Adaptive communication method and heterogeneous interconnected multi-core distributed simulation experiment system node communication architecture

Abstract

The application relates to the field of simulation technology, and particularly discloses a self-adaptive communication method and a heterogeneous interconnected multi-core distributed simulation experiment system node communication architecture, which comprises the following steps: S1, detecting a simulation entity state and judging an event triggering time; the simulation entity comprises a semi-physical simulation entity and a parallel simulation entity; S2, when the event communication is triggered, determining the simulation entities for mutual communication; and S3, performing data transmission between the determined simulation entities for mutual communication, and encoding and decoding the data. The application meets the distributed simulation demand of large-scale simulation entities, constructs an optimal communication mechanism for large-scale simulation entity scheduling management and a semi-physical model distributed simulation strategy, and further improves the virtual-real interconnection and cluster cluster control semi-physical simulation capability of a simulation experiment platform.

Description

Technical Field

[0001] This invention belongs to the field of simulation technology, and in particular relates to an adaptive communication method and a node communication architecture for a heterogeneous interconnected multi-core distributed simulation experimental system. Background Technology

[0002] The concept of adaptive swarm communication originated in the late 1990s. The US Air Force, inspired by simulated biological swarms, proposed the theoretical idea of ​​"swarm warfare," and in 2016, the US Air Force demonstrated its "swarm warfare" technical framework, emphasizing the interconnectivity of swarms. Early research focused primarily on fixed topologies for communication networks, such as star, tree, mesh, and forest topologies. Currently, it's difficult to objectively evaluate the rationality of task decomposition in distributed simulations. Jane Bachman proposed an intelligent grouping task decomposition method. This was followed by task decomposition strategies for balancing computational resource load, including equal-cost load balancing and non-equal-cost load balancing strategies. Traditional gateways use hard-coded "one-to-one" parameter conversion, which is simple in logic and clear in conversion relationships, but lacks flexibility and scalability. Domestic scholar Gao Yuan proposed a configuration-based "message-to-message" conversion method to achieve scalable adaptive message configuration, but its processing capacity is limited. To address the issue of low processing capacity, a distributed information interaction mechanism based on a shadow model was proposed. The shadow model synchronizes the behavior and state of the entity model but does not concern itself with the model's processing logic. Therefore, for the internal aspects of the simulation node, the shadow model adopts an "appearance" pattern, adhering to the model framework and interfaces required for simulation system modeling. For the internal aspects of the simulation node, the shadow model adopts a "proxy" pattern, subscribing to messages through a distributed communication message middleware. Currently, there is little research on how to achieve fast and effective adaptive communication topology in the highly dynamic, multi-tasking, and complex environment of a cluster, while reducing network traffic and improving cluster robustness. Furthermore, the understanding of the fundamental principles of evolutionary communication self-organization in swarm intelligence remains insufficient, and there is a significant gap between the multi-node distributed simulation process and actual conditions. Accurate reasoning about the swarm's combat situation and achieving effective and reliable communication decisions under actual constraints still face severe challenges, requiring the exploration of new swarm intelligence communication organizational architectures and methods.

[0003] The prior art, Chinese patent application CN201511001237.2, discloses a hardware-in-the-loop (HIL) simulation system. However, this HIL simulation system can only realize communication between HIL simulation nodes, and cannot solve the communication problem between HIL simulation nodes and HIL multi-core parallel simulation nodes in heterogeneous clusters, or between multi-core parallel simulation nodes in digital simulation systems. Summary of the Invention

[0004] The purpose of this invention is to provide an adaptive communication method and a node communication architecture for a heterogeneous interconnected multi-core distributed simulation experimental system, which partially solves or alleviates the above-mentioned shortcomings in the prior art and enables communication between virtual and real nodes.

[0005] To address the aforementioned technical problems, the present invention specifically adopts the following technical solution: A first aspect of the present invention provides an adaptive communication method applied to a heterogeneous interconnected multi-core distributed simulation experimental system. The method includes: S1 detecting the state of simulation entities and determining the timing of event triggering; the simulation entities include hardware-in-the-loop (HIL) simulation entities and parallel simulation entities; S2 determining the simulation entities communicating with each other when event communication is triggered; S3 transmitting data between the determined HIL simulation entities and encoding and decoding the data. Further, the steps for determining the event triggering time include: S11 calculating the state offset based on the current state estimate of the simulated entity and the system state estimate that met the event triggering conditions at the previous moment; and calculating the dynamic scaling factor using the state offset; S12 adjusting the automatic adjustment threshold parameter that met the event triggering conditions at the previous moment using the dynamic scaling factor and the upper and lower bounds of the automatic adjustment threshold parameter, thereby obtaining the automatic adjustment threshold parameter at the current moment; S13 adjusting the threshold using the automatic adjustment threshold parameter at the current moment, and taking the moment that meets the adjusted threshold requirement as the next moment that meets the event triggering conditions; S14 determining whether the next moment that meets the event triggering conditions belongs to Zeno behavior. If it does not belong to Zeno behavior, the simulated entity can communicate at the next moment that meets the event triggering conditions.

[0006] Furthermore, methods for calculating the dynamic scaling factor include using the formula:

[0007] , Calculate the dynamic scaling factor; where λ is the dynamic scaling factor, Δ is the normalized rate of change of state, ε is the offset correction, and α and For fuzzy rule strength coefficients, This is the estimated system state value that currently meets the event triggering conditions. This is the estimated state of the system that previously met the event triggering conditions;

[0008] Using the formula: Calculate the automatic adjustment threshold parameter at the current time; where, The threshold parameter is automatically adjusted at the current moment. Here, λ represents the automatically adjusted threshold parameter from the previous time step, and λ is the dynamically changing scaling factor. To automatically adjust the lower limit of the threshold parameter, To automatically adjust the upper limit of the threshold parameter;

[0009] Using the formula: Calculate the next time the event triggering condition is met; where t k+1 h represents the next time the event triggering condition is met, and t represents the time when the event triggering condition is met. k h is the current moment when the event triggering condition is met, h is the sampling period, Z is the number of sampling periods, and N is a natural number; For state estimation error, This is the estimated system state at the current moment. This represents the estimated system state that currently meets the event triggering conditions; T is the transpose symbol. The threshold parameter is automatically adjusted at the current moment. It is a positive definite weight matrix.

[0010] Further, the steps for determining the communicating simulation entities include: S21 constructing a communication mapping matrix and a communication topology diagram for each simulation entity; S22 using the communication mapping matrix and the communication topology diagram to calculate the final link point to determine the communicating simulation entities.

[0011] Further, the steps for constructing the communication mapping matrix and communication topology diagram for each simulation entity include: S211 selecting a neighboring node from the neighboring nodes of a simulation entity and calculating the angle formed by the simulation entity, the neighboring node, and other neighboring nodes; S212 comparing the angle with a preset Beta value and recording the number of times the angle is less than the Beta value; S213 establishing a communication link between the simulation entity and the neighboring node when the number of times the angle is less than the Beta value is equal to the number of neighboring nodes; S214 traversing all neighboring nodes of the simulation entity and repeating steps S211 to S213 to construct the communication mapping matrix and communication topology diagram of the simulation entity. Furthermore, the steps for calculating the final link point using the communication mapping matrix and the communication topology graph include: S221 calculating the second smallest eigenvalue of the Laplace matrix using the communication mapping matrix and the identity matrix; S222 forming eigenpairs by exponentially iterating the number of nodes in the communication topology graph with the second smallest eigenvalue of the Laplace matrix; S223 calculating the estimated value of the connectivity of the communication topology graph using the estimated value of the eigenvector corresponding to the second smallest eigenvalue in the eigenpair; and S224 increasing the number of connected nodes if the estimated value of the connectivity of the communication topology graph is greater than the number of nodes, thereby forming the final link point.

[0012] Furthermore, the estimated value of the connectivity of the communication topology graph is obtained using the formula: Calculate; where, L is an estimate of the connectivity. ij Let be the Laplace matrix formed by nodes i and j. Let be the element value of the Fiedler vector corresponding to the j-th node in the k-th iteration. For the element estimate of the Fiedler vector corresponding to the i-th node, N i Let i be the set of neighboring nodes of node i;

[0013] Using the formula: Calculate the element values ​​corresponding to the iterated Fiedler vector; where, Let be the element value of the Fiedler vector corresponding to the j-th node in the k-th iteration. This represents the element estimate of the Fiedler vector corresponding to the i-th node in the k-th iteration. The modulus of the element value of the Fiedler vector corresponding to the k-th iteration for each node;

[0014] Using the formula: Calculate the element-wise estimates of the Fiedler vector after the iteration; where, Let be the element value of the Fiedler vector corresponding to the i-th node in the k-th iteration. For each node, the element value W corresponding to the (k-1)th iteration of the Fiedler vector i Let be the coefficient matrix of the i-th node.

[0015] Furthermore, the update rule for the coefficient matrix W is as follows: ,

[0016] in, Let I be the coefficient matrix at the k-th iteration, L be the identity matrix, and J be a matrix of all ones. For a minimum positive value, k w Given a positive integer, n is the number of nodes, and k is the number of iterations.

[0017] This invention also provides a node communication architecture for a heterogeneous interconnected multi-core distributed simulation experimental system, including an ad hoc network data chain and an adaptive communication module set in a hardware-in-the-loop (HIL) simulation node, and an ad hoc network data chain simulation system set in a multi-core parallel simulation node. The ad hoc network data chain includes a data transmission module and an adaptive communication method deployed in an external task computer. The ad hoc network data chain simulation system is essentially a digital simulation of the ad hoc network data chain and the adaptive communication module. Communication between the HIL simulation entities simulated by the HIL simulation nodes, between the parallel simulation entities simulated by the multi-core parallel simulation nodes, and between HIL simulation entities and parallel simulation entities is achieved using the ad hoc network data chain and / or the ad hoc network data chain simulation system. Further, the HIL simulation nodes communicate using Ethernet; the multi-core parallel simulation nodes communicate using fiber optic reflected memory; the HIL simulation nodes and the multi-core parallel simulation nodes communicate using Ethernet; and the parallel simulation entities within the multi-core parallel simulation nodes communicate using shared memory.

[0018] Beneficial Effects: The heterogeneous interconnected multi-core distributed simulation experimental system provided by this invention effectively combines a dynamic inference visual simulation engine and environmental physical model, a virtual-real combined simulation experimental platform, a hardware-in-the-loop simulation system, and an RTX / WIN multi-core parallel simulation framework to form a general-purpose online UAV swarm simulation experimental platform. This platform enables UAV swarm hardware-in-the-loop simulation, digital hardware-in-the-loop simulation, and virtual-real combined swarm simulation experiments, as well as adaptive communication algorithms, swarm planning algorithms, and large-scale swarm scenario simulation verification driven by tasks. It can realize UAV mission-level scenario review experiments and swarm virtual-real combined joint simulation experiments. The heterogeneous interconnected multi-core distributed simulation experimental system possesses multi-core parallel simulation capabilities, which can overcome hardware limitations while ensuring data accuracy, achieving multi-core parallel simulation capabilities for hardware-in-the-loop simulation models and exponentially increasing the scale of swarm simulation. The experimental platform has an environmental model, which, combined with a physical simulation engine, can achieve high-fidelity simulation of environmental data in the constructed environment scenario. Using standardized data interfaces, it can realize hardware-in-the-loop simulation experiments of different adaptive communication algorithms, intelligent task planning algorithms, and path planning algorithms, reducing the R&D cost of intelligent planning and scheduling for UAV swarms, improving the verification capability of key equipment parameters, and reducing the technical difficulty of sample experiment review. The swarm simulation experimental platform is used to construct an integrated virtual-real experimental simulation environment, reducing the cost of equipment experimental research and development. It utilizes virtual and physics engines to build a near-realistic physical simulation environment and introduces physical parameter-related interferences into a semi-physical simulation model, thus establishing an integrated virtual-real simulation experimental platform. This enhances the computer's ability to simulate formal environments and shortens the experimental research and development cycle for swarm method verification. The swarm simulation experimental platform adopts standard data interfaces and can integrate different types of adaptive communication task planning algorithms and path planning algorithms. It can be used to verify different types of UAV control algorithms. The platform can perform retrospective analysis of key samples from UAV swarm simulation experiments, improving the reliability and realism of the simulation experiments. Simultaneously, it can perform task-level verification of large-scale swarm simulation experiments. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0020] Figure 1 The structural schematic diagram of a multi-core parallel heterogeneous interconnect simulation experimental platform; Figure 2 This is a schematic diagram of the structure of a multi-core simulation management system. Figure 3 This is a schematic diagram of the communication architecture of a multi-core parallel framework; Figure 4 This is a schematic diagram of a simulated data link system for an ad hoc network. Figure 5 This is a schematic diagram of a heterogeneous multi-node adaptive communication mechanism. Figure 6This is a schematic diagram illustrating whether the conditions for the Beta angle detection algorithm are met (the Beta angle detection algorithm is met). Figure 7 A diagram showing whether the conditions for the Beta angle detection algorithm are met (the conditions for the Beta angle detection algorithm are not met); Figure 8 Here is a flowchart of the cluster collaborative control method; Figure 9 Flowchart of the cluster task allocation algorithm; Figure 10 Here is a flowchart of the cluster formation control algorithm; Figure 11 This is a plane coordinate system for the northeast region. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. In this document, "multiple" means two or more, that is, it includes two, three, four, five, etc.

[0022] Example 1: This invention is based on the RTX / Window multi-core parallel simulation framework, and combines a semi-physical framework adaptive communication mechanism and a UAV cooperative control algorithm to construct a multi-core parallel heterogeneous interconnected multi-core distributed simulation experimental system, such as... Figure 1As shown, it specifically includes a hardware-in-the-loop (HIL) simulation system and a virtual system. Specifically, the HIL simulation system includes: several HIL simulation nodes, each capable of simulating half a physical entity; and several multi-core parallel simulation nodes, each capable of simulating several parallel HIL simulation entities. Specifically, the virtual system includes: a digital simulation system capable of simulating several parallel digital simulation entities; an experimental environment model for generating simulation environment modeling data; a virtual-real hybrid simulation experimental platform for rendering and extrapolating work scenarios; and a dynamic extrapolation visual simulation engine for rendering environmental information using simulation environment modeling data. The aforementioned HIL simulation entities, parallel HIL simulation entities, and parallel digital simulation entities are collectively referred to as simulation entities in the following text. Furthermore, for ease of explanation, the simulation entity in this embodiment uses a simulated aircraft (UAV) as an example. HIL simulation nodes typically connect to some real hardware devices, which may be key components of the UAV, such as flight control system hardware and sensor hardware. Taking the flight control system of a drone as an example, a real flight control computer may be used directly, utilizing its actual computing power and control logic to participate in the simulation. This makes the hardware-in-the-loop (HIL) simulation entity closer to the physical characteristics and hardware response of a real drone. Because it contains real hardware, the HIL can accurately simulate the actual working characteristics of the drone hardware, such as sensor measurement errors and flight control system response delays, providing high accuracy for verifying hardware-level algorithms and systems. Multi-core parallel simulation nodes primarily rely on the computing power of multi-core processors. Multi-core parallel simulation nodes utilize multiple processor cores to process computational tasks in parallel, fully leveraging the parallel computing performance of the hardware to improve simulation efficiency. For example, in large-scale drone swarm simulations, each core can handle a portion of the drone's computational tasks, such as dynamics calculations and sensor simulations. Through accurate mathematical models and efficient algorithms, parallel HIL can accurately simulate the dynamics and kinematics of drones. When handling large-scale drone swarm simulations, reasonable task allocation and parallel computing can ensure the computational accuracy and real-time performance of each drone model. Digital simulation systems are entirely based on mathematical models and software code to simulate various behaviors and characteristics of drones. By writing program code to describe the dynamic equations, sensor models, and communication models of the UAV, and utilizing computer computing resources for numerical calculations, the operational status of the UAV is presented in a virtual environment. Parallel digital simulation entities can flexibly adjust the simulation accuracy according to specific needs and the complexity of the model. By establishing detailed mathematical models and using advanced numerical calculation methods, it is possible to deeply simulate various behaviors of the UAV, but the realism of its simulation is lower compared to hardware-in-the-loop (HIL) simulation entities and parallel hardware-in-the-loop (PIL) simulation entities.

[0023] The heterogeneous interconnected multi-core distributed simulation experimental system provided by this invention effectively combines a dynamic inference visual simulation engine and environmental physical model, a virtual-real hybrid simulation experimental platform, a hardware-in-the-loop simulation system, and an RTX / WIN multi-core parallel simulation framework to form a general-purpose online UAV swarm simulation experimental platform. This platform enables UAV swarm hardware-in-the-loop simulation, digital hardware-in-the-loop simulation, and virtual-real hybrid swarm simulation experiments; it also allows for adaptive communication algorithms, swarm planning algorithms, and task-driven large-scale swarm scenario simulation verification. It can realize UAV mission-level scenario debriefing experiments and swarm virtual-real hybrid joint simulation experiments. The heterogeneous interconnected multi-core distributed simulation experimental system possesses multi-core parallel simulation capabilities, which can overcome hardware limitations while ensuring data accuracy, achieving multi-core parallel simulation capabilities for hardware-in-the-loop simulation models and exponentially increasing the scale of swarm simulations. The following provides a detailed description of each of the above components.

[0024] I. Hardware-in-the-loop simulation system. In this embodiment, the hardware-in-the-loop simulation system is a heterogeneous hardware-in-the-loop simulation system composed of multiple hardware-in-the-loop simulation nodes and multiple multi-core parallel simulation nodes. 1. The hardware-in-the-loop simulation nodes are mainly composed of hardware and software such as real-time simulator, flight control computer, external mission computer, communication module and hardware-in-the-loop simulation framework, which are used to build lightweight simulation nodes. External actuators such as external torque load simulator, intelligent IO device, full-quantization turntable, GPS simulator and inertial navigation simulator constitute full-quantization simulation nodes. (1) Real-time simulator (not shown in the figure) is used to build hardware-in-the-loop simulation entity model. (2) Flight control computer is used to perform motion control on the hardware-in-the-loop simulation entity built by real-time simulator, specifically for the flight control of six-degree-of-freedom aircraft. The flight control computer is composed of sensors, ARM processor and high-performance DSP processor, which mainly complete the functions of attitude information calculation, position information node, attitude control, position control and mission control. The high-performance DSP processor design features high control precision, strong wind resistance, and fast computing speed. The ARM processor has the ability to access data from various hardware peripheral interfaces. It can be used in combination with a high-performance inertial navigation sensor (IMU) and a high-precision GPS through external interfaces. The DSP processor performs fast calculations, enabling the semi-physical simulation node to have good control performance. (3) The external task computer includes an online task planning module for planning and allocating tasks for the simulated entity cluster to coordinate the control of the cluster, and an adaptive communication module for realizing communication between simulated entities. The external task computer is based on the NVIDIA series core processor and loads the cluster task allocation, trajectory planning, formation maintenance and switching algorithms into the internal core processor to realize the UAV cluster's autonomous decision-making, online task allocation and formation switching combat modes. The data communication module of the intelligent planner is mainly responsible for data interaction with hardware devices such as the flight control terminal and the self-organizing network data link, effectively ensuring the information transmission between each UAV and providing an effective data channel for the online dynamic planning and formation switching of the UAV cluster. (4) The self-organizing network data link is the physical communication link between semi-physical simulation entities. It has a high-performance wireless MESH data transmission module and provides multi-channel anti-interference, thereby achieving high-speed and high-bandwidth data transmission during long-distance communication. Through the self-organizing network data link, data interaction between various UAVs can be efficiently realized, completing the real-time data communication process within the UAV cluster and ensuring the effective online simulation verification of the virtual-physical hybrid system of the UAV cluster. It is worth noting that, due to hardware limitations, the self-organizing network data link is simulated by the self-organizing network data link simulation system using fiber optic reflection memory.2. The multi-core parallel simulation node is mainly composed of a real-time simulator, a multi-core parallel simulation framework, a self-organizing network data link simulation system, and an online task planning simulation system. It is used to build functional-level and functional-level lightweight hardware-in-the-loop agent simulation nodes. This node has the ability of multi-core simulation, which can further improve the upper limit of the hardware-in-the-loop simulation model access to the hardware-in-the-loop system. This enables the hardware-in-the-loop system to have the ability of cluster simulation. (1) Real-time simulator (not shown in the figure) is used to build parallel hardware-in-the-loop simulation entity models. The six-degree-of-freedom model of the fixed-wing UAV through hardware-in-the-loop simulation has functions such as flight control, attitude control and trajectory tracking. Multiple virtual aircraft devices can be created through this model to effectively simulate the functions of physical flight control devices. (2) The multi-core parallel simulation framework deployed in the RTX real-time system is used for the configuration of the parallel hardware-in-the-loop simulation entity test state, the control of the experimental process, and the storage and processing of experimental data. The RTX multi-core parallel simulation framework is an extension of the semi-physical integrated framework with multi-core parallel simulation capability for semi-physical models. In order to overcome the hardware limitations of the semi-physical simulation system and ensure the accuracy of the model, it is mainly used for functional level and functional level lightweight semi-physical model simulation, and provides real-time simulation results of key simulation parameters for the semi-physical system. (3) Online task planning simulation system, which is a digital simulation of the online task planning module, is used to simulate the task planning and allocation of the entity cluster for collaborative control of the cluster. The online task planning simulation system plays the same role as the online task planning module in the external task computer by constructing and loading the cluster task allocation, trajectory planning and formation control and formation change algorithm modules. (4) Self-organizing network data link simulation system, which is a digital simulation of the adaptive communication module and the self-organizing network data link, is used to realize communication between simulated entities. II. Virtual system. The virtual system consists of experimental environment model, virtual and real combined simulation experimental platform, digital simulation system and dynamic inference visual simulation engine. The virtual simulation system adopts a digital simulation mode, which can realize the same cluster simulation function as the physical system, and can freely expand the number of simulated UAVs. On the basis of saving experimental costs, it can realize large-scale UAV cluster virtual-physical combination online simulation verification. 1. The digital simulation system includes a self-organizing network data link simulation system, an online task planning simulation system, and a multi-core parallel simulation framework. (1) The multi-core parallel simulation framework deployed in the Windows system and the digital simulation model based on the multi-core parallel simulation framework are used for parallel digital simulation entity test state configuration, experimental process control, and storage and processing of test data. The digital simulation model in this embodiment is a fixed-wing UAV six-degree-of-freedom model, which has functions such as flight control, attitude control, and trajectory tracking. Multiple virtual aircraft devices can be created through this model to effectively simulate the functions of physical flight control devices.The Window multi-core parallel simulation framework is a model loading and integration framework for digital models in digital systems developed based on the Window system. It is mainly used for the simulation and operation of UAV digital models. It has the ability to run in multiple threads and can realize simulation calculations of single models and multiple models. It provides the solution results of key simulation parameters for digital systems. In the laboratory environment, both the hardware-in-the-loop simulation system and the digital simulation system can effectively simulate the actual flight process of UAV clusters. (2) Online task planning simulation system, which is a digital simulation of the online task planning module, is used to simulate the task planning and allocation of entity clusters for collaborative control of the cluster. The online task planning simulation system plays the same role as the online task planning module in the external task computer by constructing and loading algorithm modules such as cluster task allocation, trajectory planning and formation control and formation change. (3) Self-organizing network data link simulation system, which is a digital simulation of the adaptive communication module and the self-organizing network data link, is used to realize communication between simulated entities. It is the data interaction channel between each simulated UAV in the virtual simulation system. 2. Experimental environment model, used to simulate the generation of experimental environment modeling data. Specifically, it is used to simulate and generate experimental environment modeling data such as wind, rain, fog, and snow, and then connects the environmental data to the dynamic simulation engine for rendering and constructing multi-dimensional scene information. 3. A virtual-real combined simulation experimental platform is used for the simulation and rendering of working scenarios. Specifically, it is developed under the GUI framework platform based on UE5 and C++ programming, mainly including combat scenario simulation and rendering functions such as cluster control command issuance, task allocation, trajectory planning, scene construction, and data communication, effectively completing the 3D rendering and display of corresponding combat tasks. Through simulated trajectory and online data analysis, it completes the online real-time task allocation efficiency evaluation of cluster combat and the online evaluation of multi-task combat capabilities, efficiently completing the verification and evaluation of the online simulation algorithm of UAV clusters. 4. A dynamic simulation engine is used to render environmental information using modeling data from simulated experimental environments. Its main function is to construct cluster combat simulation scenarios, connect experimental environment simulation model environmental data, and render highly realistic environmental information. It uses a fiber optic reflection memory network to obtain the six degrees of freedom data of each simulation model in the hardware-in-the-loop simulation system and the digital simulation system. The cluster simulation control commands sent from the virtual-real hybrid simulation experimental platform drive the UAV cluster to perform 3D trajectory simulation and display in real time. Finally, the rendered and constructed 3D experimental scene is sent back to the virtual-real hybrid simulation experimental platform for simulation and display. III. Multi-core Parallel Simulation System. This embodiment includes two multi-core parallel simulation frameworks: a multi-core parallel simulation framework deployed on the RTX real-time system (belonging to the multi-core parallel simulation node) and a multi-core parallel simulation framework deployed on the Windows system (belonging to the digital simulation system). The two multi-core parallel simulation frameworks are used for the management of parallel semi-physical simulation entities and parallel digital simulation entities, respectively.The RTX version employs a classic real-time operating system architecture, ensuring strong real-time performance through hardware interrupts and priority scheduling, with a response cycle consistently below 1ms. The Windows 11 version, based on the standard Windows system, utilizes Windows programs to provide response services. While it cannot achieve the strong real-time performance of RTX, it achieves lower response latency through optimized scheduling algorithms and system resource management, with fluctuations controlled within 12ms. Both multi-core parallel simulation frameworks belong to the multi-core parallel simulation system. Figure 2As shown, the multi-core parallel simulation system adopts a hierarchical upper-lower-level computer structure. The upper-level computer serves as the control host for the multi-core parallel simulation system, running Windows 11 and primarily responsible for experimental state configuration, experimental process control, and the storage and processing of experimental data. The lower-level computer module is divided into two groups: one group runs Windows 7 and is equipped with a high-version RTX real-time operating subsystem supporting multi-core parallel computation; the other group runs Windows 11 to accommodate the dependency requirements of some models on higher-version systems. The upper-level computer deploys a virtual-real hybrid simulation experimental platform, in which the multi-core simulation system is integrated. Its main functions are configuring experimental states, controlling the experimental process, and storing and processing experimental data. The upper and lower-level computers communicate via reflected memory. The lower-level computer mainly runs the simulation scheduling management program and real-time simulation, responding to the upper-level computer's instructions in real time and scheduling simulation tasks in real time according to the clock signal to complete the simulation experiment. Both multi-core parallel simulation frameworks on the lower-level computer adopt an implementation scheme of "management-side task scheduling and periodic execution of the computation part," combined with a Reactor + thread pool pattern based on NIO. The Reactor pattern is an event-driven asynchronous pattern that uses one or more Reactors to handle I / O events (multi-node simulation events), suitable for high-concurrency scenarios. It separates the listening and processing of multi-node simulation events, improving system response speed. Thread pools effectively manage thread resources, avoiding the overhead of thread creation and destruction, and improving the system's concurrent processing capabilities. To ensure reasonable utilization of thread resources, the Reactor listening and scheduling thread is strongly bound to CPU0, ensuring real-time and efficient multi-node simulation events. Multi-node simulation events are distributed to corresponding event processing threads for real-time simulation, minimizing the coupling between software structure and control flow. In multi-core simulation task scheduling, the RTX version uses an RTX response service every 1ms to achieve task scheduling and data exchange. The management end issues task scheduling instructions every 1ms, and the real-time simulation computing end polls for task scheduling instructions and responds every 1ms. Communication between simulation cores strictly follows the order of first checking interrupt status and then accessing memory, ensuring efficient coordination of multi-task processing. The Windows 11 version uses a similar mechanism, but relies on Windows kernel optimization and user-mode scheduling to achieve near real-time task response. In both frameworks, the Reactor listening thread in real-time parallel simulation is responsible for monitoring the simulation process, scheduling tasks across cores, coordinating inter-core communication, allocating shared resources, handling simulation failures, and acquiring and processing data. In the RTX version, its strong real-time performance makes it more suitable for latency-sensitive, high-precision simulation tasks; while the Windows 11 version offers better system compatibility and model adaptation capabilities, making it suitable for applications with lower real-time requirements but requiring support from higher system versions. IV. Communication Architecture of Multi-Core Parallel Simulation Framework. For example... Figure 3As shown, the framework management mainly includes the division of real-time parallel simulation task modules, the allocation of simulation tasks on each core, and the real-time scheduling of simulation tasks. Real-time task scheduling is a crucial aspect of parallel simulation system design. The main purpose of task scheduling is to rationally schedule tasks to execute on various processors, minimizing program execution time while ensuring the accuracy of simulation results. The designed real-time parallel simulation task scheduling adopts an RTX clock signal-driven scheduling mechanism. The RTX clock signal is generated by the RTX software system, which has dynamic priority allocation capabilities, task switching time of less than 5.2µs, and interrupt response time of less than 5µs, making it a real-time simulation operating system. The interrupt service routine under the RTX system notifies the occurrence of the simulation event in the shortest possible time, while other high-frequency calculations and data interactions are performed within the triggered simulation task itself through the communication mechanism between the interrupt and the task. This avoids the various limitations of writing interrupt service routines and further reduces interrupt latency. On the Windows side, the real-time parallel simulation framework adopts a similar task scheduling mechanism, but since the Windows system itself is not a real-time operating system, the response time and the determinism of task scheduling are somewhat lower than RTX. Task scheduling on the Windows side primarily relies on kernel optimization and user-mode scheduling of the operating system. Through high-priority threads and fine-grained task management, near real-time task response is achieved. The response time of the Windows system is typically controlled within 12ms. While it cannot achieve the strong real-time performance of RTX, it still meets the needs of simulation scenarios with lower real-time requirements, while providing better system compatibility and model adaptation capabilities. After a simulation task is activated, the simulation results from the previous cycle are first updated to the shared memory cache. When the framework management detects updates to simulation event data on other CPU cores in the cache, it transmits the data to the host computer via reflected memory and then activates the real-time step-by-step solution tasks on other cores. Driven by external event instructions, the parallel simulation framework management periodically activates model tasks, updates simulation results, and sends them to the host computer until the simulation experiment ends. The inter-task communication models supported by parallel simulation mainly include: shared memory, semaphores, and reflected memory. The advantages of using shared memory for inter-task communication are the fastest access speed, the ability to specify the size of the data area, and the ability to freely configure the structure of the data area as needed. Using reflective memory for interaction with the host computer and external devices, the Fiber-Optic Reflective memory network employs high-speed fiber optic hardware data sharing technology. Compared to Ethernet communication networks, it not only possesses strict transmission determinism and predictability but also features simple data communication protocols, fast data transmission, strong real-time performance, and good platform versatility. V. Node Communication Architecture of the Heterogeneous Interconnected Multi-core Distributed Simulation Experiment System.Traditional adaptive communication only solves the problem of collaborative communication between simulation nodes of the same type. The data communication between virtual and physical nodes has always been a major technical challenge in engineering. The adaptive communication and distributed simulation provided in this embodiment solve the multi-point communication problem between hardware-in-the-loop (HIL) simulation nodes and between HIL and digital simulation nodes during co-simulation. Combined with a fiber optic reflective memory network, it further improves the collaborative communication efficiency between HIL simulation nodes in large-scale heterogeneous clusters. Figure 1 As shown, the node communication architecture of the heterogeneous interconnected multi-core distributed simulation experimental system includes an ad hoc network data link and an adaptive communication module (deployed in an external task computer) set in the semi-physical simulation node, and an ad hoc network data link simulation system set in the multi-core parallel simulation node. The ad hoc network data link simulation system is actually a digital simulation of the ad hoc network data link and the adaptive communication module. Communication between the semi-physical simulation entities simulated by the semi-physical simulation node, between the parallel simulation entities simulated by the multi-core parallel simulation node, and between the semi-physical simulation entities and the parallel simulation entities utilize the ad hoc network data link and / or the ad hoc network data link simulation system. To meet the distributed simulation requirements of large-scale simulation entities, an optimal communication mechanism and distributed simulation strategy for semi-physical models are constructed for large-scale simulation entity scheduling and management, further enhancing the virtual-physical interconnection and cluster control semi-physical simulation capabilities of the simulation experimental platform. The principle of its ad hoc network data link simulation system is as follows: Figure 4 As shown. 1. Adaptive Communication Method. To improve the communication performance of large-scale simulated entities in the simulation experimental platform, and addressing issues such as high latency, limited bandwidth, and frequent communication failures in large-scale entity communication, a distributed communication interaction strategy with task-demand adaptation is designed. Key adaptive communication technologies such as optimal interaction timing and minimum interaction content extraction are studied. A distributed adaptive communication mechanism under multi-task scenarios is constructed, forming the adaptive communication simulation capability of the simulation experimental platform.

[0025] Specifically, such as Figure 5As shown, the adaptive communication method between simulated entities includes: S1 detecting the state of the simulated entity and determining the event triggering timing; the simulated entities include hardware-in-the-loop (HIL) simulation entities and parallel simulation entities; S2 determining the simulation entities that communicate with each other when the event communication is triggered; S3 transmitting data between the determined simulation entities and encoding and decoding the data. Each simulation node in the figure adopts the above adaptive communication mechanism. The hardware-in-the-loop simulation nodes communicate using Ethernet; the multi-core parallel simulation nodes communicate using fiber optic reflective memory; the hardware-in-the-loop simulation nodes and the multi-core parallel simulation nodes communicate using Ethernet; the parallel simulation entities within the multi-core parallel simulation nodes communicate using shared memory. It can be seen that due to hardware limitations, the hardware-in-the-loop simulation nodes will connect to various external hardware devices. The adaptive communication mechanism between the hardware-in-the-loop nodes is constructed using traditional Ethernet. The multi-core parallel simulation nodes, to overcome hardware limitations, utilize reflective memory and adopt a multi-core simulation approach. The self-organizing network communication part uses a data link simulator to simulate the self-organizing network communication link. The adaptive communication mechanism between the multi-core parallel simulation nodes is constructed using fiber optic reflective simulation. The hardware-in-the-loop simulation node adaptive communication algorithm and the cluster task planning algorithm run on an external task computer. The multi-core parallel simulation node adaptive communication algorithm is implemented by the self-organizing network data link simulation system, which is mainly used for the construction of data links between various system simulation subsystems. The cluster task planning algorithm is completed by the online task planning simulation system, which is mainly used for the task planning and trajectory planning of each node in the cluster simulation sub-node, as well as the cluster formation maintenance and other functions.

[0026] In the adaptive discrete event-triggered communication mechanism, whether the sampled data is transmitted depends on whether the error between the current time and the previous time transmission value is greater than a fixed threshold. Although the introduction of the event triggering mechanism saves certain network resources while ensuring system performance, the fixed threshold, if not properly selected, will cause the following problems: (1) When the system is in a stable state, a large number of data packets are still transmitted. However, in actual engineering, when the system is stable and has good performance, the transmission of data packets needs to be reduced; (2) When the system does not trigger for a long time, the delay will increase, which will lead to poor system stability. In order to solve this problem, this embodiment attempts to achieve a better balance between system performance and network resource saving by dynamically and automatically adjusting the triggering threshold according to the system's operating behavior. More specifically, the step of determining the event triggering time in step S1 includes: S11 Calculate the state offset based on the current state estimate of the simulated entity and the system state estimate that meets the event triggering conditions at the previous time; and calculate the dynamic scaling factor using the state offset. The method for calculating the dynamic scaling factor includes using the formula:

[0027] , ,

[0028] Calculate the dynamic scaling factor; where λ is the dynamic scaling factor, Δ is the normalized rate of change of state, ε is the offset correction, and α and For fuzzy rule strength coefficients, This is the estimated system state value that currently meets the event triggering conditions. This is the estimated state of the system that previously met the event triggering conditions.

[0029] , , If it is a given constant, then it is a dynamic threshold triggering mechanism. Arctangent function Combined parameters and Automatically adjust threshold parameters , .like At this time , Then a smaller one is needed. Accelerate network data transmission; conversely, use larger... By reducing the transmission frequency and triggering dynamic changes in parameters, data transmission is dynamically adjusted, ultimately leading to an optimized trade-off balance between system performance and communication resources.

[0030] In particular, if or This is a fixed threshold event triggering mechanism. An upper limit is set for the automatically adjusting threshold parameter. The purpose is to limit the transmission delay when the system is in a stable state, that is, to prevent the system from not transmitting data for a long time when it is in a stable state. A small positive number is introduced. As an offset correction amount, it is used to make appropriate adjustments. The output.

[0031] S12 uses the dynamically changing scaling factor and the upper and lower bounds of the automatically adjusted threshold parameter to adjust the automatically adjusted threshold parameter that met the event triggering condition at the previous moment, thereby obtaining the automatically adjusted threshold parameter at the current moment.

[0032] Specifically, the formula is used: Calculate the automatic adjustment threshold parameter at the current moment;

[0033] in, The threshold parameter is automatically adjusted at the current moment. Here, λ represents the automatically adjusted threshold parameter from the previous time step, and λ is the dynamically changing scaling factor. To automatically adjust the lower limit of the threshold parameter, To automatically adjust the upper limit of the threshold parameter.

[0034] S13 adjusts the threshold using the automatic adjustment threshold parameter at the current moment, and uses the moment that meets the adjusted threshold requirement as the next moment that meets the event triggering condition. Specifically, it uses the formula:

[0035] Calculate the next time the event triggering condition is met; where t k+1 h represents the next time the event triggering condition is met, and t represents the time when the event triggering condition is met. k h is the current moment when the event triggering condition is met, h is the sampling period, Z is the number of sampling periods, and N is a natural number; For state estimation error, This is the estimated system state at the current moment. This represents the estimated system state that currently meets the event triggering conditions; T is the transpose symbol. The threshold parameter is automatically adjusted at the current moment. It is a positive definite weight matrix. It is an integer and .

[0036] At the start, each node is in a discrete state. When a communication event is triggered... It monotonically does not increase over time, that is: .

[0037] It can be proved by mathematical induction that: When, satisfy ; At that time, we can conclude that:

[0038] ,in: Indicates change over time Number of monitoring cycles Let c be an exponential function of the variable c, and let c be a constant whose value is an empirical value. and Represents the weight, and has .because Then there is When, satisfy ; At that time, assuming From the above formula, we can deduce that:

[0039] ;

[0040] Based on the assumption, it is derived that...

[0041] ,

[0042] And thus obtain .

[0043] That is: dynamic threshold function is a monotonically non-increasing function, and the threshold change rule can be consistent with the choice of communication trigger timing. Moreover different values of can be used to represent different communication mechanisms. When changes with time, it represents a dynamic event trigger mechanism; when when is always a constant value, it represents a static event trigger communication mechanism, that is, the traditional periodic communication mechanism.

[0044] S14 determines whether the next moment satisfying the event trigger condition belongs to Zeno behavior. If not, the simulation entity can communicate at the next moment satisfying the event trigger condition. When an event is triggered, it is necessary to determine whether there is Zeno behavior. Zeno behavior refers to the behavior of generating an infinite number of operations in a finite short time, that is, the event trigger interval tends to be infinitesimal, which will lead to frequent communication, occupying too much communication resources, and unable to achieve the purpose of saving resources. Therefore, when designing a trigger method, Zeno behavior should be avoided, otherwise communication cannot be optimized, and even the communication burden will be加重. This step identifies Zeno behavior by judging whether the next event moment is too dense; allows reasonable event-triggered communication to avoid unreasonable events from causing simulation crashes; ensures that the communication behavior of the simulation entity is both responsive in a timely manner and does not破坏时间连续性, which is the key guarantee for the stable operation of a complex event-triggered simulation system. After not belonging to Zeno behavior, it means that the simulation entity can establish a communication connection at this moment. As the number of nodes within the communication radius increases, the number of links that each node needs to maintain under the communication mechanism increases significantly, and the corresponding calculation and communication complexity will also increase significantly. In addition, algorithms that do not consider the upper bound of the vertex degree usually have the problem of the maximum vertex degree being, so it is also necessary to determine the simulation entities that communicate with each other. Step S2 for determining the simulation entities that communicate with each other includes: S21 constructs the communication mapping matrix and communication topology graph of each simulation entity. The Beta angle check algorithm is a distributed communication topology control algorithm that enables each node to maintain necessary neighbor node links and remove redundant communication links. Due to the limitation of communication resources and the absence of centralized base station and other facilities for communication resource allocation, the number of communication links that each node can maintain is limited. Therefore, the Beta angle detection algorithm must make the generated network graph have a finite upper bound of vertex degree, that is, the maximum number of edges of any node must be less than a certain constant value. To meet the distributed requirements, each node should be able to obtain information from its one-hop neighbor nodes.

[0045] The principle of the Beta angle detection algorithm is: Denote the vertex set of the graph as , vertex and vertex The edge exists if and only if for any vertex , interior angle satisfy ,in . Figure 6 The edges are shown in two dimensions. Does it meet the requirements? Conditions for corner detection algorithm. Nodes in the diagram. Located in the region Outside, that is, the corresponding All less than satisfy , indicating the edge satisfy Corner detection algorithm conditions. And... Figure 7 middle All located in within, that is That is, the edge Not satisfied Corner detection algorithm conditions. For ease of description, the above is based on the two-dimensional case; the algorithm remains effective in three-dimensional cases.

[0046] The Beta corner detection algorithm has the following advantages: Essentially, the Beta corner detection mechanism follows the principles of distributed and symmetric operation. Therefore, as the number of nodes increases, the generated network structure exhibits a certain degree of scalability and robustness. Furthermore, by modeling the communication links between nodes as bidirectional links, the symmetry of the Beta corner detection rules helps avoid unnecessary problems caused by hidden terminals due to unidirectional links. However, as the number of nodes within the communication radius increases, the number of links that each node needs to maintain under a fully connected mechanism increases significantly, leading to a substantial increase in computational and communication complexity. Moreover, algorithms that do not consider the upper bound of vertex degree often result in a maximum vertex degree of... The problem lies in the Beta angle detection algorithm mechanism, where the vertex degree has a constant upper bound, which is... The function, and related to the number of nodes. Regardless, the upper bound of the maximum vertex degree is shown in the following formula.

[0047] In two dimensions: In three dimensions: .

[0048] In summary, the Beta angle detection algorithm has the following characteristics: ① controllable upper bound of vertex degree, ② adaptive implementation, and ③ satisfies symmetry.

[0049] In this embodiment, the specific steps for constructing the communication mapping matrix and communication topology of each simulated entity using the Beta angle detection algorithm include: S211 Selecting a neighboring node from the neighboring nodes of a simulated entity and calculating the angle formed by the simulated entity, the neighboring node, and other neighboring nodes. For example, the neighboring nodes of simulated entity A are B, C, and D, with a preset Beta = 30°; selecting neighboring node B, calculating the angle ∠CAB between line AB and line AC = 25°, and the angle ∠DAB between line AB and line AD = 28°. S212 Comparing the angles with the preset Beta value and recording the number of times the angle is less than the Beta value. Continuing the example, if both angles are < 30°, the number of times is 2. S213 When the number of times the angle is less than the Beta value is equal to the number of neighboring nodes, establishing a communication link between the simulated entity and the neighboring node. Continuing the example, when the number of times equals the number of neighbors, establishing the AB link, i.e., a = 1. If there is an angle ≥ 30°, i.e., the number of times = 1 ≠ the number of neighbors, then the AB link is not established, and a = 0. S214 traverses all neighboring nodes of the simulated entity and repeats steps S211~S213 to construct the communication mapping matrix and communication topology of the simulated entity.

[0050] S22 calculates the final link point using the communication mapping matrix and communication topology graph to determine the simulated entities communicating with each other. The communication topology formed by the Beta angle detection algorithm is usually modeled as an undirected graph, where the vertices and edges represent nodes and communication links between them, respectively. Additionally, the second smallest eigenvalue of the Laplacian matrix corresponding to the communication topology graph, i.e., algebraic connectivity, is used to measure the quality of network connectivity. To better achieve data interaction, global dynamic data of nodes should be obtained as much as possible during non-periodic periods. The Beta angle detection algorithm can obtain the upper and lower bounds of algebraic connectivity. The node with the smallest vertex degree determines the upper bound of the algebraic connectivity of the network graph; therefore, the existence of this node has a greater impact on the network connectivity. Thus, it is necessary to reduce the number of low-degree nodes to improve the distributed connectivity of the network. In this embodiment, the step of calculating the final link point using the communication mapping matrix and communication topology graph includes: S221 calculating the second smallest eigenvalue of the Laplacian matrix using the communication mapping matrix and identity matrix. Combining the identity matrix (with 1 on the diagonal, representing the node itself) and the communication mapping matrix, a Laplacian matrix is ​​constructed, and its second smallest eigenvalue is calculated. The second smallest eigenvalue of the Laplace matrix is ​​a core indicator of network connectivity; a larger eigenvalue indicates better connectivity, while a value of 0 indicates a disconnected network. This step transforms the topological relationship into numerical features, quantifying global connectivity. S222 uses a power iteration method to generate feature pairs between the number of nodes in the communication topology graph and the second smallest eigenvalue of the Laplace matrix. In this step, based on the determined number of vertices N in the communication topology graph, feature pairs (eigenvalues ​​and corresponding eigenvectors) are generated using a power iteration method. S223 uses the estimated value of the eigenvector corresponding to the second smallest eigenvalue in the feature pair to calculate the estimated value of the communication topology graph connectivity; the estimated value of the communication topology graph connectivity is calculated using the formula:

[0051] Calculate; where, L is an estimate of the connectivity. ij Let be the Laplace matrix formed by nodes i and j. Let be the element value of the Fiedler vector corresponding to the j-th node in the k-th iteration. For the element estimate of the Fiedler vector corresponding to the i-th node, N i Let be the set of neighboring nodes of node i; define a coefficient matrix W such that its principal eigenvalues ​​correspond to a vector of all 1s in the Laplacian matrix: ;in, for identity matrix Denotes the minimum value, satisfying Then the eigenvectors of algebraic connectivity for The principal eigenvectors of the matrix. When the initial value... satisfy When, the iteration vector It will converge to The second largest eigenvector, namely the Laplace matrix. The algebraic connectivity vector. Therefore, each node in a multi-node system can update the corresponding element in the algebraic connectivity vector using the distributed power iteration method represented by the following formula: Calculate the element-wise estimates of the Fiedler vector after the iteration; where, Let be the element value of the Fiedler vector corresponding to the i-th node in the k-th iteration. For each node, the element value W corresponding to the (k-1)th iteration of the Fiedler vector i Let be the coefficient matrix of the i-th node. Using the formula: Calculate the element values ​​corresponding to the iterated Fiedler vector; where, Let be the element value of the Fiedler vector corresponding to the j-th node in the k-th iteration. This represents the element estimate of the Fiedler vector corresponding to the i-th node in the k-th iteration. This is the modulus of the element value corresponding to the Fiedler vector in the k-th iteration for each node. However, this exponential iteration has the following two problems: the existence of rounding errors during the iteration process, and the reliance on only the initial constraints. Iteratively obtain the initial vector It may make Gradually approaching . The computation requires access to global information, which leads to frequent global information exchange, increasing resource overhead and causing distributed characteristics to fail. To address these issues, we consider updating the global information at a specified number of iterations. At the same time, corrective measures are introduced. Iteration coefficient matrix The rules are as follows: ,in, Let I be the coefficient matrix at the k-th iteration, L be the identity matrix, and J be a matrix of all ones. For a minimum positive value, k w Given a positive integer, n is the number of nodes, and k is the number of iterations. "mod" represents the modulo operation. Therefore, each node is only considered in... This allows for large-scale data exchange, thereby reducing global data exchange. Information overhead. Under this mechanism, each node can acquire information in a distributed manner and obtain connection nodes.

[0052] S224 increases the number of connected nodes when the estimated connectivity of the communication topology graph is greater than the number of nodes, thus forming the final link. In graph theory and distributed systems, for a network consisting of n vertices, if the estimated connectivity is much greater than n, it indicates that the theoretical connectivity potential of the current network has not been fully realized, meaning the actual number of connected nodes is insufficient; conversely, if it is close to or less than n, the network connectivity is basically adapted to the scale. If the estimated connectivity > the number of vertices n, it indicates that the actual number of connected nodes in the current network is insufficient, and there are nodes that can be connected but are not, requiring an increase in the number of connected nodes. If the estimated connectivity ≤ the number of vertices n, it indicates that the number of connected nodes in the current network is already adapted to the scale, and no further expansion is needed; the existing link remains.

[0053] Step S3 employs the Huffman algorithm for data encoding and decoding. To reduce the amount of communication data transmitted and decrease latency, Huffman coding is used. This coding method is an entropy coding (weighted coding) algorithm for lossless data compression, and it is a type of variable-length coding (VLC). This method constructs the shortest codeword with the lowest average length based entirely on the probability of character occurrence; it is sometimes called the optimal coding. Huffman coding utilizes the tree structure in data structures to construct an optimal binary tree with the support of the Huffman algorithm. In computers, data storage and processing are based on bytes. A Western character is represented by one byte, while a Chinese character requires two bytes. This encoding form, where each character is represented by the same number of bytes, is called fixed-length coding. Huffman coding is a variable-length coding method that constructs the shortest average codeword based on the probability of character occurrence. In other words, if a character appears frequently in a document, its code is shorter; if a character appears infrequently, its code is longer. When the lengths of codewords in the encoding are strictly arranged in reverse order according to the probability of their corresponding symbols appearing, the average length of the encoding is minimized. In an extended binary tree, the path from the root to an outer node can be encoded by using 0 to represent moving one step to the left subtree and 1 to represent moving one step to the right subtree. The path code is 010. From the root to the node. The path codes are (00, 010, 011, 100, 101, 11). Since each path is exactly the first half of another path, no path code is a prefix of any other path code. Therefore, these codes can be used to process characters... Encoding. Let S be a string consisting of these characters. It is a character The frequency of occurrence of, among which Belongs to set If S is encoded using these codes, the length of the encoded bit string is:

[0054] For a tree with An extended binary tree with external nodes, where the external nodes are labeled as The length of its corresponding encoded bit string is: .

[0055] From the root to the external node The path length (i.e., the number of edges in the path) of a binary tree; WEP is the weighted external path length of the binary tree. To shorten the length of the encoded string, binary tree encoding must be used. The external nodes of the binary tree correspond to the characters of the string to be encoded, and the WEP is minimized. A binary tree whose WEP is minimized for a given set of frequencies is called a Huffman tree. The process of constructing a Huffman tree is as follows: First, a set of binary trees is built, each containing only one external node. Each external node represents a symbol of the string, and its weight is equal to the frequency of that symbol. Then, two binary trees with the smallest weights are continuously selected from the set and merged into a new binary tree by adding a root node, making the two binary trees the left and right subtrees respectively. The weight of the new binary tree is the sum of the weights of the two subtrees. This process continues until only one tree remains.

[0056] 1. A method for collaborative control of UAV swarms. This method enables heterogeneous UAV swarms to form specific formations, automatically fly to target areas, and safely complete various types of tasks, such as regional reconnaissance and strikes. In this embodiment, the swarm collaborative control method is deployed in an online task planning simulation system, including a hardware-in-the-loop (HIL) node with an external computing module for online task planning, a multi-core parallel simulation node, and a digital simulation system. Figure 8 As shown, the cluster collaborative control method specifically includes:

[0057] The S101 task allocation method assigns tasks to different simulated entity clusters; it plans actions for the master unit of each simulated entity cluster and calculates the follow-up action instructions for the slave units within the cluster. The purpose of this UAV swarm task allocation method is to overcome the shortcomings of existing UAV swarm task allocation algorithms in combat scenarios, such as not considering the entire UAV swarm as the task allocation object, the inability to simultaneously consider UAV swarm attributes and combat task attributes as task allocation factors, the inability to guarantee algorithm response speed, the inability to simultaneously minimize the overall cost of the allocation scheme and the variance between the values ​​of each task, and the inability to guarantee that each task is allocated at least one execution resource in all situations. Therefore, a UAV swarm combat task allocation method based on discrete particle swarm optimization is proposed.

[0058] This method generates individual rescue missions by combining information such as mission index, rescue priority, rescue location, and required arrival time. It generates mission allocation objects by combining information such as UAV swarm index, average speed of the UAV swarm, mission execution capability of the UAV swarm, and number of surviving members of the UAV swarm. A mission cost function that satisfies all the above factors is defined. The legality of different missions and UAV swarms is checked. Among legal missions and UAV swarms, the larger value is obtained by comparing the number of valid missions and UAV swarms. This value is used as the order of a full-rank sparse matrix. Random initial mission particles are generated by randomly arranging the column order of the full-rank sparse matrix. A zero matrix with the same order as the particles is initialized. The cost function is used to evaluate the value of each generated mission. The UAV swarm number and the combat mission number are used as row and column indices, respectively. The mission evaluation value is updated in this zero matrix to construct a mission cost matrix, and the mission with the highest value relative to each UAV swarm is recorded. At the beginning of each optimization iteration, mutation is achieved by generating random particles, and they are randomly cross-transferred with locally optimal particles to achieve genetic inheritance. By setting the initial number of generated particles and the number of optimization iterations, the value and variance of each particle are calculated using a task cost function during each iteration. The locally optimal particle with the highest value and lowest variance in this iteration, along with the globally optimal particle for this algorithm run, are saved. After all optimization iterations are completed, decoding rules are designed to decode the globally optimal particle to obtain a partial allocation scheme. Then, the unassigned tasks in the decoding results are supplemented by allocating the drone cluster according to the recorded highest-value task, ultimately yielding the rescue task allocation scheme for this algorithm run. Figure 9 As shown, the task allocation steps include: S1011 performing mathematical modeling of the task and the simulated entity cluster, defining the relevant attributes of the task and the simulated entity cluster. In this embodiment, mathematical modeling is performed on the combat task and the UAV cluster, defining relevant variables such as the task index, rescue priority, rescue location, and required arrival time of the rescue for the combat task, and defining relevant variables such as the UAV cluster index, the average movement speed of the UAV cluster, the ability of the UAV cluster to perform tasks, and the number of surviving members of the UAV cluster.

[0059] For the number is cluster , It has the following attributes: 1. Cluster The number of surviving drones in the middle is , 1. It is an integer greater than or equal to zero; 2. In At any moment, cluster The geometric center coordinates are , It is the matrix transpose symbol. , and They are Time Cluster The center coordinates include the north direction coordinates, the east direction coordinates, and the height coordinates perpendicular to the horizontal plane; 3. In At any moment, cluster The average flight speed is , , and Clusters The average flight speed includes the northward velocity component, the eastward velocity component, and the velocity component perpendicular to the horizontal plane; 4. Cluster Possesses The set of task type execution capabilities is ,in It is a signed number that is equal to 0 or 1. Represents cluster Not capable of executing the first The ability to perform such tasks Represents cluster Capable of executing the The ability to perform this task. For the numbered... combat mission , It has the following attributes: 1. Task type , , 1. It is a set of positive integers; 2. The task completion flag is... , It is a signed number that is equal to 0 or 1. This indicates that the task has not been completed. This indicates that the task has been completed; 3. The task execution priority is... , The larger the absolute value, the higher the priority of the combat mission; 4. In Time, Mission The geometric center coordinates of the region are , , and They are Constant Task The center coordinates include the north direction coordinates, the east direction coordinates, and the altitude coordinates perpendicular to the horizontal plane; 4. Task The start and end times are defined as follows: and , , , .

[0060] S1012 constructs the task cost function based on the characteristics of the task and the attributes related to the task and the simulation entity cluster. Let the cluster... Execute the task The task allocation pair is represented as Then, it can be achieved by designing functions. To assess the value of the rescue mission. (Known) It is a task Execution priority It is a cluster Not capable of executing the first The ability to perform such tasks It is a task If there is no completion flag, the function is defined as: ,in, It is a task Remaining effective time, known task Start execution time and end time ,exist The time threshold for executing tasks is Then the task Remaining valid time Defined as:

[0061] in, It is a cluster Execute the task The cost and time spent, for known clusters The number of drones, average flight speed, and center location are respectively , and combat mission The central position is , This is the operator for calculating the Euclidean distance between two 3D coordinate points. Defined as:

[0062] .

[0063] S1013 filters out legitimate tasks and simulation entity clusters, generating a set of allocable legitimate tasks and a set of legitimate simulation entity clusters. At any given time, a legally valid set of clusters Defined as: ,in, It is a positive integer and ; It is a positive integer and ; It is a cluster The number of drones; This represents a kind of relationship between the first task and the second task. Each task is iterated through to determine the cluster. Whether the computation has the capability to execute tasks depends on whether at least one task exists. The task was not completed. And the type is This enables the cluster For the first Execution capability of various task types ,but ,otherwise ; It is a time criterion used in clusters. Whether they can arrive at the designated location in time to carry out the mission , The definition of is: In the formula It is the present moment. and Clusters Average flight speed and center position, It is an operator for calculating the magnitude of a three-dimensional vector. It is a cluster Execute the task The cost and time spent It is a task The start time of execution; This represents a sequence of tasks from the first task to the second task. Each task is used to iterate and determine the cluster. The calculation determines whether a task can be completed within a specified time; if at least one task exists... And the task was not completed. This enables the cluster Execute the task Time criteria ,but ,otherwise .exist At any given moment, a legal and valid set of tasks Defined as: ,in, It is the present moment; It is a positive integer and ; It is a positive integer and ; It is a task The end execution time; Indicates task The executable task has not yet ended; Represents a kind of cluster from the first cluster to the second cluster. The cluster is traversed to determine the task. Whether the operation can be executed by a cluster, if the task It was not completed And at least one cluster exists. The cluster Having the ability to execute Type of task The ability is ,but ,otherwise ; It is a time criterion used in clusters. Whether they can arrive at the designated location in time to carry out the mission ; This represents a sequence of tasks from the first task to the second task. The task involves iterating through and judging each cluster. Whether the task can be successfully executed by a cluster within its lifecycle, if the task It was not completed And at least one cluster exists. Make the cluster Execute the task Time criteria ,but ,otherwise .

[0064] S1014 takes the larger value between the number of tasks and the number of resources, and uses this larger value as the order to create a full-rank sparse matrix; a random particle mathematical model is established by randomly shuffling the column order of the full-rank sparse matrix. Random particle matrix. . OK column identity matrix The original particle before random transformation is defined as: ,in, The order of the matrix is ​​the larger of the total number of elements in the set of valid clusters and the total number of elements in the set of valid tasks. The formula is: , It is the total number of legitimate clusters. It is the total number of legitimate tasks. It is the operator for calculating the number of elements in a set. This is a maximum value comparison output operator. (Full-rank sparse matrix) It is a random permutation matrix, which is defined by random permutations. Diagonal matrix of order 1 The matrix obtained by taking the column vectors in their positional order is defined as follows: ,in, The order of the matrix is ​​the larger of the total number of elements in the set of valid clusters and the total number of elements in the set of valid tasks. It is a diagonal matrix and its expression is: ,in, It is used for randomly swapping diagonal matrices. Random operators for column vectors The matrix represents A square matrix of order.

[0065] S1015 uses legal tasks as columns and legal simulation entity clusters as rows. It calculates the task cost function for each task pair under each column index and row index, fills the results into a zero matrix with the same order as the particles to obtain the task cost matrix, and records the maximum value task that each legal simulation entity cluster can execute to generate a set of backup task pairs.

[0066] Task Cost Matrix Defined as: ,in, and These are the number of valid clusters and the number of valid tasks, respectively. Number of valid clusters With the number of legitimate tasks The larger value; yes A column vector consisting of all zeros in row 1 and column 1; It is by OK The effective task cost matrix, composed of columns, is defined as follows:

[0067] index in the formula and All are integers. It is a task value evaluation function. Is to make cluster Execute the task The task allocation pair The matrix represents OK A matrix of columns. A set of alternative task pairs. It records the most valuable task for each legitimate cluster at the current moment, and it is defined as:

[0068] ,in, It is an integer. This represents a task within the set of legal tasks. This represents a specific cluster within the set of valid clusters. Is to make cluster Execute the task The task allocation pair This indicates that the set of legitimate tasks for retrieving clusters is included. The most valuable task is identified and output.

[0069] Furthermore, in step S6, the number of iterations for particle optimization in this algorithm is set to a positive integer. The number of random particles generated in each loop is a positive integer. The globally optimal particle calculated by this algorithm. Defined as:

[0070] ,in, It is the set of positive integers. This is the number of iterations for particle optimization in this algorithm. It is the loop number. It is the first The local optimum particle in the second cycle of optimization. It is the first Local optimal particle in the second cycle The random permutation matrix, Represents a matrix and Perform matrix multiplication. Indicates to and Perform matrix dot product calculation. This indicates that all iterations performed by the algorithm in this iteration are being performed. Repeat the loop and obtain the particle matrix with the highest value. This indicates that the standard deviation operation is performed only on all non-zero elements in the matrix. Indicates the first The number of particles obtained in the next iteration is not unique; the particle matrix with the smallest variance is obtained from the maximum value matrix.

[0071] S1016 performs iterative cycles, randomly genetically processes new particles, evaluates the value and variance of each particle using the task cost matrix, and records the local optimal particle and the current global optimal particle at different iteration numbers until the iteration is complete.

[0072] No. Local optimal particle in the second cycle of optimization Defined as: , .

[0073] in, This is the number of random particle iterations in this cycle; It is a random method Copying partial column data from a matrix to the matrix The genetic operation corresponding to the position of the column data ensures that each column of the matrix has only one non-zero element with a value of 1. It is the task cost matrix; It is the first The generation during the next loop The th random particle Individual particles; matrix Random particles The randomly inserted portion comes from the globally optimal particle. Gene or the local optimum particle from the previous cycle The genetic particles of a gene; It is the first Random particles A random permutation matrix; a matrix Genetic particles The permutation matrix, which is used in relation to the genetic particles The same random gene insertion order is replaced by the permutation matrix of the locally optimal particles in the previous cycle. Permutation matrix with the globally optimal particle Perform matrix column data exchange; Represents a matrix and Perform matrix multiplication. Indicates to and Perform matrix dot product calculation; This indicates that the sum of all elements in the matrix enclosed in parentheses is calculated. This indicates all generated in this loop. A random number of particles are traversed, and the particle matrix with the highest value is obtained. This indicates that the standard deviation operation is performed only on all non-zero elements in the matrix; Indicates the first The iterations obtain the particle matrix with the smallest variance from the maximum value matrix, which is not unique. Specifically, the number of iterations in the formula... ,when From time to time and Established.

[0074] S1017 decodes the globally optimal particle to obtain the set of emergency task pairs; for the cluster of simulation entities without assigned tasks, it supplements the allocation scheme in the set of backup task pairs.

[0075] For the optimal particle Decoding the matrix, the task is to map the relation matrix. Defined as:

[0076] ,in, and These are the number of valid clusters and the number of valid tasks, respectively. Number of valid clusters With the number of legitimate tasks The larger value; yes A column vector consisting of all zeros in row 1 and column 1; It is by OK The effective task pair mapping matrix, consisting of columns, is defined as follows: Index in the formula and All are integers. Is to make cluster Execute the task The task allocation pair.

[0077] For the globally optimal particle Emergency rescue mission after matrix decoding set Defined as:

[0078] , ,in, and These are the number of valid clusters and the number of valid tasks, respectively. It is an emergency relief allocation scheme matrix. Represents the emergency relief allocation scheme matrix The Line number The task pair of the column , Is to make cluster Execute the task The task allocation pair and These are the globally optimal particle matrix and its permutation matrix, respectively. It is a task-to-relationship mapping matrix. Representation matrix Inverse matrix operations, Represents a matrix With matrix Perform matrix multiplication. Represents a matrix With matrix Perform matrix dot product operations.

[0079] S1018 combines the unassigned portions of the emergency task pair set and the backup task pair set to obtain a task allocation scheme. At time 1, the final task allocation scheme of this algorithm run can be obtained, which is defined as the set of target task pairs. Specifically, it is expressed as: ,in, It is an emergency rescue mission group; It is a set of backup task pairs; It is the symbol for adding and merging sets; It is a conditional set subtraction operator. Indicates the set of standby tasks Existing elements are excluded by subtracting those from the set used in emergency rescue missions. The task pairs of the drone swarm that have been assigned tasks are used to obtain a new set.

[0080] The S102 cluster control system controls the actions of each simulated entity within the cluster and updates the individual action information of each entity, feeding back this information. Based on this feedback, the overall motion control quantity of the cluster is calculated. All followers in the cluster use the host as a reference, fixing the center of their relative coordinate system on the host. All wingmen use their coordinates in the relative coordinate system as control references. Once all wingmen are stable near the required positions, the required formation is formed. All wingmen only need to interact with the host and receive its control; they do not exchange information or even measure relative positions with each other. Compared to purely centralized control methods, this formation method is a single-point centralized control method. Figure 10 As shown, the steps of the cluster control include: S1021 solving the height difference between each simulated entity and the reference point; S1022 calculating the layered height difference between each simulated entity; S1023 determining whether layered control is needed based on the heading angle deviation and formation control deviation; S1024 controlling the horizontal position of the simulated entity according to the set formation control and the height position of the simulated entity according to the layered height difference when layered control is needed; controlling the horizontal and height positions of the simulated entity according to the set formation control when layered control is not needed; S1025 calculating the formation control command of the aircraft through the PD controller.

[0081] like Figure 11 As shown, the lead aircraft (main unit) uses It is said that the wingman used express, Let be the heading angle and velocity of the lead aircraft and wingman, respectively. Assume the lead aircraft's position in the northeast coordinate system is . The speed is The wingman's position in the northeast coordinate system is: The speed is Relative formation in reference coordinate system The heading angle can be obtained from the speed of the lead aircraft: The relative formation of the lead and wingmen will be transferred from the lead aircraft's heading reference coordinate system to the inertial coordinate system: Then the expected position of the wingman in the inertial frame is: The difference between the wingman's actual position and his desired position is: Switch to the (lead aircraft heading coordinate system): The relative speeds of the lead and wingman aircraft in the northeast coordinate system are: Switching to the lead aircraft's heading coordinate system yields: In the wingman's heading coordinate system, the x-axis component corresponds to the speed command, the y-axis component corresponds to the heading angle command, and the z-axis component corresponds to the altitude command. If the lead aircraft's status is... Based on the PD control strategy, the control commands for the wingman are obtained: .

[0082] When the main engine is turning, the heading angle control command is calculated based on the heading angle commanded by the main engine; when the main engine is not turning, the heading angle control command is calculated based on the actual heading angle of the main engine.

[0083] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. An adaptive communication method applied to a heterogeneous interconnected multi-core distributed simulation experimental system, characterized in that... include: S1 detects the state of the simulated entity and determines when an event is triggered. The simulation entities include hardware-in-the-loop simulation entities and parallel simulation entities; S2 determines the simulation entities that communicate with each other when event communication is triggered; S3 performs data transmission between simulated entities that are communicating with each other, and encodes and decodes the data; The steps to determine when an event is triggered include: S11 calculates the state offset based on the current state estimate of the simulated entity and the system state estimate that met the event triggering conditions at the previous moment; and uses the state offset to calculate the dynamic scaling factor. S12 uses the dynamically changing scaling factor and the upper and lower bounds of the automatically adjusted threshold parameter to adjust the automatically adjusted threshold parameter that met the event triggering condition at the previous moment, thereby obtaining the automatically adjusted threshold parameter at the current moment. S13 uses the automatic adjustment threshold parameter at the current moment to adjust the threshold, and takes the moment when the adjusted threshold requirement is met as the next moment when the event triggering condition is met. S14 determines whether the next time the event triggering condition is met belongs to Zeno behavior. If it does not, the simulation entity can communicate at the next time the event triggering condition is met.

2. The adaptive communication method according to claim 1, characterized in that... Methods for calculating the dynamic scaling factor include using the formula: ， ， Calculate the dynamic scaling factor; where λ is the dynamic scaling factor, Δ is the normalized rate of change of state, ε is the offset correction, and α and For fuzzy rule strength coefficients, This is the estimated system state value that currently meets the event triggering conditions. This is the estimated state of the system that previously met the event triggering conditions; Using the formula: ， Calculate the automatic adjustment threshold parameter at the current moment; where, The threshold parameter is automatically adjusted at the current moment. Here, λ represents the automatically adjusted threshold parameter from the previous time step, and λ is the dynamically changing scaling factor. To automatically adjust the lower limit of the threshold parameter, To automatically adjust the upper limit of the threshold parameter; Using the formula: ， Calculate the next time the event triggering condition is met; where t k+1 h represents the next time the event triggering condition is met, and t represents the time when the event triggering condition is met. k h is the current moment when the event triggering condition is met, h is the sampling period, Z is the number of sampling periods, and N is a natural number; For state estimation error, This is the estimated system state at the current moment. This represents the estimated system state that currently meets the event triggering conditions; T is the transpose symbol. The threshold parameter is automatically adjusted at the current moment. It is a positive definite weight matrix.

3. The adaptive communication method according to claim 1, characterized in that... The steps to determine the communicating simulation entities include: S21 constructs the communication mapping matrix and communication topology diagram for each simulation entity; S22 uses the communication mapping matrix and communication topology graph to calculate the final link point to determine the simulated entities that communicate with each other.

4. The adaptive communication method according to claim 3, characterized in that... The steps for constructing the communication mapping matrix and communication topology graph for each simulation entity include: S211 Selects a neighboring node from the neighboring nodes of a certain simulation entity and calculates the angle formed by the simulation entity and the neighboring node with the other neighboring nodes. S212 compares the included angle with a preset Beta value and records the number of times the included angle is less than the Beta value; S213 When the number of times the included angle is less than the Beta value is equal to the number of neighboring nodes, establish a communication link between the simulated entity and the neighboring node; S214 traverses all neighboring nodes of the simulation entity and repeats steps S211~S213 to construct the communication mapping matrix and communication topology graph of the simulation entity.

5. An adaptive communication method according to claim 3, characterized in that... The steps for calculating the final link point using the communication mapping matrix and communication topology graph include: S221 uses the communication mapping matrix and the identity matrix to calculate the second smallest eigenvalue of the Laplacian matrix; S222 forms feature pairs by exponential iteration between the number of nodes in the communication topology graph and the second smallest eigenvalue of the Laplace matrix; S223 uses the estimated value of the eigenvector corresponding to the second smallest eigenvalue in the feature pair to calculate the estimated value of the connectivity of the communication topology graph; S224 increases the number of connected nodes when the estimated connectivity of the communication topology graph is greater than the number of nodes, thus forming the final link.

6. An adaptive communication method according to claim 5, characterized in that... The connectivity of the communication topology graph is estimated using the following formula: ， Calculate; where, L is an estimate of the connectivity. ij Let be the Laplace matrix formed by nodes i and j. Let be the element value of the Fiedler vector corresponding to the j-th node in the k-th iteration. For the element estimate of the Fiedler vector corresponding to the i-th node, N i Let i be the set of neighboring nodes of node i; Using the formula: ， Calculate the element values ​​corresponding to the iterated Fiedler vector; where, Let be the element value of the Fiedler vector corresponding to the j-th node in the k-th iteration. This represents the element estimate of the Fiedler vector corresponding to the i-th node in the k-th iteration. The modulus of the element value of the Fiedler vector corresponding to the k-th iteration for each node; Using the formula: ， Calculate the element-wise estimates of the Fiedler vector after the iteration; where, Let be the element value of the Fiedler vector corresponding to the i-th node in the k-th iteration. For each node, the element value W corresponding to the (k-1)th iteration of the Fiedler vector i Let be the coefficient matrix of the i-th node.

7. An adaptive communication method according to claim 6, characterized in that... The update rule for the coefficient matrix W is as follows: ， in, Let I be the coefficient matrix at the k-th iteration, L be the identity matrix, and J be a matrix of all ones. For a minimum positive value, k w Given a positive integer, n is the number of nodes, and k is the number of iterations.

8. A node communication system applied to a heterogeneous interconnected multi-core distributed simulation experimental system, characterized in that: This includes a self-organizing network data link and an adaptive communication module set in the hardware-in-the-loop simulation node, as well as a self-organizing network data link simulation system set in the multi-core parallel simulation node; The self-organizing network data link includes a data transmission module and an adaptive communication method as described in any one of claims 1 to 7, deployed in an external task computer; the self-organizing network data link simulation system is actually a digital simulation of the self-organizing network data link and the adaptive communication module. The hardware-in-the-loop simulation entities simulated by the hardware-in-the-loop simulation nodes, the parallel simulation entities simulated by the multi-core parallel simulation nodes, and the hardware-in-the-loop simulation entities and parallel simulation entities communicate using a self-organizing network data link and / or a self-organizing network data link simulation system.

9. A node communication system for a heterogeneous interconnected multi-core distributed simulation experimental system according to claim 8, characterized in that: The hardware-in-the-loop simulation nodes communicate with each other via Ethernet; the multi-core parallel simulation nodes communicate with each other via fiber optic reflective memory. The hardware-in-the-loop simulation node and the multi-core parallel simulation node communicate via Ethernet. The parallel simulation entities within the multi-core parallel simulation node communicate using shared memory.

Images