A multi-directional automatic spraying process for plastic toys

Abstract

The application discloses a kind of multidirectional automatic spraying process of plastic toy, it belongs to the technical field of toy manufacturing, it includes: obtaining toy three-dimensional profile and constructing digital model, identifying feature area;Adopt path planning algorithm to generate self-intersection-free spraying track, generate compensation path to recess and hollow area;Dynamic adjustment of spraying process parameters;Automatic spraying according to planning track, real-time monitoring coating thickness and triggering compensation spraying;Thickness distribution is measured on line after spraying, to evaluate uniformity with coefficient of variation and iterative optimization parameter;Finally, solidification treatment.The application realizes the high uniformity spraying of complex surface, and the coating thickness coefficient of variation can be reduced to 8.6% or less, the adhesion reaches 0 level, and the production efficiency is increased by more than 60%.

Description

Technical Field

[0001] This invention relates to the field of toy manufacturing technology, specifically to a multi-directional automatic spraying process for plastic toys. Background Technology

[0002] As a crucial product in children's entertainment and education, the appearance quality of plastic toys directly impacts their market competitiveness. In the production process of plastic toys, surface coating not only imparts rich colors and textures but also enhances their wear resistance and lifespan. However, with consumers demanding increasing precision and consistency in toys, traditional coating processes are no longer sufficient to meet the production needs of high-quality toys.

[0003] Currently, there is a large amount of research and patent applications related to the spraying process of plastic toys both domestically and internationally. For example, utility model patent CN216988215U discloses a pigment spraying device for easily coloring resin figurines. This device uses a rotating placement platform to adjust the angle of the figurine as needed during the spraying process. Simultaneously, a motor drives a screw to adjust the height of the spraying module on the lifting plate, and a second motor adjusts the nozzle angle, enabling multi-angle spraying and improving the spraying effect. Invention patent application CN107824375A discloses a spraying machine for processing plastic toys. By changing the inclination of the input frame, the toy is indirectly pushed onto the insertion rod. The direction of the turntable is changed to be parallel to the middle layer. During the pushing process, painting and drying inside the box are carried out. After painting, the turntable flips so that the toys fall onto the conveyor frame for batch delivery, thereby improving process efficiency. Utility model patent CN222817155U discloses a plastic toy painting device. This device uses a sliding and merging masking template on a painting table to hold the plastic toy inside, achieving targeted painting and overcoming the limitations of single-color painting. Simultaneously, a drying fan blows hot air into the air duct to prevent the toy from being difficult to remove before the paint has dried. Invention patent application CN120115322A discloses a surface treatment device for toy production lines. A motor drives a turntable, allowing toy parts to rotate omnidirectionally during painting, thus improving surface treatment efficiency.

[0004] However, the aforementioned existing technologies still have the following technical problems: First, the spraying path and posture planning lack intelligence. In existing technologies, whether the workpiece angle is adjusted by a rotary table or spraying is performed by multi-angle nozzles, the movement trajectory is a preset fixed path, which cannot be adaptively adjusted according to the actual three-dimensional contour of the plastic toy. When encountering irregular structures (such as depressions, protrusions, hollows, etc.), the fixed path often leads to uneven coating thickness distribution, with serious under-spraying or over-spraying in some areas, requiring subsequent manual touch-up spraying.

[0005] Secondly, the problem of blind spots in the spraying of the clamping contact area has not been fundamentally solved. When the existing spraying device is clamped and fixed, the contact area between the clamped part and the surface of the toy cannot be covered by spraying. This "masking effect" results in exposed uncoated areas in the clamping area after demolding, which must be re-sprayed or manually repaired, seriously affecting production efficiency and product consistency.

[0006] Third, the matching between process parameters and coating systems is insufficient. Existing technologies do not disclose a systematic matching scheme between spraying parameters (such as spray distance, spray speed, air pressure, and sector angle) and coating formulations for specific plastic substrates (such as ABS resin). The lack of mathematical models for process optimization leads to significant differences in coating thickness between different batches of products, making it difficult to consistently achieve the required coating adhesion and scratch resistance. Summary of the Invention

[0007] The present invention aims to provide a multi-directional automatic spraying process for plastic toys to solve the technical problems of uneven coating thickness, missed coating in clamping and shielding areas, and poor matching of process parameters caused by irregular structures in the prior art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: A multi-directional automatic spraying process for plastic toys includes the following steps: Step S1: 3D contour scanning and modeling.

[0009] The plastic toy to be painted is placed on a multi-station turntable, and 3D point cloud data of the toy's surface is acquired using a 3D laser scanner. The raw point cloud data is filtered, denoised, and smoothed to remove outliers and noise points, resulting in a complete 3D point cloud model of the toy's surface. The Delaunay triangulation algorithm is used to reconstruct the point cloud data into a mesh, generating a digital 3D mesh model of the toy. Based on a curvature analysis algorithm, characteristic regions of the toy's surface are identified: regions with a curvature radius greater than 50 mm are classified as planar regions; regions with a curvature radius between 5 and 50 mm are classified as curvature variation regions; regions with a curvature radius less than 5 mm and whose normals converge are classified as concave regions; convex hull regions with diverging normals are classified as convex regions; and nodes with hole edges are classified as hollow edge regions.

[0010] Step S2: Intelligent planning of spraying path.

[0011] Based on the 3D mesh model generated in step S1, a path planning algorithm is used to globally sort the centroid coordinates of each triangle, generating a non-self-intersecting spraying trajectory in 3D space. The path planning algorithm preferably employs an improved traveling salesman algorithm or an ant colony optimization algorithm, with the objective function set as minimizing the total path length. A no-go zone constraint is introduced to ensure the spraying path does not cross non-spraying areas. For identified planar areas and areas with varying curvature, a zigzag reciprocating path is used to plan the nozzle movement trajectory, with the overlap between adjacent paths controlled at 30%-50%. For identified recessed areas and hollowed-out edge areas, a spiral progressive compensation path is automatically generated. The spraying speed in the compensation section is reduced by 20%-30% compared to the baseline speed to increase the paint film thickness in that area.

[0012] Step S3: Optimize dynamic process parameters.

[0013] Based on the characteristic region type identified in step S1, the spraying process parameters are dynamically adjusted. The spraying process parameters include spraying distance D∈[150,250]mm, spraying speed v∈[300,800]mm / s, atomizing air pressure P∈[0.3,0.7]MPa, fan-shaped air pressure P_f∈[0.2,0.5]MPa, and paint flow rate Q∈[80,200]mL / min.

[0014] Among them, the spraying distance D is defined as the straight-line distance from the spray gun outlet to the sprayed point on the toy surface. Its physical meaning is the travel length of the paint particles from the spray gun outlet to the workpiece surface, which directly affects the speed, atomization degree and deposition range of the paint particles when they reach the workpiece surface; the spraying speed v is defined as the tangential movement rate of the spray gun nozzle relative to the workpiece surface along the planned path. Its physical meaning is the length of the paint deposition trajectory formed on the workpiece surface per unit time, which directly affects the amount of paint accumulated per unit area; the atomizing air pressure P is defined as the compressed air pressure supplied to the spray gun to break the liquid paint into tiny droplets. Its physical meaning is the control parameter of the average particle size and distribution width of the paint particles; the fan-shaped air pressure P_f is defined as the auxiliary air pressure used to control the lateral expansion width of the paint mist jet from the spray gun. Its physical meaning is the adjustment parameter of the spray width; the paint flow rate Q is defined as the volume of liquid paint passing through the spray gun outlet per unit time. Its physical meaning is the direct control quantity of the paint deposition rate. The physical dimensions of the above five parameters are as follows: D is length (mm), v is velocity (mm / s), P and P_f are pressure (MPa), and Q is volumetric flow rate (mL / min). The physical dimensions of each parameter are strictly matched with their physical definitions.

[0015] The preferred dynamic adjustment method is a parameter optimization method based on deep reinforcement learning, specifically using the Deep Q-Network (DQN) algorithm.

[0016] The specific network structure and training process of the Deep Q-Network (DQN) algorithm are as follows: (a) Neural Network Structure: DQN adopts a fully connected feedforward neural network architecture, consisting of an input layer, three hidden layers, and an output layer. The input layer has 10 nodes, corresponding to 10 discrete states in the state space (encoding vectors of the five major feature regions and their area proportions); the first hidden layer is a fully connected layer with 128 neurons, using the ReLU function f(x)=max(0,x); the second hidden layer is a fully connected layer with 64 neurons, using the ReLU function; the third hidden layer is a fully connected layer with 32 neurons, using the ReLU function; the output layer is a fully connected layer with 32 neurons, corresponding to 32 discrete actions in the action space, with no activation function, directly outputting the Q-value estimates for each action. The network weights use the Xavier uniform initialization method, with the bias initialized to 0.

[0017] (b) Definition of state space: State space There are 10 discrete states, defined as a combination encoding of five major feature region types (planar region, curvature variation region, concave region, convex region, and hollow edge region) and their area proportions (high proportion >30%, medium proportion 10%-30%, low proportion <10%). Each state s_t is a 10-dimensional one-hot encoded vector, which physically represents the comprehensive representation of the feature type and area proportion of the currently sprayed area.

[0018] (c) Definition of action space: Action space There are 32 discrete actions, defined as a combined adjustment of spraying speed v, spraying distance D, atomizing air pressure P, and paint flow rate Q. Each action a_t is a 4-dimensional parameter adjustment vector (Δv, ΔD, ΔP, ΔQ), where Δv ∈ {-50, 0, +50} mm / s, ΔD ∈ {-10, 0, +10} mm, ΔP ∈ {-0.05, 0, +0.05} MPa, and ΔQ ∈ {-10, 0, +10} mL / min. The physical meaning of each action is an incremental adjustment of parameters based on the current process parameters.

[0019] (d) Reward Function Design: The reward function R is set to the negative value of the root mean square thickness deviation of the sprayed surface, i.e. Where h_i is the measured coating thickness (μm) at the i-th measurement point, h_target is the preset target thickness (μm), and N is the total number of measurement points. The physical meaning of this reward function is: the closer the coating thickness distribution is to the target value, the smaller the root mean square deviation, and the higher the reward value (closer to 0); the larger the deviation, the lower the reward value (the larger the negative value). The dimension of the reward function is thickness (μm), consistent with the physical dimension of coating thickness measurement.

[0020] (e) Training process: The DQN adopts an off-policy training method, uses a replay buffer to store historical experience tuples (s_t,a_t,R_t,s_{t+1}), and the capacity of the buffer is set to 10000. The parameters of the target Q network are copied from the online Q network every 100 steps to stabilize the training process. The loss function adopts the mean square error loss: , where, is the temporal difference target, γ=0.95 is the discount factor, θ is the online network parameter, is the target network parameter. The optimizer adopts the Adam algorithm, and the learning rate is set to 0.001. The training dataset is derived from the enterprise's historical production data (a total of 1562 sets of spraying parameters and corresponding coating quality data). The model is iteratively trained for 3000 epochs in a simulation environment, and each epoch contains 200 steps of interaction to obtain a parameter optimization strategy network.

[0021] Step S4: Multi-axis linkage automated spraying.

[0022] According to the spraying trajectory generated in step S2 and the spraying process parameters adjusted in step S3, a six-axis industrial robot is used as the execution carrier to automatically spray the toy. During the spraying process, the vision sensor monitors the coating thickness in real time at a sampling frequency of 50Hz. When it is detected that the coating thickness is lower than the set threshold, the system automatically triggers compensatory spraying, and the paint film thickness in this area is increased by superimposing a compensatory trajectory.

[0023] Step S5: On-line detection and closed-loop correction of coating uniformity.

[0024] After the spraying is completed, the coating thickness distribution on the toy surface is obtained by an on-line coating thickness gauge, and the spraying uniformity is evaluated by the coefficient of variation CV=σ / μ, where σ is the standard deviation (μm) and μ is the mean value (μm). The coefficient of variation CV is a dimensionless quantity, and its physical meaning is the relative dispersion degree of the coating thickness data. A smaller value indicates a more uniform coating. The set threshold is CV≤15%. If the CV value does not exceed the threshold, it is judged as qualified and sent to the subsequent curing process; if the CV value exceeds the threshold, the spraying parameters of the next workpiece are automatically corrected according to the thickness deviation distribution. After the detection of a predetermined number of workpieces (such as 50 pieces) is completed, the system automatically analyzes the main area distribution of CV exceeding the standard. For areas where more than 80% of the non-compliant workpieces have similar defects, the process knowledge base automatically reduces the corresponding spraying speed in this area by 5%-10% or increases the spraying flow rate by 5%-8% to realize the optimization iteration of parameters.

[0025] Step S6: Curing treatment.

[0026] The coated plastic toys are then transferred to an infrared-hot air composite curing oven for curing. The curing temperature is controlled in three stages: preheating stage 40-50℃ (duration 1-2min), constant temperature stage 60-70℃ (duration 5-8min), and post-curing stage 50-55℃ (duration 2-3min), so that the coating can be fully cross-linked and cured.

[0027] Compared with the prior art, the present invention has the following beneficial effects: (1) The uniformity of spraying is significantly improved. Through adaptive path planning and dynamic parameter adjustment based on a three-dimensional model, the coefficient of variation (CV) of coating thickness can be reduced to below 8.6%, which is 55.4% better than the 19.3% of the prior art. The coating thickness in difficult-to-spray areas such as recessed inner surfaces is increased from 31.5 μm to 43.8 μm, and the standard deviation is reduced from 10.4 μm to 4.1 μm.

[0028] (2) The coating adhesion is greatly enhanced. According to the cross-cut test of GB / T 9286-2021, the coating adhesion obtained by the process of the present invention reaches level 0 (the cut edge is completely smooth and there is no peeling), while the existing technology is usually only level 2 (the cut edge has a little peeling), and the adhesion level is improved by more than one level.

[0029] (3) Production efficiency is significantly improved. The spraying efficiency of the process of this invention reaches 62 pieces / hour, which is about 63.2% higher than the 38 pieces / hour of the prior art. Automatic path planning eliminates the time for manual reversal, and the composite curing process reduces the total curing time from 15-20 minutes to about 10 minutes.

[0030] (4) Overall reduction in appearance defects. The proportion of undersprayed areas decreased from 12% in the prior art to 3%, the proportion of oversprayed areas decreased from 8% to 2%, and the incidence of missing coating in clamped and masked areas decreased from 15% to 0%, resulting in an overall reduction of appearance defects of approximately 65%.

[0031] (5) High level of intelligence. After introducing deep reinforcement learning parameter optimization, the coating thickness variation coefficient can be further reduced to 5.2%, and the uniformity is improved by 39.5% compared with the basic embodiment. It also supports multi-station, multi-color switching and rapid production changeover, and is suitable for large-scale flexible production. Attached Figure Description

[0032] Figure 1 This is a flowchart illustrating the overall process of a multi-directional automatic spraying process for plastic toys according to the present invention. Figure 2 This is a schematic diagram of the equipment layout and workstation collaboration used in the multi-directional automatic spraying process for a plastic toy according to the present invention. Figure 3 This is a closed-loop quality control logic diagram of a multi-directional automatic spraying process for plastic toys according to the present invention. Figure 4 This is a schematic diagram of the multi-degree-of-freedom robot spraying path planning for a multi-directional automatic spraying process for plastic toys according to the present invention. Detailed Implementation

[0033] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0034] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0035] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0036] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0037] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0038] It should be noted that when an element is referred to as being "fixed to" or "set on" another element, it can be directly on the other element or there may be an intervening element. When an element is considered to be "connected to" another element, it can be directly connected to the other element or there may be an intervening element. The terms "vertical," "horizontal," "upper," "lower," "left," "right," and similar expressions used herein are for illustrative purposes only and do not represent the only possible implementation.

[0039] Please refer to the following: Figures 1 to 4 The following is a detailed embodiment of the multi-directional automatic spraying process for plastic toys provided by the present invention: Implementation Method 1

[0040] This embodiment provides a multi-directional automatic spraying process for plastic toys, specifically including the following operation steps: (1) Three-dimensional contour scanning and processing.

[0041] An ABS plastic toy is fixed on a multi-station turntable with 360-degree continuous rotation. A 3D laser scanner (model: HL-3D-400, scanning accuracy ±0.05mm) is activated to scan the toy around its entire circumference at a speed of 0.5 r / min, acquiring 3D point cloud data. The raw point cloud data is filtered, denoised, and smoothed to remove outliers and noise points, resulting in a complete 3D point cloud model of the toy's surface. This model is imported into 3D modeling software (such as Geomagic Control), and the Delaunay triangulation algorithm is used to reconstruct the point cloud data into a mesh, generating a digital 3D mesh model of the toy. Based on a curvature analysis algorithm, characteristic regions on the toy's surface are identified: regions with a curvature radius greater than 50 mm are classified as planar regions; regions with a curvature radius between 5 and 50 mm are classified as curvature variation regions; regions with a curvature radius less than 5 mm and whose normal directions converge are classified as concave regions; convex hull regions with diverging normal directions are classified as convex regions; and nodes with hole edges are classified as hollow edge regions.

[0042] (2) Intelligent planning of spraying path.

[0043] Based on the 3D mesh model generated in step (1), the traveling salesman algorithm (TSA) with ant colony optimization is used to globally sort the centroid coordinates of each triangle. The objective function is set to minimize the total path length, while a no-go zone constraint is introduced to ensure that the spraying path does not cross non-spraying areas. For the identified planar areas and curvature variation areas, the nozzle movement trajectory is planned according to a zigzag reciprocating path, and the overlap rate between adjacent paths is controlled at 30%-50%. For the recessed areas and hollowed-out edge areas, a spiral progressive compensation path is automatically generated. The spraying speed of the compensation section is reduced by 20%-30% compared with the base speed to increase the paint film thickness in this area. The planned path is output as the robot's instruction code.

[0044] (3) Dynamic process parameter setting.

[0045] Based on the feature region type identified in step (1), the optimal process parameters are matched from the process parameter knowledge base. This implementation method presets three parameter modes: Planar area mode: spraying distance 200mm, speed 500mm / s, atomizing air pressure 0.4MPa, fan-shaped air pressure 0.3MPa, flow rate 120mL / min; Curved transition zone mode: spraying distance 180mm, speed 400mm / s, atomizing air pressure 0.45MPa, fan-shaped air pressure 0.35MPa, flow rate 100mL / min; Recess / Edge Area Mode: Spraying distance 150mm, speed 300mm / s, atomizing air pressure 0.5MPa, fan-shaped air pressure 0.4MPa, flow rate 80mL / min.

[0046] (4) Automatic spraying operation.

[0047] The planned path code and parameter settings are imported into a six-axis industrial robot (model: ABB IRB6700). The robot, carrying a spray gun, sequentially paints each area according to the planned path. The spray gun uses a high-pressure airless spraying method (spray width 250 mm) and water-based acrylic polyurethane coating (solid content 45±2%). During the spraying process, a high-precision laser sensor monitors the paint film thickness of the scanned area in real time at a sampling frequency of 50 Hz, quickly calculates the cumulative paint film thickness of each area, and transmits the data back to the main control system. When the control system detects that the cumulative thickness of a certain scanned area is lower than the rated thickness threshold (e.g., 35 μm), the system triggers real-time compensation spraying, superimposing a compensation trajectory on the adjacent spraying path until the paint film thickness reaches the rated standard.

[0048] (5) Online detection and closed-loop correction.

[0049] After spraying, the toy is transferred to the online coating thickness gauge (model: Phynix Surfix) station. The thickness gauge measures the thickness of the entire surface of the toy in a grid pattern (grid spacing 10 mm), collecting approximately 200 measurement points. The mean thickness μ and standard deviation σ of all measurement points are calculated to obtain the coefficient of variation CV = σ / μ. If CV ≤ 15%, it is judged as qualified and sent to the next curing stage; if CV > 15%, the system automatically enters the unqualified data into the process correction database and marks the product for manual inspection. After a total of 50 tests are completed, the system automatically analyzes the main areas where CV exceeds the standard. For areas where more than 80% of the unqualified parts have the same defect, the process knowledge base automatically lowers the corresponding spraying speed (each correction increment is 5%-10%) or increases the spraying flow rate (each correction increment is 5%-8%) to achieve parameter optimization iteration.

[0050] (6) Composite curing treatment.

[0051] The coated plastic toys are then placed into an infrared-hot air composite curing oven. The oven has three temperature zones: the first zone is 45±2℃ for 1.5 minutes, primarily to allow the solvent in the coating to initially evaporate; the second zone is 65±2℃ for 6 minutes, promoting the cross-linking and curing reaction of the coating resin; and the third zone is 52±2℃ for 2 minutes, relieving internal stress in the coating. After curing, the mixture is allowed to cool naturally to room temperature (approximately 23-25℃) to obtain the finished product.

[0052] Verification of the technical effect of Implementation Method 1: Thirty ABS plastic toys were coated using the process described in Embodiment 1. Online testing results showed that the average coating thickness was 45.1 μm (top curved surface), 46.2 μm (sidewall plane), and 43.8 μm (recessed inner surface), with standard deviations of 3.2 μm, 2.5 μm, and 4.1 μm, respectively, and an overall coefficient of variation (CV) of 8.6%. According to the GB / T9286-2021 cross-cut adhesion test, the adhesion reached grade 0. The coating efficiency was 62 pieces / hour, with under-coated areas accounting for 3%, over-coated areas accounting for 2%, and a 0% rate of missed coating at the clamping and masking areas. Implementation Method 2

[0053] This implementation method, based on the first implementation method, further introduces machine learning algorithms to dynamically optimize process parameters.

[0054] In step (3) the dynamic process parameter setting stage, a reinforcement learning algorithm based on deep Q network (DQN) is used to optimize the process parameters in real time.

[0055] The specific network structure of the DQN algorithm is as follows: a fully connected feedforward neural network with 10 nodes in the input layer (corresponding to 10 discrete states in the state space), 128 neurons in the first hidden layer (ReLU activation), 64 neurons in the second hidden layer (ReLU activation), 32 neurons in the third hidden layer (ReLU activation), and 32 neurons in the output layer (corresponding to 32 discrete actions, without activation functions). The network weights are initialized uniformly using Xavier, and the biases are initialized to 0.

[0056] The state space is defined as a combination encoding of the five major feature regions of the plastic toy and their area proportions, totaling 10 discrete states. The action space is defined as the coordinated adjustment actions of spraying speed, spraying distance, atomizing air pressure, and paint flow rate, totaling 32 discrete actions. The reward function is set as the negative value of the root mean square thickness deviation of the sprayed surface, i.e. Training was conducted offline with an experience replay pool of 10,000. The target network was updated every 100 steps, with a discount factor γ = 0.95, and the loss function was the mean squared error loss. The optimizer used was Adam (learning rate 0.001). The training dataset came from the company's historical production data (a total of 1562 sets of spraying parameters and corresponding coating quality data). The model was trained iteratively for 3000 rounds in the simulation environment, with 200 interactive steps per round, to obtain the parameter optimization policy network.

[0057] In step (5), during the online inspection phase, this embodiment further introduces a threshold-based statistical process control (SPC) mechanism for coating defects. When the CV value of 20 consecutively produced plastic toys is less than 10%, the system determines that the process is stable and automatically switches to a fast inspection mode (reducing the number of measurement points from 200 to 50), shortening the thickness measurement time for each workpiece by approximately 75%, further improving the production line cycle time. When the CV value of two consecutive workpieces exceeds 12% or the thickness deviation of the same part of three consecutive workpieces exceeds 8 μm, the system triggers an early warning and automatically resumes the full-point accurate thickness measurement mode.

[0058] During the curing process, this embodiment uses an infrared thermal imager to monitor the surface temperature distribution of the toy in real time. When the surface temperature gradient of the toy exceeds 10°C, the system automatically increases the air circulation frequency or extends the dwell time of the workpiece in the curing oven by 15%-20% to ensure the consistency of curing quality.

[0059] Verification of the technical effects of Implementation Method 2: Thirty ABS plastic toys were coated using the process described in Embodiment 2. Online inspection results showed that the coefficient of variation (CV) of the coating thickness was 5.2%, an improvement of 39.5% compared to Embodiment 1 (CV = 8.6%). The coating efficiency was increased to 68 pieces / hour (rapid inspection mode saves thickness measurement time), while the adhesion remained at level 0. The overall defect incidence rate was further reduced by approximately 30% compared to Embodiment 1.

[0060] Implementation Method 3 (Comparative Experiment): To verify the technical effect of the present invention, the applicant conducted comparative spraying tests on the same model of ABS plastic toys using existing spraying processes (control group) and the process of Embodiment 1 of the present invention (experimental group). The sample size of each group was n=30, and four core indicators were collected: spraying efficiency, coating uniformity, coating adhesion and appearance defect incidence rate.

[0061] Control group process: The multi-angle spraying device disclosed in CN216988215U was used to carry out the spraying operation according to the method described in its instruction manual.

[0062] Experimental results: (1) Spraying efficiency: The average number of spraying pieces in the control group was 38 pieces / hour, while the average number of spraying pieces in the experimental group was 62 pieces / hour, with the efficiency of the experimental group increasing by 63.2%.

[0063] (2) Coating uniformity: The coating thickness of key parts of the toy surface (top curved surface, side wall plane and recessed inner surface) was sampled using an online thickness gauge. The results are shown in Table 1 below:

[0064] The coefficient of variation (CV) of the coating thickness in the experimental group was 8.6%, which was much lower than that in the control group (19.3%), and the uniformity was improved by 55.4%.

[0065] (3) Coating adhesion: According to GB / T 9286-2021 cross-cut test (100-cut method, 1mm spacing), the average dry adhesion level of the control group was 2 (a little peeling at the cut edge), while the dry adhesion level of the experimental group reached 0 (the cut edge was completely smooth and there was no peeling), and the adhesion level was improved by more than one level.

[0066] (4) Occurrence rate of appearance defects: In the control group, the proportion of under-sprayed areas was 12%, the proportion of over-sprayed areas was 8%, and the incidence of missing coating at the clamping and masking area was 15%; in the experimental group, the proportion of under-sprayed areas was 3%, the proportion of over-sprayed areas was 2%, and the incidence of missing coating at the clamping and masking area was 0%. The overall appearance defects were reduced by about 65%.

[0067] The above comparative experiments fully demonstrate the significant advancement and industrial applicability of the technical solution of this invention.

[0068] In summary, the multi-directional automatic spraying process for plastic toys provided by this invention can be widely applied to the automated spraying production of plastic toys, handicrafts, and plastic parts, and is especially suitable for mass production scenarios with complex shapes and high coating quality requirements. This process equipment has a high degree of automation, is easy to upgrade existing production lines, and has significant economic and social benefits.

[0069] Specifically, the multi-directional automatic spraying process for plastic toys provided by this invention can be widely applied to the automated spraying production of plastic toys, handicrafts, and plastic parts, and is especially suitable for mass production scenarios with complex shapes and high coating quality requirements. This process equipment has a high degree of automation, is easy to upgrade on existing production lines, and has significant economic and social benefits.

[0070] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0071] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A multi-directional automatic spraying process for plastic toys, characterized in that, Includes the following steps: Step S1: Obtain the three-dimensional contour data of the plastic toy to be sprayed and construct a digital three-dimensional model, and identify the feature areas on the surface of the model; Step S2: Use a path planning algorithm to generate a non-self-intersecting spraying trajectory in three-dimensional space, and generate a compensation spraying path for the recessed area and hollowed-out edge area in the feature region; Step S3: Dynamically adjust the spraying process parameters based on the type of the feature region. The spraying process parameters include spraying distance D∈[150,250]mm, spraying speed v∈[300,800]mm / s, atomizing air pressure P∈[0.3,0.7]MPa, fan-shaped air pressure P_f∈[0.2,0.5]MPa, and paint flow rate Q∈[80,200]mL / min; Step S4: The toy is automatically sprayed according to the spraying trajectory generated in step S2 and the spraying process parameters adjusted in step S3. At the same time, the coating thickness is monitored in real time by a vision sensor. When the coating thickness is detected to be lower than the set threshold, compensation spraying is triggered. Step S5: After spraying, measure the coating thickness distribution online, evaluate the coating uniformity with the coefficient of variation CV=σ / μ, and optimize the process parameters iteratively when the CV value exceeds the threshold. Step S6: Curing treatment is performed on the toy after the painting is completed.

2. The multi-directional automatic spraying process for plastic toys according to claim 1, characterized in that, The feature areas in step S1 include planar areas, curvature variation areas, concave areas, convex areas, and hollowed-out edge areas.

3. The multi-directional automatic spraying process for plastic toys according to claim 1, characterized in that, The path planning algorithm in step S2 adopts an improved traveling salesman algorithm or an ant colony optimization algorithm.

4. The multi-directional automatic spraying process for plastic toys according to claim 1, characterized in that, The method for dynamically adjusting the spraying process parameters in step S3 is a parameter optimization method based on deep reinforcement learning.

5. The multi-directional automatic spraying process for plastic toys according to claim 4, characterized in that, The parameter optimization method based on deep reinforcement learning adopts a deep Q-network algorithm. The state space is defined as the feature region type and its area ratio, the action space is defined as the linkage adjustment action of spraying speed, spraying distance, atomization pressure and paint flow rate, and the reward function is set as the negative value of the root mean square thickness deviation of the sprayed surface.

6. The multi-directional automatic spraying process for plastic toys according to claim 1, characterized in that, The threshold in step S5 is set to a coefficient of variation CV ≤ 15%.

7. The multi-directional automatic spraying process for plastic toys according to claim 1, characterized in that, The curing process in step S6 adopts an infrared-hot air composite curing method, and the curing temperature is controlled in three stages: preheating stage 40-50℃, constant temperature stage 60-70℃, and post-curing stage 50-55℃.

8. The multi-directional automatic spraying process for plastic toys according to claim 1, characterized in that, In step S4, the vision sensor monitors the coating thickness in real time at a sampling frequency of 50Hz, and the compensation spraying is achieved by superimposing a compensation trajectory.

9. The multi-directional automatic spraying process for plastic toys according to claim 1, characterized in that, The method for constructing a digital 3D model in step S1 includes: filtering and denoising the acquired 3D point cloud data, performing mesh reconstruction using the Delaunay triangulation algorithm, and identifying feature regions based on the curvature analysis algorithm.

10. The multi-directional automatic spraying process for plastic toys according to claim 1, characterized in that, The optimization and iteration method of process parameters in step S5 includes: after completing the inspection of a predetermined number of workpieces, analyzing the distribution of the main areas with excessive coefficient of variation, and automatically reducing the spraying speed of the corresponding area by 5%-10% or increasing the spraying flow rate by 5%-8%.

Images