A damage calculation method and device based on a bayesian network

Abstract

This invention relates to the field of machine learning technology, and provides a damage calculation method and apparatus based on Bayesian networks. The method includes: determining the network structure by using multiple factors that cause damage to friendly assets from an opposing target as event nodes; determining the prior probability of each event node based on expert prior knowledge; establishing a Bayesian network model based on the prior probabilities of each event node and the network structure; and determining the damage probability of friendly assets based on the Bayesian network model. This invention constructs a Bayesian estimation network based on target capabilities and target intentions, achieving accurate estimation of friendly asset damage in highly dynamic environments and solving the problem of difficult friendly asset damage calculation.

Description

Technical Field

[0001] This invention relates to the field of machine learning technology, and in particular to a damage calculation method and apparatus based on Bayesian networks. Background Technology

[0002] Damage estimation is the process of predicting the potential damage a target may cause to our assets, and it is a necessary step before formulating follow-up plans. Its purpose is to estimate the loss of effectiveness based on the target's threat and intent, knowing in advance the target's remaining effectiveness after damage, thus providing a reference for subsequent plans. For example, in the field of drone applications, such as drone-based simulated combat and drone firefighting, pre-estimating the damage caused by a target can effectively formulate subsequent plans. In various combat games, damage estimation can also provide valuable reference for gamers, thereby enhancing the gaming experience.

[0003] However, in actual use, it is difficult to obtain complete information about the target, and damage estimation based solely on observed situational information is often insufficiently accurate. Furthermore, the damage caused by the target can encompass many complex aspects, including the main structure, the coupling relationships between propulsion equipment and electronic devices. These factors severely impact damage estimation and the smooth execution of subsequent plans.

[0004] Therefore, how to overcome the shortcomings of existing technologies and perform damage estimation quickly and accurately in highly dynamic environments is a problem to be solved in this technical field. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a damage calculation method and apparatus based on Bayesian networks.

[0006] The present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a damage calculation method based on Bayesian networks, comprising:

[0008] The network structure is determined by using multiple factors that cause damage to our assets due to the opponent's target as event nodes;

[0009] Based on the prior knowledge of experts, determine the prior probability of each event node;

[0010] Based on the prior probabilities of each event node and the network structure, a Bayesian network model is established to determine the probability of damage to our assets.

[0011] Preferably, the event node includes one or more of the following: target type, target damage range, target location, target speed, target distance, distance between target and nearest approach point, target estimated arrival time, target threat level, target intent, and damage probability.

[0012] Preferably, the network structure is as follows: the target type is used as a premise for determining the target damage range and target speed; the target location is used as a premise for calculating the target distance and the distance between the target and the nearest approach point; the target distance and target damage range are used as a premise for inferring the target threat level; the target distance and the nearest approach point and target speed are used as a premise for calculating the target's estimated arrival time; the target's estimated arrival time is used as a premise for judging the target's intention; and the target threat level and target intention are used as a premise for determining the damage probability.

[0013] Preferably, determining the prior probability of each event node based on the expert's prior knowledge specifically includes:

[0014] Determine the prior probability P(EffectRange|TargetType) of the damage range of the target under different target types;

[0015] Determine the prior probability P(Threat|EffectRange,Distance) of the threat level of the target corresponding to different target damage ranges and different target distances;

[0016] Determine the prior probability P(Intent|TCPA) of the target's intention corresponding to different expected arrival times of the target;

[0017] Determine the prior probability P(Damage|Threat,Intent) corresponding to the damage probability when the threat level of different adversaries is different from the adversary's intention.

[0018] Preferably, determining the probability of damage to our assets based on the Bayesian network model specifically includes:

[0019] Determine the location of the enemy target, the type of the enemy target, the speed of the enemy target, and the location of our assets;

[0020] Calculate the distance to the target based on the location of the target and the location of our assets;

[0021] Based on the location of the target, the type of the target, and the location of our assets, determine the location of the nearest approach point; based on the location of the nearest approach point, calculate the estimated arrival time of the target.

[0022] By substituting the target type, target speed, target distance, distance between the target and the nearest approach point, and each prior probability into the Bayesian network model, the probability of damage to our assets is calculated.

[0023] Preferably, the step of calculating the distance to the target based on the target's location and our asset's location specifically includes:

[0024] Calculate the distance to the target

[0025] Among them, (x zz y zz , z zz (x) represents the coordinates of our assets. target y target , z target ) represents the coordinates of the target's location.

[0026] Preferably, determining the location of the nearest approach point based on the target's location, target type, and the location of our assets specifically includes:

[0027] The position of the nearest approach point is calculated using a joint equation; the joint equation includes a first equation, a second equation, a third equation, a fourth equation, a fifth equation, and a sixth equation.

[0028] The first equation is (x cpa -x zz )·(x cpa -x target ) = 0;

[0029] The second equation is (y cpa -y zz )·(y cpa -y target ) = 0;

[0030] The third-party process is (z) cpa -z zz )·(z cpa -z target ) = 0;

[0031] The fourth equation is:

[0032] The fifth equation is:

[0033] The sixth equation is:

[0034] Among them, (x cpa y cpa , z cpa(x) represents the coordinates of the nearest approach point. zz y zz , z zz (x) represents the coordinates of our assets. target y target , z target () represents the coordinates of the target's location. This represents the vector representation of the target's velocity.

[0035] Preferably, calculating the estimated arrival time of the target based on the position of the nearest approach point specifically includes:

[0036] The distance between the target and the nearest approach point is calculated based on the position of the nearest approach point.

[0037] Calculate the estimated arrival time of the target

[0038] Among them, (x cpa y cpa , z cpa (x) represents the coordinates of the nearest approach point. target y target , z target () represents the coordinates of the target's location.

[0039] Secondly, the present invention also provides a damage calculation device based on Bayesian networks, used to implement the damage calculation method based on Bayesian networks described in the first aspect, the device comprising:

[0040] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor for performing the damage calculation method based on Bayesian networks as described in the first aspect.

[0041] Thirdly, the present invention also provides a non-volatile computer storage medium storing computer-executable instructions that are executed by one or more processors to perform the method described in the first aspect.

[0042] Fourthly, a chip is provided, comprising: a processor and an interface for calling and running a computer program stored in memory, performing the method as described in the first aspect.

[0043] Fifthly, a computer program product containing instructions is provided that, when executed on a computer or processor, causes the computer or processor to perform the method as described in the first aspect.

[0044] This invention, based on a Bayesian estimation network of target capability and target intent, achieves accurate estimation of our asset damage in highly dynamic environments, solving the problem of difficulty in calculating our asset damage. Attached Figure Description

[0045] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments of the present invention will be briefly described below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0046] Figure 1 This is a schematic flowchart of a damage calculation method based on Bayesian networks provided in an embodiment of the present invention;

[0047] Figure 2 This is a schematic diagram of the causal relationship of corresponding event nodes in a damage calculation method based on Bayesian networks provided in an embodiment of the present invention;

[0048] Figure 3 This is a schematic diagram of the causal relationship of corresponding event nodes in a damage calculation method based on Bayesian networks provided in an embodiment of the present invention;

[0049] Figure 4 This is a schematic diagram of the causal relationship of corresponding event nodes in a damage calculation method based on Bayesian networks provided in an embodiment of the present invention;

[0050] Figure 5 This is a schematic diagram of the causal relationship of corresponding event nodes in a damage calculation method based on Bayesian networks provided in an embodiment of the present invention;

[0051] Figure 6 This is a schematic diagram of the Bayesian network structure in a damage calculation method based on a Bayesian network provided in an embodiment of the present invention.

[0052] Figure 7 This is a schematic diagram of the prior probability in a damage calculation method based on Bayesian networks provided in an embodiment of the present invention;

[0053] Figure 8 This is a schematic diagram of the prior probability in a damage calculation method based on Bayesian networks provided in an embodiment of the present invention;

[0054] Figure 9 This is a schematic diagram of the prior probability in a damage calculation method based on Bayesian networks provided in an embodiment of the present invention;

[0055] Figure 10This is a schematic diagram of the prior probability in a damage calculation method based on Bayesian networks provided in an embodiment of the present invention;

[0056] Figure 11 This is a schematic diagram of the Bayesian network in a damage calculation method based on a Bayesian network provided in an embodiment of the present invention;

[0057] Figure 12 This is a flowchart illustrating a damage calculation method based on Bayesian networks provided in an embodiment of the present invention.

[0058] Figure 13 This is a schematic diagram of the nearest approach point in a damage calculation method based on Bayesian networks provided in an embodiment of the present invention;

[0059] Figure 14 This is a flowchart illustrating a damage calculation method based on Bayesian networks provided in an embodiment of the present invention.

[0060] Figure 15 This is a schematic diagram of the integrated Bayesian network structure in a damage calculation method based on Bayesian networks provided in an embodiment of the present invention.

[0061] Figure 16 This is a flowchart illustrating a damage calculation method based on Bayesian networks provided in an embodiment of the present invention.

[0062] Figure 17 This is a schematic diagram of the architecture of a damage calculation device based on a Bayesian network provided in an embodiment of the present invention. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0064] Unless the context otherwise requires, throughout the specification and claims, the term "comprising" is interpreted as openly inclusive, meaning "including, but not limited to." In the description of the specification, terms such as "one embodiment," "some embodiments," "exemplary embodiment," "example," "specific example," or "some examples" are intended to indicate that a particular feature, structure, material, or characteristic associated with that embodiment or example is included in at least one embodiment or example of this disclosure. The illustrative representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics mentioned may be included in any suitable manner in any one or more embodiments or examples; that is, although they may be incorporated into embodiments or examples using the above terms for reasons such as order and position, it does not limit them to be incorporated in combination by a single embodiment or example.

[0065] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of embodiments of this disclosure, unless otherwise stated, "a plurality of" means two or more. Furthermore, for example, the description may use the prefix "A" or "B" to describe the same type of nouns as two independent entities. In this case, the corresponding features defined with "A" and "B" are used only to distinguish between similar entities and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features.

[0066] In the description of this invention, the expression “A and / or B” (where A and B are used to formally represent specific features) will be used. The corresponding expression includes the following three combinations: only A, only B, and a combination of A and B.

[0067] As used in this invention, “about,” “approximately,” or “approximately” includes the stated value and the average value within an acceptable range of deviation from a particular value, wherein the acceptable range of deviation is determined by a person skilled in the art taking into account the measurement under discussion and the error associated with the measurement of the particular quantity (i.e., the limitations of the measurement system).

[0068] Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0069] Example 1:

[0070] Damage estimation has always been a research hotspot, and many methods have been proposed, including numerical simulation methods based on physical models and prediction methods based on machine learning.

[0071] Numerical simulation methods based on physical models require the establishment of complex physical models to describe the relationship between the target structure and the load response. This involves a large number of parameters and assumptions, and the computational complexity is high. Furthermore, the establishment and validation of physical models require a large amount of experimental data, which is often difficult to obtain. Therefore, methods based on physical models have certain limitations in practical applications.

[0072] Machine learning-based prediction methods typically require a large amount of training data to build the model. However, obtaining target damage data is often very difficult. Therefore, insufficient training data may lead to unstable model performance or poor generalization ability. Furthermore, machine learning methods often struggle to interpret the model's predictions, which can limit their application in areas such as target design and maintenance. To address these issues, Embodiment 1 of this invention provides a damage calculation method based on Bayesian networks, such as... Figure 1 As shown, it includes:

[0073] In step 101, the network structure is determined by using multiple factors that cause damage to our assets due to the opposing target as event nodes; the network structure can be understood as a network used to represent the causal relationships between various event nodes. In a practical application scenario, our assets can be objects with defensive capabilities, and the opposing target can be objects with adversarial capabilities or the ability to influence our assets. It should be noted that the terms "our assets" and "opposing target" are merely descriptions used to distinguish the two opposing parties, and do not imply that the executor in this embodiment is "our side" in actual use. In actual use, taking Party A as the defender and Party B as the adversary as an example, Party A represents our assets, and Party B represents the opposing target.

[0074] In step 102, the prior probability of each event node is determined based on the expert's prior knowledge.

[0075] In step 103, a Bayesian network model is established based on the prior probabilities of each event node and the network structure, so as to determine the probability of damage to our assets based on the Bayesian network model.

[0076] The probability of damage to our assets, determined based on the Bayesian network model, can be calculated using Netica analysis software.

[0077] This embodiment, based on a Bayesian estimation network of target capability and target intent, achieves accurate estimation of friendly asset damage in highly dynamic environments, solving the problem of difficult calculation of friendly asset damage. Furthermore, the Bayesian network method has significant advantages in handling uncertain information, enabling the evaluation of damage effects on complex targets. This method is characterized by its ability to handle large amounts of uncertain and fuzzy information, making it suitable for evaluating complex systems. In this embodiment, a Bayesian network based on target threat and target intent, along with a Netica-based simulation model, is constructed, enabling rapid and accurate estimation of target damage based on prior probabilities and input target information.

[0078] In a practical application scenario, the event node includes one or more of the following: target type, target damage range, target location, target speed, target distance, distance between target and nearest approach point, target estimated arrival time, target threat level, target intent, and damage probability.

[0079] Analysis revealed that there are relationships between the various event nodes, such as... Figure 2 , Figure 3 , Figure 4 and Figure 5 The causal relationship shown, such as Figure 2 As shown, TargetType represents the target type, and the target location L can be obtained based on the target type. target (exist Figure 2 (Location in the middle) and the speed of the target V target (exist Figure 2 The middle part is Speed) and the target damage range EffectRange, i.e., TargetType→{L target V target ,EffectRange}.

[0080] like Figure 3 As shown, with L zz Given the location of our assets, the distance to the enemy target (i.e., the distance between the enemy target and our assets) can be obtained from the location of our assets and the location of the enemy target. Distance = D(L) target ,L zz If the threat level of the target is (EffectRange, Distance), then the threat level of the target can be determined by the damage range of the target (EffectRange) and the distance of the target (Distance), that is, (EffectRange, Distance) → Threat.

[0081] like Figure 4As shown, CPA represents the nearest approach point, and the distance between the enemy target and the nearest approach point is represented as range from CPA. The distance between the enemy target and the nearest approach point can be obtained from the analysis of the enemy target's position and our asset's position. The estimated arrival time of the enemy target, TCPA, can be represented as (CPA, V...). target )→TBH, the estimated arrival time of the target can be understood as the estimated time of the target's arrival at the nearest approach point, and the target's intent can be represented as TCPA→Intent.

[0082] like Figure 5 As shown, the damage probability of our assets, Damage, can be represented as (Threat, Intent) → Damage.

[0083] according to Figure 2 , Figure 3 , Figure 4 and Figure 5 The relationships between the event nodes shown are determined as follows: Figure 6 The network structure shown is as follows: the target type is used as a premise for determining the target damage range and target speed; the target location is used as a premise for calculating the target distance and the distance between the target and the nearest approach point; the target distance and target damage range are used as a premise for inferring the target threat level; the target distance and target speed are used as a premise for calculating the target's estimated arrival time; the target's estimated arrival time is used as a premise for judging the target's intention; and the target threat level and target intention are used as a premise for determining the damage probability.

[0084] In one optional implementation, determining the prior probability of each event node based on the expert's prior knowledge specifically includes:

[0085] Determine the prior probability P(EffectRange|TargetType) of the damage range of the target under different target types; in a practical application scenario, the damage range of the target is divided into three types: short, medium, and long. The prior probability P(EffectRange|TargetType) of the damage range corresponding to each type of target is as follows: Figure 7 As shown, where, Figure 7 This example illustrates the existence of three target types: TargetType, TargetType, and TargetType.

[0086] Determine the prior probability P(Threat|EffectRange,Distance) of the threat level of the target corresponding to different target damage ranges at different target distances. In a practical application scenario, target distances are categorized into three types: close, medium, and far; target threat levels are categorized into three types: low, medium, and high; and target damage ranges are categorized into three types: short, medium, and long. The prior probability P(Threat|EffectRange,Distance) of the target threat level at each target distance for each type of target damage range is obtained. Figure 8 As shown.

[0087] Determine the prior probability P(Intent|TCPA) of the target's intention corresponding to different expected arrival times. In a practical application scenario, the target's intention is divided into three levels: low, medium, and high. The prior probability P(Intent|TCPA) of the target's intention corresponding to the expected arrival time within each of these levels is then obtained. Figure 9 As shown.

[0088] Determine the prior probability P(Damage|Threat,Intent) of the damage probability corresponding to different target threat levels under different target intentions. In a practical application scenario, the damage probability is divided into two categories: suffering damage (i.e., true) and not suffering damage (i.e., false). The prior probability P(Damage|Threat,Intent) of the damage probability corresponding to each target threat level under different target intentions is as follows: Figure 10 As shown.

[0089] Based on the aforementioned prior probabilities and network structure, a Bayesian network model is established as follows: Figure 11 As shown.

[0090] In one optional embodiment, the step of determining the probability of damage to our assets based on the Bayesian network model is as follows: Figure 12 As shown, it specifically includes:

[0091] In step 201, the location of the enemy target, the type of the enemy target, the speed of the enemy target, and the location of our assets are determined.

[0092] In step 202, the distance to the target is calculated based on the target's location and the location of our assets.

[0093] In step 203, the location of the nearest approach point is determined based on the location of the target, the type of the target, and the location of our assets; the estimated arrival time of the target is calculated based on the location of the nearest approach point.

[0094] In step 204, the target type, target speed, target distance, distance between the target and the nearest approach point, and each prior probability are substituted into the Bayesian network model to calculate the damage probability of our assets.

[0095] Steps 201-204 above can be understood as being in Figure 6 The network structure shown illustrates a sequential downward solution process: determining the target type, speed, location, and our asset location; calculating the target distance and estimated arrival time; and then, based on the values ​​of the aforementioned event nodes... Figure 7 , Figure 8 , Figure 9 and Figure 10 The prior probabilities shown determine the threat level and intent of the opposing target, and then further determine the probability of damage to our assets.

[0096] Specifically, calculating the distance to the target based on the target's location and our asset's location includes: calculating the distance to the target. Among them, (x zz y zz , z zz (x) represents the coordinates of our assets. target y target , z target ) represents the coordinates of the target's location.

[0097] In practical applications, the distance to the target is converted into three values: close, medium, and far. In a specific application scenario, P(Distance = close) is D(L). target ,L zz )≤2; P(Distance=medium) is 2 <D(L target ,L zz )≤10; P(Distance=far) is 10 <D(L target ,L zz )≤40.

[0098] The nearest approach point is the point where the enemy target is closest to our assets in the direction of travel, such as... Figure 13 As shown, determining the location of the nearest approach point based on the target's location, target type, and our asset location specifically includes:

[0099] The position of the nearest approach point is calculated using a joint equation; the joint equation includes a first equation, a second equation, a third equation, a fourth equation, a fifth equation, and a sixth equation.

[0100] The first equation is (x cpa -x zz )·(x cpa -x target ) = 0; the second equation is (y cpa -y zz )·(y cpa -y target ) = 0; the third-party process is (z cpa -z zz )·(z cpa -z target ) = 0.

[0101] The fourth equation is: The fifth equation is: The sixth equation is:

[0102] Among them, (x cpa y cpa , z cpa (x) represents the coordinates of the nearest approach point. zz y zz , z zz (x) represents the coordinates of our assets. target y target , z target () represents the coordinates of the target's location. This represents the vector representation of the target's velocity. Solving the joint equations yields the position (i.e., coordinates) of the nearest approach point.

[0103] The step of calculating the estimated arrival time of the target based on the position of the nearest approach point specifically includes:

[0104] The distance between the target and the nearest approach point is calculated based on the position of the nearest approach point. Calculate the estimated arrival time of the target Among them, (x cpa y cpa , z cpa (x) represents the coordinates of the nearest approach point. target y target , z target () represents the coordinates of the target's location.

[0105] Considering that in actual use, our side may also possess air assets used for combat, primarily for protecting our ground assets, the calculation of damage probability for our ground assets would be more complex. To address this, this embodiment, based on the above-described implementation method, also provides a preferred implementation method, such as... Figure 14 As shown, it specifically includes:

[0106] In step 301, a first Bayesian network model is established between our ground assets as the defender and the opponent's target as the adversary.

[0107] In step 302, a second Bayesian network model is established between our air assets as the adversary and the enemy target as the defender.

[0108] In step 303, using the final result event nodes of the first Bayesian network model and the final result event nodes of the second Bayesian network model as premises, and the damage probability of our ground assets as the result, a comprehensive Bayesian network model is established. The structure of this comprehensive Bayesian network model is as follows: Figure 15 As shown.

[0109] In step 304, the prior probabilities of each event node in the integrated Bayesian network model are determined based on the prior knowledge of the experts.

[0110] In step 305, a first probability of damage to friendly ground assets is determined based on the first Bayesian network model; a second probability of damage to the enemy target is determined based on the second Bayesian network model; and the final probability of damage to friendly ground assets is determined based on the prior probabilities of each event node in the comprehensive Bayesian network model, the first probability, and the second probability. Specifically, determining the prior probabilities of each event node in the comprehensive Bayesian network model means determining the probability of damage to friendly ground assets under the corresponding first and second probabilities.

[0111] In a practical application scenario, the probability of damage to our assets under the corresponding first and second probability conditions is specifically manifested as follows:

[0112] Damage=Damage_1×(1-k×Damage_2)

[0113] Where Damage is the probability of damage to our ground assets, Damage_1 is the first probability, Damage_2 is the second probability, and k is a preset parameter obtained by those skilled in the art based on experience, where k is greater than 0 and less than or equal to 1.

[0114] The establishment of the first Bayesian network model and the second Bayesian network model is based on the same concept as the above-described embodiments, and will not be repeated here.

[0115] Taking the first Bayesian network model as an example, its establishment specifically includes: determining the network structure by using multiple factors that cause damage to friendly ground assets from enemy targets as event nodes; determining the prior probability of each event node based on expert prior knowledge; and establishing the first Bayesian network model based on the prior probabilities of each event node and the network structure. Similarly, the second Bayesian network model is established based on the network structure determined by using multiple factors that cause damage to enemy targets from friendly air assets as event nodes.

[0116] Example 2:

[0117] Based on the method described in Embodiment 1, this invention combines specific application scenarios and uses technical descriptions in relevant scenarios to illustrate the implementation process of the features of this invention in those scenarios.

[0118] This embodiment uses our own assets as the surface target and the aerial target as the enemy target, and uses Netica analysis software to calculate the damage to the surface target as a practical application scenario. The embodiment of this invention provides a damage calculation method based on Bayesian networks, such as... Figure 16 As shown, it specifically includes:

[0119] In step 401, the event nodes in the Bayesian network are identified. Factors (i.e., event nodes) affecting the damage estimation of water targets are analyzed and determined.

[0120] In order to make an accurate estimate of the threat level of an adversary's target, it is necessary to carefully analyze the highly dynamic environment full of incomplete information and identify relevant factors.

[0121] In real-world scenarios, various factors influence damage estimation for water targets. Based on actual scenarios, the main factors affecting damage estimation for water targets include target type, target location, target speed, target damage range, target threat level, and target intent. According to the actual simulation environment, these factors are divided into composite factors such as target type, target threat level, and target intent, as well as atomic factors such as target location, target speed, and target damage range.

[0122] In step 402, the Bayesian network structure is determined. Hierarchical relationships between nodes are established according to the causal relationships between events.

[0123] In a real-world scenario, starting from atomic factors and reasoning layer by layer, a process can ultimately derive composite factors representing the degree of damage. Based on the actual simulation environment, information such as the target's location, speed, and damage range can be obtained from the target type. The target's threat level is determined by its damage range and distance from our assets; its intent is determined by its estimated arrival time; and its estimated arrival time is determined by its speed and nearest approach point.

[0124] Accordingly, establish as Figure 2 , Figure 3 , Figure 4 and Figure 5 The hierarchical structure between the nodes shown can be integrated to obtain the following: Figure 6 The Bayesian network structure shown is (i.e., the network structure in Example 1).

[0125] Let TargetType represent the target type. Based on the target type, the target location L can be obtained. target The speed of the target V target And the target's damage range, EffectRange, i.e., TargetType→{L target ,L zz ,EffectRange}.

[0126] With L zz Indicates the location of key assets, D(L) target ,L zz If (EffectRange, Distance) represents the distance between the target and the target on the water, then the threat level of the target can be expressed as (EffectRange, Distance) → Threat.

[0127] CPA represents the nearest approach point, and the nearest approach point is as follows: Figure 13 As shown, the distance from the target location to CPA is represented as range from CPA, and the estimated arrival time TBH of the target can be represented as (CPA, V). target The target intent (Intent) can be represented as TBH→Intent.

[0128] Damage to a target on water can be represented as (Threat, Intent) → Damage.

[0129] In step 403, the prior probabilities of relevant nodes are determined: some parameters in the Bayesian network, i.e., conditional probabilities, are determined based on expert experience.

[0130] In a practical implementation scenario, based on expert experience, the probability of determining the target's damage range (near short, medium, or far long) based on the target type can be as follows: Figure 7 As shown.

[0131] The damage range of the enemy target is determined by EffectRange and the distance of the enemy target from the water target, D(L). target ,L zz The probability of both sides obtaining the opponent's target threat level as low, medium, or high is as follows: Figure 8 As shown; the probability of low, medium, or high target intent is obtained from the target's estimated arrival time TBH, as follows: Figure 9 As shown; the probability of damage to a surface target is obtained by combining the threat level and the target's intent. Figure 10 As shown.

[0132] The value of the distance between the enemy target and the surface target is determined based on the specific numerical value of the distance between the enemy target and the surface target, and needs to be calculated; according to prior knowledge: when the distance between the enemy target and the surface target is less than or equal to 2 kilometers, i.e., D(L) target ,L zz When )≤2, the target is considered relatively close, and Distance is set to close. When the distance between the target and the surface target is 2-10 kilometers, i.e., 2 <D(L target ,L zz When 10 ≤ L, the distance to the target is considered medium. The value of Distance is medium. When the distance between the target and the surface target is 10-40 kilometers, i.e., 10 < L target ,L zz When )≤40, the target is considered to be far away, and Distance takes the value far.

[0133] The estimated arrival time of the target also needs to be calculated in real time.

[0134] In step 404, a simulation calculation model is built. A Bayesian network model for damage estimation of water targets is established using Netica analysis software.

[0135] In practice, Netica analysis software is used to establish a Bayesian network model for damage estimation of water targets, such as... Figure 11 As shown; input a determined partial conditional probability value; calculate the probability of damage to the water target based on the input target information.

[0136] The parameters in the Bayesian network determined based on expert experience include: Based on the simulation environment, the types of aerial targets are mainly Class I, Class II, and Class III targets, i.e., TargetType = {Class I, Class II, Class III}; the EffectRange value for the target damage range is short, medium, and long, i.e., EffectRange = {short, medium, Long}; the Threat value for the target threat level is high, medium, and low, i.e., Threat = {high, medium, low}; the Intent value for the target intent is high, medium, and low, i.e., Intent = {high, medium, long}; and the Damage value for surface targets is true and false, i.e., Damage = {true, false}.

[0137] The target type is determined by the input target information; the probability P(EffectRange|TargetType) of obtaining the target damage range is as follows: Figure 7 As shown.

[0138] The probability P(Threat|EffectRange,Distance) of determining the threat level of an enemy target based on its damage range and real-time distance is as follows: Figure 8 As shown.

[0139] The probability P(Intent|TBH) of obtaining the target's intention from the estimated arrival time is as follows: Figure 9 As shown;

[0140] The probability P(Damage|Threat,Intent) of damage to a surface target is derived from the threat level and capability of the target. Figure 10 As shown.

[0141] After establishing the Bayesian network model, the probability of damage to our assets is calculated, specifically including: obtaining the probability of the threat level of the enemy target from the specific numerical value of the damage range and the real-time distance; calculating the nearest approach point; calculating the distance from the enemy target to the CPA; calculating the estimated arrival time (TBH) of the enemy target; determining the enemy target's intention based on the estimated arrival time, and substituting the target's intention and the enemy target's threat level into the Bayesian network model to obtain the probability of damage to our assets.

[0142] The probability of obtaining the threat level of the opposing target from the specific numerical value of the target's damage range and the real-time distance specifically includes:

[0143] The probability P(EffectRange|TargetType) of obtaining the damage range of the target based on the target type is as follows: Figure 7 As shown.

[0144] The real-time distance D(L) between the target and the critical asset (i.e., the surface target). target ,L zz ) Calculated results:

[0145] In actual implementation, the distance D(L) between the target and the surface target is used as the basis for the decision. target ,L zz The value of Distance is determined as follows:

[0146] P(Distance=close) is D(L) target ,L zz )≤2.

[0147] P(Distance = medium) is 2 <D(L target ,L zz )≤10.

[0148] P(Distance = far) is 10 <D(L target ,L zz )≤40.

[0149] The probability of deriving the threat level of an enemy target from its specific damage range and real-time distance is shown below. Figure 8 .

[0150] The calculation of the nearest approach point (CPA) specifically includes:

[0151] An illustration of the nearest point to approach (CPA) is shown below. Figure 13 CPA is the point closest to point ZZ in the direction of the target's velocity. The solution is as follows:

[0152] Let the coordinates of CPA be (x cpa ,y cpa ,z cpa ).

[0153] First, the direction vector from the target's position to CPA is perpendicular to the direction vector from CPA to point ZZ, which gives the following equation: (x cpa -x zz )·(x cpa -x target ) = 0, (y cpa -y zz )·(y cpa -y target ) = 0 and (z cpa -zz2 )·(z cpa -z target ) = 0.

[0154] Furthermore, to ensure that the calculated CPA is in the velocity direction, the direction vector from the target position to the CPA must be in the same direction as the velocity direction of the incoming weapon, as shown by the following equation. and

[0155] Solving the above equation will yield the coordinates of CPA.

[0156] The calculation of the distance from the target to the CPA specifically includes:

[0157] Calculate the distance from the target to the CPA based on the CPA's coordinates:

[0158] The calculation of the estimated arrival time (TBH) of the target specifically includes:

[0159] Define the time from the target's location to the CPA as T. CPA .

[0160] Then, T CPA =range from CPA / |V target |

[0161] TBH = T CPA

[0162] The process of determining the target's intent based on the target's estimated arrival time specifically includes:

[0163] In actual implementation, the probability P(Intent|TBH) of obtaining the target's intention from the estimated arrival time is as follows: Figure 9 As shown.

[0164] This embodiment analyzes and determines the factors affecting ship damage estimation based on the actual simulation environment, establishes hierarchical relationships between nodes according to the causal relationships between the factors, integrates the hierarchical relationships to obtain a Bayesian network structure, obtains some parameters in the Bayesian estimation network based on expert prior knowledge, and builds a simulation calculation model based on Netica software. By inputting the target information in the simulation environment, ship damage can be predicted quickly and accurately.

[0165] This embodiment achieves accurate estimation of damage to maritime targets in highly dynamic environments based on a Bayesian estimation network that considers target capabilities and intentions. Furthermore, it addresses the challenge of calculating damage to maritime targets through scientific computation methods using a Netica simulation model.

[0166] Example 3:

[0167] like Figure 17 The diagram shown is an architectural schematic of a damage calculation device based on a Bayesian network according to an embodiment of the present invention. This Bayesian network-based damage calculation device includes one or more processors 21 and a memory 22. Figure 17 Take a processor 21 as an example.

[0168] Processor 21 and memory 22 can be connected via a bus or other means. Figure 17 Taking the example of a connection between China and Israel via a bus.

[0169] The memory 22, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs and non-volatile computer-executable programs, such as the Bayesian network-based damage calculation method in Embodiment 1. The processor 21 executes the Bayesian network-based damage calculation method by running the non-volatile software program and instructions stored in the memory 22.

[0170] Memory 22 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, memory 22 may optionally include memory remotely located relative to processor 21, which can be connected to processor 21 via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0171] The program instructions / modules are stored in the memory 22 and, when executed by one or more processors 21, execute the damage calculation method based on Bayesian networks in Embodiment 1.

[0172] It is worth noting that the information interaction and execution process between the modules and units in the above-mentioned device and system are based on the same concept as the processing method embodiment of the present invention. For details, please refer to the description in the method embodiment of the present invention, and will not be repeated here.

[0173] Those skilled in the art will understand that all or part of the steps in the various methods of the embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc.

[0174] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A damage calculation method based on a Bayesian network, characterized by, The method comprises the following steps: establishing a first Bayesian network model between the target and the ground asset of the first party by taking the ground asset of the first party as the defender and the target as the opponent; establishing a second Bayesian network model between the target and the air asset of the first party by taking the air asset of the first party as the opponent and the target as the defender; establishing a comprehensive Bayesian network model by taking the final result event nodes of the first Bayesian network model and the second Bayesian network model as the premise and the damage probability of the ground asset of the first party as the result, and determining the prior probabilities of the event nodes in the comprehensive Bayesian network model based on the prior knowledge of experts; determining a first probability of the damage of the ground asset of the first party according to the first Bayesian network model, determining a second probability of the damage of the target according to the second Bayesian network model, and determining a final damage probability of the ground asset of the first party according to the prior probabilities of the event nodes in the comprehensive Bayesian network model, the first probability and the second probability, wherein the determination of the prior probabilities of the event nodes in the comprehensive Bayesian network model is the determination of the damage probability of the ground asset of the first party under the corresponding first probability and second probability. The damage probability of the ground asset of the first party under the first probability and the second probability is ; wherein, is the damage probability of the ground asset of the first party, is the first probability, is the second probability, is a preset parameter; The first Bayesian network model is established by taking multiple factors causing the damage of the ground asset of the first party by the target as the event nodes, determining the network structure, determining the prior probabilities of the event nodes based on the prior knowledge of experts, establishing the first Bayesian network model according to the prior probabilities of the event nodes and the network structure, and determining the damage probability of the ground asset of the first party, i.e. the first probability, according to the first Bayesian network model.

2. The Bayesian network-based damage calculation method according to claim 1, characterized in that, The event nodes comprise one or more of the target type, the damage range of the target, the target position, the target speed, the target distance, the distance between the target and the nearest approach point, the target estimated arrival time, the target threat degree, the target intention and the damage probability.

3. The Bayesian network-based damage calculation method according to claim 2, characterized in that, The network structure is that the target type is taken as the premise for determining the damage range of the target and the target speed, the target position is taken as the premise for calculating the target distance and the distance between the target and the nearest approach point, the target distance and the damage range of the target are taken as the premise for inferring the target threat degree, the distance between the target and the nearest approach point and the target speed are taken as the premise for calculating the target estimated arrival time, the target estimated arrival time is taken as the premise for judging the target intention, and the target threat degree and the target intention are taken as the premise for determining the damage probability.

4. The Bayesian network-based damage calculation method according to claim 3, characterized in that, The determination of the prior probabilities of the event nodes based on the prior knowledge of experts comprises: Determining a prior probability of a corresponding opposite target damage range under different opposite target types ; determining a prior probability of a threat level of a different target corresponding to a different target damage range at a different target distance ; determining a prior probability of an opposite target intent corresponding to different opposite target expected arrival times ; determining a priori probabilities of damage probabilities corresponding to different opponent target threat levels at different opponent target intentions .

5. The Bayesian network-based damage calculation method according to claim 3, characterized in that, The determination of the damage probability of the ground asset of the first party according to the Bayesian network model comprises: determining the target position, the target type, the target speed and the ground asset position of the first party; calculating the target distance according to the target position and the ground asset position of the first party; According to the position of the opposite target, the type of the opposite target and the position of the ground asset of the self, a position of a nearest approach point is determined; according to the position of the nearest approach point, a time of arrival of the opposite target is calculated; The type of the opposite target, the speed of the opposite target, the distance of the opposite target, the distance between the opposite target and the nearest approach point and each prior probability are substituted into the Bayesian network model, so that a damage probability of the ground asset of the self is calculated.

6. The Bayesian network-based damage calculation method according to claim 5, characterized in that, The distance of the opposite target is calculated according to the position of the opposite target and the position of the ground asset of the self, and specifically includes: The target distance of the opposite party is calculated ; wherein, is the coordinate of the position of our ground asset, is the coordinate of the position of the target.

7. The Bayesian network-based damage calculation method according to claim 5, characterized in that, The position of the nearest approach point is determined according to the position of the opposite target, the type of the opposite target and the position of the ground asset of the self, and specifically includes: The position of the nearest approach point is calculated by using a joint equation; the joint equation includes a first equation, a second equation, a third equation, a fourth equation, a fifth equation and a sixth equation; The first equation is ; The second equation is ; The third equation is ; The fourth equation is ; The fifth equation is ; The sixth equation is ; wherein, is the coordinate of the most recent point of approach, is the coordinate of the position of our ground asset, is the coordinate of the position of the opposing target, is the vector representation of the opposing target's velocity.

8. The Bayesian network-based damage calculation method of claim 5, wherein, The time of arrival of the opposite target is calculated according to the position of the nearest approach point, and specifically includes: calculating a distance between the opponent target and the closest approach point according to the position of the closest approach point ; Computing an opponent target estimated time of arrival ; wherein is the coordinate of the most recent approach point, is the coordinate of the opponent target position, .

9. A non-transitory computer storage medium, comprising: The computer storage medium stores computer executable instructions, and the computer executable instructions are executed by one or more processors to complete the damage calculation method based on the Bayesian network in any one of claims 1-8.

10. A Bayesian network-based damage calculation apparatus characterized by comprising: It includes: At least one processor; And a memory connected with the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the processor to execute the damage calculation method based on the Bayesian network in any one of claims 1-8.

Images