A crowd decision-making behavior simulation method for a virtual town scene

Abstract

The present invention provides a crowd decision behavior simulation method for a virtual town scene. The method includes: individuals in the virtual town use the simulation parameters, decision utility and decision factors of the crowd decision behavior simulation model to calculate the utility of the quick decision mechanism for all target points, and obtain the quick decision mechanism utility value of the default utility and advertising utility for each target point. If the quick decision mechanism utility value of a target point exceeds the dynamic quick decision mechanism threshold, the individual decides to select the target point; otherwise, pedestrians calculate the rational decision response for all target points to obtain the rational decision response value of each target point. If the rational decision response value of a target point exceeds the rational decision mechanism threshold, the individual decides to select the target point. The present invention designs an overall decision-making process suitable for virtual towns, which helps operators of virtual towns to reasonably plan the types and locations of target points.

Description

Technical Field

[0001] The present invention relates to the technical field of crowd decision-making behavior simulation, and in particular to a crowd decision-making behavior simulation method for a virtual town scene. Background Art

[0002] The crowd decision-making behavior model mainly provides support for the strategic behavior of the crowd. It is at the upper level of the intelligent body's movement behavior and is a prerequisite for the crowd's movement behavior. It plays a very important role in the simulation effect of the entire crowd. In the emergency crowd evacuation scenario, pedestrians will choose different escape exits as the surrounding environmental factors change, which will have an important impact on the overall state of the crowd (flow direction or evacuation time). In the business analysis scenario, the crowd chooses different stores in the business district based on personal factors (gender, age, identity) and environmental factors (advertising, coupons, crowd density), which is of great reference value for town operators to make correct decisions.

[0003] Decision-making in consumer purchase decision theory is an important part of crowd behavior simulation. In crowd behavior simulation, establishing an accurate decision-making model is the key to studying user purchase behavior. Faced with many possible goals, how do decision makers make up their minds to make judgments and choose a solution that meets their needs among many solutions? Some decision-making theories are now very mature.

[0004] At present, the classic decision-making frameworks include: EBK model, Howard-Sheth model, etc. The basic framework is mainly based on stimulation-evaluation-feedback. The EBK model divides decision-making into five stages, namely problem cognition, information collection, solution evaluation, solution selection, and result feedback. This process may also be affected by other factors, such as: external culture, reference groups, family influences, and personal internal motivation, personality type, demographic variables, etc. The Howard-Sheth model divides decision-making into four stages: stimulation or input factors (input variables), external factors, internal factors (internal processes), and reflection or output factors. The Howard-Sheth model believes that input factors and external factors are stimuli for purchases. It arouses and forms motivations, provides various selection plan information, affects the psychological activities of buyers (internal factors), and forms a series of mediating factors for purchase decisions together with external factors, such as selection evaluation criteria, intentions, etc. This tendency or attitude is combined with other factors, such as the limiting factors of purchase behavior, to produce purchase results. The feeling information formed by the purchase results will also be fed back to consumers, affecting their psychology and the next purchase behavior. The current decision-making frameworks are all intended to describe consumers' purchasing behavior. However, in specific scenarios such as scenic spots and amusement parks, the decision-making framework cannot fully simulate the psychological model of decision makers. The classic framework needs to be supplemented and modified to meet the needs. Related research points include: dual-system decision-making model based on neural theory and link model of decision network.

[0005] The dual-system decision-making model based on neural theory comes from the theory proposed by Evans in 1984: There are two different processing processes for human judgment and decision-making-the heuristic process and the analytical process. In the heuristic process, people will selectively extract information related to the current decision-making problem, and then enter the next analytical processing process. In the analytical processing process, people will make decisions based on the information extracted by the heuristic process, and the judgment at this time is based on logic and rules. Later, after research by many experts, the human decision-making process can be roughly divided into two modes: intuitive and logical, which can be used to simulate the two thinking modes of the human brain, namely associative thinking and logical reasoning. This theory supplements the decision-making process thinking mode.

[0006] The link model of the decision network is based on graph theory, which uses mathematical language to describe the various relationships between related nodes. In research, the link model can be used to describe the influence between nodes. For example, the decision-making of decision-makers will be influenced by close relatives and friends, and compared with weak-relationship individuals, strong-relationship individuals can better influence the purchasing behavior of word-of-mouth recipients. This theory complements the influence of social relationships.

[0007] At present, a crowd behavior simulation scheme for a virtual town based on crowd decision-making theory in the existing technology includes: studying motivational factors from multiple perspectives such as psychological mechanisms and sociological analysis, modeling the virtual crowd in the scene, revealing the crowd's response under motivation, and simulating the dynamic changes and complexity in the real-world environment through the interaction between individual behavior and dynamic scenes.

[0008] In the field of crowd evacuation, crowd decision-making behavior models mainly refer to the crowd's choice of high-level strategies, such as the choice of escape exits in specific situations. Commonly used solutions include cellular automata and discrete choice models. By establishing intelligent agents with diverse features and establishing incentive-feedback mapping relationships, individual decisions in various situations can be simulated. For researchers, the crowd's response can be analyzed under various conditions to achieve goals such as early warning.

[0009] In the field of e-commerce, crowd decision-making behavior models mainly refer to consumers' product selection behavior, especially the selection of multiple similar products. The commonly used solution is mainly the discrete choice model, which focuses on how to use features to describe a consumer's portrait while establishing an incentive-feedback mapping. The former is often a non-data-driven model, while the latter will add data-driven methods. For researchers, they can better study the impact of sales mix strategies on the crowd, thereby increasing profits.

[0010] The disadvantages of the crowd behavior simulation scheme of a virtual town based on crowd decision-making theory in the above-mentioned prior art are that the model of the scheme is not interactive. Interactivity means that researchers cannot make any interference actions on the simulation environment during the simulation, and the agents in the simulation environment cannot make any response to the interference of the simulation personnel, which greatly limits the application scope of the simulation model.

[0011] This solution cannot be applied to more flexible scenarios, such as commercial areas, small towns and other real-life scenarios. Although the existing crowd evacuation model is based on real-life scenarios, it does not support simulation related to crowd consumption decisions. Although the online retail decision-making model can simulate crowd consumption decisions, it cannot be applied to commercial park scenarios. There is still a lack of research on crowd decision-making models in offline scenarios.

[0012] This solution cannot access deep learning models. As a model with realistic simulation potential, simulators hope to use simulation as a training environment to debug deep learning models after crowd behavior and decision-making models are established. Current simulation technology does not support such a solution. Summary of the invention

[0013] The present invention provides a crowd decision-making behavior simulation method for a virtual town scene, so as to help the operator of the virtual town to reasonably plan the types and locations of target points.

[0014] In order to achieve the above object, the present invention adopts the following technical scheme.

[0015] A crowd decision-making behavior simulation method for a virtual town scene includes:

[0016] Designing a crowd decision-making behavior simulation model for a virtual town scenario, and setting simulation parameters, decision utility, and decision factors of the crowd decision-making behavior simulation model;

[0017] Individuals in the virtual town use the simulation parameters, decision utilities and decision factors of the crowd decision behavior simulation model to calculate the utility of the quick decision mechanism for all target points, and obtain the quick decision mechanism utility value of the pedestrian's default utility and advertising utility for each target point. If the quick decision mechanism utility value of a target point exceeds the dynamic quick decision mechanism threshold, the individual decides to select the target point and move toward the target point.

[0018] If there is no target point whose quick decision mechanism utility value exceeds the dynamic quick decision mechanism threshold, the pedestrian uses the simulation parameters, decision utility and decision factors of the crowd decision behavior simulation model, and the quick decision mechanism utility value to calculate the rational decision response for all target points to obtain the pedestrian's rational decision response value for each target point. If there is a target point whose rational decision response value exceeds the rational decision mechanism threshold, the individual decides to select the target point and move towards the target point.

[0019] Preferably, the crowd decision-making behavior simulation model is designed for the virtual town scene, and the simulation parameters, decision utility and decision factors of the crowd decision-making behavior simulation model are set, including:

[0020] Based on the theory of maximum utility of discrete choice, the decision utility of virtual crowds selecting targets in the virtual town scenario is divided into four parts: default cognitive utility, social influence utility, advertising influence utility and historical feedback utility V fix , three decision factors are proposed to represent the important factors affecting utility, namely: price impact factor ω p , distance influence factor ω l and type impact factor ω t ;

[0021] The simulation parameters of the crowd decision-making behavior simulation model include scenario parameters and simulation agent parameters. The scenario parameters include: the number of target points, location coordinates, characteristics, capacity and price; the agent parameters include: price sensitivity, distance sensitivity, type tendency characteristics, experience level, income level and social relationship.

[0022] Preferably, individuals in the virtual town calculate the utility of the quick decision mechanism for all target points using the simulation parameters, decision utility and decision factors of the crowd decision behavior simulation model, and obtain the quick decision mechanism utility values ​​of the pedestrians' default utility and advertising utility for each target point, including:

[0023] When individuals in the virtual town start to choose the next target point, a quick decision-making mechanism process is carried out to collect the current scene dynamic information, which includes the distance of the target point, the current advertising situation and historical feedback. The scene dynamic information is used to calculate the quick decision-making mechanism utility U for each target point according to the following formula fast ;

[0024] The fast decision-making mechanism process is defined as follows, where all calculated results are the utility values ​​of a target point to the agent:

[0025] U fast =V I +V A +Vfix

[0026] U fast represents the utility of the quick decision-making mechanism, V I represents the default response utility, V A represents the advertising response utility, V fix represents the modified feedback utility;

[0027] Default response utility V I The calculation formula is as follows:

[0028] V I =ω p +ω l +ω t

[0029] Where V I is the default response utility value, which is determined by the price factor ω p , distance factor ω l , type factor ω t Joint decision.

[0030]

[0031]

[0032]

[0033] For the price factor ω p , where p i is the price of target point i, p ave is the average price of this type of target, K p is the price sensitivity parameter of the agent, α is the adjustment coefficient, and the default value is a fixed value;

[0034] For the distance factor ω l , d i is the distance between the target point i and the agent, K d is the distance sensitivity parameter of the agent, and K corresponds to low distance sensitivity d The value is larger, β is the adjustment coefficient, and the default value is fixed;

[0035] For the type factor ω t ,This model divides the target point type into four attributes, namely, environment, function, fatigue, and characteristics.,j represents each attribute, and t j Indicates the attribute value of the target point, k tj Indicates the agent's inclination towards this attribute.

[0036] Advertising response utility V A The calculation formula is as follows:

[0037]

[0038] w a (v i , r k , t) describes the influence of advertisement k on the decision maker at time t:

[0039]

[0040] Eff ik =sen i *stren k *Qua k

[0041]

[0042] The parameters are: Eff ik Describes the intensity of the effect produced by the advertisement, sen i Used to describe the sensitivity of decision makers to advertising. k Used to describe the intensity of advertising k, Qua k describes the quality of advertisement k, N(t) simulates the trend of advertisement utility changing over time, α i is the saturation factor of advertising effectiveness, t fade The maximum duration of advertisements set in the simulation environment;

[0043] Social Response Utility V S The calculation formula is as follows:

[0044]

[0045] where w s (v self , v k ) describes the influence of individual k on the decision maker:

[0046] w s (v self , v k )=P str (v self , v k )*R(v self )

[0047] P str (v self , v k ) describes the strength of the relationship between individuals, R(v self ) describes individual cognitive ability, ω kij is the default utility of individual k for target i factor j, ω sij is the default utility of the decision maker for target i factor j;

[0048] For feedback correction utility V fix , first set it to a fixed value. If the agent selects this target point, V fix The value will be adjusted by a simple decrease in value.

[0049] Preferably, the pedestrian calculates rational decision responses for all target points using simulation parameters, decision utilities and decision factors of the crowd decision behavior simulation model, as well as utility values ​​of the rapid decision mechanism, to obtain rational decision response values ​​of the pedestrian for each target point, including:

[0050] The pedestrian runs a rational decision-making mechanism based on the calculated default utility and advertising utility, and uses the quick decision-making mechanism utility U fast As input, calculate the rational impact utility U ration =U fast +V s , V s represents the social response utility, and obtains the pedestrian's rational decision response value for each target point;

[0051] The agent will select the target point whose rational decision response value exceeds the rational decision response value among all the target points. If there is no target point whose rational decision response value exceeds the rational decision response value, the pedestrian will directly exit the simulation.

[0052] Preferably, the method further comprises:

[0053] After the pedestrian arrives at the target point, the store congestion is judged, and the decision value of the congestion is used to determine whether to insist on selecting this target point. The formula is as follows.

[0054] v c =-5*c i +K c (5)

[0055]

[0056] For the crowding decision value v c , c i is the congestion degree of target point i, K c is the crowding sensitivity parameter of pedestrians, high sensitivity corresponds to K c Larger value, capacity i is the target point capacity, which can accommodate the maximum number of agents, population i is the number of agents in the current target point. When vc is less than the crowding decision threshold, the agent directly selects the next target point. When vc is greater than or equal to the crowding decision threshold, the agent enters the target point.

[0057] Preferably, the method further comprises:

[0058] When the number of times the pedestrian chooses to enter the target point according to the utility value of the rapid decision mechanism of the target point increases, the dynamic rapid decision mechanism threshold of the rapid decision mechanism utility system is increased;

[0059] When the number of times the pedestrian chooses to enter the target point according to the rational decision mechanism threshold increases, raising the rational decision mechanism threshold of the analysis and response system;

[0060] When the pedestrian selects a target point, the pedestrian's personal historical feedback utility V for the corresponding target point is fix The value of is reduced.

[0061] It can be seen from the technical solutions provided by the above-mentioned embodiments of the present invention that the new crowd decision-making behavior model adopted in the embodiments of the present invention is based on mature crowd decision-making behavior models and consumer purchase theories, and designs an overall decision-making process suitable for a virtual town, as well as appropriate decision-making utility, which makes the entire decision-making model more realistic, greatly increases the realism of the simulation, and helps the operators of the virtual town to reasonably plan the types and locations of target points.

[0062] Additional aspects and advantages of the present invention will be given in part in the following description, which will become obvious from the following description, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0064] Figure 1 A processing flow chart of a method for simulating crowd decision-making behavior in a virtual town scenario provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0065] The embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0066] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or coupling. The term "and / or" used herein includes any unit and all combinations of one or more associated listed items.

[0067] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those generally understood by those skilled in the art in the art to which the present invention belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with the meanings in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless defined as herein.

[0068] To facilitate understanding of the embodiments of the present invention, several specific embodiments will be further explained below with reference to the accompanying drawings, and each embodiment does not constitute a limitation on the embodiments of the present invention.

[0069] The embodiment of the present invention proposes a crowd decision-making behavior simulation method for a virtual town scenario, mathematically models various aspects of crowd behavior, including personal demand perception, external environmental incentives, and individual internal decision-making mechanisms, demonstrates rationality and correctness, and simulates and verifies crowd decision-making behavior in specific scenarios.

[0070] The embodiment of the present invention proposes a crowd decision-making behavior simulation method for a virtual town scene, which simulates the decision-making process of the crowd in public places and studies the behavioral decision-making theory based on the user's psychological activity process, including personal demand perception, external environmental incentives, and individual internal decision-making mechanisms. Ultimately, it provides a high-quality simulation model foundation for public place operators.

[0071] The method of the present invention adds interactive elements, which can achieve the purpose of interfering with the simulation process by modifying part of the model data during the simulation. The agent in the simulation environment can respond to the actions of the simulation personnel according to the decision model, thereby realizing real-time interaction.

[0072] The method of the present invention starts with a discrete choice model to model an individual's internal and external incentives. In a data-free environment, it can better characterize a person's behavioral decision-making mode from the perspectives of internal incentives, social environment, external incentives, etc., thereby increasing the scope of application of the model.

[0073] A processing flow chart of a crowd decision-making behavior simulation method for a virtual town scene provided by an embodiment of the present invention is as follows: Figure 1 As shown, the processing steps include the following:

[0074] Step S10: designing the decision utility and decision factors of the crowd decision behavior simulation model.

[0075] The present invention models various aspects of crowd decision-making behavior for virtual town scenarios, including personal demand perception, external environmental incentives, and individual internal decision-making mechanisms, and constructs a crowd decision-making behavior simulation model. With the development of society and the continuous growth of population, large-scale crowd gatherings are becoming more and more frequent, such as in public places such as commercial areas, stadiums, and tourist attractions. Building virtual towns has practical significance for a variety of scenarios. In the field of public safety, building virtual towns can be used to simulate evacuation incidents in public places to determine whether the building design and crowd guidance strategies are reasonable. In the field of business decision-making, virtual towns can be used to simulate the real-time situation of the crowd to determine whether the operator's business decision is reasonable and whether the crowd guidance plan can adjust the flow of people in the area.

[0076] Based on the theory of maximum utility of discrete choice, starting from the specific virtual town scenario, the decision utility of virtual crowd selecting targets is divided into four parts: default cognitive utility, social influence utility, advertising influence utility and historical feedback utility V fix Three decision factors are proposed to represent the important factors affecting utility, namely: price impact factor ω p , distance influence factor ω l and type impact factor ω t .

[0077] In the virtual town, the target point is a location that the crowd may reach, such as a store, a square, etc., which is a closed set. The agent refers to an individual in the crowd. The agent is the decision maker when selecting the target point, and will use utility and U maximum as the selection strategy. The default cognitive utility will be affected by various decision factors, the social influence utility will be affected by other individuals with social relations, and the advertising influence utility will be affected by external advertising incentives.

[0078] The utility of the decision-making mechanism in the model decision process is defined as follows, where all calculated results are the utility values ​​of a certain target point to the agent:

[0079] U fast=V I +V A +V fix

[0080] U ration =U fast +V s

[0081] U fast Represents the utility of the quick decision-making mechanism, corresponding to the quick decision-making mechanism part of S30. V I represents the default response utility, V A represents the advertising response utility, V fix Indicates the corrected feedback utility, corresponding to the corrected feedback part of S40. ration Represents rational impact utility, corresponding to the rational response system part of S30. V S Represents social response utility.

[0082] Next, the calculation process of each module in the model flow chart is introduced.

[0083] First is the default response V I , the default response utility calculation formula is as follows:

[0084] V I =ω p +ω l +ω t

[0085] Where V I is the default response utility value, which is determined by the price factor ω p , distance factor ω l , type factor ω t Joint decision.

[0086]

[0087]

[0088]

[0089] For the price factor ω p , where p i is the price of target point i, p ave is the average price of this type of target, K p is the price sensitivity parameter of the agent. The price sensitivity of high-income people is low, and the corresponding K p The value is larger, α is the adjustment coefficient, and the default is a fixed value.

[0090] For the distance factor ω l , d i is the distance between the target point i and the agent, Kd is the distance sensitivity parameter of the agent, and K corresponds to low distance sensitivity d The value is larger, β is the adjustment coefficient, and the default value is a fixed value.

[0091] For the type factor ω t ,This model divides the target point type into four attributes, namely, environment, function, fatigue, and characteristics.,j,represents each attribute, and t j Indicates the attribute value of the target point, k tj Indicates the agent's inclination towards this attribute.

[0092] The second is the advertising response utility V A , the calculation formula is as follows:

[0093]

[0094] w a (v i , r k , t) describes the influence of advertisement k on the decision maker at time t:

[0095]

[0096] Eff ik =sen i *stren k *Qua k

[0097]

[0098] The parameters are: Eff ik Describes the intensity of the effect produced by the advertisement, sen i Used to describe the sensitivity of decision makers to advertising. k Used to describe the intensity of advertising k, Qua k describes the quality of advertisement k, N(t) simulates the trend of advertisement utility changing over time, α i is the saturation factor of advertising effectiveness, t fade The maximum duration of advertisements set in the simulation environment.

[0099] Then there is the social response utility V S , its utility is affected by agents with social relationships, and the calculation formula is as follows:

[0100]

[0101] where w s (v self , v k ) describes the influence of individual k on the decision maker:

[0102] w s (v self , v k )=P str (v self , v k )*R(v self )

[0103] P str (v self , v k ) describes the strength of the relationship between individuals, R(v self ) describes individual cognitive ability, ω kij is the default utility of individual k for target i factor j, ω sij is the decision maker’s default utility for target i factor j.

[0104] For feedback correction utility V fix , first set it to a fixed value. If the agent selects this target point, V fix The values ​​will be adjusted by simply decreasing the value to reflect less interest and more plausibility.

[0105] Step S20: designing a decision-making process for a crowd decision-making behavior simulation model.

[0106] At present, the classic decision-making framework is very mature. The present invention integrates the classic model and designs the decision-making process of the crowd decision-making behavior simulation model in a framework based on stimulus-evaluation-feedback combined with the virtual town scene. The decision-making process is mainly divided into five parts: information collection, rapid decision-making mechanism utility, decision-making, action and feedback.

[0107] In the process of information collection, virtual people in the virtual town scene are searched, and a set of candidate target points of virtual people targeting the demand is generated.

[0108] In the utility process of the rapid decision-making mechanism, external influencing factors in the purchase decision-making process of the virtual crowd are reflected, such as product stimulation (store attributes) and symbol (advertising) stimulation from the outside world.

[0109] In the decision-making process, the virtual crowd evaluates the behavior of various possible scenarios.

[0110] During the action process, the virtual crowd moves toward the selected target, reflecting the execution behavior.

[0111] In the feedback process, the results of the virtual crowd’s purchasing actions are stored in memory to provide experience for the crowd’s future decision-making process.

[0112] Step S30: designing a decision-making mechanism for a crowd decision-making behavior simulation model.

[0113] In the decision-making process, the present invention adopts a dual-system decision-making mechanism. System 1 is a fast decision-making mechanism utility system, also known as an intuitive system. System 1 usually provides a "fast decision-making mechanism utility" for a given task or situation. This response appears quickly and intuitively. System 2 is a rational response system. When System 1 has difficulty responding, detects errors, or cannot provide a response, System 2 processes and takes over the decision. The two decision-making systems simulate the intuitive and rational decision-making processes of people respectively. System 1 provides fast but incomplete decision results, and System 2 provides detailed decision analysis.

[0114] For the system 1 quick decision-making mechanism utility system, the present invention adopts a quick decision-making mechanism and adds a response threshold to simulate the process in which personal factors directly drive the agent to make decisions without detailed analysis. For the system 2 rational response system, the virtual crowd adds social factors on the basis of the quick decision-making mechanism utility, considers all destinations, and selects the target point with the greatest utility.

[0115] Finally, in order to increase the authenticity of the crowd’s selection behavior, the model adds a secondary selection mechanism. After the agent reaches the target point, the store congestion is judged. By judging the congestion decision value, it is determined whether to insist on selecting this target point. The formula is as follows.

[0116] V c =-5*c i +K c

[0117]

[0118] For the crowding decision value V c , c i is the congestion degree of target point i, K c is the crowding sensitivity parameter of the agent, high sensitivity corresponds to K c Larger value, capacity i is the target point capacity, which can accommodate the maximum number of agents, population i is the number of agents in the current target point. When vc is less than the crowding decision threshold, the agent directly selects the next target point. When vc is greater than or equal to the crowding decision threshold, the agent enters the target point.

[0119] Step S40: design decision feedback.

[0120] The decision feedback of the overall virtual crowd is reflected in three aspects: the first is the dynamic rapid decision mechanism threshold of the rapid decision mechanism utility system. The design of this threshold can simulate the agent's excitement. The higher the threshold, the higher the rapid decision mechanism utility required to trigger the rapid decision behavior. As the number of decisions increases, the threshold will increase to reflect the process of weakening excitement. The second is the rational decision mechanism threshold of the analysis response system. The design of this threshold is used to simulate the fatigue of the agent and reflect the feedback of the number of visits. The higher the threshold, the more target points the agent has entered. As the number of times the target points are entered increases, the threshold will increase, thereby controlling the tourist's tour cycle within a reasonable range.

[0121] The third is the feedback of modified utility, which is used to simulate the agent's weakening interest in the visited target points, thereby enhancing the rationality of the agent's choice.

[0122] The overall process of the crowd decision-making behavior simulation model of the present invention is described as follows:

[0123] 1. Determine simulation parameters: simulation parameters include scenario parameters and simulation agent parameters. Scenario parameters mainly include: number of target points, location coordinates, features, capacity, price, etc. Agent parameters mainly include: price sensitivity, distance sensitivity, type tendency characteristics, experience level, income level, social relations, etc. These parameters are used in the design of S10's decision utility and become one of the input variables. The remaining input variables of the decision utility calculation will be generated during the simulation operation.

[0124] 2. Start simulation: The entire simulation process is divided into four parts according to the process of S20, namely response, decision, action and feedback. The response module is the basis of the decision module and is integrated into the decision mechanism. The action part is not concerned with this invention.

[0125] 3. Response-decision-feedback process: The present invention adopts a dual-system decision mechanism. During the simulation process, when the agent starts to select a target point (for example, the simulation starts or the next target point is selected), according to the dual-system decision mechanism, the fast decision-making process is first performed, and then the rational decision-making process is performed, in which the feedback factor has an impact.

[0126] Fast decision-making mechanism process (system 1): The agent runs the fast decision-making mechanism for all target points, collecting the current scene dynamic information as input, including the distance of the target point, the current advertising situation, historical feedback, etc., and calculates the formula U fast =V I +V A +V fix , output the sum of the default utility and advertising utility of the agent for each target point, that is, the utility of the fast decision mechanism Ufast For the utility results calculated and output in this part, if the utility calculated for a target point exceeds the threshold 1 of system 1, the target point is immediately selected. If no utility exceeds the threshold 1, the rational decision process is entered.

[0127] Rational decision-making process (system 2): The agent runs a rational decision-making mechanism based on the calculated default utility and advertising utility, and converts the quick decision-making mechanism utility U fast As input, calculate the rational impact utility U ration =U fast +V S At this time, the agent will select the target point with the largest rational decision utility among all target points and whose utility value exceeds the threshold 2 of system 2. If there is no target point, the agent will directly exit the simulation, and the first decision-making process is completed.

[0128] Feedback: In the above decision-making process, the agent's threshold 1 and threshold 2 will be adjusted. The input is the agent's decision, and the result is the threshold adjustment, which is the first two aspects described in S40.

[0129] 4. Action: The agent takes action based on the result of the first decision. The action mainly refers to going to the target point. This part is not of concern to this invention.

[0130] 5. Second decision: After the agent reaches the selected target point, it determines the store congestion and determines whether to insist on selecting this target point by judging the congestion decision value, which is the formula part of S30. The input is the current store information and the output is the decision value. If the selection is successful, the agent is considered to have entered the target point. If the selection fails, the agent immediately enters the feedback process.

[0131] 6. Feedback: This part is the third aspect described in S40. If the agent selects this target point, V in Formula 1 fix Adjustments will be made, with the adjustment process being a simple decrease in the value to reflect the decrease in interest and increase plausibility.

[0132] When the number of times the pedestrian chooses to enter the target point according to the utility value of the quick decision mechanism of the target point increases, the dynamic quick decision mechanism threshold of the quick decision mechanism utility system is increased; the change of the quick decision mechanism threshold (threshold 1) does not depend on a certain target point. As long as the agent uses the quick decision mechanism to determine the target, threshold 1 will increase, and the simulated excitement will be weakened; the change of the rational decision mechanism threshold (threshold 2) does not depend on a certain target point, simulates the fatigue of the agent, reflects the feedback of the number of visits, and regulates the tourist visit cycle through threshold 2; the change of the historical feedback Vfix depends on a certain target point. After the agent selects a certain target point, its Vfix will decrease, indicating a weakened interest in the target point.

[0133] When the number of times the pedestrian chooses to enter the target point according to the rational decision mechanism threshold increases, raising the rational decision mechanism threshold of the analysis and response system;

[0134] When the pedestrian selects a target point, the pedestrian's personal historical feedback utility V for the corresponding target point is fix The value of is reduced.

[0135] The interactivity of the above crowd decision-making behavior simulation model is as follows:

[0136] Simulators can control some parameters in the model during the simulation process, such as placing advertisements for the global simulation environment, modifying target point capacity, prices, features, setting up roadblocks, etc. Under the influence of these dynamic parameters, the agent will make different decisions, reflecting the interactivity of real-time simulation.

[0137] In summary, the new crowd decision behavior model adopted in the embodiment of the present invention is based on the mature crowd decision behavior model and consumer purchase theory, and designs an overall decision process suitable for the virtual town, as well as appropriate decision utility, introduces a dual-system decision mechanism, considers the agent's own response and external social factors and advertising factors, and finally introduces a decision feedback mechanism, making the entire decision model more realistic, greatly increasing the authenticity of the simulation, and solving the problem of low authenticity of the existing agent decision model. It helps the operators of the virtual town to reasonably plan the types and locations of target points.

[0138] The present invention can realize crowd flow simulation in a virtual environment by modeling crowd behavior decisions, which has the following beneficial effects:

[0139] 1. Crowd warning:

[0140] Based on a mature crowd decision-making behavior model, we can build a virtual town environment based on real physical world data, and use the current environmental terrain, crowd information, and crowd characteristics as model inputs. The model can deduce the subsequent distribution of crowd flow, thereby predicting the possible congestion time of crowd flow in advance, and issuing early warnings to town operators. Managers can then regulate crowd flow in advance to enhance the safety of the town.

[0141] 2. Deduction of crowd control plan:

[0142] On the basis of the present invention, before making a decision on crowd control, the town operator can use the interactivity of the model to deduce the decision action in the virtual town environment. The results of multiple deductions can serve as an important reference for the town operator's decision-making, so as to ultimately make better decisions in the real physical world. For example, when a real town is or is about to be crowded, the town operator hopes to control the flow of people in one or more ways, such as placing roadblocks, assigning crowd guides, and placing store coupons. The operator can deduce a combination of one or more decision actions in the virtual environment of the model, and the results will serve as an important reference, thereby assisting the operator to make better decision actions with the assistance of deduction, and the beneficial effects brought about include easing crowd congestion, reducing the number of volunteers in the town, and increasing the overall capacity of the town to accommodate more tourists.

[0143] 3. Town operation plan deduction:

[0144] Similar to the deduction of crowd control plans, the deduction of town operation plans can mainly bring beneficial economic effects. As the operator of a town, especially in commercial scenarios such as theme parks, operators often hope to maximize the profits of the entire town through various means, including advertising, store coupons, etc. The present invention allows operators to deduce a combination of one or more decision-making actions in the virtual environment of the model, and can also access the deep learning model, use the advantages of deep learning to train, and formulate better decision-making plans, so that operators can make better decision-making actions, so that the overall benefits of the town are optimized.

[0145] The agent-based simulation model does not have interactivity. Since the simulation model developed by the present invention supports interaction, researchers can interfere with the actions of the simulation environment, and the agents in the simulation environment respond to the actions of the input environment according to the decision model, making the simulation have real-time interactivity.

[0146] The agent decision model has a narrow scope of application without data support. Since the model of the present invention is based on a mature crowd decision-making behavior model and consumer purchase theory, and models the internal and external incentives of individuals, it can better describe people's behavioral decision-making methods from the perspectives of internal incentives, social environment, external incentives, etc. in a data-free environment, thereby increasing the scope of application of the model.

[0147] Those skilled in the art can understand that the accompanying drawings are only schematic diagrams of an embodiment, and the modules or processes in the accompanying drawings are not necessarily required to implement the present invention.

[0148] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present invention can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present invention or certain parts of the embodiments.

[0149] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0150] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A crowd decision-making behavior simulation method for a virtual town scene, characterized in that: include: Designing a crowd decision-making behavior simulation model for a virtual town scenario, and setting simulation parameters, decision utility, and decision factors of the crowd decision-making behavior simulation model; Individuals in the virtual town use the simulation parameters, decision utilities and decision factors of the crowd decision behavior simulation model to calculate the utility of the quick decision mechanism for all target points, and obtain the quick decision mechanism utility values ​​of the default utility and advertising utility of pedestrians for each target point. If the quick decision mechanism utility value of a target point exceeds the dynamic quick decision mechanism threshold, the individual decides to select the target point and move toward the target point. If there is no target point whose utility value of the quick decision mechanism exceeds the threshold value of the dynamic quick decision mechanism, the pedestrian calculates the rational decision response for all target points using the simulation parameters, decision utility and decision factor of the crowd decision behavior simulation model and the utility value of the quick decision mechanism to obtain the rational decision response value of the pedestrian for each target point. If there is a target point whose rational decision response value exceeds the threshold value of the rational decision mechanism, the individual decides to select the target point and move toward the target point. Individuals in the virtual town calculate the utility of the quick decision mechanism for all target points using the simulation parameters, decision utility and decision factors of the crowd decision behavior simulation model, and obtain the quick decision mechanism utility values ​​of the pedestrians' default utility and advertising utility for each target point, including: When individuals in the virtual town start to choose the next target point, a quick decision-making mechanism process is carried out to collect the current scene dynamic information, which includes the distance of the target point, the current advertising situation and historical feedback. The scene dynamic information is used to calculate the quick decision-making mechanism utility U for each target point according to the following formula fast ; The fast decision-making mechanism process is defined as follows, where all calculated results are the utility values ​​of a target point to the agent: U fast =V I +V A +V fix U fast represents the utility of the quick decision-making mechanism, V I represents the default response utility, V A represents the advertising response utility, V fix represents the modified feedback utility; Default response utility V I The calculation formula is as follows: V I =ω p +oh l +oh t Where V I is the default response utility value, which is determined by the price factor ω p , distance factor ω l , type factor ω t Joint decision making; For the price factor ω p , where p i is the price of target point i, p ave is the average price of this type of target, K p is the price sensitivity parameter of the agent, α is the adjustment coefficient, and the default value is a fixed value; For the distance factor ω l , d i is the distance between the target point i and the agent, K d is the distance sensitivity parameter of the agent, and K corresponds to low distance sensitivity d The value is larger, β is the adjustment coefficient, and the default value is fixed; For the type factor ω t ,This model divides the target point type into four attributes, namely, environment, function, fatigue, and characteristics.,j,represents each attribute, and t j Indicates the attribute value of the target point, k tj Indicates the agent's inclination towards this attribute; Advertising response utility V A The calculation formula is as follows: w a (v i ,r k ,t) describes the influence of advertisement k on the decision maker at time t: Eff ik =late i *train k *Qua k The parameters are: Eff ik Describes the intensity of the effect produced by the advertisement, sen i Used to describe the sensitivity of decision makers to advertising. k Used to describe the intensity of advertising k, Qua k describes the quality of advertisement k, N(t) simulates the trend of advertisement utility changing over time, α i is the saturation factor of advertising effectiveness, t fade The maximum duration of advertisements set in the simulation environment; Social Response Utility V S The calculation formula is as follows: where w s (v self ,v k ) describes the influence of individual k on the decision maker: w s (v self ,v k )=P str (v self ,v k )*R(v self ) P str (v self ,v k ) describes the strength of the relationship between individuals, R(v self ) describes individual cognitive ability, ω kij is the default utility of individual k for target i factor j, ω sij is the default utility of the decision maker for target i factor j; For feedback correction utility V fix , first set it to a fixed value. If the agent selects this target point, V fix The value will be adjusted by simply subtracting the value; The pedestrian calculates rational decision responses for all target points using simulation parameters, decision utilities and decision factors of the crowd decision behavior simulation model, as well as utility values ​​of the rapid decision mechanism, to obtain rational decision response values ​​of the pedestrian for each target point, including: The pedestrian runs a rational decision-making mechanism based on the calculated default utility and advertising utility, and uses the quick decision-making mechanism utility U fast As input, calculate the rational impact utility U ratio =U fast +V S , V S represents the social response utility, and obtains the pedestrian's rational decision response value for each target point; The agent will select the target point whose rational decision response value exceeds the rational decision response value among all the target points. If there is no target point whose rational decision response value exceeds the rational decision response value, the pedestrian will directly exit the simulation.

2. The method according to claim 1, characterized in that The crowd decision-making behavior simulation model is designed for the virtual town scene, and the simulation parameters, decision utility and decision factors of the crowd decision-making behavior simulation model are set, including: Based on the theory of maximum utility of discrete choice, the decision utility of virtual crowds selecting targets in the virtual town scenario is divided into four parts: default cognitive utility, social influence utility, advertising influence utility and historical feedback utility V fix , three decision factors are proposed to represent the important factors affecting utility, namely: price impact factor ω p , distance influence factor ω l and type impact factor ω t ; The simulation parameters of the crowd decision-making behavior simulation model include scenario parameters and simulation agent parameters. The scenario parameters include: the number of target points, location coordinates, characteristics, capacity and price; the agent parameters include: price sensitivity, distance sensitivity, type tendency characteristics, experience level, income level and social relationship.

3. The method according to any one of claims 1 to 2, characterized in that: The method further comprises: After the pedestrians arrive at the target point, the store congestion is judged, and the decision value of the congestion is used to determine whether to insist on selecting this target point. The formula is as follows: v c =―5*c i +K c (5) For the crowding decision value v c , c i is the congestion degree of target point i, K c is the crowding sensitivity parameter of pedestrians, high sensitivity corresponds to K c Larger value, capacity i is the target point capacity, which can accommodate the maximum number of agents, population i is the number of agents in the current target point. When vc is less than the crowding decision threshold, the agent directly selects the next target point. When vc is greater than or equal to the crowding decision threshold, the agent enters the target point.

4. The method according to claim 3, characterized in that: The method further comprises: When the number of times the pedestrian chooses to enter the target point according to the utility value of the rapid decision mechanism of the target point increases, the dynamic rapid decision mechanism threshold of the rapid decision mechanism utility system is increased; When the number of times the pedestrian chooses to enter the target point according to the rational decision-making mechanism threshold increases, the rational decision-making mechanism threshold is increased; When the pedestrian selects a target point, the pedestrian's personal historical feedback utility V for the corresponding target point is fix The value of is reduced.

Images