CN110348708B - Ground target dynamic threat assessment method based on extreme learning machine - Google Patents

Ground target dynamic threat assessment method based on extreme learning machine Download PDF

Info

Publication number
CN110348708B
CN110348708B CN201910562265.3A CN201910562265A CN110348708B CN 110348708 B CN110348708 B CN 110348708B CN 201910562265 A CN201910562265 A CN 201910562265A CN 110348708 B CN110348708 B CN 110348708B
Authority
CN
China
Prior art keywords
enemy
target
threat
fighting
orientation
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.)
Active
Application number
CN201910562265.3A
Other languages
Chinese (zh)
Other versions
CN110348708A (en
Inventor
辛斌
刘清平
张佳
陈杰
杨庆凯
高冠强
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Institute of Technology BIT
Original Assignee
Beijing Institute of Technology BIT
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Institute of Technology BIT filed Critical Beijing Institute of Technology BIT
Priority to CN201910562265.3A priority Critical patent/CN110348708B/en
Publication of CN110348708A publication Critical patent/CN110348708A/en
Application granted granted Critical
Publication of CN110348708B publication Critical patent/CN110348708B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • 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
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0635—Risk analysis of enterprise or organisation activities
    • 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
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis

Abstract

The invention discloses a ground target dynamic threat assessment method based on an extreme learning machine, which comprehensively considers the fighting ability and the fighting intention of an enemy aiming at the ground target of the enemy, calculates the threat degree of an enemy unit in real time and assists a commander to make a quick decision in a rapidly-changing complex fighting situation. The method comprises the steps of obtaining 7 types of evaluation indexes of the enemy in the aspects of fighting intention and fighting capacity; wherein, the evaluation indexes reflecting the fighting ability comprise the hitting ability, the hitting radius and the detection radius of the enemy target; the evaluation indexes reflecting the fighting intention comprise the orientation of the enemy target, the distance from the enemy fighting group, the speed and the acceleration of the enemy target; wherein the evaluation index of the orientation of the enemy target is as follows: the number of the fighting units of each party in a semicircular range with the orientation angle of the enemy target as a central axis and the radius as a detection radius; and inputting the 7 types of evaluation indexes into a pre-trained extreme learning machine, and outputting a threat evaluation result by the extreme learning machine.

Description

Ground target dynamic threat assessment method based on extreme learning machine
Technical Field
The invention belongs to the field of threat estimation in data fusion processing, relates to a ground target threat assessment method, and more particularly relates to a ground target dynamic threat assessment method based on Extreme Learning Machine (ELM).
Background
With the rapid development of computer technology, network and communication technology and artificial intelligence technology, the traditional combat forms and modes have been greatly changed. Informationized warfare has evolved gradually into the main form of modern warfare, and various new technologies and new weapons are applied to actual combat systems. In the process of informatization combat, in the face of massive battlefield data, "know and know each other" is a key condition for capturing victory, and the purpose of threat assessment is to quantitatively assess the threat size formed by the military force deployment situation of the enemy on the enemy. Only if the threat degree of the target is accurately evaluated, scientific and reasonable military force deployment and operational planning can be carried out.
Current methods for threat assessment are broadly divided into two categories: model inference-based methods and data-driven methods. Typical model inference methods include: fuzzy reasoning, multi-attribute decision, DS (Dempster-Shafer) evidence theory, cloud model, Bayesian reasoning and the like. The commonality of the model inference type method is that the modeling is carried out on the threat assessment process, and the method is accurate and reliable, but the model is complex and the operation time is long, so that the real-time performance is poor. Typical data-driven methods include: the method comprises the following steps of considering threat assessment as a prediction problem of a nonlinear multivariate function, training a certain prediction model by using combat data, inputting assessment parameters and outputting the magnitude of a target threat value in the assessment process, wherein the method comprises the implicit algorithms of a neural network, an SVM (support vector machine) and the like. The method can realize a rapid and accurate threat assessment process under the condition of ensuring the sample to be reliable.
The extreme learning machine is a simple and efficient single-hidden-layer feedforward neural network, the core idea of the algorithm is to randomly initialize the input weight and bias of the network and construct a single-hidden-layer neural network without iteration, and compared with the traditional gradient-based feedforward neural network learning algorithm, the output weight of the network can be obtained through further calculation. The ELM has the obvious advantages of simple realization, high learning speed, less human intervention and the like, and becomes one of the popular research directions in the field of artificial intelligence at present.
Disclosure of Invention
In view of the above, the invention discloses a ground target dynamic threat assessment method based on an extreme learning machine, aiming at the problem that the traditional threat assessment method cannot meet the requirements of accuracy and real-time performance at the same time.
In order to solve the technical problem, the invention is realized as follows:
a ground target dynamic threat assessment method based on an extreme learning machine comprises the following steps:
acquiring 7 types of evaluation indexes of the enemy in the aspects of the fighting intention and the fighting capacity; wherein, the evaluation indexes reflecting the fighting ability comprise the hitting ability, the hitting radius and the detection radius of the enemy target; the evaluation indexes reflecting the fighting intention comprise the orientation of the enemy target, the distance from the enemy fighting group, the speed and the acceleration of the enemy target; the evaluation index of the orientation of the enemy target is as follows: the number of the fighting units of each party in a semicircular range with the orientation angle of the enemy target as a central axis and the radius as a detection radius;
and inputting the 7 types of evaluation indexes into a pre-trained extreme learning machine, and outputting a threat evaluation result by the extreme learning machine.
Preferably, the 7-class evaluation index input to the extreme learning machine is preprocessed data; the pre-treatment uses the min-max normalization method.
Preferably, the threat level of the enemy is divided into five threat levels of 0, 1, 2, 3 and 4, namely 'no threat', 'low threat', 'medium threat', 'large threat' and 'large threat'; the output of the extreme learning machine is 1-dimensional, the output value is classified into an integer value, and the threat level is determined according to the size of the integer value.
Preferably, the method further comprises displaying information: and converting the obtained integral value of the threat degree of the ground target to the fighting group of the party into a corresponding language description value to be displayed in the interactive interface.
Preferably, dynamically evaluating the threat level of the enemy target is realized by continuously extracting the evaluation index and inputting the evaluation index into the extreme learning machine to obtain the evaluation result until the enemy target disappears or is eliminated.
Has the advantages that:
firstly, the model structure for evaluating the ground target threat of the enemy is complex, and a plurality of factors influencing the threat evaluation result need to be considered. The invention combines the actual combat, establishes an evaluation index system based on the combat ability and the combat intention of the enemy, and takes the integrity, the non-redundancy property and the computability into consideration for the selection of the evaluation index. The assessment index system is combined with an ELM network, can ensure the rapidity of assessment while ensuring the accuracy of assessment, is suitable for rapidly changing battlefield dynamics, and has certain reference significance for helping fighters grasp battlefield situations and make reasonable combat plans.
Secondly, the invention improves the expression mode of the orientation of the enemy target, and adopts the number of the fighting units of the enemy within the semicircular range with the orientation angle of the enemy target as the central axis to express the orientation of the enemy target, so that the orientation angle information of the enemy target is avoided being directly processed, the fighting intention of the enemy target can be more intuitively displayed, the threat to the fighting groups of the enemy is further reflected, meanwhile, the processing mode is also favorable for the normalization of input data, the processing of an extreme learning machine is convenient, and the evaluation accuracy is improved.
Drawings
FIG. 1 is a flow chart of a protocol of the present invention;
FIG. 2 is a diagram illustrating evaluation criteria;
FIG. 3 is a diagram of an ELM network architecture for threat assessment.
Detailed Description
The invention is described in detail below by way of example with reference to the accompanying drawings.
The invention provides a ground target dynamic threat assessment method based on an ELM network aiming at the problem of ground target threat assessment and comprehensively considering the fighting capacity and the fighting intention of an enemy fighting unit.
The invention discloses a ground target dynamic threat assessment method based on an ELM network, which comprises the following steps:
step 1, selecting an evaluation index:
the model for evaluating the ground target threat of the enemy is complex in structure, and various factors influencing the threat evaluation result need to be considered. In order to ensure the accuracy of threat assessment as much as possible, the assessment indexes are selected according to the principles of completeness, non-redundancy and computability. As shown in fig. 2, the present invention selects corresponding evaluation indexes from two major aspects of the fighting ability and the fighting intention of the enemy ground target, and more specifically, the fighting ability is mainly determined according to the type of the enemy target, specifically, the enemy ground target is refined into the detection radius, the attack radius and the attack ability of the enemy target according to the weapon and the detection equipment equipped for the enemy target; the fighting intention mainly reflects the position and motion relation between the enemy and the my party and can be detailed into the distance between the target and the fighting group of the party, the target orientation, the speed and the acceleration. The constructed index system has complete content and can better reflect the threat degree information of the enemy target in the confronting environment of ground enemy and my; and the index system has low redundancy, and can realize quick and accurate threat assessment after being matched with an extreme learning machine.
Step 2, preprocessing of sample data:
and extracting the 7 indexes from the historical combat data, and acquiring threat information as a label to form a training sample. Selecting a normalization method as a data preprocessing means, more specifically, selecting a min-max normalization method to process initial data of each evaluation index, and converting threat degree label information into an integer value convenient for ELM network processing, specifically as follows:
strike capability: the maximum value of the hitting ability in all units of the enemy in the sample is recorded as AmaxMinimum value is denoted AminAnd the hitting ability of the enemy target (or called an evaluation unit) is marked as A, the value of the hitting ability of the unit after being normalized is as follows:
Figure BDA0002108601390000051
in the actual threat assessment, AmaxAnd AminThe value determined from the sample is used.
Second, striking radius: the maximum value of the attack radius in all units of the enemy is recorded as rmaxMinimum value is denoted as rminAnd the striking radius of the evaluation unit is recorded as r, and the normalized striking radius of the unit is as follows:
Figure BDA0002108601390000052
in the actual execution of the threat assessment,rmaxand rminThe value determined from the sample is used.
Detecting radius: the maximum value of the detection radius in all units of the enemy in the sample is recorded as RmaxMinimum value is denoted as RminAnd the detection radius of the evaluation unit is recorded as R, and the normalized detection radius of the unit is as follows:
Figure BDA0002108601390000053
in the actual threat assessment, RmaxAnd RminThe value determined from the sample is used.
Velocity and acceleration: the maximum value of the speed and the acceleration in all units of the enemy in the sample is recorded as vmaxAnd amaxMinimum value is denoted vminAnd aminAnd the speed and the acceleration of the evaluation unit are recorded as v and a, and the normalized values of the speed and the acceleration of the unit are as follows:
Figure BDA0002108601390000054
Figure BDA0002108601390000055
in the actual threat assessment, vmaxAnd amax、vminAnd aminThe value determined from the sample is used.
Distance: in this embodiment, the battlefield is limited in a rectangular area with a length X of 10 km and a width Y of 6 km, a cartesian coordinate system is established with the lower left corner of the rectangular area as the origin of coordinates, a connection line from the origin of coordinates to the lower right corner of the rectangular area is taken as the positive X-axis direction, and a connection line from the origin of coordinates to the upper left corner of the rectangular area is taken as the positive Y-axis direction. The coordinates of the enemy and my units are determined by the geometric centers of the enemy and my units.
Reading the coordinate information of an evaluation unit, and calculating the distance from the evaluation unit to the jth unit of the party, wherein the distance is expressed by Euclidean distance, and the calculation method comprises the following steps:
Figure BDA0002108601390000061
wherein (x)e,ye) Coordinates representing evaluation units, (x)j,yj) Coordinates representing the jth unit of my party. Calculating the nearest distance from the evaluation unit to the unit of the party as a distance index from the evaluation unit to the fighting group of the party, wherein the formula is as follows:
Figure BDA0002108601390000062
the above d value is calculated for each enemy unit in the sample, and the maximum value is taken as dmaxAnd the normalized value of the distance from the evaluation unit to the fighting group of the party is as follows:
Figure BDA0002108601390000063
in the actual threat assessment, dmaxThe value determined from the sample is used.
The target orientation: the orientation angle of the evaluation unit is defined as the angle between its orientation and the positive direction of the X-axis of the coordinate system. The number of the unit of our battle in the range of plus or minus 90 degrees of the orientation angle of the evaluation unit, namely, in the range of a semicircle with the orientation direction as the central axis and the radius as the detection radius reflects the fighting intention of the enemy.
Let the orientation angle of the evaluation unit be α and the coordinate be (x)p,yp) The coordinate of my square unit j is (x)j,yj) Then the vector with the evaluation unit pointing to my unit j
Figure BDA0002108601390000064
Is expressed as (x)j-xp,yj-yp). Converting orientation angle information of evaluation unit intoVector form: let the orientation angle of the evaluation unit be α, where α ∈ [ - π, π), the orientation vector is set as the unit vector with its origin at the origin. The end point coordinate (x)v,yv) Comprises the following steps:
Figure BDA0002108601390000065
yv＝sinα
then evaluate the orientation vector of the unit
Figure BDA0002108601390000066
Is expressed as (x)v,yv). Then:
Figure BDA0002108601390000067
wherein n isjA value of 1 indicates that the unit of my j is within plus or minus 90 degrees of the orientation angle of the evaluation unit and 0 indicates that it is not within this range. Wherein R represents the detection radius of the evaluation unit,
Figure BDA0002108601390000071
the specific calculation method for representing the dot product operation of the vector is as follows:
Figure BDA0002108601390000072
the number of my unit N within the range of plus or minus 90 degrees of the orientation angle of the enemy targetpComprises the following steps:
Figure BDA0002108601390000073
and if the total number of the unit of our party is N, the normalized value of the unit number of our party in the range of plus or minus 90 degrees of the orientation angle of the enemy target is as follows:
Figure BDA0002108601390000074
seventh, threat degree: the threat degree of the enemy is divided into five levels of no threat, low threat, medium threat, large threat and large threat, which are converted into five threat levels of 0, 1, 2, 3 and 4.
Step 3, the specific steps of the construction, training and testing of the ELM network are as follows:
s301, constructing a network: as shown in fig. 3, since the total number of indicators evaluating an enemy target is 7, and the final output result is the threat level of the enemy, the number of input neurons of the ELM network is determined to be 7, the number of output neurons is determined to be 1, and the number of hidden layer neurons is determined to be the optimal number through experiments. The activation function g (x) employs Sigmoid function, namely:
Figure BDA0002108601390000075
in this embodiment, if the number of output neurons is 1, different threat levels are expressed according to different values. In practice, the number of output neurons may also be designed to be the same as the number of threat levels, e.g. 5 here, each output neuron representing a threat.
S302, training of a network: samples which are subjected to data preprocessing are used as training samples and testing samples, wherein the ratio of the training samples to the testing samples is 4: 1. Given n training samples X ═ X1,x2,...,xn)∈Rn×7The label is Y ═ (Y)1,y2,...,yn)∈Rn×1. Input weight W ═ of the ELM network (W)ij)∈R7×LIs chosen randomly, where L represents the number of hidden layer neurons. The input H to the hidden layer is the same as for computing a conventional forward propagation network: g (W, x), wherein H ∈ Rn×LThe output of the network is Y' ═ H β, where β is the output weight to be found.
The training objective of the ELM network is to solve for the output weight β by minimizing the sum of the prediction error loss functions, the objective function being:
Figure BDA0002108601390000081
wherein the first term is a regular term for preventing parameter overfitting, and C in the second term is a penalty coefficient of a prediction error term. Solving the above objective function can be regarded as a least squares optimization problem, making the gradient of the objective function for β zero available:
β+CHT(Y-Hβ)＝0
the optimal solution of beta can be obtained according to the Moore-Penrose generalized inverse matrix. Considering that the number of training samples is greater than or equal to the number of hidden layer neurons, the optimal solution for β is:
Figure BDA0002108601390000082
wherein, I20Representing a 20-dimensional identity matrix.
S303, testing the network: and (4) giving m test samples, and testing the prediction accuracy of the trained network so as to evaluate the performance and classification capability of the model. The method for calculating the test accuracy comprises the following steps:
Figure BDA0002108601390000083
wherein m iscThe number of test samples representing correct prediction is higher, and the higher the test accuracy is, the stronger the performance and classification capability of the model is. When the test accuracy fails to reach the required accuracy, the network is retrained until the required accuracy is reached.
Step 4, data preprocessing: preprocessing the real-time battlefield data including the coordinates of enemy targets, the hitting capability, the hitting radius, the detection radius, the speed, the acceleration, the heading angle, the coordinates of each unit of our party and the like according to the mode in the step 2, storing the processed data in an array according to the format of sample data, and using X to belong to Rq×7Where q represents the number of evaluation units.
And 5, threat assessment based on the ELM network: and inputting the preprocessed data into a trained ELM network meeting the test precision to obtain the threat degree of the ground target to the fighting group of one party in the actual battlefield.
And step 6, information display: converting the integer value of the threat degree of the ground target to the battle group of the party into a language description value to be displayed in an interactive interface, wherein the conversion mode is as follows: 0. 1, 2, 3, 4 correspond to "no threat", "low threat", "medium threat", "large threat" and "large threat", respectively.
And continuously performing the steps 4 to 6 to dynamically evaluate the threat degree of each unit of the enemy until the target unit disappears or is eliminated.
The following describes a ground target dynamic threat assessment method based on an extreme learning machine in combination with simulation experiment results;
300 threat assessment samples were selected as a sample set, 240 of which were used as training sets and 60 of which were used as test sets. The structure of the ELM network is as follows: the number of input layer neurons is 7, the number of output layer neurons is 1, and the optimal number of hidden layer neurons is 20 determined by experiments. The training results are:
training time Training accuracy Time of measurement Test accuracy
0.947s 91.25％ 0.012s 93.30％
The test precision meets the requirement of more than 90 percent, which shows that the method has stronger classification capability.
Note: the results were obtained on a PC configured as Intel (R) Xeon (R) CPU E5-2620v4@2.10GHz,32GB memory.
In the dynamic threat assessment simulation process, it is set that both the enemy and the my have 5 ground combat platforms, and the parameters at a certain moment are as follows in the table 1:
TABLE 1
Figure BDA0002108601390000091
Figure BDA0002108601390000101
The results obtained after normalization in step 2 for each parameter are shown in table 2:
TABLE 2
Figure BDA0002108601390000102
The normalized data in the table are stored in a matrix X ∈ R5×7And inputting the real-time threat assessment result into an input layer of the ELM network, and referring to a table 3:
TABLE 3
Name of unit ELM network output result Threat assessment results
Enemy Unit 1 1 "threat Low"
Enemy unit 2 0 No threat'
Enemy unit 3 4 "great threat"
Enemy unit 4 2 "threat moderate"
Enemy unit 5 0 No threat'
The embodiments disclosed above are implemented on the premise of the technical solution of the present invention, and detailed embodiments and specific operation procedures are given, but the scope of the present invention is not limited to the embodiments. Many modifications and variations of the present invention can be made in light of the above teachings, and certain values have been set forth in this example only to better illustrate the principles and applications of the present invention, and to thereby enable better understanding and use. The invention is not limited to the specific embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (6)

1. A ground target dynamic threat assessment method based on an extreme learning machine is characterized by comprising the following steps:
acquiring 7 types of evaluation indexes of the enemy in the aspects of the fighting intention and the fighting capacity; wherein, the evaluation indexes reflecting the fighting ability comprise the hitting ability, the hitting radius and the detection radius of the enemy target; the evaluation indexes reflecting the fighting intention comprise the orientation of the enemy target, the distance from the enemy fighting group, the speed and the acceleration of the enemy target; the evaluation index of the orientation of the enemy target is as follows: the number of the fighting units of each party in a semicircular range with the orientation angle of the enemy target as a central axis and the radius as a detection radius;
the method for extracting the evaluation indexes of the orientation of the enemy target comprises the following steps:
defining a battlefield in a rectangular area with the length of X kilometers and the width of Y kilometers, establishing a Cartesian coordinate system by taking the lower left corner of the rectangular area as a coordinate origin, taking a connecting line from the coordinate origin to the lower right corner of the rectangular area as a positive X-axis direction, and taking a connecting line from the coordinate origin to the upper left corner of the rectangular area as a positive Y-axis direction; the coordinates of the enemy and my units are determined by the geometric center of the units;
let the orientation angle of the enemy object be α, where α ∈ [ - π, π), and the coordinate be (x)p,yp) The coordinate of my square unit j is (x)j,yj) Then the enemy target points to the vector of my unit j
Figure FDA0002663266260000011
Is expressed as (x)j-xp,yj-yp) (ii) a Converting the orientation angle of the enemy target into a vector form: the orientation vector is a unit vector having a starting point at the origin, and the coordinates (x) of the ending pointv,yv) Comprises the following steps:
Figure FDA0002663266260000012
yv＝sinα
the orientation vector of the enemy target
Figure FDA0002663266260000013
Is expressed as (x)v,yv) And then:
Figure FDA0002663266260000014
wherein n isjA value of 1 indicates that the unit of my j is within plus or minus 90 degrees of the enemy target heading angle, and 0 indicates that it is not within this range; wherein R represents the detection radius of the enemy target,
Figure FDA0002663266260000015
a dot product operation representing a vector;
the number of my unit N within the range of plus or minus 90 degrees of the orientation angle of the enemy targetpComprises the following steps:
Figure FDA0002663266260000021
and if the total number of the unit of our party is N, the normalized value of the unit number of our party in the range of plus or minus 90 degrees of the orientation angle of the enemy target is as follows:
Figure FDA0002663266260000022
will NGAs an evaluation index of the enemy target orientation;
and inputting the 7 types of evaluation indexes into a pre-trained extreme learning machine, and outputting a threat evaluation result by the extreme learning machine.
2. The ground target dynamic threat assessment method according to claim 1, wherein the 7-class assessment indicators input to the extreme learning machine are preprocessed data; the pre-treatment uses the min-max normalization method.
3. The method for evaluating dynamic threats of ground targets according to claim 1, wherein the index of the distance from the battle group of the same party is extracted in a manner of:
reading coordinate information of an enemy target, and calculating Euclidean distance from the enemy target to each unit in a team of our party; recording the nearest distance from the enemy target to all units of the enemy as d; d is normalized to obtain the distance index d from the enemy target to the fighting group of the partyGComprises the following steps:
Figure FDA0002663266260000023
wherein d ismaxThe maximum value of the d values calculated for all enemy targets when the network was trained.
4. The ground target dynamic threat assessment method according to claim 1, wherein the threat level of the enemy is classified as "no threat", "low threat", "medium threat", "large threat" or "large threat", which is converted into five threat levels of 0, 1, 2, 3 and 4; the output of the extreme learning machine is 1-dimensional, the output value is classified into an integer value, and the threat level is determined according to the size of the integer value.
5. The ground-based target dynamic threat assessment method of claim 1, further comprising displaying information: and converting the obtained integral value of the threat degree of the ground target to the fighting group of the party into a corresponding language description value to be displayed in the interactive interface.
6. The ground target dynamic threat assessment method according to claim 1, wherein dynamically assessing the threat level of the enemy target is achieved by continuously extracting assessment indexes and inputting the assessment indexes into the extreme learning machine to obtain assessment results until the enemy target disappears or is eliminated.
CN201910562265.3A 2019-06-26 2019-06-26 Ground target dynamic threat assessment method based on extreme learning machine Active CN110348708B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910562265.3A CN110348708B (en) 2019-06-26 2019-06-26 Ground target dynamic threat assessment method based on extreme learning machine

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910562265.3A CN110348708B (en) 2019-06-26 2019-06-26 Ground target dynamic threat assessment method based on extreme learning machine

Publications (2)

Publication Number Publication Date
CN110348708A CN110348708A (en) 2019-10-18
CN110348708B true CN110348708B (en) 2021-01-01

Family

ID=68183226

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910562265.3A Active CN110348708B (en) 2019-06-26 2019-06-26 Ground target dynamic threat assessment method based on extreme learning machine

Country Status (1)

Country Link
CN (1) CN110348708B (en)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111346382B (en) * 2020-02-21 2023-08-11 腾讯科技（深圳）有限公司 Method, device and system for determining virtual target object
CN111583083B (en) * 2020-05-13 2023-12-19 飒铂智能科技有限责任公司 Method and system for determining threat degree of non-cooperative targets in low-altitude flight to ground
CN111612673B (en) * 2020-05-13 2023-12-15 飒铂智能科技有限责任公司 Method and system for confirming threat degree of unmanned aerial vehicle to multiple places
CN112149715B (en) * 2020-08-31 2024-04-02 南京航空航天大学 Unmanned aerial vehicle air combat threat assessment method based on deep learning
CN112598046B (en) * 2020-12-17 2023-09-26 沈阳航空航天大学 Target tactical intent recognition method in multi-machine cooperative air combat
CN113393082B (en) * 2021-05-11 2024-01-09 北京机电工程研究所 Fight control method for agile response of complex scene
CN113884805B (en) * 2021-09-17 2023-10-20 国网山东省电力公司德州供电公司 Single-phase earth fault insulation degradation identification method and system based on D-S evidence theory

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10057367B2 (en) * 2016-03-02 2018-08-21 Huawei Technologies Canada Co., Ltd. Systems and methods for data caching in a communications network
CN106022517A (en) * 2016-05-17 2016-10-12 温州大学 Risk prediction method and device based on nucleus limit learning machine
CN107025598A (en) * 2017-04-06 2017-08-08 中国矿业大学 A kind of individual credit risk appraisal procedure based on extreme learning machine
CN107194155A (en) * 2017-05-02 2017-09-22 西北工业大学 A kind of threat assessment modeling method based on small data set and Bayesian network
CN108460425A (en) * 2018-03-27 2018-08-28 中国航空无线电电子研究所 Vacant lot method for situation assessment based on particular network structure
CN109740876A (en) * 2018-12-20 2019-05-10 北京冠群桦成信息技术有限公司 Target threat judgment method

Also Published As

Publication number Publication date
CN110348708A (en) 2019-10-18

Similar Documents

Publication Publication Date Title
CN110348708B (en) Ground target dynamic threat assessment method based on extreme learning machine
Xu et al. Multi-attribute decision making method for air target threat evaluation based on intuitionistic fuzzy sets
Xue et al. Panoramic convolutional long short-term memory networks for combat intension recognition of aerial targets
CN102222240B (en) DSmT (Dezert-Smarandache Theory)-based image target multi-characteristic fusion recognition method
US11574223B2 (en) Method and apparatus for rapid discovery of satellite behavior
CN112749761A (en) Enemy combat intention identification method and system based on attention mechanism and recurrent neural network
CN113128021B (en) Real-time re-decision method and system for cooperative confrontation of multiple unmanned platforms
Wei A Risk Evaluation Method for the High-Technology Project Investment Based on ET-WA Operator with 2-Tuple Linguistic Information.
CN112926739A (en) Network countermeasure effectiveness evaluation method based on neural network model
Hu et al. Robust decision making for UAV air-to-ground attack under severe uncertainty
CN113919425B (en) Autonomous aerial target allocation method and system
CN116432514A (en) Interception intention recognition strategy simulation system and method for unmanned aerial vehicle attack and defense game
CN115640755A (en) Air combat incomplete information data processing method based on extreme learning machine
CN115909027A (en) Situation estimation method and device
CN113128698B (en) Reinforced learning method for multi-unmanned aerial vehicle cooperative confrontation decision
CN115186378A (en) Real-time solution method for tactical control distance in air combat simulation environment
CN115457809A (en) Multi-agent reinforcement learning-based flight path planning method under opposite support scene
Chen et al. A Human-Machine Agent Based on Active Reinforcement Learning for Target Classification in Wargame
Zhang et al. Research on armored unit target threat assessment based on SVM
CN115563861B (en) Performance comprehensive evaluation and optimization method for intelligent tracking algorithm of radar seeker
Sova et al. Development of a methodology for training artificial neural networks for intelligent decision support systems
Yu et al. Comparative analysis and development exploration of threat assessment methods for warship air defense
Chen et al. A wireless gunshot recognition system based on tri-axis accelerometer and lightweight deep learning
Zhang Research on Weighted Model of Air Defense Weapon-Target Assignment Based on AHP and Entropy Method
Muslikhin et al. Troop camouflage detection based on deep action learning

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