A multi-crane collaborative operation simulation platform and examination system

Abstract

The present application relates to a multi-crane collaborative operation simulation platform in the field of data processing, which includes a server and several simulation operation cabins. The server is equipped with a regionalized collaborative control module and a digital twin model of the crane and the environment. The function of the regionalized collaborative control module to predict whether interference occurs between the various digital twin models based on the operation path and the real-time operation status of the crane and to provide feedback to the user can allow users to clearly know whether their operation direction is correct and how to adjust it to avoid interference with other users, so that users can perform the test operations correctly and avoid the problem of the test not being able to proceed normally due to interference between cranes; it can screen out candidates who are not only strong in personal ability, but also have strong team cooperation ability, or both, so as to meet the society's multi-level needs for talents.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a multi-crane collaborative operation simulation platform and examination system. Background Art

[0002] The existing intelligent examination system for cranes is a new type of examination data processing system for data collection and processing. Compared with the traditional manual observation of user status and scoring, the intelligent examination system for cranes adopts image processing methods, identifies user information through electronic devices, and collects images of markers during the examination process, combining them with assessment indicators to judge the assessment results.

[0003] However, the existing crane operating platform has the following defects: since cameras are set up in the examination area and multiple cameras acquire images separately, the user's operating actions may be blocked in the images from other perspectives, resulting in the examination system being unable to fully process the crane parameters and user's operating status at the examination site, resulting in inaccurate examination data processing; in addition, in the existing crane operating platform, if multiple cranes work at the same time, due to inconsistent user levels, path interference, load coordination, time synchronization and risk control between different cranes become extremely complex and changeable, and motion interference will occur between cranes, making it impossible to conduct simulated examinations. In order to ensure a normal examination, only one user can be tested at a time, and it is difficult to test the collaborative work of multiple users. It is also difficult to accurately process the crane actions and user operation data of different users, resulting in inaccurate examination evaluation results.

[0004] There is currently no effective solution to the problem that existing crane operating platforms are difficult to test the collaborative work capabilities of multiple users. Summary of the Invention

[0005] The present invention provides a multi-crane collaborative operation simulation platform and an examination system to solve the problem that existing crane operation platforms are difficult to test the collaborative working ability of multiple users.

[0006] In the first aspect, the present invention provides a multi-crane collaborative operation simulation platform, which includes a server and several simulation operation chambers. The server is equipped with a digital twin model of the crane, a digital twin model of the environment, and a regional collaborative control module. Each simulation operation chamber has a built-in VR device and an operation terminal. The VR device has a built-in physical simulation engine. The several operation terminals transmit the crane control data generated by the user's operation to their respective physical simulation engines. The physical simulation engine controls the crane's digital twin model to execute actions based on the crane control data and generates the crane's real-time operating status. The server obtains the real-time operating status of the crane corresponding to each operation terminal and transmits it to the corresponding VR device for display. Among them, the regional collaborative control module is used to collect and record the real-time operating status of the cranes in all simulation operation chambers, divide the operation area of ​​each crane digital twin model, predict the future motion trajectory of the crane digital twin model based on the crane control data and the crane's real-time operating status, and detect the operation path of the crane digital twin model in real time. Based on the operation path and the crane's real-time operating status, it predicts whether there is interference between the digital twin models and feeds it back to the user.

[0007] Furthermore, the crane control data includes joystick movements and button movements. The joystick movements include the joystick movement direction, movement amplitude, and movement duration. The button movements include the pressing and releasing time of the corresponding function button.

[0008] The real-time operating status of the crane includes: the speed and acceleration of the crane; the speed of the crane's rotation; the acceleration change of the crane; the angle change of the crane's boom; and the three-dimensional spatial coordinates of the crane's spreader.

[0009] Furthermore, the simulation operation chamber includes: a temperature control simulation unit that can precisely control the temperature range; a high-altitude shaking simulation unit installed on a six-degree-of-freedom vibration platform to simulate various shaking modes; a wind disturbance simulation unit equipped with an array of adjustable-speed blowers to simulate different wind field conditions; a visibility and light environment simulation unit with an integrated atomizer and special lighting adjustment device; and a multi-channel high-fidelity speaker array for playing real noise on the scene.

[0010] Furthermore, the simulation operation chamber has a built-in fault simulation module, which is used to collect and record the user's operation stability data and control accuracy data after the environmental control system changes the environment in the simulation operation chamber.

[0011] Furthermore, operational stability data includes: hook swing amplitude, number of equipment shakes, error rate, fault response time and number of operation interruptions;

[0012] Precision control data includes: positioning error of the load or hook, deviation of the hook's motion trajectory, speed control error of the crane, movement continuity of the crane and hook, and recovery speed of operation errors.

[0013] Secondly, the present invention also provides a multi-crane collaborative operation examination system, including a multi-crane collaborative operation simulation platform and a scoring and evaluation system. The multi-crane collaborative operation simulation platform is the multi-crane collaborative operation simulation platform described above; the digital twin model of the crane in the server is used to construct a virtual crane model that is consistent with the physical characteristics and dynamic performance of the actual crane; the digital twin model of the environment in the server is used to construct an interactive virtual operation scene; the digital twin model of the crane and the physical simulation engine built into the VR device jointly achieve real-time two-way synchronization between the virtual crane model and the simulated device status;

[0014] The scoring and evaluation system has a built-in artificial intelligence-based real-time recognition and adaptive evaluation module for operational intentions. This evaluation module introduces emergencies into the virtual exam, collects the examinee's multi-dimensional operational data, equipment status and environmental feedback, uses the Transformer model to analyze and identify the examinee's immediate operational intentions, and obtains dynamic scoring data based on this. The scoring and evaluation system also has a built-in scoring unit, which has a preset weighted scoring model that includes time efficiency, operational accuracy, the operational ability score output by the operational intention recognition module, a safety energy efficiency score that takes into account both safety and efficiency, and an emergency response score that reflects responsiveness. The weighted scoring model calculates the examinee's final total score based on crane control data, real-time crane operating status, operational stability data, control accuracy data and dynamic scoring data.

[0015] Furthermore, the regionalized collaborative control module in the server also has a built-in rotational learning function. The rotational learning function collects abnormal event data such as trajectory interference and load imbalance that occur in collaborative tasks, analyzes high-frequency collaborative problem patterns and risk scenarios, builds a problem pattern knowledge base, and generates targeted new test task templates based on the problem pattern knowledge base.

[0016] Furthermore, the AI-based real-time recognition and adaptive assessment module for operational intentions has a processing flow that includes: the emergency trigger module dynamically inserts events such as equipment failure, environmental mutation, or task change; the real-time data acquisition and feature extraction module collects the examinee's control instructions, gaze tracking data, and equipment and environmental status data, and generates a normalized feature vector; the Transformer model receives the feature vector, uses a multi-head self-attention mechanism to fuse and analyze multimodal features, and outputs the operational intention classification result and operational ability score Oc;

[0017] In the scoring unit, the safety energy efficiency score Sc is calculated by the safety efficiency dual evaluation algorithm module, which evaluates the safety indicators of the operation respectively. Sc 安全 and performance indicators Sc 效能 , and fused through dynamically adjusted weight α: ;

[0018] Emergency response score Ec 应急处置得分 Calculated by the rhythm adaptation and dynamic evaluation algorithm module, which is based on the candidate's actual reaction time Rt to the task or operation and the preset maximum allowable response time threshold Rmax: , and the performance of candidates can be classified accordingly;

[0019] The final total score is calculated as follows: , where Tf is the time taken by the examinee, Tmax is the maximum allowed time, Er is the error rate, and w1 to w5 are the weights of each scoring item.

[0020] Furthermore, the multi-crane collaborative operation examination system also includes a VR examination terminal, which integrates the functions of candidate identity recognition and automatic assignment of examination tasks.

[0021] Furthermore, the fault simulation module built into the simulation operation cabin can also preset or randomly generate a variety of fault scenarios including equipment operation failure, structural abnormality failure, control system delay failure and environmental mutation, and monitor the examinee's handling operation path, reaction time and response effectiveness in real time.

[0022] Compared with the related art, the present invention has the following beneficial effects:

[0023] 1. The regionalized collaborative control module predicts interference between digital twin models based on the operation path and real-time crane operation status and provides feedback to the user. This allows users to clearly understand whether their operation direction is correct and how to adjust it to avoid interference with other users. This allows users to correctly perform the test operations and avoids the problem of crane interference that may disrupt the test. In addition, the multi-crane collaborative operation test system based on the regionalized collaborative control module allows multiple candidates to simultaneously operate different cranes in a shared virtual scene to perform collaborative tasks. Therefore, multiple candidates operate the same object, which can reflect the collaboration between candidates in the test. Compared with traditional methods that only assess candidates' individual abilities but not their ability to collaborate with others, this invention can not only screen candidates with strong individual abilities but also strong teamwork skills, or both. This can meet society's multi-level talent needs: tasks requiring very precise operations can assign operators with strong individual abilities; tasks requiring less demanding operations but relatively large projects can assign operators with strong collaborative work abilities; tasks requiring both precise operations and relatively large and complex projects can assign operators with strong both individual abilities and collaborative work abilities.

[0024] 2. The crane's real-time operating status includes its speed and acceleration, its rotational speed, changes in its acceleration, changes in the crane boom angle, and the three-dimensional coordinates of its spreader. When using the examination system, monitoring crane control data and its real-time operating status can effectively improve the accuracy of data collection, thereby enhancing the precision of the examination assessment.

[0025] 3. During high-temperature and high-noise environments, the difficulty of assessing operational stability is increased; during periods of intense vibration, the requirements for control accuracy are increased. The system uses these changes to assess the candidate's adaptability to different environmental conditions. If the candidate maintains stable operation despite drastic environmental changes, the system will award additional points for adaptability. This direct inclusion of environmental parameters ensures fairness and authenticity in the scoring process, truly reflecting the candidate's operational and emergency response capabilities under various real-world operating conditions.

[0026] 4. The multi-crane collaborative operation exam system first constructed a 3D physical simulation and digital twin model. Candidates completed the exam by wearing VR headsets and performing computer simulations in the machine room. Compared to existing systems, this system's distinguishing feature is its environmental adaptability assessment mechanism. A specially constructed simulated operating cabin was built within the exam room, authentically recreating the on-site environment. It specifically simulated extreme operating conditions such as high temperatures and high-altitude swaying. It also simulated various fault scenarios to comprehensively assess candidates' ability to respond to failures in complex environments. Addressing the shortcomings of existing systems in constructing an operating environment and their inability to fully simulate extreme problems encountered during operation, this system innovatively introduced regionalized control and multi-crane collaborative mechanisms. This mechanism supports multiple candidates working together simultaneously and, by accumulating practical problems encountered during collaborative operation, creates a rich case library, supporting dynamic updating and learning optimization of subsequent exam items. Furthermore, the system integrates technical tools such as operation intention recognition and adaptive examination modes, dual safety and effectiveness assessment algorithms, and dynamic assessment algorithms to further enhance the accuracy of candidate operational capabilities. By applying the Transformer artificial intelligence algorithm, it effectively integrates various evaluation indicators and system parameters to generate a scientific and fair comprehensive scoring result.

[0027] The details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a system structure diagram of the multi-crane collaborative operation simulation platform in this embodiment. DETAILED DESCRIPTION

[0029] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0030] Unless otherwise defined, technical or scientific terms used in this application shall have the ordinary meanings as understood by persons of ordinary skill in the art to which this application belongs. The terms "a," "an," "the," "these," and similar expressions in this application do not denote limitations on quantity and may be singular or plural. The terms "comprise," "include," "have," and any variations thereof, as used in this application, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or device comprising a series of steps or modules (units) is not limited to the listed steps or modules (units) but may include unlisted steps or modules (units) or other steps or modules (units) inherent to the process, method, product, or device. The terms "connected," "connected," "coupled," and similar expressions used in this application are not limited to physical or mechanical connections but may include electrical connections, whether direct or indirect. As used in this application, "plurality" means two or more. "And / or" describes an association between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can mean: A exists alone; A and B exist simultaneously; or B exists alone. Generally, the character " / " indicates that the objects in the preceding and following relationship are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0031] In an embodiment of the present invention, a multi-crane collaborative operation simulation platform is provided. Figure 1 ,The multi-crane collaborative operation simulation platform includes a server and several ,simulation operation cabins.

[0032] The server houses a digital twin model of the crane, a digital twin model of the environment, and a regional collaborative control module. The model details structural parameters (such as boom length, slewing radius, hook shape, tower height, and track structure dimensions), material properties (such as component rigidity, elastic modulus, and strength parameters), and physical characteristics (such as equipment inertia, slewing damping coefficient, load limits, and dynamic response). Once modeled, these 3D models are imported into a real-time physics simulation engine like Unity3D or Unreal Engine UE5 for further tuning of the materials and physical parameters to ensure the dynamic response of the virtual model accurately matches the actual equipment. The core of the digital twin model lies in real-time mapping of the real or simulated equipment state and synchronizing it with the virtual equipment state.

[0033] Each simulated operation chamber features a built-in VR device and operator terminal. The VR device includes a built-in physics simulation engine. These include a high-resolution VR headset (such as MetaQuest Pro or HTC Vive Pro), a dual-handle controller with precise force feedback, and optional foot pedals or a simulated cockpit controller. These simulate a real-world operating environment. The VR SDK is integrated and configured, and VR interaction logic is programmed. Users wearing the VR headset experience an immersive three-dimensional virtual field of view, clearly displaying the crane's equipment status, operating environment information, and real-time task objectives. Several operator terminals transmit the crane control data generated by user operations to their respective physics simulation engines. The physics simulation engines use this data to control the crane's digital twin model, executing actions and generating the crane's real-time operating status. The server then retrieves the real-time crane operating status corresponding to each operator terminal and transmits it to the corresponding VR device for display. Crane control data includes joystick and button movements. Joystick movements include the direction, amplitude, and duration of movement, while button movements include the time it takes to press and release the corresponding function button. The crane's real-time operating status includes its speed and acceleration, its rotational speed, changes in its acceleration, changes in the crane's boom angle, and the three-dimensional coordinates of its spreader. When using a multi-crane collaborative operation testing system, monitoring crane control data and its real-time operating status can effectively improve the accuracy of data collection, thereby enhancing the accuracy of the test assessment.

[0034] The regionalized collaborative control module collects and records the real-time operating status of cranes in all simulated control rooms, divides the operating zones of each crane's digital twin model, and predicts the future motion trajectory of the crane's digital twin model based on crane control data and the crane's real-time operating status. It also monitors the crane's digital twin model's operating path in real time. Based on the operating path and the crane's real-time operating status, it predicts whether interference between the digital twin models will occur and provides feedback to the user. Low latency and high synchronization are crucial for real-time collaborative simulation. State Synchronization: The server broadcasts key information such as the crane's position, posture, and hook status to all simulated control rooms periodically or upon state changes. Input Synchronization: The simulated control room sends user operation commands (such as handle movement and button presses) to the server, which processes and updates the global state. Prediction and Smoothing: The simulated control room can use prediction algorithms (such as dead reckoning) to compensate for network latency, making other users' movements appear smoother. The function of predicting whether interference occurs between the various digital twin models based on the above-mentioned operation path and the real-time operation status of the crane and providing feedback to the user allows users to clearly know whether their operation direction is correct and how to adjust it to avoid interference with other users, thereby allowing users to perform test operations correctly and avoiding the problem of interference between cranes causing the test to be unable to proceed normally.

[0035] The regionalized collaborative control module also includes a safety monitoring module, which is used to issue an alarm and / or perform emergency braking when interference is about to occur between the various digital twin models, and can utilize the collision detection function of the physical simulation engine.

[0036] The simulated operating chamber is equipped with an environmental control system, which is used to modify the working environment within the chamber. This system includes a temperature control simulation unit, a high-altitude vibration simulation unit, a wind disturbance simulation unit, a visibility and light environment simulation unit, and an acoustic interference unit. The temperature control simulation unit controls the chamber's temperature. Equipped with an industrial-grade hot air generation system and infrared radiation heating components, it precisely and stably maintains the chamber's ambient temperature between 35°C and 60°C, simulating realistic outdoor working conditions such as scorching sunlight and extreme heat. A real-time temperature sensor ensures temperature control accuracy within ±0.5°C. All temperature data is uploaded to the system's scoring platform in real time as part of the assessment data. The high-altitude vibration simulation unit controls the chamber's vibrations. The chamber is mounted on a six-degree-of-freedom vibration simulation platform, capable of applying various vibration modes during the test, such as the continuous, gentle swaying of a crane at high altitude, or the random, impact-like vibrations of sudden equipment movement or earthquakes. The amplitude, frequency, and pattern of the sway can be dynamically adjusted in real time by the system control platform to match the pace and difficulty of the test tasks, creating a dynamic operating environment. A multi-axis accelerometer monitors the platform's sway data in real time and transmits environmental status data to the system's scoring module. The wind disturbance simulation unit simulates the wind field within the simulated operating chamber. An array of high-power, adjustable-speed blowers surrounds the chamber, simulating realistic wind conditions such as transient wind disturbances, gusts, and sustained crosswinds. When used in the multi-crane collaborative operation test system, built-in high-precision wind speed and pressure sensors collect environmental wind disturbance parameters in real time and upload them to the AI ​​scoring model for dynamic, coordinated assessment. The visibility and light environment simulation unit controls the light intensity within the simulated operating chamber. Equipped with an atomizer and special lighting control devices, it simulates rain, fog, smoke, low-visibility weather, and light interference during nighttime operations. When used in the multi-crane collaborative operation test system, the system monitors changes in light intensity and visibility in real time, using these as key data parameters for system scoring. The acoustic environment interference unit is used to play realistic construction site noise within the simulated operation chamber. Equipped with multi-channel, high-fidelity speakers, it can reproduce authentic construction site noise, including mechanical sounds from cranes, alarms, and voices from personnel. When used in the multi-crane collaborative operation test system, the system performs real-time sound intensity detection and uploads it to the scoring platform for a quantitative assessment of auditory environment interference.

[0037] The simulated operation chamber has a built-in fault simulation module. When used in the multi-crane collaborative operation test system, this module collects and records operational stability and control accuracy data after the environmental control system changes the environment within the simulated operation chamber. Operational stability data includes hook swing amplitude, equipment shake frequency, error rate, fault response time, and number of operation interruptions. Control accuracy data includes load or hook positioning error, hook trajectory deviation, crane speed control error, crane and hook movement consistency, and operational error recovery speed. The specific implementation process is as follows: All environmental parameter data is collected in real time and introduced into the simulated operation chamber at the appropriate test stage. The environmental parameter settings for each stage are carefully designed to realistically simulate the complex conditions of the construction environment. For example, the operational stability test is more challenging during high temperature and high noise environments, while the control accuracy test is more demanding during severe vibration environments. The system uses these changes to assess the examinee's ability to adapt to different environmental conditions. If the examinee maintains stable operation despite drastic environmental changes, the system will award additional points for adaptability. The direct participation of such environmental parameters ensures the fairness and authenticity of the scoring, and truly reflects the examinees' operational capabilities and emergency response capabilities under various actual working conditions. Compared with the singleness of the examination environment in the prior art, the environmental adaptability assessment mechanism of the present invention and the supporting simulation operation chamber significantly improve the comprehensive assessment capability of the multi-crane collaborative operation examination system, realize the real simulation of complex and extreme working conditions in the construction environment, dynamically adjust environmental factors to enhance the authenticity of the assessment, assess the examinees' real operational skills and emergency response capabilities under actual conditions, and improve the accuracy, fairness and comprehensiveness of the examination, thereby helping to significantly improve the actual operation level of crane operators. The fault simulation module built into the simulation operation chamber can also preset or randomly generate a variety of fault scenarios including equipment control failures, structural abnormality failures, control system delay failures and environmental mutations, and monitor the examinees' handling operation paths, reaction times and response effectiveness in real time.

[0038] Secondly, considering that existing artificially constructed operating scenarios may be incomplete and fail to fully simulate the extreme problems that candidates may encounter during operation, the platform's regionalized collaborative control module further introduces regionalized control and multi-crane collaboration mechanisms. This allows multiple candidates to take the exam simultaneously, operating different cranes in a coordinated manner. In this way, the system can accumulate various problems that arise during collaborative operation as valuable case studies. Through a rotational learning mechanism, these cases can be incorporated into subsequent exam items, continuously optimizing and enriching the exam content. Existing crane exam systems typically use a single-person, single-machine exam model with manually preset task scenarios. This results in a relatively simple operating environment and task difficulty, and fails to fully simulate the complexity and unpredictability of multi-machine collaborative operations in actual construction. In particular, in multi-crane collaborative operation scenarios, issues such as path interference, load coordination, time synchronization, and risk management between different cranes become extremely complex and variable. Existing exam systems struggle to effectively assess these issues and cannot truly test candidates' teamwork awareness, communication skills, and risk management capabilities. To address these shortcomings, this system introduces regionalized control and multi-crane collaboration mechanisms into the exam system, aiming to enable multiple candidates to take the exam simultaneously and interact with each other in real time. It also features automatic accumulation of collaborative problems and rotational learning from the case library. The development of this mechanism primarily relies on the establishment of a regionalized collaborative control mechanism. Within the virtual simulation system, a collaborative testing platform capable of supporting multiple candidates simultaneously is constructed. Each candidate is equipped with a separate VR headset and operating terminal, controlling a different virtual crane model (such as a tower crane, crawler crane, or truck crane) to simulate the collaborative operation of multiple equipment on a real-world construction site. When the system launches the test, the virtual work area is first divided into multiple logical sub-areas. Each crane has its own safe operating zone, transition zone, and restricted zone. Each zone is defined by clearly defined boundary conditions and safety distances. The system utilizes a real-time trajectory prediction algorithm to dynamically detect whether each crane's trajectory is likely to enter another crane's safe zone or cause path interference. If the trajectory prediction module identifies that the booms or loads of multiple cranes may enter a collision risk zone, the system triggers a warning signal 5-10 seconds in advance, prompting the candidate to make path corrections. When the test task requires multiple cranes to jointly lift the same load (such as an extra-long steel beam or bulky equipment), the system automatically generates a collaborative task and sets the initial position and lifting target for each crane. During operation, the system monitors data from load sensors on each crane's spreader in real time, such as hook tension, load center deviation, and spreader tilt angle, to determine whether each candidate has effectively achieved even load distribution and synchronized lifting and lowering. If uneven loads or lifting offsets occur, the system records the time and cause of the issue, as well as the specific operator's actions, creating a data record that is then fed into the AI ​​scoring model.

[0039] The present invention provides a multi-crane collaborative operation examination system, including a multi-crane collaborative operation simulation platform and a scoring and evaluation system. The multi-crane collaborative operation simulation platform is the above-mentioned multi-crane collaborative operation simulation platform; the digital twin model of the crane in the server is used to construct a virtual crane model that is consistent with the physical characteristics and dynamic performance of the actual crane, and the digital twin model of the environment in the server is used to construct an interactive virtual operation scene. The digital twin model of the crane and the physical simulation engine built into the VR device jointly realize the real-time two-way synchronization of the virtual crane model and the simulation device status; the simulation operation cabin is built-in with a unit for reproducing high temperature, high-altitude shaking, strong wind, limited visibility and complex acoustic environment, and combined with a configurable virtual fault trigger mechanism to comprehensively evaluate the candidate's operating skills and emergency response capabilities in a close-to-real harsh environment; the regionalized collaborative control module of the server built into the multi-crane collaborative operation simulation platform supports the regionalized control and dynamic learning mechanism of multi-crane collaborative operation, allowing multiple candidates to share a virtual environment. In the simulated scenario, different cranes are controlled simultaneously to perform collaborative tasks. The mechanism includes dynamic division of the working area and anti-collision warning, real-time monitoring of collaborative loads, and a rotational learning function that can automatically record and analyze collaborative errors and optimize subsequent test content accordingly; the scoring and evaluation system has a built-in real-time recognition and adaptive evaluation module for operational intentions based on artificial intelligence. This module introduces emergencies into the virtual test, collects the examinee's multi-dimensional operational data, equipment status and environmental feedback, and uses the Transformer model to analyze and identify the examinee's immediate operational intentions, and obtains dynamic scoring data accordingly; the scoring and evaluation system also has a built-in scoring unit (multi-dimensional comprehensive scoring engine), which has preset time efficiency, operational accuracy, the operational ability score (Oc) output by the operational intention recognition module, the safety energy efficiency score (Sc) that takes into account both safety and efficiency, and the emergency response score that reflects response agility. Ec 应急处置得分 The weighted scoring model calculates the candidate's final total score based on crane control data, crane real-time operating status, operation stability data, control accuracy data and dynamic scoring data.

[0040] The regionalized collaborative control module also has a built-in rotational learning function. The rotational learning function collects abnormal event data such as trajectory interference and load imbalance that occur in collaborative tasks, analyzes high-frequency collaborative problem patterns and risk scenarios, builds a problem pattern knowledge base, and automatically generates targeted new test task templates based on the problem pattern knowledge base.

[0041] The AI-based real-time recognition and adaptive assessment module for operational intentions has the following processing flow: the emergency event trigger module dynamically inserts events such as equipment failure, environmental mutation, or task change; the real-time data acquisition and feature extraction module collects the examinee's control instructions, gaze tracking data, and equipment and environmental status data, and generates a normalized feature vector; the Transformer model receives the feature vector, uses the multi-head self-attention mechanism to fuse and analyze the multimodal features, and outputs the operational intention classification result and the operational ability score (Oc).

[0042] The system also includes a VR examination terminal configured for each candidate, which integrates candidate identity recognition and automatic examination task allocation functions.

[0043] The fault simulation module built into the simulation operation cabin can also preset or randomly generate a variety of fault scenarios including equipment operation failure, structural abnormality failure, control system delay failure and environmental mutation, and monitor the examinee's handling operation path, reaction time and response effectiveness in real time.

[0044] During operation, the system also monitors load sensor data on each crane's spreader in real time, such as hook tension, load center deviation, and spreader tilt angle, to determine whether each participant is effectively achieving even load distribution and synchronized lifting and lowering movements. In advanced virtual crane simulation systems, load sensors are not physical hardware installed on the virtual equipment. Instead, they are simulated through sophisticated software algorithms and physics engine computational models. These virtual sensors are cleverly integrated into the simulation environment's core computing logic, generating and outputting various mechanical data in real time at key structural points in the crane model, providing the same performance as real sensors. First, a mechanical calculation point is set at the hook node, where the hook connects to the wire rope. This point calculates and outputs vertical tension data in real time. Using this data, the system accurately monitors the actual load weight shared by each participating crane. Second, at the connection points between the spreader and the load, given that large or irregular loads may require multiple lifting points, the system incorporates mechanical sensors at each or more key connection points between the spreader and the load. These sensor points are used to calculate the force distribution at each connection point, which is crucial for determining whether the load is evenly distributed across the spreader and whether there is any unbalanced load. Furthermore, spatial attitude detection points are located on the spreader itself, such as at each end or at other key locations that reflect its overall posture. These detection points primarily calculate the angle between the spreader and the horizontal plane, thereby monitoring any tilt in real time and ensuring a smooth lifting process. The monitoring system and its underlying structure and principles: Tension monitoring: The system relies on its built-in physics engine to accurately calculate the tension acting on the wire rope. The results are displayed in real time, clearly indicating to the observer or test-taker the specific load weight currently being shared by each crane. This is particularly important when multiple cranes are working together to lift the same load, as it allows for intuitive assessment of whether the load distribution has achieved the desired balance. Load center deviation monitoring: The system calculates the precise positional relationship between the load's geometric center and each connection point on the spreader. Based on this spatial data, the system can analyze in real time whether the load's actual center of gravity has deviated from its ideal, safest and most stable position. This monitoring effectively assesses the candidate's ability to maintain load balance during operation. Spreader tilt angle monitoring: The system applies the mathematical principles of a three-dimensional coordinate system to accurately calculate the spreader's specific position in three-dimensional space. This monitoring accurately indicates whether the spreader remains level when multiple cranes are working together to lift or move a large spreader (such as a shoulder beam). This directly impacts the assessment of the candidate's control ability during delicate operations such as synchronized lifting and lowering.

[0045] The method for recording candidate operation data is as follows:

[0046] 1. Real-time data collection.

[0047] The system collects the examinee's operational data in real time through virtual load sensors and control interfaces. The data is collected frequently to ensure a complete record of every detail of the action.

[0048] 2. Data storage and management.

[0049] Data is stored in time series format for easy analysis. Each operation is associated with a specific timestamp and device status.

[0050] 3. Abnormal event triggering.

[0051] When an abnormality is detected (such as track interference or load imbalance), the system automatically records the operation data before and after the abnormality occurs, including the type of abnormality, the time of the abnormality, the candidate's operation details, and the equipment status.

[0052] 4. Data analysis and feedback.

[0053] The recorded data is used by the AI ​​scoring model to analyze and assess the examinee's operational capabilities and the causes of errors. The system can also generate operational reports and provide feedback to help examinees improve.

[0054] The scoring and evaluation system is equipped with an AI scoring model. The training process of the custom AI scoring model includes:

[0055] 1. Collect standard handling data of professional crane operators in various fault scenarios.

[0056] 2. Establish a benchmark data set based on the scores of different levels of disposal methods given by previous experts.

[0057] 3. Extract key features (such as reaction speed, operation accuracy, decision correctness, etc.).

[0058] 4. The model is trained using supervised learning methods, combining decision tree and neural network algorithms. This trained model can more accurately assess candidates' emergency response capabilities and risk management capabilities when faced with various types of failures. By comprehensively recording candidates' operational data, the system can analyze collaborative work issues in detail, support a rotating learning mechanism, and help candidates continuously improve their operational skills and collaborative abilities.

[0059] The multi-crane collaborative operation examination system also includes a case database system, which is used to collect and record operation stability data, control accuracy data, crane control data, and problem data in the real-time operation status of the crane that does not meet the set goals, and upload the examination tasks corresponding to the problem data as cases to the cloud database, and subsequently splice new examination content based on the content of the cloud database.

[0060] Secondly, the system has established a mechanism for accumulating and rotating collaborative operation cases. During the collaborative exam, if problems such as trajectory interference, load imbalance, lifting synchronization failure, or path conflict occur, the system will automatically trigger an exception event recording function. The exception event record includes the type of exception (such as load tilt or path conflict), the candidate's operational data before and after the exception, the equipment status and trajectory, and the severity and duration of the exception. The system regularly and automatically uploads accumulated collaborative exception cases to a cloud database, forming a "collaboration problem case library." Based on this case library, the system analyzes frequently occurring collaborative operation problem patterns and risk scenarios, such as uncoordinated lifting speeds between multiple cranes leading to load tilt, improper collaborative path design leading to cross-lift collisions, or inconsistent operator rhythms leading to synchronized handling failures. By clustering and learning features from the time series data of these historical cases, the system automatically extracts typical behavioral patterns and scenario characteristics that lead to problem occurrences, thereby building a problem pattern knowledge base. Based on this collaborative case knowledge base, the system automatically generates task templates for the next batch of exams. These new task templates will systematically add targeted training subjects to address frequently occurring collaborative problems. For example, if load imbalances frequently occur, the system will automatically create dedicated synchronized load-lifting tasks. If path conflicts are frequent, the system will generate tasks that intentionally intersect but require precise coordination, training candidates in advance planning, effective communication, coordinated movement, and synchronized execution. This rotating learning mechanism ensures that exam content is constantly updated, allowing candidates to continuously tackle more challenging collaborative work scenarios that are more realistic and relevant to actual work demands during different exam periods, effectively enhancing their collaboration and adaptability in real-world construction scenarios.

[0061] By implementing the aforementioned regionalized control and multi-crane coordination mechanism, the system has achieved a significant breakthrough from a "single exam scenario" to a "complex, multi-crane collaborative scenario." This shift comprehensively enhances candidates' teamwork, risk management awareness, and operational skills in real-world operations, significantly enhancing the authenticity, systematic nature, and intelligent nature of crane operation assessments. Furthermore, the system can dynamically generate more cases for emergencies and extreme operating conditions. These cases will be incorporated into the case database for subsequent assessment training, evaluation, and analysis, further improving the practicality and coverage of the multi-crane collaborative operation examination system.

[0062] Case collection mechanisms include:

[0063] 1. Real-time monitoring of multi-crane collaborative operations, automatic marking of abnormal events (collisions, misoperations, collaborative failures, etc.); integration of operation logs with environmental data (wind speed, temperature, vibration, etc.); manual marking of special events or complex collaborative issues by staff; automatic classification and difficulty grading, removing duplicate cases and retaining representative examples.

[0064] 2. Case conversion into exam content: Structured modeling: Abstract key events into standardized exam scenarios, extracting environmental parameters and operational elements; Simulation scenario generation: Automatically reproduce case scenarios and generate variants of varying difficulty; Rotational learning mechanism: Newly emerging typical problems are automatically entered into the case library, forming a closed loop of "case-exam-re-case"; Dynamic push: Intelligently select relevant case scenarios based on the candidate's performance; The overall process forms a loop: Multi-crane collaborative exam → Case collection → Classification and grading → Structured processing → Generation of exam scenarios → Continuous optimization of new cases, ensuring that the exam content is closely integrated with real-world working conditions and continuously evolves.

[0065] In addition to the above-mentioned innovations, the multi-crane collaborative operation examination system also includes: operation intention recognition mode, safety and efficiency dual evaluation algorithm, beat adaptation and dynamic evaluation algorithm and other algorithms to further accurately evaluate the candidate's operation ability, and use the transformer artificial intelligence algorithm to string all system parameters together to finally obtain the score.

[0066] Traditional crane test systems only offer fixed task modes. While these systems incorporate unexpected events, they fail to fully and accurately assess the candidate's actual operational intentions and ability to respond to emergencies in real-world working conditions. Therefore, we have proposed an innovative operational intention recognition model. This model introduces a variety of random or preset emergencies into virtual test tasks, leveraging an artificial intelligence (AI) Transformer model to identify the candidate's operational intentions in real time and provide accurate scores, thereby objectively evaluating the candidate's true ability to respond to complex and unexpected situations.

[0067] The operational intent recognition model consists of three collaborative submodules: an emergency event triggering module, a real-time data acquisition and feature extraction module, and a Transformer-based intent recognition and scoring module. Together, these modules complete the entire process of intent recognition and capability assessment.

[0068] First, the emergency event triggering module is responsible for automatically inserting various emergency events, either pre-planned or randomly, during the candidate's normal virtual lifting operation, thereby testing the candidate's real-time reaction ability. These emergency events are diverse and cover unexpected conditions within the lifting equipment itself, such as sudden and violent hook swing, virtual sling breakage or loosening, uncontrolled boom swaying, and delayed or stalled controller response. They also include emergencies related to the operating environment, such as a sudden strong wind causing hook swing, heavy fog or light interference in the operating environment, virtual ground collapse, or sudden platform shaking. Furthermore, there are task change events, such as the sudden appearance of an obstacle in the lifting path or changes in the position or weight of the designated task target. These emergency events can be preset or randomly generated to ensure diversity and unpredictability in the test scenarios. The system has an event trigger controller that dynamically inserts these emergency events during the task process based on the task progress or the candidate's current operating status. For example, if the candidate operates smoothly for a long time, the system may automatically trigger an environmental disturbance event to examine his response to emergencies; if the candidate shows hesitation in operation or slow response, the system may trigger an equipment failure or path mutation event to test the candidate's real-time emergency decision-making ability.

[0069] Next, to accurately identify the examinee's operational intent, the real-time data acquisition and feature extraction module collects multi-dimensional data in real time and extracts effective features from it. Regarding operational data, the system collects control commands from the examinee using the VR controller and foot pedals in real time, including joystick displacement, key sequence, and button response speed. It also records helmet-mounted gaze tracking data, such as the examinee's gaze resting point and gaze shift frequency, as well as the examinee's first reaction action, action amplitude, and response delay when an event is triggered. Regarding equipment status and sensor data, the system collects data such as the virtual boom inclination angle, hook height, load tension, and swing amplitude, as well as real-time environmental sensor data including virtual wind speed, visibility, and light intensity. This collected data is processed using linear embedding and sliding time window techniques to generate normalized feature vectors, such as the operational action sequence feature vector, the equipment response timing feature vector, and the environmental disturbance change feature vector.

[0070] Finally, the Transformer-based intent recognition and scoring module is the core of the entire operation intention recognition model. First, the normalized feature vector set generated above, namely the operation action sequence, device state sequence, and environmental disturbance sequence, will be encoded into a feature representation vector of uniform length through embedding:

[0071] X feature vector = [X operation action, X device response, X environmental disturbance].

[0072] Subsequently, the X eigenvector is transformed to obtain the query matrix Q, key matrix K, and sum matrix V, and the obtained matrix is ​​input into the encoder of the Transformer model. The multi-head self-attention mechanism in the Transformer is used, and the calculation formula is:

[0073] .

[0074] Among them, Q is the query matrix, K is the key matrix, V is the sum matrix, d k is the dimension of the query vector or key vector in a single attention head, and T is the length of the time series. In the Transformer model used for real-time intent recognition in crane exams, there are several key symbols that define its structure and data processing methods. The first is the symbol d, which represents the total feature dimension of the model. This dimension refers to the vector length of the single time step input processed by the Transformer model. In actual exam scenarios, the system will collect features from multiple sources, such as control instructions, the examinee's gaze tracking data, the crane's own equipment status, and the surrounding environment status. These features from different sources initially have different dimensions, such as 64, 32, 96, and 32 respectively. In order for the model to process them uniformly, these features are concatenated and mapped to the same dimension d through a linear projection, which is set to 256 in this example.

[0075] Next, let’s look at the symbol k. It is usually implied in the formula as a subscript, representing the number of the Key subspace. Specifically, in the scaling factor d of the attention mechanism k In, d k Indicates the dimension of the query vector or key vector in a single attention head. If the model uses a multi-head attention mechanism, for example, 8 heads are set, then the total feature dimension d will be evenly divided into each head. Therefore, d k The value of is the total dimension divided by the number of heads, that is, 256 divided by 8, which is 32. The advantage of this is that each attention head only calculates the correlation in a subspace of dimension 32, which can effectively prevent the calculated inner product value from being too large, thereby avoiding the gradient saturation problem of the subsequent SoftMax function and making the model training more stable.

[0076] Finally, the symbol T defines the length of the time series. This refers to the total number of discrete time steps processed by the Transformer model during a single forward inference pass, or in other words, the length of the input sequence. In exam applications, the model needs to analyze continuous behavior over a period of time. For example, the system can be configured to sample data every 0.1 seconds, with the most recent 3 seconds of data used as an analysis window. In this case, the length of the input sequence, T, is 30. The model simultaneously examines the evolution of the behavior over these 30 time steps, outputting a real-time assessment of the candidate's operational intent and a performance score. The model is capable of deep fusion and attention analysis of multimodal features. Through this self-attention mechanism, the model automatically identifies potential patterns and anomalies in the candidate's operational data and determines their operational intent, such as a stable recovery attempt (e.g., rapid and stable hook swing), an emergency braking attempt (e.g., an abrupt stop to prevent escalation of danger), a path correction attempt (e.g., circumventing an unexpected obstacle or dangerous area), or a hesitant attempt (e.g., slow response or uncertain operation). The output layer of the Transformer model is connected to a fully connected classifier. This classifies the candidate's operational intent in real time based on the fused feature encoding vector, outputs the intent recognition result, and provides a corresponding score in a very short time (less than 1 second). For example, a clear and effective emergency response intention will receive a high score; a hesitant and uncertain intention will receive a medium score; and an incorrect or inefficient emergency response intention will receive a low score. These scores are normalized by the softmax function to produce a clear Oc (operational ability score) which is added to the previous scoring formula:

[0077] Final total score .

[0078] This exam system also innovatively incorporates a dual safety and efficiency assessment algorithm, designed to balance the safety and efficiency of the examinee's operational behavior. This algorithm first examines the examinee's operational stability and risk mitigation capabilities through the safety assessment module. This module focuses on metrics such as hook swing amplitude and frequency, boom tilt exceeding limits, peak acceleration and braking shock, and uneven load distribution and excessive stress. These key parameters are collected in real time by the sensor array in the perception layer, pre-processed, and input into the system to generate an objective safety assessment score. Meanwhile, the efficiency assessment module focuses on assessing the examinee's efficiency and cost-effectiveness in completing the task. Evaluation metrics include task completion time, path planning efficiency (including path length for each stage of lifting, transporting, and placing), estimated energy consumption during the operation (such as fuel or electricity consumption), and statistics on action repetition rate and ineffective actions. This efficiency-related data is also collected by the data acquisition module and input into the system for analysis.

[0079] The system uses a dynamic weight adjustment mechanism based on rules and statistical distribution to automatically analyze and optimize the weights of the two evaluation dimensions, safety and efficiency, in real time. Specifically, the system quantitatively evaluates key indicators during the operation process (such as the number of incorrect operations, response time, and operational efficiency), and dynamically adjusts the weights of safety and efficiency using pre-defined scoring rules and the statistical distribution of historical data. Ultimately, the system outputs a fused comprehensive evaluation score that balances the two key dimensions of safety and efficiency, resulting in a safety and efficiency score Sc, as follows:

[0080] . Among them: Sc 安全 : Indicates the score of the safety dimension, that is, the safety index, which is scored based on the examinee's operational standardization, safety awareness and risk control ability. 效能 : represents the score of the effectiveness dimension, i.e., the effectiveness index, which primarily assesses the candidate's operational efficiency, task completion quality, and time performance. α: is the safety weight parameter, calculated by the system in real time, which reflects the relative importance of safety and effectiveness in the current task environment.

[0081] The final score formula is updated as follows:

[0082] Final total score .

[0083] In addition, Honda's crane collaborative operation test system uses a simple and effective dynamic evaluation algorithm based on key reaction time to examine the performance stability and real-time adaptability of candidates under different operation rhythms. The test accuracy is higher. The system collects the candidate's action response time Rt (unit: second) in real time. The response time represents the candidate's actual reaction time to the task or operation at the moment. The system presets a maximum response time threshold Rmax allowed for the operation, which is used to measure the candidate's reaction sensitivity and rhythm adaptability. The candidate's rhythm agility emergency response score Ec 应急处置得分 Calculated by the following formula:

[0084] .

[0085] Where Rt is the actual response time of the examinee. When the examinee's actual response time Rt is closer to zero, the score is closer to 100, indicating agile and fast operation rhythm. When Rt approaches or exceeds Rmax, the score approaches zero or a negative value, indicating poor rhythm adaptability and reaction ability. Finally, the examinee's final score formula is:

[0086] Final total score .

[0087] Among them, Tf: the time required for the examinee to complete the test; Er: error rate, that is, the ratio of incorrect operation times to the total number of operations; Oc: operation ability score; Sc: safety energy efficiency score; Ec 应急处置得分 : Emergency response score; w1, w2, w3, w4, w5: The weights of each scoring indicator, which can be adjusted according to the examination objectives and focus.

[0088] Description: 1. :It reflects the time efficiency by converting the time used by the examinee to the maximum allowed time. 2. :Indicates that the accuracy of the operation is reflected through the successful operation rate. 3. : Indicates the coordination critical reaction time to evaluate agility. Ec 应急处置得分 The system further applies rule-based classification criteria to categorize examinees' tempo performance into several typical categories, such as: Fast and Precise (score > 80): Swift and stable operation; Steady (score between 50 and 80): Moderate tempo and balanced performance; Hesitant (score < 50): Slow response, with potential pauses and uncertainty. This classification result is displayed on the final score and communicated to the examinee. After each group of examinees completes the exam, the system records their task completion time and quality. The group's task completion time and quality are used to assess the group's collaborative ability, while the individual examinee's final score is used to assess their individual ability. Ultimately, the system can identify examinees with strong individual abilities but weak collaborative abilities, examinees with weak individual abilities but strong collaborative abilities, and examinees with both strong individual abilities and strong collaborative abilities. This approach addresses society's multi-faceted talent needs: tasks requiring high precision can assign operators with strong individual abilities; tasks requiring low operational requirements but large-scale projects can assign operators with strong collaborative abilities; and tasks requiring both high precision and large-scale, complex projects can assign operators with both strong individual abilities and strong collaborative abilities.

[0089] In summary, the multi-crane collaborative operation test system first constructs a three-dimensional physical simulation and digital twin model. Candidates complete the test by wearing VR headsets and performing computer simulations in the machine room. Compared to existing systems, this system's distinguishing feature is its environmental adaptability assessment mechanism. A specially constructed simulated operating cabin is located within the examination room, authentically recreating the on-site environment. It specifically simulates extreme operating conditions such as high temperatures and high-altitude swaying. It also simulates various fault scenarios to comprehensively assess candidates' ability to respond to failures in complex environments. Addressing the shortcomings of existing systems in constructing an operating environment and their inability to fully simulate the extreme challenges encountered during operation, this system innovatively introduces regionalized control and multi-crane collaborative mechanisms. This mechanism supports multiple candidates performing collaborative operations simultaneously and, by accumulating practical problems encountered during collaborative operations, forms a rich case library, supporting dynamic updating and learning optimization of subsequent exam items. Furthermore, the system integrates technical tools such as operation intention recognition and adaptive examination modes, dual safety and effectiveness assessment algorithms, and dynamic assessment algorithms to further enhance the accuracy of candidate operational competence assessments. By applying the Transformer artificial intelligence algorithm, it effectively integrates various assessment indicators and system parameters, ultimately generating a scientific and impartial comprehensive score based on scoring units.

[0090] It should be understood that the specific embodiments described herein are only used to explain this application and are not used to limit it. Based on the embodiments provided in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0091] Obviously, the accompanying drawings are merely examples or embodiments of the present application. A person skilled in the art can also apply the present application to other similar situations based on these drawings without inventive effort. Furthermore, it is understandable that, although the work involved in this development process may be complex and lengthy, certain design, manufacturing, or production changes based on the technical content disclosed in this application are merely routine technical means for a person skilled in the art and should not be considered to constitute a deficiency in the disclosure of the present application.

Claims

1. A multi-crane collaborative operation simulation platform, characterized in that: include: A server that houses a digital twin model of the crane, a digital twin model of the environment, and a regional collaborative control module; Several simulation operation chambers, each of which has a built-in VR device, a fault simulation module and an operation terminal. The fault simulation module is used to collect and record the user's operation stability data and control accuracy data after the environmental control system changes the environment in the simulation operation chamber; the fault simulation module collects and records operation stability data including hook swing amplitude, equipment shaking times, misoperation rate, fault response time, and operation interruption times, and control accuracy data including positioning error of the hoisted object or hook, hook movement trajectory deviation, crane speed control error, crane and hook movement continuity, and operation error recovery speed for weighting the examinee's final total score. The VR device has a built-in physical simulation engine. Several operation terminals transmit the crane control data generated after the user's operation to their respective physical simulation engines. The physical simulation engine controls the crane's digital twin model through the crane control data to execute actions and generate the crane's real-time operation status. The server obtains the real-time operation status of the crane corresponding to each operation terminal and transmits it to the corresponding VR device for display. Among them, the regionalized collaborative control module has a built-in regionalized control mechanism, allowing multiple users to simultaneously control different cranes to perform collaborative tasks in a shared virtual scene. The regionalized collaborative control module is used to collect and record the real-time operating status of cranes in all simulated operating compartments, divide the operating areas of each crane digital twin model, predict the future motion trajectory of the crane digital twin model based on the crane control data and the real-time operating status of the crane, and detect the operating path of the crane digital twin model in real time. Based on the operating path and the real-time operating status of the crane, it predicts whether interference occurs between the digital twin models and provides feedback to all users.

2. The multi-crane collaborative operation simulation platform according to claim 1, characterized in that: Crane control data includes joystick movements and button movements. Joystick movements include joystick movement direction, movement amplitude, and movement duration. Button actions include the pressing and releasing time of the button corresponding to the function; The real-time operating status of the crane includes: the speed and acceleration of the crane; the speed of the crane's rotation; the acceleration change of the crane; the angle change of the crane's boom; and the three-dimensional spatial coordinates of the crane's spreader.

3. The multi-crane collaborative operation simulation platform according to claim 1, characterized in that: The simulated operation chamber includes: a temperature control simulation unit that can precisely control the temperature range; a high-altitude shaking simulation unit installed on a six-degree-of-freedom vibration platform to simulate various shaking modes; a wind disturbance simulation unit equipped with an array of adjustable-speed blowers to simulate different wind field conditions; a visibility and light environment simulation unit with an integrated atomizer and special lighting adjustment device; and a multi-channel high-fidelity speaker array for playing real on-site noise.

4. A multi-crane collaborative operation examination system, characterized in that: include: A multi-crane collaborative operation simulation platform, which is the multi-crane collaborative operation simulation platform according to any one of claims 1 to 3; The digital twin model of the crane in the server is used to construct a virtual crane model that is consistent with the physical characteristics and dynamic performance of the actual crane. The digital twin model of the environment in the server is used to construct an interactive virtual operation scene. The digital twin model of the crane and the physical simulation engine built into the VR device jointly achieve real-time two-way synchronization between the virtual crane model and the simulated device status. The scoring and evaluation system has a built-in artificial intelligence-based real-time recognition and adaptive evaluation module for operational intentions. This evaluation module introduces emergencies into virtual exams, collects the examinee's multi-dimensional operational data, equipment status, and environmental feedback, and uses the Transformer model to analyze and identify the examinee's immediate operational intentions, and obtains dynamic scoring data accordingly. The scoring and evaluation system also has a built-in scoring unit, which has a preset weighted scoring model that includes time efficiency, operational accuracy, the operational ability score output by the operational intention recognition module, a safety energy efficiency score that takes into account both safety and efficiency, and an emergency response score that reflects responsiveness. The weighted scoring model calculates the examinee's final total score based on crane control data, real-time crane operating status, operational stability data, control accuracy data, and dynamic scoring data.

5. The multi-crane collaborative operation examination system according to claim 4, characterized in that: The regionalized collaborative control module within the server also has a built-in rotational learning function. The rotational learning function collects abnormal event data such as trajectory interference and load imbalance that occur in collaborative tasks, analyzes high-frequency collaborative problem patterns and risk scenarios, builds a problem pattern knowledge base, and generates targeted new test task templates based on this problem pattern knowledge base.

6. The multi-crane collaborative operation examination system according to claim 4, characterized in that: The AI-based real-time recognition and adaptive assessment module for operational intentions includes the following processing steps: the emergency event trigger module dynamically inserts equipment failure, environmental mutation, or task change events; the real-time data acquisition and feature extraction module collects the examinee's control instructions, gaze tracking data, and equipment and environmental status data, and generates a normalized feature vector; the Transformer model receives the feature vector, uses a multi-head self-attention mechanism to fuse and analyze multimodal features, and outputs the operational intention classification result and operational ability score Oc; In the scoring unit, the safety energy efficiency score Sc is calculated by the safety efficiency dual evaluation algorithm module, which evaluates the safety indicators of the operation respectively. and performance indicators , and fused through dynamically adjusted weight α: ； Emergency response score Calculated by the rhythm adaptation and dynamic evaluation algorithm module, the rhythm adaptation and dynamic evaluation algorithm module is based on the candidate's actual reaction time Rt to the task or operation and the preset maximum allowable response time threshold Rmax: ； And classify the candidates' performance accordingly; The final total score is calculated as follows: ; Among them, Tf is the time taken by the examinee, Tmax is the maximum allowed time, Er is the error rate, and w1 to w5 are the weights of each scoring item.

7. The multi-crane collaborative operation examination system according to claim 4, characterized in that: The multi-crane collaborative operation examination system also includes a VR examination terminal, which integrates the functions of candidate identity recognition and automatic assignment of examination tasks.

8. The multi-crane collaborative operation examination system according to claim 4, characterized in that: The fault simulation module built into the simulation operation cabin can also preset or randomly generate a variety of fault scenarios including equipment operation failure, structural abnormality failure, control system delay failure and environmental mutation, and monitor the examinee's handling operation path, reaction time and response effectiveness in real time.

Images