A multi-dimensional equipment simulation system

Abstract

This invention discloses a multi-dimensional equipment simulation system, relating to the field of wargaming technology. It receives simulation control commands from a visualization terminal, parses the simulation scenario data and target equipment objects associated with the commands, generates engine-driven commands, and sends them to the main simulation engine to initiate the simulation process. In response to the start of the simulation process, it captures macroscopic state event objects generated by the main simulation engine and selects key equipment nodes to be simulated from these objects according to a preset simulation dimension list. Based on the key equipment nodes to be simulated, it calls the corresponding dynamic simulation model or electromagnetic environment simulation model from the extended model library for parallel calculation according to the node type, generating microscopic correction data for the key equipment nodes to be simulated. Finally, it performs spatiotemporal registration and fusion of the macroscopic state event objects and the microscopic correction data to generate a lightweight three-dimensional situational data stream, which is then fed back to the visualization terminal.

Description

Technical Field

[0001] This invention relates to the field of wargaming technology, and in particular to an equipment simulation system that integrates multiple dimensions. Background Technology

[0002] Equipment situation simulation technology calculates and recreates the dynamic evolution of a specific scenario through multi-entity interactive simulation under preset operating rules and environmental constraints, in order to assist in the performance evaluation and decision analysis of complex systems. With the rapid development of computer hardware and software architecture and digital twin technology, modern simulation systems are accelerating their evolution towards multi-dimensional physical-level fine computing, lightweight cross-platform deployment, and three-dimensional full-view visualization.

[0003] Currently, Chinese invention patent application number 202511679343.X discloses a hierarchical multi-agent combat experiment strategy generation method and system. By decomposing the combat problem into a strategic game layer and a tactical execution layer, and utilizing an Actor-Learner distributed architecture and reinforcement learning algorithms, it effectively solves the problems of slow convergence and poor stability in strategy generation caused by the high dimensionality of the state space in large-scale simulations, thus improving the efficiency of combat decision-making. However, the aforementioned technologies, due to their excessive focus on algorithm optimization at the command and decision-making level, often neglect the physical realism of the underlying execution environment and the extensibility of the system architecture. In practical applications, they have the following shortcomings: First, the equipment simulation dimension is limited, and the underlying physical realism is insufficient. Existing intelligent simulation systems often use simple two-dimensional discrete grids and probabilistic decision models when modeling the environment, lacking continuous simulation of the equipment's microscopic physical characteristics. For example, they cannot reflect the high-precision six-degree-of-freedom aerodynamic trajectory of equipment under complex weather conditions in real time, nor do they possess high-fidelity calculations of electromagnetic wave propagation loss under complex terrain. This results in the upper-level AI decision-making being highly intelligent, but relying on underlying physical feedback that deviates from actual physical rules. First, the existing system architecture is rigid and lacks the ability to cascade heterogeneous models and visualize them. Traditional macro-level wargaming software often adopts a closed client / server architecture, making cross-platform deployment difficult. When it is necessary to introduce external high-precision dynamic or electromagnetic environment models, the significant differences in data structure and time step between the macro-level campaign simulation engine and the micro-level physical simulation model can easily lead to data spatiotemporal asynchrony and simulation logic conflicts. In addition, the traditional architecture is unable to support the efficient streaming of massive amounts of three-dimensional situational data to a lightweight web client, resulting in the visualization of the battlefield situation remaining at a low-dimensional stage. Summary of the Invention

[0004] The technical problem solved by this invention is that existing wargaming software has a rigid architecture, lacks cross-platform capabilities, and has limited equipment simulation dimensions. It is difficult to achieve lightweight deployment across operating systems and real-time interaction on the Web. Moreover, the kernel is mostly based on discrete probabilistic decision models, which makes it difficult to integrate continuous high-precision dynamic and complex electromagnetic environment data in real time during the simulation process. As a result, the simulation results lack physical support and cannot meet the practical needs of modern warfare for multi-dimensional, cross-platform, and high-fidelity simulations.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a multi-dimensional equipment simulation system, comprising a deduction instruction parsing module, a status data interception module, a multi-dimensional difference resolution module, and a spatiotemporal data fusion module. The simulation command parsing module is used to receive simulation control commands sent by the visualization terminal, parse the simulation scene data and target equipment objects associated with the simulation control commands, generate engine drive commands and send them to the main simulation engine to start the simulation process. The state data interception module is used to respond to the start of the simulation process by capturing macro-state event objects generated by the main simulation engine and selecting key equipment nodes to be simulated from the macro-state event objects according to the preset simulation dimension list. The multidimensional differential calculation module is used to call the corresponding dynamic simulation model or electromagnetic environment simulation model from the extended model library according to the node type of the key equipment node to be simulated, and perform parallel calculation to generate micro-correction data of the key equipment node to be simulated. The spatiotemporal data fusion module is used to perform spatiotemporal registration and fusion of macroscopic state event objects and microscopic correction data to generate a lightweight three-dimensional situation data stream and feed it back to the visualization terminal.

[0006] Preferably, the deduction instruction parsing module includes an instruction protocol conversion unit, a scene resource mapping unit, and a cross-platform process scheduling unit; The processing logic of the instruction protocol conversion unit includes: Establish a full-duplex WebSocket communication channel with the visualization terminal to receive JSON-formatted deduction control commands; The simulation control commands are subjected to integrity verification and deserialization parsing to extract simulation scenario data and simulation control action types from the simulation control commands. The simulation scenario data includes simulation scenario identifiers and simulation time step parameters. The simulation control action types include loading scenario, starting simulation, pausing simulation, single-step execution, and terminating simulation. The processing logic of the scene resource mapping unit includes: The scenario database on the server is retrieved based on the scenario identifier to locate the storage path of the target scenario file; The target simulation scenario file is parsed, the file structure tree is traversed, the unique identity IDs of all participating equipment entities are extracted, and an index mapping table is established between the unique identity IDs and the memory object handles inside the main simulation engine. Based on the index mapping table, the target equipment object in the simulation control command is converted into an entity handle that can be recognized by the main simulation engine. According to the simulation control action type, simulation time step parameters and entity handle, the command is reorganized according to the underlying command format requirements preset by the main simulation engine to generate the converted command sequence.

[0007] Preferably, the processing logic of the cross-platform process scheduling unit includes: Detect the type of the current server's host operating system, which includes either Windows or Linux. Depending on the type of operating system, the corresponding system-level API is called to construct the startup command-line parameters for the main inference engine; When the system is detected as a Windows operating system, the CreateProcess instruction is encapsulated and the executable file of the main inference engine is loaded. When the operating system is detected as Linux, the Wine compatibility layer environment or Docker container runtime environment is loaded, and the instance process of the main inference engine is started through the fork / exec mechanism. Initialize the standard input / output pipes of the main derivation engine and write the converted instruction sequence into the standard input pipe.

[0008] Preferably, the state data interception module includes a memory probe unit, a state semantic parsing unit, and a dimension matching and filtering unit; The processing logic of the memory probe unit includes: By using the operating system's dynamic link library injection technology, monitoring probes are injected into the running main inference engine process; Based on the memory layout feature code of the main simulation engine, locate the base address pointer of the global object list and calculate the offset of the participating equipment entity in the memory address space of the main simulation engine. At a preset interception frequency, the memory block pointed to by the base address pointer plus the offset is read within a preset period to obtain a memory snapshot of the original binary data of all participating equipment entities.

[0009] Preferably, the processing logic of the state semantic parsing unit includes: Load the underlying data structure definition template of the main simulation engine, which includes the memory byte alignment and attribute field length of the participating equipment entity; The raw binary data in the memory snapshot is mapped to the underlying data structure definition template, and the macroscopic state fields of each participating equipment entity are de-parsed. The macroscopic state fields obtained by de-parse are encapsulated into standardized macroscopic state event objects, with the timestamp of the current frame attached. The macroscopic state fields include equipment type identifier, spatial coordinates, attitude angle, velocity vector, current command, and running status flag.

[0010] Preferably, the processing logic of the dimension matching filtering unit includes: Read the preset simulation dimension list, which includes preset equipment types that need to be externally simulated with high precision and the simulation model categories associated with the preset equipment types; Traverse the macro-state event objects, extract the equipment type identifiers of the participating equipment entities in the macro-state event objects, and compare the equipment type identifiers with the preset simulation dimension list; If the equipment type identifier matches the preset simulation dimension list, the participating equipment entity will be marked as a key equipment node to be simulated, and a dynamic simulation label or an electromagnetic simulation label will be applied to the key equipment node to be simulated according to the associated simulation model category. If the equipment type identifier does not match the preset simulation dimension list, the missing participating equipment entities will be marked as ordinary nodes, and only the macroscopic status field of the ordinary nodes will be retained.

[0011] Preferably, the multidimensional differential solution module includes a model instance scheduling unit, a dynamic solution engine unit, and an electromagnetic environment calculation unit; The processing logic of the model instance scheduling unit includes: Receive the key equipment node to be simulated, and read the dynamic simulation tag or electromagnetic simulation tag; based on the equipment type identifier of the key equipment node to be simulated, instantiate the corresponding physical simulation object from the extended model library; The spatial coordinates, attitude angles, and velocity vectors in the macroscopic state field are injected as initial boundary conditions into the physical simulation object to initialize the simulation environment and align the state.

[0012] Preferably, the processing logic of the dynamics calculation engine unit includes: In response to the dynamics simulation tag, a six-degree-of-freedom kinematic model is invoked to establish a set of differential equations for the key equipment nodes to be simulated; The simulation time step parameter in the simulation scenario data is used as the macroscopic simulation step. Within the macroscopic simulation step of the main simulation engine, the differential equation system is subjected to microscopic iterative calculation. The duration of a single iteration of the microscopic iterative calculation is used as the microscopic iteration step. The high-precision trajectory and attitude change of the key equipment node to be simulated under stress are solved, and the precise spatial coordinates and precise attitude angles are output. The precise spatial coordinates and precise attitude angles are used as dynamic correction data. The processing logic of the electromagnetic environment calculation unit includes: In response to the electromagnetic simulation tag, a digital elevation model and meteorological environment parameters are loaded. Based on the spatial coordinates and radar parameters of the key equipment nodes to be simulated, the signal strength, clutter background and interference signal-to-noise ratio of electromagnetic waves in the preset environment are calculated using ray tracing algorithm or by using the algorithm. The accessibility and effectiveness of electromagnetic detection are determined, and electromagnetic decision results are generated. The electromagnetic decision results are used as electromagnetic correction data. The electromagnetic decision results include detection probability and false alarm rate. The dynamic correction data and electromagnetic correction data are used as microscopic correction data.

[0013] Preferably, the spatiotemporal data fusion module includes a time-series synchronization and alignment unit, a multi-source state merging unit, and a visual streaming inference unit; The processing logic of the timing synchronization alignment unit includes: Establish a virtual simulation clock and obtain the macro-timestamps and micro-correction data of macro-state event objects in two adjacent frames generated by the main inference engine. Calculate the ratio of macroscopic simulation step size to microscopic iteration step size. Based on the ratio and the Lagrange interpolation method, fill the time interval between two adjacent macroscopic timestamps with the microscopic correction data sequence using a preset time scale. For the integer alignment point of the macro timestamp, the corresponding micro correction data slice is extracted and strictly aligned. For the non-integer time points within the time interval, the filled micro correction data is used as a high-frequency interpolation frame to generate effective correction data. The processing logic of the multi-source state merging unit includes: Create a standard battlefield situation frame and write the macroscopic status field of the ordinary node into the standard battlefield situation frame; For the key equipment node to be simulated, a state overwrite operation is performed, replacing the spatial coordinates and attitude angles in the macroscopic state event object with the precise spatial coordinates and precise attitude angles in the effective correction data. The electromagnetic adjudication result is dynamically attached as an additional attribute field to the corresponding key equipment node to be simulated, forming a fused situational object. The fused situational object is then written into a standard battlefield situational frame. The fused situational object includes physical correction information and electromagnetic environment information.

[0014] Preferably, the processing logic of the visual streaming inference unit includes: The macroscopic state fields of ordinary nodes and the fused situational object in the standard battlefield situational frame are used as battlefield situational data. A spatial octree indexing algorithm is employed to spatially partition the battlefield situational data, specifically including: Calculate the extreme values ​​of spatial coordinates of all participating equipment entities in the battlefield situation data, and construct a global axis-aligned bounding box covering the entire battlefield; Using the global axis-aligned bounding box as the root node, recursively bisection the data along the three dimensions of length, width, and height to generate eight subspace regions. Determine the number of participating equipment entities falling into each subspace region. If the number of participating equipment entities is greater than the preset splitting threshold, continue to perform recursive cutting operations on the subspace region. If the number of participating equipment entities is less than or equal to the preset splitting threshold, the splitting is stopped, the subspace region is defined as a spatial index leaf node, the battlefield situation data is discretized and stored in the corresponding spatial index leaf node, and the spatial division of the battlefield situation data is completed. In response to the viewpoint request from the visualization terminal, a view frustum culling operation is performed, retaining only the effective situational data within the view frustum range, and serializing the retained effective situational data into a lightweight 3D situational data stream; The lightweight 3D situational data stream is pushed to the visualization terminal for rendering and display via a WebSocket full-duplex communication channel.

[0015] The beneficial effects of this invention are as follows: Considering the difficulty of acquiring data from closed-source engines, a monitoring probe is injected into the main inference engine using dynamic link library injection technology. Based on the base address pointer superimposed with offset, a memory snapshot is captured and the macroscopic state event object is decrypted. To overcome the physical distortion of probabilistic judgment, the system accurately selects key equipment nodes to be simulated according to the simulation dimension list, calls a six-degree-of-freedom kinematic model and ray tracing algorithm to perform microscopic iterative calculations, and generates multi-dimensional microscopic correction data. Facing the time barrier of macro-micro simulation, the system uses Lagrange interpolation to map to the macroscopic time axis. After extracting and aligning slices, the original coarse state is forcibly replaced by state overwrite operation, bridging the spatiotemporal misalignment gap of heterogeneous model cascade. The massive data is serialized into a lightweight three-dimensional situational data stream using spatial octree indexing algorithm and view frustum culling operation. Efficient three-dimensional display is achieved on the cross-platform web terminal via WebSocket, perfectly realizing the closed loop of macroscopic inference logic and high-fidelity physical simulation. Attached Figure Description

[0016] Figure 1 This is a basic flowchart of a multi-dimensional equipment simulation system provided as an embodiment of the present invention. Detailed Implementation

[0017] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0018] Example, refer to Figure 1 This paper presents a multi-dimensional equipment simulation system, including a deduction instruction parsing module, a status data interception module, a multi-dimensional difference resolution module, and a spatiotemporal data fusion module. The simulation command parsing module is used to receive simulation control commands sent by the visualization terminal, parse the simulation scenario data and target equipment objects associated with the simulation control commands, generate engine drive commands, and send them to the main simulation engine to start the simulation process.

[0019] The state data interception module is used to respond to the start of the simulation process by capturing macro-state event objects generated by the main simulation engine and selecting key equipment nodes to be simulated from the macro-state event objects according to the preset simulation dimension list.

[0020] The multidimensional differential calculation module is used to call the corresponding dynamic simulation model or electromagnetic environment simulation model from the extended model library according to the node type of the key equipment node to be simulated, and perform parallel calculation to generate micro-correction data of the key equipment node to be simulated.

[0021] The spatiotemporal data fusion module is used to perform spatiotemporal registration and fusion of macroscopic state event objects and microscopic correction data to generate a lightweight three-dimensional situational data stream and feed it back to the visualization terminal.

[0022] This invention injects monitoring probes into the main inference engine using dynamic link library injection technology. Based on the base address pointer superimposed with offset, it captures memory snapshots and reverse-parses macroscopic state event objects. To overcome the physical distortion of probabilistic judgment, the system accurately selects key equipment nodes to be simulated according to the simulation dimension list, calls a six-degree-of-freedom kinematic model and ray tracing algorithm for microscopic iterative calculations, and generates multi-dimensional microscopic correction data. Facing the time barrier of macro-micro simulation, the system uses Lagrange interpolation to map to the macroscopic time axis. After extracting and aligning slices, it forcibly replaces the original coarse state through state overwriting operations, bridging the spatiotemporal misalignment gap of heterogeneous model cascades. It uses spatial octree indexing algorithm and view frustum culling operation to serialize massive data into a lightweight three-dimensional situational data stream, and realizes efficient three-dimensional display on the cross-platform web terminal through WebSocket, perfectly realizing the closed loop of macroscopic inference logic and high-fidelity physical simulation.

[0023] In a specific embodiment, the deduction instruction parsing module includes an instruction protocol conversion unit, a scene resource mapping unit, and a cross-platform process scheduling unit.

[0024] The processing logic of the instruction protocol conversion unit includes: Establish a full-duplex WebSocket communication channel with the visualization terminal to receive JSON-formatted deduction control commands.

[0025] The simulation control commands are subjected to integrity verification and deserialization parsing to extract simulation scenario data and simulation control action types. The simulation scenario data includes simulation scenario identifiers and simulation time step parameters. The simulation control action types include loading scenario, starting simulation, pausing simulation, single-step execution, and terminating simulation.

[0026] The processing logic of the scene resource mapping unit includes: The scenario database on the server is retrieved based on the scenario identifier to locate the storage path of the target scenario file.

[0027] The target simulation scenario file is parsed, the file structure tree is traversed, the unique identity IDs of all participating equipment entities are extracted, and an index mapping table is established between the unique identity IDs and the memory object handles inside the main simulation engine.

[0028] Based on the index mapping table, the target equipment objects in the simulation control commands are converted into entity handles that can be recognized by the main simulation engine. According to the simulation control action type, simulation time step parameters and entity handles, the commands are reorganized according to the underlying command format requirements preset by the main simulation engine to generate the converted command sequence.

[0029] The processing logic of the cross-platform process scheduling unit includes: Detect the type of the current server's operating system, which can be either Windows or Linux.

[0030] Depending on the type of operating system, the corresponding system-level API is called to construct the startup command-line parameters for the main inference engine.

[0031] When the system is detected as a Windows operating system, the CreateProcess instruction is encapsulated and the executable file of the main inference engine is loaded.

[0032] When the operating system is detected as Linux, the Wine compatibility layer environment or Docker container runtime environment is loaded, and the instance process of the main inference engine is started through the fork / exec mechanism.

[0033] Initialize the standard input / output pipes of the main inference engine and write the converted instruction sequence into the standard input pipe to drive the inference process.

[0034] This invention addresses the pain points of traditional closed-source inference engines, which are deeply bound to a single operating system and have a closed underlying instruction format, making it impossible for web terminals to directly drive and deploy across platforms. Considering the need to achieve seamless client / server architecture integration without modifying the main inference engine source code, an inference instruction parsing module is designed to reduce instruction complexity and isolate the environment. First, it establishes a WebSocket full-duplex communication channel to receive inference control instructions in JSON format from the front end. Then, it cleverly utilizes a scene resource mapping unit to extract the terminal's unique ID and establish an index mapping table with the memory object handles within the main inference engine. This accurately translates and reassembles the web-side objects into a converted instruction sequence that conforms to the engine's underlying instruction format requirements. This addresses the limitations of operating systems... To overcome operational barriers, the idea was to dynamically adapt through a cross-platform process scheduling unit. When Windows is detected, the CreateProcess instruction is directly encapsulated. Under Linux, the Wine compatibility layer environment or Docker container runtime environment is innovatively loaded and the process is started using the fork / exec mechanism. Finally, the simulation is driven by writing instructions through the initialization of standard input / output pipes. With completely non-intrusive instruction bridging and environment virtualization, the limitations of the host system of traditional wargaming software are broken. This allows the originally rigid standalone simulation engine to be deployed as a microservice on a cloud Linux cluster in a containerized and elastic manner, and is perfectly controlled by the real-time driving interaction of the front-end web across terminals.

[0035] In a specific embodiment, the state data interception module includes a memory probe unit, a state semantic parsing unit, and a dimension matching and filtering unit.

[0036] The processing logic of the memory probe unit includes: By utilizing the operating system's dynamic link library injection technology, monitoring probes are injected into the running main inference engine process.

[0037] Based on the memory layout signature of the main simulation engine, locate the base address pointer of the global object list and calculate the offset of the participating equipment entity in the memory address space of the main simulation engine.

[0038] At a preset interception frequency, the memory block pointed to by the base address pointer plus the offset is read within a preset period to obtain a memory snapshot of the original binary data of all participating equipment entities.

[0039] In a preferred embodiment of the present invention, for the state data interception module, in order to obtain the underlying situational data of the main inference engine (such as closed-source or semi-closed-source wargaming software such as Command) in a non-intrusive and low-latency manner in the absence of an official API interface, the memory probe unit in this embodiment adopts dynamic link library injection and memory signature scanning technology. The specific implementation logic and operation steps include: During the static analysis phase, because the memory base address allocated by the main simulation engine process changes randomly and dynamically each time it runs, it is affected by the operating system's ASLR (Address Space Layout Randomization) security mechanism. Therefore, data cannot be read through hard-coded absolute addresses. It is necessary to first extract fixed memory layout signatures during the system development or adaptation phase. Specifically, using disassemblers (such as IDA Pro) and dynamic debugging tools (such as Cheat Engine), the binary executable file of the main simulation engine is reverse-engineered. By tracing the memory allocation operations when the simulation scene is loaded, the initialization assembly code segment that manages the global object linked list or contiguous object array that manages all participating equipment entities can be located.

[0040] Extract globally unique machine code instructions from the initialization assembly code snippet as memory layout signatures. The purpose of these memory layout signatures is to provide an absolutely accurate anchor point for runtime memory addressing.

[0041] During the dynamic operation phase, after the simulation process is started by the cross-platform process scheduling unit, the memory probe unit uses the operating system's low-level APIs, such as the CreateRemoteThread and WriteProcessMemory functions under Windows, or the ptrace mechanism under Linux, to covertly inject pre-written monitoring probes into the process space of the main simulation engine. The monitoring probes are DLL files or shared object files.

[0042] After successful probe injection, a separate high-priority background thread is started in the memory address space of the main inference engine to execute the pattern matching algorithm. The pattern matching algorithm traverses the code segment of the module where the main inference engine is located, ignoring wildcard bytes and strictly matching the memory layout feature code to find the absolute memory address where the memory layout feature code is located.

[0043] During the precise addressing phase, after locating the address of the memory layout signature, the probe parses the absolute address of the global pointer according to the relative addressing logic of the assembly instructions. First, it calculates the global base address pointer. The addressing logic of the global base address pointer is as follows: BasePointer = ModuleBaseAddress + StaticOffset; Where BasePointer is the global base address pointer, ModuleBaseAddress is the module base address, which is the starting address where the main inference engine executable file is loaded into memory, and StaticOffset is the static offset, which is the fixed offset length from the module base address parsed through the memory layout feature code.

[0044] For hundreds or thousands of specific participating equipment entities on the battlefield, the starting memory address of each target object is calculated by traversing the array or linked list pointed to by BasePointer. The mathematical expression for the starting memory address is: EntityAddress[i] = BasePointer + (i * EntityStructSize) +DynamicOffset; Where EntityAddress[i] is the starting address in memory, i is the index number of the participating equipment entity in the list, EntityStructSize is the fixed number of bytes occupied in memory by each participating equipment entity object as determined by the underlying data structure definition template, and DynamicOffset is the dynamic offset compensation amount that may exist when handling multi-level pointer jumps.

[0045] After establishing the accurate memory address EntityAddress[i] of all participating equipment entities, the memory probe unit periodically reads the memory blocks pointed to by these pointers by calling efficient memory copy instructions (such as memcpy) at a preset interception frequency, which is consistent with the macro rendering frame rate of the main simulation engine.

[0046] This process is completed in-place within the memory space of the main inference engine, without causing the main thread to be blocked or the inference to lag. Finally, the extracted data is assembled into a memory snapshot containing the original binary data of all participating equipment entities, and the memory snapshot is securely transmitted to the downstream state semantic parsing unit for reverse parsing through the inter-process communication mechanism.

[0047] The processing logic of the state semantic parsing unit includes: Load the underlying data structure definition template of the main simulation engine. The underlying data structure definition template includes the memory byte alignment and attribute field length of the participating equipment entities.

[0048] The raw binary data in the memory snapshot is mapped to the underlying data structure definition template, and the macroscopic status fields of each participating equipment entity are de-parsed. The macroscopic status fields obtained by de-parse are encapsulated into standardized macroscopic status event objects, with the timestamp of the current frame attached. The macroscopic status fields include equipment type identifier, spatial coordinates, attitude angle, velocity vector, current command, and running status flag.

[0049] In a preferred embodiment of the present invention, for the state semantic parsing unit in the state data interception module, in order to accurately convert the acquired unreadable underlying binary data into structured data that the system can process, this embodiment discloses in detail the underlying data structure definition template and the specific processing logic of anti-parsing: Since memory snapshots are presented as continuous, unstructured raw binary byte streams, strict memory mapping rules must be established. This embodiment pre-constructs a structure with memory byte alignment completely consistent with the main simulation engine as the underlying data structure definition template. Specifically, for each participating equipment entity, the memory address offset and data type definition of each attribute field in the underlying data structure definition template are as follows: Unique ID: Occupies a starting offset of 0x00, has a length of 4 bytes, and is a 32-bit integer (Int32).

[0050] Equipment type identifier: Occupies offset 0x04, length is 4 bytes, data type is 32-bit integer (Int32).

[0051] Spatial coordinates: Occupy offsets from 0x08 to 0x1F, containing longitude, latitude, and altitude (i.e., geodetic coordinates) based on the WGS84 coordinate system. Each dimension is 8 bytes long and the data type is double-precision floating point.

[0052] Attitude angles: occupy offsets 0x20 to 0x2B, including pitch, yaw and roll angles, each dimension is 4 bytes long, and the data type is single-precision floating point.

[0053] Running status flag: Occupies offset 0x2C, length is 1 byte, data type is character, used to identify the current command or survival status of the participating equipment entity.

[0054] After receiving the memory snapshot from the memory probe unit, the state semantic parsing unit performs cyclic slicing of the byte stream in the memory snapshot according to the total length of the underlying data structure definition template, separating the data blocks corresponding to each participating equipment entity. The total length includes memory-aligned padding bytes generated based on the operating system compilation rules. For each separated data block, the system uses pointer type casting technology or efficient memory copy functions (such as memcpy) to directly map and assign the continuous raw binary data to the instantiated structure variables of the underlying data structure definition template. Through memory-level mapping and assignment, the system successfully deparses the originally unreadable binary gibberish into the macroscopic state fields of each participating equipment entity, namely the aforementioned unique identity ID, equipment type identifier, spatial coordinates, attitude angle, and operating status flag.

[0055] The system encapsulates the macro-state fields obtained from reverse parsing into standardized macro-state event objects, along with the system timestamp of the current frame. These macro-state event objects are then output to downstream dimension matching and filtering units and multi-dimensional differential calculation modules, providing structured and high-precision real-time data support.

[0056] The processing logic for the dimension matching filtering unit includes: Read the preset simulation dimension list, which includes the preset equipment types that need to be simulated externally with high precision, and the simulation model categories associated with the preset equipment types.

[0057] Traverse the macro-state event objects, extract the equipment type identifiers of the participating equipment entities in the macro-state event objects, and compare the equipment type identifiers with the preset simulation dimension list.

[0058] If the equipment type identifier matches the preset simulation dimension list, the participating equipment entity will be marked as a key equipment node to be simulated, and a dynamic simulation label or an electromagnetic simulation label will be applied to the key equipment node to be simulated according to the associated simulation model category.

[0059] If the equipment type identifier does not match the preset simulation dimension list, the missing participating equipment entity will be marked as a normal node. Only the macroscopic status field of the normal node will be retained for basic display, and it will not enter the subsequent multidimensional difference resolution process.

[0060] This invention addresses the pain points of closed-source main inference engines lacking open data interfaces, leading to a black box effect on underlying states, and the high risk of system overload due to external high-precision calculations on all equipment. Considering the need to break down data barriers and achieve on-demand allocation of computing power without intrusion, a state data interception module is designed. The approach involves using the operating system's dynamic link library injection technology to covertly inject monitoring probes into the running process. By retrieving memory layout signatures, the base address pointer is precisely located and offset is superimposed. At a preset interception frequency, an unauthorized snapshot of the memory containing the original binary data is captured. Faced with unreadable underlying garbled code, a cleverly loaded underlying data structure definition template containing memory byte alignment and attribute field lengths is used for strict mapping, precise de-parsing, and encapsulation. As a standardized macroscopic state event object with the current frame timestamp, in order to block invalid computing power consumption from the source, the equipment type identifier is compared and traversed according to the preset simulation dimension list. The entities that are matched are upgraded and marked as key equipment nodes to be simulated and attached with dynamic simulation tags or electromagnetic simulation tags to trigger subsequent external high-precision calculations. The entities that are not matched are directly classified as ordinary nodes and only the macroscopic state fields of ordinary nodes are retained. Under the strict condition of not touching the source code of the inference engine, a closed loop from heterogeneous memory reading to semantic structured reorganization is opened up. Through precise tag-based diversion and filtering, the valuable external simulation computing power is strictly focused on high-value nodes, perfectly resolving the irreconcilable contradiction between multidimensional high-fidelity simulation and global real-time computing.

[0061] In a specific embodiment, the multidimensional differential solution module includes a model instance scheduling unit, a dynamic solution engine unit, and an electromagnetic environment calculation unit.

[0062] The processing logic of the model instance scheduling unit includes: Receive the key equipment node to be simulated and read its dynamic simulation tag or electromagnetic simulation tag. Based on the equipment type identifier of the key equipment node to be simulated, instantiate the corresponding physical simulation object from the extended model library.

[0063] The spatial coordinates, attitude angles, and velocity vectors in the macroscopic state field are injected as initial boundary conditions into the physical simulation object to initialize the simulation environment and align the state.

[0064] The processing logic of the dynamics solution engine unit includes: In response to the dynamics simulation tag, a six-degree-of-freedom kinematic model is invoked to establish a set of differential equations for the key equipment nodes to be simulated.

[0065] The simulation time step parameter in the simulation scenario data is used as the macroscopic simulation step. Within the macroscopic simulation step of the main simulation engine, the differential equation system is subjected to microscopic iterative calculation. The duration of a single iteration of the microscopic iterative calculation is used as the microscopic iteration step. The high-precision trajectory and attitude change of the key equipment node to be simulated under stress are solved, and the precise spatial coordinates and precise attitude angles are output. The precise spatial coordinates and precise attitude angles are used as dynamic correction data.

[0066] The processing logic of the electromagnetic environment calculation unit includes: In response to the electromagnetic simulation tag, a digital elevation model and meteorological environmental parameters are loaded. Based on the spatial coordinates and radar parameters of the key equipment nodes to be simulated, the reachability and effectiveness of electromagnetic detection are determined by using ray tracing algorithms or calculating the signal strength, clutter background and interference signal-to-noise ratio of electromagnetic waves in a preset environment. Electromagnetic decision results are generated and used as electromagnetic correction data. The electromagnetic decision results include the detection probability and false alarm rate.

[0067] Dynamic correction data and electromagnetic correction data are used as microscopic correction data.

[0068] This invention addresses the pain points of traditional wargaming systems, which rely excessively on coarse discrete probability adjudication models, leading to distorted equipment spatial motion trajectories and radar detection results that severely deviate from realistic complex terrain and weather patterns. Considering the need to seamlessly integrate high-fidelity continuous physical processes within the macro-level simulation framework, a multi-dimensional differential calculation module is designed. This module utilizes a model instance scheduling unit to read pre-processed triage tags, precisely injecting the extracted coarse spatial coordinates, attitude angles, and velocity vectors as initial boundary conditions into the physical simulation object, thus completing the pre-calculation state alignment. Addressing the issue that macro-level discrete commands cannot reflect the true maneuverability of equipment, a clever solution engine unit calls a six-degree-of-freedom kinematic model to establish a system of differential equations within the main simulation engine. Within the macroscopic simulation step, high-frequency microscopic iterative calculations are performed to solve for the precise spatial coordinates and attitude angles of the equipment under stress. In order to completely abandon the probabilistic dice-rolling detection detached from the three-dimensional environment, the electromagnetic environment calculation unit is used to deeply load the digital elevation model and meteorological environment parameters. With the help of the ray tracing algorithm, the signal strength, clutter background and interference signal-to-noise ratio in the real environment are rigorously calculated to generate electromagnetic adjudication results that include objective detection probability and false alarm rate. Not only are the dynamic correction data and electromagnetic correction data used as microscopic correction data without disrupting the main simulation clock progression logic, but the previously rigid and distorted macroscopic wargaming simulation has for the first time gained a solid physical foundation supported by the microscopic six-degree-of-freedom continuous dynamics and complex electromagnetic terrain attenuation model.

[0069] In a specific embodiment, the spatiotemporal data fusion module includes a time-series synchronization and alignment unit, a multi-source state merging unit, and a visual streaming inference unit.

[0070] The processing logic of the timing synchronization alignment unit includes: Establish a virtual simulation clock and obtain the macro-timestamps and micro-correction data of macro-state event objects in two adjacent frames generated by the main inference engine.

[0071] Calculate the ratio of macroscopic simulation step size to microscopic iteration step size. Based on the ratio and Lagrange interpolation, fill the time interval between two adjacent macroscopic timestamps with a preset time scale using the microscopic correction data sequence.

[0072] For the integer alignment points of the macro timestamp, the corresponding micro-correction data slices are extracted for strict state alignment. For non-integer time points within the time interval, the filled micro-correction data is used as high-frequency interpolation frames to generate effective correction data with continuous time, so as to ensure the smoothness and physical coherence of the front-end 3D situation display.

[0073] In a preferred embodiment of the present invention, for the timing synchronization alignment unit, since there is a significant difference between the macroscopic simulation step size of the main inference engine and the microscopic iteration step size of the dynamic model, the macroscopic simulation step size is set to... seconds, micro-iteration step size set to For each second, strict timing registration is required. The specific calculation process is as follows: First, the step size ratio is calculated, which is the time ratio of the macroscopic simulation step size to the microscopic iteration step size. For example, a ratio of 100 means that within the period of one macroscopic frame, the microscopic dynamics model generates 100 high-frequency trajectory point data.

[0074] Secondly, high-frequency rendering mapping operations based on Lagrange interpolation are performed. Due to the inherent time frequency asynchrony between the dynamic rendering frame rate of the front-end visualization terminal (e.g., 60 FPS) and the micro-iteration step size of the back-end dynamic model, the target rendering timestamp requested by the front-end, i.e., the current rendering moment... These data points often do not strictly correspond to a specific integer time index in the micro-corrected data sequence. Therefore, it is necessary to extract three micro-data points adjacent to the macro-time stamp. , and ,in This represents the precise spatial coordinates or precise attitude angles in the micro-correction data. The precise spatial coordinates are the local Cartesian coordinates after projection transformation, calculated using the second-order Lagrange interpolation formula at the target rendering timestamp. Precise state The mathematical expression for second-order Lagrange interpolation is: ; By applying the second-order Lagrange interpolation formula, a tightly aligned high-frequency interpolation rendering frame can be dynamically calculated from the discrete micro-corrected data sequence at the front-end rendering time. This effectively corrects the data slices, decoupling the underlying macroscopic simulation step size from the front-end visualization rendering frame rate. This not only solves the problem of two heterogeneous clocks being out of sync, but also fundamentally eliminates the problems of visual position jitter and trajectory smoothness loss caused by network latency and asynchronous sampling.

[0075] The processing logic of the multi-source state merging unit includes: Create a standard battlefield situation frame and write the macroscopic status field of ordinary nodes into the standard battlefield situation frame.

[0076] For the key equipment nodes to be simulated, a state overwrite operation is performed, replacing the spatial coordinates and attitude angles in the macroscopic state event objects with the precise spatial coordinates and precise attitude angles in the effectively corrected data.

[0077] The electromagnetic adjudication results are dynamically attached as additional attribute fields to the corresponding key equipment nodes to be simulated, forming a fused situational awareness object. This fused situational awareness object is then written into a standard battlefield situational awareness frame. The fused situational awareness object includes physical correction information and electromagnetic environment information. Spatial registration and attribute fusion of multi-source heterogeneous data are completed.

[0078] The processing logic of the visual streaming simulation unit includes: The macroscopic state fields of ordinary nodes and the fused situational object in the standard battlefield situational frame are used as battlefield situational data. A spatial octree indexing algorithm is employed to spatially partition the battlefield situational data, specifically including: Calculate the extreme values ​​of spatial coordinates of all participating equipment entities in the battlefield situation data, and construct a global axis-aligned bounding box covering the entire battlefield.

[0079] Using the global axis-aligned bounding box as the root node, recursively divide it into eight subspace regions according to the three dimensions of length, width, and height.

[0080] Determine the number of participating equipment entities falling into each subspace region. If the number of participating equipment entities is greater than the preset splitting threshold, continue to perform recursive cutting operations on the subspace region.

[0081] If the number of participating equipment entities is less than or equal to the preset splitting threshold, the splitting stops, the subspace region is defined as a spatial index leaf node, the battlefield situation data is discretized and stored in the corresponding spatial index leaf node, and the spatial division of the battlefield situation data is completed.

[0082] In response to the viewpoint request from the visualization terminal, a view frustum culling operation is performed, retaining only the valid situational data within the view frustum range. The retained valid situational data is then serialized into a lightweight 3D situational data stream conforming to the 3D Tiles standard or the GeoJSON standard.

[0083] The lightweight 3D situational data stream is pushed to the visualization terminal for rendering and display via a WebSocket full-duplex communication channel.

[0084] In a preferred embodiment of the present invention, for the visual streaming simulation unit, in order to achieve smooth rendering of massive 3D battlefield situation data on a lightweight web platform, an optimization process combining spatial octree partitioning and view frustum culling is adopted, specifically including: The constructed octree node data structure includes: the bounding box coordinate range of the current node, an array of child node pointers of length 8, and a list of participating equipment entities contained within the current node. During spatial partitioning, a preset splitting threshold and a maximum tree depth are pre-set. The preset splitting threshold is set to... Maximum tree depth is set to When the number of participating equipment entities falling into a certain subspace region is greater than the preset splitting threshold and the current depth is less than the maximum tree depth, the three-dimensional recursive binary cutting is continued according to the midpoint of the length, width and height of the corresponding subspace region. If the above conditions are not met, the cutting operation is stopped and the corresponding subspace region is defined as a spatial index leaf node to store the corresponding data.

[0085] The mathematical criterion for frustum culling is that the frustum of the front-end camera of the visualization terminal consists of six clipping planes: the near plane, far plane, left plane, right plane, top plane, and bottom plane. In three-dimensional space, the equation for each clipping plane is defined as follows: Furthermore, the normal vector of each clipping plane is set to point towards the interior of the view frustum. For one of the spatial index leaf nodes generated by the octree partitioning, the 3D center point of the axis-aligned bounding box (AABB) of the spatial index leaf node is calculated. and the half-side length of each axis of the axially aligned bounding box The axis-aligned bounding box is normal to the clipping plane. Projected radius on The mathematical expression is: ; center point Directed distance to the cutting plane The mathematical expression is: ; The criteria for performing elimination and retention are as follows: traverse the six clipping planes of the view frustum; if for one of the clipping planes, the directed distance and projected radius satisfy... If the spatial index leaf node and all battlefield situation data contained in the spatial index leaf node are completely outside the view frustum, a culling operation will be performed immediately, and the culled data will not participate in the subsequent 3D Tiles streaming and serialization transmission.

[0086] If the following condition holds true for all cutting planes If so, then the spatial index leaf node data is retained.

[0087] By implementing rejection and retention criteria, invalid rendering data can be filtered accurately and efficiently, greatly reducing the burden on network communication bandwidth and the pressure on web-based graphics rendering.

[0088] This invention addresses the issues of trajectory misalignment caused by inconsistent time steps between macro and micro engines, and the rendering stuttering and bandwidth overload that easily occur when massive amounts of 3D situational data are directly pushed to the web. Considering the need to decouple the simulation clock and rendering frame rate at the underlying level and achieve on-demand spatial data loading, a spatiotemporal data fusion module was designed. This involves establishing a virtual simulation clock to obtain the preceding and following macroscopic timestamps of two adjacent frames and calculating the step ratio. Lagrange interpolation is then used to accurately fill the time interval between two adjacent macroscopic timestamps with microscopic correction data. High-frequency interpolation frames are extracted for non-integer time points to generate effective correction data. To eliminate heterogeneous logic conflicts, a state overwrite operation is forcibly performed on the key equipment nodes to be simulated in the created standard battlefield situational frame, replacing... The original coarse coordinates are dynamically attached to electromagnetic adjudication results to form a fused situational object. Faced with the transmission bottleneck of massive fused data, the idea is to use a spatial octree indexing algorithm to construct a global axis-aligned bounding box and recursively cut and discretize it to spatial index leaf nodes according to a threshold. Then, in close coordination with the visualization terminal request to perform a view frustum culling operation, the invisible data is removed and serialized into a lightweight 3D situational data stream. Pushed via WebSocket, it crosses the spatiotemporal misalignment gap of heterogeneous model cascading. While fundamentally eliminating trajectory shifts and visual jitter caused by discrete probability inference, the spatial indexing and view frustum culling mechanism greatly reduces network communication bandwidth and front-end graphics rendering pressure, perfectly realizing high-fidelity physical situational display on cross-platform Web terminals with high-frequency smooth streaming.

[0089] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0090] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the protection scope of the present invention.

Claims

1. A multi-dimensional equipment simulation system, characterized in that, It includes a deduction instruction parsing module, a state data interception module, a multi-dimensional differential calculation module, and a spatiotemporal data fusion module: The simulation command parsing module is used to receive simulation control commands sent by the visualization terminal, parse the simulation scene data and target equipment objects associated with the simulation control commands, generate engine drive commands and send them to the main simulation engine to start the simulation process. The state data interception module is used to respond to the start of the simulation process by capturing macro-state event objects generated by the main simulation engine and selecting key equipment nodes to be simulated from the macro-state event objects according to the preset simulation dimension list. The multidimensional differential calculation module is used to call the corresponding dynamic simulation model or electromagnetic environment simulation model from the extended model library according to the node type of the key equipment node to be simulated, and perform parallel calculation to generate micro-correction data of the key equipment node to be simulated. The spatiotemporal data fusion module is used to perform spatiotemporal registration and fusion of macroscopic state event objects and microscopic correction data to generate a lightweight three-dimensional situation data stream and feed it back to the visualization terminal.

2. The equipment simulation system integrating multiple dimensions as described in claim 1, characterized in that, The simulation instruction parsing module includes an instruction protocol conversion unit, a scene resource mapping unit, and a cross-platform process scheduling unit; The processing logic of the instruction protocol conversion unit includes: Establish a full-duplex WebSocket communication channel with the visualization terminal to receive JSON-formatted deduction control commands; The simulation control commands are subjected to integrity verification and deserialization parsing to extract simulation scenario data and simulation control action types from the simulation control commands. The simulation scenario data includes simulation scenario identifiers and simulation time step parameters. The simulation control action types include loading scenario, starting simulation, pausing simulation, single-step execution, and terminating simulation. The processing logic of the scene resource mapping unit includes: The scenario database on the server is retrieved based on the scenario identifier to locate the storage path of the target scenario file; The target simulation scenario file is parsed, the file structure tree is traversed, the unique identity IDs of all participating equipment entities are extracted, and an index mapping table is established between the unique identity IDs and the memory object handles inside the main simulation engine. Based on the index mapping table, the target equipment object in the simulation control command is converted into an entity handle that can be recognized by the main simulation engine. According to the simulation control action type, simulation time step parameters and entity handle, the command is reorganized according to the underlying command format requirements preset by the main simulation engine to generate the converted command sequence.

3. The equipment simulation system integrating multiple dimensions as described in claim 2, characterized in that, The processing logic of the cross-platform process scheduling unit includes: Detect the type of the current server's host operating system, which includes either Windows or Linux. Depending on the type of operating system, the corresponding system-level API is called to construct the startup command-line parameters for the main inference engine; When the system is detected as a Windows operating system, the CreateProcess instruction is encapsulated and the executable file of the main inference engine is loaded. When the operating system is detected as Linux, the Wine compatibility layer environment or Docker container runtime environment is loaded, and the instance process of the main inference engine is started through the fork / exec mechanism. Initialize the standard input / output pipes of the main derivation engine and write the converted instruction sequence into the standard input pipe.

4. The equipment simulation system integrating multiple dimensions as described in claim 3, characterized in that, The state data interception module includes a memory probe unit, a state semantic parsing unit, and a dimension matching and filtering unit; The processing logic of the memory probe unit includes: By using the operating system's dynamic link library injection technology, monitoring probes are injected into the running main inference engine process; Based on the memory layout feature code of the main simulation engine, locate the base address pointer of the global object list and calculate the offset of the participating equipment entity in the memory address space of the main simulation engine. At a preset interception frequency, the memory block pointed to by the base address pointer plus the offset is read within a preset period to obtain a memory snapshot of the original binary data of all participating equipment entities.

5. The equipment simulation system integrating multiple dimensions as described in claim 4, characterized in that, The processing logic of the state semantic parsing unit includes: Load the underlying data structure definition template of the main simulation engine, which includes the memory byte alignment and attribute field length of the participating equipment entity; The raw binary data in the memory snapshot is mapped to the underlying data structure definition template, and the macroscopic state fields of each participating equipment entity are de-parsed. The macroscopic state fields obtained by de-parse are encapsulated into standardized macroscopic state event objects, with the timestamp of the current frame attached. The macroscopic state fields include equipment type identifier, spatial coordinates, attitude angle, velocity vector, current command, and running status flag.

6. The equipment simulation system integrating multiple dimensions as described in claim 5, characterized in that, The processing logic of the dimension matching filtering unit includes: Read the preset simulation dimension list, which includes preset equipment types that need to be externally simulated with high precision and the simulation model categories associated with the preset equipment types; Traverse the macro-state event objects, extract the equipment type identifiers of the participating equipment entities in the macro-state event objects, and compare the equipment type identifiers with the preset simulation dimension list; If the equipment type identifier matches the preset simulation dimension list, the participating equipment entity will be marked as a key equipment node to be simulated, and a dynamic simulation label or an electromagnetic simulation label will be applied to the key equipment node to be simulated according to the associated simulation model category. If the equipment type identifier does not match the preset simulation dimension list, the missing participating equipment entities will be marked as ordinary nodes, and only the macroscopic status field of the ordinary nodes will be retained.

7. The equipment simulation system integrating multiple dimensions as described in claim 6, characterized in that, The multidimensional differential solution module includes a model instance scheduling unit, a dynamic solution engine unit, and an electromagnetic environment calculation unit. The processing logic of the model instance scheduling unit includes: Receive the key equipment node to be simulated, and read the dynamic simulation tag or electromagnetic simulation tag; based on the equipment type identifier of the key equipment node to be simulated, instantiate the corresponding physical simulation object from the extended model library; The spatial coordinates, attitude angles, and velocity vectors in the macroscopic state field are injected as initial boundary conditions into the physical simulation object to initialize the simulation environment and align the state.

8. The equipment simulation system integrating multiple dimensions as described in claim 7, characterized in that, The processing logic of the dynamics solution engine unit includes: In response to the dynamics simulation tag, a six-degree-of-freedom kinematic model is invoked to establish a set of differential equations for the key equipment nodes to be simulated; The simulation time step parameter in the simulation scenario data is used as the macroscopic simulation step. Within the macroscopic simulation step of the main simulation engine, the differential equation system is subjected to microscopic iterative calculation. The duration of a single iteration of the microscopic iterative calculation is used as the microscopic iteration step. The high-precision trajectory and attitude change of the key equipment node to be simulated under stress are solved, and the precise spatial coordinates and precise attitude angles are output. The precise spatial coordinates and precise attitude angles are used as dynamic correction data. The processing logic of the electromagnetic environment calculation unit includes: In response to the electromagnetic simulation tag, a digital elevation model and meteorological environment parameters are loaded. Based on the spatial coordinates and radar parameters of the key equipment nodes to be simulated, the signal strength, clutter background and interference signal-to-noise ratio of electromagnetic waves in the preset environment are calculated using ray tracing algorithm or by using the algorithm. The accessibility and effectiveness of electromagnetic detection are determined, and electromagnetic decision results are generated. The electromagnetic decision results are used as electromagnetic correction data. The electromagnetic decision results include detection probability and false alarm rate. The dynamic correction data and electromagnetic correction data are used as microscopic correction data.

9. The equipment simulation system integrating multiple dimensions as described in claim 8, characterized in that, The spatiotemporal data fusion module includes a time-series synchronization and alignment unit, a multi-source state merging unit, and a visual streaming inference unit. The processing logic of the timing synchronization alignment unit includes: Establish a virtual simulation clock and obtain the macro-timestamps and micro-correction data of macro-state event objects in two adjacent frames generated by the main inference engine. Calculate the ratio of macroscopic simulation step size to microscopic iteration step size. Based on the ratio and the Lagrange interpolation method, fill the time interval between two adjacent macroscopic timestamps with the microscopic correction data sequence using a preset time scale. For the integer alignment point of the macro timestamp, the corresponding micro correction data slice is extracted and strictly aligned. For the non-integer time points within the time interval, the filled micro correction data is used as a high-frequency interpolation frame to generate effective correction data. The processing logic of the multi-source state merging unit includes: Create a standard battlefield situation frame and write the macroscopic status field of the ordinary node into the standard battlefield situation frame; For the key equipment node to be simulated, a state overwrite operation is performed, replacing the spatial coordinates and attitude angles in the macroscopic state event object with the precise spatial coordinates and precise attitude angles in the effective correction data. The electromagnetic adjudication result is dynamically attached as an additional attribute field to the corresponding key equipment node to be simulated, forming a fused situational object. The fused situational object is then written into a standard battlefield situational frame. The fused situational object includes physical correction information and electromagnetic environment information.

10. The equipment simulation system integrating multiple dimensions as described in claim 9, characterized in that, The processing logic of the visual streaming inference unit includes: The macroscopic state fields of ordinary nodes and the fused situational object in the standard battlefield situational frame are used as battlefield situational data. A spatial octree indexing algorithm is employed to spatially partition the battlefield situational data, specifically including: Calculate the extreme values ​​of spatial coordinates of all participating equipment entities in the battlefield situation data, and construct a global axis-aligned bounding box covering the entire battlefield; Using the global axis-aligned bounding box as the root node, recursively bisection the data along the three dimensions of length, width, and height to generate eight subspace regions. Determine the number of participating equipment entities falling into each subspace region. If the number of participating equipment entities is greater than the preset splitting threshold, continue to perform recursive cutting operations on the subspace region. If the number of participating equipment entities is less than or equal to the preset splitting threshold, the splitting is stopped, the subspace region is defined as a spatial index leaf node, the battlefield situation data is discretized and stored in the corresponding spatial index leaf node, and the spatial division of the battlefield situation data is completed. In response to the viewpoint request from the visualization terminal, a view frustum culling operation is performed, retaining only the effective situational data within the view frustum range, and serializing the retained effective situational data into a lightweight 3D situational data stream; The lightweight 3D situational data stream is pushed to the visualization terminal for rendering and display via a WebSocket full-duplex communication channel.

Images