CN109636699A - A kind of unsupervised intellectualized battle deduction system based on deeply study - Google Patents
A kind of unsupervised intellectualized battle deduction system based on deeply study Download PDFInfo
- Publication number
- CN109636699A CN109636699A CN201811313124.XA CN201811313124A CN109636699A CN 109636699 A CN109636699 A CN 109636699A CN 201811313124 A CN201811313124 A CN 201811313124A CN 109636699 A CN109636699 A CN 109636699A
- Authority
- CN
- China
- Prior art keywords
- confrontation
- simulation engine
- deeply
- training pattern
- platform
- 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.)
- Pending
Links
- 238000012549 training Methods 0.000 claims abstract description 110
- 238000004088 simulation Methods 0.000 claims abstract description 91
- 230000013016 learning Effects 0.000 claims abstract description 74
- 238000013523 data management Methods 0.000 claims description 41
- 230000007613 environmental effect Effects 0.000 claims description 31
- 238000000034 method Methods 0.000 claims description 22
- 238000012800 visualization Methods 0.000 claims description 19
- 230000008676 import Effects 0.000 claims description 17
- 230000033001 locomotion Effects 0.000 claims description 17
- 230000008569 process Effects 0.000 claims description 16
- 238000004458 analytical method Methods 0.000 claims description 13
- 230000009471 action Effects 0.000 claims description 11
- 230000000694 effects Effects 0.000 claims description 11
- 238000004422 calculation algorithm Methods 0.000 abstract description 23
- 230000000007 visual effect Effects 0.000 abstract description 8
- 238000013135 deep learning Methods 0.000 abstract description 4
- 238000005457 optimization Methods 0.000 abstract description 3
- 230000013011 mating Effects 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 10
- 238000005516 engineering process Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 238000007726 management method Methods 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 239000003990 capacitor Substances 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000001427 coherent effect Effects 0.000 description 1
- 230000002301 combined effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000010006 flight Effects 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003032 molecular docking Methods 0.000 description 1
- 210000004218 nerve net Anatomy 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
Landscapes
- Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of unsupervised intellectualized battle deduction systems based on deeply study, the intellectualized battle decision of the system is fed back by the training pattern in deeply learning platform, have coordination and mating capability between multiple agent, system independent learning ability and Continuous optimization ability is substantially improved；Fighting terminal supports manual operation and the operation of machine autonomous learning, it can be achieved that man-machine confrontation exports high-quality training sample data；It can provide the first visual angle and overall situation visual angle to show, user can directly carry out operation scenario simulation training by the first visual angle and global visual angle and discs is deduced；All environment configurations definition have generalization characteristic, and consider that the fighting capacity of battle both sides is balanced, and intellectualized algorithm is made to have Tactic selection space, to be adapted to a variety of deep learning algorithms and equipment model, and fight terminal quantity in system and support extension.
Description
Technical field
The invention belongs to fight against deducing technical field more particularly to a kind of unsupervised intelligence work based on deeply study
War deduction system.
Background technique
Intellectualized battle, which refers to, to be realized on the basis of information fighting with intellectual weapon and means with efficient commanderization, strike
Precision, operation automation, the high-tech pattern of operation that behavior intelligence is purport.In essence, intellectualized battle is people
Radiation and extension of the intelligence to information battlefield and weapon system.In terms of battle order and means, intellectualized battle includes intelligentized
The Attack Defence of command control warfare and smart weapon.The outstanding feature of intellectualized battle is smart weapon group and intelligent operation
The appearance of means.
To solve the above problems, devising the related operation deduction system based on expert system both at home and abroad, main thought is
The decision under specific situation is exported using the rule base of expert system, while emulating the game under scene between ourselves and the enemy.At present
The common operation deduction system based on expert system is based primarily upon experts database or traditional genetic algorithm etc. and carries out decision, passes through
Experts database or the new algorithmic approach enriched constantly promote the level of intelligence of operation deduction system, but hoisting power is limited.
With the rise of artificial intelligence technology, GPU server construction deep learning platform can use, using deeply
Learning algorithm by constantly trained and model iteration lifting system level of intelligence, and realizes independent learning ability and mostly intelligently
Cooperation between body.
Intensified learning is a branch in artificial intelligence machine learning areas, is merely able to for controlling one in some environment
Under uninfluenced intelligent body, by the interaction between environment, including perception and receive awards, and continuously improve its row
For, and the purpose of intensified learning, exactly select a series of action to maximize following reward.Deeply learns nerve net
The ability that network extracts complicated high dimensional data feature incorporates intensified learning, and data are transformed into low-dimensional feature, convenient at intensified learning
Reason.It is well known that military decision-making is the most complicated, activity most challenged in military field, and thereby promotes military auxiliary and determine
The generation and development of plan support technology.It is contemplated that being constantly progressive for deeply learning art will be to military intelligence auxiliary
Decision domain generates deep and great influence.
Summary of the invention
The purpose of the present invention is to provide a kind of unsupervised intellectualized battle deduction systems based on deeply study, improve
Algorithm, engine and the process of operation deduction system of the tradition based on experts database, improve the level of intelligence and learning ability of intelligent body,
Significantly promote the level of intelligence of deduction system.
To achieve the above object, the technical solution used in the present invention are as follows:
A kind of unsupervised intellectualized battle deduction system based on deeply study, the nothing based on deeply study
Supervising intellectualized battle deduction system includes that confrontation terminal, operation simulation engine, deeply learning platform and data management are flat
Platform is provided with equipment model library and model of place library in the operation simulation engine, is provided with sample in the data management platform
Database and training pattern library, in which:
The confrontation terminal, for connecting the input interface of the operation simulation engine；
The operation simulation engine, the operational environment for being inputted according to device end define information, from equipment model library
Equipment model relevant to operational environment is imported, model of place relevant to operational environment is imported from model of place library, completes just
Beginningization, the operational environment include combat mode, combat unit and operation scene；
And when combat mode is training mode, operation game between ourselves and the enemy is carried out based on timeslice and is deduced, it will be current
Environmental information is sent to the deeply learning platform, executes single step according to the movement feedback of deeply learning platform output
Decision or multistep decision, and update environmental information；
It is also used to judge currently whether meet the flat relationship of victory or defeat or termination condition according to predefined rule, if current meet victory
When bearing flat relationship or termination condition, output, which is fought, deduces analysis report, and saves Campaign Process as sample data to described
The sample database of data management platform completes training；Otherwise continue game deduction of fighting between ourselves and the enemy, until meeting victory
Bear flat relationship or termination condition；
The deeply learning platform, for importing training pattern from the training pattern library of data management platform, and
In conjunction with the environmental information that the operation simulation engine is sent, Xiang Zuozhan simulation engine output action feedback；It is also used to update operation
The corresponding training pattern of unit, and operation game deduce after output training pattern to data management platform training pattern
It is saved in library；
The data management platform, for saving the training pattern and the work that the deeply learning platform exports
The sample data that simulation engine of fighting exports；It is also used to export training pattern to the deeply learning platform.
Preferably, the environmental information includes: quantity, location information, current state, mutual spacing between ourselves and the enemy.
Preferably, the operation that the single step decision includes combat unit acts；The multistep decision includes to fight
Multiple movements that fight one battle after another of unit.
Preferably, the confrontation terminal is as device end, by the input interface of the operation simulation engine to institute
It states operation simulation engine input operational environment and defines information.
Preferably, the combat mode of the operation simulation engine further includes simulation model, the simulation model includes people
Machine confrontation and the confrontation of machine machine；
The confrontation terminal, also provides for three-dimensional visualization interface, and the operation is presented in visualization under simulation model
The environmental information of simulation engine；Also provide for confrontation operation interface, in man-machine confrontation by confrontation operation interface artificially to
The operation simulation engine input, which is fought, to be operated.
Preferably, the unsupervised intelligence based on deeply study is made when the combat mode is that machine machine is fought
War deduction system, performs the following operations:
The operation simulation engine, for accessing the three-dimensional visualization interface of the confrontation terminal, and according to device end
The operational environment of input defines information, equipment model relevant to operational environment is imported from equipment model library, from model of place library
Model of place relevant to operational environment is imported, training pattern is imported from the training pattern library of data management platform, is completed just
Beginningization；And operation game between ourselves and the enemy is carried out based on timeslice and is deduced, current environmental information is sent to the deeply
Learning platform executes single step decision according to the movement feedback of deeply learning platform output, and updates ring according to single step decision
Effect is presented in border information, the visualization for refreshing the confrontation terminal；
It is also used to judge currently whether meet the flat relationship of victory or defeat or termination condition according to predefined rule, if current meet victory
When bearing flat relationship or termination condition, output, which is fought, deduces analysis report, and saves Campaign Process as sample data to described
The sample database of data management platform completes emulation；Otherwise continue game deduction of fighting between ourselves and the enemy, until meeting victory
Bear flat relationship or termination condition；
The deeply learning platform, for importing training pattern from the training pattern library of data management platform, and
In conjunction with the environmental information that the operation simulation engine is sent, Xiang Zuozhan simulation engine output action feedback；
The data management platform, for saving the sample data of the operation simulation engine output；It is also used to described
Deeply learning platform and the operation simulation engine export training pattern.
Preferably, the unsupervised intelligence based on deeply study is made when the combat mode is man-machine confrontation
War deduction system, performs the following operations:
The operation simulation engine, for accessing the three-dimensional visualization interface and confrontation operation interface of the confrontation terminal,
And information is defined according to the operational environment that device end inputs, assembling die relevant to operational environment is imported from equipment model library
Type imports model of place relevant to operational environment from model of place library, imports from the training pattern library of data management platform
Training pattern completes initialization；And operated according to the operation of confrontation terminal, operation game between ourselves and the enemy is carried out based on timeslice and is pushed away
It drills, current environmental information is sent to the deeply learning platform, the movement exported according to deeply learning platform
Feedback executes single step decision, while receiving the operation operation for fighting terminal input, and according to single step decision and operation of fighting
Environmental information is updated, the effect of visualization of the confrontation terminal is refreshed；
It is also used to judge currently whether meet the flat relationship of victory or defeat or termination condition according to predefined rule, if current meet victory
When bearing flat relationship or termination condition, output, which is fought, deduces analysis report, and saves Campaign Process as sample data to described
The sample database of data management platform completes emulation；Otherwise continue game deduction of fighting between ourselves and the enemy, until meeting victory
Bear flat relationship or termination condition；
The deeply learning platform, for importing training pattern from the training pattern library of data management platform, and
In conjunction with the environmental information that the operation simulation engine is sent, Xiang Zuozhan simulation engine output action feedback；
The data management platform, for saving the sample data of the operation simulation engine output；It is also used to described
Deeply learning platform and the operation simulation engine export training pattern.
A kind of unsupervised intellectualized battle deduction system based on deeply study proposed by the present invention, intellectualized battle decision
It is fed back by the training pattern in deeply learning platform, has coordination and mating capability between multiple agent, substantially
Lifting system independent learning ability and Continuous optimization ability；It fights terminal and supports manual operation and the operation of machine autonomous learning, it can
Realize that man-machine confrontation exports high-quality training sample data；It can provide the first visual angle and global visual angle to show, user passes through the first view
Angle and global visual angle can directly carry out operation scenario simulation training and discs is deduced；All environment configurations definition have generalization special
Property, and consider that the fighting capacity of battle both sides is balanced, so that intellectualized algorithm is had Tactic selection space, is calculated to be adapted to a variety of deep learnings
Method and equipment model, and fight terminal quantity in system and support extension.
Detailed description of the invention
Fig. 1 is a kind of embodiment framework of the unsupervised intellectualized battle deduction system of the invention based on deeply study
Figure；
Fig. 2 is a kind of embodiment process of the unsupervised intellectualized battle deduction system of the invention based on deeply study
Block diagram；
Fig. 3 is a kind of embodiment operational effect figure that terminal is fought when fighting scene the invention shows air battle；
Fig. 4 is that a kind of embodiment of confrontation terminal when fighting scene the invention shows air battle runs global situation map.
Specific embodiment
Technical solution of the present invention is described in further details with reference to the accompanying drawings and examples, following embodiment is not constituted
Limitation of the invention.
As shown in Figure 1, being the present embodiment provides a kind of unsupervised intellectualized battle deduction system based on deeply study
The display panel control united when carrying out deployed with devices using confrontation terminal as intellectualized battle deduction system, deployment emulation simulator are soft
Part provides confrontation operation interface and three-dimensional visualization interface, and the quantity that terminal is fought in system supports extension.
Emulator using work station as intellectualized battle deduction system disposes operation simulation engine；It is serviced using GPU
Training equipment of the device as intellectualized battle deduction system, deployment depth intensified learning platform.
Storage equipment using storage dish battle array as intellectualized battle deduction system disposes data management platform, and data pipe
Sample database and training pattern library are equipped in platform, provide training sample data import and export, the function such as training pattern management
Energy.
Each equipment in the present embodiment passes through exchange network from each other and interconnects, it should be noted that each equipment mutually it
Between can using be wirelessly connected, electrical connection or mechanical connection.
As seen from the figure, it is additionally provided with equipment model library and model of place library in intellectualized battle deduction system, equips model library
Equipment model is created to combat unit, and the functions such as equipment model management are provided.Combat unit includes operation coherent element, such as
Aircraft, radar, guided missile etc..Model of place library models scene, provides the functions such as model of place management.
The equipment model library and model of place library can share a storage equipment with data management platform, can also be with
It is independently arranged in storage dish battle array, provides equipment model information and model of place information, the present embodiment for operation simulation engine
In, memory headroom is opened up in operation simulation engine, for equipment model library and model of place library to be arranged.
Wherein, the operation simulation engine of work station is designed as a Combat simulation platform unrelated with algorithm, realization side
Formula is the software package run under Linux or windows platform, and the implementation of the present embodiment is to run under Linux platform
Python software package.Operation simulation engine is provided based on decision making algorithms interfaces such as deep learnings, and passes through the decision making algorithm interface
Connection is established with the deeply learning platform, the key property of operation simulation engine is to provide: operational environment definition, equipment
The sample data importing and export, the importing of training pattern library, the access of more training platform decisions that model library is imported, fought, fight and advise
It then defines, deduces the functions such as analysis report.
In intellectualized battle deduction, the parameter configuration of operational environment and the application of algorithm model are also closely connected, if
The side that fights in belligerent Board Lot, aspect of performance has landslide, and Operation Target is easy to reach, then another party will be complete
It wins victory completely without method, the intelligent decision means under the clearly demarcated operational environment of this victory or defeat, including deeply learning algorithm
It is difficult have performance space.Therefore, it is necessary to comprehensively consider the construction feature of battlefield conditions and simulated environment in actual demand, choose
Suitable operational environment, so that the decision of battle both sides has choice, to allow algorithm decision that can embody value.
In addition, backstage training equipment of the deeply learning platform as intellectualized battle deduction system on GPU server,
The agent model that can use deeply learning algorithm training combat unit may also be combined with tactical training, confrontation examination etc.
The common intensive training combat unit of external samples data including practical combat data (common for practical flight data), output
Training pattern is to training pattern library.
Deeply learning platform is related to DQN, DDPG, the mainstreams nitrification enhancement such as A3C, PPO.These algorithms need to be into
Row transformation applies to intellectualized battle and deduces scene, and the state and control variable of simulated environment are defined as including that will output and input,
And the reward function under each state is calculated according to the parameter configuration of operational environment and victory or defeat condition, realize algorithm and simulated environment
Docking.The nitrification enhancement used in the present embodiment is DQN.
The present embodiment carries out a large amount of training tests in deeply learning platform according to specific algorithm, verify various algorithms and
Combined effect and assessment.Wherein the central principle of algorithm can be summarized as: use deep neural network approximate value function (Q function)
Or strategic function changes its state, and rewarded, passes through reward by controlling movement and the environment interaction of Intelligent target body
Recovery value/strategic function parameter, reaches training goal.
Based on the central principle of the algorithm, promote the core objective of intellectualized battle deduction system: realize algorithm training with
Intelligent decision.For this reason, it may be necessary to which the state variable space and the control variable space to simulated environment screen, obtain being suitable for calculating
Method input, output state space S={ s, s2 ..., sn } and control space C={ c1, c2 ..., cm }.
Certain state, the screening for controlling space need to consider many factors: firstly, state space can completely react battlefield
Situation is lost without information, and is controlled space and be required to complete combat unit various actions movement, while in currently emulation frame
It can be realized in frame stable based on the iterative control of timeslice；Secondly, under the premise of meeting above-mentioned condition, state space with
The dimension in control space needs to simplify as far as possible, controls the calculation amount of algorithm training in the achievable dimension of computing capability；This
Outside, the parameter configuration of operational environment can also screen space and have an impact, if environment setting is made that simplification in one aspect
Processing, corresponding state can also carry out corresponding simplify with control parameter.
As shown in connection with fig. 2, the work of each equipment in intellectualized battle deduction system is further illustrated:
Under the deployed with devices basis of intellectualized battle deduction system, Armament modeling and operation scene modeling are carried out in advance,
It saves respectively into equipment model library and model of place library, simulation engine to be fought is selected.Meanwhile rule of engagement modeling is carried out,
Judge to fight as operation simulation engine and currently whether meets the foundation of the flat relationship of victory or defeat or termination condition.
Before deduction of fighting starts, this operational environment deduced of fighting first is defined, the operational environment includes operation mould
Formula, combat unit and operation scene are used as device end using confrontation terminal in the present embodiment, are emulated and drawn by the operation
The input interface held up defines information to operation simulation engine input operational environment.
Further, confrontation terminal also provides three-dimensional visualization interface, and the operation is presented in visualization under simulation model
The environmental information of simulation engine.Confrontation operation interface is provided simultaneously, in man-machine confrontation by confrontation operation interface artificially to institute
State the input operation operation of operation simulation engine.
It should be noted that operational environment define information be not limited only to using confrontation terminal input, can also by with work
The device end or background server input operational environment of the input interface connection of war simulation engine define information.
The optional combat mode of the intellectualized battle deduction system of the present embodiment includes training mode and simulation model, and is emulated
Mode includes man-machine confrontation and the confrontation of machine machine.
S1, when combat mode be training mode when, based on deeply study unsupervised intellectualized battle deduction system, hold
The following operation of row:
S11, operation simulation engine define information according to the operational environment that device end inputs, from equipment model library import with
The relevant equipment model of operational environment imports model of place relevant to operational environment from model of place library, completes initialization.Its
In, operational environment includes combat mode, combat unit and operation scene；
Meanwhile deeply learning platform is led from the training pattern library of data management platform according to algorithm configuration information
Enter training pattern, completes initialization.
S12, operation simulation engine are based on timeslice and carry out the deduction of operation game between ourselves and the enemy, and current environmental information is sent out
It send to the deeply learning platform, single step decision or multistep is executed according to the movement feedback of deeply learning platform output
Decision, and update environmental information.
The environmental information that S13, deeply learning platform combination operation simulation engine are sent, the output of Xiang Zuozhan simulation engine
Movement feedback.
Whether S14, operation simulation engine currently meet the flat relationship of victory or defeat or termination condition according to predefined rule judgement, if
When currently meeting the flat relationship of victory or defeat or termination condition, output, which is fought, deduces analysis report, and using Campaign Process as sample data
It saves to the sample database of the data management platform, completes training；Otherwise continue game deduction of fighting between ourselves and the enemy,
Until meeting the flat relationship of victory or defeat or termination condition.
The sample data that S15, deeply learning platform are deduced by largely operation game, and combination tactical training,
External samples data including the practical flights data such as confrontation examination promote intelligent decision level, according to work after training
Performance of the war unit in this operation game deduction updates the corresponding training pattern of combat unit, and deduces in operation game
After export training pattern saved into the training pattern library of data management platform.
In above-mentioned training mode, data management platform is when deeply learning platform is initialized to its output training
Model, and the training pattern and operation simulation engine that deeply learning platform exports are saved after operation game is deduced
The sample data of output.
It should be noted that the training pattern saved in data management platform equally has guarantor in deeply learning platform
It deposits, dual memory greatly reduces data loss rate.Therefore in training mode, training pattern can be led from training pattern library
Enter, the training pattern being stored in deeply learning platform can also be called directly.
Wherein, environmental information includes: quantity, location information, current state between ourselves and the enemy, and in operational environment
One combat unit is center combat unit, the operation relationship of remaining combat unit and the center combat unit, operation relationship
Such as enemy and we's relationship, relative distance, degree of danger, whether in mutual operational state etc..
And in order to make training be more nearly true operation scene, what operation simulation engine was executed in training mode can be with
It is to be also possible to (as be based on based on the multistep decision of multiple movements that fight one battle after another based on the single step decision of operation movement
Tactics combinative movement executes multistep decision), to improve the flexibility of the execution of combat unit, the training of training pattern is imitated
Fruit is more preferable, also increases operation difficulty, has to the level of intelligence of deeply learning platform and preferably promotes effect.
S2, when combat mode be machine machine fight when, based on deeply study unsupervised intellectualized battle deduction system, hold
The following operation of row:
S21, operation simulation engine access the three-dimensional visualization interface of the confrontation terminal, and according to device end input
Operational environment defines information, imports relevant to operational environment equipment model from equipment model library, imported from model of place library and
The relevant model of place of operational environment imports training pattern from the training pattern library of data management platform, completes initialization；
Meanwhile deeply learning platform imports training pattern from the training pattern library of data management platform, completes just
Beginningization.
S22, operation simulation engine are based on timeslice and carry out the deduction of operation game between ourselves and the enemy, and current environmental information is sent out
It send to the deeply learning platform, single step decision, and root is executed according to the movement feedback of deeply learning platform output
Environmental information is updated according to single step decision, effect is presented in the visualization for refreshing the confrontation terminal.
The environmental information that S23, deeply learning platform are sent in conjunction with the operation simulation engine, Xiang Zuozhan simulation engine
Output action feedback.
Whether S24, operation simulation engine currently meet the flat relationship of victory or defeat or termination condition according to predefined rule judgement, if
When currently meeting the flat relationship of victory or defeat or termination condition, output, which is fought, deduces analysis report, and using Campaign Process as sample data
It saves to the sample database of the data management platform, completes emulation；Otherwise continue game deduction of fighting between ourselves and the enemy,
Until meeting the flat relationship of victory or defeat or termination condition.
In above-mentioned machine machine confrontation, data management platform is instructed to operation simulation engine and the output of deeply learning platform
Practice model, and saves the sample data of operation simulation engine output after operation game is deduced.
Machine machine fights the level of intelligence that can be used for detecting deeply learning platform, and by largely confrontation data to instruction
Practice model and makes analysis and judgment.
S3, when combat mode be man-machine confrontation when, it is described based on deeply study unsupervised intellectualized battle deduce system
System, performs the following operations:
The three-dimensional visualization interface and confrontation operation interface of S31, operation simulation engine access confrontation terminal, and according to equipment
The operational environment of terminal input defines information, equipment model relevant to operational environment is imported from equipment model library, from scene mould
Type library imports model of place relevant to operational environment, imports training pattern from the training pattern library of data management platform, complete
At initialization；
Meanwhile deeply learning platform imports training pattern from the training pattern library of data management platform, completes just
Beginningization.
S32, operation simulation engine are operated according to the operation of confrontation terminal, are fought between ourselves and the enemy game based on timeslice
It deduces, current environmental information is sent to the deeply learning platform, according to the dynamic of deeply learning platform output
Make feedback and execute single step decision, while receiving the operation operation for fighting terminal input, and according to single step decision and the behaviour that fights
Make update environmental information, refreshes the effect of visualization of the confrontation terminal.
The environmental information that S33, deeply learning platform combination operation simulation engine are sent, the output of Xiang Zuozhan simulation engine
Movement feedback.
Whether S34, operation simulation engine currently meet the flat relationship of victory or defeat or termination condition according to predefined rule judgement, if
When currently meeting the flat relationship of victory or defeat or termination condition, output, which is fought, deduces analysis report, and using Campaign Process as sample data
It saves to the sample database of the data management platform, completes emulation；Otherwise continue game deduction of fighting between ourselves and the enemy,
Until meeting the flat relationship of victory or defeat or termination condition.
In above-mentioned man-machine confrontation, data management platform is instructed to operation simulation engine and the output of deeply learning platform
Practice model, and saves the sample data of operation simulation engine output after operation game is deduced.It should be noted that imitative
When true mode, training pattern imported into deeply learning platform and operation simulation engine simultaneously from training pattern library, in depth
It cooperates jointly as intelligent decision basis with environmental information in intensified learning platform, as single with operation in operation simulation engine
Member matching is spare.By above-mentioned deducing maneuvers process as it can be seen that operation simulation engine can export after each training or emulation
Portion, which is fought, deduces analysis report, which deduces analysis report and mainly realize that being based on belligerent deduction process and result carries out winning rate
The statistical analysis such as statistics, Armament Demonstration and tactics research, convenient for the understanding to this deducing maneuvers process, and to such operation
Deduce the global observing of long-term operation situation.
Further, sample data of the output into data management platform after each training or emulation, can be again
It imports in operation simulation engine, for playing back Campaign Process.In each deducing maneuvers process, combat unit is according to deeply
The intelligent decision of learning platform executes single step decision or multistep decision, is needed during training by the row of agent (intelligent body)
Dynamic selection is converted into several discrete values, as the action value taken in next step.By for air battle fights scene, using working as
The parameters such as position, angle, steering angle, the distance of preceding both sides are inputted as neural network, are built deep neural network and are trained,
The probability value that output acts for next step, each step all select the movement so that maximum probability.And (depth is strong to improved DQN
Chemical learning method), if Prioritized Replay and Dueling Network are verified, it is final obtain it is reasonable as a result,
And it is cured as training pattern.Aircraft is realized that space reaches, detected, defendd, attack four core competence by great amount of samples training
Combinatorial Optimization.
Man-machine confrontation in intellectualized battle deduction system joined artificial governing factor, break training mode and machine machine pair
Anti- the case where fully relying on algorithm, realizes that man-machine confrontation exports high-quality training sample data, not only to the training mould of combat unit
Type has promotes effect well, and training pattern is made to be more nearly true warfare decision.
It is shown using confrontation terminal as front end when man-machine confrontation and operation module is fought so that air battle fights scene as an example
Flight simulator is disposed in terminal, by keyboard input control aircraft object, mock battle unit carries out man-machine confrontation.Pass through keyboard
It can control the operation such as flying speed, azimuth of aircraft.Before start, user can on map selecting object position,
Initial velocity and the azimuth of object are set.Keyboard input data (speed, azimuth) is converted into latitude and longitude coordinates, drive control
The data information of operation is uploaded to server-side, by rear end decision-making treatment after artificially controlling flare maneuver by object's position variation
Afterwards, the data processing section of the run action feedback of our aircraft to client, our aircraft after data processing is passed through
Latitude coordinate drives the location information of our aircraft to show variation, and operational effect is as shown in Figure 3.
Confrontation terminal disposes global situation visual software simultaneously, the displaying of aircraft running track will make battlefield between ourselves and the enemy
In figure, based on the exploitation of the front ends web technologies such as HTML5, JavaScript and Css, the part GIS is integrated, the page is rendered,
Cartographic information, object information and special efficacy are visualized, it is as shown in Figure 4 to run global situation situation.
Deeply study is applied to air battle and fought in typical operation scene by the present embodiment, and the operation for designing realization is deduced
System can be used for man-machine confrontation test, realize tactical confrontation scenario simulation, carries out air combat situation sensing capability for pilot and mentions
For training tool.Based on deeply learning training model construction pilot training and checking system, meets pilot's formation and make
The culture demand that the level of training of fighting improves.It is exported by typical air battle scene hands-on data and self game, summarizes and refine
Representative tactics strategy provides guidance instruction for subsequent practical air fighting, and provides for operational exertion handbook necessary
Material and training support.The case where emulation data can assess main battle weaponry is deduced simultaneously, optimizes the strategy that uses of equipment, identification
Develop the core-capacitor factor of equipment.
It should be noted that the intellectualized battle deduction system of the present embodiment can also extend to other fields of military operations, pass through
Artificial intelligence obtains training pattern, and then the training pattern by having certain level of intelligence realizes the autonomous work of combat unit
War.
In embodiment provided by the present invention, so it is easy to understand that disclosed related system can pass through others
Mode is realized.For example, the division of the platform or module, only a kind of logical function partition, can have another in actual implementation
Outer division mode, for example, multiple units or components may be combined or can be integrated into a system or some features can be with
Ignore, or does not execute.Another point, the indirect coupling or communication connection of unit or assembly can be electrical property, mechanical or other shapes
Formula.
The unit as illustrated by the separation member may or may not be physically separated, as display
Component may or may not be physical unit, it can and it is in one place, or may be distributed over multiple networks
On unit.It can some or all of the units may be selected to achieve the purpose of the solution of this embodiment according to the actual needs.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of software functional units.
Technical solution of the present invention substantially the part that contributes to existing technology or the technical solution in other words
Completely or partially it can be embodied in the form of software products, which is stored in a storage medium,
It is used including some instructions so that computer processor (processor) executes the whole of each embodiment the method for the present invention
Or part steps.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only
Memory), random access memory (RAM, Random Access Memory), magnetic or disk etc. are various can store journey
The medium of sequence code.
The above embodiments are merely illustrative of the technical solutions of the present invention rather than is limited, without departing substantially from essence of the invention
In the case where mind and its essence, those skilled in the art make various corresponding changes and change in accordance with the present invention
Shape, but these corresponding changes and modifications all should fall within the scope of protection of the appended claims of the present invention.
Claims (7)
1. a kind of unsupervised intellectualized battle deduction system based on deeply study, which is characterized in that described strong based on depth
Chemistry practise unsupervised intellectualized battle deduction system include confrontation terminal, operation simulation engine, deeply learning platform and
Data management platform is provided with equipment model library and model of place library, the data management platform in the operation simulation engine
In be provided with sample database and training pattern library, in which:
The confrontation terminal, for connecting the input interface of the operation simulation engine；
The operation simulation engine, the operational environment for being inputted according to device end define information, import from equipment model library
Equipment model relevant to operational environment imports model of place relevant to operational environment from model of place library, completes initialization,
The operational environment includes combat mode, combat unit and operation scene；
And when combat mode is training mode, operation game between ourselves and the enemy is carried out based on timeslice and is deduced, by current environment
Information is sent to the deeply learning platform, executes single step decision according to the movement feedback of deeply learning platform output
Or multistep decision, and update environmental information；
It is also used to judge currently whether meet the flat relationship of victory or defeat or termination condition according to predefined rule, if it is flat currently to meet victory or defeat
When relationship or termination condition, output, which is fought, deduces analysis report, and saves Campaign Process as sample data to the data
The sample database of platform is managed, training is completed；Otherwise continue game deduction of fighting between ourselves and the enemy, until it is flat to meet victory or defeat
Relationship or termination condition；
The deeply learning platform, for importing training pattern from the training pattern library of data management platform, and combines
The environmental information that the operation simulation engine is sent, Xiang Zuozhan simulation engine output action feedback；It is also used to update combat unit
Corresponding training pattern, and exported in training pattern library of the training pattern to data management platform after operation game is deduced
It saves；
The data management platform, training pattern and described fight for saving the deeply learning platform output are imitated
The sample data of true engine output；It is also used to export training pattern to the deeply learning platform.
2. the unsupervised intellectualized battle deduction system as described in claim 1 based on deeply study, which is characterized in that institute
Stating environmental information includes: quantity, location information, current state, mutual spacing between ourselves and the enemy.
3. the unsupervised intellectualized battle deduction system as described in claim 1 based on deeply study, which is characterized in that institute
The operation that single step decision includes combat unit is stated to act；The multistep decision include combat unit it is multiple fight one battle after another it is dynamic
Make.
4. the unsupervised intellectualized battle deduction system as described in claim 1 based on deeply study, which is characterized in that institute
Confrontation terminal is stated as device end, is inputted and is made to the operation simulation engine by the input interface of the operation simulation engine
War Environment Definition information.
5. the unsupervised intellectualized battle deduction system as described in claim 1 based on deeply study, which is characterized in that institute
The combat mode for stating operation simulation engine further includes simulation model, and the simulation model includes man-machine confrontation and the confrontation of machine machine；
The confrontation terminal, also provides for three-dimensional visualization interface, and the operation emulation is presented in visualization under simulation model
The environmental information of engine；Confrontation operation interface is also provided for, in man-machine confrontation by confrontation operation interface artificially to described
The input of operation simulation engine, which is fought, to be operated.
6. the unsupervised intellectualized battle deduction system as claimed in claim 5 based on deeply study, which is characterized in that institute
State combat mode be machine machine confrontation when, it is described based on deeply study unsupervised intellectualized battle deduction system, execute it is as follows
Operation:
The operation simulation engine for accessing the three-dimensional visualization interface of the confrontation terminal, and is inputted according to device end
Operational environment define information, relevant to operational environment equipment model is imported from equipment model library, from the importing of model of place library
Model of place relevant to operational environment imports training pattern from the training pattern library of data management platform, completes initialization；
And operation game between ourselves and the enemy is carried out based on timeslice and is deduced, current environmental information is sent to the deeply and learns to put down
Platform executes single step decision according to the movement feedback of deeply learning platform output, and updates environmental information according to single step decision,
Effect is presented in the visualization for refreshing the confrontation terminal；
It is also used to judge currently whether meet the flat relationship of victory or defeat or termination condition according to predefined rule, if it is flat currently to meet victory or defeat
When relationship or termination condition, output, which is fought, deduces analysis report, and saves Campaign Process as sample data to the data
The sample database of platform is managed, emulation is completed；Otherwise continue game deduction of fighting between ourselves and the enemy, until it is flat to meet victory or defeat
Relationship or termination condition；
The deeply learning platform, for importing training pattern from the training pattern library of data management platform, and combines
The environmental information that the operation simulation engine is sent, Xiang Zuozhan simulation engine output action feedback；
The data management platform, for saving the sample data of the operation simulation engine output；It is also used to the depth
Intensified learning platform and the operation simulation engine export training pattern.
7. the unsupervised intellectualized battle deduction system as claimed in claim 5 based on deeply study, which is characterized in that institute
State combat mode be man-machine confrontation when, it is described based on deeply study unsupervised intellectualized battle deduction system, execute it is as follows
Operation:
The operation simulation engine, for accessing the three-dimensional visualization interface and confrontation operation interface of the confrontation terminal, and root
Information is defined according to the operational environment that device end inputs, imports equipment model relevant to operational environment from equipment model library, from
Model of place library imports model of place relevant to operational environment, and training mould is imported from the training pattern library of data management platform
Type completes initialization；And operated according to the operation of confrontation terminal, operation game between ourselves and the enemy is carried out based on timeslice and is deduced, it will
Current environmental information is sent to the deeply learning platform, is held according to the movement feedback of deeply learning platform output
Row single step decision, while the operation operation of the confrontation terminal input is received, and ring is updated according to single step decision and operation of fighting
Border information refreshes the effect of visualization of the confrontation terminal；
It is also used to judge currently whether meet the flat relationship of victory or defeat or termination condition according to predefined rule, if it is flat currently to meet victory or defeat
When relationship or termination condition, output, which is fought, deduces analysis report, and saves Campaign Process as sample data to the data
The sample database of platform is managed, emulation is completed；Otherwise continue game deduction of fighting between ourselves and the enemy, until it is flat to meet victory or defeat
Relationship or termination condition；
The deeply learning platform, for importing training pattern from the training pattern library of data management platform, and combines
The environmental information that the operation simulation engine is sent, Xiang Zuozhan simulation engine output action feedback；
The data management platform, for saving the sample data of the operation simulation engine output；It is also used to the depth
Intensified learning platform and the operation simulation engine export training pattern.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811313124.XA CN109636699A (en) | 2018-11-06 | 2018-11-06 | A kind of unsupervised intellectualized battle deduction system based on deeply study |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811313124.XA CN109636699A (en) | 2018-11-06 | 2018-11-06 | A kind of unsupervised intellectualized battle deduction system based on deeply study |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109636699A true CN109636699A (en) | 2019-04-16 |
Family
ID=66067350
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811313124.XA Pending CN109636699A (en) | 2018-11-06 | 2018-11-06 | A kind of unsupervised intellectualized battle deduction system based on deeply study |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109636699A (en) |
Cited By (34)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110147883A (en) * | 2019-05-28 | 2019-08-20 | 航天科工系统仿真科技（北京）有限公司 | Training method, device, equipment and the storage medium of model for emulation of fighting |
| CN110210115A (en) * | 2019-05-30 | 2019-09-06 | 北京华如科技股份有限公司 | The design of operation simulating scheme and operation method emulated based on decision point and branch |
| CN110322017A (en) * | 2019-08-13 | 2019-10-11 | 吉林大学 | Automatic Pilot intelligent vehicle Trajectory Tracking Control strategy based on deeply study |
| CN110362258A (en) * | 2019-07-19 | 2019-10-22 | 航天科工网络信息发展有限公司 | A kind of interactive cooperation scenario system implementation method |
| CN110428057A (en) * | 2019-05-06 | 2019-11-08 | 南京大学 | A kind of intelligent game playing system based on multiple agent deeply learning algorithm |
| CN110782039A (en) * | 2019-10-11 | 2020-02-11 | 南京摄星智能科技有限公司 | Artificial intelligence instant combat guide platform based on layered structure and multiple modules |
| CN110989343A (en) * | 2019-11-25 | 2020-04-10 | 中国人民解放军国防科技大学 | Multi-stage equipment combination planning method based on reinforcement learning |
| CN111159095A (en) * | 2020-01-02 | 2020-05-15 | 中国航空工业集团公司西安航空计算技术研究所 | Heterogeneous integrated embedded intelligent computing implementation method |
| CN111488992A (en) * | 2020-03-03 | 2020-08-04 | 中国电子科技集团公司第五十二研究所 | Simulator adversary reinforcing device based on artificial intelligence |
| CN111632387A (en) * | 2020-06-12 | 2020-09-08 | 南京大学 | Command control system based on interstellar dispute II |
| CN111882027A (en) * | 2020-06-02 | 2020-11-03 | 东南大学 | Robot reinforcement learning training environment system for RoboMaster artificial intelligence challenge competition |
| CN111984751A (en) * | 2020-07-13 | 2020-11-24 | 北京华如科技股份有限公司 | System and method based on multi-domain combat visual display |
| CN112131786A (en) * | 2020-09-14 | 2020-12-25 | 中国人民解放军军事科学院评估论证研究中心 | Target detection and distribution method and device based on multi-agent reinforcement learning |
| CN112287521A (en) * | 2020-10-10 | 2021-01-29 | 东北大学 | Decision-making platform of intelligent combat equipment |
| CN112364500A (en) * | 2020-11-09 | 2021-02-12 | 中国科学院自动化研究所 | Multi-concurrency real-time countermeasure system oriented to reinforcement learning training and evaluation |
| CN112364566A (en) * | 2020-11-12 | 2021-02-12 | 中国运载火箭技术研究院 | Deduction prediction method based on typical time data characteristics |
| CN112381237A (en) * | 2020-12-09 | 2021-02-19 | 中国船舶工业系统工程研究院 | Continuous autonomous learning architecture system of water surface unmanned system based on bank and sea cooperation |
| CN112464549A (en) * | 2020-07-06 | 2021-03-09 | 中国人民解放军军事科学院评估论证研究中心 | Dynamic allocation method of countermeasure unit |
| CN112464548A (en) * | 2020-07-06 | 2021-03-09 | 中国人民解放军军事科学院评估论证研究中心 | Dynamic allocation device for countermeasure unit |
| CN112489520A (en) * | 2020-12-23 | 2021-03-12 | 中国人民解放军63983部队 | Login obstacle-breaking training system and construction method thereof |
| CN112509413A (en) * | 2020-12-23 | 2021-03-16 | 中国人民解放军63983部队 | Unmanned blue army for landing obstacle-breaking training and construction method thereof |
| CN112561032A (en) * | 2020-11-27 | 2021-03-26 | 中国电子科技集团公司第十五研究所 | Multi-agent reinforcement learning method and system based on population training |
| CN112802091A (en) * | 2021-01-28 | 2021-05-14 | 北京理工大学 | DQN-based intelligent confrontation behavior realization method under augmented reality condition |
| CN113079167A (en) * | 2021-04-12 | 2021-07-06 | 西北工业大学 | Internet of vehicles intrusion detection method and system based on deep reinforcement learning |
| CN113283124A (en) * | 2021-07-21 | 2021-08-20 | 中国人民解放军国防科技大学 | Multi-agent-based autonomous USoS participation model construction method and system |
| CN113283111A (en) * | 2021-06-11 | 2021-08-20 | 中国人民解放军国防科技大学 | Conversion method from model deduction to intelligent deduction |
| CN113633994A (en) * | 2021-07-16 | 2021-11-12 | 中国科学院自动化研究所 | Man-machine intelligent game system |
| CN113656962A (en) * | 2021-08-16 | 2021-11-16 | 中国电子科技集团公司第五十四研究所 | Strategic layer game deduction method based on information flow |
| CN113705102A (en) * | 2021-08-31 | 2021-11-26 | 湖南苍树航天科技有限公司 | Deduction simulation system, method, equipment and storage medium for sea-air cluster confrontation |
| CN113985376A (en) * | 2021-12-23 | 2022-01-28 | 南京雷电信息技术有限公司 | Radar comprehensive display and control excitation system |
| CN114154611A (en) * | 2021-11-10 | 2022-03-08 | 中国科学院自动化研究所 | Man-machine confrontation system supporting Turing test mode and intelligent agent test method |
| CN114257395A (en) * | 2021-11-01 | 2022-03-29 | 清华大学 | Customized network security situation perception method and device based on collaborative learning |
| TWI784717B (en) * | 2021-09-16 | 2022-11-21 | 李福軒 | Evaluation system of combat power prediction, assessment method thereof, computer program product with stored programs thereof, and computer readable medium with stored programs thereof |
| CN116962087A (en) * | 2023-09-20 | 2023-10-27 | 北京安天网络安全技术有限公司 | Auxiliary information distribution method, electronic equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107506830A (en) * | 2017-06-20 | 2017-12-22 | 同济大学 | Towards the artificial intelligence training platform of intelligent automobile programmed decision-making module |
| US20180032863A1 (en) * | 2016-07-27 | 2018-02-01 | Google Inc. | Training a policy neural network and a value neural network |
| CN108021754A (en) * | 2017-12-06 | 2018-05-11 | 北京航空航天大学 | A kind of unmanned plane Autonomous Air Combat Decision frame and method |
| CN108629422A (en) * | 2018-05-10 | 2018-10-09 | 浙江大学 | A kind of intelligent body learning method of knowledge based guidance-tactics perception |
| US20180293498A1 (en) * | 2016-01-27 | 2018-10-11 | Bonsai AI, Inc. | For hiearchical decomposition deep reinforcement learning for an artificial intelligence model |
-
2018
- 2018-11-06 CN CN201811313124.XA patent/CN109636699A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180293498A1 (en) * | 2016-01-27 | 2018-10-11 | Bonsai AI, Inc. | For hiearchical decomposition deep reinforcement learning for an artificial intelligence model |
| US20180032863A1 (en) * | 2016-07-27 | 2018-02-01 | Google Inc. | Training a policy neural network and a value neural network |
| CN107506830A (en) * | 2017-06-20 | 2017-12-22 | 同济大学 | Towards the artificial intelligence training platform of intelligent automobile programmed decision-making module |
| CN108021754A (en) * | 2017-12-06 | 2018-05-11 | 北京航空航天大学 | A kind of unmanned plane Autonomous Air Combat Decision frame and method |
| CN108629422A (en) * | 2018-05-10 | 2018-10-09 | 浙江大学 | A kind of intelligent body learning method of knowledge based guidance-tactics perception |
Non-Patent Citations (2)
| Title |
|---|
| 中国自动化学会: "《控制科学与工程学科发展报告》", 31 March 2018, 中国科学技术出版社 * |
| 贾连兴: "《三维可视化仿真》", 31 March 2017, 国防工业出版社 * |
Cited By (50)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110428057A (en) * | 2019-05-06 | 2019-11-08 | 南京大学 | A kind of intelligent game playing system based on multiple agent deeply learning algorithm |
| CN110147883A (en) * | 2019-05-28 | 2019-08-20 | 航天科工系统仿真科技（北京）有限公司 | Training method, device, equipment and the storage medium of model for emulation of fighting |
| CN110210115B (en) * | 2019-05-30 | 2023-01-31 | 北京华如科技股份有限公司 | Combat simulation scheme design and operation method based on decision point and branch simulation |
| CN110210115A (en) * | 2019-05-30 | 2019-09-06 | 北京华如科技股份有限公司 | The design of operation simulating scheme and operation method emulated based on decision point and branch |
| CN110362258A (en) * | 2019-07-19 | 2019-10-22 | 航天科工网络信息发展有限公司 | A kind of interactive cooperation scenario system implementation method |
| CN110322017A (en) * | 2019-08-13 | 2019-10-11 | 吉林大学 | Automatic Pilot intelligent vehicle Trajectory Tracking Control strategy based on deeply study |
| CN110782039A (en) * | 2019-10-11 | 2020-02-11 | 南京摄星智能科技有限公司 | Artificial intelligence instant combat guide platform based on layered structure and multiple modules |
| CN110782039B (en) * | 2019-10-11 | 2021-10-01 | 南京摄星智能科技有限公司 | Artificial intelligence instant combat guide platform based on layered structure and multiple modules |
| CN110989343B (en) * | 2019-11-25 | 2020-08-28 | 中国人民解放军国防科技大学 | Multi-stage equipment combination planning method based on reinforcement learning |
| CN110989343A (en) * | 2019-11-25 | 2020-04-10 | 中国人民解放军国防科技大学 | Multi-stage equipment combination planning method based on reinforcement learning |
| CN111159095A (en) * | 2020-01-02 | 2020-05-15 | 中国航空工业集团公司西安航空计算技术研究所 | Heterogeneous integrated embedded intelligent computing implementation method |
| CN111159095B (en) * | 2020-01-02 | 2023-05-12 | 中国航空工业集团公司西安航空计算技术研究所 | Heterogeneous fusion embedded intelligent computing implementation method |
| CN111488992A (en) * | 2020-03-03 | 2020-08-04 | 中国电子科技集团公司第五十二研究所 | Simulator adversary reinforcing device based on artificial intelligence |
| CN111882027A (en) * | 2020-06-02 | 2020-11-03 | 东南大学 | Robot reinforcement learning training environment system for RoboMaster artificial intelligence challenge competition |
| CN111632387A (en) * | 2020-06-12 | 2020-09-08 | 南京大学 | Command control system based on interstellar dispute II |
| CN112464549A (en) * | 2020-07-06 | 2021-03-09 | 中国人民解放军军事科学院评估论证研究中心 | Dynamic allocation method of countermeasure unit |
| CN112464548A (en) * | 2020-07-06 | 2021-03-09 | 中国人民解放军军事科学院评估论证研究中心 | Dynamic allocation device for countermeasure unit |
| CN112464548B (en) * | 2020-07-06 | 2021-05-14 | 中国人民解放军军事科学院评估论证研究中心 | Dynamic allocation device for countermeasure unit |
| CN111984751A (en) * | 2020-07-13 | 2020-11-24 | 北京华如科技股份有限公司 | System and method based on multi-domain combat visual display |
| CN112131786A (en) * | 2020-09-14 | 2020-12-25 | 中国人民解放军军事科学院评估论证研究中心 | Target detection and distribution method and device based on multi-agent reinforcement learning |
| CN112131786B (en) * | 2020-09-14 | 2024-05-31 | 中国人民解放军军事科学院评估论证研究中心 | Target detection and distribution method and device based on multi-agent reinforcement learning |
| CN112287521A (en) * | 2020-10-10 | 2021-01-29 | 东北大学 | Decision-making platform of intelligent combat equipment |
| CN112364500A (en) * | 2020-11-09 | 2021-02-12 | 中国科学院自动化研究所 | Multi-concurrency real-time countermeasure system oriented to reinforcement learning training and evaluation |
| CN112364566A (en) * | 2020-11-12 | 2021-02-12 | 中国运载火箭技术研究院 | Deduction prediction method based on typical time data characteristics |
| CN112364566B (en) * | 2020-11-12 | 2022-09-06 | 中国运载火箭技术研究院 | Deduction prediction method based on typical time data characteristics |
| CN112561032A (en) * | 2020-11-27 | 2021-03-26 | 中国电子科技集团公司第十五研究所 | Multi-agent reinforcement learning method and system based on population training |
| CN112381237A (en) * | 2020-12-09 | 2021-02-19 | 中国船舶工业系统工程研究院 | Continuous autonomous learning architecture system of water surface unmanned system based on bank and sea cooperation |
| CN112509413A (en) * | 2020-12-23 | 2021-03-16 | 中国人民解放军63983部队 | Unmanned blue army for landing obstacle-breaking training and construction method thereof |
| CN112489520A (en) * | 2020-12-23 | 2021-03-12 | 中国人民解放军63983部队 | Login obstacle-breaking training system and construction method thereof |
| CN112802091A (en) * | 2021-01-28 | 2021-05-14 | 北京理工大学 | DQN-based intelligent confrontation behavior realization method under augmented reality condition |
| CN112802091B (en) * | 2021-01-28 | 2023-08-29 | 北京理工大学 | DQN-based agent countermeasure behavior realization method under augmented reality condition |
| CN113079167A (en) * | 2021-04-12 | 2021-07-06 | 西北工业大学 | Internet of vehicles intrusion detection method and system based on deep reinforcement learning |
| CN113283111A (en) * | 2021-06-11 | 2021-08-20 | 中国人民解放军国防科技大学 | Conversion method from model deduction to intelligent deduction |
| CN113283111B (en) * | 2021-06-11 | 2022-05-27 | 中国人民解放军国防科技大学 | Conversion method from model deduction to intelligent deduction |
| CN113633994A (en) * | 2021-07-16 | 2021-11-12 | 中国科学院自动化研究所 | Man-machine intelligent game system |
| CN113283124B (en) * | 2021-07-21 | 2021-10-15 | 中国人民解放军国防科技大学 | Multi-agent-based autonomous USoS participation model construction method and system |
| CN113283124A (en) * | 2021-07-21 | 2021-08-20 | 中国人民解放军国防科技大学 | Multi-agent-based autonomous USoS participation model construction method and system |
| CN113656962B (en) * | 2021-08-16 | 2024-06-18 | 中国电子科技集团公司第五十四研究所 | Strategic layer game deduction method based on information flow |
| CN113656962A (en) * | 2021-08-16 | 2021-11-16 | 中国电子科技集团公司第五十四研究所 | Strategic layer game deduction method based on information flow |
| CN113705102B (en) * | 2021-08-31 | 2024-05-10 | 湖南苍树航天科技有限公司 | Deduction simulation system, deduction simulation method, deduction simulation equipment and deduction simulation storage medium for sea-air cluster countermeasure |
| CN113705102A (en) * | 2021-08-31 | 2021-11-26 | 湖南苍树航天科技有限公司 | Deduction simulation system, method, equipment and storage medium for sea-air cluster confrontation |
| TWI784717B (en) * | 2021-09-16 | 2022-11-21 | 李福軒 | Evaluation system of combat power prediction, assessment method thereof, computer program product with stored programs thereof, and computer readable medium with stored programs thereof |
| CN114257395A (en) * | 2021-11-01 | 2022-03-29 | 清华大学 | Customized network security situation perception method and device based on collaborative learning |
| CN114257395B (en) * | 2021-11-01 | 2022-09-27 | 清华大学 | Customized network security situation perception method and device based on collaborative learning |
| CN114154611A (en) * | 2021-11-10 | 2022-03-08 | 中国科学院自动化研究所 | Man-machine confrontation system supporting Turing test mode and intelligent agent test method |
| CN114154611B (en) * | 2021-11-10 | 2022-12-06 | 中国科学院自动化研究所 | Man-machine confrontation system supporting Turing test mode and intelligent agent test method |
| CN113985376B (en) * | 2021-12-23 | 2022-03-15 | 南京雷电信息技术有限公司 | Radar comprehensive display and control excitation system |
| CN113985376A (en) * | 2021-12-23 | 2022-01-28 | 南京雷电信息技术有限公司 | Radar comprehensive display and control excitation system |
| CN116962087B (en) * | 2023-09-20 | 2023-12-01 | 北京安天网络安全技术有限公司 | Auxiliary information distribution method, electronic equipment and storage medium |
| CN116962087A (en) * | 2023-09-20 | 2023-10-27 | 北京安天网络安全技术有限公司 | Auxiliary information distribution method, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109636699A (en) | A kind of unsupervised intellectualized battle deduction system based on deeply study | |
| CN110119773B (en) | Global situation assessment method, system and device of strategic gaming system | |
| Ontanón et al. | A survey of real-time strategy game AI research and competition in StarCraft | |
| CN110991545B (en) | Multi-agent confrontation oriented reinforcement learning training optimization method and device | |
| CN113688977B (en) | Human-computer symbiotic reinforcement learning method and device oriented to countermeasure task, computing equipment and storage medium | |
| CN105678030A (en) | Air-combat tactic team simulating method based on expert system and tactic-military-strategy fractalization | |
| CN109541960B (en) | System and method for aircraft digital battlefield confrontation | |
| CN113893539B (en) | Cooperative fighting method and device for intelligent agent | |
| Ontanón et al. | RTS AI Problems and Techniques. | |
| Zhou et al. | An air combat decision learning system based on a brain-like cognitive mechanism | |
| CN109597839A (en) | A kind of data digging method based on the avionics posture of operation | |
| Zhou et al. | Learning system for air combat decision inspired by cognitive mechanisms of the brain | |
| Wang et al. | Deep reinforcement learning-based air combat maneuver decision-making: literature review, implementation tutorial and future direction | |
| Karr et al. | Synthetic soldiers [military training simulators] | |
| CN114841068A (en) | Three-dimensional high-simulation war game deduction platform and method | |
| Wu | Intelligent warfare: Prospects of military development in the age of AI | |
| Rao et al. | A modified random network distillation algorithm and its application in USVs naval battle simulation | |
| CN117313561A (en) | Unmanned aerial vehicle intelligent decision model training method and unmanned aerial vehicle intelligent decision method | |
| Källström et al. | Design of simulation-based pilot training systems using machine learning agents | |
| CN112561032A (en) | Multi-agent reinforcement learning method and system based on population training | |
| CN117132128A (en) | Intelligent army system capable of achieving autonomous engagement and operation flow | |
| Kong et al. | Multi-ucav air combat in short-range maneuver strategy generation using reinforcement learning and curriculum learning | |
| Zhu et al. | Mastering air combat game with deep reinforcement learning | |
| Yang | A networked multi-agent combat model: Emergence explained | |
| Mendi et al. | Applications of Reinforcement Learning and its Extension to Tactical Simulation Technologies |
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 |