Multi-algorithm-driven behavior simulation model construction method and device

Abstract

The invention discloses a multi-algorithm-driven behavior simulation model construction method. The method comprises the following steps: compiling a military conceptual model; developing a meta-action model; developing a state transition model; developing an intelligent algorithm model; and integrated debugging of the interactive interface model. The invention provides a model algorithm process (including model state transition conditions, environment interaction conditions, algorithm operation conditions and the like) driven by various algorithms based on a military conceptual model, in combination with a typical combat scene and by applying a state transition method, a visual meta-action sequence combination is constructed according to the process, a visual action sequence file is generated, and the visual action sequence file is stored in a database. The action sequence file and a man-machine collaborative algorithm model are subjected to interface integration, various combat simulation platforms can be accessed through serial port communication, a shared database or a shared memory, and an innovative normal form of intelligent combat behavior design visualization, construction visualization and operation visualization is formed.

Description

Technical Field

[0001] The present invention relates to a model construction method, and in particular to a multi-algorithm driven behavior simulation model construction method and a device thereof. Background Art

[0002] The key to improving the immersion and confrontation of urban combat simulation is to build a highly realistic and robust intelligent agent behavior model. Currently, there are three widely used behavior modeling methods in the field of combat simulation: rule-based behavior modeling, finite state machine, and decision tree. The rule-based behavior modeling method has the advantages of simple logic and easy implementation, which also makes it only suitable for a single conditional conversion; the finite state machine realizes the behavior state transition of the intelligent agent based on the Markov random process. Compared with the rule-based modeling method, it can simulate more complex combat operations, but the intelligent agent behavior logic constructed by this method is single and easy to fall into a state dead loop; the decision tree behavior modeling method has the advantages of both rule-based and finite state machine modeling methods, and can drive the transfer of multiple states at the same state node through rules, but it still cannot get rid of its inherent disadvantages of being restricted by prior knowledge, not having environmental universality, and being unable to perform algorithm-level behavior simulation.

[0003] The existing rule-based behavior modeling method uses simple judgment logic, which requires a lot of manual design for specific behaviors and is not portable and reusable. The existing finite state machine-based behavior modeling method is an evolved version of the rule-based modeling method. It integrates and abstracts the rules and can improve the sophistication and robustness of the intelligent agent's behavior to a certain extent, but it still does not have universality and cannot simulate refined behavior. The behavior tree is a behavior modeling method that uses a tree structure for modeling. The number of parallel nodes is not limited, which can greatly improve the model's applicability to scenarios, but it still cannot simulate the behavior and decision-making under the equipment intelligent algorithm drive mode. Summary of the invention

[0004] The multi-algorithm driven behavior simulation model construction method proposed in the present invention mainly solves the problem that the algorithm in the current behavior model is difficult to combine with the rule-based state transition. The technical solution of the present invention aims to design a behavior modeling method that combines the state transition method after the military concept model is disassembled with the intelligent algorithm, which can realize the simulation and interaction of the algorithm-level behavior of the intelligent agent, and solve the scene adaptation and granularity problems of the model through the algorithm. The technical solution is as follows:

[0005] A method for constructing a multi-algorithm driven behavioral simulation model, characterized by:

[0006] Step 1: Military concept model writing. The military concept model is the first abstraction of real-world military activities. It is a structured and standardized description of various military entities, actions, and expected goals. It is the basis for the establishment of mathematical logic models and the reference for verification, validation, and confirmation (VV&A) of model machine simulation.

[0007] Step 2: Development of a meta-action model; the meta-action model is based on the basic actions of equipment and personnel decomposed from the combat actions involved in the military concept model;

[0008] Step 3: Development of a state transition model; the state transition model is a Markov probability transition model of the entity action process decomposed according to the military concept model for each entity involved in the combat action, including participating entity elements, action elements, and probability transition conditions;

[0009] Step 4: Develop intelligent algorithm models. The intelligent algorithm model refers to the artificial intelligence algorithm that the agent runs during the interaction with the entity and the environment during the operation of the simulation action model, including path planning, threat judgment and firepower allocation algorithms.

[0010] Step 5: Integration and debugging of interactive interface model; the interactive interface model is an interface communication program file designed for the operation process of the meta-action model, state transition model and intelligent algorithm model. Through the interactive interface model, the organic connection of various models can be realized, and the entity action sequence can be synchronized to the simulation engine to realize combat action simulation.

[0011] The present invention also discloses a multi-algorithm driven behavior simulation model construction device, which is characterized by comprising the following modules:

[0012] Military concept model writing module. The military concept model is the first abstraction of real-world military activities. It is a structured and standardized description of various military entities, actions, and expected goals. It is the basis for establishing mathematical logic models and a reference for verification, validation, and confirmation (VV&A) of model machine simulations.

[0013] Meta-action model development module; the meta-action model is the basic action of equipment and personnel decomposed from the combat actions involved in the military concept model;

[0014] State transition model development module; the state transition model is a Markov probability transition model of the entity action process decomposed according to the military concept model for each entity participating in the combat action, including participating entity elements, action elements, and probability transition conditions;

[0015] Intelligent algorithm model development module; the intelligent algorithm model refers to the artificial intelligence algorithm that the intelligent agent runs in the process of interacting with the entity and the environment during the operation of the simulation action model, including path planning, threat judgment and firepower allocation algorithms;

[0016] Interactive interface model integrated debugging module; the interactive interface model is an interface communication program file designed for the operation process of the meta-action model, state transition model and intelligent algorithm model. Through the interactive interface model, the organic connection of various models can be realized, and the entity action sequence can be synchronized to the simulation engine to realize combat action deduction.

[0017] The present invention also discloses a non-volatile storage medium, characterized in that the non-volatile storage medium includes a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute the above method.

[0018] The present invention also discloses an electronic device, characterized in that it comprises a processor and a memory; the memory stores computer-readable instructions, and the processor is used to execute the computer-readable instructions, wherein the computer-readable instructions execute the above method when they are executed.

[0019] Beneficial Effects

[0020] The present invention can simulate and interact with the algorithm-level behavior of intelligent agents, and solve the scene adaptation and granularity problems of the model through the A* algorithm, threat judgment algorithm and firepower allocation algorithm;

[0021] The present invention designs a behavior modeling method that combines the state transfer method after the military concept model is disassembled with the intelligent algorithm, which solves the problem of controlling the behavior and state of entities in fine-grained scenarios;

[0022] The present invention combines typical combat scenarios, uses state transfer methods, constructs a model algorithm process driven by multiple algorithms, constructs a visualized meta-action sequence combination according to the process, generates a visualized action sequence file, and integrates the action sequence file with the human-machine collaborative algorithm model to solve the problem of using algorithms to drive actions.

[0023] The present invention accesses various combat simulation platforms through serial port communication, shared database or shared memory, forming an innovative paradigm of intelligent combat behavior design visualization, construction visualization and operation visualization. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 Modeling schematic diagram for intelligent combat behavior model;

[0025] Combined with typical combat scenarios, the military concept model is abstracted into model entities, mission objectives, environmental interaction factors, etc. The state transition model, meta-action model construction and intelligent algorithm model development are used as intermediate steps in the construction of the simulation entity. The constructed meta-action sequence combination and intelligent algorithm data are transmitted to the combat simulation platform through an integrated interactive interface.

[0026] Figure 2 It is a composition diagram of the intelligent combat behavior simulation model;

[0027] The intelligent combat behavior simulation model proposed in the present invention consists of a military concept model, a meta-action model, a state transfer model, an intelligent algorithm model and an interactive interface model.

[0028] Figure 3 Modeling flow chart for intelligent combat behavior model.

[0029] The modeling process of the intelligent combat behavior model provided by the present invention is sequentially the writing of military concept model, the development of meta-action model, the development of state transfer model, the development of intelligent algorithm model and the integrated debugging of interactive interface. DETAILED DESCRIPTION

[0030] Example 1

[0031] A method for constructing a multi-algorithm driven behavior simulation model comprises the following steps:

[0032] Step 1: Writing a military concept model. The military concept model is the first abstraction of real military activities. It is a structured and standardized description of various military entities, actions and expected goals. It is the basis for establishing a mathematical logic model and a reference for verification, validation and confirmation (VV&A) of model machine simulation.

[0033] This step specifically includes the following:

[0034] (1) Conceptual model data preparation

[0035] A conceptual action flow chart should be formed according to the entities, environment, combat operations and other elements involved in the military concept model to provide top-level input support for model construction.

[0036] (2) Equipment model preparation

[0037] Prepare the physical models of the Red Team and Blue Team equipment involved according to the conceptual model, and conduct operational tests. After confirmation, the preparation is complete.

[0038] (3) Ammunition model preparation

[0039] Prepare the Red and Blue side ammunition models involved according to the conceptual model, and conduct operational tests. After confirmation, the preparation is complete.

[0040] Step 2: Development of a meta-action model; the meta-action model is based on the basic actions of equipment and personnel decomposed from the combat actions involved in the military concept model;

[0041] Meta-action refers to the smallest indecomposable action performed by equipment or personnel during combat simulation, such as squatting, running, jumping, crawling, raising a gun, aiming, etc.; equipment maneuvers, weapon posture changes, ammunition firing, etc.

[0042] Meta-motion development is the use of 3D animation development tools, in accordance with the simulation engine format requirements, based on the actual motion process animation file of the action to be constructed. After development is completed, its compatibility with the engine can be tested to see if it meets the requirements.

[0043] The meta-action model development process is as follows:

[0044] ①Create a new HTN file in the 3D animation development tool;

[0045] ②Create a new task (action) in the file;

[0046] ③Add conditions to the task (action);

[0047] ④Repeat steps ②-③;

[0048] ⑤Complete the development of meta-action model.

[0049] Step 3: Development of a state transition model; the state transition model is a Markov probability transition model of the entity action process decomposed according to the military concept model for each entity involved in the combat action, including participating entity elements, action elements, and probability transition conditions;

[0050] The state transition model is a Markov probability transition model constructed according to the action process of the combat behavior model, based on the equipment entity model and the meta-action model.

[0051] The Markov probability transition model abstracts the foreseeable state of the entity during the combat process and defines its state transfer matrix according to the rules of engagement. The model mainly includes state transfer objects, state space and state transfer matrix. The development process of the model is to develop and design the state space and state transfer matrix.

[0052] This step uses the Markov probability transition model to realize the interaction between simulation entities and the environment based on intelligent algorithms in different simulation systems. The method is as follows:

[0053] Given a simulation entity E that performs a given task T, according to the action plan and elements specified by the military concept model, the set of meta-actions A contained in task T can be expressed as:

[0054] A= <action1,action2,…,action >

[0055] action1, action2, …, action n Indicates the 1st to the nth action.

[0056] The state space set S of the entity can be expressed as

[0057] S= <state1,state2,…,state n >

[0058] state1 represents the first state, state n Indicates the nth state, the entity from state n-1 To state n The changing random process can be called state transfer, and the simulation step length is the state transfer step length, that is,

[0059] t transfer-delta =t simulation-delta

[0060] t transfer-delta represents the state transition step, t simulation-delta Represents the simulation step size.

[0061] The probability of a one-step transition from state i to j can be expressed as

[0062] p ij =P(X o+1 =j|X o =j) i,j∈S

[0063] The state transfer matrix P can be expressed as

[0064]

[0065] p 11 represents the one-step transition probability from state 1 to state 1, p 1n represents the probability of one-step transition from state 1 to state n, and so on, p nn represents the one-step transition probability from state n to state 1.

[0066] Step 4: Develop intelligent algorithm models. The intelligent algorithm model refers to the artificial intelligence algorithm that the agent runs during the interaction with the entity and the environment during the operation of the simulation action model, including path planning, threat judgment and firepower allocation algorithms.

[0067] The intelligent algorithm model is the computing model that is called when three types of actions, namely path planning, threat assessment, and firepower allocation, need to be performed during the state transfer process.

[0068] 1) When the entity changes state from <stationary to mobile>, the A* path planning algorithm is executed:

[0069] ① Get the current coordinates P of the entity through the interactive interface initial And the target point coordinates P dest , where P initial =(x initial ,y initial ), P dest =(x dest ,y dest ), x initial Indicates the current coordinate P initial The horizontal axis, y initial Indicates the current coordinate P initial The vertical coordinate, x dest Represents the target point coordinates P dest The horizontal axis, y dest Represents the target point coordinates P dest The vertical coordinate of

[0070] ② Establish position distance arrays C1 and C2 to store information about the nodes that the path passes through;

[0071] ③Calculate D i =f i +g i ,f i Indicates the starting point P initial The distance to the route to node i that has successfully avoided obstacles; g i Represents node i to the end point P dest The straight-line distance here refers to the straight-line distance from node i to the end point T when obstacles are not considered, that is, the remaining minimum distance. This is a minimum estimated value, D i Yes i With g i The sum of represents the current minimum total cost of node i.

[0072] ④ Set the starting point P initial Store it in array C1, traverse array C1, and find the minimum value D min The corresponding node is expanded, that is, the obstacle closest to the node is found, the node information of the obstacle is obtained, the available nodes of the obstacle are added to C1 and the expanded points in C1 are moved to C2.

[0073] ⑤ Repeat Step 2 until the line connecting the current expansion point and the end point no longer passes through any obstacles and can directly reach the end point, or stop the loop until C1 is empty;

[0074] ⑥When the entity reaches the specified location, the path planning algorithm execution ends.

[0075] 2) When the entity changes state from <Stationary-Encountering Enemy> to <Moving-Encountering Enemy>, the threat level determination algorithm is executed:

[0076] In the combat simulation process, the threat degree calculation model based on filtering and comparison is adopted:

[0077] ① Construct the membership matrix U(x1) of the threat index x1 of a certain entity target. The matrix is ​​as follows:

[0078]

[0079] Among them, m is the number of subordinate indicators of the middle indicator x1 in the threat index system, p is the number of threat level categories, r 11 The value indicating that the first indicator belongs to the level 1 threat, r 1k Indicates that the first indicator belongs to the value of level k threat, r m1 The value that indicates that the mth indicator belongs to the level 1 threat, r mp Indicates that the mth indicator belongs to the value of p-level threat. Similarly, r jk Indicates the value of the i-th subordinate indicator in indicator x1 that belongs to the k-level threat.

[0080] ② Calculate the discrimination weight α of each row index in the matrix U(x1) j , and obtain the discrimination weight vector α(x1). The calculation method is as follows:

[0081] A real number α that represents the contribution of the indicator to the target threat classification j Its value is determined by the membership degree r jk For k, the value is determined by the degree of concentration and dispersion; while r jk The degree of concentration and dispersion of k can be determined by the entropy H of the membership degree. j Quantitative description, so the real number α j It can be expressed as entropy H j The function of

[0082]

[0083] If the target threat index j belongs to the kth category, the membership degree is r jk ,k=1,…,p,j=1,…,m, and r jk satisfy Then the real number α determined by the above three formulas is j is the discrimination weight of indicator j regarding target threat. j Should meet

[0084]

[0085] Right of discrimination jThe meaning of is “distinguishing”, that is, it is a measure of whether the indicator membership can distinguish the category to which the target threat belongs and to what extent. j = 0, according to the properties of entropy, at this time there must be r j1 =r j2 =…=r jP In this case, the membership of index j is a redundant index membership that does not work for target threat classification; and the redundant index membership naturally cannot participate in the calculation of the membership of the entity target threat index.

[0086] ③ Assume that the weight vector of the existing threat indicators ω(x1)=(ω1,ω2,…,ω m ),ω1,ω2,…,ω m For each weight of the subordinate indicators of indicator x1, calculate the comparable values ​​of m indicators with respect to p levels of membership, and obtain the comparable value matrix N(x1);

[0087] The comparable value of the k-category membership of indicator j is:

[0088] n jk =ω j α j ·r jk , k=1,…,p

[0089] The k types of comparable values ​​of different indicators of target threat are comparable and directly additive; the comparable value matrix N(x1) obtained is

[0090]

[0091] ④ Calculate various comparable sum vectors M(x1) of the criterion layer index x1;

[0092] If jk is the k-type comparable value of the target's j indicator, then the k-type comparable sum of the target's threat is:

[0093]

[0094] Obviously, the greater the k-th category comparable and relative of the target threat, the greater the degree of belonging to the k-th category.

[0095] ⑤ Calculate the membership vector μ(x1) of the criterion layer indicator x1.

[0096] If M k , k=1,…,p, is the k-class comparable sum of the target threat, then the membership degree μ of the target threat belonging to the k-th class k for

[0097]

[0098] Among them, Mb represents the k-type comparable sum of the b-th target threat, M k represents the k-type comparable sum of the k-th target threat.

[0099] Obviously, the membership degree μ k satisfy

[0100]

[0101] The membership vector μ(x1) of the target threat index x1 is obtained as

[0102] μ(x1)=(μ1,μ 2, …,μ p )

[0103] ⑥ Similarly, the membership vectors of other indicators at the same level can be obtained, μ(x2), μ(x3)... etc. The membership vectors of all criterion layer indicators are used to form a new matrix for calculating the target threat degree. Repeat steps ①-⑤ to obtain the final target threat membership vector μ

[0104] μ=(μ′1,μ′2,…,μ′ p )

[0105] ⑦Threat level identification

[0106] Use the threat confidence identification criteria to determine the threat level of the target threat and give a threat confidence of no less than 80%.

[0107] μ′1+μ′2+…+μ′ k >0.8

[0108] Then we conclude that the threat level of the target is k.

[0109] 3) When the entity changes state from <Stationary-Strike> or <Mobility-Strike> or <Encounter-Strike>, the firepower allocation algorithm is executed:

[0110] (1) Call the fire attack unit entity list V and the fire attack target list T in the current simulation system task:

[0111] V=[v1,v2,v3,…,v n ]

[0112] T=[t1,t2,t3,…,t k ]

[0113] v i is the i-th fire attack unit, t i is the i-th fire strike unit, n represents the number of fire strike unit entities, and k represents the number of targets struck by fire.

[0114] (2) Record x ij is the plan of each fire attack unit for the target to be attacked, and its specific value is

[0115]

[0116] Assume that in a one-shot shooting, each firepower unit only hits one target, that is, In a fire strike, the number of fire units attacking each target shall not exceed n. j ,n j It indicates the maximum number of targets that can be hit by the jth target, and its value can be set by yourself, that is,

[0117] (3) Model for tactical mission importance index: The tactical mission importance of the jth target (j = 1, 2, ..., k) to be attacked by the i-th firepower unit (i = 1, 2, ..., n) is denoted as e ij The optimization goal proposed by the present invention is to maximize the tactical mission importance of the firepower attack target in a one-time firepower attack. The optimization model is as follows:

[0118]

[0119] (4) Damage model: The damage level of the jth blue target (j = 1, 2, ..., k) to be attacked by the i-th firepower unit (i = 1, 2, ..., n) is c ij , the goal is to maximize the importance of the tactical task of firepower attacking the target in a one-time firepower attack, so the objective function is taken as the minimum importance value, and the constraints remain unchanged, that is,

[0120]

[0121] (5) Comprehensive optimization model: Since it is necessary to optimize and evaluate the allocation plan of firepower strike, different types of quantitative indicators are standardized into a unified efficiency indicator: For indicator e ij and c ij The maximum value of For indicator e ij and c ij The minimum value of the normalized ij g ij

[0122]

[0123] After normalization, c ij is g′ ij :

[0124]

[0125] (6) The weight of each indicator can be obtained by the critic weight calculation method. Suppose the weights of z indicators are ω1, ω2, …ω z , and ω1,ω2,…ω z >0, Therefore, the firepower distribution model is:

[0126]

[0127] (7) After the calculation is completed, the weight coefficient of each strike equipment is transmitted to the simulation system, and the system executes the fire strike action according to the weight coefficient list.

[0128] The intelligent algorithm model parameterizes the A* path planning algorithm, XX threat judgment algorithm, and XX firepower allocation algorithm. Modelers can set parameters based on the combat scenario and actual equipment conditions, and can also connect to other artificial intelligence algorithms to achieve visual model development.

[0129] Step 5: Integration and debugging of interactive interface model; the interactive interface model is an interface communication program file designed for the operation of the meta-action model, state transition model and intelligent algorithm model. Through the interactive interface model, the organic connection of various models can be realized, and the entity action sequence can be synchronized to the simulation engine to realize combat action simulation.

[0130] Interactive interface integration debugging uses socket communication tools and TCP / IP protocol to test and integrate the interactive data items of the above-mentioned models, and check the effects in the simulation system.

[0131] The present invention proposes a model algorithm process driven by multiple algorithms (including model state transfer conditions, environmental interaction conditions, algorithm operation conditions, etc.) based on a military concept model and combined with typical combat scenarios by using a state transfer method. A visualized meta-action sequence combination is constructed according to the process, and a visualized action sequence file is generated. The action sequence file is integrated with the human-computer collaborative algorithm model through an interface. Various combat simulation platforms can be connected through serial port communication, shared database or shared memory, forming an innovative paradigm of visualization of intelligent combat behavior design, construction and operation.

[0132] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for constructing a multi-algorithm driven behavior simulation model, characterized by: Step 1: Compilation of military concept model. The military concept model is the first abstraction of real-world military activities. It is a structured and standardized description of various military entities, actions and expected goals. It is the basis for establishing mathematical logic models and the reference for verification, validation and confirmation of model machine simulation. Step 2: Development of a meta-action model; the meta-action model is based on the basic actions of equipment and personnel decomposed from the combat actions involved in the military concept model; Step 3: Development of a state transition model; the state transition model is a Markov probability transition model of the entity action process decomposed according to the military concept model for each entity involved in the combat action, including participating entity elements, action elements, and probability transition conditions; Step 4: Develop intelligent algorithm models. The intelligent algorithm model refers to the artificial intelligence algorithm that the agent runs during the interaction with the entity and the environment during the operation of the simulation action model, including path planning, threat judgment and firepower allocation algorithms. Step 5: Integration and debugging of interactive interface model; the interactive interface model is an interface communication program file designed for the operation of the meta-action model, state transition model and intelligent algorithm model. Through the interactive interface model, the organic connection of various models can be realized, and the entity action sequence can be synchronized to the simulation engine to realize combat action simulation.

2. The method for constructing a multi-algorithm driven behavior simulation model according to claim 1, wherein step 1 further comprises the following contents: (1) Conceptual model data preparation A conceptual action flow chart should be formed according to the entities, environment, combat actions and other elements involved in the military concept model to provide top-level input support for model construction; (2) Equipment model preparation Prepare the physical models of the Red Team and Blue Team equipment involved according to the conceptual model, and conduct operational tests. After confirmation, the preparation is complete; (3) Ammunition model preparation Prepare the Red and Blue side ammunition models involved according to the conceptual model, and conduct operational tests. After confirmation, the preparation is complete.

3. The method for constructing a multi-algorithm driven behavior simulation model according to claim 1, characterized in that: The step 2 further includes the following contents: meta-action refers to the smallest indecomposable action performed by equipment or personnel during the combat simulation process; the actions include squatting, running, jumping, crawling, raising guns, aiming, equipment maneuvers, weapon posture changes, and ammunition firing; the meta-action is developed using a three-dimensional animation development tool, in accordance with the simulation engine format requirements, based on the real motion process animation file of the action to be constructed, and after development is completed, its compatibility with the engine can be tested to see if it meets the requirements.

4. The method for constructing a multi-algorithm driven behavior simulation model according to claim 1, characterized in that: The step 3 further includes the following contents: the state transfer model is a Markov probability transfer model constructed according to the action process of the combat behavior model, based on the equipment entity model and the meta-action model; the Markov probability transfer model defines its state transfer matrix according to the engagement rules by abstracting the foreseeable state of the entity during the combat process. The model mainly includes a state transfer object, a state space and a state transfer matrix. The development process of the model is to develop and design the state space and the state transfer matrix.

5. The method for constructing a multi-algorithm driven behavior simulation model according to claim 1, characterized in that: The step 4 further includes the following contents: the intelligent algorithm model is a computational model to be called when three types of actions, namely path planning, threat judgment, and firepower allocation, need to be performed during the state transfer process; by parameterizing and encapsulating the A* path planning algorithm, XX threat judgment algorithm, or XX firepower allocation algorithm, modelers can set parameters according to the combat scenario and actual equipment conditions, and can also access other artificial intelligence algorithms to realize visual model development.

6. The method for constructing a multi-algorithm driven behavior simulation model according to claim 1, characterized in that: The step 5 further includes the following contents: the interactive interface integration debugging is to use the socket communication tool and the TCP / IP protocol to test and integrate the interactive data items of the above-mentioned various models, and check the effect in the simulation system.

7. A multi-algorithm driven behavior simulation model construction device, characterized by comprising: The following modules: Military concept model writing module, the military concept model is the first abstraction of real-world military activities, a structured and standardized description of various military entities, actions and expected goals, the basis for the establishment of mathematical logic models, and the reference for the verification, validation and confirmation of model machine simulation; Meta-action model development module; the meta-action model is the basic action of equipment and personnel decomposed from the combat actions involved in the military concept model; State transition model development module; the state transition model is a Markov probability transition model of the entity action process decomposed according to the military concept model for each entity participating in the combat action, including participating entity elements, action elements, and probability transition conditions; Intelligent algorithm model development module; the intelligent algorithm model refers to the artificial intelligence algorithm that the intelligent agent runs in the process of interacting with the entity and the environment during the operation of the simulation action model, including path planning, threat judgment and firepower allocation algorithms; Interactive interface model integration debugging module; The interactive interface model is an interface communication program file designed for the operation process of the meta-action model, the state transition model and the intelligent algorithm model. Through the interactive interface model, the organic connection of various models can be realized, and the entity action sequence can be synchronized to the simulation engine to realize the combat action deduction.

8. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored program, wherein when the program is executed, the device where the non-volatile storage medium is located is controlled to execute the method according to any one of claims 1 to 6.

9. An electronic device, characterized in that: It comprises a processor and a memory; the memory stores computer-readable instructions, and the processor is used to execute the computer-readable instructions, wherein the computer-readable instructions execute the method described in any one of claims 1 to 6 when executed.

Images