A smart accident simulation method and product for laboratory use

Abstract

This application provides an intelligent accident simulation method and product for laboratories. The intelligent accident simulation method includes: constructing a digital twin operating platform corresponding to a target laboratory; adjusting the digital twin operating platform by preset environmental variable values; collecting environmental state signals and accident trigger signals in parallel from the digital twin operating platform; obtaining a target decision signal based on the environmental state signals and the accident trigger signals; and if the target decision signal indicates a trigger, driving the digital twin operating platform to simulate and visualize the abnormal accident through a real-time event simulation trigger. Some embodiments of this application can improve the simulation training effect for abnormal accidents through a digital twin environment, significantly improving training effectiveness while saving training costs.

Description

Technical Field

[0001] This application relates to the field of training, and more specifically, the embodiments of this application relate to an intelligent accident simulation method and product for use in laboratories. Background Technology

[0002] Closed laboratories are essential infrastructure for cutting-edge scientific research, disease control, and national security. These laboratories operate with high-risk substances (e.g., highly pathogenic pathogens, hazardous chemicals, or radioactive sources), and their safe operation is crucial to personnel safety, environmental safety, and even public safety. For example, biosafety laboratories (e.g., P4 laboratories) are critical facilities for researching highly pathogenic pathogens, and their safe operation is paramount. These laboratories rely on a complex array of environmental control systems (e.g., negative pressure gradients or directional airflow), equipment systems (e.g., biosafety cabinets, autoclaves, or positive pressure protective suits), and stringent management procedures (e.g., personnel access control and sample traceability).

[0003] Currently, traditional on-site training or tabletop exercises are insufficient to simulate accident scenarios in such laboratories, and hands-on training in high-risk environments is costly and carries safety risks.

[0004] Therefore, how to improve the technical effectiveness of simulation training for abnormal accidents and reduce training costs and risks has become an urgent technical problem to be solved. Summary of the Invention

[0005] The purpose of this application is to provide an intelligent accident simulation method and product for laboratories. Some embodiments of this application construct a digital twin of a physical laboratory, integrate environmental, equipment, personnel and sample data in real time, trigger the digital twin environment to simulate abnormal accidents based on preset environmental variable values, preset abnormal accident triggering thresholds and collected information of the digital twin environment, and provide users with abnormal accident handling guidance strategies during abnormal accident simulation, thereby improving the simulation training effect of abnormal accidents through the digital twin environment, saving training costs and significantly improving the user's training and learning effect.

[0006] In a first aspect, embodiments of this application provide an intelligent accident simulation method for a laboratory. The intelligent accident simulation method includes: constructing a digital twin operating platform corresponding to a target laboratory; adjusting the digital twin operating platform by preset environmental variable values, wherein the preset environmental variable values ​​are adapted to training objectives and are used to guide the digital twin operating platform to malfunction; collecting environmental state signals and accident triggering signals in parallel from the digital twin operating platform, wherein the accident triggering signal is generated by the digital twin operating platform being triggered according to a preset abnormal environmental variable threshold; obtaining a target decision signal based on the environmental state signals and the accident triggering signal, wherein the target decision signal is used to characterize whether an abnormal accident is triggered on the digital twin platform; if the target decision signal indicates triggering, driving the digital twin operating platform to simulate and visualize the abnormal accident through a real-time event simulation trigger.

[0007] In some embodiments, obtaining the target decision signal based on the environmental state signal and the accident triggering signal includes: inputting the environmental state signal into a twin environment perception module to extract environmental semantic features; inputting the environmental semantic features into a safety knowledge graph model to obtain an anomaly diagnosis result based on preset rules; inputting the environmental semantic features and behavioral data from the digital twin operation platform into a biosafety behavior prediction model to obtain a risk assessment result based on behavior and environment interaction analysis; and generating the target decision signal based on the accident triggering signal, the anomaly diagnosis result, and the risk assessment result.

[0008] The embodiments of this application determine whether to trigger an abnormal incident simulation in the digital twin environment by using preset abnormal environmental variable thresholds and information collected from the digital twin environment. When it is confirmed that an abnormal incident simulation is to be carried out, the user is provided with processing knowledge related to the abnormal incident for learning or operation practice. Abnormal incidents and handling strategies can be simulated without causing any damage to the physical laboratory, thereby improving trainees' operational proficiency in handling abnormal incidents and enhancing training effectiveness.

[0009] In some embodiments, generating the target decision signal based on the accident triggering signal, the anomaly diagnosis result, and the risk assessment result includes: inputting the accident triggering signal, the anomaly diagnosis result, and the risk assessment result into an accident situation prediction module to obtain the target decision signal, wherein the accident situation prediction module determines the target decision signal through a gating fusion network.

[0010] Some embodiments of this application use a gating fusion network to make a final decision on whether to trigger an abnormal incident based on multiple types of input information, thereby improving the adaptability of the decision results to the actual situation and maximizing the training effect.

[0011] In some embodiments, the step of inputting the accident triggering signal, the anomaly diagnosis result, and the risk assessment result into the accident situation prediction module to obtain the target decision signal includes: inputting the accident triggering signal, the anomaly diagnosis result, and the risk assessment result into the gating fusion network, so that the gating fusion network dynamically calculates the weight of each input information according to the current task context; performing a weighted decision based on the weights to obtain the target decision signal, wherein the target decision signal is used to carry whether the abnormal accident is triggered, the type of the abnormal accident, the time of triggering the abnormal accident, and the triggering intensity.

[0012] The gating fusion network of this application dynamically determines the weight coefficients of each input based on the current task context, and then obtains the target decision signal to determine whether to trigger an abnormal incident based on the weight coefficients. Since the differentiated context information pre-set for different users or trainees is considered in the final decision, the adaptability of the target decision signal to different trainees or users can be maximized, thereby improving the training effect for various types of trainees.

[0013] In some embodiments, the current task context includes: user role, training phase, and training objective.

[0014] The embodiments of this application, by pre-setting different user roles, training stages, and training objectives for different users or trainees, can better match different users and trainees when determining abnormal accident simulations, and achieve differentiated training objectives for different training goals.

[0015] In some embodiments, the intelligent accident simulation method further includes: providing an interactive interface to a user terminal; and, in response to information input to the interactive interface, providing abnormal handling knowledge corresponding to the abnormal accident or controlling a virtual device on the digital twin platform to perform operation practice; or, when an alarm is triggered, pushing a processing video, graphic operation manual, or step-by-step checklist corresponding to the type of abnormal accident to the user terminal.

[0016] Some embodiments of this application also receive input information from users or trainees through an interactive interface, so as to provide users or trainees with corresponding abnormal incident handling knowledge based on this information, or to manipulate objects in the digital twin environment to simulate the abnormal incident handling process based on this information, thereby improving the learning effect of abnormal incident handling related knowledge or operations.

[0017] In some embodiments, the intelligent accident simulation method further includes: recording the user's learning process of the anomaly handling knowledge and evaluating the user's anomaly handling knowledge structure.

[0018] The embodiments of this application record the learning process of users or trainees in real time and evaluate the learning process, which facilitates the arrangement and planning of subsequent learning tasks and improves the final training effect.

[0019] In some embodiments, the types of abnormal accidents include at least one of the following: abnormal negative pressure gradient, filter blockage or damage, sample leakage or abnormal location, biosafety cabinet failure, autoclave malfunction, abnormal positive pressure protective clothing pressure, personnel access violation, and door opening timeout.

[0020] Secondly, some embodiments of this application provide an intelligent accident simulation device for a laboratory, the intelligent accident simulation device comprising: a digital twin platform construction module configured to construct a digital twin operation platform corresponding to a target laboratory; an adjustment module configured to adjust the digital twin operation platform by means of preset environmental variable values, wherein the preset environmental variable values ​​are adapted to training objectives and are used to guide the digital twin operation platform to malfunction; an acquisition module configured to acquire environmental state signals and accident trigger signals in parallel from the digital twin operation platform, wherein the accident trigger signal is generated by the digital twin operation platform being triggered according to a preset abnormal environmental variable threshold; a decision generation module configured to obtain a target decision signal based on the environmental state signals and the accident trigger signal, wherein the target decision signal is used to characterize whether an abnormal accident is triggered on the digital twin platform; and an abnormal accident triggering module configured to drive the digital twin operation platform to simulate and visualize an abnormal accident through a real-time event simulation trigger if the target decision signal indicates triggering.

[0021] Thirdly, some embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed, can implement the methods described in any of the embodiments included in the first aspect.

[0022] Fourthly, some embodiments of this application provide an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, can implement the method described in any of the embodiments included in the first aspect. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is an architecture diagram of an intelligent accident simulation system for biosafety laboratories provided in an embodiment of this application.

[0025] Figure 2 This is a flowchart of an intelligent accident simulation method for a biosafety laboratory, as described in an embodiment of this application.

[0026] Figure 3 This application provides a block diagram of the composition of an intelligent accident simulation device for biosafety laboratories.

[0027] Figure 4 This is a schematic diagram illustrating the composition of an electronic device provided in an embodiment of this application. Detailed Implementation

[0028] The technical solutions in the embodiments of this application will now be described with reference to the accompanying drawings.

[0029] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0030] Some embodiments of this application can set different training tasks and training objectives according to different users. The training content includes the abnormal event handling process of experimental operation, tutorials and related professional knowledge guidance. The abnormal accident simulation provided by the embodiments of this application is triggered by users or trainees through pre-set abnormal environmental variable thresholds and preset environmental variable values. The abnormal accident simulation can be visualized through the established laboratory digital twin environment, and the process of visualizing the abnormal accident can receive interactive information from users or trainees and provide corresponding abnormal accident handling knowledge, thereby enabling users or trainees to learn relevant knowledge through the abnormal event handling process and improving the training effect.

[0031] Please refer to Figure 1 , Figure 1 The present application provides an intelligent accident simulation system for a laboratory, which includes: a digital twin operation platform 110, a training content setting module 120, an automatic training environment adjustment module 130, a real-time monitoring system 140, a real-time event simulation triggering module 150, a twin environment perception module 160, a real-time event state prediction module 170, a safety knowledge graph 180, and a biosafety behavior model 190.

[0032] The training content setting module is used to set the training objectives and content, while the training environment automatic adjustment module is used to generate preset environmental variable values ​​based on the information set in the training content setting module, and guide the accident triggering process of the digital twin operation platform based on these preset environmental variable values.

[0033] For example, in some embodiments of this application, the training content setting module is used to define the identity of the trainees participating in the training (e.g., new lab technicians, safety administrators, or maintenance engineers); to set specific training task lists for different roles, including: routine procedure training (e.g., donning and doffing positive pressure protective clothing or standard operation of biosafety cabinets) and special emergency response training (e.g., specific handling procedures for abnormal accidents such as sample leakage or loss of negative pressure); and to clarify the knowledge, skills, and behavioral goals to be achieved for each task, and to preset scoring rules (e.g., operation step points or response time points).

[0034] For example, in some embodiments of this application, the training environment automatic adjustment module is used to set preset environmental variable values. These preset environmental variable values ​​can be adjusted according to the training objectives provided by the training content setting module. The preset environmental variable values ​​are used to set the initial state or evolution logic of key parameters in the digital twin environment. In some embodiments of this application, the training environment automatic adjustment module is also used to deploy soft triggers in the virtual environment. For example, setting a virtual opening time threshold for a door or setting a decay curve for the efficiency of a virtual fan. In some embodiments of this application, the training environment automatic adjustment module is also used to control certain variables to automatically evolve into abnormal states or combinations over time or through user operations after training begins, providing conditions for accident simulation.

[0035] The real-time monitoring system 140 receives an accident trigger signal from the digital twin platform (the accident trigger signal is determined by comparing the real-time collected data with the threshold of abnormal environmental variables), processes the received data, and outputs it to the real-time event state prediction module 170.

[0036] The twin environment perception module 160 is used to fuse environmental signals and user behavior data, extract features, and provide feature vectors to the security knowledge graph module. It is understood that this twin environment perception module is at least used to perform multi-source data fusion and feature extraction.

[0037] The security knowledge graph module performs logical reasoning based on rules and input feature vectors, and outputs structured diagnostic results, i.e., anomaly diagnostic results.

[0038] The biosafety behavior model is used to identify risks based on input feature vectors and virtual identity behavior data, and output risk assessment results.

[0039] The real-time event scenario prediction module integrates the accident triggering signal, the abnormal diagnosis results output by the safety knowledge graph module, and the risk assessment results output by the biosafety behavior model. It uses a gating network for dynamic weighting to finally obtain the target decision signal, which can be used to determine whether an abnormal accident has been triggered.

[0040] The real-time event simulation triggering module 150 is used to receive the decision of the real-time event scenario prediction module 170, call the accident scenario library, generate execution instructions, and send the execution instructions to the digital twin operation platform to simulate abnormal accidents.

[0041] Figure 1 The system's workflow includes:

[0042] The first step is to set different training tasks and objectives based on the different training users. The training content includes normal biosafety knowledge and abnormal event handling procedures for experimental operations, tutorials and related professional knowledge guidance. Accident simulation is used to automatically trigger some abnormal accidents according to the objectives during the learning process of users through the digital twin system. When users face abnormal accident events, the system can guide trainees or users to learn relevant abnormal accident handling knowledge or procedures through abnormal event handling procedures.

[0043] The second step is automatic adjustment of the training environment. This step sets the variables of the digital twin environment (i.e., sets preset environmental variable values) according to the training objectives. These environmental variables will evolve into an abnormal value or an abnormal combination of values ​​after a certain period of time, triggering an abnormal accident in the digital twin operation platform.

[0044] The third step involves the digital twin platform operating based on the preset environmental variable values ​​from the second step, and transmitting the real-time collected variable values ​​back to the real-time monitoring system.

[0045] The fourth step involves monitoring environmental data within the digital twin environment, manipulating audio or video data, and providing the data to the twin environment perception module. The fifth step involves inputting the collected monitoring data into the security knowledge graph module for abnormal event prediction and reasoning to obtain abnormal diagnosis results. Simultaneously, the data is sent to the biosafety behavior model for multimodal analysis to infer abnormal situation characteristics and obtain risk assessment results. The sixth step involves fusing the collected raw data, the results from the security knowledge graph, and the biosafety behavior model in the accident situation prediction module to determine whether the abnormal accident should be punished. The seventh step, if an abnormal accident is constituted, sends the abnormal accident signal to the real-time event simulation trigger to generate an abnormal accident signal. This module then sends the signal to the user terminal (e.g., a VR terminal or a PC terminal) through the digital twin system. The eighth step, after receiving the abnormal accident signal, the user can learn relevant abnormal handling knowledge through the security knowledge graph in a two-way interactive manner via VR or PC.

[0046] The seventh step involves the digital twin platform recording the user's learning process regarding handling abnormal incidents and assessing their knowledge structure in this area.

[0047] The following is combined with Figure 2 This application provides an exemplary embodiment of an intelligent accident simulation method for a laboratory, which can be operated on... Figure 1 In the system.

[0048] like Figure 2 As shown, embodiments of this application provide an intelligent accident simulation method for a laboratory, the intelligent accident simulation method comprising:

[0049] S210, construct a digital twin operation platform corresponding to the target laboratory.

[0050] S220, the digital twin operating platform is adjusted by preset environmental variable values, wherein the environmental variable values ​​are adapted to the training objectives, and the preset environmental variable values ​​are used to guide the digital twin operating platform to malfunction.

[0051] S230, environmental status signals and accident trigger signals are collected in parallel from the digital twin operation platform, wherein the accident trigger signal is generated by the digital twin operation platform based on a preset abnormal environmental variable threshold.

[0052] S240, a target decision signal is obtained based on the environmental state signal and the accident triggering signal, wherein the target decision signal is used to characterize whether an abnormal accident is triggered on the digital twin platform.

[0053] S250, if the target decision signal indicates a trigger (i.e., indicates a penalty for an abnormal incident), the digital twin operation platform is driven by a real-time event simulation trigger to simulate and visualize the abnormal incident.

[0054] The following example illustrates... Figure 2 The implementation process of the relevant steps.

[0055] In some embodiments of this application, S240 includes:

[0056] The first step is to input the environmental state signal into the twin environment perception module to extract environmental semantic features.

[0057] The second step is to input the environmental semantic features into the security knowledge graph model to obtain anomaly diagnosis results based on preset rules.

[0058] For example, in some embodiments of this application, the security knowledge graph module is a knowledge system based on graph structure for storing and reasoning security rules, where nodes represent laboratory entities (such as rooms, equipment, and personnel) and edges represent relationships between entities (such as being located, controlling, or belonging to) or security rule logic.

[0059] Exemplary operating processes of the security knowledge graph module in some embodiments of this application include:

[0060] The input to the safety knowledge graph module is features received from the twin environment perception module, such as {region: core area, indicator: pressure difference, value: -38Pa, trend: continuously rising}.

[0061] The rules set for the safety knowledge graph module include: if the pressure difference in the core area lasts for 5 seconds > -40Pa, the abnormal event is determined to be a negative pressure loss alarm, and the level is emergency.

[0062] The reasoning output of the safety knowledge graph module (as an example of anomaly diagnosis results): {Event ID: E001, Type: Loss of negative pressure, Level: Emergency, Confidence: 0.98, Suggested action: [Check the ventilation fan, confirm that the access control is closed]}.

[0063] The third step involves inputting the environmental semantic features and behavioral data from the digital twin operation platform into the security behavior prediction model to obtain risk assessment results based on behavior and environment interaction analysis.

[0064] The security behavior prediction model of some embodiments of this application is a multimodal behavior analysis model based on deep learning (such as Transformer, LSTM) to assess the risk of user operations under specific environmental situations.

[0065] Exemplary working process:

[0066] The first input to the safety behavior prediction model (i.e., the environmental context) is features from the twin environmental perception module, such as {environmental state: loss of negative pressure, critical equipment: BSC-01, status: low airflow}.

[0067] The second input to the safety behavior prediction model (i.e., behavioral data) is a sequence of behaviors from the digital twin operation platform, such as a user clicking a fan switch (turning it off) or a user attempting to open a contaminated area access control.

[0068] Fusion analysis and prediction of safety behavior prediction models: analyzing the patterns, temporal sequence, and correlation with environmental context of behavioral sequences. For example, identifying shutting down the blower when negative pressure is lost as a high-risk operation, or attempting to open a door when airflow is abnormal as a violation.

[0069] Output of the safety behavior prediction model (as an example of risk assessment results): {Overall risk score: 0.85, high-risk behavior sequence: [shut down the fan], inferred intent: may have attempted to leave but the operation sequence was incorrect, recommended intervention: immediately lock the access control and prompt the correct evacuation procedure}.

[0070] The fourth step is to generate the target decision signal based on the accident triggering signal, the anomaly diagnosis result, and the risk assessment result.

[0071] In some embodiments of this application, the fourth step includes: inputting the accident triggering signal, the anomaly diagnosis result, and the risk assessment result into the accident situation prediction module to obtain the target decision signal, wherein the accident situation prediction module determines the target decision signal through a gating fusion network. For example, inputting the accident triggering signal, the anomaly diagnosis result, and the risk assessment result into the accident situation prediction module to obtain the target decision signal includes: inputting the accident triggering signal, the anomaly diagnosis result, and the risk assessment result into the gating fusion network, so that the gating fusion network dynamically calculates the weight of each input information according to the current task context (e.g., the current task context includes: user role, training stage, and training objective); performing a weighted decision based on the weights to obtain the target decision signal, wherein the target decision signal is used to carry whether the abnormal accident is triggered, the type of the abnormal accident triggered, the time of triggering the abnormal accident, and the trigger intensity.

[0072] For example, in some embodiments of this application, the importance weights of the three inputs from the safety knowledge graph, the safety behavior prediction model, and the underlying monitoring signal (i.e., the accident trigger signal, which determines whether the collected value is greater than a preset environmental variable threshold) are automatically adjusted according to the real-time context of the current training (e.g., whether the user is a novice or an expert, whether the current training stage is the initial learning or the final assessment, and whether the goal of this training is to familiarize oneself with the process or to improve resilience).

[0073] Example:

[0074] For a novice learner, the system relies more on the explicit rules (high weight) of the security knowledge graph to ensure that they first master the standard procedures.

[0075] For a senior employee undergoing stress assessment, the system focuses more on the safety behavior prediction model's evaluation of their judgment under pressure (high weight), while appropriately reducing the response to simple rule triggers (reducing the weight corresponding to the safety knowledge graph module).

[0076] When an accident trigger signal indicates that an extremely dangerous physical condition has been simulated, the weight of that signal is instantly increased regardless of the user's identity, to ensure that a safe response is prioritized.

[0077] The real-time event prediction model in some embodiments of this application uses a gated fusion network to make a final decision on whether to trigger an abnormal event based on multiple types of input information, thereby improving the adaptability of the decision results to the actual situation, maximizing the training effect, and better achieving the differentiated training objectives for different users.

[0078] The gating fusion network of this application dynamically determines the weight coefficients of each input based on the current task context, and then obtains the target decision signal to determine whether to trigger an abnormal incident based on the weight coefficients. Since the differential context information pre-set for different users or trainees is combined in the final decision, the adaptability of the target decision signal to different trainees or users can be maximized, thereby improving the training effect for various types of trainees.

[0079] The embodiments of this application, by pre-setting different user roles, training stages, and training objectives for different users or trainees, can better match different users and trainees when determining abnormal accident simulations, and achieve differentiated training objectives for different training goals.

[0080] The embodiments of this application determine whether to trigger an abnormal incident simulation in the digital twin environment by using preset abnormal environmental variable thresholds and information collected from the digital twin environment. When it is confirmed that an abnormal incident simulation is to be carried out, the user is provided with processing knowledge related to the abnormal incident for learning or operation practice. Abnormal incidents and handling strategies can be simulated without causing any damage to the physical laboratory, thereby improving trainees' operational proficiency in handling abnormal incidents and enhancing training effectiveness.

[0081] To enhance trainees' ability to handle abnormal incidents, some embodiments of this application provide an intelligent incident simulation method that further includes: providing an interactive interface to a user terminal; and, in response to information input to the interactive interface, providing abnormal incident handling knowledge corresponding to the abnormal incident or controlling virtual devices on the digital twin platform to perform operational exercises (e.g., adjusting fan speed to restore negative pressure, executing a two-person, two-lock door opening procedure, etc.); or, when an alarm is triggered, pushing a processing video, graphic operation manual, or step-by-step checklist corresponding to the type of abnormal incident to the user terminal (e.g., the eight-step method for replacing a HEPA filter).

[0082] Some embodiments of this application also receive input information from users or trainees through an interactive interface, so as to provide users or trainees with corresponding abnormal incident handling knowledge based on this information, or to manipulate objects in the digital twin environment to simulate the abnormal incident handling process based on this information, thereby improving the learning effect of abnormal incident handling related knowledge or operations.

[0083] In order to evaluate the learning effect of trainees in a timely manner, in some embodiments of this application, the intelligent accident simulation method further includes: recording the user's learning process of the anomaly handling knowledge and evaluating the user's anomaly handling knowledge structure.

[0084] In order to evaluate the learning effect of trainees or users, the embodiments of this application also record the learning process of users or trainees in real time and evaluate the learning process, so as to facilitate the arrangement and planning of subsequent learning tasks and improve the final training effect.

[0085] It should be noted that, in some embodiments of this application, the types of abnormal accidents include at least one of the following: abnormal negative pressure gradient, filter blockage or damage, sample leakage or abnormal location, biosafety cabinet failure, autoclave malfunction, abnormal positive pressure protective clothing pressure, violation of personnel access rights, and door opening timeout.

[0086] The following uses a high-level biosafety laboratory as an example to illustrate the intelligent accident simulation method for laboratories provided in this application.

[0087] The intelligent accident simulation method for laboratories provided in the embodiments of this application can be used to simulate abnormal accidents in high-level biosafety laboratories. This method constructs a digital twin of the physical laboratory, integrates environmental, equipment, personnel, and sample data in real time, dynamically triggers simulations of various accident scenarios based on preset rules, and visualizes the evolution process in a three-dimensional environment. The system integrates a large model to provide intelligent voice Q&A and handling guidance, and supports collaborative monitoring, operation, and training assessment on VR terminals, PC terminals, or multiple backend terminals. Some embodiments of this application can upgrade BSL-4 laboratories from passive monitoring to proactive simulation and early warning, from experience-based handling to intelligent decision-making, and from single training to multi-terminal collaborative drills, significantly improving the laboratory's safety management and emergency response capabilities.

[0088] Some embodiments of this application provide intelligent accident simulation methods for laboratories, including:

[0089] Step 1: Multi-source heterogeneous data acquisition and twin construction

[0090] 1. Data Acquisition: By integrating with the existing control system in the laboratory through an IoT gateway, data signals are collected in real time to simulate events and build a basic dataset.

[0091] 2. Environmental data: Pressure difference, temperature, humidity, and filter resistance in each area (clean area, semi-contaminated area, contaminated area) serve as event trigger signals for event simulation.

[0092] 3. Equipment data: negative pressure, airflow speed, and HEPA status of the biosafety cabinet; temperature, pressure, and time of the autoclave; internal pressure, oxygen supply, and battery life of the positive pressure protective suit; door magnetic status, etc., as event trigger signals for event simulation.

[0093] 4. Security data: Personnel access control permissions, real-time location, and sample RFID tag location information are used as event trigger signals for event simulation.

[0094] 5. Twin Construction: Based on the collected data, a 3D visual digital twin model corresponding to the physical laboratory at a 1:1 scale is constructed in a game engine (such as Unity3D or Unreal Engine). This model dynamically reflects the color status of each area (for example, using three different colors, red, yellow and blue, to represent three different states), the operating status of the equipment, and the real-time location of personnel and samples.

[0095] Step 2: Simulate and dynamically trigger abnormal events based on the knowledge graph module.

[0096] 1. The reasoning and rule base establishment process of the knowledge graph module includes pre-setting multi-level alarm rules and thresholds based on laboratory safety regulations.

[0097] Emergency Alarm (Red): Loss of negative pressure (core area > -40Pa), ventilation failure or protective clothing air supply failure are emergency alarms and require immediate triggering of audible and visual alarm linkage equipment (such as closing area air valves).

[0098] Important alarm (yellow): If the differential pressure deviates from the set value (±10Pa), the filter resistance rises to 1.5 times the initial value, or the temperature and humidity exceed the limits, the system will prompt for inspection and maintenance.

[0099] Warning / Reminder (Blue): Such as reminders when the equipment maintenance cycle is due or when disinfection is completed.

[0100] 2. Accident Scenario Simulation Library: Pre-sets various triggerable accident scenarios, including but not limited to the following:

[0101] Negative pressure backflow simulation: Simulates the dynamic process of air from the contaminated area flowing back into the clean area due to the buffer zone door opening time exceeding the limit (e.g., >30 seconds).

[0102] Sample leakage simulation: If a sample remains outside the biosafety cabinet for an extended period of time (e.g., >5 minutes), the system will automatically trigger area isolation and initiate the disinfection process.

[0103] Equipment failure simulation: Simulates airflow failure caused by damage to the HEPA filter of a biosafety cabinet (sudden drop in resistance ≤50Pa).

[0104] Personnel violation simulation: Simulates a person with low privileges attempting to enter a high-privilege area or a single person attempting to open a double-lock access control system.

[0105] 3. Dynamic Triggering: The system backend can trigger a single accident simulation semaphore or a batch of accident simulation semaphores as needed to output an accident signal and compare it with the rule base. If the accident triggering conditions are met, the semaphore and event code are sent to the user terminal (e.g., VR and PC), thus dynamically demonstrating the evolution process of the abnormal accident in the digital twin operation platform.

[0106] Step 3: Multimodal intelligent interaction and handling guidance.

[0107] 1. AI Voice Question Answering: By integrating a large language model, the system allows users to ask questions via voice or text, such as: What is the cause of the current loss of negative pressure in the core area? or What is the standard procedure for handling sample leakage? The system generates answers based on the current context and knowledge base.

[0108] 2. Visualized handling guidance: When an alarm is triggered, the system automatically pushes the corresponding handling video, graphic operation manual or step-by-step checklist (such as the eight-step method for replacing HEPA filter) to the VR and PC terminals, allowing users to learn professional knowledge of abnormal accident handling in an immersive digital twin environment.

[0109] 3. Virtual Operation Simulation: On the VR / PC platform, users can practice operating virtual devices, such as adjusting the fan speed to restore negative pressure or executing a two-person, two-lock door opening procedure. The system will judge the correctness of the operation and provide feedback.

[0110] Step 4: A multi-terminal collaborative training and assessment system.

[0111] 1. Multi-terminal synchronization: Supports real-time data synchronization and interaction across four terminals: large screen (global situation monitoring), VR (immersive operation and handling), PC (refined management and configuration), and backend (data management and simulation configuration).

[0112] 2. Role-playing and task list:

[0113] Individual assessment: Set a task list, which contains one or more event simulation lists, such as completing a standard autoclave operating system operation, and automatically score based on the degree of operational standardization and completion.

[0114] Team / Dual-person Collaboration Assessment: Collaborative tasks are set up, such as multiple people taking on the roles of control room administrator and lab operator, jointly handling a sample leak emergency. The system evaluates team communication, process execution, and collaboration efficiency.

[0115] 3. Training Debriefing: The system records the entire training process and supports post-training playback and debriefing, which can be used to analyze and improve trainees' professional learning plans.

[0116] The system corresponding to the above method may also include the following modules:

[0117] 1. Data Acquisition and Interface Module: Responsible for communicating with sensors, controllers, and business systems (such as access control and sample tracking) in the physics laboratory to acquire sample data from the environment and equipment.

[0118] 2. Digital Twin Engine and Visualization Module: Based on a 3D engine, a virtual laboratory is built and driven to achieve real-time, dynamic, and visual presentation of the environment, equipment, personnel, and sample status.

[0119] 3. Accident Simulation and Alarm Management Module: It has a built-in rule engine and accident scenario library, and is responsible for logical judgment, accident simulation triggering, alarm classification and multi-terminal notification.

[0120] 4. Intelligent Interaction and Guidance Module: Integrates a large model and multimedia knowledge base, providing voice / text Q&A, handling guidance push, and virtual operation guidance functions.

[0121] 5. Training Assessment and Management Module: Provides role management, task configuration, process editing, automatic scoring and exercise review functions, and supports one-person and multi-terminal collaborative training.

[0122] 6. System Management and Configuration Module: Responsible for the background configuration and management of user permissions, device models, alarm rules, system parameters, etc.

[0123] It should be noted that some embodiments of this application Figure 1 The system can adopt a cloud-based collaborative architecture, deploying a digital twin engine, core accident simulation logic, large model services, and database on cloud or local servers. Users can access the system through a browser, VR headset, or PC client.

[0124] The accident simulation implementation process in some embodiments of this application includes: creating corresponding agents for key entities (such as doors, fans, and samples) in the system. For example, when the door agent detects that the opening time has exceeded the limit, the agent sends an event to the environmental agent. The environmental agent calculates the pressure difference change based on the airflow dynamics model and triggers the alarm agent to issue a notification. This process is visualized in real time in the twin.

[0125] Intelligent guidance is implemented by structuring the laboratory's SOPs (Standard Operating Procedures) and storing them in a knowledge base. When a specific alarm is triggered, the system uses RAG (Retrieval Enhanced Generation) technology to combine relevant SOP fragments with a large language model to generate context-sensitive, natural language handling instructions.

[0126] Application Examples:

[0127] 1. Example 1: Simulation of Emergency Response to Loss of Negative Pressure

[0128] Trigger: The system detects that the negative pressure in the core area has been above the -40Pa threshold for 5 seconds.

[0129] Twin response: The core area turns red and flashes in the 3D model. The large-screen alarm dashboard displays the loss of negative pressure in the core area.

[0130] Intelligent guidance: A video prompts for action on the VR device, accompanied by an AI voice prompt: Warning: Loss of negative pressure in the core area. Please immediately check the operating status of the A-3 supply fan and the resistance of the B-2 exhaust filter. Suggested actions: 1. Confirm all doors are closed; 2. Proceed to the fan control panel…

[0131] Training and assessment: In the training mode, trainees need to complete a series of virtual operations in VR according to the instructions. The system records the response time and the accuracy of the operation and scores them.

[0132] 2. Example 2: Assessment of Two-Person Collaborative Entry into Contaminated Area

[0133] Task assignment: Two trainees will play the roles of main operator and supervisor in VR, respectively, and their task is to enter the contaminated area to collect samples.

[0134] Process execution: Two people need to virtually collaborate to complete the following process: permission application, multi-level approval, swiping cards simultaneously in front of the virtual access control (double lock for both people), and checking the interlocking door status after entering.

[0135] System assessment: The system monitors the entire process. If the supervisor fails to fulfill their review responsibilities or the main operator attempts to open the door alone, points will be deducted and an alarm will be triggered, emphasizing the BSL-4 safety standards.

[0136] like Figure 3 As shown, some embodiments of this application provide an intelligent accident simulation device for a laboratory. It should be understood that this device is similar to the one described above. Figure 2 Corresponding to the method embodiments, it can execute the various steps involved in the above method embodiments. The specific functions of the device can be found in the description above; to avoid repetition, detailed descriptions are appropriately omitted here. The device includes at least one software function module that can be stored in a memory or embedded in the device's operating system in the form of software or firmware. This intelligent accident simulation device for laboratories includes:

[0137] The digital twin platform building module 310 is configured to build a digital twin operating platform corresponding to the target laboratory.

[0138] The adjustment module 320 is configured to adjust the digital twin operating platform by means of preset environmental variable values, wherein the preset environmental variable values ​​are determined by training objectives, and the preset environmental variable values ​​are used to guide the digital twin operating platform to malfunction, and the digital twin platform is constructed for a target laboratory.

[0139] The acquisition module 330 is configured to acquire environmental status signals and accident trigger signals in parallel from the digital twin operation platform, wherein the accident trigger signal is generated by the digital twin operation platform based on a preset abnormal environmental variable threshold.

[0140] The decision generation module 340 is configured to obtain a target decision signal based on the environmental state signal and the accident triggering signal, wherein the target decision signal is used to characterize whether an abnormal accident is triggered on the digital twin platform.

[0141] The abnormal incident triggering module 350 is configured to drive the digital twin operation platform to simulate and visualize abnormal incidents through a real-time event simulation trigger if the target decision signal indicates that an abnormal incident has occurred.

[0142] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the device described above can be referred to the corresponding process in the aforementioned method, and will not be elaborated further here.

[0143] Some embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed, can implement the method described in any of the embodiments of the intelligent accident simulation method for laboratory described above.

[0144] like Figure 4 As shown, some embodiments of this application provide an electronic device 400, which exemplarily includes a memory 410, a processor 420, and a computer program stored in the memory 410 and executable on the processor 420. When the processor 420 reads the program via a bus 430 and executes the computer program, it can implement the intelligent accident simulation method for laboratory use as described in the above embodiments.

[0145] Processor 420 can process digital signals and may include various computing architectures. For example, it may be a complex instruction set computer architecture, a reduced instruction set computer architecture, or an architecture that implements multiple instruction set combinations. In some examples, processor 420 may be a microprocessor.

[0146] Memory 410 can be used to store instructions executed by processor 420 or data related to the execution of instructions. These instructions and / or data may include code used to implement some or all of the functions of one or more modules described in the embodiments of this application. The processor 420 of the embodiments of this disclosure can be used to execute the instructions in memory 410 to implement… Figure 2 The method shown. Memory 410 includes dynamic random access memory, static random access memory, flash memory, optical memory, or other memory well known to those skilled in the art.

[0147] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0148] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0149] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0150] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application. It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0151] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0152] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A method for intelligent accident simulation in a laboratory, characterized in that, The intelligent accident simulation method includes: Construct a digital twin operation platform corresponding to the target laboratory; The digital twin operating platform is adjusted by setting preset environmental variable values, wherein the preset environmental variable values ​​are adapted to the training objectives and are used to guide the digital twin operating platform to malfunction. The digital twin operation platform collects environmental status signals and accident trigger signals in parallel, wherein the accident trigger signal is generated by the digital twin operation platform based on a preset abnormal environmental variable threshold. A target decision signal is obtained based on the environmental state signal and the accident triggering signal, wherein the target decision signal is used to characterize whether an abnormal accident is triggered on the digital twin platform; If the target decision signal is triggered, the digital twin operation platform is driven by a real-time event simulation trigger to simulate and visualize the abnormal event.

2. The intelligent accident simulation method as described in claim 1, characterized in that, The step of obtaining the target decision signal based on the environmental state signal and the accident triggering signal includes: The environmental state signal is input into the twin environment perception module to extract environmental semantic features; The environmental semantic features are input into the security knowledge graph model to obtain anomaly diagnosis results based on preset rules; The environmental semantic features and behavioral data from the digital twin operation platform are input into the biosafety behavior prediction model to obtain risk assessment results based on behavior and environment interaction analysis; The target decision signal is generated based on the accident triggering signal, the anomaly diagnosis result, and the risk assessment result.

3. The intelligent accident simulation method as described in claim 2, characterized in that, The step of generating the target decision signal based on the accident triggering signal, the anomaly diagnosis result, and the risk assessment result includes: The accident trigger signal, the anomaly diagnosis result, and the risk assessment result are input into the accident situation prediction module to obtain the target decision signal, wherein the accident situation prediction module determines the target decision signal through a gating fusion network.

4. The intelligent accident simulation method as described in claim 3, characterized in that, The step of inputting the accident triggering signal, the anomaly diagnosis result, and the risk assessment result into the accident situation prediction module to obtain the target decision signal includes: The accident trigger signal, the anomaly diagnosis result, and the risk assessment result are input into the gating fusion network so that the gating fusion network dynamically calculates the weight of each input information according to the current task context, wherein the current task context includes: user role, training stage, and training objective; The target decision signal is obtained by performing a weighted decision based on the weights, wherein the target decision signal is used to carry whether the abnormal event is triggered, the type of the abnormal event, the time of triggering the abnormal event, and the triggering intensity.

5. The intelligent accident simulation method as described in any one of claims 2-4, characterized in that, The intelligent accident simulation method also includes: Provide an interactive interface to the user terminal; and, in response to information input to the interactive interface, provide anomaly handling knowledge corresponding to the abnormal incident or control the virtual device in the digital twin platform for operational practice; or When an alarm is triggered, a processing video, graphic operation manual, or step-by-step checklist corresponding to the type of abnormal incident is pushed to the user terminal.

6. The intelligent accident simulation method as described in claim 5, characterized in that, The intelligent accident simulation method also includes: Record the user's learning process of the exception handling knowledge and evaluate the user's exception handling knowledge structure.

7. The intelligent accident simulation method as described in any one of claims 1-6, characterized in that, The types of abnormal incidents include at least one of the following: abnormal negative pressure gradient, filter blockage or damage, sample leakage or abnormal location, biosafety cabinet failure, autoclave malfunction, abnormal positive pressure protective clothing pressure, violation of personnel access rights, and door opening timeout.

8. An intelligent accident simulation device for laboratory use, characterized in that, The intelligent accident simulation device includes: The digital twin platform construction module is configured to build a digital twin operation platform corresponding to the target laboratory; The adjustment module is configured to adjust the digital twin operating platform by means of preset environmental variable values, wherein the preset environmental variable values ​​are adapted to the training objectives and are used to guide the digital twin operating platform to malfunction. The acquisition module is configured to acquire environmental status signals and accident trigger signals in parallel from the digital twin operation platform, wherein the accident trigger signal is generated by the digital twin operation platform based on a preset abnormal environmental variable threshold. The decision generation module is configured to obtain a target decision signal based on the environmental state signal and the accident triggering signal, wherein the target decision signal is used to characterize whether an abnormal accident is triggered on the digital twin platform; The abnormal incident triggering module is fast and is configured to drive the digital twin operation platform to simulate and visualize abnormal incidents if the target decision signal indicates that an abnormal incident has occurred.

9. A computer-readable storage medium having a computer program stored thereon, said computer program being executed to perform the method as described in any one of claims 1-7.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein, When the processor executes the program, it can implement the method as described in any one of claims 1-7.

Images