A ship emergency exercise virtual simulation interaction method, device, equipment and medium

Abstract

The application provides a ship emergency exercise virtual simulation interaction method, device, equipment and medium, solves the technical problems of insufficient environment authenticity, low cooperation efficiency, insufficient immersion, single evaluation and poor system expansibility in traditional exercises through three-dimensional virtual environment construction, multi-person cooperative interaction, dynamic scene generation and intelligent team evaluation technology. The method is suitable for crew team training, maritime rescue cooperative exercise and ship safety verification, and can significantly improve the team emergency response capability and training efficiency.

Description

Technical Field

[0001] This invention belongs to the field of ship emergency drill simulation technology, specifically relating to a virtual simulation interaction method, device, equipment, and medium for ship emergency drills. Background Technology

[0002] Emergency drills for ships are a core component of ensuring maritime operational safety. However, traditional field drills are limited by high costs, significant safety risks, and difficulties in scenario reproduction, making it difficult to meet the demands for high-frequency, multi-scenario training. With the advancement of virtual reality (VR) and simulation technologies, emergency drill systems based on three-dimensional virtual environments have become a research hotspot in the industry. These systems simulate emergency scenarios through digital means, effectively lowering the training threshold. However, existing technologies still have key shortcomings: at the collaborative interaction level, most systems only support single-user operation or simple network synchronization, lacking refined operation permission management, conflict resolution mechanisms, and cross-platform compatibility, making it difficult to support multi-role collaboration in complex emergency scenarios; in terms of task flow control, traditional systems rely on preset scripts to drive drills, unable to dynamically adjust step priorities or handle abnormal states based on real-time situations, resulting in insufficient realism and flexibility in drills.

[0003] Furthermore, existing technologies have significant shortcomings in terms of intelligence: First, situation analysis largely relies on static rules or simple threshold judgments, lacking dynamic prediction capabilities based on multi-source data fusion, making it difficult to support real-time decision-making in complex emergency scenarios; second, team evaluation relies on manual scoring or basic indicator statistics, lacking machine learning-driven objective evaluation models and personalized feedback strategies, and cannot be deeply integrated with the exercise process to achieve dynamic difficulty adjustment; third, emergency scenario generation relies on manual configuration, case simulation flexibility is poor, and debriefing functions are mostly limited to operation playback, lacking data-driven in-depth analysis and automated case generation mechanisms. To address these issues, this invention proposes a virtual simulation interactive method for ship emergency drills. By integrating the Photon Server collaborative framework, situation analysis, dynamic task flow control, and intelligent evaluation system, it achieves a highly realistic, highly interactive, and intelligent ship emergency training solution. Summary of the Invention

[0004] The purpose of this invention is to provide a virtual simulation interaction method, device, equipment, and medium for ship emergency drills, aiming to solve the problems of low environmental simulation, poor multi-role collaboration efficiency, reliance on experience for situation assessment, and strong subjectivity in team evaluation in traditional ship emergency drills. By using behavior tree technology, the invention achieves intelligent adaptive adjustment of the drill process, significantly improving the emergency decision-making ability of crew members in complex sea conditions and the efficiency of cross-platform collaborative training.

[0005] To achieve the above objectives, the technical solution of the present invention is as follows:

[0006] On the one hand, a multi-person collaborative simulation interaction method for ship emergency safety drills in a three-dimensional virtual environment is provided, including the following steps:

[0007] Based on the ship production design model, a three-dimensional virtual ship emergency drill environment is constructed, integrating ship structure, pipeline system and equipment layout, and simulating flame spread, water flow and structural deformation effects through a dynamic physics engine, supporting multi-scale perspective switching and visualization rendering.

[0008] A multi-user collaborative interactive simulation interface is set up for the three-dimensional virtual ship emergency drill environment;

[0009] An emergency task process framework is constructed based on behavior trees, and a behavior tree task process model is constructed through reinforcement learning and anomaly detection mechanisms. The behavior tree task process model is used to display the simulation process of emergency drills using a visual flowchart.

[0010] Construct an emergency situation analysis model to analyze the evolution of the accident in real time and predict the future situation. Use heat maps to display the intensity of the fire and visualize the best evacuation route.

[0011] It generates individual and team evaluation metrics based on machine learning models, provides real-time visual feedback, and enables collaborative optimization of individuals and teams;

[0012] In conjunction with the emergency task process steps, task points are set at corresponding positions in the three-dimensional virtual ship emergency drill environment. The emergency task process is integrated through the behavior tree task process model to generate differentiated emergency case simulation scenarios.

[0013] On the other hand, a virtual simulation interactive device for ship emergency drills is provided, the device comprising:

[0014] The 3D virtual ship emergency environment construction module is used to build a 3D virtual ship emergency drill environment based on the ship production design model. It integrates the ship structure, pipeline system and equipment layout, and simulates flame spread, water flow and structural deformation effects through a dynamic physics engine. It supports multi-scale perspective switching and visualization rendering.

[0015] A multi-user collaborative interactive simulation interface module is used to set up a multi-user collaborative interactive simulation interface for the three-dimensional virtual ship emergency drill environment.

[0016] The behavior tree task process construction module is used to build an emergency task process framework based on behavior trees, and to build a behavior tree task process model through reinforcement learning and anomaly detection mechanisms. The behavior tree task process model is used to display the simulation process of emergency drills using a visual flowchart.

[0017] The emergency situation analysis and prediction module is used to build emergency situation analysis models, analyze the evolution of accidents in real time and predict future situations, use heat maps to display fire intensity, and visualize the best evacuation routes.

[0018] The intelligent assessment and collaborative optimization module is used to generate individual and team assessment metrics based on machine learning models, provide real-time visual feedback, and perform collaborative optimization for individuals and teams.

[0019] The emergency case generation and simulation module is used to combine emergency task process steps, set task points at corresponding positions in the three-dimensional virtual ship emergency drill environment, integrate the emergency task process through the behavior tree task process model, and generate differentiated emergency case simulation scenarios.

[0020] This invention achieves efficient collaboration, dynamic response, and scientific evaluation of ship emergency drills by organically integrating behavior tree-driven intelligent task flow control, multi-person collaborative interaction mechanism, and real-time situation analysis system, significantly improving the realism of the drills and training effectiveness.

[0021] In another aspect, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the virtual simulation interaction method for ship emergency drills described in any of the preceding descriptions.

[0022] In another aspect, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the virtual simulation interaction method for ship emergency drills described in any of the preceding claims.

[0023] The beneficial effects of this invention are as follows:

[0024] This invention provides a highly realistic and intelligent training platform for multi-user collaborative ship emergency drills through a behavior tree-driven virtual simulation method, device, equipment, and medium. The system constructs a 3D virtual environment based on ship design data and simulates effects such as flames and water flow using a dynamic physics engine, enhancing the realism of the training. It supports multi-user collaborative operation, intelligently assigns permissions, and automatically resolves conflicts to ensure smooth collaborative drills. Task flows are automatically managed through behavior trees, with dynamic priority adjustments and automatic handling of abnormal situations, improving the flexibility of the drills. Real-time emergency situation analysis and visualization of fire and evacuation routes aid decision-making. A machine learning-driven evaluation system provides personalized feedback, analyzes team collaboration issues, and generates training reports. The system supports multi-scenario configuration and case reviews, facilitating the optimization of training programs and improving the team's emergency response capabilities. Attached Figure Description

[0025] Figure 1 is a flowchart of a virtual simulation interaction method for ship emergency drills according to the present invention;

[0026] Figure 2 is a flowchart illustrating the implementation process of constructing a three-dimensional virtual ship emergency drill environment according to the present invention;

[0027] Figure 3 is a flowchart of the implementation of the multi-person collaborative interaction simulation interface of the present invention;

[0028] Figure 4 is a flowchart illustrating the implementation of the behavior tree-driven task flow model of this invention.

[0029] Figure 5 is a flowchart illustrating the implementation of the emergency situation analysis model constructed according to the present invention;

[0030] Figure 6 is a flowchart of the implementation of the intelligent team evaluation and feedback system of the present invention;

[0031] Figure 7 is a flowchart illustrating the implementation process of generating emergency case simulation scenarios according to the present invention;

[0032] Figure 8 is a flowchart of the emergency accident analysis process of the present invention;

[0033] Figure 9 is a flowchart of the finite state machine for accident evolution according to the present invention;

[0034] Figure 10 is a flowchart of the conflict arbitration system of the present invention;

[0035] Figure 11 is a logic diagram of the virtual simulation interaction method for ship emergency drills according to an embodiment of the present invention;

[0036] Figure 12 is a flowchart of another virtual simulation interaction method for ship emergency drills according to an embodiment of the present invention;

[0037] Figure 13 is a structural block diagram of a virtual simulation interactive device for ship emergency drills according to an embodiment of the present invention;

[0038] Figure 14 is an internal structural diagram of a computer device according to an embodiment of the present invention. Detailed Implementation

[0039] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] Example 1

[0041] Embodiment 1 of this application provides a virtual simulation interaction method for ship emergency drills, as shown in Figure 1, including the following steps:

[0042] Step S1) Construct a three-dimensional virtual ship emergency drill environment

[0043] Based on the ship production design model, a highly realistic three-dimensional virtual emergency environment is constructed. The specific implementation process is as follows: Figure 2 As shown.

[0044] Step S11) Basic 3D Scene Construction

[0045] A basic 3D scene is constructed based on the ship production design model. The ship's configuration data is converted into a precise 3D mesh model using 3D modeling software. Level of Detail (LOD) technology is used to optimize the model at multiple levels, reducing model complexity while ensuring visual quality. Specifically, this includes: integrating macroscopic elements such as the ship's main structure and cabin layout; integrating key subsystems such as piping systems and electrical wiring; and arranging important facilities such as fire-fighting equipment and emergency devices.

[0046] Build an interactive basic scene that allows users to move freely and switch perspectives.

[0047] Step S12) Integrating a dynamic physics engine

[0048] The system integrates environmental sensor data to create a dynamic physics engine that simulates dynamic physical effects such as flame spread, water flow, and structural deformation.

[0049] Environmental sensor data includes temperature, smoke concentration, and oxygen concentration. Temperature sensor data drives flame intensity parameters, smoke detector data is fed back to the particle system, and oxygen concentration data is used to calculate the combustion rate, thereby driving the spread of the virtual flame. A multi-layered flame effect is achieved using VFX (Visual Effects) combined with shader programming. The flame diffusion path is simulated based on fluid dynamics algorithms, dynamic flame shapes are generated through noise textures, and the impact of thermal radiation on surrounding objects is calculated in real time.

[0050] A complete structural deformation simulation system was built using the Unity engine's damage plugin. By introducing a dynamic fracture determination mechanism based on material mechanical properties, and combining real-time physical calculations with preset typical fracture modes, accurate dynamic simulation of hull damage was achieved. Simultaneously, stress models based on material properties were established for key equipment such as valves and pipe fittings. By setting differentiated deformation parameters, the progressive damage effect from slight deformation to complete failure was visualized, fully simulating the structural damage evolution process from the equipment level to the hull level. This simulation system supports real-time physical calculations and can accurately reflect the structural response under different impact loads, providing realistic and reliable structural damage simulation effects for emergency drills.

[0051] Step S13) Construct an interactive ship equipment model

[0052] A library of interactive operation models for ship equipment has been constructed, with standardized operating logic and status feedback mechanisms established for each type of equipment, including:

[0053] 1. Valve-type equipment: Supports rotation, switching, and other operations, and is associated with the fluid system status;

[0054] 2. Firefighting equipment: Operable equipment such as fire extinguishers and fire hydrants, simulating actual usage effects;

[0055] 3. Electrical control box: Enables visualization and interactive control of the electrical system status;

[0056] 4. Emergency communication device: simulates shipboard broadcasting, intercom and other communication functions.

[0057] Step S14) Emergency Incident Evolution

[0058] An emergency incident triggering and evolution mechanism is set up. Based on the fault tree analysis (FTA) method, the triggering conditions of accidents such as fire, collision, and oil spill are defined, and the accident evolution path is constructed using a finite state machine (FSM).

[0059] First, a multi-level fault tree model is established, decomposing top-level accidents into quantifiable bottom-level events, which are then logically connected to form a complete causal network. A ScriptableObject data structure is used to store the probability parameters and association rules of each node. When a bottom-level event meets the triggering condition, the system calculates the probability of the accident occurring from the bottom up based on preset logical relationships.

[0060] As shown in Figure 8, this flowchart uses a tree structure to intuitively display the accident analysis process. The top shows the main accident types such as fire, collision, and oil spill, with branches branching down to various inducing factors. For example, a fire accident can be broken down into secondary nodes such as electrical short circuit and oil leakage. Electrical short circuit can be further subdivided into specific causes such as insulation aging and overload. Each node is labeled with its probability weight and linked using AND / OR logic. In the Unity editor, this analysis model monitors the equipment status and environmental parameters in the scene in real time. When signals such as a drop in insulation value or abnormal oil pressure are detected, the corresponding fault assessment process is immediately initiated.

[0061] The accident evolution process is dynamically simulated using a Finite State Machine (FSM), as shown in Figure 9. The state machine model includes key states such as warning, outbreak, and spread. The transition conditions between states include threshold judgments for physical parameters such as temperature and pressure, as well as consideration of the intervention effects of emergency operations. Visual configuration is achieved through PlayMaker, and complex business logic is implemented using C# code. When the accident enters a new development stage, the system automatically adjusts the physics engine parameters and particle effects, while simultaneously triggering the corresponding emergency task process, forming a complete "monitoring-assessment-response" closed loop. This dual modeling mechanism based on FTA and FSM ensures both the scientific nature of the accident triggering and the realism of the evolution process.

[0062] Step S15) Visualization and Rendering

[0063] It features a multi-scale visualization rendering system that integrates ray tracing technology to achieve highly realistic rendering and lifelike lighting effects; it supports multi-scale perspective switching from a global top-down view to detailed device operations; it optimizes GPU instantiation technology to improve rendering efficiency and achieve realistic rendering of physical effects such as smoke, flames, and fluids; and it supports VR device access to provide an immersive visual experience.

[0064] Step S2) Set up the multi-user collaborative interaction simulation interface

[0065] Based on Photon Server technology, a multi-user collaborative simulation framework is implemented, and operation collaboration interfaces are implemented for common interactive objects. The specific implementation process is shown in Figure 3.

[0066] Step S21) Distributed simulation architecture setup

[0067] A distributed simulation architecture is built based on Photon Server, deploying a master-slave server cluster. A dynamic load balancing algorithm automatically allocates user connections to the optimal node. Quadtree spatial partitioning technology divides the ship scene into multiple logical regions, each managed by an independent server node, enabling segmented scene processing. An innovative TCP / UDP hybrid transmission protocol is adopted, with critical commands transmitted using reliable TCP and real-time operation data using UDP with forward error correction, ensuring synchronization latency is kept within 150ms even with hundreds of concurrent users.

[0068] Step S22) Access Control System

[0069] A hierarchical management mechanism for operation permissions is implemented, defining three roles—Commander, Operator, and Observer—based on a role-based access control model. Commanders possess a global perspective and supreme control, allowing them to forcibly take over any device. Operators receive specific device operation permissions based on task assignments. Observers only have scene browsing and data display permissions. A dynamic token pool management mechanism enables intelligent allocation of device control rights, employing a time-sensitive and priority-based preemption algorithm to ensure timely execution of critical operations. The system monitors the permission status of each role in real time, immediately triggering a security interception mechanism when unauthorized operations are detected, and visually displaying permission change notifications in the 3D interface.

[0070] Step S23) State synchronization system

[0071] A real-time synchronization algorithm for device operation status is implemented, employing incremental status update technology to transmit changed data. By comparing the status differences before and after device operation in real time, only the changed data attributes are extracted and transmitted, significantly reducing network load. The system maintains a status change flag for each device and uses a circular buffer to store multiple frames of historical status, ensuring rapid recovery in case of data loss. Based on a priority queue management mechanism, the system dynamically adjusts the synchronization frequency according to the criticality of the devices, optimizing resource utilization while ensuring real-time performance.

[0072] To address network transmission reliability, a layered protection system is constructed by combining the UDP protocol and a forward error correction mechanism. At the transport layer, the UDP protocol ensures low latency, and a forward error correction strategy that generates a redundant check packet every five data packets effectively mitigates network packet loss of less than 20%. The system incorporates an intelligent compensation mechanism that automatically switches to full synchronization mode when continuous packet loss is detected, ensuring operational continuity in multi-user collaborative scenarios.

[0073] Step S24) Conflict Arbitration System

[0074] A conflict resolution and arbitration system is established, featuring a triple conflict resolution mechanism: timestamp priority, determining operation timing accurate to milliseconds; permission level verification, automatically granting priority to users with higher permissions; and operation compliance checks, filtering operations that violate security procedures. When a conflict occurs, the system dynamically allocates control of the conflicting device to users who meet the arbitration criteria in real time, ensuring that critical operations can continue. Simultaneously, in the 3D virtual scene, the system highlights the conflicting device and pushes a visual conflict notification interface to users who have not obtained operating rights. This interface displays the cause of the conflict and the arbitration result in detail, and provides interactive options for "operation rollback" and "re-application." The conflict arbitration system process is shown in Figure 10.

[0075] Step S25) Cross-platform communication protocol

[0076] Build a cross-platform network communication protocol, define a binary communication protocol based on Protobuf, and support data interoperability between multiple terminals.

[0077] Step S3) Construct a behavior tree-driven task flow model

[0078] An automated task flow-driven mechanism prototype model is built based on behavior trees, which can reasonably control the switching of steps. A visual flowchart is used to show the exercise rule deduction process of emergency drills. The specific implementation process is shown in Figure 4.

[0079] Step S31) Construct an emergency task process framework based on behavior tree, define main task nodes and sub-task nodes using a hierarchical behavior tree structure, and realize task status sharing through the blackboard system.

[0080] Step S32) Set up an emergency decision-making knowledge graph and construct a graph database containing knowledge such as International Maritime Organization regulations and ship emergency manuals. The graph database enables knowledge retrieval and reasoning.

[0081] Step S33) Set up a dynamic weight adjustment algorithm, which dynamically adjusts task priority weights based on reinforcement learning to achieve automatic insertion and execution of urgent tasks. By analyzing key indicators such as remaining task time and risk level, the algorithm automatically adjusts execution weights to ensure that critical tasks are handled first. The core of the algorithm uses an improved Q-learning algorithm to define the state space. It contains 20-dimensional task features, and the action space A contains 8 weight adjustment strategies.

[0082] ;

[0083] in For the remaining time of the task, For mission risk level, For resource utilization rate, For 4-dimensional environmental dynamic factors, For 5-dimensional team collaboration status, It is an 8-dimensional task execution feature. It represents 20 dimensions.

[0084] Action space A includes eight weight adjustment strategies for dynamically intervening in task execution priority. Different actions are triggered under different conditions, including: emergency order insertion, resource preemption, task splitting, collaborative acceleration, degraded execution, external support request, process reengineering, and contingency plan switching.

[0085] Taking into account time, risk, and resource factors, we define R(s,a) to represent the reward function. In state space s, after performing action a, the immediate reward value returned by the environment. The reward function is as follows:

[0086] ;

[0087] Where tanh represents the tanh function. This is an adjustable hyperparameter that controls the contribution ratio of each item. (The parameter is...) .

[0088] The output includes: task weight table, resource allocation instructions, collaborative operation guidance, and decision tracing log, which are used to guide the task execution engine and visualization system.

[0089] In actual operation, the algorithm is deeply integrated with the task execution engine. When an emergency is detected, the system quickly adjusts the weights and ensures the execution order through a global queue. The system provides a visual interface that displays the weight change curve in real time to assist command and decision-making. Through a continuous learning mechanism, the algorithm can automatically adapt to the changing needs of different training scenarios.

[0090] Step S34) Construct an abnormal state detection and recovery mechanism, employing an improved LSTM network integrating multi-scale feature extraction and temporal attention mechanisms to predict deviations during task execution. This network enhances its ability to capture locally dependent features by introducing convolutional modules to preprocess the input multi-dimensional temporal data (including 12 key indicators such as completion progress, resource consumption rate, and equipment status). Simultaneously, a temporal attention module is embedded in the network's hidden layers to dynamically focus on key time steps and anomaly-sensitive indicators, improving the sensitivity to early deviation signals.

[0091] The network receives real-time status data from the blackboard system in step S31. After processing through three hidden layers and attention weighting, it outputs the predicted trajectory of the status at multiple future time points. By comparing the predicted values ​​with the actual execution, the system calculates the deviation index and determines whether it exceeds the threshold. Once an anomaly is detected, a tiered response mechanism is activated: for routine deviations, it automatically rolls back to the most recent valid checkpoint; for complex anomalies, it generates a structured diagnostic report, pushes it to the command terminal via the message queue in step S35, and triggers manual intervention. During the recovery process, the system monitors the recovery trend of indicators in real time, adaptively adjusts the rollback depth and resource allocation, and works in conjunction with the dynamic weight adjustment algorithm in step S33 to proactively avoid potential cascading failures.

[0092] Step S35) Set up a multi-task parallel processing engine, use the Actor model to achieve concurrent task execution, and decouple task scheduling and execution through message queues.

[0093] The Actor model encapsulates each task in ship emergency drills as an independent Actor entity. For example, tasks such as firefighting, evacuation, and equipment operation each correspond to a dedicated Actor. Each Actor maintains its own task state machine and execution logic, achieving concurrent processing through lightweight threads, eliminating the need for manual thread synchronization management. The system dynamically creates Actor instances based on task type; for example, in a fire scenario, a "FireActor" is generated to simulate fire spread, while a "ValveActor" controls pipeline valve operations. Actors interact strictly through message passing, avoiding the risks associated with shared memory. Leveraging the Actor model's fault tolerance mechanism, when a single task crashes abnormally, the monitoring strategy automatically restarts the affected Actors, ensuring the continued operation of the overall system.

[0094] The system uses RabbitMQ as its core message middleware, establishing an asynchronous communication channel between the scheduler and execution units. The scheduler transforms task instructions into standardized messages and delivers them to the corresponding Actor's dedicated queue, without needing to know the specific execution node's location. The message header contains a task priority flag, and the queues are automatically sorted by weight, ensuring that urgent tasks are dispatched first. Execution units obtain instructions by subscribing to the queues and return results through callback queues after processing. This setup achieves physical separation between scheduling and execution, supporting dynamic scaling—when the load surges, execution nodes can be added horizontally to consume backlogged messages without modifying the scheduling logic.

[0095] Step S36) Set up an emergency plan automatic matching system, which uses case reasoning technology to match similar accident scenarios from the historical exercise database, automatically recommends the optimal plan and initializes the task process.

[0096] The system uses a feature extraction engine to perform multi-dimensional analysis of the current accident scenario, including key features such as accident type (fire, collision, etc.), environmental parameters (wind speed, ship position), and equipment status (valve opening, power load), and constructs a standardized feature vector. This vector is then matched with cases in the historical exercise database for similarity. After matching, the system uses an integrated analysis module to evaluate the handling effectiveness of similar cases, generating a recommended list of contingency plans based on comprehensive indicators such as response time, resource consumption, and completion rate. Once a contingency plan is selected, the system automatically parses the standard operating procedures in the plan, maps them to a sequence of behavior tree nodes defined in step S31, and generates an executable task flow through the initialization engine.

[0097] Step S4) Construct an emergency situation analysis model

[0098] Analyze the current emergency situation and update the development of the current scenario in real time.

[0099] Step S41) Set up a multi-source data fusion engine to integrate heterogeneous data such as device status data and user operation logs from the virtual environment. Use a stream processing framework to achieve real-time data aggregation and preprocessing. The specific implementation process is shown in Figure 5. Construct a spatiotemporal data matrix:

[0100] ;

[0101] The input parameters are wind speed, etc. v (m / s), wave level s (1-9) Cabin temperature T (°C), smoke concentration σ (mg / m 3 ), valve opening α (%), pipeline pressure p (kPa), power load P (kW), population density ρ (person / m) 2 ), incumbency β (%), sampling period τ is 5 seconds, data standardization processing adopts:

[0102] ;

[0103] in For the new standardized data values, The original data values, The mean, 1 represents the standard deviation.

[0104] Step S42) Construct an accident evolution prediction model using a hybrid architecture combining convolutional neural networks (CNN) and long short-term memory networks (LSTM). Input the current accident parameters and output the prediction results of future multi-dimensional situations. The model input layer receives preprocessed multivariate time series data. (n features, m time steps), where n is the number of features (e.g., temperature, pressure, wind speed), and m is the time step size. Processed using the following hybrid path:

[0105] Spatial feature path:

[0106] ;

[0107] in, The convolution kernel weight matrix is... For bias terms, Enter data for the current moment.

[0108] Temporal feature path:

[0109] ;

[0110] in, and These represent the cellular state and the hidden state, respectively. and The outputs are the forget gate and the input gate, respectively. and For the corresponding weights and biases.

[0111] Dynamic feature fusion:

[0112] ;

[0113] Among them, the weighting coefficient .

[0114] σ It is the Sigmoid activation function. , This is the attention weight matrix.

[0115] Situation prediction output:

[0116] Thermal distribution , indicating future time period Inner position The thermal intensity at that location.

[0117] Smoke diffusion: , representing the gradient of smoke concentration, is measured by the partial derivatives of smoke concentration in the x and y directions to determine the rate and trend of change of smoke in these directions.

[0118] Escape routes: This indicates that evacuation paths are generated based on minimizing the risk integral. Numbers are used to obtain path On which path is the minimum value obtained? Indicates the summation of risks. The final escape route is an ordered sequence.

[0119] Step S43) Set up a real-time situation visualization system, overlay a situation layer on the virtual scene, use a heat map to show the fire intensity, and use a path planning algorithm to show the best evacuation route.

[0120] Step S5) Set up an intelligent team assessment and feedback system

[0121] Based on machine learning models, objective team evaluation metrics are generated and visual feedback is provided. The specific implementation process is shown in Figure 6.

[0122] Step S51) Construct a multi-dimensional evaluation index system, define individual and team indicators, and use the analytic hierarchy process (AHP) to determine the weight of each indicator. Individual indicators include operational accuracy and response speed, while team indicators include communication density and role complementarity. Weight allocation is calculated using an expert scoring matrix; for example, technical ability accounts for 70% of the individual score, while communication efficiency dominates the team collaboration with a weight of 55%, ensuring that the indicators align with actual business needs. Indicator settings must meet three principles: observability, quantifiability, and non-overlapping.

[0123] Step S52) Set up a machine learning-driven evaluation model, train the evaluation model based on the algorithm, input operational data, and output individual ability scores and team collaboration index.

[0124] This model relies on a multimodal data fusion and hybrid algorithm architecture. Input data includes structured operation logs (task time, error rate), unstructured voice interactions (collaborative positivity extracted through sentiment analysis), and spatial behavior data (frequency of distance changes between members). By transforming the raw data through feature engineering into multidimensional feature vectors, a comprehensive score including sub-dimensions such as technical ability and adaptability is output to obtain individual ability scores. The team collaboration index calculates team cohesion and identifies bottleneck nodes by analyzing communication frequency and task dependencies among members. Model training employs a two-stage strategy: first, pre-training using historical labeled data, and then fine-tuning online using a reinforcement learning framework combined with expert feedback.

[0125] Step S53) Construct a visual feedback system, overlay an evaluation panel on the virtual scene, use charts to display indicators, and support data export and automatic generation of text summaries.

[0126] Step S54) Set up the team collaboration effectiveness analysis module, construct a network diagram of relationships between users based on the analysis method, identify key nodes and information silos, and provide suggestions for optimizing the collaboration mode.

[0127] Step S55) Set up a personalized assessment report generator, customize the report content according to the user role, and use a template for automatic formatting.

[0128] Step S6) Generate emergency case simulation scenarios

[0129] By combining the emergency scenario process steps, task points are set at corresponding positions in the target 3D emergency environment. The task process is integrated through the behavior tree process model. Influencing factors are input into the emergency situation template and linked to the team evaluation and feedback system to form a complete emergency scenario case. The specific implementation process is shown in Figure 7.

[0130] Step S61) Define task points in conjunction with the emergency scenario process steps, and deploy virtual operation stations that match the accident response process in the three-dimensional emergency environment.

[0131] Based on the International Maritime Organization's emergency response standards, key mission positions are precisely located within a 3D ship environment. Taking a ship fire scenario as an example, mission nodes are set up, including fire source location points (requiring operation of infrared thermal imagers), fire pump start stations (requiring two-person collaborative operation), and escape route signs (requiring regular inspection). Each node is associated with ship structural constraints (such as the width of narrow corridors and the opening and closing status of fire doors) and equipment interaction logic (such as fire hydrant water pressure simulation). A physics engine simulates the fire smoke diffusion path and the ship's tilt angle to ensure that environmental physical parameters (such as smoke visibility <1 meter) conform to real sea conditions.

[0132] Step S62) Integrate the task process through the behavior tree process model, map the standard operation steps in the emergency plan to the behavior tree nodes, and establish execution dependencies.

[0133] Implement the digital mapping of the maritime emergency plan using a modular behavior tree. Select node disposal strategy branches (such as sealing the cabin and extinguishing the fire or abandoning the ship), sequence nodes define the operation chain (such as triggering the fire alarm, starting the sprinkler system, and evacuating passengers), and parallel nodes coordinate multi-disciplinary collaboration (such as the turbine crew's emergency repair and the deck crew's rescue being synchronized). Nodes are pre-set with maritime rule constraints (such as a certain ship only executes cabin sealing when the fire area > 2 square meters), and cross-regional linkage is achieved through the event bus mechanism (such as a fire in the engine room triggering a broadcast instruction on the bridge).

[0134] Step S63) Input impact factors in the emergency situation template, configure variables such as environmental parameters, equipment status, and personnel distribution to generate differentiated drill scenarios.

[0135] Step S64) Associate the team evaluation feedback system, transmit the operation data during the drill to the evaluation model in real-time, and dynamically adjust the case difficulty and feedback strategy.

[0136] Step S65) Set the case review and export function, generate a complete case file containing operation trajectories, situation changes, and evaluation results, and support case playback and data export.

[0137] As shown in Figure 11, taking the Unity3D virtual simulation engine as an example, the virtual simulation interaction method for ship emergency drills includes the following steps:

[0138] 1) Simulation start: The user first needs to log in to the system, and then select the drill mode, task type, and participating role according to actual needs. The system supports two modes: single-person training and multi-person collaboration. The user can choose different task scenarios such as fire disposal and collision emergency, and assume specific position roles such as the captain and the engineer.

[0139] 2) Multi-person collaboration: The system ensures that the drill environments of all participants are highly consistent through distributed synchronization technology. This technology realizes timeline synchronization, three-dimensional scene synchronization, spatial coordinate synchronization, and equipment status synchronization, providing reliable technical support for multi-person collaborative training.

[0140] 3) System loading: Automatically load various required resources when the user enters the scene. These resources include core components such as a visual operation interface, a high-precision ship virtual scene, a real-time emergency situation analysis system, virtual character models, a process behavior tree system, and a distributed timer.

[0141] 4) User Operations: During user operations, the system supports interaction between various external input devices and the virtual environment. The scene includes a variety of interactive objects, such as rotatable valves, pickable fire extinguishers, electrical control boxes with status visualization capabilities, and emergency communication devices. All operations must be performed according to the guidelines of the process behavior tree.

[0142] 5) Situation Analysis: The system monitors equipment status data and user operations in real time, performs situation analysis through a hybrid neural network, and generates corresponding escape routes.

[0143] 6) Behavior Tree Update: The behavior tree system employs a dynamic update mechanism. User actions and situational changes will affect the execution path of the behavior tree in real time. The system features a multi-branch decision structure. For example, if the correct measures are taken promptly in the early stages of an incident, the behavior tree will terminate normally; otherwise, improper handling will trigger situational escalation, entering a secondary handling process. Users must complete the task requirements of the new branch within a specified time.

[0144] 7) Evaluation and Feedback: Through the robot's learning model, the system evaluates multi-person collaborative operations, including individual and team metrics. Individual metrics include operational accuracy and response speed, while team metrics include communication density and role complementarity. Within the individual metrics, each person's operational steps are assigned a total score of 100, allocated according to the importance of each step and scored based on the accuracy of the operation. Even if the situation escalates, the user can still receive full marks as long as they successfully complete the task.

[0145] Among team metrics, collaboration and communication efficiency carries the most weight, ensuring that the metrics align with actual business needs. Metric settings must adhere to three principles: observability, quantifiability, and non-overlapping.

[0146] Case Study: The final case study phase systematically integrates the entire exercise process. By setting task nodes in a 3D scenario and combining a behavior tree process model with situational influencing factors, a complete emergency case library is formed. This case library is linked to the evaluation and feedback system, providing a reference for subsequent training.

[0147] Example 2

[0148] Embodiment 2 of this application provides a virtual simulation interaction method for ship emergency drills, such as... Figure 12 As shown, it includes the following steps:

[0149] S10. Based on the ship production design model, a three-dimensional virtual ship emergency drill environment is constructed, integrating ship structure, pipeline system and equipment layout, and simulating flame spread, water flow and structural deformation effects through a dynamic physics engine, supporting multi-scale perspective switching and visualization rendering.

[0150] S20. Set up a multi-person collaborative interactive simulation interface for the three-dimensional virtual ship emergency drill environment;

[0151] S30. Construct an emergency task process framework based on behavior tree, and construct a behavior tree task process model through reinforcement learning and anomaly detection mechanism. The behavior tree task process model is used to display the simulation process of emergency drills using a visual flowchart.

[0152] S40. Construct an emergency situation analysis model to analyze the evolution of the accident and predict the future situation in real time. Use heat maps to display the intensity of the fire and visualize the best evacuation route.

[0153] S50 generates individual and team evaluation metrics based on machine learning models, provides real-time visual feedback, and performs collaborative optimization for individuals and teams;

[0154] S60. In conjunction with the emergency task process steps, set task points at corresponding positions in the three-dimensional virtual ship emergency drill environment, integrate the emergency task process through the behavior tree task process model, and generate differentiated emergency case simulation scenarios.

[0155] This invention achieves a comprehensive technical effect of high simulation, high intelligence, and high collaboration through a virtual simulation interaction method for ship emergency drills. The method constructs a three-dimensional virtual drill environment based on a ship production design model, combining a dynamic physics engine to simulate physical effects such as flame spread, water flow, and structural deformation, significantly enhancing the realism and immersion of emergency training. By setting up a multi-user collaborative interactive simulation interface, the system supports real-time online collaboration among multiple users, intelligently allocating operation permissions and automatically resolving operational conflicts, ensuring the stability and smoothness of the collaborative drill process. A behavior tree-based task flow control mechanism automatically manages emergency tasks, and combined with reinforcement learning and anomaly detection algorithms, it dynamically adjusts task priorities and handles anomalies, improving the flexibility and autonomous decision-making capabilities of the drills. The system further constructs an emergency situation analysis model, predicts the development trend of accidents in real time, and visually displays the fire distribution and optimal evacuation routes in the form of heat maps, providing intelligent assistance for command and decision-making. A machine learning-based evaluation mechanism generates quantitative evaluation indicators for individuals and teams, automatically analyzes collaborative performance, and outputs personalized training reports, achieving intelligent feedback and training optimization. In addition, the system can quickly generate differentiated simulation scenarios based on different emergency procedures, support case review and multi-difficulty exercise configuration, facilitate continuous improvement of training strategies and enhance team emergency response capabilities, thereby building a virtual simulation platform for ship emergency drills that integrates high simulation, strong interactivity, intelligent evaluation and replayability.

[0156] Furthermore, the aforementioned construction of a three-dimensional virtual ship emergency drill environment based on the ship production design model integrates the ship's structure, piping system, and equipment layout, and simulates flame spread, water flow, and structural deformation effects through a dynamic physics engine, supporting multi-scale perspective switching and visualization rendering, including:

[0157] A basic 3D scene is constructed based on the ship production design model. The ship setting data is converted into a 3D mesh model. By refining the complexity of the 3D mesh model, the ship structure, pipeline system and equipment layout are integrated to form an interactive basic scene.

[0158] A dynamic physics engine is implemented by integrating environmental sensor data, and the dynamic physics engine is used to simulate the dynamic physical effects of flame spread, water flow, and structural deformation.

[0159] Build a library of interactive operation models for ship equipment, and set up equipment operation logic including valves, fire-fighting equipment, electrical control boxes, and emergency communication devices;

[0160] An emergency incident triggering and evolution mechanism is set up. The triggering conditions for fire, collision and oil spill accidents are defined based on the fault tree analysis method (FTA), and the accident evolution path is constructed using a finite state machine (FSM).

[0161] Build a multi-scale visualization rendering system, integrate ray tracing to achieve visualization rendering, and support multi-scale perspective switching from global scene top view to device operation.

[0162] Furthermore, the provision of a multi-user collaborative interactive simulation interface for the three-dimensional virtual ship emergency drill environment includes:

[0163] A distributed simulation architecture is constructed, a master-slave server cluster is deployed, and a regional partitioning method is used to achieve scene block management;

[0164] A hierarchical management mechanism for operation permissions is set up, defining three types of roles—commander, operator, and observer—based on a role-based access control model. The token system is used to dynamically allocate and seize the right to operate the equipment.

[0165] Incremental state update is used to transmit changed data, and UDP / TCP protocol and forward error correction mechanism are combined to ensure operational continuity in network packet loss environment;

[0166] When multiple users operate the same device at the same time, the system resolves conflicts based on timestamp priority and operation legality verification algorithm, and provides visual conflict prompts and manual retry function;

[0167] Build a cross-platform network communication protocol, define a binary communication protocol based on Protobuf, and support data interoperability between multiple terminals.

[0168] Furthermore, the emergency task process framework is constructed based on behavior trees, and a behavior tree task process model is built through reinforcement learning and anomaly detection mechanisms. This behavior tree task process model is used to visualize the simulation process of emergency drills using flowcharts, including:

[0169] An emergency task process framework is built based on behavior tree, and a hierarchical behavior tree structure is used to define the main task node and sub-task node. Task status sharing is achieved through the blackboard system.

[0170] An emergency decision-making knowledge graph is set up, and a graph database containing at least the International Maritime Organization norms and ship emergency manual knowledge is constructed. Knowledge retrieval and reasoning are performed through the graph database.

[0171] The priority weights of emergency tasks are dynamically adjusted based on the reinforcement learning algorithm. The execution order of emergency tasks is adjusted in real time by adjusting the priority weights of emergency tasks according to their urgency.

[0172] Construct an abnormal state detection and recovery mechanism, predict task execution deviations through a long short-term memory network, and trigger a plan rollback or issue a manual intervention prompt when an abnormality is detected.

[0173] A multi-task parallel processing engine is set up to enable concurrent execution of tasks, and task scheduling and execution are decoupled through message queues;

[0174] An automatic emergency plan matching system is set up to match similar accident scenarios from the historical drill database and recommend the optimal accident scenario. The simulation process of the emergency drill is then displayed using a visual flowchart based on the optimal accident scenario.

[0175] Furthermore, the construction of the emergency situation analysis model, which analyzes the evolution of the accident in real time and predicts future trends, uses heat maps to display fire intensity, and visualizes the optimal evacuation route includes:

[0176] A multi-source data fusion engine is set up to integrate heterogeneous data such as device status data and user operation logs in the virtual environment, and a stream processing framework is used to realize real-time data aggregation and preprocessing.

[0177] Construct an accident evolution prediction model, based on a hybrid neural network that takes current accident parameters as input and outputs predictions of future trends;

[0178] Set up a real-time situation visualization system, overlay situation layers in the virtual scene, use heat maps to show the fire intensity, and use path planning algorithms to show the best evacuation route;

[0179] Furthermore, the generation of individual and team evaluation metrics based on machine learning models, providing real-time visual feedback, and collaboratively optimizing individuals and teams includes:

[0180] Construct a multi-dimensional evaluation index system, define individual and team indicators, and use the analytic hierarchy process to determine the weight of each indicator.

[0181] Set up a machine learning-driven evaluation model, train the evaluation model based on the algorithm, input operational data, and output individual ability scores and team collaboration index;

[0182] Build a visual feedback system that overlays evaluation panels in a virtual scene, displays indicators using charts, and supports data export and automatic generation of text summaries;

[0183] Set up a team collaboration effectiveness analysis module, construct a user relationship network diagram based on analysis methods, identify key nodes and information silos, and provide suggestions for optimizing collaboration patterns;

[0184] Set up a personalized assessment report generator to customize report content based on preset templates according to user roles.

[0185] Furthermore, the step of combining emergency task process steps, setting task points at corresponding positions in the three-dimensional virtual ship emergency drill environment, and integrating the emergency task process through the behavior tree task process model to generate differentiated emergency case simulation scenarios includes:

[0186] Task points are defined in conjunction with emergency scenario process steps, and virtual operating positions matching the accident response process are deployed in the three-dimensional virtual ship emergency drill environment;

[0187] The task flow is integrated through the behavior tree task flow model, and the standard operating steps in the emergency plan are mapped to behavior tree nodes and execution dependencies are established.

[0188] Input influencing factors into the emergency situation template, and configure environmental parameters, equipment status, and personnel distribution variables to generate differentiated drill scenarios;

[0189] The associated team evaluation and feedback system transmits operational data during the exercise to the evaluation model in real time, dynamically adjusting the difficulty of the exercise scenario and the feedback strategy.

[0190] Generate differentiated emergency case simulation scenarios that include operation trajectories, situational changes, and assessment results. These differentiated emergency case simulation scenarios support case playback and data export.

[0191] In one embodiment, such as Figure 13 As shown, a virtual simulation interactive device for ship emergency drills is provided, comprising:

[0192] The 3D virtual ship emergency environment construction module is used to build a 3D virtual ship emergency drill environment based on the ship production design model. It integrates the ship structure, pipeline system and equipment layout, and simulates flame spread, water flow and structural deformation effects through a dynamic physics engine. It supports multi-scale perspective switching and visualization rendering.

[0193] A multi-user collaborative interactive simulation interface module is used to set up a multi-user collaborative interactive simulation interface for the three-dimensional virtual ship emergency drill environment.

[0194] The behavior tree task process construction module is used to build an emergency task process framework based on behavior trees, and to build a behavior tree task process model through reinforcement learning and anomaly detection mechanisms. The behavior tree task process model is used to display the simulation process of emergency drills using a visual flowchart.

[0195] The emergency situation analysis and prediction module is used to build emergency situation analysis models, analyze the evolution of accidents in real time and predict future situations, use heat maps to display fire intensity, and visualize the best evacuation routes.

[0196] The intelligent assessment and collaborative optimization module is used to generate individual and team assessment metrics based on machine learning models, provide real-time visual feedback, and perform collaborative optimization for individuals and teams.

[0197] The emergency case generation and simulation module is used to combine emergency task process steps, set task points at corresponding positions in the three-dimensional virtual ship emergency drill environment, integrate the emergency task process through the behavior tree task process model, and generate differentiated emergency case simulation scenarios.

[0198] Specific limitations regarding the virtual simulation interactive device for ship emergency drills can be found in the limitations of the virtual simulation interactive method for ship emergency drills mentioned above, and will not be repeated here. Each module in the aforementioned virtual simulation interactive device for ship emergency drills can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in the computer device in hardware form, or stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0199] This invention provides a highly realistic and intelligent training platform for multi-user collaborative ship emergency drills using a behavior tree-driven virtual simulation method. The system constructs a 3D virtual environment based on ship design data and simulates effects such as flames and water flow through a dynamic physics engine, enhancing the realism of the training. It supports multi-user collaborative operation, intelligently assigns permissions, and automatically resolves conflicts to ensure smooth collaborative drills. Task flows are automatically managed through behavior trees, with dynamic priority adjustments and automatic handling of anomalies, improving the flexibility of the drills. Real-time emergency situation analysis and visualization of fire and evacuation routes aid decision-making. A machine learning-driven evaluation system provides personalized feedback, analyzes team collaboration issues, and generates training reports. The system supports multi-scenario configuration and case reviews, facilitating the optimization of training programs and improving the team's emergency response capabilities.

[0200] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 14 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores virtual simulation interactive data for ship emergency drills. The network interface communicates with external terminals via a network connection. When the computer program is executed by the processor, it implements a virtual simulation interactive method for ship emergency drills.

[0201] Those skilled in the art will understand that Figure 14 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0202] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps:

[0203] Based on the ship production design model, a three-dimensional virtual ship emergency drill environment is constructed, integrating ship structure, pipeline system and equipment layout, and simulating flame spread, water flow and structural deformation effects through a dynamic physics engine, supporting multi-scale perspective switching and visualization rendering.

[0204] A multi-user collaborative interactive simulation interface is set up for the three-dimensional virtual ship emergency drill environment;

[0205] An emergency task process framework is constructed based on behavior trees, and a behavior tree task process model is constructed through reinforcement learning and anomaly detection mechanisms. The behavior tree task process model is used to display the simulation process of emergency drills using a visual flowchart.

[0206] Construct an emergency situation analysis model to analyze the evolution of the accident in real time and predict the future situation. Use heat maps to display the intensity of the fire and visualize the best evacuation route.

[0207] It generates individual and team evaluation metrics based on machine learning models, provides real-time visual feedback, and enables collaborative optimization of individuals and teams;

[0208] In conjunction with the emergency task process steps, task points are set at corresponding positions in the three-dimensional virtual ship emergency drill environment. The emergency task process is integrated through the behavior tree task process model to generate differentiated emergency case simulation scenarios.

[0209] For specific limitations on the steps implemented by the processor when executing a computer program, please refer to the limitations on the virtual simulation interaction method for ship emergency drills mentioned above, which will not be repeated here.

[0210] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0211] Based on the ship production design model, a three-dimensional virtual ship emergency drill environment is constructed, integrating ship structure, pipeline system and equipment layout, and simulating flame spread, water flow and structural deformation effects through a dynamic physics engine, supporting multi-scale perspective switching and visualization rendering.

[0212] A multi-user collaborative interactive simulation interface is set up for the three-dimensional virtual ship emergency drill environment;

[0213] An emergency task process framework is constructed based on behavior trees, and a behavior tree task process model is constructed through reinforcement learning and anomaly detection mechanisms. The behavior tree task process model is used to display the simulation process of emergency drills using a visual flowchart.

[0214] Construct an emergency situation analysis model to analyze the evolution of the accident in real time and predict the future situation. Use heat maps to display the intensity of the fire and visualize the best evacuation route.

[0215] It generates individual and team evaluation metrics based on machine learning models, provides real-time visual feedback, and enables collaborative optimization of individuals and teams;

[0216] In conjunction with the emergency task process steps, task points are set at corresponding positions in the three-dimensional virtual ship emergency drill environment. The emergency task process is integrated through the behavior tree task process model to generate differentiated emergency case simulation scenarios.

[0217] For specific limitations on the steps implemented when a computer program is executed by a processor, please refer to the limitations on the virtual simulation interaction method for ship emergency drills mentioned above, which will not be repeated here.

[0218] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0219] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A virtual simulation interactive method for ship emergency drills, characterized in that, Includes the following steps: Based on the ship production design model, a three-dimensional virtual ship emergency drill environment is constructed, integrating ship structure, pipeline system and equipment layout, and simulating flame spread, water flow and structural deformation effects through a dynamic physics engine, supporting multi-scale perspective switching and visualization rendering. A multi-user collaborative interactive simulation interface is set up for the three-dimensional virtual ship emergency drill environment; An emergency task process framework is constructed based on behavior trees, and a behavior tree task process model is constructed through reinforcement learning and anomaly detection mechanisms. The behavior tree task process model is used to display the simulation process of emergency drills using a visual flowchart. Construct an emergency situation analysis model to analyze the evolution of the accident in real time and predict the future situation. Use heat maps to display the intensity of the fire and visualize the best evacuation route. It generates individual and team evaluation metrics based on machine learning models, provides real-time visual feedback, and enables collaborative optimization of individuals and teams; In conjunction with the emergency task process steps, task points are set at corresponding positions in the three-dimensional virtual ship emergency drill environment. The emergency task process is integrated through the behavior tree task process model to generate differentiated emergency case simulation scenarios. The step of constructing an emergency task process framework based on behavior trees, and building a behavior tree task process model through reinforcement learning and anomaly detection mechanisms, wherein the behavior tree task process model is used to visualize the simulation process of emergency drills using flowcharts, includes: An emergency task process framework is built based on behavior tree, and a hierarchical behavior tree structure is used to define the main task node and sub-task node. Task status sharing is achieved through the blackboard system. An emergency decision-making knowledge graph is set up, and a graph database containing at least the International Maritime Organization norms and ship emergency manual knowledge is constructed. Knowledge retrieval and reasoning are performed through the graph database. The priority weights of emergency tasks are dynamically adjusted based on the reinforcement learning algorithm. The execution order of emergency tasks is adjusted in real time by adjusting the priority weights of emergency tasks according to their urgency. Construct an abnormal state detection and recovery mechanism, predict task execution deviations through a long short-term memory network, and trigger a plan rollback or issue a manual intervention prompt when an abnormality is detected. A multi-task parallel processing engine is set up to enable concurrent execution of tasks, and task scheduling and execution are decoupled through message queues; An automatic emergency plan matching system is set up to match similar accident scenarios from the historical drill database and recommend the optimal accident scenario. The simulation process of the emergency drill is displayed using a visual flowchart according to the optimal accident scenario. The step of combining emergency task process steps, setting task points at corresponding positions in the three-dimensional virtual ship emergency drill environment, and integrating the emergency task process through the behavior tree task process model to generate differentiated emergency case simulation scenarios includes: Task points are defined in conjunction with emergency scenario process steps, and virtual operating positions matching the accident response process are deployed in the three-dimensional virtual ship emergency drill environment; The task flow is integrated through the behavior tree task flow model, and the standard operating steps in the emergency plan are mapped to behavior tree nodes and execution dependencies are established. Input influencing factors into the emergency situation template, and configure environmental parameters, equipment status, and personnel distribution variables to generate differentiated drill scenarios; The associated team evaluation and feedback system transmits operational data during the exercise to the evaluation model in real time, dynamically adjusting the difficulty of the exercise scenario and the feedback strategy. Generate differentiated emergency case simulation scenarios that include operation trajectories, situational changes, and assessment results. These differentiated emergency case simulation scenarios support case playback and data export.

2. The virtual simulation interactive method for ship emergency drills according to claim 1, characterized in that, The aforementioned three-dimensional virtual ship emergency drill environment, based on a ship production design model, integrates ship structure, piping systems, and equipment layout. It simulates flame spread, water flow, and structural deformation effects using a dynamic physics engine, supporting multi-scale perspective switching and visualization rendering. This includes: A basic 3D scene is constructed based on the ship production design model. The ship setting data is converted into a 3D mesh model. By refining the complexity of the 3D mesh model, the ship structure, pipeline system and equipment layout are integrated to form an interactive basic scene. A dynamic physics engine is implemented by integrating environmental sensor data, and the dynamic physics engine is used to simulate the dynamic physical effects of flame spread, water flow, and structural deformation. Build a library of interactive operation models for ship equipment, and set up equipment operation logic including valves, fire-fighting equipment, electrical control boxes, and emergency communication devices; An emergency incident triggering and evolution mechanism is set up. The triggering conditions for fire, collision and oil spill accidents are defined based on the fault tree analysis method, and the accident evolution path is constructed using a finite state machine. Build a multi-scale visualization rendering system, integrate ray tracing to achieve visualization rendering, and support multi-scale perspective switching from global scene top view to device operation.

3. The virtual simulation interactive method for ship emergency drills according to claim 1, characterized in that, The multi-user collaborative interactive simulation interface for the three-dimensional virtual ship emergency drill environment includes: A distributed simulation architecture is constructed, a master-slave server cluster is deployed, and a regional partitioning method is used to achieve scene block management; A hierarchical management mechanism for operation permissions is set up, defining three types of roles—commander, operator, and observer—based on a role-based access control model. The token system is used to dynamically allocate and seize the right to operate the equipment. Incremental state update is used to transmit changed data, and UDP / TCP protocol and forward error correction mechanism are combined to ensure operational continuity in network packet loss environment; When multiple users operate the same device at the same time, the system resolves conflicts based on timestamp priority and operation legality verification algorithm, and provides visual conflict prompts and manual retry function; Build a cross-platform network communication protocol, define a binary communication protocol based on Protobuf, and support data interoperability between multiple terminals.

4. The virtual simulation interactive method for ship emergency drills according to claim 1, characterized in that, The construction of the emergency situation analysis model, which analyzes the evolution of the accident in real time and predicts future trends, uses heat maps to display fire intensity, and visualizes the best evacuation routes includes: A multi-source data fusion engine is set up to integrate heterogeneous data such as device status data and user operation logs in the virtual environment, and a stream processing framework is used to realize real-time data aggregation and preprocessing. Construct an accident evolution prediction model, based on a hybrid neural network that takes current accident parameters as input and outputs predictions of future trends; A real-time situation visualization system is set up, overlaying situation layers on a virtual scene, using heat maps to display fire intensity, and using path planning algorithms to display the optimal evacuation route.

5. The virtual simulation interactive method for ship emergency drills according to claim 1, characterized in that, The process of generating individual and team evaluation metrics based on machine learning models, providing real-time visual feedback, and collaboratively optimizing individuals and teams includes: Construct a multi-dimensional evaluation index system, define individual and team indicators, and use the analytic hierarchy process to determine the weight of each indicator. Set up a machine learning-driven evaluation model, train the evaluation model based on the algorithm, input operational data, and output individual ability scores and team collaboration index; Build a visual feedback system that overlays evaluation panels in a virtual scene, displays indicators using charts, and supports data export and automatic generation of text summaries; Set up a team collaboration effectiveness analysis module, construct a user relationship network diagram based on analysis methods, identify key nodes and information silos, and provide suggestions for optimizing collaboration patterns; Set up a personalized assessment report generator to customize report content based on preset templates according to user roles.

6. A virtual simulation interactive device for ship emergency drills, characterized in that, The apparatus for implementing the virtual simulation interactive method for ship emergency drills according to any one of claims 1 to 5, the apparatus comprising: The 3D virtual ship emergency environment construction module is used to build a 3D virtual ship emergency drill environment based on the ship production design model. It integrates the ship structure, pipeline system and equipment layout, and simulates flame spread, water flow and structural deformation effects through a dynamic physics engine. It supports multi-scale perspective switching and visualization rendering. A multi-user collaborative interactive simulation interface module is used to set up a multi-user collaborative interactive simulation interface for the three-dimensional virtual ship emergency drill environment. The behavior tree task process construction module is used to build an emergency task process framework based on behavior trees, and to build a behavior tree task process model through reinforcement learning and anomaly detection mechanisms. The behavior tree task process model is used to display the simulation process of emergency drills using a visual flowchart. The emergency situation analysis and prediction module is used to build emergency situation analysis models, analyze the evolution of accidents in real time and predict future situations, use heat maps to display fire intensity, and visualize the best evacuation routes. The intelligent assessment and collaborative optimization module is used to generate individual and team assessment metrics based on machine learning models, provide real-time visual feedback, and perform collaborative optimization for individuals and teams. The emergency case generation and simulation module is used to combine emergency task process steps, set task points at corresponding positions in the three-dimensional virtual ship emergency drill environment, integrate the emergency task process through the behavior tree task process model, and generate differentiated emergency case simulation scenarios.

7. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

Images