AI-based pneumatic fault practical training analogue simulation method and system
By introducing AI-based pneumatic fault training simulation method in pneumatic training teaching, the problems of high cost and limited learning effects of traditional training models are solved, efficient fault diagnosis and repair training are achieved, and user experience and learning effects are improved.
Patent Information
- Application Number
- CN202510243063.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-03
- Publication Date
- 2025-05-16
AI Technical Summary
The traditional pneumatic training teaching model has high cost, long time and equipment loss risks. The existing virtual simulation technology lacks intelligent guidance and diversified training scenarios in fault diagnosis, resulting in limited user experience and learning effects.
Using AI-based pneumatic fault training simulation method, a virtual simulation laboratory scenario model is built by establishing a real-time operating state database and fault database, a machine learning algorithm is used to diagnose and identify simulated fault scenarios, and fault repair guidelines are provided, and virtual fault diagnosis and scoring reports are finally generated.
It effectively improves the efficiency and learning experience of pneumatic fault training, improves the fault diagnosis ability and operation skills of trainees, provides personalized feedback and scores, and enhances the pertinence of learning.
Smart Images
Figure CN120014900A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of industrial automation and educational technology, and specifically, relates to an AI-based pneumatic fault training simulation method and system. Background Art
[0002] Pneumatic training is widely used in the fields of industry and vocational education, aiming to cultivate trainees' ability to operate and maintain pneumatic equipment. However, the traditional practical teaching mode has problems such as high cost, long time consumption and high risk of equipment loss. In addition, the fault simulation in the pneumatic system is relatively complicated, especially in the case of multiple fault combinations, which is difficult to simulate realistically, making it difficult for trainees to systematically master fault diagnosis skills in complex scenarios. With the development of virtual reality and artificial intelligence technology, pneumatic training systems based on virtual simulation technology have gradually become an effective solution. However, the existing virtual simulation technology lacks intelligent guidance and diversified training scenarios in fault diagnosis, resulting in limited user experience and learning effects. Therefore, there is an urgent need for a pneumatic fault training simulation method that can flexibly simulate multiple fault scenarios and enhance teaching effects through AI. Summary of the invention
[0003] The first purpose of the invention is to overcome the shortcomings and deficiencies in the prior art and to provide an AI-based pneumatic fault training simulation method that can effectively improve the efficiency and learning experience of pneumatic fault training and improve the fault diagnosis ability of trainees.
[0004] The second purpose of the present invention is to provide an AI-based pneumatic fault training simulation system.
[0005] The purpose of the present invention is achieved through the following technical solution: an AI-based pneumatic fault training simulation method, comprising the steps of:
[0006] S1. Establish a real-time operating status database for pneumatic equipment and a fault database;
[0007] S2. Construct a virtual simulation laboratory scene model, wherein the virtual simulation laboratory model includes a virtual pneumatic test bench and pneumatic equipment model, and a pneumatic training virtual platform;
[0008] S3, based on the virtual simulation laboratory scenario model, combined with the real-time operation status database and fault database, trigger the corresponding simulated fault scenario;
[0009] S4. Use machine learning algorithms to diagnose and identify simulated fault scenarios and provide fault repair guidance;
[0010] S5. Generate virtual fault diagnosis and scoring reports to complete pneumatic fault training.
[0011] Preferably, in step S1, the fault database includes a base layer, a component layer and a system layer, the base layer is used to define common types of faults, the component layer is used to define fault types of specific components, and the system layer is used to define cross-component fault combinations; the annotation information of each layer of faults includes fault conditions and fault repair methods.
[0012] Preferably, step S2 specifically includes:
[0013] S21. Construct a virtual laboratory environment model through 3D modeling software, including simulating the environmental configuration and physical properties of the actual pneumatic laboratory;
[0014] S22. Construct a virtual pneumatic test bench and pneumatic equipment model using 3D modeling software, including key components of simulated pneumatic equipment;
[0015] S23. Build a virtual platform for pneumatic training based on Unity, including setting the physical parameters and behavioral parameters of key components.
[0016] Preferably, step S3 specifically includes:
[0017] S31, collecting real-time operating status data of pneumatic equipment through a virtual simulation laboratory scene model to simulate the normal operating status of actual pneumatic equipment;
[0018] S32. The trainee manually selects or randomly generates fault conditions preset in the fault library. When the real-time operating status data meets the selected fault conditions, the corresponding simulated fault scenario is automatically triggered.
[0019] Preferably, the fault conditions include single fault conditions and combined fault conditions, and the combined fault conditions are defined using Boolean logic; the simulated fault scenarios include single fault scenarios and compound fault scenarios, and the compound fault scenarios are designed using a multi-threaded architecture, wherein each fault scenario is processed by an independent thread; when the real-time operating status data meets the combined fault conditions, the compound fault scenario is triggered.
[0020] Preferably, the control of the virtual pneumatic test bench and the pneumatic equipment model includes realization through a computer terminal mode and an immersive VR equipment mode.
[0021] The control of the computer mode specifically includes using the mouse, keyboard and virtual pneumatic test bench to interact with the virtual control panel in the pneumatic equipment model to select different components of the virtual pneumatic equipment as the test objects;
[0022] The control of the immersive VR device mode specifically includes using gestures to select different components of the virtual pneumatic device as detection objects through a head-mounted display and a handle.
[0023] Preferably, step S4 specifically includes:
[0024] S41, extracting characteristic parameters from the real-time operating status data, and using the pre-trained classification model to classify faults according to the input characteristic parameters, and outputting the fault type;
[0025] S42. Based on the identified fault type, the user is guided to perform fault troubleshooting and repair operations according to the fault repair method preset in the fault database.
[0026] Preferably, the classification model includes a decision tree and a neural network, and the training of the classification model optimizes parameters by minimizing the cross entropy loss function:
[0027]
[0028] Among them, y i is the actual fault type, It is the probability of the fault type predicted by the classification model. The pre-trained machine learning model can automatically identify the fault type based on the data collected in real time.
[0029] Preferably, step S5 specifically includes:
[0030] S51. Use a weight-based multi-factor scoring model to score the user's fault repair process:
[0031] S=ω1S accuracy +ω2S efficiency +ω3S precision ,
[0032] Among them, S is the total score, S accuracy is the fault identification accuracy, S efficiency is the fault handling efficiency, S precision is the operation accuracy, ω1, ω2, ω3 are weight coefficients;
[0033] S52: Output a fault diagnosis report, which includes a fault identification process, a solution and improvement suggestions, and then upload the fault diagnosis report to a training management database.
[0034] AI-based pneumatic fault training simulation system, including:
[0035] The acquisition module is used to establish a real-time operating status database of pneumatic equipment and a fault database;
[0036] A modeling module is used to construct a virtual simulation laboratory scene model, wherein the virtual simulation laboratory model includes a virtual pneumatic test bench and pneumatic equipment model, and a pneumatic training virtual platform;
[0037] The simulation module is used to trigger the corresponding simulated fault scenarios based on the virtual simulation laboratory scenario model in combination with the real-time operation status database and the fault database;
[0038] AI training module, which uses machine learning algorithms to diagnose and identify simulated fault scenarios and provide fault repair guidance;
[0039] The scoring module is used to generate virtual fault diagnosis and scoring reports.
[0040] Compared with the prior art, the present invention has the following advantages and effects:
[0041] (1) The present invention provides an AI-based pneumatic fault training simulation method, which constructs a virtual simulation laboratory scene model, simulates the fault of a pneumatic test bench in a virtual environment, diagnoses and identifies the simulated fault scene and provides fault repair guidance through a machine learning algorithm, and finally generates a virtual fault diagnosis report to provide data support for teachers. This method can effectively improve the efficiency and safety of pneumatic fault training, and improve the fault diagnosis ability and operating skills of trainees; the present invention is suitable for teaching environments such as vocational education institutions, colleges and universities, and corporate training, and has a wide range of applications.
[0042] (2) The present invention constructs a highly simulated virtual pneumatic test bench and pneumatic equipment model, and the user can achieve touch control through a mouse and keyboard or interact naturally with the model through a VR device, thereby obtaining an immersive training experience, thereby enhancing the user's actual operation sense and learning effect. Compared with the traditional training mode, the present invention is more interactive and scalable.
[0043] (3) The present invention simulates various pneumatic equipment failure scenarios by building a multi-level, modular fault library into the system. Users can practice various failure scenarios in a safe virtual environment and obtain fault repair guidance through a classification model, thereby optimizing the learning effect.
[0044] (4) The present invention provides an AI-based pneumatic fault training simulation system, which performs real-time analysis of user operation data through the AI training module, can record the user's learning path and provide personalized feedback and scoring. Compared with the traditional fixed learning process, this system effectively improves the targeted learning and helps users steadily master the troubleshooting skills of pneumatic systems. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] Figure 1 It is a flow chart of the AI-based pneumatic fault training simulation method of the present invention. DETAILED DESCRIPTION
[0046] The present invention will be further described in detail below in conjunction with embodiments and drawings, but the embodiments of the present invention are not limited thereto.
[0047] Example 1
[0048] like Figure 1 The figure shows a flow chart of the AI-based pneumatic fault training simulation method, including the following steps:
[0049] S1. Establish a real-time operating status database for pneumatic equipment and a fault database;
[0050] S2. Construct a virtual simulation laboratory scene model, wherein the virtual simulation laboratory model includes a virtual pneumatic test bench and pneumatic equipment model, and a pneumatic training virtual platform;
[0051] S3, based on the virtual simulation laboratory scenario model, combined with the real-time operation status database and fault database, trigger the corresponding simulated fault scenario;
[0052] S4. Use machine learning algorithms to diagnose and identify simulated fault scenarios and provide fault repair guidance;
[0053] S5. Generate virtual fault diagnosis and scoring reports to complete pneumatic fault training.
[0054] Specifically, in this embodiment, sensors are used to obtain real-time data on the operating status of cylinders, valves, etc., including but not limited to equipment temperature, air pressure, working hours, etc., and abnormal values are detected and removed, and then these real-time data are stored in a real-time operating status database.
[0055] Then, based on the historical fault data and known fault modes of the pneumatic equipment, a modular and hierarchical fault library is constructed. In step S1, the fault database includes a basic layer, a component layer, and a system layer. The basic layer is used to define common types of faults, such as pipeline leakage and cylinder jamming. The component layer is used to define the fault types of specific components, such as solenoid valve jamming and cylinder failure. The system layer is used to define cross-component fault combinations; for example, the coordination failure between the pneumatic and electronic control systems or the simultaneous failure of multiple components. Each layer of faults is stored in a structured manner in the database, and the annotation information of each layer of faults includes the fault conditions and fault repair methods.
[0056] Step S2 specifically includes:
[0057] S21. Construct a virtual laboratory environment model through 3D modeling software, including simulating the environmental configuration and physical properties of the actual pneumatic laboratory; the virtual laboratory environment model has the same environmental configuration as the actual pneumatic training laboratory, such as walls, floors, and lights;
[0058] S22. Use 3D modeling software (such as Blender or SolidWorks) to build a virtual pneumatic test bench and pneumatic equipment model, including simulating key components of pneumatic equipment (such as cylinders, solenoid valves, and pressure sensors);
[0059] S23. Build a virtual platform for pneumatic training based on Unity, including setting the physical parameters (such as size, material properties) and behavioral parameters (such as action response, fault characteristics) of key components to ensure realistic simulation effects.
[0060] Specifically, the virtual laboratory environment model, including basic elements such as walls, floors, and lights, is the same as that of an actual pneumatic laboratory; the virtual simulation pneumatic test bench and pneumatic equipment model are the same in appearance and parameters as the equipment used in actual pneumatic training, and simulate the actual operating state. The virtual control panel model is the same in appearance and parameters as the control panel of the test bench, allowing users to operate and monitor the test bench.
[0061] Step S3 specifically includes:
[0062] S31, collecting real-time operating status data of pneumatic equipment through a virtual simulation laboratory scene model to simulate the normal operating status of actual pneumatic equipment;
[0063] S32. The trainee manually selects or randomly generates fault conditions preset in the fault library. When the real-time operating status data meets the selected fault conditions, the corresponding simulated fault scenario is automatically triggered.
[0064] The fault conditions include single fault conditions and combined fault conditions, and the combined fault conditions are defined using Boolean logic; the simulated fault scenarios include single fault scenarios and compound fault scenarios, and the compound fault scenarios are designed using a multi-threaded architecture, where each fault scenario is processed by an independent thread; when the real-time operating status data meets the combined fault conditions, the compound fault scenario is triggered.
[0065] Specifically, by collecting the operating data of the pneumatic equipment in real time and comparing it with the pre-defined fault conditions in the fault library, when the operating data meets the specific fault conditions, the system will automatically trigger the corresponding fault scenario in the virtual environment. In this embodiment, in simulation mode, specific abnormal behaviors can also be triggered by importing simulated fault data or actively setting fault parameters. For example, when a pipeline ruptures, the system generates a gas leakage visual effect based on a particle system accompanied by an alarm sound; when the cylinder is stuck, the cylinder movement is slow and the sensor feedback information is distorted, thereby truly reproducing the abnormal behavior in the equipment. This design can not only support fault detection of real data, but also effectively verify the performance and diagnostic logic of fault scenarios in a simulation environment.
[0066] In order to meet different training needs, the present invention also supports users to freely combine multiple fault scenarios. This function is realized through combinational logic and multi-threaded simulation technology. Trainees (users) can select multiple fault conditions, and the system uses Boolean logic (such as AND / OR operations) to define these combination conditions and combine them into composite fault scenarios. For example, when condition A (air pressure is lower than the threshold) and condition B (temperature rises too fast) are met at the same time, the system triggers a combined fault scenario (composite fault scenario). In order to realize the synchronous simulation of multiple fault scenarios, the system adopts a multi-threaded architecture, and each fault scenario is processed by an independent thread to ensure the real-time and smoothness of the simulation effect.
[0067] The control of the virtual pneumatic test bench and the pneumatic equipment model includes realization through a computer terminal mode and an immersive VR equipment mode.
[0068] The control of the computer mode specifically includes using the mouse, keyboard and virtual pneumatic test bench to interact with the virtual control panel in the pneumatic equipment model to select different components of the virtual pneumatic equipment as the test objects;
[0069] The control of the immersive VR device mode specifically includes using gestures to select different components of the virtual pneumatic device as detection objects through a head-mounted display and a handle.
[0070] Specifically, in the computer mode, the pneumatic training virtual platform provides real-time feedback on test results and operation effects, allowing users to learn efficiently on computer devices. In the immersive VR device mode, the pneumatic training virtual platform combines real 3D sound and visual effects to provide users with a highly immersive experience. Users can move freely in the virtual simulation laboratory and observe equipment and fault scenarios from multiple angles, significantly improving learning effects and participation, and stimulating the enthusiasm of trainees.
[0071] In the present invention, the user achieves touch control through a mouse and keyboard or interacts naturally with the model through a VR device, thereby obtaining an immersive training experience, enhancing the user's actual operation sense and learning effect, and being more interactive and scalable than traditional training modes.
[0072] Step S4 specifically includes:
[0073] S41, extracting characteristic parameters from the real-time operating status data, and using the pre-trained classification model to classify faults according to the input characteristic parameters, and outputting the fault type;
[0074] S42. Based on the identified fault type, the user is guided to perform fault troubleshooting and repair operations according to the fault repair method preset in the fault database.
[0075] Specifically, the present invention uses machine learning algorithms to analyze and predict fault scenarios to identify and infer fault types and causes. The pneumatic training virtual platform extracts characteristic parameters (such as pressure P, temperature T, cylinder displacement x, and gas flow Q) from the equipment operation in real time, and inputs these characteristic parameters into a pre-trained classification model (such as a decision tree or neural network) to perform fault classification. The classification model constructs a nonlinear relationship between characteristic parameters and outputs possible fault types, such as cylinder failure or sensor signal distortion, to help users quickly locate the cause of the fault, and provide specific repair suggestions based on the model output to improve fault diagnosis efficiency.
[0076] The system provides users with a guided fault handling solution based on the fault type identification results. When the system identifies a specific fault (such as a cylinder jam fault), it will provide users with detailed repair steps based on the pre-set fault repair process:
[0077] First, explain the cause of the fault. For example, the cylinder may be stuck due to insufficient air pressure or foreign objects blocking the cylinder. Then, the pneumatic training virtual platform guides the user to perform troubleshooting and repair operations in sequence. The user will check each item according to the system prompts and take corresponding repair measures. For example, the pneumatic training virtual platform may prompt the user to check whether the cylinder is moving normally, adjust the air pressure to the specified range, and reset the cylinder. In each repair step, the platform will monitor the user's operation and confirm the correctness of the user's operation through data feedback. After the user completes each step, the pneumatic training virtual platform will detect based on the real-time collected data, verify whether the operation meets the predetermined standards, and provide timely feedback information. If the step is executed properly, the pneumatic training virtual platform will continue to guide to the next step. Otherwise, the pneumatic training virtual platform will provide specific correction suggestions based on the error type to ensure that the user can effectively complete the repair process.
[0078] The classification model includes a decision tree and a neural network. The training of the classification model optimizes the parameters by minimizing the cross entropy loss function:
[0079]
[0080] Among them, y i is the actual fault type, It is the probability of the fault type predicted by the classification model. The pre-trained machine learning model can automatically identify the fault type based on the data collected in real time.
[0081] Step S5 specifically includes:
[0082] S51. Use a weight-based multi-factor scoring model to score the user's fault repair process:
[0083] S=ω1S accuracy +ω2Sefficiency +ω3S precision ,
[0084] Among them, S is the total score, S accuracy is the fault identification accuracy, S efficiency is the fault handling efficiency, S precision is the operation accuracy, ω1, ω2, ω3 are weight coefficients;
[0085] S52: Output a fault diagnosis report, which includes a fault identification process, a solution and improvement suggestions, and then upload the fault diagnosis report to a training management database.
[0086] Specifically, after generating scores based on the accuracy and efficiency of user operations, a detailed diagnostic report is output, including the fault identification process, solutions, and improvement suggestions, and then the fault diagnosis report is uploaded to the training management database. After the pneumatic fault virtual training is completed, the system automatically cleans up the real-time data and user interaction records related to the training session, including equipment operation data, fault diagnosis results, and repair step execution status, to ensure system performance and data security.
[0087] At the same time, the system archives the user's learning path and performance data, including the training tasks completed by the user, fault repair accuracy, operation efficiency, and progress during the training process. These data will be stored in a structured manner for subsequent analysis and use. By analyzing historical learning data, the system can provide personalized support for the next training and optimize subsequent training content. For example, based on the user's performance, the system can recommend training areas that need to be strengthened or provide relevant advanced training content. This personalized support improves the pertinence and effectiveness of training. By saving the user's progress data, the system can track the user's improvement in multiple training sessions, ensuring that each training session can be optimized according to the user's learning needs, thereby providing continuous learning feedback and growth paths to help users continuously improve their fault diagnosis and repair skills.
[0088] Example 2
[0089] AI-based pneumatic fault training simulation system, including:
[0090] The acquisition module is used to establish a real-time operating status database of pneumatic equipment and a fault database;
[0091] A modeling module is used to construct a virtual simulation laboratory scene model, wherein the virtual simulation laboratory model includes a virtual pneumatic test bench and pneumatic equipment model, and a pneumatic training virtual platform;
[0092] The simulation module is used to trigger the corresponding simulated fault scenarios based on the virtual simulation laboratory scenario model in combination with the real-time operation status database and the fault database;
[0093] AI training module, which uses machine learning algorithms to diagnose and identify simulated fault scenarios and provide fault repair guidance;
[0094] The scoring module is used to generate virtual fault diagnosis and scoring reports.
[0095] Specifically, in this embodiment, the AI training module performs real-time analysis of user operation data, can record the user's learning path and provide personalized feedback and scoring. Compared with the traditional fixed learning process, this system effectively improves the pertinence of learning and helps users steadily master the troubleshooting skills of pneumatic systems. The AI training module uses machine learning and data analysis technology to improve the user's learning effect. The simulation module is used to simulate fault scenarios in various pneumatic systems and to provide a way for users to interact with the system, ensuring that users can operate the experimental environment naturally and smoothly and obtain timely feedback.
[0096] The above embodiments are preferred implementations of the present invention and are not intended to limit the present invention. Any other changes or other equivalent replacement methods that do not deviate from the technical solutions of the present invention are included in the protection scope of the present invention.
Claims
1. AI-based pneumatic fault training simulation method, characterized in that: Includes steps: S1. Establish a real-time operating status database for pneumatic equipment and a fault database; S2. Construct a virtual simulation laboratory scene model, wherein the virtual simulation laboratory model includes a virtual pneumatic test bench and pneumatic equipment model, and a pneumatic training virtual platform; S3, based on the virtual simulation laboratory scenario model, combined with the real-time operation status database and fault database, trigger the corresponding simulated fault scenario; S4. Use machine learning algorithms to diagnose and identify simulated fault scenarios and provide fault repair guidance; S5. Generate virtual fault diagnosis and scoring reports to complete pneumatic fault training.
2. The AI-based pneumatic fault training simulation method according to claim 1 is characterized in that: In step S1, the fault database includes a basic layer, a component layer and a system layer. The basic layer is used to define common types of faults, the component layer is used to define fault types of specific components, and the system layer is used to define cross-component fault combinations; the annotation information of each layer of faults includes fault conditions and fault repair methods.
3. The AI-based pneumatic fault training simulation method according to claim 1 is characterized in that: Step S2 specifically includes: S21. Construct a virtual laboratory environment model through 3D modeling software, including simulating the environmental configuration and physical properties of the actual pneumatic laboratory; S22. Construct a virtual pneumatic test bench and pneumatic equipment model using 3D modeling software, including key components of simulated pneumatic equipment; S23. Build a virtual platform for pneumatic training based on Unity, including setting the physical parameters and behavioral parameters of key components.
4. The AI-based pneumatic fault training simulation method according to claim 1 is characterized in that: Step S3 specifically includes: S31, collecting real-time operating status data of pneumatic equipment through a virtual simulation laboratory scene model to simulate the normal operating status of actual pneumatic equipment; S32. The trainee manually selects or randomly generates fault conditions preset in the fault library. When the real-time operating status data meets the selected fault conditions, the corresponding simulated fault scenario is automatically triggered.
5. The AI-based pneumatic fault training simulation method according to claim 4 is characterized in that: The fault conditions include single fault conditions and combined fault conditions, and the combined fault conditions are defined by Boolean logic; the simulated fault scenarios include single fault scenarios and compound fault scenarios, and the compound fault scenarios are designed by multi-threaded architecture, wherein each fault scenario is processed by an independent thread; When the real-time operating status data meets the combined fault conditions, the compound fault scenario is triggered.
6. The AI-based pneumatic fault training simulation method according to claim 1 is characterized in that: The control of the virtual pneumatic test bench and the pneumatic equipment model includes realization through a computer terminal mode and an immersive VR equipment mode. The control of the computer mode specifically includes using the mouse, keyboard and virtual pneumatic test bench to interact with the virtual control panel in the pneumatic equipment model to select different components of the virtual pneumatic equipment as the test objects; The control of the immersive VR device mode specifically includes using gestures to select different components of the virtual pneumatic device as detection objects through a head-mounted display and a handle.
7. The AI-based pneumatic fault training simulation method according to claim 1 is characterized in that: Step S4 specifically includes: S41, extracting characteristic parameters from the real-time operating status data, and using the pre-trained classification model to classify faults according to the input characteristic parameters, and outputting the fault type; S42. Based on the identified fault type, the user is guided to perform fault troubleshooting and repair operations according to the fault repair method preset in the fault database.
8. The AI-based pneumatic fault training simulation method according to claim 7 is characterized in that: The classification model includes a decision tree and a neural network. The training of the classification model optimizes the parameters by minimizing the cross entropy loss function: Among them, y i is the actual fault type, It is the probability of the fault type predicted by the classification model. The pre-trained machine learning model can automatically identify the fault type based on the data collected in real time.
9. The AI-based pneumatic fault training simulation method according to claim 1 is characterized in that: Step S5 specifically includes: S51. Use a weight-based multi-factor scoring model to score the user's fault repair process: S=ω1S accuracy +ω2S efficiency +ω3S precision , Among them, S is the total score, S accuracy is the fault identification accuracy, S efficiency is the fault handling efficiency, S precision is the operation accuracy, ω1, ω2, ω3 are weight coefficients; S52: Output a fault diagnosis report, which includes a fault identification process, a solution and improvement suggestions, and then upload the fault diagnosis report to a training management database.
10. AI-based pneumatic fault training simulation system, characterized by: include: The acquisition module is used to establish a real-time operating status database of pneumatic equipment and a fault database; A modeling module is used to construct a virtual simulation laboratory scene model, wherein the virtual simulation laboratory model includes a virtual pneumatic test bench and pneumatic equipment model, and a pneumatic training virtual platform; The simulation module is used to trigger the corresponding simulated fault scenarios based on the virtual simulation laboratory scenario model in combination with the real-time operation status database and the fault database; AI training module, which uses machine learning algorithms to diagnose and identify simulated fault scenarios and provide fault repair guidance; The scoring module is used to generate virtual fault diagnosis and scoring reports.