A method for intelligent painting of automobile parts based on neural network

Abstract

The present invention relates to an intelligent painting method for automobile parts based on neural network, comprising: constructing a parts model according to spatial feature data, extracting a model index data set of the parts model, inputting the model index data set into an original neural network model, obtaining original control parameters, modeling a paint surface according to the original control parameters, obtaining an original paint surface model, extracting original paint layer parameters, obtaining standard paint layer parameters, feedback-adjusting the original neural network model according to the original paint layer parameters and the standard paint layer parameters, obtaining a target neural network, inputting a current feature set into the target neural network, obtaining target control parameters, and spraying the paint surface according to the target control parameters. The present invention solves the problem that it is difficult for a machine to achieve accurate spraying of various parts of automobile parts during the process of automatic spraying of paint, which often results in uneven spraying and large color difference.

Description

Technical Field

[0001] The invention relates to the technical field of automobile spraying, and in particular to an intelligent spraying method for automobile parts based on a neural network. Background Art

[0002] With the development of the economy, the number of cars in my country is increasing year by year. Driven by the demand for cars, the spraying process of auto parts has also attracted people's attention.

[0003] Auto parts spraying refers to spraying a layer of paint from the surface of the parts to achieve the anti-corrosion and aesthetic effects of the auto parts. Compared with ordinary manual painting, mechanized automatic painting technology can better control the start-up, spray volume and spray trajectory of the spray gun, so it has the advantages of stability, safety and high spraying efficiency. However, it is difficult for machines to achieve accurate spraying of various parts of auto parts during the process of automatic spraying of paint, which often causes problems such as uneven spraying and large color difference. Summary of the invention

[0004] The present invention provides an intelligent painting method for automobile parts based on neural network, the main purpose of which is to improve the problem that it is difficult to achieve accurate spraying of various parts of automobile parts during the process of automatic spraying of paint by machines, which often causes uneven spraying and large color difference.

[0005] To achieve the above object, the present invention provides a method for realizing intelligent painting of automobile parts based on a neural network, comprising:

[0006] Collecting pre-built spatial feature data of automobile parts, and building a part model according to the spatial feature data;

[0007] Extracting a model index data set of the component model according to a predetermined feature index set;

[0008] Inputting the model indicator data set into a pre-built original neural network model to obtain original control parameters, wherein the original neural network includes an input layer, a hidden layer, and an output layer;

[0009] Performing paint surface modeling on the component model according to the original control parameters to obtain an original paint surface model, and extracting original paint layer parameters of the original paint surface model;

[0010] The standard paint layer parameters corresponding to the component model are obtained, and the pre-built difference calculation formula is used to feedback-adjust the original neural network model according to the original paint layer parameters and the standard paint layer parameters to obtain the target neural network, wherein the difference calculation formula is as follows:

[0011]

[0012] Among them, c i,j,k represents the difference value of the index parameter of the standard paint layer parameter and the original paint layer parameter at the corresponding spatial coordinate (i, j, k), n represents the coordinate serial number of the standard paint layer parameter and the original paint layer parameter in the unit area containing the spatial coordinate (i, j, k), N represents the number of coordinate serial numbers in the unit area, x represents the index parameter of the standard paint layer parameter at the spatial coordinate (i, j, k), y represents the index parameter of the original paint layer parameter at the spatial coordinate (i, j, k), and N spatial coordinates (i, j, k) are equidistantly distributed in the unit area;

[0013] Acquire a current feature set of the component to be painted, input the current feature set into the target neural network, and obtain target control parameters;

[0014] The paint surface of the parts to be painted is sprayed according to the target control parameters to complete the intelligent painting of automobile parts.

[0015] Optionally, the collecting of pre-built spatial feature data of automobile parts includes:

[0016] Constructing a local coordinate system according to the center position of the automobile parts;

[0017] A feature point set of the automobile component is identified, and spatial feature data of the feature point set is extracted in the local coordinate system.

[0018] Optionally, inputting the model indicator data set into a pre-built original neural network model to obtain original control parameters includes:

[0019] Normalizing each model indicator data in the model indicator data set to obtain a standard indicator data set;

[0020] Performing weight initialization on the input layer and the hidden layer in the original neural network to obtain initial input layer weights and initial hidden layer weights;

[0021] According to the correspondence between the feature indicators in the feature indicator set and the nodes in the input layer, the standard indicator data set is input into the original neural network model, and the standard indicator data set is weighted summed using the initial input layer weights to obtain the original input sum value;

[0022] Activating the original input sum value using a pre-built activation function to obtain an original hidden value;

[0023] Performing weighted summation on the original hidden values ​​using the initial hidden layer weights to obtain an original hidden sum value;

[0024] The activation function is used to activate the original hidden sum value to obtain the original control parameter.

[0025] Optionally, performing paint surface modeling on the component model according to the original control parameters to obtain the original paint surface model includes:

[0026] Sequentially extracting original trajectory control parameters, original distance control parameters, original angle control parameters, original speed control parameters and original pressure control parameters from the original control parameters;

[0027] Calculate the spraying area according to the original trajectory control parameter, the original distance control parameter and the original angle control parameter;

[0028] Calculating the amount of paint sprayed in the spraying area according to the original speed control parameter and the original pressure control parameter;

[0029] Calculating the paint surface parameters of the spraying area according to the sprayed paint amount;

[0030] The local coordinate system is used to perform paint surface modeling on the component model according to the paint surface parameters to obtain the original paint surface model.

[0031] Optionally, the calculating the amount of paint sprayed in the spraying area according to the original speed control parameter and the original pressure control parameter includes:

[0032] A set of unit areas selected in the spraying area according to a predetermined spacing;

[0033] Calculating the amount of paint sprayed in each unit area of ​​the unit area set according to the original speed control parameter and the original pressure control parameter;

[0034] The paint spraying amount of each unit area in the unit area set is fitted to obtain the paint spraying amount of the spraying area.

[0035] Optionally, the calculating the amount of paint sprayed on each unit area in the unit area set according to the original speed control parameter and the original pressure control parameter includes:

[0036] Calculate the paint spraying speed according to the preset paint spraying speed formula;

[0037] According to a preset unit paint quantity calculation formula, the paint quantity spraying speed is used to calculate the spray paint quantity of each unit area in the unit area set.

[0038] Optionally, the paint spraying speed formula is as follows:

[0039]

[0040] Among them, vp represents the paint spraying speed, f represents the nozzle pressure, t represents the time, m represents the paint mass, p represents the nozzle pressure, d represents the nozzle diameter, and ρ represents the paint density.

[0041] Optionally, the unit paint amount calculation formula is as follows:

[0042]

[0043] Wherein, q represents the amount of paint sprayed in each unit area of ​​the unit area set, L represents the length of the unit area along the spraying trajectory in the spraying area, l represents the side length of the unit area, and v y It represents the moving speed of the nozzle along the spraying trajectory, s represents the area of ​​the unit area, and S represents the area of ​​the spraying area.

[0044] Optionally, the using of a pre-constructed difference calculation formula to perform feedback adjustment on the original neural network model according to the original paint layer parameters and the standard paint layer parameters to obtain a target neural network includes:

[0045] Extracting corresponding index parameters from the original paint layer parameters and the standard paint layer parameters according to the preset paint layer index;

[0046] Calculating the difference value according to the original paint layer parameters and the corresponding index parameters in the standard paint layer parameters by using the difference calculation formula;

[0047] The original neural network model is subjected to negative feedback adjustment according to the difference value to obtain a target neural network.

[0048] Optionally, spraying the paint surface of the component to be painted according to the target control parameter includes:

[0049] Calibrate the target control parameters according to a preset start time to obtain a standard control parameter set;

[0050] The pre-built nozzle is regulated according to the standard control parameter set to spray the parts to be painted.

[0051] The present invention mainly utilizes a neural network to establish a correct correspondence between automobile parts and target control parameters for regulating spraying. Before establishing the correct correspondence between the automobile parts and the target control parameters, it is first necessary to collect the features of the automobile parts. Therefore, it is necessary to first construct a part model based on the spatial feature data of the automobile parts, and then classify and extract the model index data set of the part model according to a predetermined feature index set. After obtaining the model index data set, it is necessary to input it into a pre-constructed original neural network model to obtain the original control parameters. In the process of adjusting the original neural network model, it is necessary to adjust the spraying effect corresponding to the original control parameters and the automobile parts. Therefore, it is necessary to model the paint surface on the component model according to the original control parameters to obtain the original paint surface model, and then extract the original paint layer parameters of the original paint surface model, so as to realize the conversion of the spraying effect corresponding to the original control parameters, and then convert the correct spraying effect of the automobile parts into the standard paint layer parameters, and then the original paint layer parameters and the standard paint layer parameters can be used to feedback adjust the original neural network model to obtain the target neural network, and finally, the current feature set is input into the target neural network to obtain the target control parameters, and the paint surface of the component to be painted can be sprayed according to the target control parameters. Therefore, the present invention can solve the problem that it is difficult for machinery to achieve accurate spraying of various parts of automobile parts in the process of automatic spraying of paint, which often causes uneven spraying and large color difference. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A schematic diagram of a process flow of a method for intelligent painting of automobile parts based on a neural network provided by an embodiment of the present invention;

[0053] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0054] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0055] Reference Figure 1 FIG. 1 is a flow chart of a method for intelligent painting of automobile parts based on a neural network according to an embodiment of the present invention. In this embodiment, the method for intelligent painting of automobile parts based on a neural network includes:

[0056] S1. Collecting pre-built spatial feature data of automobile parts, and building a part model according to the spatial feature data.

[0057] It can be explained that the automobile parts can be parts that need to be painted, such as engines, oil pumps, water pumps, and body frames. The spatial feature data refers to the relative position coordinate data of each feature point on the surface of the automobile parts.

[0058] In the embodiment of the present invention, the collecting of pre-built spatial feature data of automobile parts includes:

[0059] Constructing a local coordinate system according to the center position of the automobile parts;

[0060] A feature point set of the automobile component is identified, and spatial feature data of the feature point set is extracted in the local coordinate system.

[0061] Explainably, the local coordinate system may be a three-dimensional space coordinate system whose origin is the center position, and the center position may be the center of gravity of the component. The feature point set may be a point set with features on an automobile component, such as a line inflection point, a shell lift vertex, etc. on a component.

[0062] S2. Extracting a model index data set of the component model according to a predetermined feature index set.

[0063] In the embodiment of the present invention, the feature index set refers to a set of appearance feature indexes of a component, such as shape, size, surface roughness, etc. The model index data set refers to specific data of the component model under each feature index, such as shape is a cuboid, size is 20cm long, 10cm wide, 10cm high, etc.

[0064] S3. Inputting the model indicator data set into a pre-built original neural network model to obtain original control parameters, wherein the original neural network includes an input layer, a hidden layer and an output layer.

[0065] Explainable, the original neural network model can be a BP neural network, used for negative feedback regulation training, and can include an input layer, a hidden layer and an output layer, each layer including multiple nodes. The nodes of the input layer can be shape input nodes, size input nodes, surface roughness input nodes, etc. The nodes of the output layer can be trajectory output nodes, distance output nodes, angle output nodes, speed output nodes and pressure output nodes. The trajectory output node can output the parameters of the movement trajectory of the nozzle during the painting process over time; the distance output node can output the parameters of the closest distance value of the nozzle to the surface of the component during the painting process over time; the angle output node can output the parameters of the spray angle of the nozzle during the painting process over time; the speed output node can output the parameters of the movement speed of the nozzle during the painting process over time; the pressure output node can output the parameters of the paint spraying pressure of the nozzle during the painting process over time.

[0066] Furthermore, the original control parameter refers to the parameter output from the output layer node for the first time after the model indicator data set is input into the original neural network model.

[0067] In the embodiment of the present invention, the step of inputting the model indicator data set into the pre-built original neural network model to obtain the original control parameters includes:

[0068] Normalizing each model indicator data in the model indicator data set to obtain a standard indicator data set;

[0069] Performing weight initialization on the input layer and the hidden layer in the original neural network to obtain initial input layer weights and initial hidden layer weights;

[0070] According to the correspondence between the feature indicators in the feature indicator set and the nodes in the input layer, the standard indicator data set is input into the original neural network model, and the standard indicator data set is weighted summed using the initial input layer weights to obtain the original input sum value;

[0071] Activating the original input sum value using a pre-built activation function to obtain an original hidden value;

[0072] Performing weighted summation on the original hidden values ​​using the initial hidden layer weights to obtain an original hidden sum value;

[0073] The activation function is used to activate the original hidden sum value to obtain the original control parameter.

[0074] Furthermore, the activation function may adopt an S-shaped curve formula, which is:

[0075]

[0076] Among them, y represents the value after activation, and x represents the value before activation.

[0077] S4. Paint surface modeling is performed on the component model according to the original control parameters to obtain an original paint surface model, and original paint layer parameters of the original paint surface model are extracted.

[0078] It can be understood that the original paint layer parameters refer to data such as the paint thickness and paint usage at various positions in the original paint surface model.

[0079] In the embodiment of the present invention, the paint surface modeling is performed on the component model according to the original control parameters to obtain the original paint surface model, including:

[0080] Sequentially extracting original trajectory control parameters, original distance control parameters, original angle control parameters, original speed control parameters and original pressure control parameters from the original control parameters;

[0081] Calculate the spraying area according to the original trajectory control parameter, the original distance control parameter and the original angle control parameter;

[0082] Calculating the amount of paint sprayed in the spraying area according to the original speed control parameter and the original pressure control parameter;

[0083] Calculating the paint surface parameters of the spraying area according to the sprayed paint amount;

[0084] The local coordinate system is used to perform paint surface modeling on the component model according to the paint surface parameters to obtain the original paint surface model.

[0085] It can be explained that the spraying area refers to the paint area sprayed by the nozzle to the fixed automobile parts when the nozzle is fixed, and generally the spraying area can be circular, elliptical, etc. The paint surface parameter refers to the amount of paint sprayed on the paint surface in the spraying area.

[0086] Further, the amount of paint sprayed in the spraying area is calculated according to the original speed control parameter and the original pressure control parameter, including:

[0087] A set of unit areas selected in the spraying area according to a predetermined spacing;

[0088] Calculating the amount of paint sprayed in each unit area of ​​the unit area set according to the original speed control parameter and the original pressure control parameter;

[0089] The paint spraying amount of each unit area in the unit area set is fitted to obtain the paint spraying amount of the spraying area.

[0090] Optionally, the spacing may be 1 cm. The unit area set may be a set of square areas, and the side length of the unit area may also be 1 cm. Since each unit area has four spacing areas adjacent to the four sides, the spacing area may take the middle value of the spray paint amount of the left and right unit areas, then take the middle value of the spray paint amount of the upper and lower unit areas, and finally take the average of the two middle values, so as to achieve a fitting effect and obtain the spray paint amount of the entire spray area. Finally, the spray paint amounts of all spray areas are combined together to obtain the spray paint amount of the entire automotive component.

[0091] Optionally, the index parameters such as the amount of paint sprayed in the unit area can also be linearly fitted, and the average of the above two intermediate values ​​is used as the amount of paint sprayed at the center point of the unit area, and then the amount of paint sprayed in all unit areas is fitted.

[0092] In detail, the calculation of the amount of paint sprayed in each unit area of ​​the unit area set according to the original speed control parameter and the original pressure control parameter includes:

[0093] Calculate the paint spraying speed according to the preset paint spraying speed formula;

[0094] According to a preset unit paint quantity calculation formula, the paint quantity spraying speed is used to calculate the spray paint quantity of each unit area in the unit area set.

[0095] Furthermore, the paint spraying speed refers to the paint spraying speed of the nozzle.

[0096] Explainably, the paint spraying speed formula can be obtained by the momentum theorem, as shown below:

[0097]

[0098] Among them, v p represents the paint spraying speed, f represents the nozzle pressure, t represents the time, m represents the paint mass, p represents the nozzle pressure, d represents the nozzle diameter, and ρ represents the paint density.

[0099] In detail, the unit paint amount calculation formula is as follows:

[0100]

[0101] Wherein, q represents the amount of paint sprayed in each unit area of ​​the unit area set, L represents the length of the unit area along the spraying trajectory in the spraying area, l represents the side length of the unit area, and v y It represents the moving speed of the nozzle along the spraying trajectory, s represents the area of ​​the unit area, and S represents the area of ​​the spraying area.

[0102] It is understandable that the amount of paint sprayed in each unit area can be calculated by first calculating the area ratio of the unit area to the spraying area, then calculating the amount of paint sprayed in the spraying area according to the paint spraying speed and the time it takes to pass through the unit area, and finally calculating the amount of paint sprayed in the unit area according to the area ratio and the amount of paint sprayed in the spraying area. Here, the spraying area under different pressures should be approximately constant, and the amount of paint sprayed at different positions in the same spraying area in the same period should be approximately the same.

[0103] S5. Obtain standard paint layer parameters corresponding to the component model, and use a pre-constructed difference calculation formula to perform feedback adjustment on the original neural network model according to the original paint layer parameters and the standard paint layer parameters to obtain a target neural network.

[0104] It can be explained that the standard paint layer parameters refer to the specified spray paint quantity at each position in the component model, for example: the paint surface thickness, paint usage and other data specified at each position.

[0105] In detail, the difference calculation formula is as follows:

[0106]

[0107] Among them, c i,j,k represents the difference value of the index parameter between the standard paint layer parameter and the original paint layer parameter at the corresponding spatial coordinate (i, j, k), n represents the coordinate number of the standard paint layer parameter and the original paint layer parameter in the unit area containing the spatial coordinate (i, j, k), N represents the number of coordinate numbers in the unit area, x represents the index parameter of the standard paint layer parameter at the spatial coordinate (i, j, k), y represents the index parameter of the original paint layer parameter at the spatial coordinate (i, j, k), and N spatial coordinates (i, j, k) are equidistantly distributed in the unit area.

[0108] Explainably, the index parameter may be a paint thickness index, a paint chromaticity index, a paint usage index, etc., and each unit area may select a certain number of spatial coordinates spaced at a certain distance for calculation. When the difference value is greater than a preset difference threshold, it indicates that the calculation of the corresponding unit area still needs feedback adjustment; when the difference value is less than or equal to the difference threshold, it indicates that the calculation of the corresponding unit area has completed feedback adjustment. When all unit areas in the automotive parts have completed feedback adjustment, it indicates that the target neural network is obtained.

[0109] In the embodiment of the present invention, the pre-constructed difference calculation formula is used to feedback-adjust the original neural network model according to the original paint layer parameters and the standard paint layer parameters to obtain the target neural network, including:

[0110] Extracting corresponding index parameters from the original paint layer parameters and the standard paint layer parameters according to the preset paint layer index;

[0111] Calculating the difference value according to the original paint layer parameters and the corresponding index parameters in the standard paint layer parameters by using the difference calculation formula;

[0112] The original neural network model is subjected to negative feedback adjustment according to the difference value to obtain a target neural network.

[0113] It is understandable that the paint layer index refers to an index describing the amount of paint sprayed at various positions in the automotive parts, such as paint thickness index, paint chromaticity index, paint usage index, etc. The paint chromaticity can be calculated by the amount of paint sprayed. Generally, the greater the amount of paint sprayed, the darker the paint chromaticity.

[0114] It should be understood that in the process of performing negative feedback adjustment on the original neural network model, the difference value should be required to be less than or equal to a preset difference threshold. When it is greater than the difference threshold, it indicates that negative feedback adjustment is required for the original neural network model to optimize the internal weight parameters.

[0115] S6. Obtain a current feature set of the component to be painted, input the current feature set into the target neural network, and obtain target control parameters.

[0116] It should be understood that the current feature set refers to the appearance feature set of the part to be painted, such as shape, size, surface roughness, etc. The target control parameters refer to the corresponding optimal trajectory control parameters, optimal distance control parameters, optimal angle control parameters, optimal speed control parameters and optimal pressure control parameters of the part to be painted.

[0117] In an embodiment of the present invention, after training, the target neural network should have the function of realizing specific trajectory control parameters, distance control parameters, angle control parameters, speed control parameters and pressure control parameters corresponding to appearance features such as different shapes, sizes, surface roughness, etc.

[0118] S7. Spray paint on the parts to be painted according to the target control parameters to complete intelligent painting of automobile parts.

[0119] In an embodiment of the present invention, the step of spraying paint on the component to be painted according to the target control parameter includes:

[0120] Calibrate the target control parameters according to a preset start time to obtain a standard control parameter set;

[0121] The pre-built nozzle is regulated according to the standard control parameter set to spray the parts to be painted.

[0122] It can be explained that various parameters in the target control parameters should be values ​​that change with time, so various parameters at the same time should be calibrated. For example: the trajectory control parameters, distance control parameters, angle control parameters, speed control parameters and pressure control parameters in the spraying process start to change at the same time and end at the same time.

[0123] The embodiment of the present invention is to solve the problem described in the background technology. The present invention mainly uses a neural network to establish a correct correspondence between automobile parts and target control parameters for controlling spraying. Before establishing the correct correspondence between the automobile parts and the target control parameters, it is necessary to first collect the characteristics of the automobile parts. Therefore, it is necessary to first build a part model based on the spatial feature data of the automobile parts, and then extract the model index data set of the part model according to a predetermined feature index set classification. After obtaining the model index data set, it is necessary to input it into a pre-built original neural network model to obtain the original control parameters. In the process of adjusting the original neural network model, it is necessary to correspond to the original control parameters. Feedback adjustment is performed based on the difference between the spraying effect of the automobile parts and the correct spraying effect of the automobile parts. Therefore, it is necessary to model the paint surface on the part model according to the original control parameters to obtain the original paint surface model, and then extract the original paint layer parameters of the original paint surface model, so as to realize the conversion of the spraying effect corresponding to the original control parameters, and then convert the correct spraying effect of the automobile parts into the standard paint layer parameters, and then the original paint layer parameters and the standard paint layer parameters can be used to feedback adjust the original neural network model to obtain the target neural network, and finally, the current feature set is input into the target neural network to obtain the target control parameters, and the paint surface of the parts to be painted is sprayed according to the target control parameters. Therefore, the present invention can solve the problem that it is difficult for machines to achieve accurate spraying of various parts of automobile parts during the process of automatic spraying of paint, which often causes uneven spraying and large color differences.

[0124] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for intelligent painting of automobile parts based on neural network, characterized in that: The method comprises: Collecting pre-built spatial feature data of automobile parts, and building a part model according to the spatial feature data; Extracting a model index data set of the component model according to a predetermined feature index set; Inputting the model indicator data set into a pre-built original neural network model to obtain original control parameters, wherein the original neural network includes an input layer, a hidden layer, and an output layer; Performing paint surface modeling on the component model according to the original control parameters to obtain an original paint surface model, and extracting original paint layer parameters of the original paint surface model; The standard paint layer parameters corresponding to the component model are obtained, and the pre-built difference calculation formula is used to feedback-adjust the original neural network model according to the original paint layer parameters and the standard paint layer parameters to obtain the target neural network, wherein the difference calculation formula is as follows: Among them, c i,j,k represents the difference value of the index parameter of the standard paint layer parameter and the original paint layer parameter at the corresponding spatial coordinate (i, j, k), n represents the coordinate serial number of the standard paint layer parameter and the original paint layer parameter in the unit area containing the spatial coordinate (i, j, k), N represents the number of coordinate serial numbers in the unit area, x represents the index parameter of the standard paint layer parameter at the spatial coordinate (i, j, k), y represents the index parameter of the original paint layer parameter at the spatial coordinate (i, j, k), and N spatial coordinates (i, j, k) are equidistantly distributed in the unit area; Acquire a current feature set of the component to be painted, input the current feature set into the target neural network, and obtain target control parameters; The paint surface of the parts to be painted is sprayed according to the target control parameters to complete the intelligent painting of automobile parts.

2. The intelligent painting method for automobile parts according to claim 1, characterized in that: The collecting of pre-built spatial feature data of automobile parts includes: Constructing a local coordinate system according to the center position of the automobile component; A feature point set of the automobile component is identified, and spatial feature data of the feature point set is extracted in the local coordinate system.

3. The intelligent painting method for automobile parts according to claim 2, characterized in that: The step of inputting the model indicator data set into the pre-built original neural network model to obtain the original control parameters includes: Normalizing each model indicator data in the model indicator data set to obtain a standard indicator data set; Performing weight initialization on the input layer and the hidden layer in the original neural network to obtain initial input layer weights and initial hidden layer weights; According to the correspondence between the feature indicators in the feature indicator set and the nodes in the input layer, the standard indicator data set is input into the original neural network model, and the standard indicator data set is weighted summed using the initial input layer weights to obtain the original input sum value; Activating the original input sum value using a pre-built activation function to obtain an original hidden value; Performing weighted summation on the original hidden values ​​using the initial hidden layer weights to obtain an original hidden sum value; The activation function is used to activate the original hidden sum value to obtain the original control parameter.

4. The intelligent painting method for automobile parts according to claim 3, characterized in that: The step of performing paint surface modeling on the component model according to the original control parameters to obtain the original paint surface model includes: Sequentially extracting original trajectory control parameters, original distance control parameters, original angle control parameters, original speed control parameters and original pressure control parameters from the original control parameters; Calculate the spraying area according to the original trajectory control parameter, the original distance control parameter and the original angle control parameter; Calculating the amount of paint sprayed in the spraying area according to the original speed control parameter and the original pressure control parameter; Calculating the paint surface parameters of the spraying area according to the sprayed paint amount; The local coordinate system is used to perform paint surface modeling on the component model according to the paint surface parameters to obtain the original paint surface model.

5. The intelligent painting method for automobile parts according to claim 4, characterized in that: The calculating the amount of paint sprayed in the spraying area according to the original speed control parameter and the original pressure control parameter includes: A set of unit areas selected in the spraying area according to a predetermined spacing; Calculating the amount of paint sprayed in each unit area of ​​the unit area set according to the original speed control parameter and the original pressure control parameter; The paint spraying amount of each unit area in the unit area set is fitted to obtain the paint spraying amount of the spraying area.

6. The intelligent painting method for automobile parts according to claim 5, characterized in that: The step of calculating the amount of paint sprayed in each unit area of ​​the unit area set according to the original speed control parameter and the original pressure control parameter includes: Calculate the paint spraying speed according to the preset paint spraying speed formula; According to a preset unit paint quantity calculation formula, the paint quantity spraying speed is used to calculate the spray paint quantity of each unit area in the unit area set.

7. The intelligent painting method for automobile parts according to claim 6, characterized in that: The paint spraying speed formula is as follows: Among them, v p represents the paint spraying speed, f represents the nozzle pressure, t represents the time, m represents the paint mass, p represents the nozzle pressure, d represents the nozzle diameter, and ρ represents the paint density.

8. The intelligent painting method for automobile parts according to claim 6, characterized in that: The unit paint volume calculation formula is as follows: Wherein, q represents the amount of paint sprayed in each unit area of ​​the unit area set, L represents the length of the unit area along the spraying trajectory in the spraying area, l represents the side length of the unit area, and v y It represents the moving speed of the nozzle along the spraying trajectory, s represents the area of ​​the unit area, and S represents the area of ​​the spraying area.

9. The intelligent painting method for automobile parts according to claim 3, characterized in that: The method of using the pre-constructed difference calculation formula to feedback-adjust the original neural network model according to the original paint layer parameters and the standard paint layer parameters to obtain a target neural network includes: Extracting corresponding index parameters from the original paint layer parameters and the standard paint layer parameters according to the preset paint layer index; Calculating the difference value according to the original paint layer parameters and the corresponding index parameters in the standard paint layer parameters by using the difference calculation formula; The original neural network model is subjected to negative feedback adjustment according to the difference value to obtain a target neural network.

10. The intelligent painting method for automobile parts according to claim 1, characterized in that: The step of spraying paint on the component to be painted according to the target control parameters includes: Calibrate the target control parameters according to a preset start time to obtain a standard control parameter set; The pre-built nozzle is regulated according to the standard control parameter set to spray the parts to be painted.

Images