CN106200673A - Integration flight maneuver control method automatically - Google Patents
Integration flight maneuver control method automatically Download PDFInfo
- Publication number
- CN106200673A CN106200673A CN201610593079.2A CN201610593079A CN106200673A CN 106200673 A CN106200673 A CN 106200673A CN 201610593079 A CN201610593079 A CN 201610593079A CN 106200673 A CN106200673 A CN 106200673A
- Authority
- CN
- China
- Prior art keywords
- flight
- state
- operational order
- aircraft
- sequence
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 25
- 230000010354 integration Effects 0.000 title claims abstract description 7
- 230000008569 process Effects 0.000 claims abstract description 10
- RZVHIXYEVGDQDX-UHFFFAOYSA-N 9,10-anthraquinone Chemical compound C1=CC=C2C(=O)C3=CC=CC=C3C(=O)C2=C1 RZVHIXYEVGDQDX-UHFFFAOYSA-N 0.000 abstract description 5
- 238000010801 machine learning Methods 0.000 description 6
- 108090000623 proteins and genes Proteins 0.000 description 6
- 230000009471 action Effects 0.000 description 5
- 230000006399 behavior Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000002068 genetic effect Effects 0.000 description 3
- 239000000203 mixture Substances 0.000 description 3
- 238000009825 accumulation Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 229910052709 silver Inorganic materials 0.000 description 1
- 239000004332 silver Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/10—Simultaneous control of position or course in three dimensions
- G05D1/101—Simultaneous control of position or course in three dimensions specially adapted for aircraft
Landscapes
- Engineering & Computer Science (AREA)
- Aviation & Aerospace Engineering (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
Abstract
The invention belongs to Intelligent flight control system technology, it is provided that a kind of integration flight maneuver control method automatically, including: step 1: judge that aircraft current flight state has met maneuvering target, if meeting, then finishing control process；Otherwise enter step 2；Step 2: initialize decision model；Step 3: decision model generates one and attempts to be connected original state and the operational order sequence of final state；Step 4: run model aircraft according to operational order sequence, obtains state of flight track；Step 5: give a mark to state of flight track；Step 6: according to operational order sequence and state of flight track score replaceme diiion model thereof；Step 7: if there being the calculating time to remain, then enter step 3；Otherwise enter step 8；Step 8: first operational order in the operational order sequence of highest scoring is exported to flight control system by decision model；Step 9: wait that aircraft has performed operational order, subsequently into step 1.
Description
Technical field
The invention belongs to Intelligent flight control system technology, be specifically related to the flight of a kind of automatic maneuver based on machine learning
Control method.
Background technology
In all kinds of aerial missions, the maneuverability of aircraft plays vital effect for reaching task object.Pass
Maneuver autopilot process is divided into system flight maneuver control method reference locus generate (being responsible for by Guidance Law) and track is followed (by controlling
System rule is responsible for) two parts realize.Due to the versatility requirement of algorithm design, existing reference locus generating algorithm is generally based on
Pilot or the written experience of expert, compile the program for band parameter or director data table by the most motor-driven PATH GENERATION
Lattice.When motor-driven order received by aircraft, algorithm generates reference according to given parameters caller or by the method for interpolation of tabling look-up
Track, then transfers to control law to complete track and follows.
The shortcoming of existing flight maneuver control method includes:
1, the PATH GENERATION of various maneuvers all must by understand the expert of flight operation process or pilot from
Line is worked out, and this makes aircraft not possess the probability of motor-driven task completing online to write the most after deliberation；
2, the parametric variable of PATH GENERATION too much (can not be generally limited to 2～3 parameters), otherwise can cause program
Branch is too much or tables of data scale is excessive, it is difficult to establishment；
3, design owing to track following algorithm (control law) combines concrete aircraft type, if Track Pick-up is calculated
Method does not consider that can concrete type catch up with the problem of reference locus, then aircraft may be caused not catch up with the reference locus of generation, enter
And motor-driven task cannot be completed；If PATH GENERATION considers the Capability Requirement of concrete type, then the versatility of algorithm is just
Can be the lowest, cause to redesign a PATH GENERATION for every kind of aircraft.
Machine learning can obtain the feasible sequence of operation of given control task by computer Automatic Optimal.Hence with
Machine learning techniques, can effectively reduce the design difficulty of flight maneuver controller, and improve the versatility of control algolithm.
Summary of the invention
Goal of the invention: a kind of integration flight maneuver control method automatically is provided, utilizes machine learning algorithm, it is achieved general
Automatic flight maneuver controller in various aircraft types, all kinds of motor-driven task and various initial flight state.
Technical scheme: integration flight maneuver control method automatically, including:
Step 1: after receiving motor-driven order, it is judged that aircraft current flight state has met maneuvering target, if meeting,
Then finishing control process；If being unsatisfactory for, then enter step 2；
Step 2: record current flight state is original state, and the dbjective state that motor-driven order specifies is final state, just
Beginningization decision model；
Step 3: decision model is random or attempts to be connected original state and the behaviour of final state by specific program generation one
Make job sequence；
Step 4: give model aircraft by the operational order sequence of generation and perform in order, obtains corresponding the flying of job sequence
Row state trajectory；
Step 5: give evaluator by state of flight track and mark, evaluator is according to whether reach flight restriction bar
Part, gives a mark to state of flight track；
Step 6: according to described operational order sequence and the score replaceme diiion model of the state of flight track of correspondence thereof；
Step 7: determine whether that the calculating time remains, if remaining without the time, then enters step 8；If having the time to remain, then
Enter step 3；
Step 8: first operational order in the operational order sequence of highest scoring is exported to flight control by decision model
System processed；
Step 9: wait that aircraft has performed operational order, subsequently into step 1.
Beneficial effect:
The high-speed computation ability automatic calculation utilizing computer can realize the command operating sequence of maneuvering target, and it is right to eliminate
The requested knowledge of flight maneuver specific operation process, reduces the design difficulty of automatic flight maneuver controller, improves motor-driven
The intelligence of flight controller and general geological coodinate system, promote maneuvering control device design experiences and transmit in different type of machines and accumulate.
Accompanying drawing explanation
Fig. 1 is present invention integration flight maneuver control system functional block diagram automatically.
Fig. 2 is the present invention automatic flight maneuver control algolithm flow chart.
Detailed description of the invention
Automatically flight maneuver control system is formed (see Fig. 1) by model aircraft, decision model and evaluator.In given calculating
In the case of time, aircraft original state and maneuvering target, this system can control aircraft automatic calculation and complete motor-driven task
(see Fig. 2).Noting, flight maneuver controller does not replace the control action that control law is dynamic to aircraft, but at higher level
The upper aircraft directly handling band control law.Advantage of this is that flight maneuver controller can inherit the behaviour of band control law aircraft
The steady function such as characteristic and flight envelope protection.Need the accuracy ensureing model aircraft in motor-driven sequence performs the period herein,
In case the operational order sequence generated under simulated environment can not produce intended flight path in true environment.
It is as follows that automatic flight maneuver control system specifically controls process:
Step 1: after receiving motor-driven order, it is judged that aircraft current flight state has met maneuvering target, if meeting,
Then finishing control process；If being unsatisfactory for, then enter step 2；
Aircraft state of flight at a time can be expressed as the set of variables such as position, speed, attitude, overload, angular velocity
The multi-C vector become.State of flight track refers to the sequence of aircraft state value composition in multiple time steps.
Meet maneuvering target and refer to reach the target flight state of motor-driven order defined, and do not violate the mistake of motor-driven order
Range request (such as transships restriction, attitude angle limits, location window restriction etc.).
Step 2: record current flight state is original state, and the dbjective state that motor-driven order specifies is final state, just
Beginningization decision model；
Decision model is to can be according to the score Automatic Optimal decision instruction of the result of decision or the calculation of decision instruction sequence
The general name of method.Typical decision model includes genetic algorithm[1], intensified learning[2]And monte carlo search tree[3]Deng.
[1]Fogel,D.B.Evolutionary computation:toward a new philosophy of
machine intelligence,volume 1.John Wiley&Sons.2006.
[2]Leslie P.K.,Michael L.L,and Andrew W.M.Reinforcement Learning:A
Survey.Journal of Artificial Intelligence Research 1996,4:237-285.
[3]S.Gelly,D.Silver,Combining online and offline knowledge in UCT[C],
ICML'07,2007.
Step 3: decision model is random or attempts to be connected original state and the behaviour of final state by specific program generation one
Make job sequence；
Operational order refers to send the instructions such as overload, angular velocity or control surface deflection and wait that it performs the mistake of a period of time
Journey.For example, it is possible to then at the uniform velocity push rod 0.1 degree is waited numbered No. 1 action in 0.1 second, by 0.1 degree of pull bar and wait 0.1 second
Numbered No. 2 actions, will not carry out any operation and wait numbered No. 3 actions in 0.1 second, then the continuous print of pilot longitudinally flies
Row operation is just represented by the combination of 1,2, No. 3 actions, and this combination can be encoded to digital command sequence with 1,2,3.
Stochastic generation operational order sequence refers to that the mode according to stochastic sampling repeatedly chooses multiple order number, composition behaviour
The method making job sequence；According to specific program generate operational order sequence refer to according to non-homogeneous probability distribution or certain fix
Rule repeatedly chooses multiple order number, the method for composition operational order sequence.
Step 4: give model aircraft by the operational order sequence of generation and perform in order, obtains corresponding the flying of job sequence
Row state trajectory；
As described in step 3, if the mobile operation process of human pilot can be conceptualized as being moved by limited multiple standards
Make the job sequence formed, then under conditions of aircraft is identical with atmospheric condition, from the beginning of same initial flight state, fly
Machine performs identical job sequence, all should obtain same or analogous state of flight track.In all possible track, if
A track is had can ideally to reach maneuvering target, then this track is called optimal trajectory, and claims the operational order of its correspondence
Sequence is Optimum Operation job sequence.Decision model can utilize the high-speed computational capability of computer to check many groups at short notice
The state of flight track that operational order sequence pair is answered.In extreme circumstances, if all instruction combinations are all tried one time, then must energy
Try out optimal trajectory.In the case of applying mechanically machine learning algorithm and certain experiences (probability distribution or rule), decision model can
Likely instruct combination traveling through, and find optimum or feasible trajectory quickly.
Step 5: give evaluator by state of flight track and mark, evaluator is according to whether reach flight restriction bar
Part, gives a mark to state of flight track；
Flight restriction condition is to include whether that reaching maneuvering target (such as, for somersault is motor-driven, just reaches maneuvering target
Refer to the angle of pitch is increased by 360 degree) and (such as, any point in state of flight track is all whether to meet flight safety restriction
Can not strikes obstacles or ground, it is impossible to beyond speed, the envelope of load factor of aircraft, it is impossible to break through attitude angle protection domain etc.)
Deng requiring in the general designation of interior flight quality leveling factors.
Step 6: according to described operational order sequence and the score replaceme diiion model of the state of flight track of correspondence thereof.
Replaceme diiion model refers to that the score according to flight path changes some parameter in decision model, to adjust certainly
The process of the operational order sequence create-rule of plan model.Such as, genetic algorithm can (i.e. operational order sequence pair should according to gene
Digital coding) score decide whether to eliminate some member in gene bank, thus ensure that the most outstanding gene can be gathered around
There is bigger probability to raise up seed, and then realize whole gene bank to the direction evolution generating high score gene.Herein, it is updated
Decision model is exactly the gene bank of genetic algorithm.
Step 7: determine whether that the calculating time remains, if remaining without the time, then enters step 8；If having the time to remain, then
Enter step 3.
Step 8: first operational order in the operational order sequence of highest scoring is exported to flight control by decision model
System processed.
Although decision model generates a complete operational order sequence, but if waits that aircraft performs in sequence
All operations instructs, then the error or the actual environment interference accumulation in Long time scale that are likely to be due to model cause actual flying
Row track is bigger with expected trajectory deviation.Have employed rolling optimization mechanism herein, update after each operational order performs
The current state of aircraft, make machine learning algorithm based on current flight state continuous updating optimum mobile operation sequence, to reach
To the purpose allowing aircraft complete maneuvering flight in the environment of change adaptively.
Step 9: wait that aircraft has performed operational order, subsequently into step 1.
Claims (1)
1. integration flight maneuver control method automatically, it is characterised in that including:
Step 1: after receiving motor-driven order, it is judged that aircraft current flight state has met maneuvering target, if meeting, then ties
Beam control process；If being unsatisfactory for, then enter step 2；
Step 2: record current flight state is original state, and the dbjective state that motor-driven order specifies is final state, initializes
Decision model；
Step 3: decision model is random or refers to by one operation attempting linking original state and final state of specific program generation
Make sequence；
Step 4: give model aircraft by the operational order sequence of generation and perform in order, obtains the flight shape that job sequence is corresponding
State track；
Step 5: give evaluator by state of flight track and mark, evaluator, according to whether reach flight restriction condition, is given
State of flight track is given a mark；
Step 6: according to described operational order sequence and the score replaceme diiion model of the state of flight track of correspondence thereof；
Step 7: determine whether that the calculating time remains, if remaining without the time, then enters step 8；If there being the time to remain, then enter
Step 3；
Step 8: first operational order in the operational order sequence of highest scoring is exported and control system to flight by decision model
System；
Step 9: wait that aircraft has performed operational order, subsequently into step 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610593079.2A CN106200673B (en) | 2016-07-26 | 2016-07-26 | Integrated automatic flight maneuver control method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610593079.2A CN106200673B (en) | 2016-07-26 | 2016-07-26 | Integrated automatic flight maneuver control method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106200673A true CN106200673A (en) | 2016-12-07 |
| CN106200673B CN106200673B (en) | 2019-10-18 |
Family
ID=57496472
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610593079.2A Active CN106200673B (en) | 2016-07-26 | 2016-07-26 | Integrated automatic flight maneuver control method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106200673B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109933053A (en) * | 2017-12-15 | 2019-06-25 | 海鹰航空通用装备有限责任公司 | A kind of unmanned aerial vehicle (UAV) control method and unmanned plane based on maneuver chain |
| CN110712765A (en) * | 2019-10-30 | 2020-01-21 | 北京航空航天大学 | Aircraft abnormal operation positioning method based on operation spectrum |
| CN111273680A (en) * | 2020-02-27 | 2020-06-12 | 成都飞机工业（集团）有限责任公司 | Method for controlling maneuvering of rib bucket of flying wing layout unmanned aerial vehicle |
| EP3786752A1 (en) * | 2019-08-29 | 2021-03-03 | The Boeing Company | Automated aircraft system with goal driven action planning |
| CN112543899A (en) * | 2019-12-26 | 2021-03-23 | 深圳市大疆创新科技有限公司 | Control method and control device for movable carrier and computer readable storage medium |
| CN114488906A (en) * | 2022-02-15 | 2022-05-13 | 广东汇天航空航天科技有限公司 | Flight control method, flight control unit and flight control system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103823380A (en) * | 2014-03-14 | 2014-05-28 | 北京航空航天大学 | Helicopter overall design method based on consideration of flying quality |
| CN104267614A (en) * | 2014-09-15 | 2015-01-07 | 南京航空航天大学 | Unmanned aerial vehicle real-time simulation system and developing method thereof |
| CN104932527A (en) * | 2015-05-29 | 2015-09-23 | 广州亿航智能技术有限公司 | Aircraft control method and device |
| CN105404152A (en) * | 2015-12-10 | 2016-03-16 | 中国人民解放军海军航空工程学院 | Flight quality prediction method for simulating subjective evaluation of pilot |
| CN105652891A (en) * | 2016-03-02 | 2016-06-08 | 中山大学 | Unmanned gyroplane moving target autonomous tracking device and control method thereof |
| CN105676863A (en) * | 2016-04-06 | 2016-06-15 | 谭圆圆 | Unmanned aerial vehicle control method and control device |
-
2016
- 2016-07-26 CN CN201610593079.2A patent/CN106200673B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103823380A (en) * | 2014-03-14 | 2014-05-28 | 北京航空航天大学 | Helicopter overall design method based on consideration of flying quality |
| CN104267614A (en) * | 2014-09-15 | 2015-01-07 | 南京航空航天大学 | Unmanned aerial vehicle real-time simulation system and developing method thereof |
| CN104932527A (en) * | 2015-05-29 | 2015-09-23 | 广州亿航智能技术有限公司 | Aircraft control method and device |
| CN105404152A (en) * | 2015-12-10 | 2016-03-16 | 中国人民解放军海军航空工程学院 | Flight quality prediction method for simulating subjective evaluation of pilot |
| CN105652891A (en) * | 2016-03-02 | 2016-06-08 | 中山大学 | Unmanned gyroplane moving target autonomous tracking device and control method thereof |
| CN105676863A (en) * | 2016-04-06 | 2016-06-15 | 谭圆圆 | Unmanned aerial vehicle control method and control device |
Non-Patent Citations (1)
| Title |
|---|
| PIETER ABBEEL: ""An Application of reinforcement learning to aerobatic helicopter flight"", 《INTERNATIONAL CONFERENCE ON NEURAL INFORMATION PROCESSING SYSTEMS》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109933053A (en) * | 2017-12-15 | 2019-06-25 | 海鹰航空通用装备有限责任公司 | A kind of unmanned aerial vehicle (UAV) control method and unmanned plane based on maneuver chain |
| CN109933053B (en) * | 2017-12-15 | 2022-03-11 | 海鹰航空通用装备有限责任公司 | Unmanned aerial vehicle control method based on maneuvering action chain and unmanned aerial vehicle |
| EP3786752A1 (en) * | 2019-08-29 | 2021-03-03 | The Boeing Company | Automated aircraft system with goal driven action planning |
| US11527165B2 (en) | 2019-08-29 | 2022-12-13 | The Boeing Company | Automated aircraft system with goal driven action planning |
| CN110712765A (en) * | 2019-10-30 | 2020-01-21 | 北京航空航天大学 | Aircraft abnormal operation positioning method based on operation spectrum |
| CN112543899A (en) * | 2019-12-26 | 2021-03-23 | 深圳市大疆创新科技有限公司 | Control method and control device for movable carrier and computer readable storage medium |
| WO2021128184A1 (en) * | 2019-12-26 | 2021-07-01 | 深圳市大疆创新科技有限公司 | Control method and control apparatus for movable carrier, and computer-readable storage medium |
| CN111273680A (en) * | 2020-02-27 | 2020-06-12 | 成都飞机工业（集团）有限责任公司 | Method for controlling maneuvering of rib bucket of flying wing layout unmanned aerial vehicle |
| CN114488906A (en) * | 2022-02-15 | 2022-05-13 | 广东汇天航空航天科技有限公司 | Flight control method, flight control unit and flight control system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106200673B (en) | 2019-10-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106200673A (en) | Integration flight maneuver control method automatically | |
| US10394238B2 (en) | Multi-mode mission planning and optimization for autonomous agricultural vehicles | |
| CN107063255B (en) | Three-dimensional route planning method based on improved drosophila optimization algorithm | |
| KR101339480B1 (en) | Trajectory planning method for mobile robot using dual tree structure based on rrt | |
| CN112286218B (en) | Aircraft large-attack-angle rock-and-roll suppression method based on depth certainty strategy gradient | |
| CN112550314B (en) | Embedded optimization type control method suitable for unmanned driving, driving control module and automatic driving control system thereof | |
| Al Dabooni et al. | Heuristic dynamic programming for mobile robot path planning based on Dyna approach | |
| CN110658816B (en) | Mobile robot navigation and control method based on intelligent component | |
| Palm et al. | Particle swarm optimization of potential fields for obstacle avoidance | |
| JP7116799B2 (en) | Work area zone demarcation device for autonomous mobile work machine | |
| CN111176324B (en) | Method for avoiding dynamic obstacle by multi-unmanned aerial vehicle distributed collaborative formation | |
| CN117109574A (en) | Agricultural transportation machinery coverage path planning method | |
| CN113485323B (en) | Flexible formation method for cascading multiple mobile robots | |
| CN116227622A (en) | Multi-agent landmark coverage method and system based on deep reinforcement learning | |
| MacArthur et al. | Compliant formation control of a multi-vehicle system | |
| CN113959446B (en) | Autonomous logistics transportation navigation method for robot based on neural network | |
| Zhang et al. | Research on the local path planning of an orchard mowing robot based on an elliptic repulsion scope boundary constraint potential field method | |
| Southey et al. | Approaching evolutionary robotics through population-based incremental learning | |
| CN115373415A (en) | Unmanned aerial vehicle intelligent navigation method based on deep reinforcement learning | |
| Rybak et al. | Development of an algorithm for managing a multi-robot system for cargo transportation based on reinforcement learning in a virtual environment | |
| Cao et al. | Multi-robot learning dynamic obstacle avoidance in formation with information-directed exploration | |
| Tian et al. | Fruit Picking Robot Arm Training Solution Based on Reinforcement Learning in Digital Twin | |
| Kuyucu et al. | Incremental evolution of fast moving and sensing simulated snake-like robot with multiobjective GP and strongly-typed crossover | |
| Henderson et al. | Cost adaptation for robust decentralized swarm behaviour | |
| Li et al. | Distributed Circle Formation Control for Quadrotors Based on Multi-agent Deep Reinforcement Learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |