Multi-scene cooperative operation three-dimensional simulation deduction and intelligent scheduling system

Abstract

The application relates to the field of three-dimensional simulation scheduling, and discloses a three-dimensional simulation deduction and intelligent scheduling system for multi-scene cooperative operation. The system comprises a scene fusion modeling module, a dynamic deduction period calculation module, a resource dimension analysis module and a deduction strategy generation module. The scene fusion modeling module generates a three-dimensional cooperative operation space model containing spatial coordinate mapping relationships of different operation scenes and resource flow transfer paths between scenes based on multi-scene parameters configured by a user; the dynamic deduction period calculation module obtains the remaining time from the current time to a preset deduction end time, and calculates an initial deduction period by calling a first dynamic mapping relationship; the resource dimension analysis module identifies the number of resource types participating in cooperative operation, and generates a period adjustment factor by calling a second dynamic mapping relationship; and the deduction strategy generation module obtains an adaptive deduction period by weightedly fusing the two, and formulates a three-dimensional simulation deduction strategy accordingly, so that the system is adapted to dynamic changes of multiple scenes, and efficient cooperation between simulation and scheduling is realized.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional simulation scheduling technology, specifically a three-dimensional simulation deduction and intelligent scheduling system for multi-scenario collaborative operations. Background Technology

[0002] In modern logistics, engineering construction, intelligent manufacturing, and other fields, multi-scenario collaborative operations have become an important mode for improving operational efficiency. The matching 3D simulation and intelligent scheduling technologies are also gradually becoming one of the key technologies supporting the implementation of these operational modes. However, current market-based systems still have many limitations in practical applications and struggle to meet the complex needs of multi-scenario collaborative operations.

[0003] In the scenario modeling stage, most existing systems construct 3D models for single operational scenarios, with models for different scenarios operating independently and lacking a unified spatial coordinate mapping mechanism. This leads to inaccurate spatial relationships between different scenarios during multi-scenario collaborative operations. For example, in engineering construction scenarios, the 3D models of the design scenario and the construction scenario may not be accurately connected due to different coordinate systems, causing deviations between the movement paths of construction equipment in simulations and the actual design plans. Furthermore, existing systems tend to plan resource flow paths between scenarios in a simplistic manner, typically considering only the unidirectional flow of a single resource and neglecting the cross-flow requirements of multiple resources across different scenarios. This results in frequent resource congestion or idleness during simulations, failing to realistically simulate the dynamic resource flow process in actual operations.

[0004] Regarding the setting of dynamic simulation cycles, existing technologies generally adopt a fixed cycle mode. Once the simulation cycle is determined, the entire simulation process continues according to that cycle without adjustment based on the simulation progress. This approach has significant drawbacks: when the remaining time between the current moment and the preset simulation end time is long, a fixed cycle leads to an overly lengthy simulation process, wasting a large amount of computational resources on repetitive simulation steps and reducing simulation efficiency; conversely, when the remaining time is short, a fixed cycle may prevent the simulation task from being completed within the preset time, resulting in incomplete simulation results that cannot provide effective reference for subsequent scheduling.

[0005] In terms of resource dimension analysis and inference strategy formulation, existing systems do not provide a comprehensive analysis of resources participating in collaborative operations. They often focus only on a few core resource types, such as equipment and personnel, while neglecting other important resource types such as raw materials and information data. Even when some systems consider multiple resource types, they do not link the number of resource types to the inference cycle, making it impossible to adjust inference parameters according to changes in the number of resource types. For example, when the number of resource types participating in collaborative operations increases significantly, existing systems still perform inferences according to the original cycle, resulting in an inability to fully cover the scheduling status of all resources in each cycle. Consequently, the inference results fail to reflect the actual situation of multi-resource collaborative operations. Furthermore, the generation of inference strategies in existing systems lacks flexibility, mostly based on preset fixed rules, failing to effectively integrate the inference cycle with resource conditions. This results in inference strategies that cannot adapt to the dynamic changes in resources under multiple scenarios, thus affecting the accuracy of intelligent scheduling. Summary of the Invention

[0006] The purpose of this invention is to provide a three-dimensional simulation and intelligent scheduling system for multi-scenario collaborative operations, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides a three-dimensional simulation and intelligent scheduling system for multi-scenario collaborative operations, the system comprising:

[0008] The scene fusion modeling module is used to generate a three-dimensional collaborative work space model based on user-configured multi-scene parameters. The three-dimensional collaborative work space model includes the spatial coordinate mapping relationship of different work scenes and the resource flow path between scenes.

[0009] The dynamic simulation cycle calculation module is used to obtain the remaining time from the current time to the preset simulation end time, and to calculate the initial simulation cycle based on the remaining time by calling the first dynamic mapping relationship.

[0010] The resource dimension analysis module is used to identify the number of resource types participating in collaborative operations, and to generate a periodic adjustment factor based on the number of resource types by calling the second dynamic mapping relationship.

[0011] The simulation strategy generation module is used to perform weighted fusion processing on the initial simulation period and the period adjustment factor to generate an adaptive simulation period, and to formulate a three-dimensional simulation simulation strategy based on the adaptive simulation period.

[0012] Preferably, the dynamic extrapolation cycle calculation module is specifically used for:

[0013] Extract the spatiotemporal constraint parameters of each operation scenario in the three-dimensional collaborative operation space model;

[0014] The remaining duration and the spatiotemporal constraint parameters are input into a dynamic time slicing algorithm to generate the initial simulation cycle, which is negatively correlated with the complexity of the task.

[0015] Preferably, the resource dimension analysis module is specifically used for:

[0016] The types of interface protocols for heterogeneous resources within the stated resource types are statistically analyzed.

[0017] Based on the aforementioned interface protocol type, a third dynamic mapping relationship is invoked to generate a resource complexity coefficient;

[0018] The periodic adjustment factor is output by performing a convolution operation between the number of resource types and the resource complexity coefficient.

[0019] Preferably, the system further includes:

[0020] The real-time trajectory acquisition module is used to synchronously acquire the running trajectory data of physical entities in various work scenarios;

[0021] The state space construction module is used to reconstruct the running trajectory data in tensor space to generate a multidimensional state space with a dimension equal to the number of feature parameters. The coordinate axes of the multidimensional state space correspond to the rate of change of the dynamic features of the trajectory.

[0022] Preferably, the state space construction module is further used for:

[0023] Mark the characteristic coordinates of historical anomalous events in the multidimensional state space;

[0024] The spatial clustering density value of the feature coordinates is calculated based on the density clustering algorithm.

[0025] An abnormal state warning domain is generated based on the area where the spatial aggregation density value exceeds a preset threshold.

[0026] Preferably, the system further includes:

[0027] The deduction deviation analysis module is used during the execution of the three-dimensional simulation deduction strategy:

[0028] Real-time capture of the actual position coordinates of physical entities in the multi-dimensional state space;

[0029] A three-dimensional difference analysis was performed between the actual location coordinates and the predicted coordinates, considering the cumulative offset in the time domain, the similarity of spatial distribution, and the matching degree of dynamic event sequences.

[0030] The output is a 3D inference deviation dataset containing the deviation dimension.

[0031] Preferably, the deduction deviation analysis module is specifically used for:

[0032] The cumulative offset in the time domain is dynamically normalized and aligned using a sliding time window.

[0033] The curvature matching calculation is performed using the manifold distance algorithm to assess spatial distribution similarity.

[0034] Phase synchronization is evaluated using spatiotemporal entropy analysis on dynamic event sequences.

[0035] Preferably, the system further includes:

[0036] The collaborative optimization module is used to: input the three-dimensional inference deviation dataset into a graph neural network;

[0037] The resource flow path weights in the three-dimensional collaborative operation space model are updated based on the output nodes of the graph neural network.

[0038] Generate cross-scenario resource scheduling instructions based on the updated resource flow path weights.

[0039] Preferably, the collaborative optimization module is further used for:

[0040] When the time-domain cumulative offset of a specific job scenario exceeds the tolerance threshold,

[0041] Send a periodic recalibration request to the inference strategy generation module;

[0042] The dynamic simulation cycle calculation module is triggered to recalculate the initial simulation cycle.

[0043] Preferably, the scene fusion modeling module is further used for:

[0044] Continuously receive resource scheduling instructions sent by the collaborative optimization module;

[0045] The resource flow paths between scenes in the three-dimensional collaborative operation space model are dynamically updated according to the resource scheduling instructions.

[0046] The updated spatial coordinate mapping relationship is synchronized to the inference strategy generation module.

[0047] Compared with the prior art, the beneficial effects of the present invention are:

[0048] The 3D collaborative work space model generated by the scene fusion modeling module not only integrates the spatial coordinate mapping relationships of different work scenarios but also clarifies the resource flow paths between scenarios. This makes multiple scenarios no longer independent simulation units but form an organically linked whole. In practical applications, whether it is warehousing, transportation, and distribution scenarios in logistics transportation or design, construction, and operation and maintenance scenarios in engineering construction, the spatial positions between different scenarios can be accurately matched, and the flow of resources between scenarios can be carried out in an orderly manner according to preset paths. This avoids simulation deviations caused by scenario fragmentation or ambiguous resource flow paths, making the 3D simulation more closely match the spatial layout and resource flow patterns of actual operations, and more realistically reproducing the entire process of multi-scenario collaborative operations.

[0049] The dynamic simulation cycle calculation module calculates the initial simulation cycle based on the remaining time, breaking the limitations of the traditional fixed cycle. During the simulation, the system can sense the time change from the current time to the preset simulation end time in real time. When the remaining time is long, the initial simulation cycle can be set to a relatively reasonable length, making full use of the time for detailed simulation and ensuring that the simulation of each operation can be fully carried out. When the remaining time is short, the initial simulation cycle will be adjusted accordingly to prioritize the completion of the simulation of core operation stages and avoid simulation interruption or incomplete results due to insufficient time. This time-based dynamic adjustment method allows the simulation process to better adapt to time constraints, improves the timeliness and feasibility of the simulation, and ensures that effective simulation results are output within the preset time.

[0050] The resource dimension analysis module generates a cycle adjustment factor by identifying the number of resource types, ensuring that the adjustment of the simulation cycle fully matches the resource scale. When there are many resource types participating in collaborative operations, the cycle adjustment factor will appropriately optimize the initial simulation cycle, extending or adjusting the cycle structure to ensure that the simulation process covers the scheduling status of all resource types, avoiding incomplete scheduling schemes due to missing resource types. When there are few resource types, the cycle adjustment factor will streamline the initial simulation cycle, reducing unnecessary simulation steps and improving simulation efficiency. This dynamic adjustment based on the number of resource types ensures that the simulation cycle is highly consistent with the actual resource situation, avoiding resource waste or insufficient simulation caused by cycles that are too long or too short.

[0051] The simulation strategy generation module weighted and merges the initial simulation cycle with a cycle adjustment factor to generate an adaptive simulation cycle, and formulates a simulation strategy accordingly, further enhancing the system's adaptability and flexibility. In multi-scenario collaborative operations, time and resources are often dynamically changing. For example, the addition of a work scenario may lead to changes in the type and quantity of resources, or the simulation time may be shortened due to unforeseen circumstances. In such cases, the adaptive simulation cycle will be quickly adjusted, allowing the simulation strategy to be optimized in a timely manner. This adaptive simulation strategy can effectively cope with dynamic changes in multiple scenarios, ensuring that the system can output a reasonable simulation plan regardless of adjustments to time constraints or resource conditions, providing precise guidance for intelligent scheduling. At the same time, the entire system operation requires minimal human intervention, reducing the impact of human factors on simulation and scheduling, making the simulation and intelligent scheduling process of multi-scenario collaborative operations more efficient and reliable. It can play a positive role in multi-scenario collaborative operations in various industries, promoting the precision and intelligent development of collaborative operations. Attached Figure Description

[0052] Figure 1 This is a timing diagram of the three-dimensional simulation and intelligent scheduling system for multi-scenario collaborative operations described in this invention.

[0053] Figure 2 A flowchart for generating the periodic adjustment factor;

[0054] Figure 3 A flowchart for real-time trajectory acquisition and state space construction;

[0055] Figure 4 A flowchart generated for a 3D extrapolation deviation dataset;

[0056] Figure 5 A flowchart for generating cross-scenario resource scheduling instructions. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. 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.

[0058] Please see Figure 1 This invention provides a three-dimensional simulation and intelligent scheduling system for multi-scenario collaborative operations, the system comprising:

[0059] The scene fusion modeling module generates a 3D collaborative operation space model based on user-configured multi-scene parameters. This model includes spatial coordinate mapping relationships of different operation scenarios and resource flow paths between scenarios. The dynamic simulation cycle calculation module obtains the remaining time from the current time to the preset simulation end time, and calculates the initial simulation cycle based on the remaining time by calling the first dynamic mapping relationship. The resource dimension analysis module identifies the number of resource types participating in the collaborative operation, and generates a cycle adjustment factor based on the number of resource types by calling the second dynamic mapping relationship. The simulation strategy generation module performs weighted fusion processing on the initial simulation cycle and the cycle adjustment factor to generate an adaptive simulation cycle, and formulates a 3D simulation strategy based on the adaptive simulation cycle.

[0060] Example 1: See Figure 2 The dynamic simulation cycle calculation module begins by extracting the spatiotemporal constraint parameters of each work scenario from the 3D collaborative work space model. These parameters include key constraints such as time window limits, spatial capacity limits, and resource usage time slots for each work scenario. The time window limit defines the operable time range of the scenario, the spatial capacity limits the number of physical entities the scenario can accommodate, and the resource usage time slot specifies the specific time period during which resources can be called upon. The extraction process is achieved by parsing the metadata of the spatial model. The system identifies and reads the constraint definition information embedded in the model and converts it into a computable parameter format. The combination of remaining time and spatiotemporal constraint parameters is completed through a dynamic time slicing algorithm. This algorithm employs an adaptive time segmentation strategy, allocating time slice lengths according to the complexity and strictness of the constraints. For work scenarios with complex and strict constraints, the algorithm allocates shorter time slices to ensure simulation accuracy; for scenarios with simple and lenient constraints, it allocates longer time slices to improve simulation efficiency. The algorithm outputs an initial simulation cycle that is negatively correlated with the complexity of the work. This means that the more complex the work environment, the shorter the simulation cycle used by the system, in order to capture dynamic changes more precisely.

[0061] The resource dimension analysis module requires identifying the number of resource types participating in collaborative operations. The system traverses all operational scenarios, statistically analyzing different types of resource entities, including equipment resources, human resources, and technical resources. Each resource type is categorized and registered in detail according to its functional characteristics and usage requirements, establishing a complete resource type list. Based on resource type identification, the module further statistically analyzes the interface protocol types of heterogeneous resources. Heterogeneous resources refer to resource entities with different technical specifications and communication requirements; they may employ different data exchange protocols, control command sets, or communication standards. The system detects the interface protocols supported by each resource type, including real-time communication protocols, data exchange protocols, and control command protocols, and records the specific technical parameters and usage conditions of each protocol.

[0062] Based on the types of interface protocols, the module uses a third dynamic mapping relationship to generate resource complexity coefficients. This third dynamic mapping relationship is a pre-defined correspondence table that maps different types and numbers of interface protocols to specific complexity values. This mapping relationship considers factors such as the technical complexity, implementation difficulty, and interoperability requirements of the protocols, ensuring that the complexity coefficients accurately reflect the actual complexity of the resources. The module performs a convolution operation between the number of resource types and the resource complexity coefficients. The convolution operation uses a weighted summation function as the convolution kernel, where the weight coefficients are dynamically adjusted according to the importance of the resource type and the interface complexity. Higher weights are assigned to critical resource types and complex interface protocols, while lower weights are assigned to secondary resources and simple interfaces. The output of the operation is a period adjustment factor, which quantifies the impact of the resource dimension on the deduction period, providing a basis for subsequent deduction period adjustments.

[0063] The entire implementation process is automated, requiring no manual intervention from parameter extraction to result output. The system has built-in data validation to ensure the accuracy and completeness of all input parameters. Fault tolerance is employed during computation, automatically using backup calculation methods when encountering abnormal data. All intermediate results and final outputs are logged in the system log, supporting subsequent auditing and optimization. The dynamic time-slicing algorithm supports parameter adjustment, allowing users to modify the time-slice allocation strategy according to specific needs. Resource complexity coefficient calculation supports custom mapping relationships, allowing different complexity evaluation standards to be used for different application scenarios. The weight coefficients of convolution operations are configurable, meeting the differentiated definitions of resource importance for different projects.

[0064] The system also provides real-time monitoring capabilities, continuously monitoring the execution status and result quality of each stage during the computation process. When anomalies are detected or the computation results exceed expectations, the system automatically triggers an alarm mechanism, prompting operators to check and intervene. Simultaneously, the system supports pausing and resuming the computation process, allowing adjustments and optimizations without system downtime. All algorithms and parameters used in the implementation process have undergone thorough testing and verification to ensure stability and reliability under different scenarios. The system adopts a modular design, with each computation stage relatively independent, facilitating individual maintenance and upgrades.

[0065] Example 2: See Figure 3The real-time trajectory acquisition module is responsible for synchronously acquiring the trajectory data of physical entities in various work scenarios. This module relies on a sensor network deployed at the work site, including high-precision positioning sensors, inertial measurement units (IMUs), and condition monitoring equipment. The positioning sensors employ multi-source fusion positioning technology, combining GPS, BeiDou, and other satellite positioning signals with base station positioning data to provide centimeter-level position information. The IMU records motion parameters such as acceleration and angular velocity of the entities, while the condition monitoring equipment collects operational status and working mode data. The acquisition frequency is dynamically adjusted according to the entity's motion characteristics; millisecond-level acquisition intervals are used for high-speed moving entities, while the acquisition frequency is appropriately reduced for low-speed or stationary entities to save storage space. All acquired data is accompanied by precise timestamps and spatial coordinate information, forming a trajectory dataset containing multi-dimensional attributes such as time-series position, velocity, acceleration, and direction of motion.

[0066] The state space construction module performs tensor space reconstruction processing on the trajectory data, converting the original time-series trajectory data into a high-order tensor representation. The order of the tensor is equal to the number of feature parameters, which include dynamic features such as the rate of change of position, velocity, acceleration, and direction of motion. The reconstruction algorithm employs a sliding time window mechanism, segmenting the continuous time series and calculating the feature statistics within each time window. For the rate of change of position feature, the ratio of displacement to time within the window is calculated; for the rate of change of velocity feature, the differential of the velocity vector is calculated; and for the rate of change of acceleration feature, the gradient of the acceleration vector is calculated. Each feature parameter is standardized to eliminate dimensional differences and ensure that the features are numerically comparable.

[0067] A multi-dimensional state space is constructed based on feature parameters, with coordinate axes corresponding to the rates of change of various trajectory dynamic features. Each coordinate axis represents a feature change dimension, and the coordinate values ​​indicate the magnitude and direction of the feature's change at any given time. Each point in the space corresponds to the entity's motion state at a certain moment, and the change in the point's position reflects the evolution of the entity's motion state. The number of dimensions in the state space is determined according to actual needs, typically containing 6-12 main feature dimensions to ensure the completeness of the state description while avoiding the curse of dimensionality. Within the multi-dimensional state space, the system annotates the feature coordinates of historical abnormal events. These historical abnormal events come from recorded anomalies such as safety accidents, equipment failures, and operational errors during previous operations. Each abnormal event corresponds to one or more feature coordinate points, which are determined by analyzing the rate of change of trajectory features at the time the abnormality occurred. The annotation process uses an event-feature association algorithm to establish a mapping relationship between abnormal events and their corresponding state space coordinates, forming a set of feature coordinates for historical abnormal events.

[0068] This algorithm calculates the spatial clustering density of feature coordinates using a density clustering method. It employs a grid-based density estimation approach, dividing the state space into several grid cells and counting the number of feature coordinate points within each cell. For each grid cell, it calculates its density connectivity with neighboring cells, identifying the set of density-reachable grid cells. The density value of each grid cell is then calculated using a kernel density estimation method, and a Gaussian kernel function is used for density smoothing to obtain a continuous spatial density distribution. The density value reflects the degree of clustering of anomalous events in the state space; higher density indicates more frequent occurrences of anomalous events in that region.

[0069] Anomaly warning domains are generated based on areas where spatial density values ​​exceed a preset threshold. The threshold is set based on the statistical distribution characteristics of historical data, typically taking the upper quartile of the density distribution or adjusted according to specific application requirements. Warning domains are generated using a region growing algorithm, starting with grid cells exceeding the threshold and gradually merging adjacent high-density cells to form connected warning areas. Each warning domain records its spatial extent, peak density, and anomaly event type, providing a basis for subsequent real-time monitoring and early warning.

[0070] The entire implementation process adopts a distributed computing architecture. The real-time trajectory acquisition module is deployed on edge computing nodes to achieve local data processing and rapid response; the state space construction module runs on a cloud computing platform, providing powerful computing capabilities and storage resources. The two modules communicate via a high-speed data bus to ensure the real-time and reliable transmission of data. The system also provides a visual monitoring interface that displays the state space construction process and the distribution of warning domains in real time, supporting users to intuitively understand and interactively control the system's operating status.

[0071] Example 3: See Figure 4 The system monitors the actual motion state of physical entities in a multi-dimensional state space in real time and performs multi-dimensional comparative analysis with the predicted state to form a comprehensive deviation dataset reflecting the quality of the prediction. The operation of the prediction deviation analysis module begins with the real-time acquisition of actual position coordinates. This process is achieved by mapping the real-time acquired trajectory data of physical entities to a pre-constructed multi-dimensional state space. The system continuously receives real-time trajectory data from the sensor network, including parameters such as timestamps, spatial coordinates, motion speed, and orientation angles. After preprocessing and feature extraction, these data are used to calculate their corresponding coordinates in the state space. The coordinate calculation employs a nearest neighbor matching algorithm, calculating the similarity between the real-time feature vector and the state space basis vectors to determine the most matching coordinate position. Each coordinate point is accompanied by a confidence assessment, reflecting the reliability of the matching result.

[0072] The temporal cumulative offset analysis employs a sliding time window for dynamic warping and alignment. The system sets a variable-length time window, the size of which is dynamically adjusted according to the entity's motion characteristics. For entities with stable motion, a longer time window is used to capture long-term trends; for entities with variable motion, a shorter time window is used to capture rapid changes. The warping and alignment process uses a dynamic time warping algorithm, which can eliminate the nonlinear differences between the actual trajectory and the expected trajectory on the time scale. The algorithm finds the optimal warping path, aligns the two time series on the time axis, and calculates the cumulative offset after alignment. The offset calculation considers the combined effects of positional deviation, velocity deviation, and orientation deviation, forming an overall deviation metric in the time domain.

[0073] Spatial distribution similarity analysis employs a manifold distance algorithm for curvature matching. This algorithm treats the actual and expected coordinate distributions as low-dimensional manifolds embedded in a high-dimensional space. The similarity is assessed by calculating the geodesic distance between the two manifolds. Curvature matching focuses on analyzing the local and global bending characteristics of the manifolds, comparing the geometric consistency of the two distributions. The calculation process includes three steps: manifold reconstruction, curvature estimation, and distance measurement. Manifold reconstruction is achieved through a local linear embedding method, mapping high-dimensional data to a low-dimensional manifold; curvature estimation uses a principal curvature calculation method to characterize the local bending degree of the manifold; and the distance measurement combines geodesic distance and curvature difference to form a comprehensive spatial similarity index.

[0074] The dynamic event sequence matching analysis employs spatiotemporal entropy analysis for phase synchronization assessment. The system abstracts key events occurring during the simulation process (such as task initiation, resource interaction, and state transitions) into event sequences. Each event includes information such as a timestamp, spatial location, and event type. Spatiotemporal entropy analysis calculates the degree of disorder in the event sequence across time and space, reflecting the regularity of event distribution through entropy values. Phase synchronization assessment uses a cross-recursive graph method to analyze the synchronization degree between the actual and expected event sequences in terms of timing and order. Evaluation indicators include parameters such as phase lag, synchronization strength, and coupling direction, comprehensively characterizing the matching of the two event sequences across the spatiotemporal dimensions.

[0075] The results of the three-dimensional difference analysis are integrated using the following formula:

[0076]

[0077] in: Indicates the overall deviation. This represents the number of samples analyzed. This represents the actual state coordinates of the i-th sample. This represents the expected state coordinates of the i-th sample. Represents manifold distance, Represents the spacetime entropy value. , , These represent the weighting coefficients for temporal offset, spatial similarity, and event matching, respectively. The weighting coefficients are allocated based on the importance of the specific application scenario, and their sum is 1.

[0078] The output is a 3D extrapolation deviation dataset containing the deviation dimension. This dataset is stored in a hierarchical structure, comprising a raw data layer, a feature extraction layer, and a comprehensive index layer. The raw data layer records the actual collected trajectory data and the expected extrapolation data; the feature extraction layer stores the feature parameters and intermediate calculation results for each dimension; the comprehensive index layer contains the final deviation indexes for three dimensions: temporal cumulative offset, spatial distribution similarity, and dynamic event sequence matching degree. Each index is accompanied by quality assessment information and a timestamp, supporting subsequent traceability analysis and visualization.

[0079] The dataset generation process employs a pipelined architecture, with each analysis stage processed in parallel and data exchange and status synchronization achieved through message queues. The system incorporates multiple quality control points to validate intermediate results, and an automatic recalculation mechanism is triggered for abnormal data. The final output dataset uses a standardized format and supports both real-time push and batch export methods to meet the needs of different application scenarios.

[0080] Example 4: See Figure 5 The system processes 3D inference deviation data to optimize resource scheduling strategies and triggers a recalibration process for the inference cycle when significant deviations are detected. The entire process is based on the analysis results of a graph neural network, enabling dynamic updates to the 3D collaborative work space model and intelligent decision-making for cross-scenario resource scheduling. The collaborative optimization module first receives a 3D inference deviation dataset from the inference deviation analysis module. This dataset contains deviation indices in three dimensions: temporal cumulative offset, spatial distribution similarity, and dynamic event sequence matching degree. The module preprocesses the input data, including data cleaning, format conversion, and feature extraction. The cleaning process removes outliers and missing data, the format conversion adjusts the data to an input format acceptable to the graph neural network, and feature extraction extracts key feature parameters from the original data as network input.

[0081] The preprocessed 3D projection deviation dataset is input into a graph neural network for analysis. The graph neural network is constructed based on the structural features of a 3D collaborative work space model, where nodes represent various work scenarios and resource entities, and edges represent the relationships between scenarios and resource flow paths. The network employs a message passing mechanism, where nodes transmit deviation information and update their own states through edges. The network structure includes an input layer, multiple hidden layers, and an output layer. The hidden layers perform feature transformation and information aggregation, while the output layer generates the updated weights for each node. The training process uses the backpropagation algorithm to optimize network parameters by minimizing the prediction error.

[0082] The weights of resource flow paths in the 3D collaborative work space model are updated based on the output nodes of a graph neural network. The weight update process considers the importance of nodes, connection strength, and the degree of deviation impact. Higher update magnitudes are assigned to critical nodes and connections with high deviation impact, while lower update magnitudes are used for secondary nodes and connections with low deviation impact. The updated weights reflect the actual efficiency and reliability of each resource flow path under the current working environment (see Table 1).

[0083] Table 1: Graph Neural Network Training Parameter Configuration Table.

[0084]

[0085] Cross-scenario resource scheduling instructions are generated based on the updated resource flow path weights. The instruction generation process employs an optimization algorithm, comprehensively considering path weights, resource requirements, time constraints, and cost factors. The algorithm outputs scheduling instructions containing detailed information such as resource allocation schemes, path selection suggestions, and execution time schedules. The instruction format uses a standardized template to ensure compatibility with downstream systems. Instruction transmission is achieved through a message queue, ensuring the reliability and timeliness of instruction delivery. When the accumulated offset in the time domain exceeds the tolerance threshold, the system initiates a recalibration process. The tolerance threshold is set based on historical data and job requirements, typically taking the upper limit of the normal offset distribution. The detection process uses a sliding window statistical method to continuously monitor the offset change trend within multiple time windows. Once it is confirmed that the offset continuously exceeds the threshold, the system immediately sends a periodic recalibration request to the inference strategy generation module. The request includes a detailed deviation analysis report, impact assessment results, and recalibration recommendation parameters.

[0086] The system triggers a dynamic simulation cycle calculation module to recalculate the initial simulation cycle. The recalculation process considers current deviations, changes in operational status, and adjustments to environmental factors. The system first collects the latest spatiotemporal constraint parameters and resource status information, then uses a dynamic time-slicing algorithm to recalculate the simulation cycle. The newly calculated cycle is typically shorter than the original cycle to accommodate higher accuracy requirements and more complex operational environments. After recalculation, the system automatically updates the simulation strategy to ensure the simulation process remains consistent with the current situation. The system monitors the optimization effect in real time, evaluating the effectiveness of the optimization strategy by comparing performance indicators before and after optimization. Monitoring indicators include key performance parameters such as resource utilization, task completion time, and energy efficiency. Based on monitoring results, the system dynamically adjusts optimization parameters and strategies to achieve continuous performance improvement. The system provides a comprehensive management interface, allowing operators to view optimization process details, adjust optimization parameters, and intervene in optimization decisions. The interface displays real-time deviation data, optimization progress, and effect indicators, and supports historical data queries and trend analysis. All operation records and system status are logged in detail, supporting post-event auditing and problem troubleshooting.

[0087] Example 5: The scene fusion modeling module continuously receives resource scheduling instructions from the collaborative optimization module via an asynchronous messaging mechanism. Instruction transmission adopts a publish-subscribe model, with the module acting as a subscriber listening to message topics. When a new instruction is published, the receiving and processing flow is immediately triggered. The receiving process includes two steps: integrity verification and decoding / parsing. Integrity verification ensures that instruction data is not damaged or lost during transmission through checksum verification; instructions that fail verification are required to be retransmitted. Decoding / parsing converts binary instruction data into internally processable structured data, extracting key information such as instruction type, priority, and execution parameters. Instruction priorities are divided into four levels: urgent, high, medium, and low, with different processing strategies applied to instructions of different levels.

[0088] Based on the received resource scheduling instructions, the module dynamically updates the resource flow paths between scenes in the 3D collaborative work space model. Update operations include various types such as path weight adjustment, path addition / deletion, and path attribute modification. Path weight adjustment updates the weight values ​​of existing paths according to the weight coefficients in the instructions, employing a smooth transition algorithm to avoid system oscillations caused by sudden weight changes. Path addition / deletion operations add new resource flow paths or remove invalid paths according to the instructions. Adding a path requires initializing its weight attributes and connectivity relationships, while removing a path requires handling related dependencies. Path attribute modification updates auxiliary attributes such as path capacity limits, time constraints, and cost coefficients. All update operations employ a transaction processing mechanism to ensure atomicity and consistency, automatically rolling back to the previous stable state in the event of an error during the update process.

[0089] Updating spatial coordinate mapping is a crucial step in the implementation process. The module recalculates the spatial coordinate mapping between scenes based on changes in resource flow paths. The calculation process considers factors such as the relative positions of scenes, the geometric characteristics of paths, and the spatial distribution of resources, employing a multidimensional scaling analysis method to map high-dimensional spatial relationships to a three-dimensional coordinate system. The updated mapping maintains the topological connectivity between scenes while optimizing spatial layout to improve resource flow efficiency. For large-scale scene sets, an incremental update algorithm is used, recalculating only the mapping relationships of the affected parts to reduce computational overhead.

[0090] The updated spatial coordinate mapping is synchronized to the inference strategy generation module. The synchronization process employs a version control mechanism, assigning a unique version identifier to each update and recording information such as update time, update content, and operator. Data transmission uses a differential synchronization method, transmitting only the changed data to reduce network bandwidth consumption. The synchronization protocol ensures reliable data transmission, employing an acknowledgment and retransmission mechanism to ensure the receiver correctly receives all updated data. Upon receiving the updated data, the inference strategy generation module immediately updates its internal strategy calculation parameters, ensuring that subsequent inferences are based on the latest spatial model.

[0091] The entire implementation process establishes a comprehensive monitoring and logging mechanism. Modules monitor performance metrics for command reception, model updates, and data synchronization in real time, including processing latency, throughput, and error rate. Monitoring data is used for performance optimization and fault diagnosis, automatically triggering an early warning mechanism when performance degradation or an increase in errors is detected. All operations are logged in detail, including command content, update operations, synchronization status, and other detailed information, supporting post-event auditing and problem tracing. Log data is stored in a structured format for easy querying and analysis.

[0092] The system provides a management interface for monitoring and intervening in the implementation process. Through this interface, users can view the current model status, update history, synchronization status, and other information, and can also manually trigger model updates or data synchronization operations. The interface supports parameter adjustment, allowing modification of configuration parameters such as update strategies and synchronization frequencies to adapt to different application scenarios and performance requirements. Access control mechanisms ensure that only authorized personnel can operate the system, guaranteeing system security. Asynchronous messaging ensures reliable command transmission, transaction processing ensures consistent model updates, difference synchronization reduces data transmission volume, and version control supports status traceability. These features enable the system to adapt to complex operating environments and large-scale scene configurations, providing accurate and reliable spatial model support for 3D simulation and deduction.

[0093] 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 process, method, article, or apparatus.

[0094] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A three-dimensional simulation and intelligent scheduling system for multi-scenario collaborative operations, characterized in that, include: The scene fusion modeling module is used to generate a three-dimensional collaborative work space model based on user-configured multi-scene parameters. The three-dimensional collaborative work space model includes the spatial coordinate mapping relationship of different work scenes and the resource flow path between scenes. The dynamic simulation cycle calculation module is used to obtain the remaining time from the current time to the preset simulation end time, and to calculate the initial simulation cycle based on the remaining time by calling the first dynamic mapping relationship. The resource dimension analysis module is used to identify the number of resource types participating in collaborative operations, and to generate a periodic adjustment factor based on the number of resource types by calling the second dynamic mapping relationship. The simulation strategy generation module is used to perform weighted fusion processing on the initial simulation cycle and the cycle adjustment factor to generate an adaptive simulation cycle, and formulate a three-dimensional simulation simulation strategy based on the adaptive simulation cycle. The dynamic deduction period calculation module is specifically used for: Extract the spatiotemporal constraint parameters of each operation scenario in the three-dimensional collaborative operation space model; The remaining duration and the spatiotemporal constraint parameters are input into a dynamic time slicing algorithm to generate the initial simulation cycle that is negatively correlated with the complexity of the task. The resource dimension analysis module is specifically used for: The types of interface protocols for heterogeneous resources within the stated resource types are statistically analyzed. Based on the aforementioned interface protocol type, a third dynamic mapping relationship is invoked to generate a resource complexity coefficient; The number of resource types is convolved with the resource complexity coefficient to output the period adjustment factor; The real-time trajectory acquisition module is used to synchronously acquire the running trajectory data of physical entities in various work scenarios; The state space construction module is used to reconstruct the running trajectory data in tensor space to generate a multidimensional state space with a dimension equal to the number of feature parameters. The coordinate axes of the multidimensional state space correspond to the rate of change of the dynamic features of the trajectory. The deduction deviation analysis module is used during the execution of the three-dimensional simulation deduction strategy: Real-time capture of the actual position coordinates of physical entities in the multi-dimensional state space; A three-dimensional difference analysis was performed between the actual location coordinates and the predicted coordinates, considering the cumulative offset in the time domain, the similarity of spatial distribution, and the matching degree of dynamic event sequences. Output a 3D inference deviation dataset containing the deviation dimension; The collaborative optimization module is used to: input the three-dimensional inference deviation dataset into a graph neural network; The resource flow path weights in the three-dimensional collaborative operation space model are updated based on the output nodes of the graph neural network. Generate cross-scenario resource scheduling instructions based on the updated resource flow path weights.

2. The three-dimensional simulation and intelligent scheduling system for multi-scenario collaborative operations according to claim 1, characterized in that, The state space construction module is also used for: Mark the characteristic coordinates of historical anomalous events in the multidimensional state space; The spatial clustering density value of the feature coordinates is calculated based on the density clustering algorithm. An abnormal state warning domain is generated based on the area where the spatial aggregation density value exceeds a preset threshold.

3. The three-dimensional simulation and intelligent scheduling system for multi-scenario collaborative operations according to claim 1, characterized in that, The deduction deviation analysis module is specifically used for: The cumulative offset in the time domain is dynamically normalized and aligned using a sliding time window. The curvature matching calculation is performed using the manifold distance algorithm to assess spatial distribution similarity. Phase synchronization is evaluated using spatiotemporal entropy analysis on dynamic event sequences.

4. The three-dimensional simulation and intelligent scheduling system for multi-scenario collaborative operations according to claim 1, characterized in that, The collaborative optimization module is also used for: When the time-domain cumulative offset of a specific job scenario exceeds the tolerance threshold, Send a periodic recalibration request to the inference strategy generation module; The dynamic simulation cycle calculation module is triggered to recalculate the initial simulation cycle.

5. The three-dimensional simulation and intelligent scheduling system for multi-scenario collaborative operations according to any one of claims 1-4, characterized in that, The scene fusion modeling module is also used for: Continuously receive resource scheduling instructions sent by the collaborative optimization module; The resource flow paths between scenes in the three-dimensional collaborative operation space model are dynamically updated according to the resource scheduling instructions. The updated spatial coordinate mapping relationship is synchronized to the inference strategy generation module.

Images