Mechanical arm fault prediction system based on digital twinning

Abstract

The invention discloses a mechanical arm fault prediction system based on digital twinning, and belongs to the technical field of mechanical arms. The system comprises a digital twin model construction module, an experimental simulation module, a data collection module, a model training module, a model test module, a fault prediction module, a prediction result output module, a result analysis module, an alarm prompt module, an aging prediction module, a durability analysis module, a digital model display module and a durability identification module for each part. The mechanical arm fault prediction platform is constructed on the basis of the digital twinning technology by means of cooperation of the multiple component state identification modules, the pre-arranged plan execution module and the pre-arranged plan setting storage module, and the faults, the aging degree and the durability of all the components of the mechanical arm can be predicted.

Description

Technical Field

[0001] The invention relates to a mechanical arm fault prediction system based on digital twins, and belongs to the technical field of mechanical arms. Background Art

[0002] Robotic arms are the most widely used automated mechanical devices in the field of robotics technology. They are widely used in industrial manufacturing, medical treatment, entertainment services, military, semiconductor manufacturing, and space exploration.

[0003] Robotic arms are divided into multi-joint robotic arms, rectangular coordinate robotic arms, spherical coordinate robotic arms, polar coordinate robotic arms, cylindrical coordinate robotic arms, etc. according to their different structural forms. Digital twins are a technology that matches the digital model of a physical entity with its real entity. By establishing a digital twin of a device, the operating status of the device can be monitored and simulated in real time. Most of the research on robotic arms is still in the field of fault diagnosis. It is impossible to predict the faults, aging and durability of the various components of the robotic arm, and it is impossible to take corresponding measures in time to reduce the occurrence of faults and improve production efficiency.

[0004] Currently, there are few studies on integrating digital twins into the prediction of robotic arm failure, aging and durability. However, robotic arm failure prediction technology based on digital twins is an inevitable trend of future development. Summary of the invention

[0005] The present invention provides a robotic arm fault prediction system based on digital twin, which is used to build a robotic arm fault prediction platform based on digital twin technology, so as to predict the failure, aging degree and durability of various components of the robotic arm.

[0006] The technical solution of the present invention is:

[0007] A mechanical arm fault prediction system based on digital twins, including a digital twin model construction module, an experimental simulation module, a data collection module, a model training module, a model testing module, a fault prediction module, a prediction result output module, a result analysis module, an alarm prompt module, an aging prediction module, a durability analysis module, a digital model display module, a durability identification module for each component, a status identification module for each component, a plan execution module, and a plan setting storage module;

[0008] The digital twin model building module is used to perform three-dimensional modeling of the shape, size, and assembly relationship of the physical robotic arm in the three-dimensional modeling software; the established three-dimensional model is imported into the smart factory virtual simulation software, the behavior attributes of each joint of the robotic arm and the physical attributes of each component are defined, and the construction of the virtual robotic arm model is completed;

[0009] The experimental simulation module is used for experimental environment setting and physical robot arm working condition setting, wherein the experimental environment setting is used to configure the simulation environment parameters during the experiment; the physical robot arm working condition setting is used to set the working position, speed, and torque of the physical robot arm;

[0010] The data collection module acquires the sensor data of the physical robot arm in real time when it is running by establishing a standardized data interface between the experimental simulation module and the data collection module, and summarizes the data to form an operation log document;

[0011] The model training module is used to train the LSTM model according to the operation log historical data exported from the data collection module to obtain a trained LSTM model;

[0012] The model validation module is used to test and validate the LSTM model constructed in the model training module, and use the validation set to evaluate the model performance after each training round to monitor the overfitting of the model, and optimize the prediction effect of the model by adjusting the hyperparameters;

[0013] The fault prediction module is used to receive sensor data of the physical robot arm in real time and preprocess the data; the preprocessed data is input into the trained LSTM model according to the set time step, and the model captures the features in the time series through its hidden unit and outputs the prediction result through the prediction result output module;

[0014] The prediction result output module is used to output the fault prediction result generated by the fault prediction module;

[0015] The result analysis module is used to compare the output of the LSTM model after completing the prediction with the preset alarm threshold: if the model output result exceeds the set threshold, the system will trigger an alarm prompt;

[0016] The alarm prompt module is used to issue an alarm prompt when the result analysis module triggers the alarm prompt;

[0017] The aging prediction module is used to train and verify the hidden semi-Markov model based on the sensor data on the key components of the physical robotic arm collected by the data collection module, and obtain the trained hidden semi-Markov model to predict the aging of each component of the robotic arm based on the real-time data;

[0018] The durability analysis module is used to predict the average duration of each component of the robot arm according to the aging degree of each component of the robot arm calculated by the aging prediction module, and analyze the durability of each component of the robot arm by calculating the percentage of the average duration in each state to the total working time of the robot arm;

[0019] The digital model display module is used to display the three-dimensional image of the virtual robotic arm through VC as a display terminal for information visualization;

[0020] The durability identification module for each component is used to mark the durability percentage of each component of the robot arm obtained by the durability analysis module on the corresponding component of the virtual robot arm, so that the management personnel can analyze and view it through the digital model display module;

[0021] The component status identification module is used to mark the results of the prediction result output module and the result analysis module on the components of the virtual robotic arm;

[0022] The plan execution module is used to execute the corresponding plan when a certain fault of the robot arm is detected;

[0023] The plan setting storage module is used to set and store the corresponding emergency plan according to the type and content of the robot arm fault.

[0024] The experimental simulation module controls the operation of the experimental equipment, which includes a bottom plate 1, a test box 2 is fixedly connected to the top of the bottom plate 1, a temperature control chamber 3 is provided inside the test box 2, water inlets 4 are fixedly connected to both sides of the test box 2, a water outlet 5 is fixedly connected to the side of the water inlet 4, and one end of the water inlet 4 and the water outlet 5 both extend into the temperature control chamber 3; a light lamp 15 is fixedly connected to the top of the inner cavity of the test box 2, atomizing heads 16 are fixedly connected to the top of the test box 2 and on both sides of the light lamp 15, and the bottom ends of the atomizing heads 16 extend into the interior of the test box 2; a mixing chamber 10 is fixedly connected to one side of the test box 2, and a storage tank 11 is fixedly connected to the top of the mixing chamber 10 A switch valve 12 is fixedly connected to the bottom of the storage tank 11, and the bottom end of the switch valve 12 extends into the mixing chamber 10. A pump 13 is fixedly connected to one side of the experimental box 2, and one end of the pump 13 extends into the mixing chamber 10; the mixing chamber 10 is used to communicate with one end of the connecting pipe II14 and one end of the connecting pipe I9. Fixed frames 6 are fixedly connected to both sides of the inner cavity of the experimental box 2, and the fixed frames 6 are rotatably connected to the nozzles 7. A servo motor 8 is fixedly connected to one side of the fixed frame 6, and the output end of the servo motor 8 is fixedly connected to the outer side of the nozzle 7. The end of the nozzle 7 is connected to the other end of the connecting pipe I9 and the connecting pipe II14 that are fixedly connected and extend into the interior of the experimental box 2.

[0025] The beneficial effects of the present invention are:

[0026] First, the present invention is provided with a fault prediction module, an aging prediction module and a durability analysis module. When predicting and detecting faults of a robotic arm in use, the data of the robotic arm is collected and input into the twin LSTM model deployed, and the fault prediction result generated by the LSTM type is output. The fault prediction result is analyzed to determine whether an alarm prompt is required. If it is determined that an alarm prompt is required, an alarm prompt alarm is issued to the management personnel, and the aging degree of each component of the robotic arm is predicted according to the working conditions of the robotic arm and the state data of the robotic arm itself. According to the aging degree prediction results of each component of the robotic arm, the durability of each component of the robotic arm is analyzed, the durability of each component is marked on the virtual robotic arm, and the prediction results and the detection results are marked on each component of the virtual robotic arm, so that the management personnel can view the virtual robotic arm through the display device and monitor the virtual robotic arm, so as to deal with the robotic arm in time and reduce the impact on the work.

[0027] Second, the present invention is provided with an experimental box, a temperature regulating chamber and an air chamber, and the robot arm is installed inside the bottom plate. Then, the lighting lamp adjusts the lighting conditions such as the lighting intensity and the lighting time, sprays atomized water into the inner cavity of the experimental box through the atomizing head, adjusts the humidity of the inner cavity of the experimental box, inputs the temperature transfer liquid into the temperature regulating chamber through the water inlet, and extracts the temperature transfer liquid in the temperature regulating chamber through the water outlet, and reheats or cools down. By opening the switch valve, the simulated dust stored in the storage tank or the debris generated during work falls into the mixing chamber through the switch valve, and then the pump is started to inhale the outside air, and then the air is discharged into the mixing chamber through the pump, so that the air carries the dust or debris in the tube into the connecting tube through the connecting tube, and then enters the nozzle through the connecting tube, and then is sprayed into the inner cavity of the experimental box through the nozzle, and the nozzle is driven to rotate by the output end of the servo motor, and the spray angle of the nozzle is adjusted to collect the data of the robot arm performing simulated work, so as to predict the failure and aging degree of the robot arm.

[0028] Third, the present invention dynamically changes the temperature, humidity, dust, and light-related working conditions in the experimental box through the experimental environment setting module and the experimental condition setting module, which compensates for the single working environment of the robotic arm and can simulate the working conditions of the robotic arm in various different environments. Changes in environmental information directly act on the physical robotic arm and directly affect the data collected by the robotic arm sensors, thereby exploring the occurrence of robotic arm failures and aging progress under different environmental conditions.

[0029] Fourth, the present invention provides an innovative solution by effectively combining digital twin technology with fault prediction and aging prediction related technologies. Through digital twin technology, the system can collect the operation data of the robot arm in real time and combine it with the virtual model. The working process of the robot arm, the predicted fault information display, the aging percentage of each component of the robot arm, etc. are all presented to the management personnel in a visual and clear way, which greatly reduces the management personnel's dependence on complex technical details, allowing them to focus more on decision-making and management, so as to take corresponding measures more quickly point-to-point, thereby improving the overall maintenance efficiency and reducing the risk of misoperation or multiple operations. In this way, not only the equipment maintenance process is simplified, but also it helps enterprises to extend the life of equipment more effectively and reduce operating costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a schematic diagram of the module structure of the present invention;

[0031] Figure 2 A schematic diagram of constructing a twin model of the robotic arm of the present invention;

[0032] Figure 3 It is a cross-sectional view of the internal structure of the experimental equipment of the present invention;

[0033] Figure 4 It is a side view of the internal structure of the experimental equipment of the present invention;

[0034] Figure 5 This is a schematic diagram of an experimental box of the experimental equipment of the present invention after one side panel is opened;

[0035] Figure 6 This is a schematic diagram of the structure of the nozzle fixing frame of the present invention;

[0036] Figure 7 Some temperature, current and vibration time series data;

[0037] The numbers in the figure are: 1. bottom plate; 2. experimental box; 3. temperature control chamber; 4. water inlet; 5. water outlet; 6. fixing frame; 7. nozzle; 8. servo motor; 9. connecting pipe I; 10. mixing chamber; 11. storage tank; 12. switch valve; 13. pump; 14. connecting pipe II; 15. light lamp; 16. atomizing head. DETAILED DESCRIPTION

[0038] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. It should be noted that the embodiments in this application and the features in the embodiments can be combined with each other arbitrarily without conflict.

[0039] Example 1: Figure 1-7 As shown, a mechanical arm fault prediction system based on digital twins includes a digital twin model construction module, an experimental simulation module, a data collection module, a model training module, a model testing module, a fault prediction module, a prediction result output module, a result analysis module, an alarm prompt module, an aging prediction module, a durability analysis module, a digital model display module, a durability identification module for each component, a status identification module for each component, a plan execution module, and a plan setting storage module;

[0040] The digital twin model construction module is used to perform accurate and high-precision three-dimensional modeling of the shape, size, assembly relationship, etc. of the physical robotic arm in the three-dimensional modeling software (SOLIDWORKS); import the established three-dimensional model into the smart factory virtual simulation software (Visual Components, VC), define the behavior attributes of each joint of the robotic arm and the physical attributes of each component, and complete the construction of the virtual robotic arm model;

[0041] See also Figure 2 , further explained by building a twin model of the robotic arm, the robotic arm is geometrically modeled using 3D modeling software. According to the shape and size of the actual robotic arm, a 3D model of each joint is created in the 3D software SOLIDWORKS, and assembled according to the hierarchy and assembly relationship to form a complete 3D model of the robotic arm. This 3D model is imported into the twinning software Visual Components for secondary modeling, including the modeling of physical properties and behavioral properties. Physical property modeling involves the mass, material, system properties, speed of each joint, and center of gravity of the robotic arm to ensure that these properties are consistent with the actual robotic arm. Behavioral property modeling covers action scripts, communication interfaces, connection scripts, controllers, and data statistics modules. After completing these property modeling, the robotic arm should have the same motion behavior as the physical robotic arm. Data is transmitted through the data interface to achieve virtual and real mutual control of the robotic arm.

[0042] The experimental simulation module is used for experimental environment settings and physical robot arm working condition settings. The experimental environment settings are used to configure the simulated environmental parameters during the experiment, such as the temperature, humidity, dust, and light of the space; the physical robot arm working condition settings are used to set the physical robot arm working position, speed, and torque. This module can simulate various extreme or common working environments to help test the performance and durability of the robot arm under different environmental conditions;

[0043] The data collection module, by establishing a standardized data interface between the experimental simulation module and the data collection module, acquires the sensor data (position sensor, acceleration sensor, temperature sensor, current sensor, vibration sensor) of the physical robotic arm in real time during operation through the external data acquisition software KEP Servere and the embedded OPC UA server data acquisition module developed based on ESP32, and summarizes it into an operation log document. In specific applications, the above method is used to collect data for each component on the robotic arm;

[0044] The model training module is used to train the LSTM model according to the operation log history data exported from the data collection module to obtain a trained LSTM model. By applying the above technical solution, it can be seen that by adopting the LSTM model, the time series data collected by the robot arm has a better prediction ability;

[0045] The model validation module is used to test and validate the LSTM model constructed in the model training module. The model performance is evaluated using the validation set after each training round to monitor the overfitting of the model. At the same time, the prediction effect of the model is optimized by adjusting the hyperparameters: time step, number of LSTM layers, number of hidden units, and learning rate.

[0046] The specific training and verification process is: pre-processing the collected data, including: processing of missing values ​​and outliers; standardizing or normalizing the processed data to improve the robustness and convergence speed of model training. The completed data is divided into a training set and a verification set, where the verification set is used for the model testing module. Exemplarily, the present invention constructs a training set and a verification set based on the collected temperature and current time series data at a preset time, such as Figure 7Part of the temperature, current, and vibration data is shown. For the data in the training set and validation set, select features that are helpful for fault prediction, such as data from specific sensors, and use the sliding window method to generate time series features to extract useful features from the original data. Then, define the input layer, select the appropriate time step, and input the time series data into the LSTM layer. The LSTM layer consists of multiple hidden units to capture time-dependent features; then, further process the LSTM output through the fully connected layer, and the output layer uses multi-classification output and the "softmax" activation function. Use the Categorical Cross-Entropy loss function. Set the batch size and training epoch, where the batch size controls the number of samples used each time the model parameters are updated, and the training epoch determines the number of times the model iterates over the entire data set.

[0047] The fault prediction module is used to receive sensor data of the physical robotic arm in real time and preprocess the data (such as normalization); the preprocessed data is input into the trained LSTM model according to the set time step, and the model captures the features in the time series through its hidden units and outputs the prediction results through the prediction result output module.

[0048] The prediction result output module is used to output the fault prediction result generated by the fault prediction module;

[0049] The result analysis module is used to compare the output of the LSTM model after completing the prediction with the preset alarm threshold: if the model output result exceeds the set threshold, for example, the predicted value of a key parameter (such as temperature, current, position, vibration) exceeds the safe range, the system will trigger an alarm prompt;

[0050] The alarm prompt module is used to send an alarm prompt to the management personnel through sound and light signals, system pop-up windows or SMS notifications when the alarm prompt is triggered by the result analysis module;

[0051] The aging prediction module is used to train and verify the hidden semi-Markov model based on the sensor data (such as temperature and vibration) on the key components of the physical robotic arm (such as joints and motors) collected by the data collection module, to obtain a trained hidden semi-Markov model, and further predict the aging of each component of the robotic arm through real-time data.

[0052] Specifically, the hidden semi-Markov model training and verification are as follows:

[0053] According to the data collected by the data collection module (temperature, vibration, etc.), the data are classified according to each component and stored in the database. The abnormal data are removed and the missing data are supplemented to maintain the data integrity of each time step. The characteristic data related to aging (temperature change rate, amplitude change rate) are calculated and extracted. The hidden state of the robot arm aging process is divided into three states: "normal state", "mild aging" and "severe aging", and a probability distribution model of the observed data is established for each state. In the "normal state" and "mild aging" states, the state duration distribution is defined by normal distribution, and in the "severe aging" stage, the state duration distribution is defined by exponential distribution. After completing the state and duration modeling, the preprocessed historical data is used to train the hidden semi-Markov model HSMM to estimate the model's state transition matrix, state duration distribution, and observation probability distribution parameters. The expectation maximization (EM) algorithm is used to optimize the model parameters so that the model can accurately describe the data generation process. Finally, the cross-validation method is used to evaluate the performance of the HSMM model and compare the actual aging data with the model prediction results to determine the effectiveness of the model.

[0054] The durability analysis module is used to predict the average duration of each component of the robot arm according to the aging degree of each component of the robot arm calculated by the aging prediction module, and analyze the durability of each component of the robot arm by calculating the percentage of the average duration in each state to the total working time of the robot arm;

[0055] The digital model display module is used to display the three-dimensional image of the virtual robotic arm through VC as a display terminal for information visualization;

[0056] The durability identification module for each component is used to mark the durability percentage of each component of the robot arm obtained by the durability analysis module on the corresponding component of the virtual robot arm, so that the management personnel can analyze and view it through the digital model display module;

[0057] The component status identification module is used to mark the results of the prediction result output module and the result analysis module on the components of the virtual robotic arm;

[0058] The plan execution module is used to execute the corresponding plan when a certain fault of the robotic arm is detected;

[0059] The plan setting storage module is used to set and store the corresponding emergency plan according to the type and content of the robot arm fault.

[0060] like Figure 3-Figure 6As shown, the experimental simulation module controls the operation of the experimental equipment, and the experimental equipment includes a bottom plate 1, the top of the bottom plate 1 is fixedly connected to an experimental box 2, a temperature control chamber 3 is provided inside the experimental box 2, water inlets 4 are fixedly connected to both sides of the experimental box 2, a water outlet 5 is fixedly connected next to the water inlet 4, and one end of the water inlet 4 and the water outlet 5 both extend into the temperature control chamber 3; a lighting lamp 15 is fixedly connected to the top of the inner cavity of the experimental box 2, atomizing heads 16 are fixedly connected to the top of the experimental box 2 and on both sides of the lighting lamp 15, and the bottom ends of the atomizing heads 16 extend into the interior of the experimental box 2; a mixing chamber 10 is fixedly connected to one side of the experimental box 2, and a storage tank is fixedly connected to the top of the mixing chamber 10 11, a switch valve 12 is fixedly connected to the bottom of the storage tank 11, and the bottom end of the switch valve 12 extends to the inside of the mixing chamber 10, and a pump 13 is fixedly connected to one side of the experimental box 2, and one end of the pump 13 extends to the inside of the mixing chamber 10; the mixing chamber 10 is used to communicate with one end of the connecting pipe II14 and one end of the connecting pipe I9, and the two sides of the inner cavity of the experimental box 2 are fixedly connected with a fixing frame 6, and the fixing frame 6 is rotatably connected with a nozzle 7, and one side of the fixing frame 6 is fixedly connected with a servo motor 8, and the output end of the servo motor 8 is fixedly connected to the outer side of the nozzle 7, and the end of the nozzle 7 is connected to the other end of the connecting pipe I9 and the connecting pipe II14 that are fixedly connected and extend to the inside of the experimental box 2.

[0061] Before simulating and collecting data for the robot arm, different environmental conditions are set for the robot arm to perform simulation experiments. The specific operation process can be: install the robot arm inside the bottom plate 1, then adjust the light conditions such as light intensity and light time with the light lamp 15, spray atomized water into the inner cavity of the experimental box 2 through the atomizing head 16, adjust the humidity of the inner cavity of the experimental box 2, input temperature transfer liquid into the temperature control chamber 3 through the water injection port 4, extract the temperature transfer liquid in the temperature control chamber 3 through the water outlet 5, and reheat or cool down the temperature, and open the switch valve 12 , so that the simulated dust stored in the storage tank 11 or the debris generated during operation falls into the mixing chamber 10 through the switch valve 12, and then the pump 30 is started to suck in the outside air, and then the air is discharged into the mixing chamber 10 through the pump 13, so that the air carries the dust or debris in the mixing chamber 10 into the connecting pipe 9 through the connecting pipe 14, and then enters the nozzle 7 through the connecting pipe 9, and then is sprayed into the inner cavity of the experimental box 2 through the nozzle 7, and the nozzle 7 is driven to rotate by the output end of the servo motor 8 to adjust the spray angle of the nozzle 7.

[0062] Furthermore, the robot arm is modeled by three-dimensional software, and then the model is imported into VC for physical property and behavior property modeling to complete the construction of the twin model. The robot arm performs simulation work in the inner cavity of the experimental box 2. During the simulation work, the operation data of the robot arm in the required time period is collected, and the data is divided into a training set and a validation set. The training set data is used to train the LSTM prediction model, and the validation set data is used to verify the trained prediction model. The prediction results of the prediction model are evaluated, and the prediction model is optimized according to the evaluation results to obtain the optimal LSTM prediction model; when the optimal LSTM prediction model is used to predict faults for the real-time collected data, the real-time data of the robot arm collected is input into the prediction model, and the generated fault prediction results are output. The fault prediction results are analyzed to determine whether an alarm prompt is required. If it is determined that an alarm prompt is required, an alarm prompt alarm is issued to the management personnel;

[0063] Furthermore, the aging degree of each component of the robotic arm is predicted based on the robotic arm's own status data, and the durability of each component of the robotic arm is identified based on the predicted results of the aging degree of each component of the robotic arm. The durability of each component is marked on the virtual robotic arm, and the predicted results are marked on each component of the virtual robotic arm so that the management personnel can view the virtual robotic arm through the display device and monitor the virtual robotic arm.

[0064] The specific implementation modes of the present invention are described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the above implementation modes, and various changes can be made within the knowledge scope of ordinary technicians in this field without departing from the purpose of the present invention.

Claims

1. A robotic arm fault prediction system based on digital twin, characterized in that: It includes a digital twin model construction module, an experimental simulation module, a data collection module, a model training module, a model testing module, a fault prediction module, a prediction result output module, a result analysis module, an alarm prompt module, an aging prediction module, a durability analysis module, a digital model display module, a durability identification module for each component, a status identification module for each component, a plan execution module, and a plan setting storage module; The digital twin model building module is used to perform three-dimensional modeling of the shape, size, and assembly relationship of the physical robotic arm in the three-dimensional modeling software; the established three-dimensional model is imported into the smart factory virtual simulation software, the behavior attributes of each joint of the robotic arm and the physical attributes of each component are defined, and the construction of the virtual robotic arm model is completed; The experimental simulation module is used for experimental environment setting and physical robot arm working condition setting, wherein the experimental environment setting is used to configure the simulation environment parameters during the experiment; the physical robot arm working condition setting is used to set the working position, speed, and torque of the physical robot arm; The data collection module acquires the sensor data of the physical robot arm in real time when it is running by establishing a standardized data interface between the experimental simulation module and the data collection module, and summarizes the data to form an operation log document; The model training module is used to train the LSTM model according to the operation log historical data exported from the data collection module to obtain a trained LSTM model; The model validation module is used to test and validate the LSTM model constructed in the model training module, and use the validation set to evaluate the model performance after each training round to monitor the overfitting of the model, and optimize the prediction effect of the model by adjusting the hyperparameters; The fault prediction module is used to receive sensor data of the physical robot arm in real time and preprocess the data; the preprocessed data is input into the trained LSTM model according to the set time step, and the model captures the features in the time series through its hidden unit and outputs the prediction result through the prediction result output module; The prediction result output module is used to output the fault prediction result generated by the fault prediction module; The result analysis module is used to compare the output of the LSTM model after completing the prediction with the preset alarm threshold: if the model output result exceeds the set threshold, the system will trigger an alarm prompt; The alarm prompt module is used to issue an alarm prompt when the result analysis module triggers the alarm prompt; The aging prediction module is used to train and verify the hidden semi-Markov model based on the sensor data on the key components of the physical robotic arm collected by the data collection module, and obtain the trained hidden semi-Markov model to predict the aging of each component of the robotic arm based on the real-time data; The durability analysis module is used to predict the average duration of each component of the robot arm according to the aging degree of each component of the robot arm calculated by the aging prediction module, and analyze the durability of each component of the robot arm by calculating the percentage of the average duration in each state to the total working time of the robot arm; The digital model display module is used to display the three-dimensional image of the virtual robotic arm through VC as a display terminal for information visualization; The durability identification module for each component is used to mark the durability percentage of each component of the robot arm obtained by the durability analysis module on the corresponding component of the virtual robot arm, so that the management personnel can analyze and view it through the digital model display module; The component status identification module is used to mark the results of the prediction result output module and the result analysis module on the components of the virtual robotic arm; The plan execution module is used to execute the corresponding plan when a certain fault of the robotic arm is detected; The plan setting storage module is used to set and store the corresponding emergency plan according to the type and content of the robot arm fault.

2. The digital twin-based robotic arm fault prediction system according to claim 1, characterized in that: The experimental simulation module controls the operation of the experimental equipment, and the experimental equipment comprises a bottom plate (1), the top of the bottom plate (1) is fixedly connected to an experimental box (2), a temperature regulating chamber (3) is provided inside the experimental box (2), water injection ports (4) are fixedly connected to both sides of the experimental box (2), a water outlet (5) is fixedly connected to the side of the water injection port (4), and one end of each of the water injection port (4) and the water outlet (5) extends into the temperature regulating chamber (3); a lighting lamp (15) is fixedly connected to the top of the inner chamber of the experimental box (2), atomizing heads (16) are fixedly connected to the top of the experimental box (2) and on both sides of the lighting lamp (15), and the bottom ends of each of the atomizing heads (16) extend into the interior of the experimental box (2); a mixing chamber (10) is fixedly connected to one side of the experimental box (2), and a storage tank (11) is fixedly connected to the top of the mixing chamber (10). The bottom of the storage tank (11) is fixedly connected to a switch valve (12), the bottom end of which extends into the mixing chamber (10); one side of the experimental box (2) is fixedly connected to a pump (13), one end of which extends into the mixing chamber (10); the mixing chamber (10) is used to communicate with one end of a connecting pipe II (14) and one end of a connecting pipe I (9); both sides of the inner cavity of the experimental box (2) are fixedly connected to a fixing frame (6), the fixing frame (6) is rotatably connected to a nozzle (7), one side of the fixing frame (6) is fixedly connected to a servo motor (8), the output end of the servo motor (8) is fixedly connected to the outer side of the nozzle (7), and the end of the nozzle (7) is connected to the other end of the connecting pipe I (9) and the connecting pipe II (14) which are fixedly connected and extend into the interior of the experimental box (2).

Images