A method, medium, and system for generating smoke in a smoke simulation training system.

Abstract

This invention provides a smoke screen generation method, medium, and system for a smoke screen simulation training system, belonging to the field of smoke screen simulation training technology. The method includes: dividing the actual training field environment into a first grid based on a scenario where a smoke screen launcher releases smoke in a windless environment; acquiring multiple smoke screen rules to form a smoke screen rule set; establishing a simulated training field and dividing the simulated training field into a second grid corresponding to the first grid; setting up a simulated smoke screen launcher in the simulated training field and setting the initial smoke screen launched by the simulated smoke screen launcher; using each smoke screen rule in the smoke screen rule set to perform preliminary evolution on the smoke screen that has reached a first stable state, obtaining multiple smoke screen states at different times; scoring the multiple smoke screen states at different times, and using the smoke screen rule corresponding to the smoke screen state with the highest score as the target rule; and further evolving the smoke screen state corresponding to the target rule to obtain the generated simulated smoke screen.

Description

Technical Field

[0001] This invention belongs to the field of smoke screen simulation training technology, and specifically relates to a smoke screen generation method, medium and system for a smoke screen simulation training system. Background Technology

[0002] In modern military, aviation, and aerospace fields, smoke screen simulation training systems have significant application value. By simulating actual smoke screen scenarios, they can help military pilots and other personnel better master smoke screen tactics, skills, and emergency response capabilities. Currently, training for specific airborne smoke screen tactics units generally utilizes smoke screen simulation training systems. A crucial aspect of these systems is the generation of simulated smoke screens launched by smoke screen launchers. Smoke screen simulation is typically achieved using computer graphics and physical simulation methods. In computer graphics, particle systems are a common approach. By generating a large number of particles in three-dimensional space and setting relevant attributes for each particle, such as position, velocity, and color, the appearance and behavior of smoke can be simulated. By updating and rendering the particles, realistic smoke screen effects can be created. In physical simulation, a commonly used method is based on fluid dynamics. Smoke is treated as a fluid, and its motion and changes can be simulated by solving fluid dynamics equations. This method involves complex mathematical calculations and a large number of physical parameters, but it can provide more realistic smoke effects.

[0003] A cellular automaton is a mathematical model used to simulate a system of interacting cells based on simple rules. It consists of a two- or three-dimensional discrete lattice space, with one cell at each lattice point, and these cells update and evolve their states according to specific rules. Typically, cellular automata evolve in discrete time steps, with each time step determining the current state of the cell based on the states of its neighbors.

[0004] Cellular automata can simulate many different phenomena and systems, such as those in physics, biology, and sociology. They are widely used in simulating the evolution and behavior of complex systems, as well as in research and development in fields such as artificial life, artificial intelligence, and computer graphics.

[0005] Currently, although there are studies on using cellular automata to simulate smoke screens, there is a lack of research on smoke screens launched by specific smoke screen launch vehicles. At the same time, the rules for the evolution of cellular automata of smoke screens are often preset, which leads to a large gap between the diffusion of simulated smoke screens and actual smoke screens as the simulation progresses, and thus cannot obtain good smoke screen simulation training. Summary of the Invention

[0006] In view of this, the present invention provides a smoke screen generation method, medium and system for a smoke screen simulation training system. It utilizes cellular automata to simulate smoke screens, solving the technical problem that the rules for the evolution of cellular automata of smoke screens are often preset, resulting in a large gap between the diffusion of simulated smoke screens and actual smoke screens as the simulation progresses, thus failing to obtain good smoke screen simulation training.

[0007] This invention is implemented as follows:

[0008] The first aspect of the present invention provides a method for generating smoke in a smoke simulation training system, comprising the following steps:

[0009] S10. Based on the scenario of a smoke screen launcher releasing smoke in the actual training ground environment when there is no wind, the actual training ground environment is divided into a first grid, and multiple smoke screen rules are obtained to form a smoke screen rule set.

[0010] S20. Establish a simulated training field based on the actual training field environment when there is no wind, and divide the simulated training field into a second grid corresponding to the first grid.

[0011] S30. Set up a simulated smoke screen launching vehicle in the simulated training field, and set the initial smoke screen launched by the simulated smoke screen launching vehicle. The initial smoke screen contains multiple smoke screen particles. The smoke screen particles are launched from the simulated launching vehicle according to the rules of fluid dynamics and reach a first stable state. The first stable state is the state when the average speed of the smoke screen particles is lower than the stable speed threshold.

[0012] S40. Using each smoke rule in the smoke rule set, the smoke that has reached the first stable state is initially evolved to obtain multiple smoke states at different times.

[0013] S50. Score multiple smoke screen states at different times, and take the smoke screen rule corresponding to the smoke screen state with the highest score as the target rule.

[0014] S60. The smoke screen state corresponding to the target rule is further evolved to generate a simulated smoke screen.

[0015] Wherein, the side length of each small cube in the first grid is 1mm; the stable velocity threshold is 1mm / s.

[0016] Based on the above technical solution, the smoke generation method of the smoke simulation training system of the present invention can be further improved as follows:

[0017] The smoke particles are cubes, and the side length of each cube is equal to the side length of each small cube in the second grid.

[0018] The step of obtaining multiple smoke screen rules to form a smoke screen rule set specifically includes:

[0019] Divide the scene of the smoke screen launcher releasing smoke in the actual training environment into multiple consecutive frames according to the time interval;

[0020] A first smoke matrix is ​​established for each frame of the consecutive frames. The first smoke matrix is ​​a three-dimensional matrix. Each element of the three-dimensional matrix corresponds one-to-one with each small cube of the first grid. If there is smoke in a small cube in the first grid, the element of the three-dimensional matrix corresponding to this small cube is set to 1. If there is no smoke in a small cube in the first grid, the element of the three-dimensional matrix corresponding to this small cube is set to 0.

[0021] Using the principle of cellular automata, a neural network is used to obtain the multiple evolutionary rules with the highest matching degree of the multiple consecutive frames, which are used as the smoke screen rule set.

[0022] The step of using each smoke rule in the smoke rule set to perform preliminary evolution of the smoke that has reached the first stable state to obtain multiple smoke states at different times specifically includes:

[0023] Initialize the smoke state and time step using each smoke rule in the smoke rule set;

[0024] Evolutionary calculations are performed for each smoke screen rule and time step to obtain multiple smoke screen states at different times;

[0025] The initial evolution time is 10 seconds.

[0026] Typically, a time step of 1 millisecond is used.

[0027] In the step of scoring multiple smoke screen states at different times, the scoring criteria considered are the density and diffusion degree of the smoke screen.

[0028] The step of scoring multiple smoke screen states at different times specifically includes:

[0029] Establish a smoke screen state dataset, including initial state smoke screen and continuous smoke screen states at specified intervals, denoted as the diffused smoke screen set. The initial state smoke screen is the smoke screen released by the smoke screen launcher in the actual training field environment when it reaches the first stable state in the windless environment.

[0030] Establish a smoke screen matrix set, including the initial state smoke screen matrix corresponding to the initial state smoke screen and the matrix corresponding to each smoke screen state in the diffusion smoke screen set, denoted as the diffusion matrix;

[0031] Establish a matching matrix set, transform the smoke screen of the first stable state and the multiple smoke screen states at each time step into multiple state matrices, and form a matching matrix set;

[0032] The first similarity analysis calculates the similarity between the initial state smoke screen matrix and the matrix corresponding to the smoke screen in the first stable state, and uses this as the first similarity.

[0033] The second similarity analysis calculates the similarity between each state matrix in the matching matrix set and the diffusion matrix at the same time, and uses this as the second similarity.

[0034] The score is calculated using the following formula:

[0035] Generally, the specified duration is 0.1 seconds.

[0036] Furthermore, the calculation method for the first similarity and the second similarity is a pre-similarity method.

[0037] In the step of continuing to evolve the smoke screen state corresponding to the target rule to generate a simulated smoke screen, the influence of wind on the simulated smoke screen when it exists in the simulated training field is also considered. The calculation method adopted is to use fluid dynamics calculation method to perform fluid dynamics calculation on the simulated smoke screen immediately after each evolution step, and use the simulated smoke screen after the fluid dynamics calculation as the output.

[0038] A second aspect of the present invention provides a computer-readable storage medium storing program instructions, which, when executed, are used to perform the smoke generation method of the smoke simulation training system described above.

[0039] A third aspect of the present invention provides a smoke screen generation system for a smoke screen simulation training system, comprising the aforementioned computer-readable storage medium.

[0040] Compared with existing technologies, the beneficial effects of the smoke generation method, medium, and system of the smoke simulation training system provided by this invention are:

[0041] 1. Using cellular automata, the diffusion of smoke screens in windless conditions is simulated with high accuracy;

[0042] 2. By using a neural network to obtain the multiple evolutionary rules with the highest matching degree among the multiple consecutive frames, the cellular automata rules can be inferred from the actual smoke screen diffusion situation and used as smoke screen evolution rules, which effectively improves the accuracy of smoke screen evolution; thus eliminating the need for manual pre-setting of evolutionary rules, but rather inferring the most suitable evolutionary rules from the actual situation.

[0043] 3. The initial smoke screen is evolved using multiple evolution rules with the highest matching degree, resulting in multiple evolved smoke screens. These multiple evolved smoke screens can be scored and optimized to obtain the optimal smoke screen evolution rule. Using this smoke screen evolution rule, the resulting Mooney smoke feast is not much different from the actual smoke screen diffusion.

[0044] 4. By utilizing the smoke screen state at each step of the evolution process, and employing fluid dynamics calculations to determine the wind force's impact on the smoke screen state, the wind force influence settings for the evolving smoke screen can be adjusted to better reflect actual training conditions. Attached Figure Description

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

[0046] Figure 1 The flowchart illustrates a smoke screen generation method for a smoke screen simulation training system provided by this invention. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] Furthermore, 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 technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0049] like Figure 1 The diagram shown is a flowchart of a smoke screen generation method for a smoke screen simulation training system provided by the first aspect of the present invention. This method includes the following steps:

[0050] S10. Based on the scenario of the smoke launcher releasing smoke in the actual training ground environment when there is no wind, the actual training ground environment is divided into the first grid, and multiple smoke rules are obtained to form a smoke rule set.

[0051] S20. Establish a simulated training field based on the actual training field environment when there is no wind, and divide the simulated training field into a second grid corresponding to the first grid.

[0052] S30. Set up a simulated smoke screen launcher in the simulated training field and set the initial smoke screen launched by the simulated smoke screen launcher. The initial smoke screen contains multiple smoke screen particles. The smoke screen particles are launched from the simulated launcher according to the rules of fluid dynamics and reach the first stable state. The first stable state is the state when the average velocity of the smoke screen particles is lower than the stable velocity threshold.

[0053] S40. Using each smoke rule in the smoke rule set, perform a preliminary evolution of the smoke that has reached the first stable state to obtain multiple smoke states at different times.

[0054] S50. Score multiple smoke screen states at different times, and take the smoke screen rule corresponding to the smoke screen state with the highest score as the target rule.

[0055] S60. Continue to evolve the smoke screen state corresponding to the objective rule to generate a simulated smoke screen.

[0056] In the above technical solution, the smoke particles are cubes, and the side length of the smoke particle cube is equal to the side length of each small cube in the second grid.

[0057] The step of obtaining multiple smoke screen rules to form a smoke screen rule set in the above technical solution specifically includes:

[0058] Divide the scene of the smoke launcher releasing smoke in the actual training environment into multiple consecutive frames according to the time interval;

[0059] Establish a first smoke matrix. For each frame of a continuous series of frames, establish a first smoke matrix. The first smoke matrix is ​​a three-dimensional matrix. Each element of the three-dimensional matrix corresponds one-to-one with each small cube in the first grid. If there is smoke in a small cube in the first grid, the element of the three-dimensional matrix corresponding to that small cube is set to 1. If there is no smoke in a small cube in the first grid, the element of the three-dimensional matrix corresponding to that small cube is set to 0.

[0060] Using the principle of cellular automata, a neural network is used to obtain the multiple evolutionary rules with the highest matching degree in multiple consecutive frames, which are used as the smoke screen rule set.

[0061] The specific implementation method for this step is as follows:

[0062] Data preprocessing

[0063] Before implementing the principles of cellular automata, data preprocessing is necessary. Specifically, each frame in the first smokescreen matrix needs to be transformed into a set of feature vectors to facilitate processing by machine learning algorithms. The feature vectors can be constructed using the following methods:

[0064] 1) Take the state (0 or 1) of each element in the first smoke matrix as a feature and construct a feature vector of length n, where n is the number of small cubes in the first grid.

[0065] 2) Take the state of each element in the first smoke matrix and the states of several surrounding elements (e.g., the 8 surrounding elements) as a feature, and construct a feature vector of length n.

[0066] Construction of neural networks

[0067] After data preprocessing, neural networks can be used to implement the principles of cellular automata. Specifically, convolutional neural networks (CNNs) can be used. CNNs can efficiently process image data, making them well-suited for processing data in the first smokescreen matrix.

[0068] The basic structure of a CNN includes convolutional layers, pooling layers, and fully connected layers. The convolutional and pooling layers extract features from the image, while the fully connected layers map the extracted features to the output layer to obtain the final prediction result.

[0069] Specifically, the following CNN structure can be used:

[0070] 1) Input layer: Input feature vector.

[0071] 2) Convolutional layer 1: Convolution operation is performed using a 5×5 convolutional kernel to obtain 64 feature maps.

[0072] 3) Pooling layer 1: Pooling operation is performed using a 2×2 pooling kernel to obtain 32 feature maps.

[0073] 4) Convolutional layer 2: Convolution operation is performed using a 3×3 convolution kernel to obtain 32 feature maps.

[0074] 5) Pooling layer 2: Pooling operation is performed using a 2×2 pooling kernel to obtain 16 feature maps.

[0075] 6) Fully connected layer 1: The 16 feature maps are unfolded into a vector, and then 128 neurons are used for fully connected operation.

[0076] 7) Fully connected layer 2: Uses multiple neurons to perform fully connected operations to obtain multiple output results.

[0077] Training Model

[0078] After constructing the neural network, the backpropagation (BP) algorithm can be used to train the model. The BP algorithm can improve the model's prediction accuracy by optimizing the loss function to update the weights and biases in the neural network.

[0079] Specifically, the following steps can be used for training:

[0080] 1) Divide the dataset into training set and test set.

[0081] 2) Input the training set into the neural network and calculate the prediction results.

[0082] 3) Calculate the error between the predicted result and the actual result, and then use the BP algorithm to update the weights and biases in the neural network.

[0083] 4) Use a test set to test the model's prediction accuracy.

[0084] 5) If the model's prediction accuracy meets the requirements, stop training; otherwise, return to step 2 to continue training.

[0085] Acquire multiple evolutionary rules

[0086] After training the model, a neural network can be used to predict the smoke screen state in the next frame. Specifically, the feature vector of the current frame can be input into the neural network to obtain the prediction result. The prediction result can be represented as a smoke screen matrix, where each element has a state of 0 or 1.

[0087] After obtaining the prediction results, the principles of cellular automata can be used to obtain multiple evolutionary rules.

[0088] Specifically, the following methods can be used:

[0089] 1) Transform the prediction results into the initial state of a cellular automaton.

[0090] 2) The evolution of the smoke screen was simulated using a cellular automata algorithm, resulting in multiple evolutionary rules.

[0091] 3) Use a neural network to evaluate the matching degree of each evolutionary rule, and select the multiple evolutionary rules with the highest matching degree as the smokescreen rule set.

[0092] The following methods can be used to evaluate the degree of fit of evolutionary rules:

[0093] 1) Transform each evolution rule into a smokescreen matrix.

[0094] 2) Compare the smoke screen matrix with the actual smoke screen matrix and calculate the similarity between them.

[0095] 3) Use similarity as the matching degree of evolutionary rules, and select the multiple evolutionary rules with the highest matching degree as the smoke screen rule set.

[0096] In summary, the principles of neural networks and cellular automata can be used to implement a method for determining smoke effects. Specifically, the data in the first smoke matrix can be transformed into feature vectors, then a CNN can be used to build a model, a backpropagation (BP) algorithm can be used to train the model, and finally, the principles of cellular automata can be used to obtain multiple evolutionary rules.

[0097] In the above technical solution, the step of using each smoke rule in the smoke rule set to perform preliminary evolution of the smoke reaching the first stable state to obtain multiple smoke states at different times is as follows:

[0098] Initialize the smoke state and time step using each smoke rule in the smoke rule set;

[0099] Evolutionary calculations are performed for each smoke screen rule and time step to obtain multiple smoke screen states at different times;

[0100] The initial evolution takes 10 seconds.

[0101] In step S30, we have obtained the smoke screen that has reached the first stable state, that is, the state where the average velocity of the smoke particles is lower than the stable velocity threshold. Next, we will use each smoke rule in the smoke screen rule set to perform preliminary evolution on these smoke screens, obtaining multiple smoke screen states at different times.

[0102] First, we need to define the smoke screen rule set. The smoke screen rule set consists of multiple smoke screen rules obtained from the scenario of a smoke launcher releasing smoke in an actual 3D training environment. Each smoke screen rule contains a set of parameters used to describe the evolution of the smoke screen. Assume we have N smoke screen rules, denoted as R1, R2, ..., R... N .

[0103] Next, we will apply each smoke rule from the smoke rule set to each smoke screen that reaches the first stable state, and perform evolution. Assume we have M smoke screens that have reached the first stable state, denoted as S1, S2, ..., S... M For each smoke screen S i We will apply each smoke rule in the smoke rule set to obtain multiple smoke states at different times.

[0104] The specific implementation method is as follows:

[0105] 1. For each smoke screen S iThe initialization time is t = 0.

[0106] 2. For each smoke rule R j Initialize the smoke screen state S i The evolution result is S ij (t=0)=S i .

[0107] 3. For each time step Δt, perform the following operations:

[0108] a. For each smoke rule R j Calculate the smoke screen state S ij The evolution result S of the next time step of (t) ij (t+Δt).

[0109] b. Repeat step a until the preset maximum time step is reached or the smoke screen state meets the termination condition.

[0110] c. Set the smoke screen state S ij (t+Δt) represents the smoke screen state S. ij The evolution results of (t).

[0111] 4. Repeat step 3 until the evolution results for all time steps are obtained.

[0112] Through the above steps, we can obtain multiple smoke screen states S at different times. ij (t), where i = 1, 2, ..., M represents the i-th smoke screen that reaches the first stable state, and j = 1, 2, ..., N represents the j-th smoke screen rule.

[0113] In practice, the evolution of smoke screen rules can be simulated using numerical methods, such as the Euler method or the Runge-Kutta method. The specific evolution process needs to be calculated based on the specific form of the smoke screen rules.

[0114] In the above technical solution, the scoring criteria for evaluating multiple smoke screen states at different times are the density and diffusion degree of the smoke screen.

[0115] Next, we will score these smoke screen states to determine the smoke screen rule corresponding to the highest-scoring smoke screen state as the target rule.

[0116] The purpose of the scoring is to objectively and quantitatively evaluate the smoke screen condition based on its characteristics and pre-defined scoring criteria. The scoring criteria can be tailored to specific needs, such as smoke density, diffusion degree, and duration. In this paper, we will use smoke density and diffusion degree as the scoring criteria, and the specific implementation method is as follows:

[0117] 1. Define the density scoring function D(S) for the smoke screen. ij (t) and diffusion score function S(S) ij (t) are used to evaluate the smoke state S. ij The density and diffusion degree of (t). These scoring functions can be defined according to actual needs, such as statistical methods based on the number and distribution of smoke particles.

[0118] 2. For each smoke state S ij (t), calculate its density score D(S) ij (t) and diffusion score S(S) ij (t)).

[0119] 3. Based on the preset weights, the density score D(S) is... ij (t) and diffusion score S(S) ij We perform a weighted summation of (t) to obtain the comprehensive score F(S). ij (t)).

[0120] F(S ij (t))=w D ·D(S ij (t))+w S ·S(S ij (t))

[0121] Among them, w D and w S These represent the weights of the density score and the diffusion score, respectively.

[0122] 4. For each smoke rule R j Calculate the corresponding smoke screen state S ij The overall score F(S) of (t) ij The average value of (t)).

[0123]

[0124] in, Smoke screen rule R j The average rating.

[0125] 5. Select the smoke rule with the highest average score from all smoke rules as the target rule.

[0126]

[0127] Through the above steps, we can score multiple smoke screen states at different times and select the smoke screen rule corresponding to the smoke screen state with the highest score as the target rule.

[0128] In practical implementation, the definitions of the density scoring function and the diffusion degree scoring function need to be determined based on actual needs. Image processing and computer vision technologies can be considered for the analysis and evaluation of smoke screen conditions. The scoring criteria need to be adjusted and optimized according to the actual application scenario.

[0129] The step of scoring multiple smoke screen states at different times in the above technical solution specifically includes:

[0130] Establish a smoke screen state dataset, including the initial state smoke screen and continuous smoke screen states at specified intervals, denoted as the diffused smoke screen set. The initial state smoke screen is the smoke screen released by the smoke screen launcher in the actual training field environment when it reaches the first stable state under windless conditions.

[0131] Establish a smoke screen matrix set, including the initial state smoke screen matrix corresponding to the initial state smoke screen and the matrix corresponding to each smoke screen state in the diffusion smoke screen set, denoted as the diffusion matrix;

[0132] Establish a matching matrix set, transform the smoke screen of the first stable state and the multiple smoke screen states at each time step into multiple state matrices, and form a matching matrix set;

[0133] The first similarity analysis calculates the similarity between the initial smoke screen matrix and the matrix corresponding to the smoke screen in the first stable state, and uses this as the first similarity.

[0134] The second similarity analysis calculates the similarity between each state matrix in the matching matrix set and the diffusion matrix at the same time, and uses this as the second similarity.

[0135] The score is calculated using the following formula:

[0136] Furthermore, in the above technical solution, the calculation method for the first similarity and the similarity in the second similarity is a pre-similarity.

[0137] In the above technical solution, the smoke screen state corresponding to the target rule is further evolved. In the step of generating the simulated smoke screen, the influence of wind in the simulated training field on the simulated smoke screen is also considered. The calculation method adopted is to use fluid dynamics calculation method to perform fluid dynamics calculation on the simulated smoke screen immediately after each evolution step, and use the simulated smoke screen after the fluid dynamics calculation as the output.

[0138] Methods involving fluid dynamics calculations can utilize fluid dynamics models such as the Navier-Stokes equations for solution. Numerical methods, such as the finite difference method and the finite element method, can also be used to solve fluid dynamics problems. Specific algorithms and formulas are as follows:

[0139] 1. Navier-Stokes equations:

[0140] The Navier-Stokes equations are the fundamental equations describing fluid motion, and their general form is:

[0141]

[0142] Where u is the fluid velocity vector, θ is time, p is pressure, ρ is fluid density, ν is dynamic viscosity, and f is external force vector.

[0143] 2. Finite Difference Method: The finite difference method is a commonly used numerical method for discretizing partial differential equations. In fluid dynamics calculations, the finite difference method can be used to discretize and solve the Navier-Stokes equations. The basic idea of ​​the finite difference method is to approximate the derivatives in the partial differential equations using finite differences, and then solve the difference equations.

[0144] 3. Difference equation:

[0145] During the evolution of the simulated smoke screen, the state of the simulated smoke screen can be updated using difference equations. The general form of the difference equation is:

[0146] τ ω+1 =τ ω +ΔξF(τ ω )

[0147] Where, τ ω+1 This is the smoke screen state at the next time step, τ ω This represents the smoke screen state at the current time step, where Δξ is the time step size, and F(τ) is the smoke screen state at the current time step. ω ) is the smoke screen state update amount calculated based on fluid dynamics, and ω represents the evolution step size.

[0148] 4. Termination Condition Determination: During the simulated smoke screen evolution, it is necessary to determine whether a termination condition has been met. The termination condition could be that the smoke screen's evolution time reaches a set upper limit, or that the smoke screen's density falls below a set threshold. The termination condition can be determined using logical judgments and conditional statements.

[0149] A second aspect of the present invention provides a computer-readable storage medium storing program instructions, which, when executed, are used to perform the smoke generation method of the smoke simulation training system described above.

[0150] A third aspect of the present invention provides a smoke screen generation system for a smoke screen simulation training system, comprising the aforementioned computer-readable storage medium.

[0151] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for generating smoke in a smoke simulation training system, characterized in that, Includes the following steps: S10. Based on the scenario of a smoke screen launcher releasing smoke in the actual training ground environment when there is no wind, the actual training ground environment is divided into a first grid, and multiple smoke screen rules are obtained to form a smoke screen rule set. S20. Establish a simulated training field based on the actual training field environment when there is no wind, and divide the simulated training field into a second grid corresponding to the first grid. S30. Set up a simulated smoke screen launching vehicle in the simulated training field, and set the initial smoke screen launched by the simulated smoke screen launching vehicle. The initial smoke screen contains multiple smoke screen particles. The smoke screen particles are launched from the simulated smoke screen launching vehicle according to the rules of fluid dynamics and reach a first stable state. The first stable state is the state when the average speed of the smoke screen particles is lower than the stable speed threshold. S40. Using each smoke rule in the smoke rule set, the smoke that has reached the first stable state is initially evolved to obtain multiple smoke states at different times. S50. Score multiple smoke screen states at different times, and take the smoke screen rule corresponding to the smoke screen state with the highest score as the target rule. S60. Continue to evolve the smoke screen state corresponding to the target rule as the generated simulated smoke screen; The step of obtaining multiple smoke screen rules to form a smoke screen rule set specifically includes: Divide the scene of the smoke launcher releasing smoke in the actual training environment into multiple consecutive frames according to the time interval; A first smoke matrix is ​​established for each frame of the consecutive frames. The first smoke matrix is ​​a three-dimensional matrix. Each element of the three-dimensional matrix corresponds one-to-one with each small cube of the first grid. If there is smoke in a small cube in the first grid, the element of the three-dimensional matrix corresponding to this small cube is set to 1. If there is no smoke in a small cube in the first grid, the element of the three-dimensional matrix corresponding to this small cube is set to 0. The principle of cellular automata is used to obtain the multiple evolutionary rules with the highest matching degree of the multiple consecutive frames using a neural network, which are used as the smoke screen rule set; The step of using each smoke rule in the smoke rule set to perform preliminary evolution of the smoke that has reached the first stable state to obtain multiple smoke states at different times specifically includes: Initialize the smoke state and time step using each smoke rule in the smoke rule set; Evolutionary calculations are performed for each smoke screen rule and time step to obtain multiple smoke screen states at different times; The initial evolution time is 10 seconds.

2. The smoke screen generation method of a smoke screen simulation training system according to claim 1, characterized in that, The smoke particles are cubes, and the side length of the smoke particle cube is equal to the side length of each small cube in the second grid.

3. The smoke screen generation method of a smoke screen simulation training system according to claim 1, characterized in that, In the step of scoring multiple smoke screen states at different times, the scoring criteria considered are the density and diffusion degree of the smoke screen.

4. The smoke screen generation method of a smoke screen simulation training system according to claim 1, characterized in that, The step of scoring multiple smoke screen states at different times specifically includes: Establish a smoke screen state dataset, including initial state smoke screen and continuous smoke screen states at specified intervals, denoted as the diffused smoke screen set. The initial state smoke screen is the smoke screen released by the smoke screen launcher in the actual training field environment when it reaches the first stable state in the windless environment. Establish a smoke screen matrix set, including the initial state smoke screen matrix corresponding to the initial state smoke screen and the matrix corresponding to each smoke screen state in the diffusion smoke screen set, denoted as the diffusion matrix; Establish a matching matrix set, transform the smoke screen of the first stable state and the multiple smoke screen states at each time step into multiple state matrices, and form a matching matrix set; The first similarity analysis calculates the similarity between the initial state smoke screen matrix and the matrix corresponding to the smoke screen in the first stable state, and uses this as the first similarity. The second similarity analysis calculates the similarity between each state matrix in the matching matrix set and the diffusion matrix at the same time, and uses this as the second similarity. The score is calculated using the following formula: .

5. The smoke screen generation method of a smoke screen simulation training system according to claim 1, characterized in that, In the step of continuing to evolve the smoke screen state corresponding to the target rule to generate a simulated smoke screen, the influence of wind on the simulated smoke screen when it exists in the simulated training field is also considered. The calculation method adopted is to use fluid dynamics calculation method to perform fluid dynamics calculation on the simulated smoke screen immediately after each evolution step, and use the simulated smoke screen after the fluid dynamics calculation as the output.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program instructions, which, when executed, are used to perform a smoke screen generation method for a smoke screen simulation training system according to any one of claims 1-5.

7. A smoke screen generation system for a smoke screen simulation training system, characterized in that, Includes the computer-readable storage medium as described in claim 6.

Images