Method for optimizing mechanical arm cleaning of digital twin system through particle swarm optimization

Abstract

The invention discloses a method for optimizing mechanical arm cleaning of a digital twin system through a particle swarm algorithm, and belongs to the technical field of digital technologies and engineering. The method mainly aims at a ship intelligent workshop and comprises the steps that data conversion and arrangement are conducted on the state, collected by a sensor, of a mechanical arm through a data fusion interaction module, and a self-adaptive function is generated through converted information; parameter information of the digital twin system needing to be optimized serves as optimization parameters in the particle swarm algorithm; and reading the adaptive function by adopting a particle swarm algorithm, and mapping parameters of the digital twin system to a target function value for evaluating the performance of each particle. When the method is used in a shipyard, the labor cost can be reduced, the working efficiency can be improved, and the accident rate can be reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of digital technology and engineering, and in particular to a method for optimizing robot arm cleaning of a digital twin system using a particle swarm algorithm. Background Art

[0002] Digital twins make full use of data such as physical models, sensor updates, and operation history, integrate multi-disciplinary, multi-physical quantity, multi-scale, and multi-probability simulation processes, complete mapping in virtual space, and thus reflect the entire life cycle of the corresponding physical equipment; it is a technology that uses digital technology to comprehensively and accurately simulate and map physical entities. The structure, behavior, and performance of physical entities are twinned, and simulation is used to simulate their operating status under different conditions. In smart factories, various sensors such as position sensors, pressure sensors, and flow sensors are installed at key locations such as the joints of the robotic arm, nozzles, and cleaning fluid delivery pipelines. How to transmit the real-time data collected by the sensor to the digital twin model at a high speed and stably and ensure the consistency of the model with the state of the physical entity is a problem that needs to be solved in the digital twin process.

[0003] The current digital twin system has problems such as large amount of data to be read and various contents, which makes the operation of the visual twin system unstable. It is usually necessary to optimize the data imported into the visual twin system, but the traditional particle swarm algorithm has weak global optimization ability in the early stage and cannot find suitable seeds; and weak local optimization ability in the later stage and cannot improve the convergence accuracy. Summary of the invention

[0004] In order to solve the problems existing in the prior art, the present invention provides a method for optimizing the cleaning of the robot arm of the digital twin system by using a particle swarm algorithm, establishes an improved particle swarm algorithm mathematical model, and improves the traditional particle swarm algorithm by three weight coefficients; compared with the traditional algorithm, the improved algorithm has a strong global optimization capability in the early stage, is easy to obtain suitable seeds, and has a strong local optimization capability in the later stage, is easy to improve the convergence accuracy, thereby meeting the stability of the algorithm. The data collected by the sensor system can be stably transmitted to the visual twin system.

[0005] The present invention uses a robotic arm motor driver, sensor, inverter, servo amplifier and control circuit, and combines the cleaning system, the intelligent algorithm fusion system for the real-time operation of the parts to be cleaned and the robotic arm, and the robotic arm intelligent network communication and monitoring intelligent system to form a method in which the sensor communication module collects data, the main controller module analyzes and processes data, the intelligent system performs signal processing and outputs instructions to the robotic arm cleaning equipment, the robotic arm executes actual actions, and finally realizes digital twin visualization in the form of a smart robotic arm cleaning system, thereby accurately calculating the cleaning efficiency of the smart factory, monitoring the working information of the equipment, reducing the waste of water and electricity resources and environmental pollution, and reducing accidents caused by production.

[0006] The specific technical solution adopted by the present invention is as follows:

[0007] Step 1: The parts are placed in the area to be cleaned, the robotic arm hardware model issues instructions, and the joints of the robotic arm start to move;

[0008] Step 2: At the same time, sensors at each joint of the robotic arm begin to collect signals;

[0009] Step 3: The sensor starts working after receiving the signal, identifies and detects the working status of each joint of the robot arm, and transmits the working status information of the robot arm to the data fusion and interaction module;

[0010] Step 4: The robot arm cleaning model is equipped with a sensor to flush the cleaning agent onto the surface of the part, and waits for the surface cleaning of the part to be completed. The sensor transmits and stores the cleaning speed information to the data fusion and interaction module;

[0011] Step 5: After the robot arm finishes cleaning, the information is transmitted to the virtual space. The virtual space receives the signal and feeds it back to the robot arm drying system. The robot arm drying system receives the instruction and starts working until the surface of the part is dried. The drying status and part status information collected by the sensor are transmitted to the data fusion interaction module.

[0012] Step 6: The working state of the robot arm, the drying state, the cleaning speed, and the part state collected by the sensor are converted and sorted through the data fusion interaction module, and the converted information is used to generate an adaptive function;

[0013] Step 7: Determine the parameter information of the digital twin system that needs to be optimized as the optimization parameter in the particle swarm algorithm; use the particle swarm algorithm to read the adaptive function and map the parameters of the digital twin system to the objective function value to evaluate the performance of each particle;

[0014] Initialize the particle swarm, and each particle is assigned its best position according to its individual pbest and the best position for the entire group gbest to adjust its speed and position;

[0015] Particle position update expression:

[0016] (1)

[0017] Where: is the j-th dimension position of the optimal position of the i-th particle in the current iteration; is the j-th dimension position of the optimal position of the i-1th particle in the current iteration; is the j-th velocity component of the velocity of the i-1th particle at the current iteration;

[0018] Particle group velocity update expression:

[0019] (2)

[0020] Where: To improve the inertia weight, it represents the degree of trust in the current speed direction; is the improved individual learning factor and the improved population learning factor; is a random value in [0, 1]; pbest i is the individual optimal position of the i-th particle; gbest i is the optimal position of the entire group when the i-th particle is in the individual optimal position; is the j-th velocity component of the velocity of the i-th particle at the current iteration; Improved inertia weight function for:

[0021] (3)

[0022] Where: is the maximum inertia weight; is the minimum inertia weight; is the current iteration number; is the total number of iterations;

[0023] The improved individual learning factor is:

[0024] (4)

[0025] The improved population learning factor is:

[0026] (5)

[0027] Where: is the maximum learning factor; is the minimum learning factor;

[0028] According to the updated position, the fitness value of each particle, that is, the value of the objective function, is calculated; according to the fitness value of each particle, the individual position of each particle and the global position of the entire particle swarm are updated;

[0029] According to the iterative update of the current population, determine whether the iterative data is optimal. If it meets the accuracy requirements, stop the iteration and save the global optimal solution. Otherwise, re-judge;

[0030] Step 8: Transfer the global optimal solution to the virtual space;

[0031] Step 9: The operation information, system function, adaptive function, and algorithm optimization information transmitted in the virtual space are used to generate a digital twin platform;

[0032] Step 10: Scan the physical model through sensor technology and directly import it into the digital twin platform to generate a shape twin model;

[0033] Step 11: The visualization system displays the shape twin model and the information optimized using the improved particle swarm algorithm.

[0034] The operating logic of the above method is: the robot arm hardware model and the robot arm software model use physical signal mapping to convey information; the robot arm software and the robot arm cleaning model use signal transmission to convey information; the robot arm hardware model, the robot arm software model, the robot arm cleaning model and the particle swarm optimization algorithm module use algorithm embedding to transmit information; the parameters obtained by the particle swarm optimization algorithm module are fed back to the improved particle swarm optimization algorithm; at the same time, the robot arm hardware model, the robot arm software model, and the robot arm cleaning model are introduced into the improved particle swarm optimization algorithm, the working status of each part of the smart factory is received by the sensor, and the improved particle swarm optimization algorithm is used to optimize the data, and the optimized information is stably transmitted to the smart factory visualization system, and finally displayed on the visualization large screen.

[0035] The hardware model of the robot arm includes a robot arm motor driver module, a robot arm sensor module, a robot arm servo amplifier module, and a robot arm control circuit module. The motor driver module converts electrical energy into mechanical energy to control the rotational and linear motion of the motor, provide energy for the robot arm, and realize the six-degree-of-freedom movement of each joint of the robot arm; the robot arm sensor module is used to identify the temperature, angle, speed, displacement, and light of the robot arm and the cleaning parts to collect and acquire information; the robot arm servo amplifier module is used to transmit the motor signal and amplify or reduce the signal; the control circuit module is used to receive and process the command signal from the operator or the upper control system, and generate the control signal according to the motion state of the robot arm, control the speed, position and strength of the servo motor, etc., and control the output of the servo motor by executing the driver module. It receives the control signal from the controller module, uses the servo amplifier to amplify it and transmits it to the servo motor to drive the robot arm to perform precise position and speed control, and uses the communication module to exchange the state of the robot arm with the upper computer system.

[0036] The robotic arm cleaning model includes a cleaning device module, a control system module, a liquid supply system module, and a blow-drying system module. It is mainly used to spray, brush, blow dry, etc. the parts to be cleaned. It uses the movement of the robotic arm to perform comprehensive, efficient, and automated cleaning operations on the target objects, thereby achieving the purpose of intelligent cleaning.

[0037] The robot software model includes a sensor communication module, a main control program, an execution servo driver, and a robot execution action. The smart factory robot collects, analyzes, processes, and finally integrates data through the sensor communication module; the sorted data is mapped to the large screen of the visualization system through the particle swarm optimization algorithm module; the main control program is used to receive and process the signal, thereby outputting the processed signal instruction, and the output signal instruction is amplified and speed-regulated by the execution servo driver, and the feedback electrical signal is transmitted to the motor, so that the robot performs the corresponding action. Compared with the prior art, the present invention has the following advantages:

[0038] Determine the optimization goal of maximizing system performance of the digital twin system; select optimization variables and determine the parameter information of the digital twin system that needs to be optimized. These variables will become the optimization parameters in the particle swarm algorithm; map the parameters of the digital twin system to the fitness function value to evaluate the performance of each particle. Read the improved weight function to obtain the individual optimal and global optimal solutions, use the improved inertia weight function formula to calculate the particle fitness, use the improved individual learning factor function formula to gradually increase the number of particles and improve the local optimization ability of the particle swarm, and use the improved population learning factor function formula to gradually reduce the number of particles and improve the global optimization ability of the particle swarm. Use the improved particle swarm velocity update function to calculate the particle swarm velocity.

[0039] The larger the inertia weight coefficient is in the early stage, the stronger the global search is. The larger the inertia weight coefficient is in the later stage, the stronger the local search is, and the convergence accuracy is easy to improve. The improved inertia weight function is used to solve the imbalance between global optimality and local optimality when the inertia coefficient is a constant. The improved individual learning factor overcomes the problem that when the individual learning factor is a constant, it is easy to fall into the local optimality, causing the algorithm to tend to local search and lack the ability of global search. The improved population learning factor solves the problem that the population learning factor is too large, ignoring the differences between particles, causing the algorithm to enter the global optimal solution too early and unable to effectively perform local search. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly describe the implementation method of the present invention or the prior art solution, the implementation method or the prior art solution will be described below in the form of drawings and briefly introduced. It can be seen from the drawings that the drawings described below are the implementation methods of the present invention. People in this field can expand on them based on the drawings without involving creative work.

[0041] Figure 1 It is the overall system architecture diagram.

[0042] Figure 2 It is a flow chart of the method of optimizing the robot arm cleaning of the digital twin system using the particle swarm algorithm.

[0043] Figure 3 It is the structural block diagram of the robot arm cleaning system.

[0044] Figure 4 It is a structural block diagram of the cleaning parts warehousing system.

[0045] Figure 5 It is a flowchart of the improved particle swarm algorithm.

[0046] Figure 6 This is the simulation result of the improved particle swarm algorithm.

[0047] Figure 7 This is a large-screen visualization of the cleaning process of the robot arm in a smart factory. DETAILED DESCRIPTION

[0048] It is hereby declared that the implementation methods in the present invention can be combined with each other without conflict. The following is a detailed description of the present invention in combination with the accompanying drawings and the implementation methods.

[0049] In order to make the implementation method of the present invention more clearly described, the following is a detailed technical solution description of the drawings in the implementation method. It is obvious that the described implementation scheme is only a part of the present invention, not all implementation methods. The following will be described in some of the following drawings on the implementation method, but it does not limit the application or use of the present invention. For the implementation method of the present invention, if the personnel in the technical field do not make creative work, they all belong to the protection scope of the present invention.

[0050] It should be noted that the expressions used in the present invention are only for better describing the specific implementation method, and are not intended to limit the implementation methods of this invention. Unless otherwise specified in the present invention, singular expressions also mean plural expressions.

[0051] It should be noted that the parts, systems, operating methods, relative positions, front and rear arrangements, numerical values ​​and mathematical expressions described in the present invention do not limit the scope of protection of the present invention. In addition, in order to clearly express the implementation method of the present invention, the dimensions in the drawings of the present invention are not made in proportion. The techniques, schemes, equipment and drawings that have been clearly understood by ordinary people in the field will not be discussed in detail, but in special cases, the equipment, techniques, schemes and block diagrams described shall be regarded as part of the authorization specification. The implementation methods in the present invention are only exemplary. It should be noted in the present invention that the numbers and letters in the block diagram represent similar items, so once defined in the drawings, the subsequent drawings do not need to be discussed in detail.

[0052] It should be noted that the orientations indicated in the present invention are only based on the directions and positions shown in the accompanying drawings, and are only for the purpose of making the description of the scheme in the present invention clearer. Similarly, the devices or equipment in the present invention do not represent the operation sequence and operation steps. Therefore, it cannot be said that the protection scope of the present invention is limited.

[0053] It should be noted that the order of arrangement of the parts in the present invention is only for the convenience of distinction. If there is no special explanation, the above description has no substantive meaning, and therefore cannot be used to indicate that the object of protection of the present invention is limited.

[0054] The overall system architecture of the present invention is as follows Figure 1As shown in the figure, the mechanical arm hardware model and the mechanical arm software model use physical signal mapping to convey information; the mechanical arm software and the mechanical arm cleaning model use signal transmission to convey information; the mechanical arm hardware model, the mechanical arm software model, the mechanical arm cleaning model and the particle swarm optimization algorithm module use algorithm embedding to convey information; the parameters obtained by the particle swarm optimization algorithm module are fed back to the improved particle swarm optimization algorithm; at the same time, the mechanical arm hardware model, the mechanical arm software model and the mechanical arm cleaning model are introduced into the improved particle swarm optimization algorithm, the working status of each part of the smart factory is received by the sensor, and the improved particle swarm optimization algorithm is used to optimize the data, and the optimized information is stably transmitted to the smart factory visualization system, and finally displayed on the visualization large screen.

[0055] The hardware model of the robot arm includes a robot arm motor driver module, a robot arm sensor module, a robot arm servo amplifier module, and a robot arm control circuit module. The motor driver module converts electrical energy into mechanical energy to provide energy for the robot arm and realize the six-degree-of-freedom movement of the robot arm; the robot arm sensor module is used to identify the temperature, angle, speed, displacement, and light of the robot arm and the cleaning parts to collect and acquire information; the robot arm servo amplifier module is used to transmit the motor signal and amplify or reduce the signal; the control circuit module is used to receive and process the command signal from the operator or the upper control system, and generate the control signal according to the motion state of the robot arm, control the speed, position and strength of the servo motor, etc., and control the output of the servo motor through the execution driver module. It receives the control signal from the controller module, uses the servo amplifier to amplify it and transmits it to the servo motor to drive the robot arm for precise position and speed control, and uses the communication module to exchange the state of the robot arm with the upper computer system.

[0056] The robotic arm cleaning model includes a cleaning device module, a control system module, a liquid supply system module, and a blow-drying system module. It is mainly used to spray, brush, blow dry, etc. the parts to be cleaned. It uses the movement of the robotic arm to perform comprehensive, efficient, and automated cleaning operations on the target objects, thereby achieving the purpose of intelligent cleaning.

[0057] The robot software model includes a sensor communication module, a main control program, an execution servo driver, and a robot execution action. The smart factory robot collects, analyzes, processes, and finally integrates data through the sensor communication module; the sorted data is mapped to the large screen of the visualization system through the particle swarm optimization algorithm module; the main control program is used to receive and process the signal, and then the processed signal instruction is output, and the output signal instruction is amplified and speed-regulated by the execution servo driver, and the feedback electrical signal is transmitted to the motor, so that the robot performs the corresponding action.

[0058] Improve the particle swarm algorithm module, including inertia weight function, population factor function, and individual factor function. The particle swarm algorithm optimizes the robot arm cleaning method of the digital twin system, as follows: Figure 2 As shown, the specific process includes:

[0059] Step 1: The parts are placed in the area to be cleaned, the robotic arm hardware model issues instructions, and the joints of the robotic arm start to move;

[0060] Step 2: At the same time, sensors at each joint of the robotic arm begin to collect signals;

[0061] Step 3: The sensor starts working after receiving the signal, identifies and detects the working status of each joint of the robot arm, and transmits the working status information of the robot arm to the data fusion and interaction module;

[0062] Step 4: The robot arm cleaning model is equipped with a sensor to flush the cleaning agent onto the surface of the part, and waits for the surface cleaning of the part to be completed. The sensor transmits and stores the cleaning speed information to the data fusion and interaction module;

[0063] Step 5: After the robot arm finishes cleaning, the information is transmitted to the virtual space. The virtual space receives the signal and feeds it back to the robot arm drying system. The robot arm drying system receives the instruction and starts working until the surface of the part is dried. The drying status and part status information collected by the sensor are transmitted to the data fusion interaction module.

[0064] Step 6: The working state of the robot arm, the drying state, the cleaning speed, and the part state collected by the sensor are converted and sorted through the data fusion interaction module, and the converted information is used to generate an adaptive function;

[0065] Step 7: Determine the parameter information of the digital twin system that needs to be optimized as the optimization parameter in the particle swarm algorithm; use the particle swarm algorithm to read the adaptive function and map the parameters of the digital twin system to the objective function value to evaluate the performance of each particle. The specific process is as follows: Figure 5 As shown; initialize the particle swarm, and then randomly generate a certain number of particles, each particle represents a parameter combination, that is, the parameter configuration of the digital twin system;

[0066] According to the current position, speed and historical optimal position of the particle swarm, the position and speed of the particles are updated through the iterative update rule of the PSO algorithm; the search is realized by updating the speed and position of each particle; each particle is updated according to its individual optimal position pbes t and the optimal position of the entire group gbest to adjust its speed and position;

[0067] Particle position update expression:

[0068] (1)

[0069] Where: is the j-th dimension position of the optimal position of the i-th particle in the current iteration; is the j-th dimension position of the optimal position of the i-1th particle in the current iteration; is the j-th velocity component of the velocity of the i-1th particle at the current iteration;

[0070] Particle group velocity update expression:

[0071] (2)

[0072] Where: To improve the inertia weight, it represents the degree of trust in the current speed direction; is the improved individual learning factor and the improved population learning factor; is a random value in [0, 1]; pbest i is the individual optimal position of the ith particle, gbest i is the optimal position of the entire group when the i-th particle is in the individual optimal position, is the j-th velocity component of the velocity of the i-th particle at the current iteration;

[0073] To solve the imbalance between global optimum and local optimum when the inertia coefficient is a constant, improve the inertia weight function for:

[0074] (3)

[0075] Where: is the maximum inertia weight; is the minimum inertia weight; is the current iteration number; is the total number of iterations;

[0076] The larger the inertia weight coefficient is in the early stage, the stronger the global superiority is. The larger the inertia weight coefficient is in the later stage, the stronger the local superiority is, which is easy to improve the convergence accuracy. Read the improved weight function, obtain the individual optimal and global optimal solutions, and use the improved inertia weight function formula to calculate the particle fitness.

[0077] In order to overcome the problem that when the individual learning factor is a constant, it is easy to fall into the local optimum, which causes the algorithm to tend to local search and lack the ability of global search; the improved individual learning factor is adopted as follows:

[0078] (4)

[0079] In order to solve the problem that the population learning factor is too large and the differences between particles are ignored, the algorithm will enter the global optimal solution too early and cannot effectively perform local search; the improved population learning factor is adopted as follows:

[0080] (5)

[0081] Where: is the maximum learning factor; is the minimum learning factor;

[0082] The improved individual learning factor function formula is used to gradually increase the number of particles, improving the local optimization ability of the particle swarm. The improved population learning factor function formula is used to gradually reduce the number of particles, improving the global optimization ability of the particle swarm. The particle swarm speed update function is used to calculate the particle swarm speed.

[0083] According to the updated position, the fitness value of each particle, that is, the value of the objective function, is calculated; according to the fitness value of each particle, the individual position of each particle and the global position of the entire particle swarm are updated.

[0084] Based on the iterative update of the current population, determine whether the iterative data is optimal. If it meets the accuracy requirements, stop the iteration and save the global optimal solution. Otherwise, re-judge.

[0085] When the above method is used, by improving the individual factor function, the global optimal solution is reached after 4 iterations, and the convergence speed is 8% higher than that of the traditional particle swarm algorithm. The parameters corresponding to the global optimal position are the optimized digital twin system configuration parameters, such as Figure 6 shown.

[0086] Step 8: Continuously optimize the adaptive function through the particle swarm algorithm, and finally transfer the global optimal solution to the virtual space;

[0087] Step 9: The operation information, system function, adaptive function, and algorithm optimization information transmitted in the virtual space are used to generate a digital twin platform;

[0088] Step 10: Scan the physical model through sensor technology and directly import it into the digital twin platform to generate a shape twin model;

[0089] Step 11: The visualization system can obtain information from the physical space and use the improved particle swarm algorithm to optimize in real time until it is stable; the visualization system displays the shape twin model and the information optimized in real time using the improved particle swarm algorithm.

[0090] The above method can be used to display the digital twin of the robot arm cleaning system, cleaning parts storage system, etc.:

[0091] Figure 3In the invention, the robot arm cleaning system includes a robot arm unit and a cleaning unit, wherein the robot arm unit includes a robot arm intelligent hardware system and a robot arm intelligent algorithm software system; the cleaning unit includes a cleaning intelligent hardware and software system and an intelligent control system for parts to be cleaned. The system of the invention uses Ethernet network technology for data communication, sensor communication and monitoring intelligent system for information collection and aggregation, particle swarm optimization algorithm to maintain the stability of the system, network technology to issue instructions and real-time changes of data, and a digital twin visualization intelligent cleaning platform to display the actual operation status of each part of the robot arm, as well as all the states of the cleaning parts from entering the site to leaving the factory, and includes order information tracking of the cleaning parts manufacturer, shipping status tracking of parts of the supporting manufacturer, tracking of oil pollution level, tracking of warehouse space occupancy after entering the shipyard, tracking of total inventory, tracking of incoming quantity, tracking of cleaned quantity, tracking of to-be-cleaned quantity, tracking of outgoing quantity, etc. The invention uses a digital twin platform to display all information status in the smart factory, and serves the purpose of data tracking, data monitoring and management.

[0092] Figure 4 In the cleaning parts warehousing system, the digital twin smart factory robot arm cleaning visualization platform monitors the real-time status of the cleaning parts, and the data optimized by the particle swarm optimization algorithm is fed back to the visualization system; the data information instructions are issued by the cleaning parts monitoring intelligent system, and after the data is optimized by the particle swarm algorithm system, it is transmitted to the visualization platform to obtain the manufacturer information of the parts to be cleaned and the smart factory warehouse information. The digital twin visualization intelligent system can coordinate the relevant information of the manufacturer and the smart factory, further increase the manufacturer's parts cleaning efficiency, and at the same time reduce the low utilization rate of the smart factory warehouse.

[0093] The visualization screen is displayed as follows Figure 7 As shown, the order information, delivery status, cleaning level, cleaning steps, drying degree, warehousing information, outbound quantity, and parameter indicators and data analysis of the parts to be cleaned are released.

[0094] Finally, it should be noted that the above is merely the technical solution of the present invention and does not limit it; even though the above real-time method is described in detail, it does not limit it; technicians in this field can modify the case described or replace the technology with the same features; but no matter the modification or replacement, it should not deviate from the technical essence of the present invention.

Claims

1. The method of optimizing the robot arm cleaning of the digital twin system by particle swarm algorithm is characterized by: The following steps are involved: Step 1: The parts are placed in the area to be cleaned, the robotic arm hardware model issues instructions, and the joints of the robotic arm start to move; Step 2: At the same time, sensors at each joint of the robotic arm begin to collect signals; Step 3: The sensor starts working after receiving the signal, identifies and detects the working status of each joint of the robot arm, and transmits the working status information of the robot arm to the data fusion and interaction module; Step 4: The robot arm cleaning model is equipped with a sensor to flush the cleaning agent onto the surface of the part, and waits for the surface cleaning of the part to be completed. The sensor transmits and stores the cleaning speed information to the data fusion and interaction module; Step 5: After the robot arm finishes cleaning, the information is transmitted to the virtual space. The virtual space receives the signal and feeds it back to the robot arm drying system. The robot arm drying system receives the instruction and starts working until the surface of the part is dried. The drying status and part status information collected by the sensor are transmitted to the data fusion interaction module. Step 6: The working state of the robot arm, the drying state, the cleaning speed, and the part state collected by the sensor are converted and sorted through the data fusion interaction module, and the converted information is used to generate an adaptive function; Step 7: Determine the parameter information of the digital twin system that needs to be optimized as the optimization parameter in the particle swarm algorithm; use the particle swarm algorithm to read the adaptive function and map the parameters of the digital twin system to the objective function value to evaluate the performance of each particle; initialize the particle swarm, and each particle is optimized according to its individual best position. pbest and the best position for the entire group gbest to adjust its speed and position; Particle position update expression: (1) Where: is the j-th dimension position of the optimal position of the i-th particle in the current iteration; is the j-th dimension position of the optimal position of the i-1th particle in the current iteration; is the j-th velocity component of the velocity of the i-1th particle at the current iteration; Particle group velocity update expression: (2) Where: To improve the inertia weight, it represents the degree of trust in the current speed direction; is the improved individual learning factor and the improved population learning factor; is a random value in [0, 1]; pbest i is the individual optimal position of the i-th particle; gbest i is the optimal position of the entire group when the i-th particle is in the individual optimal position; is the j-th velocity component of the velocity of the i-th particle at the current iteration; Improved inertia weight function for: (3) Where: is the maximum inertia weight; is the minimum inertia weight; is the current iteration number; is the total number of iterations; The improved individual learning factor is: (4) The improved population learning factor is: (5) Where: is the maximum learning factor; is the minimum learning factor; According to the updated position, the fitness value of each particle, that is, the value of the objective function, is calculated; according to the fitness value of each particle, the individual position of each particle and the global position of the entire particle swarm are updated; According to the iterative update of the current population, the iterative data is judged. If the accuracy requirement is met, the iteration is stopped to save the global optimal solution, otherwise it is re-judged; Step 8: Transfer the global optimal solution to the virtual space; Step 9: The operation information, system function, adaptive function, and algorithm optimization information transmitted in the virtual space are used to generate a digital twin platform; Step 10: Scan the physical model through sensor technology and directly import it into the digital twin platform to generate a shape twin model; Step 11: The visualization system displays the shape twin model and the information optimized using the improved particle swarm algorithm.

Images