CN113627749B - Damage efficiency evaluation and fire planning method based on convolutional neural network - Google Patents

Damage efficiency evaluation and fire planning method based on convolutional neural network Download PDF

Info

Publication number
CN113627749B
CN113627749B CN202110832886.6A CN202110832886A CN113627749B CN 113627749 B CN113627749 B CN 113627749B CN 202110832886 A CN202110832886 A CN 202110832886A CN 113627749 B CN113627749 B CN 113627749B
Authority
CN
China
Prior art keywords
damage
target
missile
probability
aiming
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
CN202110832886.6A
Other languages
Chinese (zh)
Other versions
CN113627749A (en
Inventor
熊芬芬
吴巍
张�成
任成坤
李超
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.)
Pla 63863 Unit
Beijing Institute of Technology BIT
Original Assignee
Pla 63863 Unit
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 Pla 63863 Unit, Beijing Institute of Technology BIT filed Critical Pla 63863 Unit
Priority to CN202110832886.6A priority Critical patent/CN113627749B/en
Publication of CN113627749A publication Critical patent/CN113627749A/en
Application granted granted Critical
Publication of CN113627749B publication Critical patent/CN113627749B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • 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
    • G06Q10/0637—Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals

Abstract

The invention provides a damage efficiency evaluation and fire planning method based on a convolutional neural network, which can fully consider the influence of various uncertain random factors and improve the rationality and precision of fire planning. Meanwhile, the damage probability evaluation process is regarded as a regression problem from an image to a numerical value, a convolutional neural network model for damage evaluation is trained, and the problem that firepower planning consumes time is solved. A convolutional neural network is introduced, the geometric shape and coordinates of the two-dimensional projection of the target and the corresponding aiming point coordinates are converted into an image, and the damage probability of the target under any given aiming point is predicted by constructing the convolutional neural network based on an image processing mode, so that the damage efficiency evaluation process is very visual, and meanwhile, the guarantee is provided for quickly planning firepower.

Description

Damage efficiency evaluation and fire planning method based on convolutional neural network
Technical Field
The invention relates to damage performance evaluation and fire planning, in particular to a damage performance evaluation and fire planning method based on a Convolutional Neural Network (CNN).
Background
The existing damage efficiency evaluation and fire planning method has the following defects:
(1) When the damage performance is evaluated, the concept of the damage radius is basically based on, the difference is far from the actual situation, the damage performance evaluation is not ideal, and the fire power planning scheme cannot achieve the expected effect in the actual situation.
(2) When the damage performance is evaluated, uncertainty of the target position caused by detection errors is not considered, and the detection errors may be very large in practice, so that the damage performance is seriously influenced.
(3) When the damage evaluation method is used for multiple (more than 4) edge-shaped surface targets, the effective area of the surface target is not well divided, and meanwhile, when the damage efficiency is calculated, a damage amplitude worker is often similar to a circle or an ellipse, the difference from a real damage amplitude worker (such as a shoe-shaped gold ingot, a crescent and the like) is large, and the error is large when the damage evaluation is carried out.
Disclosure of Invention
The invention provides a damage efficiency evaluation and fire planning method based on a convolutional neural network, which can fully consider the influence of various uncertain random factors and improve the reasonability and the accuracy of fire planning. Meanwhile, the damage probability evaluation process is regarded as a regression problem from an image to a numerical value, a convolutional neural network model for damage evaluation is trained, and the problem that firepower planning consumes time is solved.
The invention discloses a damage performance evaluation method based on a convolutional neural network, which comprises the following steps:
step 1, constructing a damage efficiency evaluation agent model based on a convolutional neural network;
the damage performance evaluation agent model is input as an image, the image is obtained by converting target information in a data form and distribution information of corresponding missile aiming points, and the target information comprises a target category, a geometric dimension and coordinates;
the damage efficiency evaluation agent model outputs damage probability corresponding to target information and missile aiming point distribution information;
step 2, obtaining a plurality of groups of different target information and damage probabilities under corresponding missile aiming point distribution information;
respectively converting a plurality of groups of different target information and corresponding missile aiming point distribution information into images to obtain a plurality of groups of images;
corresponding the multiple groups of images to respective damage probabilities one by one to serve as training samples and test samples;
step 3, training the damage performance evaluation agent model by using a training sample, and testing the precision of the trained damage performance evaluation agent model by using a test sample until the trained damage performance evaluation agent model meets the set precision to obtain the trained damage performance evaluation agent model;
and 4, converting target information to be evaluated and corresponding missile aiming point distribution information into images, inputting the images into the trained damage efficiency evaluation agent model to obtain corresponding damage probability, and finishing damage efficiency evaluation.
Preferably, in the step 2, the specific way of obtaining damage probabilities under a plurality of groups of different target information and corresponding missile aiming point distribution information is as follows:
under the condition of a plurality of groups of different target information and corresponding missile aiming point distribution information, respectively obtaining a damage matrix representing whether the missile damages the target or not according to the target category and the geometric overall dimension; obtaining a scattering law of the target position according to the target coordinates; obtaining a dispersion law of the missile aiming points according to the distribution information of the missile aiming points;
and obtaining a plurality of groups of different target information and damage probabilities corresponding to the distribution information of the corresponding missile aiming points according to the damage matrix, the distribution law of the target positions and the distribution law of the missile aiming points.
Preferably, for a point target, the damage probability of the K successive-fire missiles to the target is as follows:
Figure GDA0003880521810000021
wherein, P k The damage probability of the K-th missile to the target is K =1,2,3 \ 8230k;
for a planar target, under k independent shots, the damage probability to the target is as follows:
Figure GDA0003880521810000031
wherein N is the number of missile landing points and target positions sampled according to the scattering law,
Figure GDA0003880521810000032
the damaged area ratio of the target opposite to the jth sampling point under the combined action of the target opposite to the k-missile is shown, and j =1,2,3 \8230N.
Preferably, for a point target, the damage probability of a single missile to the target is calculated in the following way:
Figure GDA0003880521810000033
wherein, (x, y) represents coordinates of a point on the ground;
f (x, y) represents the spread law of the target position;
l (x, y) represents the dispersion law of the missile aiming points;
g (x, y) is a damage matrix converted by a damage amplitude transformer, and represents whether a certain position of a target opposite to the missile is damaged or not, wherein the damage is 1, and otherwise, the damage is 0;
for a surface target, the surface target is cut into a plurality of rectangular surface elements according to a circumscribed rectangle, the rectangular surface elements in the surface target are recorded as effective surface elements, and the damage probability of a single missile to the target is as follows:
Figure GDA0003880521810000034
wherein N is the number of missile landing points and target positions sampled according to the scattering law; f j (S) the proportion of the damaged area of the opposite target of the jth sampling point under the action of the single missile on the opposite target; m represents the number of active bins, (x) i ,y i ) The coordinates of the center point of the ith effective surface element are shown, i =1,2,3 \8230M.
Preferably, a plurality of effective surface elements are combined into a large surface element, and the large surface element is used as a new effective surface element to calculate the damage probability of the single missile on the target.
The invention also provides a fire planning method, which comprises the following steps:
the firepower planning problem under the expected damage efficiency is summarized into that the optimal aiming point coordinate is searched under the condition of solving the given bomb consumption, so that the total damage probability is maximum;
solving the fire planning problem, comprising the sub-steps of:
step 11, setting initial bullet consumption;
step 12, converting the target information and the distribution information of the corresponding missile aiming points into images, calling the trained damage performance evaluation agent model in the damage performance evaluation method, and calculating damage probabilities under different missile aiming point distribution information to obtain the maximum damage probability under the current missile consumption;
and step 13, judging whether the maximum damage probability under the current bomb consumption meets the given expected damage efficiency, if so, the aiming point coordinate corresponding to the maximum damage probability is the optimal aiming point coordinate, if not, the bomb consumption is increased by 1, the step 12 is returned to be executed until the given expected damage efficiency is met, the current bomb consumption, the optimal aiming point coordinate and the current damage probability are output, and the firepower planning is completed.
Preferably, the fire planning problem given the expected damage performance is:
Figure GDA0003880521810000041
wherein (x) i ,y i ) Is the target position of the ith missile, x l ,x u ,y l And y u The range of the aiming point is set according to the actual target position, the spokesman, the reconnaissance error and the warhead damage spokesman.
Preferably, a genetic algorithm is used to solve the fire planning problem given the expected damage performance.
Has the advantages that:
the method introduces the convolutional neural network into the damage efficiency evaluation, converts the geometric shape and the coordinates of the two-dimensional projection of the target and the corresponding aiming point coordinates into an image, and predicts the damage probability of the target under any given aiming point by constructing the convolutional neural network based on an image processing mode, so that the damage efficiency evaluation process is very intuitive, and meanwhile, the guarantee is provided for quickly planning firepower. Uncertainty of a target position and a missile aiming point is introduced into damage efficiency evaluation, and an obtained evaluation result is more in line with a real situation.
In the damage efficiency evaluation of the invention, for a polygonal surface target, when the damage efficiency is evaluated by a damage spotter, the damage spotter is designed according to the actual distribution of the fragments of the warhead under different working conditions, and the damage spotter which is more accurate than a circle and an ellipse is formed by fitting the distribution quantity of the fragments of each area in a ground net target file generated by reading simulation or experimental data. Firstly, a surface target is cut into a plurality of rectangular surface elements according to a circumscribed rectangle, and then the rectangle in the surface target is identified and marked as an effective surface element. At this time, the damage probability of the current missile to the target can be obtained only by calculating the proportion of the number of damaged effective surface elements to the total number of effective surface elements.
In the damage efficiency evaluation of the invention, for the long-distance acting missile, the convolution filter principle is used for reference, a data extraction mode is designed, the number of horizontal (m) effective surface elements and vertical (n) effective surface elements can be flexibly combined, a plurality of effective surface elements (m x n) are combined into a large surface element, the large surface element is used as a new effective surface element to carry out damage probability calculation of a single missile on a target, when the calculation is carried out, all the effective surface elements in the large surface element are considered to be damaged as long as the large surface element is judged to be damaged, and only the large surface element needs to be traversed without traversing each effective surface element, so that the calculation efficiency can be greatly improved.
According to the fire planning method, as the damage probability evaluation is carried out based on the damage efficiency evaluation agent model, optimization only needs to continuously call the damage efficiency evaluation agent model, and the problem that the fire planning calculation is time-consuming is effectively solved.
According to the fire planning method, the optimization problem is solved by adopting intelligent optimization algorithms such as a genetic algorithm and the like according to given set conditions, so that the optimal scheme of fire planning is obtained, and the acquisition of an optimal global optimal solution is ensured.
Drawings
FIG. 1 is a diagram of a convolutional neural network architecture of the present invention.
Fig. 2 (a) is a schematic diagram of an effective surface element in the effective area division of the surface object according to the present invention.
Fig. 2 (b) is a schematic diagram of a large surface element combined by a plurality of effective surface elements (m × n) when the effective area of the surface object is divided.
Fig. 3 is a flow chart of the fire planning of the present invention.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings.
The invention discloses a damage performance evaluation method based on a convolutional neural network, which comprises the following steps:
step 1, constructing a damage efficiency evaluation agent model based on a convolutional neural network, which comprises the following specific steps:
considering that the most time-consuming part in the whole fire planning is damage performance evaluation, especially multi-source uncertainties such as detection, guidance and damage need to be considered, after optimization iteration is involved, the whole fire planning faces the problem of large calculated amount, and the requirement of rapid battlefield planning cannot be met. Therefore, by utilizing the deep learning technology in the field of image processing, a damage performance evaluation proxy model based on the convolutional neural network is constructed by designing a convolutional layer, a pooling layer and a linear layer of the network, and the structural diagram of the convolutional neural network is shown in fig. 1.
The damage performance evaluation agent model is input as an image, the image is obtained by converting target information in a data form and distribution information of corresponding missile aiming points, and the target information comprises a target category, a geometric dimension and coordinates;
and outputting the damage probability corresponding to the target information and the missile aiming point distribution information by the damage efficiency evaluation agent model.
Step 2, generating a certain number of training samples and testing samples aiming at different projectile consumption and different types of targets, converting the geometric overall dimension, the types and the missile aiming point distribution of the striking targets into image formats, and corresponding to the damage probabilities under respective striking conditions one by one, specifically:
under the condition of a plurality of groups of different target information and corresponding missile aiming point distribution information, respectively obtaining a damage matrix representing whether the missile damages the target or not according to the target category and the geometric overall dimension; obtaining a scattering law of the target position according to the target coordinates; obtaining a distribution law of the missile aiming points according to the distribution information of the missile aiming points;
obtaining a plurality of groups of different target information and damage probabilities corresponding to the distribution information of the corresponding missile aiming points according to the damage matrix, the distribution law of the target positions and the distribution law of the missile aiming points;
respectively converting a plurality of groups of different target information and corresponding missile aiming point distribution information into images to obtain a plurality of groups of images;
corresponding the multiple groups of images to respective damage probabilities one by one to serve as training samples and test samples;
wherein for a single small target of a certain function, of relatively small size, such as: an aircraft, a naval vessel, a tank. Obtaining a spread standard deviation sigma of the target position from the target detection error (TEP) t = TEP, obtaining shot point scattering rule or CEP according to guided missile CEP, obtaining shot point distribution standard deviation
Figure GDA0003880521810000071
Both dispersion laws are considered to follow a zero-mean normal distribution.
For the long-distance acting missile, the concept of a coordinate damage matrix is introduced, and the killing effect on the target is further accurately evaluated. The damage probability of a single missile to a target is as follows:
Figure GDA0003880521810000072
wherein, (x, y) represents a target position; g (x, y) is a damage matrix converted by a damage amplitude transformer, and represents whether a certain position of a target opposite to the missile is damaged or not, wherein the damage is 1, and otherwise, the damage is 0; f (x, y) represents the spread law of the target position; l (x, y) represents the dispersion law of the missile aiming points.
For polygonal-surface targets, because the target area is large, an effective damage matrix can hardly be obtained, and therefore, the damage effectiveness is generally evaluated by calculating the proportion of the damaged area to the total area by a damage spotter.
When the damage effectiveness is evaluated by a damage amplitude transformer, the damage amplitude transformer is changed to be similar to a geometric shape with a great difference from the real situation, such as a circle or an ellipse, and the like, the damage amplitude transformer is designed according to the actual distribution of the fragments of the warhead under different working conditions (different drop speeds, drop angles and the like of missiles), and the damage amplitude transformer which is more accurate than the circle and the ellipse is formed by fitting the distribution quantity of the fragments of each area in a ground net target file generated by reading simulation or experimental data. Firstly, a surface target is cut into a plurality of rectangular surface elements according to a circumscribed rectangle, and then rectangles in the surface target are identified and marked as effective surface elements, as shown in fig. 2 (a). At this time, the damage probability of the current missile to the target can be obtained only by calculating the proportion of the number of damaged effective surface elements to the total number of effective surface elements.
The damage probability of a single missile to a target is calculated in the following mode:
Figure GDA0003880521810000081
g (x-x ', y-y') is a coordinate damage law representing single shooting, and represents whether a certain position of the missile opposite to the target is damaged or not, the damage is 1, otherwise, the damage is 0, wherein x 'and y' represent the central surface element coordinates of the surface target.
The damage probability of a single missile to a target can be realized by Monte Carlo integration:
Figure GDA0003880521810000082
wherein N is the number of sampling missile aiming points and target positions according to the scattering law; f j (S) the proportion of the damaged area of the opposite target of the jth sampling point under the action of the single missile on the opposite target; m represents the number of active bins, (x) i ,y i ) The coordinates of the center point of the ith active bin are indicated.
For the long-distance acting missile, because the damage area is very large, if effective surface elements are used for calculating whether the missile is damaged one by one, the calculation efficiency is very low, so that the data extraction mode is designed by using the principle of a convolution filter for reference, and the missile can be used for flexibly designing the data extraction modeThe number of the horizontal (M) and vertical (n) active surface elements is movably combined, a plurality of active surface elements (M × n) are combined into a large surface element, as shown in fig. 2 (b), when the large surface element is judged to be damaged in calculation, all the active surface elements in the large surface element are considered to be damaged, and the effective surface elements do not need to be traversed, and only the M effective surface elements need to be traversed s (M s <M) large face elements, and the calculation efficiency can be greatly improved.
In a real battlefield, only one missile or the same missile is usually not adopted, damage efficiency evaluation of multiple (types of) missiles to a target needs to be considered at the moment, the missiles are subjected to a continuous shooting mode by default, and a system error (aiming error) does not exist. For a point target, the damage probability of k consecutive-launched missiles to the target is as follows:
Figure GDA0003880521810000083
wherein, P i The damage probability of the ith missile to the target is calculated by the formula (1).
For a planar target, under k independent shots, the damage of effective surface elements in the planar target is calculated without superposition, and the damage probability of the target is as follows: the total average relative kill area is:
Figure GDA0003880521810000091
wherein N is the number of missile aiming point and target position sampled according to the dispersion law,
Figure GDA0003880521810000092
and the damaged area ratio of the j-th sampling point to the opposite target is shown under the combined action of the k-missile to the opposite target.
And 3, training the damage performance evaluation agent model by using the training sample, extracting target information and missile aiming point distribution information from the image of the training sample by using the convolutional neural network for training, and testing the precision of the trained damage performance evaluation agent model by using the test sample.
And 4, converting target information (size and category) to be evaluated and missile aiming point distribution information into an image according to a pre-designed method, and predicting damage probability by using the trained damage performance evaluation agent model to finish damage performance evaluation.
Uncertainty of a target position and a missile aiming point is introduced into the damage efficiency evaluation agent model, and an obtained evaluation result is more consistent with a real situation. Furthermore, a convolutional neural network is introduced, the geometric shape and coordinates of the two-dimensional projection of the target and the corresponding aiming point coordinates are converted into an image, and the damage probability of the target under any given aiming point is predicted by constructing the convolutional neural network based on an image processing mode, so that the damage efficiency evaluation process is very visual, and meanwhile, the guarantee is provided for quickly planning firepower.
Therefore, on the basis of the damage performance evaluation method based on the convolutional neural network, the invention also provides a fire planning method, and for both a single point target and a planar target, the fire planning essence can be summarized to find the optimal aiming point coordinate (x) under the condition of solving the given shot consumption k m ,y m ) (m =1, 2.. K) to maximize the overall damage probability.
Specifically, the fire planning problem given the expected damage performance is:
Figure GDA0003880521810000093
wherein x is l ,x u ,y l And y u The range of the aiming point is set according to the actual target position, the spotter, the reconnaissance error and the fighter who damages the warhead.
If it is required to solve the fire planning problem given the expected damage performance, the procedure shown in fig. 3 can be followed, first, the initial shot size k = n is given 0 ，n 0 To evaluate the starting point of the shot consumption, n can be given according to the damage requirement of a certain target in the actual planning 0 Default valueBy default, n 0 And =1. Then, a firepower planning model shown in the formula (6) needs to be called to solve the optimal aiming point and the maximum damage probability under the current bomb usage, namely, a solver converts target information (size and category) and aiming points into images, then the trained damage efficiency evaluation agent model is continuously called to calculate the damage probability under different aiming points, and further the aiming point coordinate enabling the damage probability to be maximum is found. And if the maximum damage probability under the current projectile consumption does not meet the given expectation, continuously calling the expression (6) to solve the projectile consumption k +1 until the optimized maximum damage probability meets the expectation, and outputting the current projectile consumption k, the optimal aiming point coordinate and the current damage probability.
According to given question setting conditions, in order to ensure the acquisition of an optimal global optimal solution, an optimal scheme of fire planning is obtained by solving the optimization problem shown in the formula (6) by adopting an intelligent optimization algorithm such as a genetic algorithm and the like. Because the damage probability evaluation is carried out based on the damage efficiency evaluation agent model, the optimization only needs to continuously call the damage efficiency evaluation agent model, and the problem that the firepower planning calculation is time-consuming is effectively solved.
In summary, the above embodiments are merely exemplary embodiments of the present invention, and any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (8)

1. A damage performance evaluation method based on a convolutional neural network is characterized by comprising the following steps:
step 1, constructing a damage efficiency evaluation agent model based on a convolutional neural network;
the damage performance evaluation agent model is input as an image, the image is obtained by converting target information in a data form and distribution information of corresponding missile aiming points, and the target information comprises a target category, a geometric dimension and coordinates;
the damage efficiency evaluation agent model outputs damage probability corresponding to target information and missile aiming point distribution information;
step 2, obtaining a plurality of groups of different target information and damage probabilities under corresponding missile aiming point distribution information;
respectively converting a plurality of groups of different target information and corresponding missile aiming point distribution information into images to obtain a plurality of groups of images;
corresponding the multiple groups of images to respective damage probabilities one by one to be used as training samples and testing samples;
step 3, training the damage performance evaluation agent model by using a training sample, and testing the precision of the trained damage performance evaluation agent model by using a test sample until the trained damage performance evaluation agent model meets the set precision to obtain the trained damage performance evaluation agent model;
and 4, converting target information to be evaluated and corresponding missile aiming point distribution information into images, inputting the images into the trained damage efficiency evaluation agent model to obtain corresponding damage probability, and finishing damage efficiency evaluation.
2. The damage performance evaluation method of claim 1, wherein in step 2, the damage probability under the multiple groups of different target information and corresponding missile aiming point distribution information is obtained by:
under the condition of a plurality of groups of different target information and corresponding missile aiming point distribution information, respectively obtaining a damage matrix representing whether the missile damages the target or not according to the target category and the geometric overall dimension; obtaining a scattering law of the target position according to the target coordinates; obtaining a dispersion law of the missile aiming points according to the distribution information of the missile aiming points;
and obtaining a plurality of groups of different target information and damage probabilities corresponding to the distribution information of the corresponding missile aiming points according to the damage matrix, the distribution law of the target positions and the distribution law of the missile aiming points.
3. The damage performance evaluation method of claim 2, wherein for a point target, the damage probability of K consecutive missiles to the target is:
Figure FDA0003880521800000021
wherein, P k The damage probability of the K-th missile to the target is K =1,2,3 \ 8230k;
for a surface target, under k independent shots, the damage probability to the target is as follows:
Figure FDA0003880521800000022
wherein N is the number of missile landing points and target positions sampled according to the scattering law,
Figure FDA0003880521800000023
and j =1,2,3 \8230n, which represents the ratio of the damaged area of the target opposite to the jth sampling point under the combined action of the k-missile and the target opposite to the jth sampling point.
4. The damage performance assessment method of claim 3, wherein for point targets, the damage probability of a single missile to a target is calculated by:
Figure FDA0003880521800000024
wherein, (x, y) represents coordinates of a point on the ground;
f (x, y) represents the spread law of the target position;
l (x, y) represents the dispersion law of the missile aiming points;
g (x, y) is a damage matrix converted by a damage amplitude transformer, and represents whether a certain position of a target opposite to the missile is damaged or not, wherein the damage is 1, and otherwise, the damage is 0;
for a surface target, the surface target is cut into a plurality of rectangular surface elements according to a circumscribed rectangle, the rectangular surface elements in the surface target are recorded as effective surface elements, and the damage probability of a single missile to the target is as follows:
Figure FDA0003880521800000031
wherein N is the number of missile landing points and target positions sampled according to the scattering law; f j (S) the proportion of the damaged area of the opposite target of the jth sampling point under the action of the single missile on the opposite target; m represents the number of active bins, (x) i ,y i ) The coordinates of the center point of the ith effective surface element are shown, i =1,2,3 \8230M.
5. The method of claim 4, wherein the effective surface elements are combined into a large surface element, and the large surface element is used as a new effective surface element to calculate the damage probability of the single missile on the target.
6. A fire planning method, comprising the steps of:
the firepower planning problem under the expected damage efficiency is summarized into the condition of solving the given bomb consumption, and the optimal aiming point coordinate is searched, so that the total damage probability is maximum;
solving the fire planning problem, comprising the following substeps:
step 11, setting initial shot consumption;
step 12, converting the target information and the distribution information of the corresponding missile aiming points into images, calling the trained damage performance evaluation agent model in the damage performance evaluation method according to any one of claims 1 to 5, and calculating the damage probability under the distribution information of different missile aiming points to obtain the maximum damage probability under the current missile consumption;
and step 13, judging whether the maximum damage probability under the current bomb consumption meets the given expected damage efficiency, if so, the aiming point coordinate corresponding to the maximum damage probability is the optimal aiming point coordinate, if not, the bomb consumption is increased by 1, the step 12 is returned to be executed until the given expected damage efficiency is met, the current bomb consumption, the optimal aiming point coordinate and the current damage probability are output, and the firepower planning is completed.
7. A fire planning method according to claim 6, wherein the fire planning problem given the expected damage performance is:
Figure FDA0003880521800000041
wherein (x) i ,y i ) Is the target position of the ith missile, x l ,x u ,y l And y u The range of the aiming point is set according to the actual target position, the spotter, the reconnaissance error and the damage spotter of the warhead.
8. A fire planning method according to claim 6 or 7, wherein a genetic algorithm is used to solve the fire planning problem for a given expected damage performance.
CN202110832886.6A 2021-07-22 2021-07-22 Damage efficiency evaluation and fire planning method based on convolutional neural network Active CN113627749B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110832886.6A CN113627749B (en) 2021-07-22 2021-07-22 Damage efficiency evaluation and fire planning method based on convolutional neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110832886.6A CN113627749B (en) 2021-07-22 2021-07-22 Damage efficiency evaluation and fire planning method based on convolutional neural network

Publications (2)

Publication Number Publication Date
CN113627749A CN113627749A (en) 2021-11-09
CN113627749B true CN113627749B (en) 2022-11-25

Family

ID=78380601

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110832886.6A Active CN113627749B (en) 2021-07-22 2021-07-22 Damage efficiency evaluation and fire planning method based on convolutional neural network

Country Status (1)

Country Link
CN (1) CN113627749B (en)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114358413B (en) * 2021-12-31 2023-02-10 北京理工大学 Fire distribution method considering target grouping and attack resolution
CN114048637B (en) * 2022-01-11 2022-04-08 中国人民解放军96901部队 Meta-normal-form-based rapid target damage calculation method
CN115345035B (en) * 2022-10-18 2023-06-27 中国人民解放军96901部队 Optimization method for aiming point of body target multidirectional striking space
CN116486272B (en) * 2023-06-14 2023-09-05 南京理工大学 Multi-dimensional index damage evaluation method and system based on complex scene
CN117575426A (en) * 2024-01-17 2024-02-20 北京理工大学 Rapid damage probability assessment method, device and medium

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104143028B (en) * 2014-08-12 2017-02-22 中国人民解放军装甲兵工程学院 Armored equipment damage rule analysis method based on simulation experiment
CN106407596B (en) * 2016-10-11 2019-11-22 中国人民解放军军械工程学院 Process model building emulation mode is injured in air defence missile hit
CN108694281B (en) * 2018-05-14 2022-09-16 西安工业大学 Target damage probability calculation method based on multi-dimensional angle fragment scattering penetration
CN112784437B (en) * 2021-02-04 2022-11-29 中国人民解放军63863部队 System for evaluating damage efficiency of air-defense missile to helicopter target

Also Published As

Publication number Publication date
CN113627749A (en) 2021-11-09

Similar Documents

Publication Publication Date Title
CN113627749B (en) Damage efficiency evaluation and fire planning method based on convolutional neural network
Przemieniecki Mathematical methods in defense analyses
Jaiswal Military operations research: quantitative decision making
CN109409695B (en) System efficiency evaluation index system construction method and system based on correlation analysis
US8046203B2 (en) Method and apparatus for analysis of errors, accuracy, and precision of guns and direct and indirect fire control mechanisms
CN109614729A (en) A kind of equipment Efficacy assessment result rapid analysis method based on meta-model
CN114912741A (en) Effectiveness evaluation method and device for combat system structure and storage medium
CN113177583B (en) Aerial target clustering method
Radovanović et al. Analysis of accuracy and precision of shooting with home: Made automatic rifles using the AHP method
CN113063419A (en) Unmanned aerial vehicle path planning method and system
CN116050515B (en) XGBoost-based parallel deduction multi-branch situation prediction method
CN115310345B (en) Intelligent algorithm evaluation method and device based on artificial potential field
KR101914234B1 (en) Method and apparatus for evaluating threat of surface targets
CN112818522B (en) Ammunition consumption measuring and calculating method and device
CN115761522A (en) Damage analysis method and damage evaluation system based on space-based information
CN110930054A (en) Data-driven battle system key parameter rapid optimization method
CN114038513A (en) Method, device and terminal for predicting mass concentration of hydrogen sulfide in coal-fired boiler
CN113837644B (en) Equipment combat effectiveness and contribution rate integrated evaluation method based on grey correlation
CN115374392B (en) Method for acquiring object parameters of target event, electronic equipment and storage medium
Zhang Research on Weighted Model of Air Defense Weapon-Target Assignment Based on AHP and Entropy Method
Chen et al. A comprehensive assessment method based on improved principal component analysis
RU136204U1 (en) DEVICE FOR SELECTION OF HIGH-SPEED MANEUVERING OBJECTIVES TO TEAM POINTS OF AIR DEFENSE GROUPS
Zhou et al. Research on Prediction Method of First Hit Ratio of Artillery Based on Data Driven
Singer et al. Probabilistic Risk Assessment for Acceptance Testing
Li et al. Modeling method of combat mission based on OODA loop

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