CN110109653A

CN110109653A - A kind of land battle war game intelligent engine and its operation method

Info

Publication number: CN110109653A
Application number: CN201910393123.9A
Authority: CN
Inventors: 程恺; 刘满; 余晓晗; 郝文宁; 陈刚; 刘斌; 别林; 陈亮
Original assignee: Army Engineering University of PLA
Current assignee: Army Engineering University of PLA
Priority date: 2019-05-13
Filing date: 2019-05-13
Publication date: 2019-08-09
Anticipated expiration: 2039-05-13
Also published as: CN110109653B

Abstract

The present invention discloses a kind of land battle war game intelligent engine and its operation method, is divided into general technological system and operation frame, technological frame includes the analysis mining module of war game data, using the soft optimization algorithm of Multi-attribute synthetic evaluation as the decision-making module of core；Operation frame includes enemy's situation processing module, multiple action modules and operational process frame.The dynamic data that the intelligent war game engine of design is generated by reading war game simulation system in real time, the static data of itself and war game is subjected to data fusion, generate attribute data, again using the soft optimization algorithm of Multi-attribute synthetic evaluation as the decision-making module of core to attribute data and war game primitive rule make inferences or operation after, the action command of chess piece is finally provided, and transfers the command to war game simulation system.Land battle war game intelligent engine of the invention, required computing resource is low, and real time execution walks son flexibly, and it is horizontal to have higher autonomous intelligence.

Description

A kind of land battle war game intelligent engine and its operation method

Technical field

The present invention relates to a kind of land battle war game intelligent engine and its operation method, for war game platform, integrated data excavates, The soft preferred method of Multi-attribute synthetic evaluation, designs a kind of intelligent engine.

Background technique

By means of the development of computer technology, traditional stringent formula war game starts to digitize.China's computer war game technology It starts late, becomes the research hotspot of each colleges and universities and incorporated business in recent years, research contents is concentrated mainly on the sheet for illustrating war game Matter and feature, research war game Cartographic Technique, chess piece computer aided algorithm, rule of inference system etc., some companies or research institution's design Have developed the war game simulation system of the different scales such as Tactics-level, campaign grade, the different synthesis arm of the services.Traditional war game is deduced more automatically It is manual operation, everybody fights mode, and with Internet of Things, the development of big data, artificial intelligence technology, intelligent war game starts into For the hot spot of research.War game intelligent engine refers to the situation letter that can receive deduction platform for war game simulation platform development Breath, an intelligence software module of output deduction movement after handling by analysis, it plays the simulation mankind and carries out war game simulation Effect.The technological approaches that war game intelligent engine is realized mainly has, three kinds of rule-based reasoning, searching algorithm and data mining approach. The intelligent engine of Process Based makes inferences operation according to war game rule, and the movement generally exported is relatively more fixed, lacks and becomes The property changed, moving law is easy to be grasped by opponent；Based on the intelligent engine of searching algorithm because war game state space is huge, need A large amount of computing resources are consumed, the calculating time is longer, and real-time is not strong；Intelligent engine based on data mining needs to excavate to be gone through in the past History deduces the Heuristics in data, more demanding to the quality and quantity of data.

In recent years, it has held serial war game simulation with local Institutions of higher leaning inside army and has competed, the accumulation of post-games computer system A large amount of competition data, these data is based on the intelligent war game engine that data knowledge excavates provide data resource.Currently, On excavation, the feature extraction of war game data the problems such as, certain methods are proposed, but based on theoretical research and inquiring into, are extracted Feature mainly analyze and determine that situation uses for mankind commander, complete set can mature operation dug based on data knowledge The war game intelligent engine realization of pick does not make substantial progress also.

Summary of the invention

Goal of the invention: sub- fixation is walked for existing rule-based war game intelligent engine, the war game intelligence based on searching algorithm Energy engine calculates time long disadvantage, and design realizes land battle war game intelligent engine, reduces required computing resource, improve The real-time of operation and the flexibility for walking son, it is horizontal to have higher autonomous intelligence.

Technical solution: the present invention devises war game entity operation intelligent decision making model frame first, as shown in Figure 1. Regular data, environmental data and the historical data in war game data are excavated in static data excavation, are extracted related to decision Static knowledge data.Dynamic data combining carries out data fusion to the real-time situation of knowledge data and deduction system, is formed The relevant attribute data of decision.The comprehensive war game primitive rule of multi-Agent action Sequence Decision, situation data and attribute data, it is complete It is generated at the order of action of war game operation entity.Order of action is performed after being sent to Tactics-level war game simulation system, deduces system System can pass out new situation data.So circulation realizes the intelligent decision of war game simulation system.

A kind of land battle war game intelligent engine carries out under the guidance of war game entity operation intelligent decision making model frame Refinement simultaneously programs realization.The engine is divided into general technological system and operation frame, and technological frame includes war game data mining Fusion Module, using the soft optimization algorithm of Multi-attribute synthetic evaluation as the decision-making module of core；Operation frame include enemy's situation processing module, Multiple action modules and operational process frame.

Static data in war game data is extracted through data mining algorithm and is closed by the excavation Fusion Module of the war game data Key information data element, basis of formation knowledge data carry out data fusion with the dynamic data that war game simulation system generates in real time, It generates and meets the three classes attribute data of mankind's intuitivism apprehension: enemy's situation attribute, my feelings attribute and environment attribute；According to deduce process and Enemy and we troops situation, then using the soft optimization algorithm of Multi-attribute synthetic evaluation be the decision-making module of core to enemy's situation attribute, my feelings attribute Comprehensive operation is carried out with environment attribute, the action command of chess piece is finally provided, and transfer the command to war game simulation system, drives Deduction system operation.

War game data can be divided into environmental data, regular data, historical data and situation data in the way of generation.Environment Data, regular data, historical data are static data；Situation data are dynamic data；Environmental data refers to war game simulation system The data that the quantization means of battlefield surroundings are generated；Regular data refers to war game simulation system to operation entity, rule of engagement Description and the data that generate；Historical data refers to the deduction data that war game simulation system is recorded during multiple deduce Summation；Situation data refer to generated during war game simulation description battlefield current intelligence data, including time data, Enemy's status data, our status data and take control point data by force.Fig. 2 describes war game data mining fusion process frame, is divided into 5 A step: (1) data select；(2) data prediction；(3) data conversion；(4) data mining；(5) data fusion.Data selection Data to be treated are selected depending on the application；Data prediction filters out incomplete competition data；Data conversion will count According to the format for being converted to processing；Data mining excavates valuable information according to requirement of engineering, and the information that the present invention excavates has: from Real-time enemy's situation information is excavated in the situation data of real-time war game simulation system, intervisibility letter is excavated from regular data and environmental data Control temperature is ceased and takes by force, it is for statistical analysis to the position of stage each chess piece each in historical data, extract historical position and upper Out-of-the-car position information.Data fusion is i.e. by the enemy's situation data fusion in real-time situation information into historical probabilities data, thus raw The attribute data of Cheng Gengneng reflection battlefield enemy situation.Finally, the data that war game data mining Fusion Module generates are classified as 3 classes Attribute data: (1) enemy's situation attribute.Observable being determined property of enemy's chess piece is analyzed, ratio is passed through to not observable enemy chess piece It matches historical data and carries out probability analysis, form real-time hostile location prediction after comprehensive.After comprehensive intervisibility situation, calculate to enemy Observation degree and by enemy the two evaluation attributes of observation degree.(2) my feelings attribute.By being carried out to our chess piece position in competition data Probability analysis forms the pre- in-position probability value of our chess piece, an evaluation attributes as our lower moved further position.We Chess piece is got on or off the bus location information, the selection for truck position above and below our chess piece.(3) environment attribute.It motor-driven can be arrived for lower step All positions, my less advantageous is evaluated with a control point distance is taken by force by landform and current location, control hot value is taken in formation by force, is made For an evaluation attributes of our lower moved further position.

It is described using the soft optimization algorithm of Multi-attribute synthetic evaluation as the decision-making module of core, be according to analysis mining module export " enemy's situation attribute ", " my feelings attribute " and " environment attribute " three classes attribute information and match situation data, integrated use more belong to Property the soft optimization algorithm of overall merit and after war game primitive rule judged, export chess piece order of action, and order of action is passed It is defeated to arrive war game simulation system.

The soft optimization algorithm of Multi-attribute synthetic evaluation is mentioned after merging Multi-attribute synthetic evaluation algorithm and soft optimization algorithm A kind of algorithm suitable for war game decision out.War game entity action type have infantry's chess piece get on or off the bus movement, it is motor-driven, shooting, Take control by force, in these movements, maneuver is most difficult to decision, because it is closely related with the selection of map location, such as tank chess piece one A motor-driven six lattice of most multipotency of motor-driven stage, then the selection of motor-driven terminal may have 96, and a terminal can at least generate a machine Dynamic scheme.

Some position can evaluate the significance level of operation entity by three attribute, enemy's situation attribute: chess piece moves A possibility that dynamic terminal is observed by enemy is the smaller the better, we is less susceptible to be attacked；My feelings attribute: it counts previous and walks Sub- experience, the more position of our dwell times are better；Environment attribute: it consumes one's strength less and is got over closer to the position for taking control point by force It is good.The selection of motor-driven terminal is typical Multi-attribute synthetic evaluation select permeability, can be solved with correlation theory.

Multi-attribute synthetic evaluation (Multi-attribute Comprehensive Evaluation, MCE) is according to evaluation Object and evaluation purpose, the evaluation index of certain feature of describing system is chosen from different sides, establishes index system, and pass through Certain mathematical model (or algorithm) realizes systemic overall merit.Weighted mean method is using most common by multiple evaluations Index value synthesizes the comprehensive evaluation value of a globality.For the Multi-attribute synthetic evaluation of the motor-driven terminal of war game entity, enable

Wherein, U_iFor the weighted synthetical evaluation value of i-th of final position (or i-th of scheme), x_ijFor i-th terminal J-th of attribute value, V_j(x_ij) it is x_ijStandardization evaluation of estimate, ω_jFor the weight coefficient of j-th of attribute value.

The preferred standard of traditional scheme is the selection highest scheme max (U of comprehensive evaluation value_i), but this preferred method It is not suitable for military issue selection.

In order to avoid traditional Multi-attribute synthetic evaluation optimization algorithm is in tactful preferably upper fixed defect, machine is introduced herein Learn the preferred method of soft probability in subject, proposes the soft optimization algorithm of Multi-attribute synthetic evaluation, i.e., be converted to comprehensive evaluation value The probability value of temperature can be regulated and controled, and the size according to probability carrys out selection scheme.

Preferred criteria is changed to:

K=random_choice (p) (3)

Wherein, p_iIt indicates by evaluation of estimate U_iThe probability value of conversion, τ > 0, referred to as " temperature factor " have the work of regulation probability With p is by p_iThe discrete probability distribution array of composition, random_choice () are according to probability selection coordinate function, and k is last It is preferred that the scheme serial number gone out.

As τ → 0, the probability value difference of different schemes is big, and it is maximum that the scheme finally selected is intended to comprehensive evaluation value Scheme；As τ →+∞, the probability value gap of different schemes is small, and the scheme finally selected is intended to randomly choose；As 1≤τ≤e When, quality and variation can be preferably taken into account, τ is bigger, then variability is stronger.

Scientific evaluation is carried out to position attribution each in war game as long as can be seen that, it is soft excellent using Multi-attribute synthetic evaluation Algorithm is selected, the motor-driven final position of war game entity can be calculated, then calls existing coordinates measurement algorithm (such as limiting distance The method that floods afterwards), so that it may export complete motor-driven order.

The excavation Fusion Module of the war game data, using historical data and map datum as data source, through data mining After fusion, multiple data forms are generated, the generation method of all kinds of tables is described in detail below.The historical data of every game is 8 Data form, record are competed initial troops' state and situation data during the games.Map datum includes map each six The information such as number, high level, the landforms of angle lattice, " hexagonal grid " is identical with the object that " point " refers in this patent.Data analysis mining Process has following data table generation: chess piece historical position probability table S [c, n, l, i], boarding position list L_up, out-of-the-car position List L_down, take by force control Thermometer (be divided into vehicle master and take control Thermometer by forceVehicle takes control Thermometer by forceVehicle master It is secondary to take control Thermometer by forceSoldier master takes control Thermometer by forceSoldier time takes control Thermometer by forceSoldier's primary and secondary is taken by force Control Thermometer), observation ability table (be divided into my hostile observation table C_E2O[n, i, j], I to enemy observe table C_O2E[n, i, J]) and real-time enemy's situation information list E.Into after data fusion, following 3 class data: (1) enemy's situation attribute are formed.Merge chess piece history After location probability table S [c, n, l, i], real-time enemy's situation information list E and observation ability table generate to enemy observation degree matrix M and by Oppose observation degree matrix(2) my feelings attribute.Our some chess piece is extracted from chess piece historical position probability table S [c, n, l, i] is middle Our historical position probability tables are formed in the location probability of current generationDuring direct reference data analysis mining The boarding position list L of generation_up, out-of-the-car position list L_down；(3) environment attribute.During direct reference data analysis mining What is generated takes control Thermometer by force.

Chess piece historical position probability table S [c, n, l, i] generation method.Original historical data information is more scattered, it is necessary to It will be by screening, integration, the process extracted.Screening is exactly to consider match integrality, grading factors of competing to obtain, and is filtered out incomplete Competition data.Integration is exactly that multiple information tables of all competition datas are integrated into 1, and essential record data category has scenario ID, match ID, chess piece ID, chess piece title, chess piece color, chess piece stage initial position, chess piece stage final position, stage obtain Point, the stage lose points, final score of compete, death mark totally 11 elements.Extract is exactly that the data after integration are carried out statistical Analysis, extracts each chess piece in the probability distribution in each stage final position, ultimately forms table S [c, n, l, i], and c is chess piece face Color index, n are chess piece name index, and l is phase index, and i is chess piece location index, and table content indicates that the n chess piece of c color exists Rest on the probability of the position i on map is how many after the l stage.

Boarding position list L_upWith out-of-the-car position list L_downGeneration method.With chess piece historical position probability table generation side Method is similar, by the process of screening, integration, extraction.Screening is exactly to consider match integrality, grading factors of competing to obtain, and is filtered out not Complete competition data.Integration is exactly that the information of all competition datas got on the bus and got off about soldier is integrated into 1 table, Essential record data category has scenario ID, match ID, chess piece ID, chess piece color, chess piece boarding position, chess piece out-of-the-car position totally 6 A element.Extraction is exactly that boarding position in the data after integration and out-of-the-car position are carried out number statistics, and number is greater than x position It sets number and is stored in boarding position list L respectively_upWith out-of-the-car position list L_downIt is interior.Boarding position list L_upWhen record soldier gets on the bus Map hexagonal grid number, out-of-the-car position list L_downMap hexagonal grid number when record soldier gets off.

Take control temperature table generating method by force.Take by force control Thermometer according to certain point with take control point distance by force and orographic factor is comprehensive is formed The big tables of data with map table etc. characterizes the position significance level of each point on map.Because landform to vehicle (comprising tank and Two class chess piece of battlebus) it is different with the physical demands of soldier influence, generation take by force control Thermometer time zone separating vehicles take by force control Thermometer and Soldier takes control Thermometer by force.There are two take control point by force on General maps: mainly taking control point by force with secondary and takes control point by force.According to preferentially taking which is accounted for by force Take control point by force, can calculate and main take control Thermometer by force, secondary take control Thermometer by force and primary and secondary takes control Thermometer by force.Main control Thermometer of taking by force indicates excellent First take by force to account for and mainly take control point by force, it is secondary take control Thermometer by force and indicate preferentially to take by force to account for time take control point by force, primary and secondary, which takes control Thermometer by force, to be indicated preferentially to take control by force Primary and secondary takes any one in control point by force.

Vehicle master takes control Thermometer by forceGeneration method.For certain point P (x, y) on map, main control point of taking by force is O_first(x_first, y_first), then the vehicle main control temperature of the point may be expressed as:

Wherein τ_commonFor the physical strength value that vehicle needs to consume by the common landform of a lattice, τ (x, y) be vehicle by P (x, Y) the physical strength value of point consumption, L (P, O_first) be P point arrive mainly take by force control put lattice quantity distance,Mainly to take control point and side by force Boundary's maximum lattice quantity distance.The vehicle main control temperature for finding out all hexagonal grids on map forms vehicle master and takes control Thermometer by force

Vehicle takes control Thermometer by forceGeneration method.For certain point P (x, y) on map, secondary control point of taking by force is O_second(x_second, y_second), then the vehicle secondary control temperature of the point may be expressed as:

Wherein τ_commonFor the physical strength value that vehicle needs to consume by the common landform of a lattice, τ (x, y) be vehicle by P (x, Y) the physical strength value of point consumption, L (P, O_second) be P point to it is secondary take by force control put lattice quantity distance,For it is secondary take by force control point with Boundary maximum lattice quantity distance.The vehicle secondary control temperature for finding out all hexagonal grids on map forms vehicle and takes control Thermometer by force

Vehicle takes control Thermometer by forceGeneration method.For certain point P (x, y) on map, the master of the point is found out Take control temperature by forceTake control temperature by force with secondaryAfterwards, primary and secondary is taken control temperature by force and be may be expressed as:

The vehicle primary and secondary control temperature for finding out all hexagonal grids on map forms vehicle primary and secondary and takes control Thermometer by force

Soldier master takes control Thermometer by forceGeneration method.Soldier is preferable to terrain adaptability when advancing, and takes control by force asking Orographic factor is not considered when temperature.For certain point P (x, y) on map, main control point of taking by force is O_first(x_first, y_first), then should Soldier's master control temperature of point may be expressed as:

Wherein, L (P, O_first) be P point arrive mainly take by force control put lattice quantity distance,Mainly to take control point and boundary by force most Big lattice quantity distance.The soldier's master control temperature for finding out all hexagonal grids on map forms soldier master and takes control Thermometer by force

Soldier time takes control Thermometer by forceGeneration method.For certain point P (x, y) on map, secondary control point of taking by force is O_second(x_second, y_second), then soldier's secondary control temperature of the point may be expressed as:

Wherein L (P, O_second) be P point to it is secondary take by force control put lattice quantity distance,Take control point and boundary by force to be secondary Maximum lattice quantity distance.The soldier's secondary control temperature for finding out all hexagonal grids on map forms soldier time and takes control Thermometer by force

Soldier time takes control Thermometer by forceGeneration method.For certain point P (x, y) on map, the master of the point is found out Take control temperature by forceTake control temperature by force with secondaryAfterwards, primary and secondary is taken control temperature by force and be may be expressed as:

The soldier's primary and secondary control temperature for finding out all hexagonal grids on map forms soldier's primary and secondary and takes control Thermometer by force

Observation ability table generating method.Observation ability table is divided into my hostile observation table C_E2O[n, i, j], I to enemy observe table C_O2E[n, i, j]).If chess piece is located in close terrain, viewing distance is constant, but observed distance halves.When in close terrain Chess piece when being observed that enemy's chess piece, the case where chess piece in close terrain is not observed there are enemy's chess piece, because This needs to consider respectively that my hostile observation observes enemy with me.Using our chess piece type n and Position Number i as line index, enemy Chess piece Position Number j is column index, if meeting my hostile observation condition, otherwise it is 0 that corresponding crosspoint content, which is 1, in table. All map hexagonal grids are traversed, my hostile observation table C is calculated_E2O[n, i, j].With the position of enemy chess piece type n and our chess piece Setting number i is line index, and enemy chess piece Position Number j is column index, if meeting me to enemy's observation condition, table accordingly intersects Point content is 1, is otherwise 0.All map hexagonal grids are traversed, calculate me to enemy observation table C_O2E[n, i, j].

Real-time enemy's situation information list E generation method.War game simulation system, being capable of output-response war in real time during deduction The situation data of field situation.Comprising can be by enemy's piece information that we observes in situation information, not observable piece information It is not shown in situation data.Enemy's chess piece type and location information are extracted, real-time enemy's situation information list E is formed.

Enemy's situation attribute data generation method.Comprising to enemy observation degree matrix M and by enemy's observation degree matrixEnemy is called first Feelings and hostile location update module function are by enemy's piece information in chess piece historical position probability table S [c, n, l, i] and in real time Enemy's situation information list E is merged, and real-time hostile location probability tables are generatedVehicle point P (x, y) to enemy's observation degree and by Enemy's observation degree can indicate are as follows:

Wherein i is that the hexagonal grid of P point is numbered,For 6 chess pieces of l stage c color shape after real-time situation data fusion At real-time hostile location probability matrix,For we the position i to enemy vehicle observation matrix,It is us in i Position to enemy soldier's observation matrix,It is enemy to the observation matrix of our vehicle, T is matrix transposition.In to map I All accessibility hexagonal grids are asked to enemy's observation degree in one motor-driven stage of side, and not accessibility hexagonal grid is set as enemy's observation degree 0, being formed with big matrixes such as map sizes is to enemy observation degree matrix M.It is all in our in map in a motor-driven stage can The hexagonal grid of arrival is asked by enemy's observation degree, and not accessibility hexagonal grid is set as 1 to enemy's observation degree, and the matrix eventually formed is to be opposed Observation degree matrix

My feelings attribute data generation method.Include our historical position probability tablesBoarding position list L_upWith Out-of-the-car position list L_down.Our historical position probability tablesIt is straight by chess piece color c, chess piece title n and deduction stage l It connects and is extracted from chess piece historical position probability table S [c, n, l, i], boarding position list L_upWith out-of-the-car position list L_down? It is generated in the data analysis mining stage.

Environmental attribute data generates.(it is divided into vehicle master comprising taking control Thermometer by force and takes control Thermometer by forceVehicle takes control by force ThermometerVehicle primary and secondary takes control Thermometer by forceSoldier master takes control Thermometer by forceSoldier time takes control temperature by force TableSoldier's primary and secondary takes control Thermometer by force), generated in the data analysis mining stage.

The operation frame of the war game intelligent engine is with technological frame for guidance, in conjunction with the situation of specific deduction system Data and war game rule, Programming is input with situation data, and multiple action commands are the software module of output, including enemy Feelings processing module, multiple action modules and operational process frame.

The enemy's situation processing module does analysis processing to the situation information of war game simulation system input, main to complete 2 Function: (1) enemy and we's strength updates: mainly according to situation information, the chess piece type and quantity that statistical updating enemy lives, and enemy Dead chess piece and quantity, the type and quantity of the visible chess piece of enemy, the sightless chess piece of enemy and quantity, the chess that we lives Son and quantity, our dead chess piece and quantity, information such as relative military strength between ourselves and the enemy, as programmed decision foundation；(2) enemy Location updating: by enemy's piece information in chess piece historical position probability table S [c, n, l, i] and real-time enemy's situation information list E into Row fusion, i.e., be set as 0 for the hostile location probability of our observable position, observable hostile location probability is set as 1, then Standardization processing is carried out, hostile location probability matrix is formed

The multiple action module is that the movement for being able to carry out chess piece each in war game simulation system is decomposed, often The decision of a movement, which executes, is used as a subprogram, calls for main program.The decision-making foundation of each movement is mainly according to war game The primitive rule of deduction system is judged, or provides objectives position according to the soft optimization algorithm of Multi-attribute synthetic evaluation. Shared fire takes control movement, movement of getting on the bus, movement of getting off, soldier's movement, battlebus movement, the mobile totally 7 classes movement of tank by force.

Shoot, take by force the generation rule of control movement.The rule shot and take control movement generation by force is relatively easy, according to war game simulation Rule is carried out this movement after chess piece meets shooting or takes control condition by force.

It gets on the bus the generation rule of movement of getting off.The difficult point of movement of getting on the bus and get off is to select the position got on or off the bus.It gets on the bus Rule be after satisfaction gets on the bus condition, the position of people and battlebus is located at the list L that gets on the bus_upIt is interior, then execute movement of getting on the bus.It gets off dynamic The rule of work is after satisfaction gets off condition, and the position of manned battlebus is located at the list L that gets off_downInterior or manned battlebus distance Take control point by force and be less than or equal to certain numerical value (present invention is set as 3), then executes movement of getting off.

Soldier's moving algorithm.Soldier is mobile, our historical position probability in my feelings attribute of overall meritWith Take control temperature totally 2 factors in environment attribute by force, generates chess piece movement routine.Because soldier's target is smaller, it is not easy by enemy attack, and Soldier is needed to detect enemy's situation to forward position, so not considering enemy's situation attribute.According to the soft optimization algorithm of Multi-attribute synthetic evaluation, consider ground All the points on figure, soldier's mobile integrated evaluations matrix may be expressed as:

Wherein, F be soldier can maneuvering range matrix (reaching is 1, is not reached as 0),For soldier's historical position Probability matrix (extracts) from historical position probability tables S [c, n, l, i],Take control temperature matrix by force for soldier's primary and secondary, is Matrix dot product,It is weighted factor, is according to circumstances manually set.

It is preferred then to carry out soft probability:

(x, y)=random_choice (p) (14)

Wherein, sum () is the summation of matrix all elements, and p is the discrete probability distribution big with body of a map or chart etc., random_ Chioce () is according to probability selection serial number function, and (x, y) is the last mobile terminal preferably gone out.Finally, calling coordinates measurement letter Number (subprogram write according to the method that floods (breadth first search) after existing limiting distance) generates chess piece movement routine.

Battlebus moving algorithm.Battlebus is mobile, is opposed observation degree matrix with reference in enemy's situation attributeWe in my feelings attribute Historical position probabilityTemperature totally 3 factors are controlled with taking by force in environment attribute, generate chess piece movement routine.According to more attributes The soft optimization algorithm of overall merit, considers the evaluation of all the points on map, and battlebus mobile integrated evaluations matrix may be expressed as:

Wherein, F be battlebus can maneuvering range matrix,Take control temperature matrix by force for vehicle primary and secondary,It is gone through for battlebus History location probability matrix (extracts) from historical position probability tables S [c, n, l, i],It is matrix to be opposed observation degree matrix Dot product,It is weighted factor, is according to circumstances manually set.

Then soft probability is preferred:

(x, y)=random_choice (p) (17)

Tank moving algorithm.Tank can carry out fire flexibility during motor-driven, and control process is relative complex, need to divide Not Ji Suan observed fire point and terminal, target mobile shooting and mobile without target has also been distinguished in the selection of observed fire point Shooting.

When not having enemy's situation information in situation information, tank is executed without target mobile shooting, and strategy is divided into following steps:

Step 1: pressing the soft optimization algorithm of Multifactor Comprehensive Evaluation, and consideration is taken control temperature, our historical position probability by force and opposed 3 factors of observation degree calculate the mobile terminal of tank referring to battlebus path

Step 2: tank is calculated from starting point to terminalIntermediate all possible points passed through.Based on to enemy's observation degree Calculation method, calculate these point to enemy observation degree (point that cannot pass through its to enemy observation degree be 0), formation observation degree matrix M, Consider this factor, by the soft optimization algorithm of multi-factor comprehensive, seeks observed fire point

Step 3: calling coordinates measurement function, generates the path of tank point from starting point to observed fire, and executes motor-driven Movement；

Step 4: situation information is updated, discovery enemy mover then calls shooting function to be shot at, no to then follow the steps Five；

Step 5: calling coordinates measurement function, generates path of the tank from observed fire point to terminal, and executes motor-driven dynamic Make.

When there is enemy's situation in situation information, tank execution has target mobile shooting, and strategy is divided into following steps:

Step 1: being ranked up enemy mover by threat degree, forms enemy's situation list；

Step 2: selection goal.Select a chess piece in enemy's situation list that it is removable to calculate tank as target in order With the presence or absence of that can shoot a little in dynamic range, such as it is not present, then selects next chess piece as target from enemy's situation list, and judge With the presence or absence of can shoot a little.So circulation executes step 3, if enemy's situation list target is not until finding the target that can be shot It can shoot, then stopping executing has target shooting, starts to execute and shoot without target observations.

Step 3: observed fire point is calculated.To the goal of selection, calculate minimum apart from the sum of tank and target range Point set, therefrom select nearest point with a distance from tank as observed fire point.

Step 4: calling coordinates measurement function, is generated to the path of observed fire point, executes maneuver.Reach observation After shooting point, after updating situation information, fire is executed.

Step 5: pressing the soft optimization algorithm of Multifactor Comprehensive Evaluation, and consideration is taken control temperature, our historical position probability by force and opposed 3 factors of observation degree calculate the mobile terminal of tank referring to battlebus pathCoordinates measurement function is called, is generated To the path of terminal, maneuver is executed.

Tank moving algorithm without target mobile shooting process and having target mobile shooting process to integrate, will be formed above The whole submodule for judging decision and exporting control command.

The operational process frame is the main program flow of entire war game intelligent engine, covers war game intelligent engine from starting The whole process that war game simulation is deduced to end.It comprises the steps of:

Step 1: initialization battle environment.It is communicated with war game simulation system, clearly deduces and start, and initialize intelligence Variable in energy war game engine；

Step 2: situation, amendment enemy's situation are updated.The situation information of war game simulation system push is read, updates and is pushed away in engine Drill stage variable, our chess piece list, the list of enemy's chess piece, the variable for taking control three-point state information by force etc. for condition judgement.It calls Enemy's situation processing module analyzes situation data, completes enemy and we's strength and updates and enemy's real time position probability updating, generation enemy Set probability matrix in orientationAnd it is saved in database；

Whether step 3: judging to deduce terminates.The foundation of judgement is to deduce to have arrived ending phase or our chess piece is complete Portion is dead or enemy is over match, if terminated, jumps to step 10 eight and destroys battle environment；If do not tied Beam then jumps to next step；

Step 4: judge whether that control can be taken by force.According to rule of inference, each chess piece is successively inquired, judges that we is at chess piece Can no arrival, which be taken control point by force and be executed, takes control movement (calling takes control action module by force) by force, if can take control by force, jumps to step Control movement is taken in five execution by force；If being unsatisfactory for taking control condition by force, step 6 is jumped to；

Step 5: control is taken in execution by force.The order taking control by force and acting is sent to war game simulation system.War game simulation system feedback order After running succeeded, step 2 is jumped to；

Step 6: judge whether to get on the bus.Each soldier's chess piece is successively inquired, judges whether our soldier meets into battlebus Condition (calling get on the bus action module), if can get on the bus, otherwise jump procedure seven jumps to step 8；

Step 7: execution is got on the bus.The order that soldier gets on the bus is sent to war game simulation system.War game simulation system feedback order After running succeeded, step 2 is jumped to；

Step 8: judge whether to get off.Each soldier's chess piece is successively inquired, judges whether our soldier meets the item got off Part (calling get off action module), if can get off, jumps to step 9, otherwise, jumps to step 10；

Step 9: execution is got off.The order that soldier gets off is sent to war game simulation system.War game simulation system feedback order After running succeeded, step 2 is jumped to；

Step 10: judge whether to shoot.Each chess piece is successively inquired, judges whether to meet shooting condition by chess piece that (calling is penetrated Hit action module), if can shoot, step 11 is jumped to, step 12 is otherwise jumped to；

Step 11: shooting is executed.Chess piece fire command is sent to war game simulation system.War game simulation system feedback order After running succeeded, step 2 is jumped to；

Step 12: judge whether soldier can be motor-driven.Successively whether inquiry soldier can be motor-driven, if can be motor-driven, Step 13 is jumped to, step 14 is otherwise jumped to；

Step 13: soldier is mobile.Soldier's mobile module is called, war game is transmitted to after generation soldier's movement routine order and pushes away It drills system and jumps to step 2 after deduction system feedback execution is ordered successfully；

Step 14: judge that can battlebus motor-driven.Successively whether inquiry battlebus can be motor-driven, if can be motor-driven, jumps Step 15 is gone to, step 10 six is otherwise jumped to；

Step 15: battlebus is motor-driven.Battlebus mobile module is called, war game is transmitted to after generation battlebus movement routine order and pushes away It drills system and jumps to step 2 after deduction system feedback execution is ordered successfully；

Step 10 six: judge that can tank motor-driven.Successively whether inquiry tank can be motor-driven, if can be motor-driven, jumps Step 10 seven is gone to, step 2 is otherwise jumped to；

Step 10 seven: tank is mobile.Tank mobile module is called, complete tank has target to penetrate without target shooting or tank It hits, and motor-driven or fire command is transmitted to war game simulation system, tank mobile module jumps to step 2 after being finished；

Step 10 eight: battle environment is destroyed.Clearly deducing with war game simulation system communication terminates, and discharges engine occupancy Memory source.

War game intelligent engine is realized based on python language, removes hostile location probability matrixIn situation information processing stage Outside generating, other tables of data are all generated in data preprocessing phase in advance.Main program flow can be briefly described are as follows: initialize first Environment is fought, situation information is then updated and updates enemy's situation, whether terminate, be not finished and then enter main actions if then judging to deduce Judge link；It successively inquires our all chess pieces and whether can execute and the movement such as take control by force, shoot, get on the bus, getting off, is motor-driven, it can be with It executes, then calls respective algorithms module, generate chess piece movement, final program returns to situation update module, executes next circulation. There is the difference of priority between difference movement, takes control movement highest priority by force, inquire first in a cycle, tank movement is excellent First grade is minimum, in a cycle last inquiry.

The utility model has the advantages that compared with prior art, land battle war game intelligent engine and its operation method provided by the invention have Following advantage:

1) data form, war game can be generated as by excavating match historical data and map datum, the knowledge excavated Intelligent engine at runtime can matrix operation, arithmetic speed is fast, strong real-time；

2) whole process does not set any fixation and walks sub- point, only carries out data mining to the deduction data of specific map and scenario, whole A method is suitable for the different maps and scenario of different war game system or same war game system, portable good；

3) chess piece shift position is provided based on the soft Optimal Decision-making algorithm of Multi-attribute synthetic evaluation, exploration appropriate is manually set After the factor, quality and variation are taken into account in chess piece shift position, breach general intelligence engine and walk the fixed disadvantage of son.

Detailed description of the invention

Fig. 1 is war game entity operation intelligent decision making model frame diagram of the invention；

Fig. 2 is war game data mining fusion process frame diagram of the invention；

Fig. 3 is the war game intelligent engine Run-time scenario figure of the embodiment of the present invention；

Fig. 4 is the tank mobile shooting action flow chart of the embodiment of the present invention；

Fig. 5 is the war game intelligent engine main program flow chart of the embodiment of the present invention.

Specific embodiment

Combined with specific embodiments below, the present invention is furture elucidated, but protection scope of the present invention is not limited to that:

Embodiment: " the brandreth attack group war game simulation system " used with " 2017 First Nationwide war game simulation contest " is soldier The deduction that chess is deduced platform (being not included in the present invention), collected during the games with " 2017 First Nationwide war game simulation contest " Data are data source (being not included in the present invention), the use of map are " urban settlment ", are carried out using python programming language Programming, introduces an application example of this war game intelligent engine.

System Run-time scenario.The present invention is unable to isolated operation, need to be designed for specific war game simulation system, recombinant Get up and run together, realizes that intelligence deduces function.Fig. 3 is the war game intelligent engine Run-time scenario figure of the embodiment of the present invention.Include Three long-range war game simulation system, interface platform and war game intelligent engine parts, triplicity use could complete entire war game Intelligence is deduced.Long-range war game simulation system and interface platform are the back-up environment of war game intelligent engine operation, be other companies or The designed software systems of research, are not protected by the invention patent.Long-range war game simulation system is traditional artificial war game simulation Platform operates in server end, is accessed and is watched by browser.Interface platform is the interface software designed for intelligent engine, is risen To the effect with long-range war game simulation system communication and data transmission, it is erected between war game simulation system and intelligent engine One bridge block.War game intelligent engine plays the role of " brain ", war game situation data is read by interface platform, by intelligent soldier After chess engine calculates, output chess piece movement is transferred to war game simulation system through interface platform, so that whole system be driven effectively to transport Row.

Intelligent war game design is guidance with war game entity operation intelligent decision making model frame, as shown in Figure 1.Static number Regular data, environmental data and the historical data in war game data are excavated according to excavating, extract static state relevant to decision Knowledge data.Dynamic data combining carries out data fusion to the real-time situation of knowledge data and deduction system, forms decision phase The attribute data of pass.The comprehensive war game primitive rule of multi-Agent action Sequence Decision, situation data and attribute data, complete war game The order of action of operation entity generates.Order of action is performed after being sent to Tactics-level war game simulation system, and deduction system can incite somebody to action New situation data pass out.So circulation realizes the intelligent decision of war game simulation system.In specific design process by with Gradually implement lower 3 parts: (1) static data excavates, and generates static data table.The static data realized in Fig. 1 excavates Function；(2) write function treatment submodule, including enemy's situation processing module, take by force control action module, fire module, get on the bus it is dynamic Make module, action module of getting off, soldier's mobile module, battlebus mobile module, tank mobile module.Realize the dynamic in Fig. 1 The rule-based legal action of data fusion and war game Agent action decision part judges and soft based on Multi-attribute synthetic evaluation 2 subfunctions of preferred motor-driven program decisions；(3) main program module is write, recited function handles submodule, realizes system function Energy.Realize the decision process control and action sequence subfunction of war game Agent action decision part in Fig. 1.

(1) static data excavates, and generates static data table, calls for function treatment submodule.

For the data that the soft optimization algorithm of Multi-attribute synthetic evaluation needs to use, collection is pushed away using data mining technology It drills data to be excavated, there is following data table generation: chess piece historical position probability table S [c, n, l, i], boarding position list L_up, out-of-the-car position list L_down, take by force control Thermometer (be divided into vehicle master and take control Thermometer by forceVehicle takes control Thermometer by forceVehicle primary and secondary takes control Thermometer by forceSoldier master takes control Thermometer by forceSoldier time takes control Thermometer by forceSoldier's primary and secondary takes control Thermometer by force), observation ability table (be divided into my hostile observation table C_E2O[n, i, j], I am right Enemy's observation table C_O2E[n, i, j]).Its basic process handled is by preceding 4 steps in Fig. 2: (1) data select；(2) data are located in advance Reason；(3) data conversion；(4) data mining.5th data fusion step is ongoing dynamic number during war game simulation According to treatment process, realized in writing function treatment submodule.

It generates chess piece historical position probability table S [c, n, l, i].By screening, integration, extraction process to original history ratio Match data are handled.Screening is exactly to consider match integrality, grading factors of competing to obtain, and selects the high competition data of quality.It is whole Closing is exactly that multiple information tables of all competition datas are integrated into 1, and essential record data category has scenario ID, match ID, chess Sub- ID, chess piece color, chess piece stage initial position, chess piece stage final position, totally 11 elements such as final score of competing.It extracts Be exactly the data after integration are for statistical analysis, extract each chess piece in the probability distribution in each stage final position, finally It is formed table S [c, n, l, i], c is chess piece color index, and n is chess piece name index, and l is phase index, and i is chess piece position rope Draw, table content indicates that the n chess piece of c color rests on the probability of the position i on map is how many after the l stage.Table 1 is opened up Table S [c, n, l, i] local circumstance is shown, the 1st is classified as chess piece color index, and the 2nd is classified as chess piece name index, and the 3rd is classified as rank Segment index, the number of the first row are chess piece location index.

Table 1

colour	name	stage	…	50032	50034	50036	…
								0	10	1	…	0	0.04310	0	…
1	10	2	…	0	0	0	…
								0	10	3	…	0.01724	0.00862	0.02586	…

Generate boarding position list L_upWith out-of-the-car position list L_down.With chess piece historical position probability table generation method class Seemingly, by the process of screening, integration, extraction.Screening is exactly to consider match integrality, grading factors of competing to obtain, and selects quality height Competition data.Integration is exactly that the information of all competition datas got on the bus and got off about soldier is integrated into 1 table, mainly Record data category has scenario ID, match ID, chess piece ID, chess piece color, chess piece boarding position, chess piece out-of-the-car position to want for totally 6 Element.Extraction is exactly that boarding position in the data after integration and out-of-the-car position are carried out to number statistics, and position of the number greater than 3 times is compiled It is stored in boarding position list L number respectively_upWith out-of-the-car position list L_downIt is interior.

It generates vehicle master and takes control Thermometer by forceFor certain point P (x, y) on map, main control point of taking by force is O_first (x_first, y_first), then the vehicle main control temperature of the point may be expressed as:

Wherein τ_commonFor the physical strength value that vehicle needs to consume by the common landform of a lattice, τ (x, y) be vehicle by P (x, Y) the physical strength value of point consumption, L (P, O_first) be P point arrive mainly take by force control put lattice quantity distance,Mainly to take control point and side by force Boundary's maximum lattice quantity distance.The vehicle main control temperature for finding out all hexagonal grids on map forms vehicle master and takes control Thermometer by forceTable 2 shows that local vehicle master takes control Thermometer by forceMain control point of taking by force is 1, is started far from temperature after taking control point by force Become smaller, such as 0.95,0.9 point, there are the points that village landform (the physical strength value of consumption vehicle is big), such as temperature are 0.5.

Table 2

…	…	…	…	…	…
						…	0.95	0.95	0.95	0.9	…
…	0.95	1	0.5	0.9	…
						…	0.5	0.95	0.9	0.9	…
…	…	…	…	…	…

It generates vehicle and takes control Thermometer by forceIt is secondary to take control point by force for certain point (or hexagonal grid) P (x, y) on map For O_second(x_second, y_second), then the vehicle secondary control temperature of the point may be expressed as:

It generates vehicle and takes control Thermometer by forceFor certain point (or hexagonal grid) P (x, y) on map, the point is found out Master take control temperature by forceTake control temperature by force with secondaryAfterwards, primary and secondary is taken control temperature by force and be may be expressed as:

It generates soldier master and takes control Thermometer by forceSoldier advance when it is preferable to terrain adaptability, ask take by force control temperature when Do not consider orographic factor.For certain point (or hexagonal grid) P (x, y) on map, main control point of taking by force is O_first(x_first, y_first), Then soldier's master control temperature of the point may be expressed as:

It generates soldier time and takes control Thermometer by forceIt is secondary to take control point by force for certain point (or hexagonal grid) P (x, y) on map For O_second(x_second, y_second), then soldier's secondary control temperature of the point may be expressed as:

It generates soldier time and takes control Thermometer by forceFor certain point (or hexagonal grid) P (x, y) on map, the point is found out Master take control temperature by forceTake control temperature by force with secondaryAfterwards, primary and secondary is taken control temperature by force and be may be expressed as:

Generate my hostile observation table C_E2O[n, i, j] and I to enemy observe table C_O2E[n, i, j]).With our chess piece type n and Position Number i is line index, and enemy chess piece Position Number i is column index, corresponding in table if meeting my hostile observation condition Crosspoint content is 1, is otherwise 0.All map hexagonal grids are traversed, my hostile observation table C is calculated_E2O[n, i, j].Table 3 is enemy The part of table is observed me, and first is classified as chess piece type, and 1 represents vehicle, and enemy mover (vehicle or soldier) is in number 10025 Not it is observed that our vehicle, enemy mover are observed that our vehicle in 10037 hexagonal grid of number in hexagonal grid.

Table 3

bop_type	grid_pos	10025	10027	10029	10031	10033	10035	10037
									1	40022	0	0	0	0	0	0	1
1	40024	0	0	0	0	0	1	1
									1	40026	0	0	1	1	1	1	1

Using the Position Number i of enemy chess piece type n and our chess piece as line index, enemy chess piece Position Number j is column rope Draw, if meeting me to enemy's observation condition, otherwise it is 0 that the corresponding crosspoint content of table, which is 1,.All map hexagonal grids are traversed, are counted Me is calculated to enemy observation table C_O2E[n, i, j].

(2) it writes function treatment submodule, including enemy's situation processing module, takes control action module by force, fire module, gets on the bus Action module, action module of getting off, soldier's mobile module, battlebus mobile module, tank mobile module.Mentioned skill in technical solution Art frame: the soft optimization algorithm of data fusion step and Multi-attribute synthetic evaluation in data mining Fusion Module is technical principle, Correlation method will be applied in each function treatment submodule.

Enemy's situation processing module.Analysis processing is done to the situation information of war game simulation system input, mainly completes 2 functions: (1) enemy and we's strength updates: mainly according to situation information, the chess piece type and quantity that statistical updating enemy lives, and enemy's death Chess piece and quantity, the type and quantity of the visible chess piece of enemy, the sightless chess piece of enemy and quantity, chess piece that we lives and Quantity, our dead chess piece and quantity, information such as relative military strength between ourselves and the enemy, as programmed decision foundation；(2) hostile location It updates: using the technology of data fusion, by enemy's piece information in chess piece historical position probability table S [c, n, l, i] and in real time Enemy's situation information list E is merged, i.e., the hostile location probability of our observable and unmatched square chess is set as 0, can be observed Hostile location probability be set as 1, then carry out standardization processing (by map all the points location probability equal proportion convert, make 1) location probability summation is equal to, form hostile location probability matrix

Take control action module by force.It completes 2 functions: (1) successively inquiring our chess piece, sentenced according to war game primitive rule It is disconnected, it sees whether to meet and takes control condition by force；(2) meet our chess piece for taking control condition by force if it exists, generate and send the order for taking control by force.

Fire module.It completes 2 functions: (1) successively inquiring our chess piece, according to war game primitive rule and situation number Judged according to middle enemy's situation information, sees whether meet shooting condition；(2) if meeting shooting condition, the life of shooting is generated and sent It enables.

It gets on the bus action module.It completes 2 functions: (1) inquiring us with the presence or absence of taxi war game that can get on the bus.It gets on the bus Condition is chess piece after satisfaction gets on the bus primitive rule, and the position of people and battlebus is located at the list L that gets on the bus_upIt is interior；(2) exist can more than The vehicle taxi war game period of the day from 11 p.m. to 1 a.m, generates and sends order of getting on the bus.

It gets off action module.It completes 2 functions: (1) inquiring us with the presence or absence of taxi war game that can get off.It gets off dynamic The condition of work is after satisfaction gets off primitive rule, and the position of manned battlebus is located at the list L that gets off_downInterior or manned battlebus Distance takes control point by force and is less than or equal to 3；(2) there is the taxi war game period of the day from 11 p.m. to 1 a.m that can get off, generate and send order of getting off.

Soldier's mobile module.It inquires whether our soldier's chess piece meets maneuvering condition according to war game primitive rule, meets machine After dynamic condition, generates chess piece movement routine and send the order.The mobile terminal of soldier presses the soft optimization algorithm of Multi-attribute synthetic evaluation It generates.Our historical position probability in my feelings attribute of overall meritTake control temperature by force with soldier's primary and secondary in environment attribute TableTotally 2 factors.According to the soft optimization algorithm of Multi-attribute synthetic evaluation, all the points on map, soldier's mobile integrated are considered Evaluations matrix may be expressed as:

Wherein, F be soldier can maneuvering range matrix,It is soldier's historical position probability matrix (from historical position probability C is pressed in table S [c, n, l, i], tri- indexes of n, l extract),Take control temperature matrix by force for soldier's primary and secondary, is matrix dot Multiply,It is weighted factor, is according to circumstances manually set.

Then soft probability is preferred:

(x, y)=random_choice (p) (14)

Battlebus mobile module.It inquires whether our battlebus chess piece meets maneuvering condition according to war game primitive rule, meets machine After dynamic condition, generates battlebus movement routine and send the order.It is mobile that battlebus is generated by the soft optimization algorithm of Multi-attribute synthetic evaluation Terminal, the attribute of consideration are in enemy's situation attribute by enemy's observation degree matrixOur historical position probability in my feelings attributeTake control Thermometer by force with vehicle primary and secondary in environment attributeTotally 3 factors.

It generates by enemy's observation degree matrixVehicle can be indicated in point P (x, y) to enemy's observation degree and by observation degree is opposed are as follows:

Wherein i is that the hexagonal grid of P point is numbered,Real-time hostile location probability tables, c, n are generated in enemy's situation update module It is respectively color with l, chess piece title, deduces phase index,For we the position i to enemy vehicle observation matrix,For we the position i to enemy soldier's observation matrix,It is enemy to the observation matrix of our vehicle, T is matrix Transposition.All accessibility hexagonal grids are asked by enemy's observation degree, not accessibility hexagonal in our in a motor-driven stage in map Lattice are set as 1 to enemy's observation degree, and the matrix eventually formed is by enemy's observation degree matrix

According to the soft optimization algorithm of Multi-attribute synthetic evaluation, the evaluation of all the points on map, the evaluation of battlebus mobile integrated are considered Matrix may be expressed as:

Then soft probability is preferred:

(x, y)=random_choice (p) (17)

Tank mobile module.Tank has the ability of motor-driven middle shooting, and the order that tank mobile module generates contains motor-driven life It enables and fire command, this module point different situations has multiple orders to export, the tank that specific steps can refer to Fig. 4 embodiment is mobile Fire flow chart.Wherein the calculation method of the mobile terminal of tank, observed fire point is as the mobile endpoint algorithm of battlebus, this In be not described in detail.

Tank mobile shooting is divided into two kinds of situations whether there is or not the enemy's situation information being observed that according in situation information: tank without Target mobile shooting and there is target mobile shooting.When not having enemy's situation information in situation information, tank is executed to be penetrated without target movement It hits, strategy is divided into following steps:

Step 1: pressing the soft optimization algorithm of Multifactor Comprehensive Evaluation, and control temperature is taken in consideration by forceOur historical position probabilityWith by enemy observation degreeTotally 3 factors calculate the mobile terminal of tank referring to battlebus path

Step 2: patrol tank is from starting point to terminalIntermediate all possible points passed through.Based on to enemy's observation degree Calculation method, calculate these points to enemy's observation degree, the point that cannot pass through is 0 to enemy's observation degree, formation and the big observation such as map Matrix M is spent, this factor is only considered, by the soft optimization algorithm of Multifactor Comprehensive Evaluation, seeks observed fire point

Step 3: calling coordinates measurement function, the path of tank point from starting point to observed fire is generated, to war game simulation System sends motor-driven order；

Step 4: after war game simulation system feedback tank reaches observed fire point, situation information is updated, if situation information The enemy mover that middle appearance can be shot then calls shooting function to be shot at, otherwise executes step 5；

Step 5: calling coordinates measurement function, generates tank from observed fire pointTo terminalPath, And execute maneuver.

Step 1: being ranked up enemy mover by the threat degree that expertise is evaluated, and forms enemy's situation list；

Step 3: observed fire point is calculated.To the goal of selection, calculate minimum apart from the sum of tank and target range Point set, therefrom select nearest point with a distance from tank as observed fire point

Step 4: coordinates measurement function is called, observed fire point is generated toPath, execute maneuver.It arrives Up to after observed fire point, fire command is sent to war game simulation system.

Step 5: pressing the soft optimization algorithm of Multifactor Comprehensive Evaluation, and control temperature is taken in consideration by forceOur historical position probabilityWith by enemy observation degreeTotally 3 factors calculate the mobile terminal of tank referring to the mobile terminal generating process of battlebusIt adjusts With coordinates measurement function, it is generated to the path of terminal, sends motor-driven order to war game simulation system.

Tank movement is that comprehensive have target tank mobile shooting and complete without one formed after target tank mobile shooting Whole functional module, the mobile process of entire tank refer to the tank mobile shooting action flow chart of Fig. 4 embodiment.

(3) main program module is write, recited function handles submodule, realizes system function.

Main program module is the main program of this war game intelligent engine, it is controlled whole by calling each function treatment submodule Running body process realizes the function that war game is intelligently deduced.It comprises the steps of:

Step 2: situation, amendment enemy's situation are updated.The situation information of war game simulation system push is read, updates and is pushed away in engine Drill stage variable, our chess piece list, the list of enemy's chess piece, the variable for taking control three-point state information by force etc. for condition judgement.It calls Enemy's situation processing module analyzes situation data, completes enemy and we's strength and updates and enemy's real time position probability updating, generation enemy Set probability matrix in orientationAnd it is saved in database that (boarding position table, out-of-the-car position table, chess piece historical position are general in database Rate table, soldier take by force control Thermometer, vehicle take by force control Thermometer, my hostile observation table, I to enemy's observation table in data mining rank Duan Tiqian is generated, and is static data)；

Step 3: judge whether game terminates.The foundation of judgement is to deduce to have arrived ending phase or our chess piece is complete Portion is dead or enemy is over match, if terminated, jumps to step 10 eight and destroys battle environment；If do not tied Beam then jumps to next step；

Step 10 seven: tank is mobile.Tank mobile module is called, completing tank without target mobile shooting or tank has mesh Mobile shooting is marked, and motor-driven or fire command is transmitted to war game simulation system, tank mobile module jumps to after being finished Step 2；

Fig. 5 illustrates the war game intelligent engine main program flow chart of the embodiment of the present invention, is each function in dashed rectangle Module.Brief operational process can be described as: initialization battle environment first, then updates situation information and updates enemy's situation, and Judge whether game terminates afterwards, is not finished and then judges link into main actions.Each function sub-modules are successively called, us is inquired Whether all chess pieces, which can execute, the movement such as is taken control by force, shoots, gets on the bus, getting off, is motor-driven, can be executed, then generates chess piece movement simultaneously Order is executed to the transmission of war game simulation system, final program returns to situation update module, executes next circulation.Different movements Between have the difference of priority, take control movement highest priority by force, inquire first in the circulating cycle, the mobile priority of tank is minimum, is following It is finally inquired in ring.

The present invention supports several maps, and parameter can be according to artificial settings flexible modulation.This example is " cities and towns using map Settlement place ", probability temperature τ=1, not only ensure that the quality of generation strategy in this way, but also take into account the diversity of strategy.Separately Outside, in the selection that 3 classes take control Thermometer by force, primary and secondary is selected to take control Thermometer by force.The setting of major parameter is as shown in table 4.

Table 4

Claims

1. a kind of land battle war game intelligent engine, it is characterised in that: commented including war game data mining Fusion Module, with more attribute synthesis The soft optimization algorithm of valence is decision-making module, enemy's situation processing module and the multiple action modules of core；

Static data in war game data is extracted key message through data mining algorithm by the war game data mining Fusion Module Data element, basis of formation knowledge data, the dynamic data generated in real time with war game simulation system are merged, and attribute number is generated According to；According to the process of deduction and enemy and we troops situation, then using the soft optimization algorithm of Multi-attribute synthetic evaluation as the decision-making module pair of core Attribute data is carried out comprehensive operation or is judged based on war game primitive rule, finally provides the action command of chess piece, and will Order is transferred to war game simulation system, driving deduction system operation.

2. land battle war game intelligent engine as described in claim 1, it is characterised in that: at war game data mining Fusion Module data Reason process can be divided into 5 parts: (1) data select；(2) data prediction；(3) data conversion；(4) data mining；(5) number According to fusion；Data selection selects data to be treated depending on the application；Data conversion is the format for converting data to processing； Data mining excavates valuable information according to requirement of engineering, and the information of excavation has: from the situation number of real-time war game simulation system Real-time enemy's situation information is excavated according to middle, intervisibility information is excavated from regular data and environmental data and takes control temperature by force, to historical data In each stage each chess piece position it is for statistical analysis, extract historical position and get on or off the bus location information；Data fusion is By the enemy's situation data fusion in real-time situation information into historical probabilities data, so that battlefield enemy's situation can more be reflected by generating Attribute data.Finally, the data that war game data mining Fusion Module generates are classified as 3 generic attribute data:: (1) enemy's situation attribute；It is right The analysis of observable being determined property of enemy's chess piece carries out probability analysis by match historical data to not observable enemy chess piece, Real-time hostile location prediction is formed after comprehensive；After comprehensive intervisibility situation, calculate to enemy's observation degree and opposed observation degree this two A evaluation attributes；(2) my feelings attribute；By carrying out probability analysis to our chess piece position in competition data, our chess piece is formed Pre- in-position probability value, an evaluation attributes as our lower moved further position；Our chess piece is got on or off the bus location information, is used The selection of truck position above and below our chess piece；(3) environment attribute；For all positions that lower step motor-driven can arrive, by landform and work as My less advantageous is evaluated with control point distance is taken by force in front position, and control hot value is taken in formation by force, as moved further position under us One evaluation attributes.

3. land battle war game intelligent engine as described in claim 1, it is characterised in that: described soft preferably with Multi-attribute synthetic evaluation Algorithm is the decision-making module of core, proposes the soft optimization algorithm of Multi-attribute synthetic evaluation, i.e., being converted to comprehensive evaluation value can adjust The probability value of temperature is controlled, and the size according to probability carrys out selection scheme.

For the Multi-attribute synthetic evaluation of the motor-driven terminal of war game entity, enable

Wherein, U_iFor the weighted synthetical evaluation value of i-th of final position (or i-th of scheme), x_ijFor the jth of i-th of terminal A attribute value, V_j(x_ij) it is x_ijStandardization evaluation of estimate, ω_jFor the weight coefficient of j-th of attribute value.

It is preferred that quasi- are as follows:

K=random_choice (p) (3)

Wherein, p_iIt indicates by evaluation of estimate U_iThe probability value of conversion, τ > 0, referred to as " temperature factor " have the function of probability, p For by p_iThe discrete probability distribution array of composition, random_choice () are according to probability selection coordinate function, and k is last preferred Scheme serial number out；The soft optimization algorithm of Multi-attribute synthetic evaluation is integrated according to enemy's situation attribute, my feelings attribute and environment attribute Evaluation, makes each hexagonal grid on map have a comprehensive evaluation value, and the comprehensive evaluation value formation one of all hexagonal grids is comprehensive Evaluations matrix is closed, synthetic evaluation matrix is after the conversion of the probability of soft optimization algorithm, under being used as according to one hexagonal grid of probability selection Walk the destination that chess piece is advanced.

4. land battle war game intelligent engine as described in claim 1, it is characterised in that: the war game data mining Fusion Module, Using historical data and map datum as data source, data handling procedure has the generation of following data table: chess piece historical position is general Rate table S [c, n, l, i], boarding position list L_up, out-of-the-car position list L_down, take by force control Thermometer (be divided into vehicle master take by force control heat Spend tableVehicle takes control Thermometer by forceVehicle primary and secondary takes control Thermometer by forceSoldier master takes control Thermometer by forceSoldier time takes control Thermometer by forceSoldier's primary and secondary takes control Thermometer by force), observation ability table (is divided into hostile I observes table C_E2O[n, i, j], I to enemy observe table C_O2E[n, i, j]) and real-time enemy's situation information list E；The data ultimately produced It is divided into 3 seed types: (1) enemy's situation attribute；Merge chess piece historical position probability table S [c, n, l, i], real-time enemy's situation information list E With generation after observation ability table to enemy observation degree matrix M and by enemy's observation degree matrix(2) my feelings attribute；From chess piece history bit It sets and extracts our location probability of some chess piece in the current generation in probability table S [c, n, l, i] to form our historical position general Rate tableThe boarding position list L generated during direct reference data analysis mining_up, out-of-the-car position list L_down； (3) environment attribute；What is generated during direct reference data analysis mining takes control Thermometer by force.

5. land battle war game intelligent engine as claimed in claim 4, it is characterised in that: chess piece historical position probability table S [c, n, L, i] in, c is chess piece color index, and n is chess piece name index, and l is phase index, and i is chess piece location index, table content table The probability for showing that the n chess piece of c color rests on the position i on map after the l stage is how many；Boarding position list L_upRecord scholar Map hexagonal grid number when soldier gets on the bus, out-of-the-car position list L_downMap hexagonal grid number when record soldier gets off；

Vehicle master takes control Thermometer by forceGeneration method are as follows: for certain point P (x, y) on map, main control point of taking by force is O_first (x_first,y_first), then the vehicle main control temperature of the point may be expressed as:

Wherein τ_commonFor the physical strength value that vehicle needs to consume by the common landform of a lattice, τ (x, y) is that vehicle passes through P (x, y) point The physical strength value of consumption, L (P, O_first) be P point arrive mainly take by force control put lattice quantity distance,Mainly to take control point and boundary by force most Big lattice quantity distance；The vehicle main control temperature for finding out all hexagonal grids on map forms vehicle master and takes control Thermometer by force

Vehicle takes control Thermometer by forceGeneration method are as follows: for certain point P (x, y) on map, it is secondary take by force control point be O_second (x_second,y_second), then the vehicle secondary control temperature of the point may be expressed as:

Wherein τ_commonFor the physical strength value that vehicle needs to consume by the common landform of a lattice, τ (x, y) is that vehicle passes through P (x, y) point The physical strength value of consumption, L (P, O_second) be P point to it is secondary take by force control put lattice quantity distance,Take control point and boundary by force to be secondary Maximum lattice quantity distance；The vehicle secondary control temperature for finding out all hexagonal grids on map forms vehicle and takes control Thermometer by force

Vehicle takes control Thermometer by forceGeneration method are as follows: for certain point P (x, y) on map, the master for finding out the point takes by force Control temperatureTake control temperature by force with secondaryAfterwards, primary and secondary is taken control temperature by force and be may be expressed as:

Soldier master takes control Thermometer by forceGeneration method are as follows: soldier advance when it is preferable to terrain adaptability, ask take by force control heat Orographic factor is not considered when spending；For certain point P (x, y) on map, main control point of taking by force is O_first(x_first,y_first), then the point Soldier's master control temperature may be expressed as:

Wherein, L (P, O_first) be P point arrive mainly take by force control put lattice quantity distance,Mainly to take control point and boundary maximum lattice by force Quantity distance；The soldier's master control temperature for finding out all hexagonal grids on map forms soldier master and takes control Thermometer by force

Soldier time takes control Thermometer by forceGeneration method are as follows: for certain point P (x, y) on map, it is secondary take by force control point be O_second(x_second,y_second), then soldier's secondary control temperature of the point may be expressed as:

Wherein L (P, O_second) be P point to it is secondary take by force control put lattice quantity distance,Take control point and boundary maximum lattice by force to be secondary Quantity distance；The soldier's secondary control temperature for finding out all hexagonal grids on map forms soldier time and takes control Thermometer by force

Soldier time takes control Thermometer by forceGeneration method are as follows: for certain point P (x, y) on map, the master for finding out the point takes by force Control temperatureTake control temperature by force with secondaryAfterwards, primary and secondary is taken control temperature by force and be may be expressed as:

6. land battle war game intelligent engine as claimed in claim 4, it is characterised in that: the method for observation ability table generating method Are as follows: observation ability table is divided into my hostile observation table C_E2O[n, i, j], I to enemy observe table C_O2E[n,i,j])；With our chess piece class Type n and Position Number i is line index, and enemy chess piece Position Number j is column index, if meeting my hostile observation condition, table In corresponding crosspoint content be 1, be otherwise 0；All map hexagonal grids are traversed, my hostile observation table C is calculated_E2O[n,i,j]； Using the Position Number i of enemy chess piece type n and our chess piece as line index, enemy chess piece Position Number j is column index, if meeting To enemy's observation condition, then otherwise it is 0 that the corresponding crosspoint content of table, which is 1, for I；All map hexagonal grids are traversed, it is right to calculate me Enemy's observation table C_O2E[n,i,j]。

7. land battle war game intelligent engine as claimed in claim 4, it is characterised in that: enemy's situation attribute data includes to enemy's observation degree Matrix M and by enemy observation degree matrixCall enemy's situation and hostile location update module function by chess piece historical position probability first Enemy's piece information and real-time enemy's situation information list E are merged in table S [c, n, l, i], generate real-time hostile location probability TableVehicle can be indicated in point P (x, y) to enemy's observation degree and by observation degree is opposed are as follows:

Wherein i is that the hexagonal grid of P point is numbered,It is formed after real-time situation data fusion for 6 chess pieces of l stage c color Real-time hostile location probability matrix,For we the position i to enemy vehicle observation matrix,It is us in the position i To enemy soldier's observation matrix,It is enemy to the observation matrix of our vehicle, T is matrix transposition；In to map we one It is to enemy observation degree matrix M that all accessibility hexagonal grids, which seek the matrix formed to enemy's observation degree, in a motor-driven stage；To map It is to be opposed observation degree matrix that all accessibility hexagonal grids, which seek the matrix formed by enemy's observation degree, in our upper in a motor-driven stage

8. land battle war game intelligent engine as described in claim 1, it is characterised in that: the enemy's situation processing module pushes away war game The situation information for drilling system input does analysis processing, mainly completes 2 functions: (1) enemy and we's strength updates: according to situation information, The chess piece type and quantity that statistical updating enemy lives, the chess piece and quantity of enemy's death, the type sum number of the visible chess piece of enemy Amount, the sightless chess piece of enemy and quantity, the chess piece and quantity that we lives, the chess piece and quantity of our death, between ourselves and the enemy The information such as relative military strength, as programmed decision foundation；(2) hostile location update: by chess piece historical position probability table S [c, n, L, i] in enemy's piece information and real-time enemy's situation information list E merge, i.e., it is our hostile location of observable position is general Rate is set as 0, and observable hostile location probability is set as 1, then carries out standardization processing, forms hostile location probability matrix

The multiple action module is that the movement for being able to carry out chess piece each in war game simulation system is decomposed, Mei Gedong The decision of work, which executes, is used as a subprogram, calls for main program；The decision-making foundation of each movement is mainly according to war game simulation The primitive rule of system is judged, or provides objectives position according to the soft optimization algorithm of Multi-attribute synthetic evaluation；It is shared Fire takes control movement, movement of getting on the bus, movement of getting off, soldier's movement, battlebus movement, the mobile totally 7 classes movement of tank by force.

9. a kind of operation method of land battle war game intelligent engine, which is characterized in that the operational process is war game intelligent engine Main program flow comprising the steps of:

Step 1: initialization battle environment；It is communicated with war game simulation system, clearly deduces and start, and initialize intelligent soldier Variable in chess engine；

Step 2: situation, amendment enemy's situation are updated；The situation information of war game simulation system push is read, updates in engine and deduces rank Segment variable, our chess piece list, the list of enemy's chess piece, the variable for taking control three-point state information by force etc. for condition judgement.Call enemy's situation Processing module analyzes situation data, completes enemy and we's strength and updates and enemy's real time position probability updating, generation enemy position Set probability matrixAnd it is saved in database；

Whether step 3: judging to deduce terminates；The foundation of judgement is to deduce to have arrived ending phase or our chess piece is all dead It dies or enemy is over match, if terminated, jump to step 10 eight and destroy battle environment；If it has not ended, then Jump to next step；

Step 4: judge whether that control can be taken by force；According to rule of inference, each chess piece is successively inquired, judges whether we arrives chess piece Take control movement by force up to taking control point by force and can execute, if control can be taken by force, jumps to step 5 execution and take control movement by force；If discontented Foot takes control condition by force, then jumps to step 6；

Step 5: control is taken in execution by force；The order taking control by force and acting is sent to war game simulation system；War game simulation system feedback order executes After success, step 2 is jumped to；

Step 6: judge whether to get on the bus；Each soldier's chess piece is successively inquired, judges whether our soldier meets the item into battlebus Part, if can get on the bus, otherwise jump procedure seven jumps to step 8；

Step 7: execution is got on the bus；The order that soldier gets on the bus is sent to war game simulation system；War game simulation system feedback order executes After success, step 2 is jumped to；

Step 8: judge whether to get off；Each soldier's chess piece is successively inquired, judges whether our soldier meets the condition got off, If can get off, step 9 is jumped to, otherwise, jumps to step 10；

Step 9: execution is got off；The order that soldier gets off is sent to war game simulation system；War game simulation system feedback order executes After success, step 2 is jumped to；

Step 10: judge whether to shoot；Each chess piece is successively inquired, judges whether chess piece meets shooting condition, if can penetrate It hits, then jumps to step 11, otherwise jump to step 12；

Step 11: shooting is executed；Chess piece fire command is sent to war game simulation system；War game simulation system feedback order executes After success, step 2 is jumped to；

Step 12: judge whether soldier can be motor-driven；Successively whether inquiry soldier can be motor-driven, if can be motor-driven, jumps To step 13, step 14 is otherwise jumped to；

Step 13: soldier is mobile；Soldier's mobile module is called, is transmitted to war game simulation system after generating the order of soldier's movement routine System jumps to step 2 after deduction system feedback execution is ordered successfully；

Step 14: judge that can battlebus motor-driven；Successively whether inquiry battlebus can be motor-driven, if can be motor-driven, jumps to Otherwise step 15 jumps to step 10 six；

Step 15: battlebus is motor-driven；Battlebus mobile module is called, is transmitted to war game simulation system after generating the order of battlebus movement routine System jumps to step 2 after deduction system feedback execution is ordered successfully；

Step 10 six: judge that can tank motor-driven；Successively whether inquiry tank can be motor-driven, if can be motor-driven, jumps to Step 10 seven, otherwise jumps to step 2；

Step 10 seven: tank is mobile；Tank mobile module is called, completing tank without target shooting or tank has target shooting, And motor-driven or fire command is transmitted to war game simulation system, tank mobile module jumps to step 2 after being finished；

Step 10 eight: battle environment is destroyed；Clearly deducing with war game simulation system communication terminates, and discharges the memory of engine occupancy Resource.