AnyLogic-based strategic game simulation and deduction platform

Abstract

According to the method, a strategic game simulation and deduction platform is established based on an AnyLogic tool, the problem of possibility of occurrence of future items in a strategic game is mainly solved, the probability of decision selection of a game player and the probability of option implementation results are calculated, and possible results of conflicts are simulated and deduced. The platform mainly comprises a data import module, a game evolution graph construction module and a strategic game simulation and deduction module. Wherein the data import module imports initial data for strategic game analysis from an interface; the game evolution graph construction module is used for constructing a game evolution graph; and the strategic game simulation and deduction module is used for carrying out strategic game simulation and deduction according to the game evolution graph, and comprises three modes of ''everyone game'', ''machine-machine game'' and ''man-machine game''. Through the platform provided by the invention, the functions of initial data import, game evolution graph construction, strategic game analysis and deduction and strategic game simulation can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of strategic game analysis, and in particular to a strategic game simulation and deduction platform based on AnyLogic. Background Art

[0002] Conflicts often involve multiple players, and the actions and decisions of the players will have an impact on other parties, which in turn will affect their own decisions, thus forming a complex system. Agent-Based Modeling (ABM) is a computational modeling method used to study the behavior of complex systems. It explores the changes and development of the overall system by modeling and simulating individual agents in the system. In the actual strategic game process, the position, influence and importance of each player will change with the implementation of the influencing strategy. The analysis of probability distribution is difficult to conduct dynamic analysis of conflicts, and it is necessary to simulate and emulate strategic games. Therefore, how to simulate and deduce strategic games on the construction of a simulation analysis platform is a problem that needs to be solved urgently. AnyLogic combines multiple modeling methods such as system dynamics, discrete events, and agents, supports arbitrary combination of multi-method modeling, and fully supports object-oriented and hierarchical modeling. The present invention uses AnyLogic as a research and development platform and combines strategic game analysis models to build a strategic game simulation and deduction platform. Summary of the invention

[0003] The purpose of the present invention is to provide a strategic game simulation and deduction platform based on AnyLogic, which is used to realize data import, game evolution map construction, strategic game analysis and deduction, and strategic game simulation functions.

[0004] To achieve the above objectives, the present invention provides a strategic game simulation and deduction platform based on AnyLogic, which is characterized by comprising

[0005] Data import module, used to import initial data for strategic game analysis;

[0006] Game evolution graph construction module, used to construct game evolution graph;

[0007] The strategic game simulation and deduction module is used to simulate and deduces strategic games based on the game evolution graph, including three modes: "human-human game", "machine-machine game" and "human-machine game".

[0008] Furthermore, the game evolution graph construction module includes a game party submodule, an option submodule and a game evolution graph establishment submodule;

[0009] The game party includes a game party attribute setting unit and a decision interface, wherein the game party attributes include name, options owned, and risk preference; the decision interface is used to implement decision selection of options based on prospect theory;

[0010] The option submodule includes an option attribute setting unit and an implementation interface. Option attributes include name, game party name, status, position, influence, and importance. The implementation interface is used to implement the option.

[0011] The game evolution graph establishment submodule includes a node setting unit, an option movement unit, a result movement unit, a state calculation interface, a utility calculation interface, an add node interface, a delete node interface, an add movement interface, and a delete movement interface.

[0012] Furthermore, the nodes include root nodes, decision nodes, implementation nodes and final nodes; the attributes of the decision nodes include the name of the player, utility, and status, the utility attribute is calculated through the utility calculation interface, and the status attribute is calculated through the status calculation interface; the attributes of the implementation node include options, time, and location, which are used to represent the option object currently implemented.

[0013] Further, the strategic game simulation and deduction module includes a graph database, an algorithm submodule and a game controller;

[0014] Graph database, based on the game evolution graph, acts as a database at the data level, used for calling and storing data in the game analysis process, represents the perception of each player on the future conflict trend at the ABM modeling level, and refers to the graphical representation of strategic ideas at the strategic planning level;

[0015] The algorithm submodule is used to provide the algorithm required to be executed in the entire strategic game analysis process, and gradually promote the strategic game simulation through interaction with the game evolution graph and the game controller;

[0016] The game controller is used to control the strategic game process. During the strategic game simulation process, it locates the current node on the game evolution graph and calls the relevant algorithms in the algorithm submodule to realize the evolution of the strategic game.

[0017] Furthermore, the algorithm submodule includes a state generation algorithm, a utility calculation algorithm, a decision selection algorithm, and an option implementation algorithm;

[0018] The state generation algorithm is used to assign the state of each decision node in the game evolution graph;

[0019] Utility calculation algorithm, used to calculate the utility value of each player at each decision node;

[0020] Decision selection algorithm, which is used to calculate the psychological utility and selection probability of each option based on prospect theory at the decision node;

[0021] The option implementation algorithm is used to estimate the probability of the option implementation results based on the relative strength of each game player at the implementation node.

[0022] Therefore, the present invention adopts agent-based modeling (ABM) to establish a strategic game deduction and simulation experimental platform on AnyLogic, which can realize the game evolution map construction, strategic game analysis and deduction, and strategic game simulation functions. The platform has the following beneficial effects:

[0023] First, analyze the heterogeneity of intelligent agents. Analyzing the behaviors and decision-making processes of different intelligent agents can better capture the heterogeneity between intelligent agents. By modeling intelligent agents, their perception, decision-making, interaction and other behaviors can be studied;

[0024] Second, visualize evolution and dynamics. This platform analyzes the evolution and dynamics of strategic games, simulates the interaction between intelligent agents, and visualizes the evolution path and state changes of conflicts at different time points.

[0025] Third, predict the conflict results. This platform simulates the evolution trend and experimental results of conflicts in different scenarios by setting agent attributes, environmental parameters and behavioral rules.

[0026] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 UML class diagram of the analysis platform constructed for the present invention.

[0028] Figure 2 This is the framework diagram of the algorithm submodule in the strategic game analysis and deduction module.

[0029] Figure 3 This is the game evolution graph constructed in the embodiment.

[0030] Figure 4 It is a diagram of the “machine-machine game” interface in the embodiment.

[0031] Figure 5 It is a simulation statistical chart of “machine-machine game” in the embodiment.

[0032] Figure 6 This is a diagram of the “human-computer game” interface in the embodiment.

[0033] Figure 7 This is a diagram of the “Everyone Plays Game” interface in the embodiment. DETAILED DESCRIPTION

[0034] In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art make various changes or modifications to the present invention, and these equivalent forms also fall within the scope limited by the claims attached to the application.

[0035] 1. The overall framework of the strategic game analysis platform

[0036] based on Figure 1 As shown in the UML class diagram, the classes of this platform mainly include players (Actor), options (Option), game evolutionary graph (GameEvolutionaryGraph), game controller (GameController), decision node (DecisionNode), option movement (OptionMove), implementation node (ImplementationNode), result movement (ResultMove), and two functional interfaces implementation (Implementation) and decision (Decision). Among them, the players have multiple options, and there is an association relationship between the options. The players realize the decision-making function, and the options realize the implementation function; the construction of the game evolution graph depends on the players and options, and the game controller depends on the game evolution graph, which is mainly used to control the process of strategic game analysis, deduction and simulation; the game evolution graph is formed by the combination of decision nodes, option movement, implementation nodes and result movement.

[0037] The simulation and deduction platform proposed in the present invention mainly includes a data import module, a game evolution graph construction module, and a strategic game simulation and deduction module;

[0038] Data import module, imports initial data for strategic game analysis from the interface;

[0039] Game evolution graph construction module, used to construct game evolution graph;

[0040] The strategic game simulation and deduction module is used to simulate and deduces strategic games based on the game evolution graph, including three modes: "human-human game", "machine-machine game" and "human-machine game".

[0041] 2. Data import module

[0042] It is mainly used to import initial data from expert evaluation data. The initial data includes the game parties, options, option status, the game party's position on the option, the game party's importance on the option, and the game party's influence on the option.

[0043] (1) Game parties

[0044] It includes a player attribute setting unit and a decision interface, wherein the player attributes include name, owned options, and risk preference; the decision interface is used to implement option decision selection based on prospect theory;

[0045] (2) Options

[0046] It includes an option attribute setting unit and an implementation interface, wherein the option attributes include name, actor name, state, position, influence, and importance, and the type of the state attribute is a dictionary, which represents the key-value pairs of all possible states and state values ​​of the option, the position attribute is the key-value pair of the actor and the actor's position on the option, the influence attribute is the key-value pair of the actor and the actor's influence on the option, and the importance attribute is the key-value pair of the actor and the actor's importance on the option; the implementation interface is used to implement the option;

[0047] 3. Game evolution graph construction module

[0048] The construction of the game evolution graph depends on the game parties and options, which includes a node setting unit, an option movement (option_move) unit, a result movement (result_move) unit, and a state calculation (generate_state) interface, a utility calculation (generate_utility) interface, an add node (add_node) interface, a delete node (del_node) interface, an add movement (add_move) interface, and a delete movement (del_move) interface;

[0049] The nodes include root nodes (root_node), decision nodes (decision_node), implementation nodes (implementation_node) and final nodes (final_node); the attributes of decision nodes include player names (Actor_name), utility (utility), and state (state), and the utility attribute is a key-value pair of the player and the utility value of the player at the decision node, which can be calculated through the utility calculation interface, and the state attribute is a key-value pair of options and option states, which can be calculated through the state calculation interface; the attributes of implementation nodes include options (option), time (time), and place (place), which are used to represent the currently implemented option object; the option moving unit includes name setting, psychological utility calculation, and option probability calculation subunits; the result moving unit includes name setting and result probability setting subunits, wherein the result probability is calculated by the implementation interface of the option;

[0050] The state calculation interface is used to calculate the state of each option under each decision node; the utility calculation interface is used to calculate the utility size of each decision node according to the state of the options under each decision node; the add node interface, delete node interface, add mobile interface and delete mobile interface are used to add nodes, delete nodes, add mobile nodes and delete mobile nodes respectively.

[0051] 4. Strategic game simulation and deduction module

[0052] like Figure 2 As shown, it includes a graph database, an algorithm submodule and a game controller; wherein,

[0053] Graph database, based on the game evolution graph, acts as a database at the data level, used for calling and storing data in the game analysis process, represents the perception of each player on the future conflict trend at the ABM modeling level, and refers to the graphical representation of strategic ideas at the strategic planning level;

[0054] The algorithm submodule is used to provide the algorithm required to be executed in the entire strategic game analysis process. Through the interaction with the game evolution graph and the game controller, the strategic game simulation is gradually promoted. The initial data is imported through the data import module.

[0055] The game controller is used to control the process of strategic game. During the simulation of strategic game, the current node is located on the game evolution diagram, and the relevant algorithms in the algorithm submodule are called to realize the evolution of strategic game. When it is at the decision node, the decision algorithm is called. The decision algorithm calculates the psychological utility and option probability of all options, and then returns the selected option according to the Monte Carlo simulation. The option is implemented at the node, and the implementation algorithm is called. The implementation algorithm obtains the probability of the option implementation result, and then returns the implementation result according to the Monte Carlo simulation, so that the conflict development context evolves forward. Finally, it is determined whether the final node is reached. If the final node is not reached, the above process is repeated. If the final node is reached, the simulation is stopped.

[0056] The algorithm submodule includes state generation algorithm, utility calculation algorithm, decision selection algorithm, and option implementation algorithm. The following is an introduction to each algorithm:

[0057] (1) State Generation Algorithm

[0058] The state generation (generate_state) algorithm is executed after the planner builds the game evolution graph or updates the graph each time. It is used to assign the state (state) of each decision node in the game evolution graph. State is a dictionary with option name as key and option state as value, which indicates the state of all options at the current decision node. As shown in Table 1, the algorithm mainly traverses all decision nodes, assigns the state value of the decision node to the subsequent decision node, and updates the state value of the subsequent decision node according to the options and option states between the two decision nodes, so as to update the state value of all decision nodes in the game evolution graph.

[0059] Table 1 State generation algorithm

[0060]

[0061] (2) Utility calculation algorithm

[0062] The utility calculation (generate_utility) algorithm is used to calculate the utility value of each player at each decision node, as shown in Table 2. First, each decision node (dn) in the game evolution graph is traversed, and each player (a) is traversed for each decision node. Then, the state and state value of each option under the current decision node are traversed, and the three critical values ​​of importance (imp), position (pos) and state value (statevalue) are obtained. According to the obtained importance, position and state value, the utility value of each option is calculated, and the utility value of the player at the decision node is obtained by adding up the utility values ​​of all options (uti). Finally, the player name is used as the key and the utility value is assigned to the utility dictionary of the decision node.

[0063] Table 2 Utility generation algorithm

[0064]

[0065] (3) Decision selection algorithm

[0066] The decision selection algorithm is implemented based on prospect theory, and is used to calculate the psychological utility and selection probability of each option at the decision node based on prospect theory. The input of the algorithm is the current decision node dn and the game evolution graph GEG, and the output is a dictionary dict with each option as the key and the selection probability as the value after the decision node. As shown in Table 3, first traverse each choice (rm) after the current decision node, then traverse all the final nodes (fn) after the choice, and then calculate the utility difference (△uti) between the current decision node (dn) and the final node (fn), traverse all the moves between the current option and the final node, and judge if the move belongs to the option move, then take the previous decision node of the option move as input, run decision(), and perform recursive operations. If the move belongs to the result move, take the probability value of the result move (p_of_result), and all the moves on the path The product of the probability values ​​of the movement is obtained to obtain the probability (p) from om to fn. According to the model of prospect theory, the value v(△uti) and the distortion probability w(p) are obtained, and the psychological utility of a single path is obtained. The psychological utility of all paths after om is accumulated to obtain the psychological utility (pu) of the om. Finally, according to the relative size of the psychological utility, the probability of each option (p_om) is obtained and assigned to the attribute p_of_option of each om. At the same time, a dictionary dict with om as the key and probability p_om as the value is returned. The return value is mainly for the convenience of recursive operations.

[0067] Table 3 Decision selection algorithm

[0068]

[0069]

[0070] (3) Option Implementation Algorithm

[0071] The option implementation algorithm is used to estimate the probability of the option implementation result according to the relative strength of each player at the implementation node. As shown in Table 4, firstly, the implemented option (option) is extracted according to the attributes of the implementation node (in), then all the results of the option after the implementation node (rm) are traversed, and then all the players (a) are traversed to obtain the player's position (pos), importance (imp), influence (ifl) and option state (statevalue) on the option. The player's support power (pow) on the current result is calculated according to the formula, and the support power of all players is accumulated to obtain pow_all_actor. Finally, the result probability (p) is calculated according to the relative size of the implementation results of each option, and it is stored in the attribute p_of_result of rm.

[0072] Table 4 Option implementation algorithm

[0073]

[0074] Example

[0075] Taking the conflict between country A and country B as an example, conflict analysis data is used as initial data input, and strategic game simulation and deduction are performed on the AnyLogic-based strategic game simulation and deduction platform constructed by the present invention.

[0076] 1. Data import: Import initial conflict analysis data, which includes game parties, game party options, option status, game party's position on the option, game party's importance on the option, and game party's influence on the option.

[0077] 2. Construct a game evolution graph. Figure 3 , a game evolution map is constructed based on the role-playing method. First, different roles are assigned to planners, and planners are made familiar with the role identity, intentions, interests and strength; then, the research topic is introduced, and the deduction process, rules and other precautions are made clear to the planners; then, the planners perform role-playing interactions. At different decision nodes, the role of the decision node thinks and analyzes the options that may be faced, and adds option movement branches to the system. Add an implementation node after one of the option branches, and then the planners collectively discuss the possible different results of the option, and add result movement branches to the system. Follow this order until all paths are analyzed; finally, the planners and experts in related fields discuss and remove unlikely paths, and finally form a game evolution map that represents the future development trend of the conflict between A and B.

[0078] 3. Strategic game simulation and deduction.

[0079] In this embodiment, two analysis forms are mainly designed: unmanned strategic game simulation and human-involved strategic game deduction. The unmanned strategic game simulation adopts the "machine-machine game" mode, and the human-involved strategic game deduction includes the "human-human game" and "human-machine game" modes. Through deduction and simulation, strategic ideas can be further tested and improved, the evolution trend of future conflicts can be analyzed, and planners can reach a consensus on the development of future conflicts.

[0080] (1) Machine-to-machine game. Figure 4 , adopting the method of "machine-machine game", through the intelligent agents of each player, according to the decision rules established by prospect theory, recursive reasoning autonomous game can be carried out, and the probability distribution of the final outcome can be analyzed. The results are as follows Figure 5First, set the number of simulations, then start the simulation, starting from RN and ending at FN, use red lines to indicate the path, and red flag icons to indicate the current location, and finally count the number of times the final node is reached, as shown below: Figure 5 As can be seen from the figure, the number of arrivals of FN1, FN2, FN6, FN7, FN11, FN12, FN18, FN23, and FN24 is relatively high, mainly in the evolutionary path of C's high-intensity intervention when B is at a disadvantage, and the use of XXX weapons when A is at a disadvantage. Among them, the number of arrivals of FN23 and FN24 is also relatively high. Although A will most likely choose large-scale conventional military operations, the evolutionary paths of FN23 and FN24 are relatively short, and the probability of the entire evolutionary path is relatively large, so the probability of FN23 and FN24 arriving is also relatively high. The number of arrivals of FN3, FN8, FN13, FN16, FN19, and FN22 is relatively small, mainly in the evolutionary path of A choosing to compromise and withdraw troops and B choosing to compromise and surrender, indicating that the possibility of A and B choosing to compromise is relatively small under the current circumstances.

[0081] (2) Human-computer game. There are two forms of strategic game deduction with human participation: one is "human-computer game" and the other is "human-human game". In "human-computer game", humans mainly make strategic decisions at the decision point to decide which option to choose, and the intelligent agent decides which option to choose according to the rules. The implementation result of the option is determined by the probability of the result. Figure 6 , Party C is played by a human, and Party A and Party B are played by intelligent agents. Through multiple interactions between humans and machines, it helps to analyze the possible judgments and decisions of Party C in different scenarios.

[0082] (3) Everyone plays the game. In the “Everyone plays the game”, the players play different roles, make decisions at key decisions, and simulate possible future development trends. This requires the players to have a deep understanding of the entire conflict and be familiar with the relevant information of the players. They are usually experts in the field. Figure 7 Different players play different roles and the game is played in a turn-based system. Each player can observe, judge, make decisions and act, and at the same time modify the game evolution map to analyze and judge possible future development trends.

[0083] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. AnyLogic-based strategic game simulation and deduction platform, characterized by: include: Data import module, used to import initial data for strategic game analysis; Game evolution graph construction module, used to construct game evolution graph; The strategic game simulation and deduction module is used to simulate and deduces strategic games based on the game evolution graph, including three modes: "human-human game", "machine-machine game" and "human-machine game".

2. The AnyLogic-based strategic game simulation and deduction platform according to claim 1, characterized in that: The game evolution graph construction module includes a game party submodule, an option submodule and a game evolution graph establishment submodule; The game party includes a game party attribute setting unit and a decision interface, wherein the game party attributes include name, options owned, and risk preference; the decision interface is used to implement decision selection of options based on prospect theory; The option submodule includes an option attribute setting unit and an implementation interface. Option attributes include name, game party name, status, position, influence, and importance. The implementation interface is used to implement the option. The game evolution graph establishment submodule includes a node setting unit, an option movement unit, a result movement unit, a state calculation interface, a utility calculation interface, an add node interface, a delete node interface, an add movement interface, and a delete movement interface.

3. The AnyLogic-based strategic game simulation and deduction platform according to claim 2, characterized in that: The nodes include root nodes, decision nodes, implementation nodes and final nodes; the attributes of decision nodes include the name of the player, utility, and status. The utility attribute is calculated through the utility calculation interface, and the status attribute is calculated through the status calculation interface; the attributes of implementation nodes include options, time, and location, which are used to represent the option object currently implemented.

4. The AnyLogic-based strategic game simulation and deduction platform according to claim 3, characterized in that: The strategic game simulation and deduction module includes a graph database, an algorithm submodule, and a game controller; Graph database, based on the game evolution graph, acts as a database at the data level, used for calling and storing data in the game analysis process, represents the perception of each player on the future conflict trend at the ABM modeling level, and refers to the graphical representation of strategic ideas at the strategic planning level; The algorithm submodule is used to provide the algorithm required to be executed in the entire strategic game analysis process, and gradually promote the strategic game simulation through interaction with the game evolution graph and the game controller; The game controller is used to control the strategic game process. During the strategic game simulation process, it locates the current node on the game evolution graph and calls the relevant algorithms in the algorithm submodule to realize the evolution of the strategic game.

5. The AnyLogic-based strategic game simulation and deduction platform according to claim 4, characterized in that: The algorithm submodule includes state generation algorithm, utility calculation algorithm, decision selection algorithm, and option implementation algorithm; The state generation algorithm is used to assign the state of each decision node in the game evolution graph; Utility calculation algorithm, used to calculate the utility value of each player at each decision node; Decision selection algorithm, which is used to calculate the psychological utility and selection probability of each option based on prospect theory at the decision node; The option implementation algorithm is used to estimate the probability of the option implementation results based on the relative strength of each game player at the implementation node.

Images