Spraying system and method based on artificial intelligence algorithm
Through the spraying system based on artificial intelligence algorithm, the deep neural network model is used to determine the spraying quality and adjust the parameters, the problems of low efficiency and poor quality consistency of traditional artificial spraying are solved, and efficient automation and quality improvement of the spraying system are achieved, reducing health harm.
Patent Information
- Application Number
- CN202411207607.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-30
- Publication Date
- 2024-11-12
AI Technical Summary
During traditional manual spraying, the spraying efficiency is low and the quality consistency is poor. When robots perform automatic spraying operations, they are prone to sudden paint flow changes and unclear boundaries, which cannot effectively solve the problem of unclear spraying boundaries and contours.
The spraying system based on artificial intelligence algorithms is adopted, including robotics, visual recognition devices, content storage devices, dynamic adjustment devices, spray guns, information extraction modules and instruction generation modules, and spray quality judgment and parameter adjustment are carried out through the deep neural network model, and real-time monitoring and Optimize the spraying process to ensure that the spraying quality meets process requirements.
It improves the operating efficiency and spray quality of the spray system, effectively solves problems such as sudden paint flow and unclear boundaries, improves the degree of automation and material utilization, and reduces the harm to the health of sprayers.
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of automated painting, and in particular relates to a painting system and method based on an artificial intelligence algorithm. Background Art
[0002] Since spray painting can protect equipment from corrosion by corrosive media such as moisture and oxygen, extend the life of equipment, beautify the appearance of equipment, and enhance the aesthetics of equipment, it is an indispensable part of modern manufacturing. The initial spraying method was manual spraying, which not only seriously endangers the health of sprayers, but also affects the quality and efficiency of spraying due to factors such as the proficiency of the operator. With the continuous improvement of equipment spraying requirements, clear regulations have been made on the basic units, area ratios, spot configurations and coating requirements in the spraying process, especially the wet film thickness, the clarity of the clumps, and the degree of pollution-free adjacent color blocks. The use of intelligent technology to achieve fast and efficient spraying of patterns is a development trend in the manufacturing industry.
[0003] Faced with the requirements of pattern spraying with multiple colors, clear boundaries and no color mixing, manual spraying is currently mostly used. In the process of manual spraying, first draw the baseline and positioning lines according to the design drawings; design and manufacture the templates with corresponding numbers according to the design drawings and place them in the corresponding positions of the equipment; tightly combine the template with the surface of the equipment to avoid paint splashing; spray the main spots first, and then spray the modified spots. The manual spraying method has problems such as low spraying efficiency and high labor intensity. The quality control of spraying and the effective use of materials all depend on the experience accumulation of sprayers. At the same time, the spraying industry belongs to harmful types of work. The paint splashed during the spraying process and the waste gas produced seriously endanger the health of the sprayers. Although with the development of technology, robots are used for automatic spraying operations, in the face of complex surface structures of equipment, industrial robots are prone to sudden changes in the fluidity and viscosity of the paint during the spraying process (i.e., paint flow mutation phenomenon), resulting in defects and problems such as color mixing at the equipment spraying interface.
[0004] The existing robot spraying control method, device, equipment and storage medium obtains the time taken to open the spray gun and the time taken to close the spray gun, and calculates the opening movement distance and the closing movement distance according to the time taken to open the spray gun, the time taken to close the spray gun and the preset spraying speed of the spraying robot. This solution avoids the situation of less or more spraying at the edge of the spray gun as much as possible, but it cannot effectively solve the problem of unclear spraying boundaries and contours when facing problems such as uneven spraying surfaces and complex structures.
[0005] There is also a spraying robot control method, which identifies the target object to be processed, and when the characteristic information in the electronic tag carried by the target object to be processed is identified, the spraying control program is called, and based on the control program and the paint number required by the target object to be processed, the spraying robot is controlled to change color and spray. This solution can improve the automation of the spraying operation of the spraying robot, but it cannot solve the problems such as sudden changes in paint flow during the execution of the spraying program. Summary of the invention
[0006] The present invention provides a spraying system and method based on an artificial intelligence algorithm. The technical problem to be solved is: to solve the problems of low spraying efficiency and poor quality consistency in the traditional manual spraying process, and at the same time to effectively improve the problems of sudden paint flow and unclear boundaries that occur during the robot's automatic spraying operation, thereby improving the operating efficiency and spraying quality of the spraying system.
[0007] In order to solve the above technical problems, the present invention provides a spraying system based on artificial intelligence algorithm, which is characterized by: comprising a manipulator, a visual recognition device, a content storage device, a dynamic adjustment device, a spray gun, an information extraction module, a data analysis module and an instruction generation module;
[0008] The manipulator can move to a specified position in space so that the manipulator and the surface of the equipment to be sprayed are kept in a fixed position;
[0009] A visual recognition device, fixed on the manipulator, is used to extract and recognize features on the surface of the equipment before and during spraying, and transmit the image feature data after visual recognition to the content storage device for storage;
[0010] An information extraction module is used to extract the equipment surface spraying image data in the content storage device to obtain the boundary and contour information of the different color spraying units on the equipment surface, and send it to the data analysis module;
[0011] The data analysis module is used to analyze the spraying image data of the equipment surface to obtain the boundary and contour information of different colors on the equipment surface, so as to obtain the overall quality information of the spraying in the spraying image data area, timely discover the quality problems in the spraying process, and timely feedback to the instruction generation module;
[0012] The instruction generation module is used to evaluate the quality problems that arise, adjust the spatial position of the robot and the flow rate of the paint, and feed back the adjusted instructions to the dynamic adjustment device, thereby controlling the spray gun to spray the paint onto the equipment surface at a preset spraying speed, controlling the flow rate of the paint during the spraying process, and the position of the spray robot, thereby controlling the posture of the spray gun.
[0013] A spraying method based on artificial intelligence algorithm, characterized in that the specific steps are as follows:
[0014] S1. Design a deep neural network model, comprising a single input layer, a single output layer, and a plurality of hidden layers between the single input layer and the single output layer;
[0015] The input layer can receive various information of the spray target;
[0016] The number and structure of hidden layers should be designed according to the complexity and diversity of the spraying task;
[0017] The output layer should be able to output the adjusted paint flow rate and robot spatial position parameters;
[0018] S2. Determine whether the coating quality in the area of the sprayed image data meets the process requirements, obtain the boundary clarity, contour clarity, and layer uniformity information of different colors, and establish clear spray quality judgment standards to automatically evaluate whether the spray quality meets the process requirements;
[0019] S3. Obtain the shapes of different regions of each spray unit image, the paint flow rate, paint temperature, and manipulator spatial position parameters of different color spraying processes when spraying different color paints, and take the ambient temperature and humidity as part of the reference parameters;
[0020] S4. Use the spraying effects and corresponding parameters that meet the process requirements in steps S2 and S3 as labeled data to train the network, optimize the network weights, and monitor the loss function during the training process and the performance on the validation set to adjust the learning rate and avoid overfitting;
[0021] S5. Based on the collected data and the established spraying quality judgment standard, a supervised learning intelligent algorithm is used to predict and adjust the paint flow rate and the spatial position of the manipulator; the number of supervised learning times is positively correlated with the complexity of the spraying surface pattern, ensuring that the network can fully learn the spraying rules of various complex patterns; the specific adjustment strategy is: in each learning, the spraying image boundary clarity, spraying image contour clarity, spraying layer uniformity, spraying coating thickness, manipulator position information, spraying color, paint flow rate, paint temperature, ambient temperature and humidity, and whether the spraying meets the process requirements at a certain time of past data collection are used as a complete data set as input content to ensure that the network can learn the mapping relationship between the complete spraying process and results;
[0022] S6. Apply the adjusted parameters to the new spraying task, and evaluate the spraying quality here, forming a closed loop and continuously optimizing the adjustment strategy;
[0023] S7. Apply adjustment strategies in real time during the spraying process to ensure that the spraying quality always remains at an acceptable level.
[0024] Beneficial effects: The present invention can use artificial intelligence algorithms to visually identify the surface spraying quality of equipment, evaluate and plan the spraying path and spraying trajectory, improve the degree of automation of robot spraying operations, effectively solve problems such as paint flow mutations and unclear boundaries that occur during the execution of the spraying program, and improve the operating efficiency and spraying quality of the spraying system. DETAILED DESCRIPTION
[0025] In order to make the purpose, content and advantages of the present invention more clear, the specific implementation methods of the present invention are further described in detail below.
[0026] The present invention proposes a spraying system based on an artificial intelligence algorithm, which includes a manipulator, a visual recognition device, a content storage device, a dynamic adjustment device, a paint delivery pipeline, a spray gun, an information extraction module, a data analysis module and an instruction generation module.
[0027] The manipulator can execute a preset execution program and move to a specified position in space, so that the manipulator and the surface of the equipment to be sprayed remain in a fixed position;
[0028] The visual recognition device is fixed on the manipulator and is used to extract and recognize features on the surface of the equipment before and during spraying. Specifically, the device is used to: first, recognize the position information of each spray unit in the spraying target; second, collect and obtain the effect image of the spraying on the surface of the equipment during and after spraying, and extract and recognize features of the spraying pattern on the surface of the equipment based on the effect image, and finally transmit the image feature data after visual recognition to the content storage device for storage;
[0029] An information extraction module is used to extract the equipment surface spraying image data in the content storage device to obtain the boundary and contour information of the different color spraying units on the equipment surface, and send it to the data analysis module;
[0030] The data analysis module is used to analyze the spraying image data of the equipment surface to obtain the boundary and contour information of different colors on the equipment surface, so as to obtain the overall spraying quality information in the spraying image data area, timely discover the quality problems such as sagging, unclear boundaries and contours caused by sudden changes in paint flow during the spraying process, and timely feedback to the instruction generation module;
[0031] The instruction generation module is used to evaluate quality problems such as sagging, unclear boundaries and contours through artificial intelligence algorithms, adjust the spatial position of the robot and the flow rate of the paint, and feed back the adjusted instructions to the dynamic adjustment device, thereby controlling the spray gun to spray the paint onto the equipment surface at a preset spraying speed, controlling the flow rate of the paint during the spraying process, and the position of the spray robot, thereby controlling the posture of the spray gun.
[0032] The dynamic adjustment device is connected to the manipulator and can control the displacement of the manipulator: the dynamic adjustment device can receive instructions from the instruction generation module and control the spray gun to spray the paint onto the equipment surface at a preset spray speed; at the same time, it controls the flow rate of the paint during the spraying process and the position of the spray manipulator to control the posture of the spray gun, that is, the distance between the spray gun and the equipment spraying surface and the adjustment of the spatial position of the spray gun. At the same time, the dynamic adjustment device can detect and sense the flow rate of the paint through the flow rate sensor installed on the device.
[0033] The paint delivery pipeline is connected to the spray gun and fixed on the manipulator to deliver the paint to the spray gun; there is a temperature sensor in the pipeline to detect the temperature of the paint in the paint delivery pipeline.
[0034] The spray gun is connected to the robot, and the paint is sprayed onto the equipment surface through the spray gun.
[0035] The artificial intelligence algorithm is implemented by using a deep neural network that has completed multiple learnings to perform the intelligent evaluation, and the number of learnings of the multiple learnings is proportional to the complexity of the surface pattern of the spraying equipment.
[0036] According to a second aspect of the present invention, a personalized spraying method based on an artificial intelligence algorithm is provided, the method comprising the following steps:
[0037] S1. Design a deep neural network model, including a single input layer, a single output layer, and multiple hidden layers between the single input layer and the single output layer.
[0038] The input layer can receive a variety of information about the spraying target, including the position coordinates and rotation angle of the robot when each spraying unit image is formed, the spraying shape of the spraying area, the type of spraying color, the paint flow rate, the ambient temperature and humidity, and whether the spraying effect image meets the process requirements.
[0039] The number and structure of hidden layers should be designed according to the complexity and diversity of spraying tasks. Multi-layer convolutional neural networks (CNNs) are used to process spraying effect images and extract image features. For spraying tasks with many colors and complex shapes, the number of hidden layers and the number of neurons in each layer should be increased to enhance the network learning and generalization capabilities. For non-image data such as robot position information, spraying color, paint flow rate, and paint temperature, fully connected layers (Dense Layers) are used for feature learning and conversion, and batch normalization is used to prevent overfitting.
[0040] The output layer should be able to output the adjusted paint flow rate, robot spatial position parameters (including robot spatial coordinates, rotation angle), etc.
[0041] S2. Determine whether the paint spraying quality in the image data area after spraying meets the process requirements, obtain information such as boundary clarity, contour clarity, layer uniformity, etc. of different colors, and establish clear spraying quality judgment standards, such as boundary clarity threshold, coating thickness range, etc., so as to automatically evaluate whether the spraying quality meets the process requirements.
[0042] S3. Obtain the shapes of different regions of each spray unit image, the paint flow rate, paint temperature, and manipulator spatial position parameters of different color spraying processes when spraying different colors of paint, and take the ambient temperature and humidity as part of the reference parameters.
[0043] S4. Use the spraying effects and corresponding parameters in S2 and S3 that meet the process requirements as labeled data to train the network. At the same time, use the back propagation algorithm and gradient descent to optimize the network weights, and monitor the loss function during training and the performance on the validation set to adjust the learning rate and avoid overfitting.
[0044] S5. Based on the collected data and the established spray quality judgment criteria, a supervised learning intelligent algorithm is used to predict and adjust the paint flow rate and the spatial position of the robot. The number of supervised learning times is positively correlated with the complexity of the spray surface pattern, ensuring that the network can fully learn the spraying rules of various complex patterns. The specific adjustment strategy is: in each learning, the spray image boundary clarity, spray image contour clarity, spray layer uniformity, spray coating thickness, robot position information, spray color, paint flow rate, paint temperature, ambient temperature and humidity, and whether the spraying meets the process requirements at a certain time of past data collection are used as a complete data set as input content to ensure that the network can learn the mapping relationship between the complete spraying process and results, thereby improving the learning effect and accuracy.
[0045] S6. Apply the adjusted parameters to new spraying tasks and evaluate the spraying quality here to form a closed loop and continuously optimize the adjustment strategy.
[0046] S7. Apply adjustment strategies in real time during the spraying process to ensure that the spraying quality always remains at an acceptable level.
[0047] The present invention adopts a deep neural network that has completed multiple learnings to implement the artificial intelligence algorithm to perform the intelligent evaluation. In order to ensure the reliability and stability of the intelligent evaluation results of the spraying quality, the following two construction measures are taken for the deep neural network that has completed multiple learnings: the number of learning times of the multiple learnings is proportional to the complexity of the surface pattern of the spraying equipment; in each learning performed on the deep neural network, the area shape known at a certain data collection moment in the past, the paint flow rate of the spraying process, the paint temperature, the spatial position of the manipulator, and the boundary and contour clarity of the spraying unit are used as the input content of the deep neural network after completing multiple learnings, thereby ensuring the effect of each learning.
[0048] The number of the multiple hidden layers between the single input layer and the single output layer of the present invention is positively correlated with the shape and color type of the spraying unit in the set area, thereby realizing intelligent evaluation of the spraying quality of different shapes and colors based on the artificial intelligence algorithm.
[0049] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A spraying system based on artificial intelligence algorithm, characterized by: It includes a manipulator, a visual recognition device, a content storage device, a dynamic adjustment device, a spray gun, an information extraction module, a data analysis module and an instruction generation module; The manipulator can move to a specified position in space so that the manipulator and the surface of the equipment to be sprayed are kept in a fixed position; A visual recognition device, fixed on the manipulator, is used to extract and recognize features on the surface of the equipment before and during spraying, and transmit the image feature data after visual recognition to the content storage device for storage; An information extraction module is used to extract the equipment surface spraying image data in the content storage device to obtain the boundary and contour information of the different color spraying units on the equipment surface, and send it to the data analysis module; The data analysis module is used to analyze the spraying image data of the equipment surface to obtain the boundary and contour information of different colors on the equipment surface, so as to obtain the overall quality information of the spraying in the spraying image data area, timely discover the quality problems in the spraying process, and timely feedback to the instruction generation module; The instruction generation module is used to evaluate the quality problems that arise, adjust the spatial position of the robot and the flow rate of the paint, and feed back the adjusted instructions to the dynamic adjustment device, thereby controlling the spray gun to spray the paint onto the equipment surface at a preset spraying speed, controlling the flow rate of the paint during the spraying process, and the position of the spray robot, thereby controlling the posture of the spray gun.
2. The spraying system based on artificial intelligence algorithm according to claim 1, characterized in that: The visual recognition device is used to firstly identify the position information of each spraying unit in the spraying target; secondly, it is used to collect and obtain the effect image of the equipment surface spraying during and after the spraying process, and extract and identify the features of the surface spraying pattern of the equipment according to the effect image, and finally transmit the image feature data after visual recognition to the content storage device for storage.
3. The spraying system based on artificial intelligence algorithm according to claim 1, characterized in that: The quality issues mentioned include sagging caused by sudden changes in paint flow, and unclear boundaries and outlines.
4. The spraying system based on artificial intelligence algorithm according to claim 1, characterized in that: The dynamic adjustment device is connected to the manipulator and can control the displacement of the manipulator; the dynamic adjustment device can detect and sense the flow rate of the paint through a flow rate sensor.
5. The spraying system based on artificial intelligence algorithm according to claim 1, characterized in that: The dynamic adjustment device can receive instructions from the instruction generation module and control the spray gun to spray the paint onto the equipment surface at a preset spraying speed; at the same time, it controls the flow rate of the paint during the spraying process and the position of the spraying robot to control the posture of the spray gun, that is, the distance between the spray gun and the equipment spraying surface and the adjustment of the spatial position of the spray gun.
6. The spraying system based on artificial intelligence algorithm according to claim 1, characterized in that: There is a temperature sensor in the paint delivery pipeline to detect the temperature of the paint in the paint delivery pipeline.
7. The spraying system based on artificial intelligence algorithm according to claim 1, characterized in that: A paint delivery pipeline and a spray gun are also fixed on the manipulator, and the paint delivery pipeline is connected to the spray gun to deliver the paint to the spray gun.
8. A spraying method based on artificial intelligence algorithm, characterized in that: The specific steps are as follows: S1. Design a deep neural network model, comprising a single input layer, a single output layer, and a plurality of hidden layers between the single input layer and the single output layer; The input layer can receive various information of the spray target; The number and structure of hidden layers should be designed according to the complexity and diversity of the spraying task; The output layer should be able to output the adjusted paint flow rate and robot spatial position parameters; S2. Determine whether the coating quality in the area of the sprayed image data meets the process requirements, obtain the boundary clarity, contour clarity, and layer uniformity information of different colors, and establish clear spray quality judgment standards to automatically evaluate whether the spray quality meets the process requirements; S3. Obtain the shapes of different regions of each spray unit image, the paint flow rate, paint temperature, and manipulator spatial position parameters of different color spraying processes when spraying different color paints, and take the ambient temperature and humidity as part of the reference parameters; S4. Use the spraying effects and corresponding parameters that meet the process requirements in steps S2 and S3 as labeled data to train the network, optimize the network weights, and monitor the loss function during the training process and the performance on the validation set to adjust the learning rate and avoid overfitting; S5. Based on the collected data and the established spraying quality judgment standard, a supervised learning intelligent algorithm is used to predict and adjust the paint flow rate and the spatial position of the manipulator; the number of supervised learning times is positively correlated with the complexity of the spraying surface pattern, ensuring that the network can fully learn the spraying rules of various complex patterns; the specific adjustment strategy is: in each learning, the spraying image boundary clarity, spraying image contour clarity, spraying layer uniformity, spraying coating thickness, manipulator position information, spraying color, paint flow rate, paint temperature, ambient temperature and humidity, and whether the spraying meets the process requirements at a certain time of past data collection are used as a complete data set as input content to ensure that the network can learn the mapping relationship between the complete spraying process and results; S6. Apply the adjusted parameters to the new spraying task, and evaluate the spraying quality here, forming a closed loop and continuously optimizing the adjustment strategy; S7. Apply adjustment strategies in real time during the spraying process to ensure that the spraying quality always remains at an acceptable level.
9. A spraying method based on artificial intelligence algorithm according to claim 8, characterized in that: The various information of the spraying target includes the position coordinates and rotation angle of the robot when each spraying unit image is formed, the spraying shape of the spraying area, the type of spraying color, the paint flow rate, the ambient temperature and humidity, and whether the spraying effect image meets the process requirements.
10. The spraying method based on artificial intelligence algorithm according to claim 8, characterized in that: The number of times the deep neural network learns multiple times is proportional to the complexity of the surface pattern of the spraying system.
Citation Information
Cited By
Accurate positioning and spraying thickness control system in airplane paint spraying hangar
CN119293101A
Large casting multi-robot cooperative spraying control method based on self-adaption
CN119387122A
