CN116167230A - Group intelligent simulation test system and method for water surface unmanned cluster - Google Patents
Group intelligent simulation test system and method for water surface unmanned cluster Download PDFInfo
- Publication number
- CN116167230A CN116167230A CN202310193282.0A CN202310193282A CN116167230A CN 116167230 A CN116167230 A CN 116167230A CN 202310193282 A CN202310193282 A CN 202310193282A CN 116167230 A CN116167230 A CN 116167230A
- Authority
- CN
- China
- Prior art keywords
- simulation
- intelligent
- data
- scene
- unmanned
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
- 238000004088 simulation Methods 0.000 title claims abstract description 269
- 238000012360 testing method Methods 0.000 title claims abstract description 62
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 33
- 238000000034 method Methods 0.000 title claims abstract description 29
- 239000003795 chemical substances by application Substances 0.000 claims abstract description 56
- 238000007726 management method Methods 0.000 claims description 62
- 238000013461 design Methods 0.000 claims description 24
- 238000012549 training Methods 0.000 claims description 24
- 230000003993 interaction Effects 0.000 claims description 22
- 238000004891 communication Methods 0.000 claims description 21
- 230000000007 visual effect Effects 0.000 claims description 18
- 238000013480 data collection Methods 0.000 claims description 17
- 230000008447 perception Effects 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 15
- 238000004458 analytical method Methods 0.000 claims description 12
- 230000000694 effects Effects 0.000 claims description 11
- 238000007405 data analysis Methods 0.000 claims description 10
- 230000002452 interceptive effect Effects 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000012216 screening Methods 0.000 claims description 7
- 238000012876 topography Methods 0.000 claims description 7
- 238000012544 monitoring process Methods 0.000 claims description 6
- 238000010141 design making Methods 0.000 claims description 5
- 238000010998 test method Methods 0.000 claims description 5
- 238000012795 verification Methods 0.000 claims description 5
- 230000010354 integration Effects 0.000 claims description 4
- 230000005484 gravity Effects 0.000 claims description 2
- 239000000463 material Substances 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 claims description 2
- 239000007787 solid Substances 0.000 abstract description 2
- 230000002787 reinforcement Effects 0.000 description 12
- 230000009471 action Effects 0.000 description 8
- 238000011160 research Methods 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 239000010410 layer Substances 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000012827 research and development Methods 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 241000192710 Microcystis aeruginosa Species 0.000 description 1
- 241000112598 Pseudoblennius percoides Species 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 239000011324 bead Substances 0.000 description 1
- 230000003542 behavioural effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000013079 data visualisation Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000004941 influx Effects 0.000 description 1
- 239000002346 layers by function Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a group intelligent simulation test system and a method thereof for a water surface unmanned cluster, which are used for providing a simulation test system for an unmanned ship cluster, instantiating a solid model of multiple intelligent agents into a group intelligent simulation entity in a wanted space by a scene customization module, carrying out simulation test on the unmanned ship cluster, verifying the function condition of the unmanned ship in the cluster scene, and realizing unmanned ship cluster test under complex environments such as typical sea conditions. During simulation, the number of unmanned ship models is copied according to the number of unmanned ship intelligent bodies, each unmanned ship intelligent body corresponds to one unmanned ship model, each unmanned ship intelligent body is an independent individual in a simulation environment, positions, roles and tasks are compiled for each unmanned ship intelligent body through a scene customizing module, each unmanned ship intelligent body calculates and judges according to the positions, roles and tasks where the unmanned ship intelligent body is located, and therefore a cluster environment of multiple intelligent bodies is formed, and simulation tests under the scene of the multiple intelligent bodies are achieved.
Description
Technical Field
The invention belongs to the technical field of simulation systems, and particularly relates to a group intelligent simulation test system and method for a water surface unmanned cluster.
Background
The unmanned ship can replace or assist some ships to execute tasks in some high-risk areas, so that risks are reduced to the greatest extent, cost is reduced, safety is guaranteed, and more importantly, the future combat style of sea combat is greatly influenced or even thoroughly subverted from combat concepts and war designs. According to the different task modules carried, the unmanned ship can cooperate with the manned ship or independently complete various military tasks such as anti-water Mine Combat (MCM), anti-diving combat (ASW), offshore safety (MS), water combat (SUW), support special army combat (SOF), electronic combat (EW), support offshore interception combat (MIO) and the like. Unmanned boats are becoming a medium-hard force for offshore equipment, unmanned systems and key technologies thereof related to the field are also becoming open beads in offshore intelligent manufacturing, and the development of the whole unmanned boat field is decisively influenced.
Autonomous navigation is well known to be a requisite feature capability and core value of unmanned boats, and a prerequisite for this is to achieve autonomous adaptation to the surrounding environment and autonomous behavioral decisions. If the unmanned ship can truly exert the maximum performance of unmanned cooperation, capability tests on situation awareness capability, group communication capability, group control capability, group game capability and the like of the unmanned ship are required.
However, in the current research of unmanned ships, because the test conditions are higher, the research is usually only stopped at the theoretical level, or a large amount of cost is required to be input for carrying out the test once, so that the research and development difficulty is greatly improved. At present, simulation systems for unmanned aerial vehicles or unmanned vehicles are also proposed, but the simulation systems only simulate a single intelligent agent, cannot realize group intelligent simulation tests and cannot realize offshore complex environment simulation tests, so that the current research on unmanned boats generally lacks experimental verification, cannot guarantee the effect of research technology, and cannot guarantee the game effect of systematic unmanned clusters integrating various researches.
In addition, the key of realizing autonomous navigation of the unmanned ship at present is a deep learning technology, and the deep learning has an important characteristic of needing training, and the training of the deep learning model of the unmanned ship is based on training data to train a single intelligent agent. However, the performance of the unmanned ship on the offshore tasks in the real environment is often unsatisfactory, and the main reason is that in the real environment, on one hand, the offshore environment is complex and severe, the change is often unusual, and the autonomous control capability of the unmanned ship is greatly reduced due to the mutual influx of a great amount of environment information and the mutual influence of various environment targets. On the other hand, the offshore tasks in the real environment usually require multi-agent cooperation and game countermeasure, while the training is performed by a single agent, so that the training effect cannot meet the real requirements. The unmanned ship model is built by combining the deep reinforcement learning with the reinforcement learning, and the unmanned ship can realize the autonomous control capability of multi-agent game scenes under complex marine environments, such as situation awareness capability, group communication capability, group control capability, group game capability and the like in continuous cyclic calculation and repeated attempts by utilizing the capability of the deep reinforcement learning for self-iterative learning exploration. However, the foregoing also mentions that deep reinforcement learning requires a large number of attempts to be made, but there is not as much opportunity in reality to enable unmanned boats to make a large number of attempts. In this regard, the intelligent group simulation test system for the unmanned water surface clusters is provided, the simulation test system is based on various capacities of simulation tests on unmanned ship models constructed by the deep reinforcement learning model, and meanwhile, accumulation of simulation experience is realized in the process of the simulation test, so that the intelligent group simulation test system can provide learning requirements for the deep reinforcement learning.
Disclosure of Invention
The invention aims to solve the problems and provides a group intelligent simulation test system and method for a water surface unmanned cluster.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
the group intelligent simulation test system for the unmanned water surface clusters comprises a simulation application module, a simulation engine and a database, wherein the simulation application module comprises a scene customization module, a scene operation management module, a scene state display module and a data collection analysis module; the simulation engine comprises a perception data simulation engine, a three-dimensional visual simulation engine, a distributed interactive real-time simulation engine and a simulation interconnection interaction component;
the databases comprise a simulation database, a wanted database, an environment database, a model database and a rule database;
the scene customizing module is used for carrying out the design making and the simulation scene setting based on the simulation database and the design database, instantiating the entity model of the multi-agent into group intelligent simulation entities in the design space, editing the attribute of the simulation entities and the command communication relation between the simulation entities, and generating simulation design;
the scene operation management module is used for managing software and hardware resources of the simulation system, controlling the simulation operation process, monitoring the operation state of the simulation system and managing simulation operation data.
The scene state display module is used for displaying the wanted state potential in an image and visual way;
the data collection and analysis module is used for simulating data collection, arrangement and analysis;
the perception data simulation engine is used for simulating the perception data simulation on the entity carrier;
the three-dimensional visual simulation engine is used for simulating and generating an operation scene of the crowd-sourced intelligent entity, and constructing a high-resolution virtual environment suitable for the crowd-sourced intelligent platform, such as a virtual battlefield environment, comprising terrain, water surface, materials, light shadow and gravity effect by utilizing corresponding data of the environment database;
the distributed interactive real-time simulation engine is used for providing basic management services for all contents in a simulation scene by running the support service, including time management, object management, identification management, event management, interactive management and model management, statement management, battlefield management and the like;
the simulation interconnection interaction component is used for providing a communication physical link simulation, a discrete event simulation, a data interaction and integration simulation and a heterogeneous interconnection interaction method;
the model database at least comprises an unmanned ship model, the rule database comprises rule data for testing the capability to be tested of the intelligent group, the unmanned ship model comprises a physical model and an algorithm model, and the rule data is optimized by using experience data and mainly comprises the algorithm model.
In the above group intelligent simulation test system facing the unmanned water surface cluster, the intelligent group capability to be tested comprises any one or more of situation awareness capability, cluster communication capability, group control capability and group game capability;
the rule data comprises any one or a combination of a plurality of collaborative behavior rules, target search rules, decision rules, target threat rules and engagement rules.
In the above group intelligent simulation test system facing the unmanned water surface cluster, the model database further comprises an intelligent body model of a non-unmanned ship;
and the intelligent body model of the non-unmanned ship comprises an unmanned plane model and an unmanned vehicle model.
In the group intelligent simulation test system oriented to the unmanned water surface vehicle, the data collection and analysis module is used for collecting simulation data in real time, on one hand, the simulation scene is presented in a three-dimensional scene running state display and simulation data display mode, on the other hand, the simulation data is preprocessed, and an operator verification scheme is assisted to conduct real-time and post-hoc data analysis;
the data collection analysis module is connected to the experience pool and is also used for screening the acquired simulation data, screening the experience data in the simulation data and putting the simulation data into the experience pool for training of the unmanned ship model.
In the group intelligent simulation test system facing the unmanned water surface cluster, the simulation database comprises a simulation scheme, data classification and data results;
the wanted database comprises grouping data, deployment data, deduction management and pilot control;
the environment database comprises equipment performance, geographical environment, meteorological hydrology and space coordinates.
In the group intelligent simulation test system facing the unmanned water surface vehicle, the scene state display module performs virtual environment display, two-dimensional situation display, three-dimensional situation display and gesture playback according to display elements and display methods customized by users;
and the two-dimensional situation display includes: digital map operation, situation playback control, display content control, simulation entity information inquiry, topography information inquiry and calculation;
the three-dimensional situation display comprises: three-dimensional model management, special effect management, geomorphic management, display content control and three-dimensional playback control.
In the group intelligent simulation test system facing the unmanned water surface cluster, the scene customization module is used for generating simulation design comprising design management, grouping and configuration, design and task planning;
the scene operation management module comprises simulation resource management, simulation operation setting, simulation operation control and simulation state monitoring.
A group intelligent simulation test method for a water surface unmanned cluster comprises the following steps:
s1, a scene customizing module performs design making and simulation scene setting based on a simulation database and a design database according to demonstration requirements;
s2, determining the number of the agents as N according to the demonstration task, determining the N agents as an countermeasure mode or a matching mode, copying the N agent models, and instantiating each agent model as a group intelligent simulation entity in a wanted space;
the N intelligent body models comprise at least two unmanned ship models;
s3, the scene customization module edits the simulation entity attribute and the command communication relation between the simulation entities to generate simulation thinking taking the intelligent group of the unmanned ship as a main body;
s4, simulating virtual environments and environment perception of corresponding visual angles for each intelligent agent in the intelligent group by a simulation engine, and realizing simulation interaction among the intelligent agents;
s5, starting simulation deduction, and performing simulation operation control by a scene operation management module according to intelligent group rules in a simulation process, and displaying a desired state by a state display module;
s6, in the operation process, the data collection analysis module collects simulation data in real time to perform data visualization presentation, and meanwhile, whether experience data in the collected simulation data are screened or not is determined according to user selection, and the experience data are stored in an experience pool for training the unmanned ship.
In the above-mentioned group intelligent simulation test method facing the unmanned water surface vehicle, in step S3, the scene customization module performs grouping and configuration and task planning for N agents according to the demonstration task, so as to construct a multi-agent game countermeasure scene or a multi-agent cooperation cooperative task scene.
In the above-mentioned group intelligent simulation test method facing the unmanned water surface cluster, in step S1, a simulation demonstration is performed according to demonstration requirements to verify scene design, and a real-scene modeling is performed on application scenes including topography, landform, ground object and weather; according to the simulation scene design, a three-dimensional visual simulation engine is relied on to conduct basic situation deployment of intelligent group demonstration, and intelligent group rules are defined according to a rule database while situation deployment is conducted;
and S6, training the unmanned ship model by using experience data in the experience pool, and storing the unmanned ship model after the training into a database to update the unmanned ship model for the next simulation test.
The invention has the advantages that:
providing a simulation test system aiming at the unmanned ship cluster, instantiating a solid model of multiple intelligent agents into group intelligent simulation entities in a wanted space by a scene customization module, performing simulation test on the unmanned ship cluster, verifying the function condition of the unmanned ship under the cluster scene, and realizing unmanned ship cluster test under the complex environments such as typical sea conditions;
during simulation, the number of unmanned ship models is duplicated according to the number of unmanned ship intelligent bodies, each unmanned ship intelligent body corresponds to one unmanned ship model, each unmanned ship intelligent body is an independent individual in a simulation environment, positions, roles and tasks are compiled for each unmanned ship intelligent body through a scene customizing module, each unmanned ship intelligent body calculates and judges according to the positions, roles and tasks, so that a multi-intelligent-body cluster environment is formed, and simulation tests under a multi-intelligent-body scene are realized;
aiming at the problems of difficult autonomous control, weak environment adaptability and the like caused by a plurality of marine environment information elements and complex sea and battlefield environments, a group-based intelligent simulation platform is provided, experience data are screened from collected simulation data, and the unmanned ship based on deep reinforcement learning is trained by using the experience data generated by simulation, so that the unmanned ship can adapt to the marine environment with high complexity, and meanwhile, although the training aims at a single agent, the experience data are collected from a scene containing multiple agents, so that the training process can consider the cooperation and the countermeasure among the agents, and the adaptability of a learning algorithm to the scene of the multiple agents is improved.
Drawings
FIG. 1 is a schematic diagram of a system architecture of a group intelligent simulation test system for a water surface unmanned cluster;
FIG. 2 is a second architecture diagram of the intelligent simulation test system for the water surface unmanned cluster;
fig. 3 is a schematic diagram of a demonstration flow of the intelligent group simulation test system for the unmanned water surface cluster.
Detailed Description
The invention will be described in further detail with reference to the drawings and the detailed description.
The group intelligent simulation test system for the unmanned water surface clusters is mainly used for building, running, collecting and analyzing typical group intelligent scenes, can provide platform support for research and development and iteration of key technologies such as group intelligent situation awareness, group communication, group control, group game and the like, and can display research results of group intelligent related technologies; meanwhile, the experimental data for training the unmanned ship model can be acquired based on data collection and analysis of the simulation process, and the unmanned ship model is mainly a deep reinforcement learning network with self-iterative learning exploration capability, so that the unmanned ship model can be suitable for complex typical offshore environments of multiple intelligent agents, and the unmanned ship intelligent agent with stronger capability and closer to actual demands is obtained.
As shown in fig. 1, the system mainly comprises a simulation application module, a simulation engine and a database, wherein the simulation application module comprises a scene customization module, a scene operation management module, a scene state display module and a data collection analysis module; the simulation engine comprises a perception data simulation engine, a three-dimensional visual simulation engine, a distributed interactive real-time simulation engine and a simulation interconnection interaction component; the database provides registration, discovery, acquisition, management and use of various resources, forms a basic support of the whole system, comprises a simulation database, a wanted database, an environment database, a model database and a rule database, and provides efficient data storage management and space information abstraction and expression for intelligent group simulation.
The simulation database comprises a simulation scheme, data classification and data results; the wanted database comprises grouping data, deployment data, deduction management and pilot control; the environmental database includes geographic environments, weather hydrology, spatial coordinates, and may also include equipment and its capabilities.
The scene customization module is used for generating simulation thinking comprising thinking management, grouping and compiling, thinking and editing and task planning, wherein the grouping and compiling provides support for building a simulation environment of group intelligence; the scene customizing module is used for carrying out the design making and the simulation scene setting based on the simulation database and the design database, instantiating the entity model of the multi-agent into group intelligent simulation entities in the design space, editing the attribute of the simulation entities and the command communication relation between the simulation entities, and generating simulation design;
the scene operation management module comprises simulation resource management, simulation operation setting, simulation operation control and simulation state monitoring. The scene operation management module is responsible for transforming background things in a scene, such as the sun rises and falls according to a rule, such as simulating rainfall at a specified time, such as water bloom caused by the striking of an intelligent body. The actions and the states before and after the actions of each intelligent agent are realized by the intelligent agent through calculation, action execution and the like, so that an offshore environment game scene with multiple intelligent agents participating and each intelligent agent having their own roles is constructed.
The scene state display module is used for visually displaying the wanted situation, and can flexibly select and customize display elements and display methods, and comprises two parts of two-dimensional situation display and three-dimensional situation display. The two-dimensional situation display function includes: digital map operation, situation playback control, display content control, simulation entity information inquiry, topography information inquiry and calculation and the like. The three-dimensional situation display function comprises: three-dimensional model management, special effect management, large topography management, display content control, three-dimensional playback control and the like.
The data collection and analysis module is used for simulating data collection, arrangement and analysis, and comprises the functions of data monitoring, data preprocessing, data analysis, data visual display and the like. And the training device is connected with the experience pool and used for storing the experience data screened by the sorting analysis into the experience pool for training of the model.
The data collection analysis module collects simulation data in real time, on one hand, the simulation scene is presented in a three-dimensional scene running state display and simulation data display mode, and on the other hand, the simulation data is preprocessed, so that an operator is assisted to verify a scheme to conduct real-time and post-hoc data analysis; in yet another aspect, if the user selects the empirical data screening, the simulation data is screened, and the empirical data in the screening is put into an empirical pool for training the unmanned ship. When the data is screened, an unmanned ship is used as a target, any unmanned ship can be selected as a target, or an unmanned ship at a leading position can be selected as a target, and the empirical data of the unmanned ship, namely the action data of the unmanned ship, the perception data observed by the unmanned ship before and after each action and the perception data received from other intelligent agents, such as situation data, environment observation data and the like of each intelligent agent before and after the unmanned ship executes the current action, are screened. The navigation experience data of the unmanned ship under the complex offshore environment such as typical sea conditions are obtained in the simulation process, and the simulation is that the situation awareness capability, the group communication capability, the group control capability, the group game capability and the like are realized under the group intelligent game scene, so that the experience data comprise information, states and the like of other intelligent groups in the intelligent group, namely, other intelligent groups are combined, the training of the unmanned ship is not limited to a single intelligent body, and thus the unmanned ship which is continuously trained can adapt to the game countering scene of multiple intelligent bodies under the real complex sea environment.
The perception data simulation engine is used for simulating the perception data simulation on the entity carrier, and generates element data such as the position, the direction and the like along with the movement of the corresponding intelligent body. The maneuvering model is based on an open-source physx physical operation engine and meets the geometric relation of the three-dimensional model and the constraint requirements of the geographic environment.
The three-dimensional visual simulation engine is used for simulating and generating an operation scene of the crowd intelligent entity according to the control of the scene operation management module, and constructing a high-resolution virtual environment suitable for the crowd intelligent platform by utilizing corresponding data of an environment database. The method is suitable for sea, land and air combined simulation large scenes and a geocentric coordinate system, can construct various combat target models, interference models and the like, and has the capability of constructing multi-dimensional virtual battlefield environments such as land, sea, air, sky, electricity and the like in a 'what you see is what you get' mode.
And the distributed interactive real-time simulation engine provides interface standards, communication structures, management structures and the like for integrated simulation of each system. The system is used for providing basic management services for all contents in the simulation scene through operation support services, including time management, object management, identification management, event management, interaction management, model management, statement management and battlefield management, and realizing operation management under a unified framework.
And the simulation interconnection interaction component is used for providing communication physical link simulation, discrete event simulation, data interaction and integration simulation and heterogeneous interconnection interaction methods. The communication physical link simulation engine, the data interaction and integration simulation engine and the heterogeneous interconnection interaction engine are provided, and a foundation is provided for constructing an operation environment of group intelligent simulation. During simulation, the simulation system copies the number of unmanned ship models according to the number of unmanned ship intelligent bodies, each unmanned ship intelligent body corresponds to one unmanned ship model, each unmanned ship intelligent body is an independent individual in a simulation environment, positions, roles and tasks are allocated to each unmanned ship intelligent body through a scene customizing module, equipment is carried on each unmanned ship intelligent body, and the equipment carried on each unmanned ship can at least meet the performance required by the allocated roles and tasks. Each unmanned ship intelligent body performs own calculation and judgment according to the position, the role and the task of the unmanned ship intelligent body, communication interaction among the intelligent bodies is realized through the simulation engine, and finally respective actions are executed based on the calculation, the judgment and the interaction, so that a cluster environment of multiple intelligent bodies is formed, and a simulation test under the scene of the multiple intelligent bodies is realized.
In addition, the simulation interconnection interaction component is also provided with a discrete event simulation engine, namely, the discrete event factors of the real sea and battlefield environment are considered, so that the simulation is more comprehensive and real.
Although the unmanned ship is used as a main body in the scheme, other types of intelligent body models can be arranged in the model database besides the unmanned ship model, such as an unmanned plane model and an unmanned vehicle model, and a large sea, land and air combined simulation scene can be constructed. The unmanned ship model comprises a physical model and an algorithm model, and the training of the unmanned ship model by using the empirical data is mainly based on a deep reinforcement learning network in the algorithm model.
The rules database includes rules data for testing the ability of the intelligent community to be tested. The intelligent group to-be-tested capability comprises situation awareness capability, group communication capability, group control capability and group game capability, and the corresponding rule data comprises a collaborative behavior rule, a target search rule, a decision rule, a target threat rule, a fight rule and the like. The user can select one of the capabilities to perform simulation test, and can also select a plurality of the capabilities to perform simulation test simultaneously, and the simulation system extracts corresponding rule data according to the simulation test task and distributes the rule data to the corresponding intelligent agent.
Further, as shown in fig. 2, the system further has a standard protocol layer on the overall architecture, and the standard protocol layer mainly refers to a standard protocol to be followed when data and model construction is performed. The data standard protocol consists of a distributed simulation data standard, a coding standard, an equipment data standard, a wanted data protocol, a regular data protocol and the like. The model classification criteria consists of entity resolution, equipment entity classification criteria, and action simulation granularity. The modeling standard protocol is composed of a model interface protocol and an interactive interface protocol. And by the constraint of a standard protocol, a unified space coordinate reference system and standard data format definition are provided for data construction of an upper data layer, a functional layer and data application of an application layer.
When the unmanned aerial vehicle is put into use, the unmanned aerial vehicle model subjected to preliminary training can be used for testing various capacities of the unmanned aerial vehicle in a group intelligent scene. Meanwhile, in the simulation test process, a user can select to extract experience data of the simulation process to an experience pool at the same time, the unmanned ship model uses the experience data to perform self iteration and learning exploration, and through continuous simulation tests, the autonomous navigation capacity of the unmanned ship in a complex marine environment and a multi-agent scene is continuously enhanced, so that the unmanned ship model has a good motion control effect before a small amount or no real marine test exists. Through the provision of this platform, both unmanned ship ability's test and for unmanned ship degree of depth reinforcement study network's study can reduce the demand to marine real scene to reduce test degree of difficulty and test cost when guaranteeing the research and development effect.
Specifically, the demonstration verification effect of the simulation test system is approximately as shown in fig. 3:
s1, a scene customizing module carries out simulation demonstration to verify scene design according to demonstration requirements, wherein the scene customizing module comprises simulation design and management, simulation model design and management and the like.
Specifically, based on a simulation database and a desired database, desired production and simulation scene setting are carried out, application scenes including topography, landform, ground object and weather are subjected to real-scene modeling, basic situation deployment of intelligent group demonstration is carried out by depending on a three-dimensional visual simulation engine according to simulation scene design, intelligent group rules are defined according to a rule database while situation deployment is carried out, and rule conditions of intelligent groups in the simulation process, such as intelligent group perception, cluster communication, group control, group game and the like are defined.
S2, then, determining the number of the intelligent agents as N by a scene customizing module according to a demonstration task, determining the number of the N intelligent agents as an countermeasure mode or a matching mode, taking 8-to-8 unmanned ship games as an example, copying 16 unmanned ship models for a total of N=16 unmanned ship intelligent agents, and instantiating each unmanned ship model as a group intelligent simulation entity in a wanted space.
S3, a scene customization module is used for grouping and collocating 16 intelligent agents and task planning according to demonstration tasks, editing simulation entity attributes and command communication relations among the simulation entities, equipping each intelligent agent on the simulation entity according to roles, tasks and the like, arranging initial positions of each intelligent agent to prepare for entering game demonstration, and generating simulation thinking based on intelligent groups of multiple unmanned boats so as to construct a multi-intelligent-agent game countermeasure scene.
S4, simulating virtual environments and environment perception of corresponding visual angles for each intelligent agent in the intelligent group by a simulation engine, and realizing simulation interaction among the intelligent agents; the three-dimensional visual simulation shows simulation environments of corresponding visual angles for the intelligent agents in the two groups, and the perceptron simulation engine simulates the perceptrons of the intelligent agents in the two groups to obtain perceptron simulation data under the demonstration environment;
s5, starting simulation deduction, and performing simulation operation control by the scene operation management module according to intelligent group rules in the simulation process, and displaying the wanted state by the state display module.
In the deduction process, the data collection analysis module collects simulation data in real time, on one hand, the simulation data are presented to the spectator in a three-dimensional scene running state display and simulation data display mode, and on the other hand, the simulation data are preprocessed, so that an operator verification scheme is assisted to conduct real-time and post-hoc data analysis. Meanwhile, if the user sets experience data screening, the experience data in the collected simulation data are screened and stored in an experience pool. And then training the deep reinforcement learning network of the unmanned ship by using the experience data in the experience pool, and storing the unmanned ship after the training into a database to update the unmanned ship model for the next simulation test, thereby continuously improving the performance of the unmanned ship model.
The simulation system can be used for carrying out capability test on situation awareness, trunking communication, group control, group game and the like of the unmanned ship in the intelligent group, and is not limited to testing of a single intelligent agent. The method can also be used for accumulating experience data, thereby continuously improving the autonomous navigation capability of the unmanned ship model taking the deep reinforcement learning network as a learning model, and enabling the unmanned ship after continuous research, development, experiment, iteration and upgrading to be suitable for multi-agent game scenes under high-complexity marine environment.
The specific embodiments described herein are offered by way of example only to illustrate the spirit of the invention. Those skilled in the art may make various modifications or additions to the described embodiments or substitutions thereof without departing from the spirit of the invention or exceeding the scope of the invention as defined in the accompanying claims.
Claims (10)
1. The group intelligent simulation test system for the unmanned water surface clusters is characterized by comprising a simulation application module, a simulation engine and a database, wherein the simulation application module comprises a scene customization module, a scene operation management module, a scene state display module and a data collection analysis module; the simulation engine comprises a perception data simulation engine, a three-dimensional visual simulation engine, a distributed interactive real-time simulation engine and a simulation interconnection interaction component;
the databases comprise a simulation database, a wanted database, an environment database, a model database and a rule database;
the scene customizing module is used for carrying out the design making and the simulation scene setting based on the simulation database and the design database, instantiating the entity model of the multi-agent into group intelligent simulation entities in the design space, editing the attribute of the simulation entities and the command communication relation between the simulation entities, and generating simulation design;
the scene operation management module is used for managing software and hardware resources of the simulation system, controlling the simulation operation process, monitoring the operation state of the simulation system and managing simulation operation data.
The scene state display module is used for displaying the wanted state potential in an image and visual way;
the data collection and analysis module is used for simulating data collection, arrangement and analysis;
the perception data simulation engine is used for simulating the perception data simulation on the entity carrier;
the three-dimensional visual simulation engine is used for simulating and generating an operation scene of the crowd intelligent entity, and constructing a high-resolution virtual environment suitable for the crowd intelligent platform by utilizing corresponding data of the environment database, wherein the high-resolution virtual environment comprises terrain, water surface, materials, light shadow and gravity effect;
the distributed interactive real-time simulation engine is used for providing basic management services for all contents in a simulation scene by running the support service, including time management, object management, identification management, event management, interaction management and model management;
the simulation interconnection interaction component is used for providing a communication physical link simulation, a discrete event simulation, a data interaction and integration simulation and a heterogeneous interconnection interaction method;
the model database at least comprises an unmanned ship model, and the rule database comprises rule data for testing the capability to be tested of the intelligent group.
2. The water surface unmanned cluster-oriented group intelligent simulation test system according to claim 1, wherein the intelligent group to-be-tested capability comprises any one or a combination of a plurality of situation awareness capability, cluster communication capability, group control capability and group game capability;
the rule data comprises any one or a combination of a plurality of collaborative behavior rules, target search rules, decision rules, target threat rules and engagement rules.
3. The water surface unmanned cluster-oriented population intelligent simulation test system of claim 2, wherein the model database further comprises an agent model of a non-unmanned boat;
and the intelligent body model of the non-unmanned ship comprises an unmanned plane model and an unmanned vehicle model.
4. The intelligent simulation test system for the water surface unmanned cluster group according to claim 1, wherein the data collection and analysis module is used for collecting simulation data in real time, presenting a simulation scene in a three-dimensional scene running state display and simulation data display mode, preprocessing the simulation data on the other hand, and assisting an operator verification scheme to analyze real-time and post-hoc data; and screening the simulation data, namely screening the experience data, and putting the simulation data into an experience pool for training the unmanned ship.
5. The intelligent simulation test system for the water surface unmanned cluster according to claim 1, wherein the simulation database comprises a simulation scheme, data classification and data result;
the wanted database comprises grouping data, deployment data, deduction management and pilot control;
the environment database comprises geographic environment, meteorological hydrology and space coordinates.
6. The intelligent simulation test system for the water surface unmanned cluster group, according to claim 1, wherein the scene state display module performs virtual environment display, two-dimensional situation display, three-dimensional situation display and gesture playback according to display elements and display methods customized by users;
and the two-dimensional situation display includes: digital map operation, situation playback control, display content control, simulation entity information inquiry, topography information inquiry and calculation;
the three-dimensional situation display comprises: three-dimensional model management, special effect management, geomorphic management, display content control and three-dimensional playback control.
7. The water-surface-oriented unmanned cluster group intelligent simulation test system according to claim 1, wherein the scene customization module is configured to generate simulation assumptions including a thinking management, a grouping compilation, a thinking compilation, and a task planning;
the scene operation management module comprises simulation resource management, simulation operation setting, simulation operation control and simulation state monitoring.
8. The intelligent group simulation test method for the unmanned water surface clusters is characterized by comprising the following steps of:
s1, according to demonstration requirements, carrying out design making and simulation scene setting based on a simulation database and a design database;
s2, determining the number of the agents as N according to a demonstration task, determining the N agents as an countermeasure mode or a matching mode, copying the N agent models, and instantiating each agent model as a group intelligent simulation entity in a wanted space;
the N intelligent body models comprise at least two unmanned ship models;
s3, editing the attribute of the simulation entity and the command communication relation between the simulation entities to generate a simulation design based on the intelligent group of the unmanned ships;
s4, simulating virtual environments and environment perception of corresponding visual angles for each intelligent agent in the intelligent group by a simulation engine, and realizing simulation interaction among the intelligent agents;
s5, starting simulation deduction, and performing simulation operation control by a scene operation management module according to intelligent group rules in a simulation process, and displaying a desired state by a state display module;
in the running process, the data collection analysis module collects simulation data in real time to perform data visual presentation, and meanwhile, empirical data in the collected simulation data are screened and stored into an empirical pool for training the unmanned ship.
9. The method for intelligent simulation test of a water surface unmanned cluster-oriented community according to claim 8, wherein in step S3, the scene customization module performs grouping and task planning for N agents according to the demonstration task to construct a multi-agent game countermeasure scene or a multi-agent cooperation cooperative task scene.
10. The intelligent simulation test method for the water surface unmanned cluster group, according to claim 9, is characterized in that in step S1, simulation demonstration is performed according to demonstration requirements to verify scene design, and application scenes including topography, landform, ground feature and weather are subjected to real-scene modeling; according to the simulation scene design, a three-dimensional visual simulation engine is relied on to conduct basic situation deployment of intelligent group demonstration, and intelligent group rules are defined according to a rule database while situation deployment is conducted;
training the unmanned ship by using experience data in the experience pool, and storing the unmanned ship after the training in a database to update the unmanned ship model for the next simulation test.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310193282.0A CN116167230A (en) | 2023-02-27 | 2023-02-27 | Group intelligent simulation test system and method for water surface unmanned cluster |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310193282.0A CN116167230A (en) | 2023-02-27 | 2023-02-27 | Group intelligent simulation test system and method for water surface unmanned cluster |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116167230A true CN116167230A (en) | 2023-05-26 |
Family
ID=86421820
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310193282.0A Withdrawn CN116167230A (en) | 2023-02-27 | 2023-02-27 | Group intelligent simulation test system and method for water surface unmanned cluster |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116167230A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116842758A (en) * | 2023-08-28 | 2023-10-03 | 中国民航管理干部学院 | Simulation platform and method for civil unmanned aerial vehicle air traffic service algorithm verification |
| CN117669776A (en) * | 2024-01-31 | 2024-03-08 | 北京云中盖娅科技有限公司 | Combined simulation system and method for sea, land and air clusters |
-
2023
- 2023-02-27 CN CN202310193282.0A patent/CN116167230A/en not_active Withdrawn
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116842758A (en) * | 2023-08-28 | 2023-10-03 | 中国民航管理干部学院 | Simulation platform and method for civil unmanned aerial vehicle air traffic service algorithm verification |
| CN116842758B (en) * | 2023-08-28 | 2024-03-19 | 中国民航管理干部学院 | Simulation platform and method for civil unmanned aerial vehicle air traffic service algorithm verification |
| CN117669776A (en) * | 2024-01-31 | 2024-03-08 | 北京云中盖娅科技有限公司 | Combined simulation system and method for sea, land and air clusters |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116167230A (en) | Group intelligent simulation test system and method for water surface unmanned cluster | |
| CN110694256A (en) | Novel emergency computer war game deduction system and method | |
| Zhu et al. | A knowledge-based approach to data integration for soil mapping | |
| CN106845032B (en) | The construction method of multimode navigation three-dimensional dynamic visual simulation platform | |
| Odonkor et al. | Distributed operation of collaborating unmanned aerial vehicles for time-sensitive oil spill mapping | |
| CN109597839B (en) | Data mining method based on avionic combat situation | |
| CN112307622A (en) | Autonomous planning system and planning method for generating military forces by computer | |
| CN101118654A (en) | Machine vision computer simulation emulation system based on sensor network | |
| CN114882759B (en) | Virtual-real hybrid integrated simulation intelligent ship multichannel interaction simulation system and method | |
| CN115525769B (en) | Global-oriented battlefield environmental data organization method and related device | |
| CN109541960A (en) | A kind of system and method for the confrontation of aircraft digital battlefield | |
| CN116522570A (en) | Intelligent unmanned cluster system area coverage relay communication application simulation and test system | |
| CN107748502A (en) | The passive spatial perception exchange method of entity in operation emulation based on discrete event | |
| CN116820121B (en) | Unmanned aerial vehicle group joint investigation strategy generation method and terminal | |
| CN109238271B (en) | Line fitting method based on time | |
| Villanueva et al. | On building support of digital twin concept for smart spaces | |
| Dompke | Computer generated forces-background, definition and basic technologies | |
| CN114662213A (en) | Model-based visual missile defense penetration probability verification method and device | |
| CN110705021B (en) | Data-driven test driving method | |
| Biswal et al. | Mobile-Based Weather Forecasting and Visualization Using Augmented Reality | |
| Chilkunda et al. | UAV-based scenario builder and physical testing platform for autonomous vehicles | |
| Gorsich | The use of gaming engines for design requirements | |
| CN117787091A (en) | Simulation method and device for target object | |
| Campos | Image Manipulation and Classification: An Application to Fire Detection | |
| Chen et al. | Design of Simulation Experimental Platform for Airport Bird Control Linkage System Based on Improved YOLOv5 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20230526 |
|
| WW01 | Invention patent application withdrawn after publication |