CN109711678A - A kind of heterogeneous sensor intelligent task planing method based on machine learning - Google Patents
A kind of heterogeneous sensor intelligent task planing method based on machine learning Download PDFInfo
- Publication number
- CN109711678A CN109711678A CN201811495205.6A CN201811495205A CN109711678A CN 109711678 A CN109711678 A CN 109711678A CN 201811495205 A CN201811495205 A CN 201811495205A CN 109711678 A CN109711678 A CN 109711678A
- Authority
- CN
- China
- Prior art keywords
- rule
- method based
- machine learning
- heterogeneous sensor
- learning
- 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
Landscapes
- Feedback Control In General (AREA)
- Aiming, Guidance, Guns With A Light Source, Armor, Camouflage, And Targets (AREA)
Abstract
The heterogeneous sensor intelligent task planing method based on machine learning that the present invention provides a kind of, include the following steps: (1) rule setting: the format judged using condition establishes switching on and shutting down and radiation rule, operating mode switching law and interference protection measure applying rules；(2) quantification treatment: above-mentioned switching on and shutting down and radiation rule, operating mode switching law and interference protection measure applying rules are traversed respectively and are quantified as parameter group；(3) learning model；(4) scene quantifies；(5) rule-based reasoning；(6) confirmation feedback.The present invention combines the planning inference method based on expertise with the artificial intelligence approach based on machine learning, improve the self-learning capability of expert system, the reasoning to rule other than knowledge base can be achieved, it is excessive to expert system dependence to solve conventional method, the problem of lacking adaptivity and corresponding self-organizing, self-learning capability, is conducive to find new rule, forms new experience.
Description
Technical field
The heterogeneous sensor intelligent task planing method based on machine learning that the present invention relates to a kind of, belongs to area defense and refers to
Wave control field.
Background technique
Region class Command & Decision System is the core of performance system equipment collective Efficacy, and basic task is based on comprehensive state
Gesture completes the coordinated control to each resource according to the operating status of system, comprising to early warning radar, guidance radar, infrared acquisition
The planning control of the heterogeneous sensors system such as system.Commanding and decision-making has wanting for real-time, correctness, completeness and expandability
It asks, in decision process, when certain rule is most suitable for current situation, the rule can be enabled rapidly, sufficiently catch opportunity of combat, work as institute
It is regular when being unsuitable for, should also there be certain tactics strategy.Meanwhile it should be able to also be real according to the experience after Attack Defence manoeuvre
When the new DECISION KNOWLEDGE of increase into system.
Traditional command and control system is to the work opportunity planning of sensing system, mode planning etc. generally according to combat experience
It is got with expertise, planing method is based primarily upon fixed criterion, generally by finding decision-making party to traversal regular one by one
Case is insufficient in flexibility and autonomous extendibility.Due to the complexity of modern system operation, non-linear and ambiguity, so that with
It is not able to satisfy adaptive, the self study demand of intelligent Combat Command System toward the method for the planning modeling using unalterable rules.
Summary of the invention
In order to solve the above technical problems, the present invention provides a kind of, the heterogeneous sensor intelligent task based on machine learning is advised
The method of drawing is somebody's turn to do the heterogeneous sensor intelligent task planing method based on machine learning and solves conventional method to expert system dependence
Property it is excessive, lack adaptivity and the problem of corresponding self-organizing, self-learning capability, be conducive to find new rule, form fresh warp thread
It tests.
The present invention is achieved by the following technical programs.
A kind of heterogeneous sensor intelligent task planing method based on machine learning provided by the invention, including walk as follows
It is rapid:
(1) rule setting: the format judged using condition establishes switching on and shutting down and radiation rule, operating mode switching law
With interference protection measure applying rules；
(2) quantification treatment: by above-mentioned switching on and shutting down and radiation rule, operating mode switching law and interference protection measure with rule
It then traverses respectively and is quantified as parameter group；
(3) it learning model: using obtained parameter group as training sample, is trained, is obtained using machine learning algorithm
Rule-based reasoning model；
(4) scene quantifies: obtaining the component value of the input of target and equipment state corresponding to rule-based reasoning model；
(5) rule-based reasoning: above-mentioned component value is substituting in rule-based reasoning model, program results are obtained；
(6) confirmation feedback: confirm and correct program results, while by the program results obtained after amendment together with corresponding defeated
Enter component to be added into parameter group together, re-starts step (3).
Quantization in the step (2) refers to the digital table of target corresponding in rule, equipment state and program results
Show, wherein program results value is integer.
Traversal in the step (2) refers to switching on and shutting down and radiation rule, operating mode switching law or anti-interference arranges
Apply in applying rules, according to all situations that may occur of rule enumerate come.
Machine learning algorithm uses the method based on extreme learning machine in the step (3).
In the method based on extreme learning machine, activation primitive uses Sigmoid function.
In the method based on extreme learning machine, the number of hidden nodes is nine.
In the step (5), rule-based reasoning model is calculated output result and obtains planning knot after round
Fruit.
The beneficial effects of the present invention are: by based on expertise planning inference method with based on the artificial of machine learning
Intelligent method combines, and improves the self-learning capability of expert system, it can be achieved that solving to reasoning regular other than knowledge base
The problem of conventional method is excessive to expert system dependence, lacks adaptivity and corresponding self-organizing, self-learning capability, favorably
In the new rule of discovery, form new experience.
Detailed description of the invention
Fig. 1 is flow diagram of the invention.
Specific embodiment
Be described further below technical solution of the present invention, but claimed range be not limited to it is described.
A kind of heterogeneous sensor intelligent task planing method based on machine learning as shown in Figure 1, the specific steps are as follows:
Step 1: mode sensor Plan Rule designs.
In system operation, in order to guarantee that the radar etc. for needing active radiation signal to carry out target measurement can measure in due course
Target position, and can be reduced the chance to stick one's chin out, it answers strict control to be switched on and radiate opportunity, generally works as and distribute to sensor
Target is just radiated when reaching its maximum detectable range.For infrared detection system, due to its will not active radiation energy, can
It scans for monitoring as monitoring system, when there is Target indication information, tracing mode can be transferred to, carry out target in a small range
Tracking.And when sensor determines it by the timing of antiradiation latch, radiation should be closed, if but other are heavy for the sensor face at this time
Point target is tracked, then needs to judge the highest priority whether by other sensors tenacious tracking, further according to the highest priority
It is tracked situation and judges whether sensor needs to close radiation or implement intermittent radiation mode.After target enters range, it is used for
The radar of guidance should be transferred to essence with mode, discharge interference when detecting target, determine mesh using the passive location technology of radar
Cursor position uses corresponding interference protection measure according to interference type, according to the above knowledge, is established such as using the format of If-Then
Lower rule:
A) switching on and shutting down and radiation rule
{ there are targets relative to sensor distance < sensor maximum range+target velocity × biography by regular k+1:If
Sensor is switched on/and open radiated time }
Then { working sensor state=open radiation/booting }
Else { working sensor state=pass radiation/shutdown }
Wherein, for infrared detection sensor, since actively radiation signal, working condition are not startup and shutdown
Two kinds.
For the radar class sensor of active radiation signal, there are also following rules:
Regular k+2:if { there are antiradiation threats }
Then if { there are highest priority tracing tasks }
Then if { highest priority is tracked by collaboration }
Then { radar operation mode=pass radiation }
Else { radar operation mode=intermittent radiation }
Else { radar operation mode=pass radiation }
Else { radar operation mode=" keeping status " }
Wherein, when radar intermittent radiation, essence cannot be transferred to state.
B) operating mode switching law
Regular k+1:If { there are Target indication information }
Then { infrared system operating mode=tracing mode
Guidance radar operating mode=outer guidance }
Else { infrared system operating mode=search pattern
Guidance radar operating mode=while with while sweep；
Regular k+2:if { there are target relative distance < permission maximum distance+target velocity × system times }
Then { guidance radar operating mode=essence with }
Else { guidance radar operating mode=slightly with }
C) interference protection measure applying rules
For guidance radar, there is following interference protection measure application method:
Regular k+1:if { target discharges self-defence type noise jamming }
Then { Radar cross-section redaction mode=passive tracking }
Regular k+2:If { target release storm rainfall far from typhoon }
Then { Radar cross-section redaction mode=" frequency agility " and " sidelobe cancellation " and " sidelobe blanking " }
Regular k+3:If { target discharges self-defence type false target jamming profile }
Then { Radar cross-section redaction mode=" frequency agility " and " sidelobe cancellation " and " sidelobe blanking " and decoy }
Regular k+4:If { target release pull-off jamming }
Then { Radar cross-section redaction mode=" broadband " and " phase code " }
Regular k+5:If { target release decoy jamming }
Then { Radar cross-section redaction mode=" MTI " and " MTD " and " broadband waveform tracking " }
Step 2: above-mentioned rule is carried out quantification treatment.By in first step a) item switching on and shutting down and radiation rule for into
The above rule is carried out quantification treatment by row explanation.Quantization method is as follows: input variable: x=(x1, x2, x3, x4, x5)=(main
Whether dynamic or passive system, target enter its investigative range, if having ARM to attack it, and if have and has tracked highest priority, weight
Whether point tracking target is tracked by collaboration), wherein each component can value be 0 or 1, x1 be 0 expression sensor be Active Radar,
It is passive radar or infrared that x1, which is 1 expression sensor, other component values are 1 to be illustrated as, and 0 is expressed as "No".Output valve y takes
Value is that 1 expression is not switched on (or non-radiating), and 2 indicate booting (or radiation).
Since battlefield variation is dynamically, for the value of above-mentioned input parameter, can be blurred according to the actual situation
Processing, if whether sensor is uncertain ARM attack, then the entry value can be subordinate to letter using maximum according to the behavioural characteristic of target
Several methods, fuzzy value are the decimal between 0~1；
Step 3: using the rule after quantization as N number of training sample (xi,yi), wherein xi=[xi1,xi2,…,xi5]T∈
R5, yi∈ R has the ELM of L hidden node that can indicate for one are as follows:
Wherein wi=[wi1,wi2,…,win]TIt is the input weight for connecting i-th of hiding node layer；biIt is i hidden layer
The deviation of node；βi=[βi1,βi2,…,βim]TIt is the output weight for connecting i hiding node layers；wi·xiIndicate wiAnd xi's
Inner product.Activation primitive g () can be the non-constant continuous function of any bounded, usually " Sigmoid ", " Sine " or
" RBF " etc..It is trained study using ELM algorithm, acquires corresponding factor betai, wiAnd bi。
Step 4: threatening target and equipment state according to real-time, each input is determined using the method for maximum membership function
Then component value calculates decision value using the formula in third step, and carries out round processing.
Step 5: being judged by user result, and by after the modified result of mistake, quantify as new rule merging
In rule base afterwards, third step learning training is re-started, obtains new inference pattern.
Claims (7)
1. a kind of heterogeneous sensor intelligent task planing method based on machine learning, characterized by the following steps:
(1) rule setting: the format judged using condition is established switching on and shutting down and radiation rule, operating mode switching law and resisted
Jamming countermeasure applying rules；
(2) quantification treatment: by above-mentioned switching on and shutting down and radiation rule, operating mode switching law and interference protection measure applying rules point
It does not traverse and is quantified as parameter group；
(3) it learning model: using obtained parameter group as training sample, is trained using machine learning algorithm, obtains rule
Inference pattern；
(4) scene quantifies: obtaining the component value of the input of target and equipment state corresponding to rule-based reasoning model；
(5) rule-based reasoning: above-mentioned component value is substituting in rule-based reasoning model, program results are obtained；
(6) confirmation feedback: confirm and correct program results, while by the program results obtained after amendment together with corresponding input point
Amount is added into parameter group together, re-starts step (3).
2. the heterogeneous sensor intelligent task planing method based on machine learning as described in claim 1, it is characterised in that: institute
The quantization in step (2) is stated, is referred to by target corresponding in rule, equipment state and program results digital representation, wherein planning
As a result it is rounded numerical value.
3. the heterogeneous sensor intelligent task planing method based on machine learning as described in claim 1, it is characterised in that: institute
The traversal in step (2) is stated, is referred to switching on and shutting down and radiation rule, operating mode switching law or interference protection measure applying rules
In, according to all situations that may occur of rule enumerate come.
4. the heterogeneous sensor intelligent task planing method based on machine learning as described in claim 1, it is characterised in that: institute
It states machine learning algorithm in step (3) and uses the method based on extreme learning machine.
5. the heterogeneous sensor intelligent task planing method based on machine learning as claimed in claim 4, it is characterised in that:
In the method based on extreme learning machine, activation primitive uses Sigmoid function.
6. the heterogeneous sensor intelligent task planing method based on machine learning as claimed in claim 4, it is characterised in that:
In the method based on extreme learning machine, the number of hidden nodes is nine.
7. the heterogeneous sensor intelligent task planing method based on machine learning as described in claim 1, it is characterised in that: institute
It states in step (5), rule-based reasoning model is calculated output result and obtains program results after round.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811495205.6A CN109711678B (en) | 2018-12-07 | 2018-12-07 | Heterogeneous sensor intelligent task planning method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811495205.6A CN109711678B (en) | 2018-12-07 | 2018-12-07 | Heterogeneous sensor intelligent task planning method based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109711678A true CN109711678A (en) | 2019-05-03 |
| CN109711678B CN109711678B (en) | 2021-02-12 |
Family
ID=66254091
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811495205.6A Active CN109711678B (en) | 2018-12-07 | 2018-12-07 | Heterogeneous sensor intelligent task planning method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109711678B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111489610A (en) * | 2020-05-06 | 2020-08-04 | 西安爱生技术集团公司 | Anti-radiation unmanned aerial vehicle simulation training system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104581939A (en) * | 2015-01-04 | 2015-04-29 | 中国科学院信息工程研究所 | Queuing behavior detection method and system based on multiple heterogeneous sensors |
| CN105515184A (en) * | 2015-12-04 | 2016-04-20 | 国网河南省电力公司电力科学研究院 | Wireless sensor network-based cooperative monitoring system of multi-sensor and multi-parameter distribution network |
| CN105827731A (en) * | 2016-05-09 | 2016-08-03 | 包磊 | Intelligent health management server, system and control method based on fusion model |
| CN105893945A (en) * | 2016-03-29 | 2016-08-24 | 中国科学院自动化研究所 | Target identification method for remote sensing image |
| CN108184241A (en) * | 2017-12-19 | 2018-06-19 | 重庆工商大学 | Towards the isomery directional sensor network node scheduling method of different priorities target |
-
2018
- 2018-12-07 CN CN201811495205.6A patent/CN109711678B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104581939A (en) * | 2015-01-04 | 2015-04-29 | 中国科学院信息工程研究所 | Queuing behavior detection method and system based on multiple heterogeneous sensors |
| CN105515184A (en) * | 2015-12-04 | 2016-04-20 | 国网河南省电力公司电力科学研究院 | Wireless sensor network-based cooperative monitoring system of multi-sensor and multi-parameter distribution network |
| CN105893945A (en) * | 2016-03-29 | 2016-08-24 | 中国科学院自动化研究所 | Target identification method for remote sensing image |
| CN105827731A (en) * | 2016-05-09 | 2016-08-03 | 包磊 | Intelligent health management server, system and control method based on fusion model |
| CN108184241A (en) * | 2017-12-19 | 2018-06-19 | 重庆工商大学 | Towards the isomery directional sensor network node scheduling method of different priorities target |
Non-Patent Citations (1)
| Title |
|---|
| 吴小勇: "反潜体系的搜索能力优化方法研究", 《中国博士学位论文全文数据库 工程科技Ⅱ辑》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111489610A (en) * | 2020-05-06 | 2020-08-04 | 西安爱生技术集团公司 | Anti-radiation unmanned aerial vehicle simulation training system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109711678B (en) | 2021-02-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Stover et al. | A fuzzy-logic architecture for autonomous multisensor data fusion | |
| López et al. | Fuzzy reasoning for multisensor management | |
| Konda et al. | Decentralized function approximated q-learning in multi-robot systems for predator avoidance | |
| Zhang et al. | Research on decision-making system of cognitive jamming against multifunctional radar | |
| CN109711678A (en) | A kind of heterogeneous sensor intelligent task planing method based on machine learning | |
| Tkach et al. | Switching between collaboration levels in a human–robot target recognition system | |
| Young et al. | A survey of research on control of teams of small robots in military operations | |
| Das | Modeling intelligent decision-making command and control agents: An application to air defense | |
| He et al. | Learning-based airborne sensor task assignment in unknown dynamic environments | |
| Arik et al. | Enabling cognition on electronic countermeasure systems against next-generation radars | |
| Azimirad et al. | The comprehensive review on JDL model in data fusion networks: techniques and methods | |
| Liu et al. | Evolutionary algorithm-based attack strategy with swarm robots in denied environments | |
| Gibson et al. | An autonomous fuzzy logic architecture for multisensor data fusion | |
| Abu et al. | Simulation of soil PH Control system using fuzzy logic Method | |
| CN114200960B (en) | Unmanned aerial vehicle cluster search control optimization method for improving sparrow algorithm based on tabu list | |
| Straub | Anti-Drone and Anti-Autonomy: Achieving Drone Control via System Logic Analysis | |
| Naseri et al. | Evaluation of data fusion in radars network and determination of optimum algorithm | |
| Anderson et al. | Sensor resource management driven by threat projection and priorities | |
| Czuba | Artificial Intelligence-Based Cognitive Radar Architecture | |
| McWhorter et al. | Machine learning aided electronic warfare system | |
| De Freitas et al. | Response surface modeling for networked radar resource allocation | |
| Singh et al. | Network threat ratings in conventional dread model using fuzzy logic | |
| Zhang et al. | Design of Cognitive Jamming Decision-Making System Against MFR Based on Reinforcement Learning | |
| Ke et al. | A method of task allocation and automated negotiation for multi robots | |
| Das et al. | Agent-based Decision Making for Integrated Air Defence Systems |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20221220 Address after: 550009 No.7 Honghe Road, economic and Technological Development Zone, Guiyang City, Guizhou Province Patentee after: Aerospace Jiangnan Group Co.,Ltd. Patentee after: JIANGNAN ELECTROMECHANICAL DESIGN Research Institute Address before: 550009 Guizhou Aerospace Industrial Park, No.7 Honghe Road, Xiaohe District, Guiyang City, Guizhou Province Patentee before: JIANGNAN ELECTROMECHANICAL DESIGN Research Institute |
|
| TR01 | Transfer of patent right |