature across different regions, this poses a severe challenge to the planning of the robotic arm's painting trajectory. Existing technologies typically utilize intelligent optimization algorithms such as particle swarm optimization (PSO) to plan the trajectory parameters of the robotic arm in order to find the optimal solution for the painting path. However, when applied to painting such complex curved surfaces, traditional PSO algorithms employ a global optimization method to determine the trajectory parameters of the entire painting path. When dealing with component surfaces with drastic curvature changes, they lack the ability to perceive local geometric features, making it difficult to achieve rapid convergence while simultaneously ensuring the painting quality of high-curvature areas. This leads to problems such as inaccurate spray gun posture, uneven coating thickness, and even missed spraying when handling rounded corners or irregular curved surfaces, severely affecting painting quality and control accuracy. Summary of the Invention

[0004] To address the technical problems of low coating quality and control precision, this application provides an intelligent coating method and system for electronic product parts. This system enables high-precision adjustment of coating control using local geometric information within an optimization window, effectively solving the problems of low coating quality and low control precision in the coating of complex curved surface parts.

[0005] In a first aspect, this application provides an intelligent spraying method for electronic product components. The spraying method includes: acquiring the curvature characteristics of various points on the surface of the component to be sprayed, and planning a spraying trajectory containing multiple control points; performing local optimization on multiple current control points within a current optimization window; converting the global optimal solution of the current optimization window into control commands to drive a robotic arm to perform a spraying action; sliding the current optimization window along the spraying trajectory and repeating the local optimization until all control points on the spraying trajectory are traversed; wherein, performing local optimization within the current optimization window includes: initializing the current optimal... The system constructs a basic inertia weight and multiple particles, where each particle includes trajectory parameters of the current control point. The basic inertia weight is negatively correlated with the average curvature characteristic of the current control point. A fitness function is constructed, which is negatively correlated with the projection distortion degree and the standard deviation of the estimated coating thickness of each current control point. The projection distortion degree is determined based on the curvature characteristic and the spray gun incident angle, which is the angle between the spray gun axis and the normal of the current control point. The particle swarm is iteratively updated within the feasible region of the robotic arm using the basic inertia weight and the fitness function until the global optimum is obtained.

[0006] By leveraging the global exploration and local development capabilities of the curvature feature dynamic adjustment algorithm, the robotic arm is forced to prioritize ensuring the verticality of the spray gun posture and reduce the thickness variance at complex curvature locations. This significantly improves the coating uniformity and spraying quality at complex curved surfaces while maintaining full-path spraying efficiency.

[0007] Preferably, obtaining the curvature characteristics of each position point on the surface of the part to be sprayed includes: discretizing the three-dimensional model of the part to be sprayed into a set of position points, calculating the maximum normal curvature and minimum normal curvature at any position point, and determining the curvature characteristics of the position point based on the maximum normal curvature and minimum normal curvature.

[0008] Preferably, the method for initializing the basic inertia weights includes: ; in, Based on the inertia weight, and These are the minimum inertia weight and the maximum inertia weight, respectively. For inertial sensitivity, This represents the average curvature characteristic of the current control point.

[0009] Maintaining a large inertia weight in flat regions avoids getting trapped in local optima and improves search speed, while rapidly reducing the inertia weight in high curvature regions enhances local optimization accuracy, thus ensuring the convergence efficiency of the optimization algorithm when dealing with regions of different geometric complexities.

[0010] Preferably, the trajectory parameters include the spray gun position, the spray gun attitude angle, and the moving speed. The spray gun position is the three-dimensional Cartesian coordinate of the spray gun tip in the workpiece coordinate system, and the spray gun attitude angle includes roll angle, pitch angle, and yaw angle.

[0011] Preferably, the projection distortion is calculated by obtaining the spray gun incident angle at any current control point; the projection distortion of the current control point is negatively correlated with the cosine value of the spray gun incident angle and positively correlated with the curvature feature.

[0012] It quantifies the coating deformation risk caused by non-perpendicular spraying on surfaces with different curvatures. During the optimization process, it can force the spray gun axis to be as close as possible to the surface normal direction. In particular, it can apply greater penalty force in high curvature areas, thereby effectively avoiding paint sagging or insufficient coverage caused by improper spraying angle and ensuring spraying quality.

[0013] Preferably, the estimated coating thickness at each current control point is obtained through a prediction model. The input parameters of the prediction model include the standard spray radius of the spray gun, the preset paint flow rate, and the trajectory parameters of the control point. The output is the coating thickness at the control point.

[0014] Preferably, the training process of the prediction model includes: recording the trajectory parameters, standard spray radius, and preset paint flow rate of any control point in a historical spraying operation as a training sample; measuring the actual coating thickness of the control point as the sample label of the training sample; inputting the standardized training sample into the prediction model; using the mean squared error loss of the output result and the sample label as the loss function; using the gradient descent method for backpropagation; iteratively updating the prediction model until the loss function is less than the preset convergence threshold, thus completing the training.

[0015] Preferably, the fitness function satisfies the following relation: ; in, For the first One particle; This represents the number of current control points in the particle. For the first The particle in the first Projection distortion at the current control point For the first Normalized rate of change of curvature at the current control point For the first The standard deviation of the estimated coating thickness per particle For the first The average of the estimated coating thickness per particle. The target value for coating thickness, and These are the first coefficient and the second coefficient, respectively. The second coefficient is positively correlated with the average curvature characteristic of the current control point, and the first coefficient is a preset value. This represents the maximum coating thickness, used for standardization.

[0016] By introducing a normalized rate of curvature change to weight the projection distortion degree and making the weight coefficient positively correlated with the average curvature feature, dynamic weighted optimization of multiple objectives is realized. This can automatically amplify the impact of projection distortion on fitness values ​​in locations with drastic curvature changes, forcing the sacrifice of some speed or path smoothness in that region in exchange for a better jet attitude.

[0017] Preferably, the step of converting the trajectory parameters of the first N current control points in the global optimal solution into control commands includes: performing smooth interpolation on the trajectory parameters of the first N current control points in the global optimal solution to generate an interpolation point sequence; and using the inverse kinematics solver of the robotic arm to convert the smoothed interpolation point sequence into rotation angle values ​​of the servo motors of each joint of the robotic arm, wherein the rotation angle value of each servo motor corresponds to the control command.

[0018] It eliminates the discontinuity between discrete optimization points, prevents the robotic arm from shaking or experiencing sudden speed changes during execution, protects the mechanical equipment, and ensures the continuity and stability of the spray gun movement, thereby improving the surface smoothness of the final coating.

[0019] In a second aspect, this application also provides an intelligent spraying system for electronic product parts, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an intelligent spraying method for electronic product parts according to the first aspect of this application is implemented.

[0020] The technical solution of this application has the following beneficial technical effects: By acquiring the surface curvature features of the parts and using a sliding optimization window for local optimization, and dynamically adjusting the search strategy of the particle swarm optimization algorithm using the basic inertia weights negatively correlated with the curvature features, a multi-objective fitness function is constructed based on the projection distortion degree and the estimated coating thickness, thus realizing adaptive trajectory planning for complex curved surface parts. It can automatically reduce the inertia weights in complex areas with large curvature for refined search, and force optimization of the spray gun posture by penalizing projection distortion, while it can converge quickly in flat areas. This effectively solves the problems of uneven coating thickness, many spray dead corners, and low control precision in the spraying of irregular curved surfaces of electronic products, and significantly improves the surface quality of automated spraying. Attached Figure Description

[0021] Figure 1 This is a flowchart of an intelligent spraying method for electronic product parts according to an embodiment of this application.

[0022] Figure 2 This is a flowchart illustrating local optimization of multiple current control points within the current optimization window according to an embodiment of this application.

[0023] Figure 3 This is a diagram showing the comparison of the optimization process of the global optimization algorithm and the local optimization algorithm according to the embodiments of this application.

[0024] Figure 4 This is a structural block diagram of an intelligent spraying system for electronic product parts according to an embodiment of this application. Detailed Implementation According to a first aspect of this application, this application provides an intelligent spraying method for electronic product parts. Figure 1 This is a flowchart of an intelligent spraying method for electronic product components according to an embodiment of this application. Figure 1 As shown, the intelligent spraying method for electronic product parts includes steps S101 to S104, which are described in detail below.

[0025] S101, obtain the curvature characteristics of each position on the surface of the part to be sprayed, and plan the spraying trajectory containing multiple control points.

[0026] In one embodiment, to achieve precision spraying of electronic product parts with complex geometric features, such as the rounded corners of mobile phone casings or irregular curved surfaces, it is necessary to digitally characterize the geometric properties of the part surface. Specifically, obtaining the curvature characteristics of each position point on the surface of the part to be sprayed includes: discretizing the three-dimensional model of the part to be sprayed into a set of position points; for any position point, fitting the local neighborhood surface at that position point, thereby obtaining the maximum normal curvature and minimum normal curvature at that position point; and determining the curvature characteristics of that position point based on the maximum and minimum normal curvature.

[0027] Location point Curvature characteristics at The mean curvature is used to represent this, and its calculation formula is as follows: ; in, and They are the location points Maximum and minimum normal curvature. Curvature characteristics. Capable of quantitatively characterizing the surface of a part at a given location. The degree of curvature, for example, the curvature feature approaches 0 in flat areas, while the value of the curvature feature increases significantly in areas with geometric abrupt changes such as the R-angle.

[0028] In other alternative embodiments, location points Curvature characteristics at It can also be expressed as the sum of squares, and its calculation formula is as follows: .

[0029] Furthermore, a spraying trajectory containing multiple control points is planned on the surface of the part to be sprayed. This spraying trajectory can be a rough path generated based on the normal direction of each position point, or multiple control points can be selected at equal intervals on the spraying trajectory. That is, a spraying trajectory contains a series of discrete control point sequences for subsequent optimization and adjustment.

[0030] In this way, by discretizing the model and extracting curvature features, the complex surface of the part is transformed into a digital representation of its geometric properties, laying a data foundation for subsequent adaptive control based on geometric features.

[0031] S102, perform local optimization on multiple current control points within the current optimization window.

[0032] In one embodiment, in order to accurately obtain the trajectory parameters at each control point in the spraying trajectory, this application introduces a sliding optimization window to perform local optimization of the trajectory parameters at each control point. The optimization window slides along the control points in the spraying trajectory.

[0033] Figure 2 This is a flowchart of local optimization of multiple current control points within the current optimization window according to an embodiment of this application. The local optimization includes steps S201 to S204, which are described in detail below.

[0034] S201, initialize the basic inertia weights and multiple particles of the current optimization window. The particles include the trajectory parameters of each current control point. The basic inertia weights are negatively correlated with the average curvature characteristics of the current control point.

[0035] In this embodiment, the current optimization window contains multiple current control points. In order to accurately obtain the trajectory parameters of the current control points, the basic inertia weight of the current optimization window is determined based on the average curvature characteristics of the current control points. In complex regions with large curvature, the inertia weight is reduced to enhance the fine search capability in the local area and avoid missing the best trajectory parameters. In flat regions with small curvature, the inertia weight is increased to improve the global search speed and accelerate the local optimization speed.

[0036] The methods for initializing the basic inertia weights include: ; in, Based on the inertia weight, and These are the minimum inertia weight and the maximum inertia weight, respectively. For inertial sensitivity, This represents the average curvature characteristic of the current control point. Among these, the minimum inertia weight... The value is 0.4, which represents the maximum inertia weight. The value is 0.9; inertial sensitivity The value is 2mm, used to adjust the sensitivity of the basic inertia weight to changes in curvature. The dimensions are Inertial sensitivity The product of the mean curvature feature It is a dimensionless numerical value.

[0037] Meanwhile, each particle includes trajectory parameters for each current control point. The initialization method for the multiple particles includes adding multiple random perturbations to the global optimal solution of the previous optimization window to obtain multiple initialized particles, wherein the number of particles is preset. The trajectory parameters include the spray gun position, spray gun attitude angle, and moving speed, used to control the pose of the robotic arm at the corresponding current control point. The spray gun position is the three-dimensional Cartesian coordinate of the spray gun tip in the workpiece coordinate system, and the spray gun attitude angle includes roll angle, pitch angle, and yaw angle.

[0038] S202, Construct the fitness function.

[0039] The fitness function is used to evaluate the spraying quality of particles. A higher fitness function value indicates a higher spraying quality when spraying the current optimization window based on that particle. The fitness function is negatively correlated with the projection distortion degree of each current control point and the standard deviation of the estimated coating thickness.

[0040] The projection distortion is determined based on curvature characteristics and the spray gun incident angle. Specifically, the spray gun incident angle at any current control point is obtained, where the spray gun incident angle is the angle between the spray gun axis and the normal to the current control point. The projection distortion of the current control point is negatively correlated with the cosine of the spray gun incident angle and positively correlated with the curvature characteristics. At any current control point, the more tilted the spray gun, the larger the spray gun incident angle, and the smaller the cosine of the spray gun incident angle. This results in more severe distortion of the sprayed shape on the part surface. Furthermore, if the curvature characteristics of the current control point are large, it indicates that the current control point is on a curved, complex surface, which further increases the distortion of the sprayed shape. This distortion affects the spraying quality at adjacent control points. Therefore, the first... The particle in the first Projection distortion at the current control point Satisfying the relation: ; in, For the first The curvature characteristics of the current control point For the first The spray gun incident angle at the current control point The curvature influence factor is used to adjust the degree of influence of curvature on distortion. The value of can be 1, and its dimension is mm. It is a dimensionless numerical value.

[0041] The estimated coating thickness at each current control point is obtained through a prediction model. The input parameters of the prediction model include the standard spray radius of the spray gun, the preset paint flow rate, and the trajectory parameters of the control point. The output is the coating thickness at the control point. The prediction model can be a BP neural network or a multinomial model. Before inputting the input parameters into the prediction model, each input parameter needs to be standardized to eliminate the dimensions of each input parameter.

[0042] To ensure the prediction model can output accurate coating thickness estimates, it needs to be trained. The training process includes: recording the trajectory parameters, standard spray radius, and preset paint flow rate of any control point in a historical spraying operation as a training sample; measuring the actual coating thickness of the control point as the sample label; inputting the standardized training sample into the prediction model; using the mean squared error of the output result and sample labels as the loss function; performing backpropagation using gradient descent to iteratively update the prediction model until the loss function is less than a preset convergence threshold, completing the training. The convergence threshold is 0.01.

[0043] After obtaining the projection distortion and estimated coating thickness of each current control point, the fitness function... Satisfying the relation: ; in, For the first One particle; This represents the number of current control points in the particle. For the first The particle in the first Projection distortion at the current control point For the first Normalized rate of change of curvature at the current control point For the first The standard deviation of the estimated coating thickness per particle For the first The average of the estimated coating thickness per particle. The target value for coating thickness, and These are the first coefficient and the second coefficient, respectively. The second coefficient is positively correlated with the average curvature characteristics of the current control point, and the first coefficient is a preset value. This represents the maximum coating thickness, used for standardization.

[0044] Understandably, the first The standard deviation of the estimated coating thickness per particle Its dimensions are the same as those of the coating thickness. The maximum coating thickness is used to... Standardize to eliminate The dimension of the value makes both parts of the fitness function dimensionless values.

[0045] It should be noted that, in order to ensure high-quality coating results in complex areas of the part surface, the second coefficient... The coefficient is positively correlated with the average curvature characteristics of the current control point. This means that in complex regions with high curvature, the emphasis on projection distortion is increased during the optimization process. Regardless of whether the region is complex or flat, the first coefficient is a preset value of 1, ensuring that all optimization windows place equal emphasis on coating quality. Used to constrain the uniformity of coating thickness, Used to ensure the accuracy of coating thickness, The higher the value, the better the coating quality.

[0046] Second coefficient Satisfying the relation: ;in, As the benchmark coefficient, The average curvature characteristic of the current control point. The weighting factor is 1 mm, which makes the weighting factor increase. It is a dimensionless numerical value; the baseline coefficient is 0.1.

[0047] Second coefficient The importance of projection distortion in the current local optimization is determined based on the global characteristics of all control points within the current optimization window. Furthermore, for each current control point within the current optimization window, if the rate of curvature change on its adjacent sides is large, it indicates that it is located at the transition position between flat and complex regions. This position is more affected by projection distortion and is more prone to paint accumulation or missed spraying dead corners. Therefore, the projection distortion at this position should be reduced as much as possible, and the importance of projection distortion at this position should be increased during the local optimization process.

[0048] No. Rate of curvature change at the current control point Satisfying the relation: ;in, For the first The curvature characteristics of the current control point and The first one in the spraying trajectory The curvature characteristics of the previous and next adjacent control points of the current control point. It should be noted that if... If it is the first control point in the spraying trajectory, then The value of is 0. ;like If it is the last control point in the spraying trajectory, then The value of is 0. The first Rate of curvature change at the current control point The ratio of the curvature change rate to the sum of the curvature change rates at all current control points is used as the first... Normalized rate of curvature change at the current control point .

[0049] S203 uses the basic inertia weights and fitness function to iteratively update the particle swarm within the feasible region of the robotic arm until the global optimum is obtained.

[0050] Using the basic inertia weights and fitness function of the current optimization window, a particle swarm optimization algorithm is employed to iteratively update the position and velocity of the particle swarm within the feasible region of the robotic arm until a convergence condition is met, thus obtaining the global optimum of the current optimization window. The convergence condition is that the number of iterations equals the maximum number of iterations or the change in fitness is less than a change threshold. The maximum number of iterations is 50, and the change threshold is 0.01. The feasible region of the robotic arm is the set of all trajectory parameters that the robotic arm can reach.

[0051] Thus, through the aforementioned dynamically adjusted inertia weights and multi-objective fitness functions, rapid planning is possible in planar regions, while the search is automatically tightened and geometric adaptability is emphasized in complex regions, reducing projection distortion and thus outputting high-quality trajectory parameters.

[0052] S103 converts the global optimal solution of the current optimization window into control commands to drive the robotic arm to perform the spraying action.

[0053] In one embodiment, in order to convert the planned Cartesian space trajectory into the actual movement of the robotic arm and to ensure the smoothness of the robotic arm's movement and prevent jitter caused by abrupt changes in attitude between discrete control points, trajectory smoothing and kinematic transformation are required.

[0054] Since the control points are obtained through sampling, in order to obtain continuous and smooth control commands, the process of converting the global optimal solution of the current optimization window into control commands includes: smoothing the trajectory parameters in the global optimal solution to generate an interpolation point sequence; and using the inverse kinematics solver of the robotic arm to convert the smoothed interpolation point sequence into the rotation angle values ​​of the servo motors of each joint of the robotic arm. ,in, This represents the number of degrees of freedom for the robotic arm. The rotation angle value of each joint servo motor corresponds to the control command, which drives the robotic arm to perform the spraying action via the controller. Smooth interpolation employs either a linear interpolation algorithm or a nonlinear interpolation algorithm.

[0055] S104, slide the current optimization window along the spraying trajectory and repeat the local optimization until all control points in the spraying trajectory have been traversed.

[0056] In one embodiment, after issuing control commands for each current control point, the current optimization window is slid forward by N control points along a preset spraying trajectory, where N is the number of control points within the optimization window.

[0057] At this point, the curvature characteristics of the newly entered control points in the optimization window are read, and steps S102 to S103 are repeated. This process continues until the optimization window covers and processes all control points in the spraying trajectory.

[0058] By optimizing the sliding mechanism of the window, local optimization of the entire spraying trajectory is achieved, avoiding the problems of high computational cost and low control accuracy in the global optimization of the entire spraying trajectory in traditional algorithms. Please refer to [link / reference]. Figure 3 The diagram shows the comparison of the optimization process of the global optimization and the local optimization algorithms according to the embodiments of this application. It can be seen that the convergence speed of local optimization is better than that of global optimization, and the fitness function value of local optimization is better than that of global optimization. This indicates that local optimization can fully consider local geometric information to make high-precision adjustments for intelligent spraying.

[0059] Thus, through the rolling optimization mechanism, the spraying control can be adjusted with high precision using local geometric information within the optimization window, effectively solving the problems of low spraying quality and low control precision in the spraying of complex curved parts.

[0060] According to a second aspect of this application, this application also provides an intelligent spraying system for electronic product parts. Figure 4 This is a structural block diagram of an intelligent spraying system for electronic product parts according to an embodiment of this application. Figure 4As shown, the system 50 includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the intelligent spraying method for electronic product parts according to the first aspect of this application. The system also includes other components well-known to those skilled in the art, such as a communication bus and a communication interface. Their configuration and functions are known in the art and will not be described further here.

[0061] It should be noted that, for those skilled in the art, several modifications and improvements can be made without departing from the concept of this application, and these all fall within the scope of protection of this application.

Claims

1. A smart spraying method for electronic product components, characterized in that, Spraying methods include: Obtain the curvature features of each point on the surface of the part to be sprayed, and plan a spraying trajectory containing multiple control points; perform local optimization on multiple current control points within the current optimization window; convert the global optimal solution of the current optimization window into control commands to drive the robotic arm to perform the spraying action; slide the current optimization window along the spraying trajectory and repeat local optimization until all control points in the spraying trajectory are traversed. The local optimization within the current optimization window includes: initializing the basic inertia weights and multiple particles for the current optimization window. These particles include the trajectory parameters of each current control point. The basic inertia weights are negatively correlated with the average curvature characteristics of the current control points. A fitness function is constructed, which is negatively correlated with the projection distortion degree and the standard deviation of the estimated coating thickness for each current control point. The projection distortion degree is determined based on the curvature characteristics and the spray gun incident angle, which is the angle between the spray gun axis and the normal to the current control point. The particle swarm is iteratively updated within the feasible region of the robotic arm using the basic inertia weights and the fitness function until the global optimum is obtained. Satisfying the relation: ； For the first One particle; This represents the number of current control points in the particle. For the first The particle in the first Projection distortion at the current control point For the first Normalized rate of change of curvature at the current control point For the first The standard deviation of the estimated coating thickness per particle For the first The average of the estimated coating thickness per particle. The target value for coating thickness, and These are the first coefficient and the second coefficient, respectively. The second coefficient is positively correlated with the average curvature characteristic of the current control point, and the first coefficient is a preset value. This is the maximum coating thickness, used for standardization. No. The particle in the first Projection distortion at the current control point Satisfying the relation: ； , The first The curvature characteristics of the current control point and the spray gun incident angle The curvature influence factor is used to adjust the degree of influence of curvature on distortion. The value of is 1, and its dimension is mm. It is a dimensionless numerical value.

2. The intelligent spraying method for electronic product parts according to claim 1, characterized in that, The process of obtaining the curvature characteristics of various points on the surface of the part to be coated includes: The three-dimensional model of the part to be painted is discretized into a set of position points. The maximum and minimum normal curvature at any position point are calculated, and the curvature characteristics of that position point are determined based on the maximum and minimum normal curvature.

3. The intelligent spraying method for electronic product parts according to claim 1, characterized in that, The methods for initializing the basic inertia weights include: ； in, Based on the inertia weight, and These are the minimum inertia weight and the maximum inertia weight, respectively. For inertial sensitivity, This represents the average curvature characteristic of the current control point.

4. The intelligent spraying method for electronic product parts according to claim 1, characterized in that, The trajectory parameters include the spray gun position, spray gun attitude angle, and moving speed. The spray gun position is the three-dimensional Cartesian coordinate of the spray gun tip in the workpiece coordinate system. The spray gun attitude angle includes roll angle, pitch angle, and yaw angle.

5. The intelligent spraying method for electronic product parts according to claim 1, characterized in that, The projection distortion is calculated as follows: obtain the spray gun incident angle of any current control point; the projection distortion of the current control point is negatively correlated with the cosine value of the spray gun incident angle and positively correlated with the curvature feature.

6. The intelligent spraying method for electronic product parts according to claim 1, characterized in that, The estimated coating thickness at each current control point is obtained through a prediction model. The input parameters of the prediction model include the standard spray radius of the spray gun, the preset paint flow rate, and the trajectory parameters of the control point. The output is the coating thickness at the control point.

7. The intelligent spraying method for electronic product parts according to claim 6, characterized in that, The training process of the prediction model includes: recording the trajectory parameters, standard spray radius, and preset paint flow rate of any control point in a historical spraying operation as a training sample; measuring the actual coating thickness of the control point as the sample label of the training sample; inputting the standardized training sample into the prediction model; using the mean squared error loss of the output result and the sample label as the loss function; using the gradient descent method for backpropagation; iteratively updating the prediction model until the loss function is less than the preset convergence threshold, thus completing the training.

8. The intelligent spraying method for electronic product parts according to claim 1, characterized in that, Converting the trajectory parameters of the first N current control points in the global optimal solution into control commands includes: performing smooth interpolation on the trajectory parameters of the first N current control points in the global optimal solution to generate an interpolation point sequence; and using the inverse kinematics solver of the robotic arm to convert the smoothed interpolation point sequence into rotation angle values ​​of the servo motors of each joint of the robotic arm, wherein the rotation angle values ​​of each servo motor correspond to the control commands.

9. An intelligent spray coating system for electronic product parts, characterized in that, It includes a processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement a smart spraying method for electronic product parts according to any one of claims 1 to 8.

Images