A model calling method and system based on digital twin city

Abstract

The present application provides a model calling method and system based on digital twin cities, which belongs to the field of digital twin cities and is used to solve the problem in related technologies that the calling of digital twin models cannot meet the user's needs for efficient and personalized use of digital twin cities. The method and system can combine the historical calling data of the model to accurately analyze and predict the model access needs of visiting users, and determine the calling strategy based on the calling priority of the digital twin model obtained by analysis, so that the calling strategy meets the model access needs based on user behavior and demand prediction, which is conducive to improving the user access experience.

Description

Technical Field

[0001] The present application relates to the field of digital twin cities, and in particular to a model calling method and system based on digital twin cities. Background Art

[0002] With the rapid development of information technology, digital twin cities, as an emerging urban development concept and technical system, are gradually becoming an important tool for urban planning, management and operation. By building a digital model corresponding to the real city, digital twin cities can reflect the operation status of the city in real time, provide decision-making support for city managers, and provide more convenient services for residents.

[0003] In the application of digital twin cities, the efficiency and accuracy of model calling have a crucial impact on user experience and system performance. Traditional digital twin model calling methods often lack in-depth analysis of user behavior and needs. When loading the model, they do not fully consider the areas and models that users may be concerned about, resulting in the loaded model not matching the actual needs of users. This not only wastes a lot of computing resources and network bandwidth, but may also cause users to experience problems such as freezes and slow loading during use, seriously affecting the user experience.

[0004] At the same time, existing model calling methods usually use fixed rules to determine the loading order and priority of models, without taking into account the behavioral differences and personalized needs of different users. Different users have different ways of using and focusing on digital twin models due to their age, occupation, interest preferences and other factors. For example, urban planners may pay more attention to the overall layout and infrastructure construction of the city, while ordinary residents may pay more attention to surrounding living facilities and traffic conditions. Therefore, how to accurately predict users' access needs to digital twin models based on their personalized needs and behavioral characteristics, and reasonably determine the loading order and priority of models, is an urgent problem to be solved in the current field of digital twin cities.

[0005] In addition, in practical applications, the amount of data in digital twin models is often very large. How to efficiently store and manage these models under limited computing resources and network bandwidth conditions is also a key issue that needs to be solved. Traditional model storage and management methods cannot meet the real-time and high-efficiency requirements of digital twin cities for model calls, which can easily lead to model call delays and affect the overall performance of the system.

[0006] In summary, the existing digital twin model calling methods and systems have many shortcomings and cannot meet users' needs for efficient and personalized use of digital twin cities. Summary of the invention

[0007] The present application provides a model calling method and system based on a digital twin city, which can accurately predict model access requirements according to user behavior and needs, and reasonably determine the model loading order and priority, so as to improve the user access experience.

[0008] In a first aspect, the present application provides a model calling method based on a digital twin city. The method comprises:

[0009] Acquire historical model call data based on the trigger of the access instruction of the access user, wherein the historical model call data carries a timestamp, a user identifier, a called digital twin model, and operation behavior information on the digital twin model;

[0010] Determine, in combination with the historical call data of the model, other users whose user similarity to the accessing user is higher than a similarity threshold as similar users;

[0011] Determining call priority range data of each digital twin model based on model history call data carrying user identifiers of the accessing user and similar users;

[0012] Determining the calling priority data of each digital twin model in the calling priority range data based on the model historical calling data carrying the user identification of the accessing user;

[0013] The calling strategy information of the digital twin model is determined according to the calling priority data.

[0014] By adopting the above technical solution, it is possible to combine the historical call data of the model to accurately analyze and predict the model access needs of the accessing users, and determine the calling strategy based on the calling priority of the digital twin model obtained by analysis, so that the calling strategy meets the model access needs based on user behavior and demand prediction, which is conducive to improving the user access experience.

[0015] Further, the determining, in combination with the historical model call data, other users whose user similarity to the accessing user is higher than a similarity threshold as similar users includes:

[0016] Calculate the user feature similarity between the accessing user and each other user based on the user feature tag carried by the user identifier;

[0017] Analyze the user behavior similarity between the accessing user and each other user based on the historical call data of the model;

[0018] Determine the comprehensive similarity between the accessing user and each other user according to the user feature similarity and the user behavior similarity;

[0019] Other users whose comprehensive similarity is higher than a preset similarity threshold are determined as the similar users.

[0020] Further, the analyzing the user behavior similarity between the accessing user and each other user based on the model history call data includes:

[0021] Determine user behavior characteristics of the accessing user and each other user based on the historical call record of the model, wherein the user behavior characteristics represent a feature vector of the digital twin model called with a timestamp and the operation behavior information on the digital twin model;

[0022] For each other user, let their user behavior similarity be ,but , where and Respectively represent the access user and other users in Dimension value on the behavioral characteristic dimension.

[0023] Further, determining the comprehensive similarity between the accessing user and each other user according to the user feature similarity and the user behavior similarity includes:

[0024] For each other user, let their user feature similarity be , the user behavior similarity is The comprehensive similarity is ,but , where and All are preset calculation weights greater than zero.

[0025] Further, the determining of the calling priority range data of each digital twin model based on the model historical calling data carrying the user identification of the accessing user and similar users includes:

[0026] Constructing model history call data carrying user identifiers of the accessing user and similar users as a first call data set;

[0027] Assume that the jth digital twin model is , the first weighted access frequency of the jth digital twin model determined based on the first call data set is ,but , where The user weight for the preset user ID, The number of times the user identifies the jth digital twin model, is the model weight of the preset j-th digital twin model;

[0028] Calculate the frequency average of weighted visit frequency and frequency standard deviation , determine the weighted access frequency range as , where are all positive integers;

[0029] Mapping the weighted access frequency ranges to determine priority range data , the priority range data value is positively correlated with the weighted access frequency range value;

[0030] The determining of the calling priority data of each digital twin model in the calling priority range data based on the model historical calling data carrying the user identification of the accessing user includes:

[0031] Constructing model history call data carrying the user identifier of the accessing user as a second call data set;

[0032] The second weighted access frequency of the j-th digital twin model is determined based on the second call data set: ,but , where The user weight for the preset user ID, The number of times the user identifies the jth digital twin model, is the model weight of the preset j-th digital twin model;

[0033] Calculate the frequency average of weighted visit frequency and frequency standard deviation , the calling priority data of the digital twin model is ,but

[0034] ;

[0035] ;

[0036] ;

[0037] In the formula, To calculate the calling priority, is an intermediate variable.

[0038] In the second aspect, the present application provides a model calling system based on digital twin cities. The system includes: a data acquisition module, a similarity recognition module, a range determination module, a data processing module and a strategy determination module;

[0039] The data acquisition module is used to acquire model history call data based on the triggering of the access instruction of the access user, and the model history call data carries a timestamp, a user identifier, a called digital twin model, and operation behavior information on the digital twin model;

[0040] The similarity identification module is used to determine other users whose user similarity with the accessing user is higher than a similarity threshold as similar users in combination with the historical call data of the model;

[0041] The range determination module is used to determine the call priority range data of each digital twin model based on the model history call data carrying the user identification of the access user and similar users;

[0042] The data processing module is used to determine the calling priority data of each digital twin model in the calling priority range data based on the model historical calling data carrying the user identification of the accessing user;

[0043] The strategy determination module is used to determine the calling strategy information of the digital twin model according to the calling priority data.

[0044] Furthermore, the similarity identification module is further configured to determine that other users whose user similarity with the accessing user is higher than a similarity threshold as similar users in combination with the model history call data include:

[0045] Calculate the user feature similarity between the accessing user and each other user based on the user feature tag carried by the user identifier;

[0046] Analyze the user behavior similarity between the accessing user and each other user based on the historical call data of the model;

[0047] Determine the comprehensive similarity between the accessing user and each other user according to the user feature similarity and the user behavior similarity;

[0048] Other users whose comprehensive similarity is higher than a preset similarity threshold are determined as the similar users.

[0049] Furthermore, the similarity identification module is further configured such that the analyzing the user behavior similarity between the accessing user and each other user based on the model history call data includes:

[0050] Determine user behavior characteristics of the accessing user and each other user based on the historical call record of the model, wherein the user behavior characteristics represent a feature vector of the digital twin model called with a timestamp and the operation behavior information on the digital twin model;

[0051] For each other user, let their user behavior similarity be ,but , where and Respectively represent the access user and other users in Dimension value on the behavioral characteristic dimension.

[0052] Furthermore, the similarity identification module is further configured to determine the comprehensive similarity between the accessing user and each other user according to the user feature similarity and the user behavior similarity, including:

[0053] For each other user, let their user feature similarity be , the user behavior similarity is The comprehensive similarity is ,but , where and All are preset calculation weights greater than zero.

[0054] Furthermore, the range determination module is further configured to determine the call priority range data of each digital twin model based on the model history call data carrying the user identification of the access user and similar users, including:

[0055] Constructing model history call data carrying user identifiers of the accessing user and similar users as a first call data set;

[0056] Assume that the jth digital twin model is , the first weighted access frequency of the jth digital twin model determined based on the first call data set is ,but , where The user weight for the preset user ID, The number of times the user identifies the jth digital twin model, is the model weight of the preset j-th digital twin model

[0057] Calculate the frequency average of weighted visit frequency and frequency standard deviation , determine the weighted access frequency range as , where are all positive integers;

[0058] Mapping the weighted access frequency ranges to determine priority range data , the priority range data value is positively correlated with the weighted access frequency range value;

[0059] The data processing module is further configured to determine the calling priority data of each digital twin model in the calling priority range data based on the model historical calling data carrying the user identification of the accessing user, including:

[0060] Constructing model history call data carrying the user identifier of the accessing user as a second call data set;

[0061] The second weighted access frequency of the j-th digital twin model is determined based on the second call data set: ,but , where The user weight for the preset user ID, The number of times the user identifies the jth digital twin model, is the model weight of the preset j-th digital twin model;

[0062] Calculate the frequency average of weighted visit frequency and frequency standard deviation , the calling priority data of the digital twin model is ,but

[0063] ;

[0064] ;

[0065] ;

[0066] In the formula, To calculate the calling priority, is an intermediate variable.

[0067] In summary, this application at least has the following beneficial effects:

[0068] A model calling method and system based on digital twin city are provided, which can accurately predict the calling priority data of accessing users for each digital twin model in combination with the historical calling data of the model, so that the determined calling strategy information can meet the personalized needs of users and improve the user access experience.

[0069] It should be understood that the contents described in the Summary of the Invention are not intended to limit the key or important features of the embodiments of the present application, nor are they intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] The above and other features, advantages and aspects of the embodiments of the present application will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, wherein:

[0071] Figure 1 A flow chart of a model calling method based on a digital twin city in an embodiment of the present application is shown;

[0072] Figure 2 A block diagram of a model calling system based on a digital twin city in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0073] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0074] In addition, the term "and / or" in this article is only a description of the association relationship between the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0075] The present application provides a model calling method and system based on a digital twin city, which can accurately predict model access requirements according to user behavior and needs, and reasonably determine the model loading order and priority, so as to improve the user access experience.

[0076] In the first aspect, the embodiment of the present application discloses a model calling method based on a digital twin city. The method can be executed by a local server. In a specific application scenario, all digital twin models of the digital twin city are configured on a cloud server, and the local server is connected to the cloud server in communication. The local server is configured with a basic framework model of the digital twin city, and the digital twin model is preloaded and called on the local server for fast calling and presentation. The calling method described in the embodiment of the present application may refer to a calling strategy for calling the digital twin model from the cloud server to the local server.

[0077] Figure 1 A flowchart of a model calling method based on a digital twin city in an embodiment of the present application is shown.

[0078] Reference Figure 1 , the method specifically comprises the following steps:

[0079] S110: Acquire historical model call data based on the trigger of the access instruction of the access user.

[0080] The model historical call data refers to the historical call records of all digital twin models of the digital twin city for all users. It reflects when and by which user each digital twin model was called and what specific operations were performed. Therefore, the model historical call data carries a timestamp (T), user identifier (UID), the called digital twin model (M) and the operation behavior information of the digital twin model (such as zooming in, viewing details, etc.).

[0081] S120: Determine, in combination with the historical model call data, other users whose user similarity to the accessing user is higher than a similarity threshold as similar users.

[0082] In the method of this step, each user identifier carries multiple user feature tags, such as age, occupation, interest preferences, etc. These user feature tags can be obtained from multiple channels such as user registration information and browsing history. The user feature tags can be constructed into high-dimensional feature vectors to represent the user's feature form. The timestamp in the model history call record, the called digital twin model, and the operation behavior information on the digital twin model can be constructed as user behavior features. The user behavior features are expressed as high-dimensional feature vectors, which can express the number of times the user calls the digital twin model at a specific time and under a specific operation. In this way, the model history call data of each user's user identifier can be constructed into user behavior features in the form of high-dimensional vectors.

[0083] Based on this, the method of this step specifically includes: calculating the user feature similarity between the visiting user and each other user based on the user feature tag carried by the user identifier; analyzing the user behavior similarity between the visiting user and each other user based on the model history call data; determining the comprehensive similarity between the visiting user and each other user based on the user feature similarity and the user behavior similarity; and determining other users whose comprehensive similarity is higher than a preset similarity threshold as the similar users.

[0084] In the method of this step, the similarity of the user characteristics of the accessing user and other users is determined by the similarity of the high-dimensional feature vector representing the user form. In one example, the cosine similarity formula is used to calculate the similarity of the user characteristic labels. Assuming that user feature labels are represented by vectors, and Respectively represent the access user and other users in The value of a feature label dimension (such as age value, occupation classification code, etc.): .

[0085] In the method of this step, the user behavior similarity analysis of the accessing user and each other user based on the model history call data includes: determining the user behavior characteristics of the accessing user and each other user based on the model history call record, the user behavior characteristics representing the feature vector of the digital twin model with the timestamp call and the operation behavior information of the digital twin model; for each other user, set its user behavior similarity to ,but , where and Respectively represent the access user and other users in Dimension value on the behavioral characteristic dimension.

[0086] In a specific example, determining the comprehensive similarity between the access user and each other user according to the user feature similarity and the user behavior similarity includes: for each other user, setting the user feature similarity to , the user behavior similarity is The comprehensive similarity is ,but , where and All are preset calculation weights greater than zero.

[0087] In another example, weights are assigned to behavioral similarity and feature label similarity. and , calculate the comprehensive similarity ( ): .

[0088] S130: Determine the calling priority range data of each digital twin model based on the model history calling data carrying the user identifications of the accessing user and similar users.

[0089] The method of this step includes: constructing the model history call data carrying the user identification of the access user and similar users as the first call data set; assuming that the j-th digital twin model is , the first weighted access frequency of the jth digital twin model determined based on the first call data set is ,but , where The user weight for the preset user ID, The number of times the user identifies the jth digital twin model, is the model weight of the preset j-th digital twin model; calculates the average frequency of weighted access frequency and frequency standard deviation , determine the weighted access frequency range as , where are all positive integers; the weighted access frequency range is mapped to determine the priority range data , the priority range data value is positively correlated with the weighted access frequency range value. In the embodiment of the present application, The specific priority range data may be 0-100, and the specific mapping means may be linear mapping or non-linear mapping, which is not specifically limited here.

[0090] S140: Determine the calling priority data of each digital twin model in the calling priority range data based on the model history calling data carrying the user identification of the accessing user.

[0091] The method of this step includes: constructing model history call data carrying the user identifier of the access user as a second call data set; determining the second weighted access frequency of the j-th digital twin model based on the second call data set as ,but , where The user weight for the preset user ID, The number of times the user identifies the jth digital twin model, is the model weight of the preset j-th digital twin model; calculates the average frequency of weighted access frequency and frequency standard deviation , the calling priority data of the digital twin model is ,but

[0092] ;

[0093] ;

[0094] ;

[0095] In the formula, To calculate the calling priority, is an intermediate variable.

[0096] Of course, linear interpolation algorithms or other algorithms may also be used, which are not listed here one by one.

[0097] S150: Determine the calling strategy information of the digital twin model according to the calling priority data.

[0098] After determining the calling priority data of each digital twin model for accessing users, the digital twin models to be preloaded and called to the local server can be determined based on the calling priority data, so as to improve the loading and calling speed of the digital twin models when users access them, which is conducive to improving the user access experience.

[0099] A specific calling strategy information may be: loading and calling the digital twin model with higher calling priority data to the local server until the storage space of the local server available for loading and calling is full.

[0100] Another specific calling strategy information can be: marking the digital twin model with the highest calling priority data as the model to be loaded and called, and marking the digital twin model whose spatial distance to the model to be loaded and called within the digital twin city is less than a preset distance as a non-loaded and called model, and repeating this method until the size of all the models to be loaded and called fills the storage space available for loading and calling.

[0101] In summary, this method can combine the historical call data of the model to accurately analyze and predict the model access needs of the accessing users, and determine the call strategy based on the call priority of the digital twin model obtained by analysis, so that the call strategy meets the model access needs based on user behavior and demand prediction, which is conducive to improving the user access experience.

[0102] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the embodiments of the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present application.

[0103] In a second aspect, the embodiment of the present application discloses a model calling system based on a digital twin city. The system can be included in a local server or implemented as a local server.

[0104] Figure 2 A block diagram of a model calling system based on a digital twin city in an embodiment of the present application is shown.

[0105] The system includes: a data acquisition module 210, a similarity identification module 220, a range determination module 230, a data processing module 240 and a strategy determination module 250;

[0106] The data acquisition module 210 is used to acquire model history call data based on the triggering of the access instruction of the access user, and the model history call data carries a timestamp, a user identifier, a called digital twin model, and operation behavior information on the digital twin model;

[0107] The similarity identification module 220 is used to determine other users whose user similarity with the accessing user is higher than a similarity threshold as similar users in combination with the model history call data;

[0108] The range determination module 230 is used to determine the call priority range data of each digital twin model based on the model history call data carrying the user identification of the access user and similar users;

[0109] The data processing module 240 is used to determine the calling priority data of each digital twin model in the calling priority range data based on the model historical calling data carrying the user identification of the accessing user;

[0110] The strategy determination module 250 is used to determine the calling strategy information of the digital twin model according to the calling priority data.

[0111] Furthermore, the similarity identification module 220 is further configured to determine other users whose user similarity with the accessing user is higher than a similarity threshold as similar users in combination with the model history call data, including:

[0112] Calculate the user feature similarity between the accessing user and each other user based on the user feature tag carried by the user identifier;

[0113] Analyze the user behavior similarity between the accessing user and each other user based on the historical call data of the model;

[0114] Determine the comprehensive similarity between the accessing user and each other user according to the user feature similarity and the user behavior similarity;

[0115] Other users whose comprehensive similarity is higher than a preset similarity threshold are determined as the similar users.

[0116] Furthermore, the similarity identification module 220 is further configured to analyze the user behavior similarity between the accessing user and each other user based on the model history call data, including:

[0117] Determine user behavior characteristics of the accessing user and each other user based on the historical call record of the model, wherein the user behavior characteristics represent a feature vector of the digital twin model called with a timestamp and the operation behavior information on the digital twin model;

[0118] For each other user, let their user behavior similarity be ,but , where and Respectively represent the access user and other users in Dimension value on the behavioral characteristic dimension.

[0119] Furthermore, the similarity identification module 220 is further configured to determine the comprehensive similarity between the accessing user and each other user according to the user feature similarity and the user behavior similarity, including:

[0120] For each other user, let their user feature similarity be , the user behavior similarity is The comprehensive similarity is ,but , where and All are preset calculation weights greater than zero.

[0121] Furthermore, the range determination module 230 is further configured to determine the call priority range data of each digital twin model based on the model history call data carrying the user identification of the access user and similar users, including:

[0122] Constructing model history call data carrying user identifiers of the accessing user and similar users as a first call data set;

[0123] Assume that the jth digital twin model is , the first weighted access frequency of the jth digital twin model determined based on the first call data set is ,but , where The user weight for the preset user ID, The number of times the user identifies the jth digital twin model, is the model weight of the preset j-th digital twin model;

[0124] Calculate the frequency average of weighted visit frequency and frequency standard deviation , determine the weighted access frequency range as , where All are positive integers;

[0125] Mapping the weighted access frequency ranges to determine priority range data , the priority range data value is positively correlated with the weighted access frequency range value;

[0126] The data processing module 240 is further configured to determine the calling priority data of each digital twin model in the calling priority range data based on the model historical calling data carrying the user identification of the accessing user, including:

[0127] Constructing model history call data carrying the user identifier of the accessing user as a second call data set;

[0128] The second weighted access frequency of the j-th digital twin model is determined based on the second call data set: ,but , where The user weight for the preset user ID, The number of times the user identifies the jth digital twin model, is the model weight of the preset j-th digital twin model;

[0129] Calculate the frequency average of weighted visit frequency and frequency standard deviation , the calling priority data of the digital twin model is ,but

[0130] ;

[0131] ;

[0132] ;

[0133] In the formula, To calculate the calling priority, is an intermediate variable.

[0134] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the described system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0135] In summary, this application at least has the following beneficial effects:

[0136] A model calling method and system based on digital twin city are provided, which can accurately predict the calling priority data of accessing users for each digital twin model in combination with the historical calling data of the model, so that the determined calling strategy information can meet the personalized needs of users and improve the user access experience.

[0137] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the aforementioned disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

Claims

1. A model calling method based on digital twin city, characterized in that: include: Acquire historical model call data based on the trigger of the access instruction of the access user, wherein the historical model call data carries a timestamp, a user identifier, a called digital twin model, and operation behavior information on the digital twin model; Determine, in combination with the historical call data of the model, other users whose user similarity to the accessing user is higher than a similarity threshold as similar users; Determining call priority range data of each digital twin model based on model history call data carrying user identifiers of the accessing user and similar users; Determining the calling priority data of each digital twin model in the calling priority range data based on the model historical calling data carrying the user identification of the accessing user; Determining calling strategy information of the digital twin model according to the calling priority data; The determining of the calling priority data of each digital twin model in the calling priority range data based on the model historical calling data carrying the user identification of the accessing user includes: Constructing model history call data carrying the user identifier of the accessing user as a second call data set; The second weighted access frequency of the j-th digital twin model is determined based on the second call data set: ,but , where The user weight for the preset user ID, The number of times the user identifies the jth digital twin model, is the model weight of the preset j-th digital twin model; Calculate the frequency average of weighted visit frequency and frequency standard deviation , the calling priority data of the digital twin model is ,but ； ； ； In the formula, To calculate the calling priority, is the intermediate variable, The upper limit of the range of the calling priority range data, The lower limit of the calling priority range data. is the hyperbolic tangent function, and are the frequency mean and frequency standard deviation determined based on the pre-constructed first call data set, respectively.

2. The method according to claim 1, characterized in that The determining, in combination with the historical model call data, that other users whose user similarity to the accessing user is higher than a similarity threshold are similar users comprises: Calculate the user feature similarity between the accessing user and each other user based on the user feature tag carried by the user identifier; Analyze the user behavior similarity between the accessing user and each other user based on the historical call data of the model; Determine the comprehensive similarity between the accessing user and each other user according to the user feature similarity and the user behavior similarity; Other users whose comprehensive similarity is higher than a preset similarity threshold are determined as the similar users.

3. The method according to claim 2, characterized in that The analyzing the user behavior similarity between the accessing user and each other user based on the model history call data includes: Determine user behavior characteristics of the accessing user and each other user based on the historical call record of the model, wherein the user behavior characteristics represent a feature vector of the digital twin model called with a timestamp and the operation behavior information on the digital twin model; For each other user, let their user behavior similarity be ,but , where Respectively represent the access user and other users in Dimension value on the behavioral characteristic dimension.

4. The method according to claim 2, characterized in that: Determining the comprehensive similarity between the accessing user and each other user according to the user feature similarity and the user behavior similarity includes: For each other user, let their user feature similarity be , the user behavior similarity is The comprehensive similarity is ,but , where and All are preset calculation weights greater than zero.

5. The method according to claim 2, characterized in that: The method of determining the calling priority range data of each digital twin model based on the model history calling data carrying the user identification of the accessing user and similar users includes: Constructing model history call data carrying user identifiers of the accessing user and similar users as a first call data set; Assume that the jth digital twin model is , the first weighted access frequency of the jth digital twin model determined based on the first call data set is ,but , where The user weight for the preset user ID, The number of times the user identifies the jth digital twin model, is the model weight of the preset j-th digital twin model; Calculate the frequency average of weighted visit frequency and frequency standard deviation , determine the weighted access frequency range as , where All are positive integers; Mapping the weighted access frequency ranges to determine priority range data , the priority range data value is positively correlated with the weighted access frequency range value.

6. A model calling system based on digital twin city, characterized in that: It includes a data acquisition module (210), a similarity identification module (220), a range determination module (230), a data processing module (240) and a strategy determination module (250); The data acquisition module (210) is used to acquire model history call data based on the triggering of an access instruction of an access user, wherein the model history call data carries a timestamp, a user identifier, a called digital twin model, and operation behavior information on the digital twin model; The similarity identification module (220) is used to determine other users whose user similarity with the accessing user is higher than a similarity threshold as similar users in combination with the model history call data; The range determination module (230) is used to determine the call priority range data of each digital twin model based on the model history call data carrying the user identification of the access user and similar users; The data processing module (240) is used to determine the calling priority data of each digital twin model in the calling priority range data based on the model historical calling data carrying the user identification of the accessing user; The strategy determination module (250) is used to determine the calling strategy information of the digital twin model according to the calling priority data; The data processing module (240) is further configured to determine the calling priority data of each digital twin model in the calling priority range data based on the model historical calling data carrying the user identification of the accessing user, including: Constructing model history call data carrying the user identifier of the accessing user as a second call data set; The second weighted access frequency of the j-th digital twin model is determined based on the second call data set: ,but , where The user weight for the preset user ID, The number of times the user identifies the jth digital twin model, is the model weight of the preset j-th digital twin model; Calculate the frequency average of weighted visit frequency and frequency standard deviation , the calling priority data of the digital twin model is ,but ； ； ； In the formula, To calculate the calling priority, is the intermediate variable, The upper limit of the range of the calling priority range data, The lower limit of the calling priority range data. is the hyperbolic tangent function, and are the frequency mean and frequency standard deviation determined based on the pre-constructed first call data set, respectively.

7. The system according to claim 6, characterized in that The similarity identification module (220) is further configured to determine, in combination with the model history call data, that other users whose user similarity with the accessing user is higher than a similarity threshold are similar users, including: Calculate the user feature similarity between the accessing user and each other user based on the user feature tag carried by the user identifier; Analyze the user behavior similarity between the accessing user and each other user based on the historical call data of the model; Determine the comprehensive similarity between the accessing user and each other user according to the user feature similarity and the user behavior similarity; Other users whose comprehensive similarity is higher than a preset similarity threshold are determined as the similar users.

8. The system according to claim 7, characterized in that The similarity identification module (220) is further configured to analyze the user behavior similarity between the accessing user and each other user based on the model history call data, including: Determine user behavior characteristics of the accessing user and each other user based on the historical call record of the model, wherein the user behavior characteristics represent a feature vector of the digital twin model called with a timestamp and the operation behavior information on the digital twin model; For each other user, let their user behavior similarity be ,but , where Respectively represent the access user and other users in Dimension value on the behavioral characteristic dimension.

9. The system according to claim 7, characterized in that The similarity identification module (220) is further configured to determine the comprehensive similarity between the accessing user and each other user according to the user feature similarity and the user behavior similarity, including: For each other user, let their user feature similarity be , the user behavior similarity is The comprehensive similarity is ,but , where and All are preset calculation weights greater than zero.

10. The system according to claim 7, characterized in that The range determination module (230) is further configured to determine the call priority range data of each digital twin model based on the model history call data carrying the user identification of the access user and similar users, including: Constructing model history call data carrying user identifiers of the accessing user and similar users as a first call data set; Assume that the jth digital twin model is , the first weighted access frequency of the jth digital twin model determined based on the first call data set is ,but , where The user weight for the preset user ID, The number of times the user identifies the jth digital twin model, is the model weight of the preset j-th digital twin model; Calculate the frequency average of weighted visit frequency and frequency standard deviation , determine the weighted access frequency range as , where All are positive integers; Mapping the weighted access frequency ranges to determine priority range data , the priority range data value is positively correlated with the weighted access frequency range value.

Images