A digital twin data generation method and device based on multi-source heterogeneous data

Abstract

The present invention relates to the field of digital twin data acquisition, and in particular to a digital twin data generation method and device based on multi-source heterogeneous data. Applied to the digital twin metadata generation system, the server function attribute data acquisition step: obtain the function classification label and classification confidence corresponding to each activity vector. According to the function classification labels and classification confidences corresponding to all activity vectors, the functional attribute data of the newly connected server is determined. The data standardization platform in the present invention can perform unified standardization processing on attribute data of different formats to improve the availability of metadata provided to other digital twin construction platforms in the later stage. In addition, the functional attributes of the server are obtained through the type and quantity characteristics of the executed operations reflected in the log information. The functional attributes of the newly connected server can be determined more accurately, thereby reducing the problem of low accuracy of human judgment results.

Description

Background Art

[0002] The digital twin of an entity object, that is, by defining the data structure and attribute information of the entity object, realizes the accurate description and digital expression of various entity objects in the real world, laying a solid data foundation for the next step of the twin of the business scenario and the application of simulation and artificial intelligence technology in the digital twin scenario. Digital twin refers to the digital expression of entity objects with physical form or physical boundaries, which is an important step in digital twin. Among them, the data used to describe the digital twin includes business metadata and other metadata. Among them, the business metadata of the digital twin mainly includes three categories: common attributes, proprietary attributes and extended attributes. Among them, the common attributes include data of the dimensions of identification, time and space, geometry, ability, relationship, and state. Among them, the ability is used to describe the behavior of the entity, which is a measure of the level of world exploration, perception, cognition, change, etc. Specifically, the ability is also the function ability, which is used to represent the professional skills of the entity to exert influence, produce effects or achieve goals in a specific field or environment. For example, the function ability of a bus is to carry passengers, the function ability of a teacher is to educate, the function ability of a drug is to cure diseases, and the function ability of a dam is to regulate water flow, flood control and power generation. An entity can have one or more functions. For example, in the process of building a digital twin of the corresponding data center, in order to be able to generate a more complete and accurate digital twin of each server entity, it is necessary to collect data in the above-mentioned attribute dimensions to generate corresponding metadata.

[0003] The construction of digital twin cities involves converting a large number of real-world physical objects into corresponding digital twins. Based on the different characteristics of real-world physical objects and referring to the relevant classification of real-world physical objects, digital twins can be divided into eight categories: natural entities, management entities, buildings and artificial structures, vehicles, objects and equipment, natural persons, legal persons and other organizations, and other entities. The above-mentioned servers belong to the category of objects and equipment among the eight categories.

[0004] In the prior art, with the development of technologies such as the Internet of Things, cloud computing, big data, and artificial intelligence, it will be necessary to build multiple corresponding data centers to support the corresponding functional requirements. Usually there are a large number of servers in a data center, and different servers may perform different tasks, such as data collection tasks, video rendering tasks, storage tasks, and computing tasks. As a result, different servers correspond to different capability labels. Even a server performs multiple tasks at the same time, that is, it corresponds to multiple different capability labels at the same time. Therefore, there will be a high professional threshold and difficulty in determining the capability label corresponding to each server in the data center, especially for some newly connected servers. It is difficult to accurately determine the capability type of the server through simple manual judgment. Summary of the invention

[0005] In view of the above technical problems, the technical solution adopted by the present invention is:

[0006] According to one aspect of the present invention, a method for generating digital twin data based on multi-source heterogeneous data is provided. The method is applied to a digital twin metadata generation system. The digital twin metadata generation system includes a data standardization platform and multiple initial metadata collection platforms. The multiple initial metadata collection platforms are all connected to the data standardization platform for communication. The multiple initial metadata collection platforms are respectively used to collect attribute data corresponding to different entity objects. There are differences in the formats of attribute values ​​collected by multiple initial metadata collection platforms for the same attribute. The data standardization platform is used to perform data standardization processing on the acquired attribute data of different entities to generate metadata in a preset format.

[0007] The multiple initial metadata collection platforms include a server metadata collection platform for collecting server attribute data, and the server attribute data includes server function attribute data. The server metadata collection platform obtains the function attribute data of the newly connected server according to the following steps:

[0008] Based on the log information of multiple historical periods of the newly connected server, the server activity vector corresponding to each historical period is generated respectively. a =(A a 1 , A a 2 , …, A a i , …, A a z ), A a is the server activity vector corresponding to the a-th historical period, A a i is the total number of times the server executes the i-th type of operation in the a-th historical period, z is the total number of types of operations that can be executed during server operation, i=1, 2…z.

[0009] Each activity vector is input into the target classification model, and the functional classification label and classification confidence corresponding to each activity vector are input. The target classification model uses the classification label corresponding to the highest classification confidence as the output classification result.

[0010] According to the function classification labels and classification confidences corresponding to all activity vectors, the matching value corresponding to each type of function is generated. Among them, the matching value B corresponding to the b-th type of function in the function classification label is b The following conditions must be met:

[0011] B b =P b Y1 / P sum , where Pb Y1 The total number of classification labels corresponding to the b-th type of function whose classification confidence in the function classification label is greater than the first confidence threshold Y1. P sum The total number of function classification labels.

[0012] If B b > P1, then the classification label corresponding to the b-th type of function in the function classification label is determined as the function attribute data of the newly connected server. P1 is the first proportion threshold.

[0013] Furthermore, the steps for the server metadata collection platform to obtain the function attribute data of the newly connected server also include:

[0014] If B b ≤ P1, then perform a multi-functional type judgment process to generate at least two classification labels corresponding to the newly connected server.

[0015] The multi-functional type judgment process includes:

[0016] Generate a classification result confidence C based on the function classification labels and classification confidences corresponding to all activity vectors. C satisfies the following condition: C = P Y2 / P sum . P Y2 Is the total number of classification labels whose classification confidence in the function classification label is greater than the second confidence threshold Y2, where Y2 < Y1.

[0017] If C > P2, then obtain the classification label vector corresponding to the newly connected server. The classification label vector is formed by converting the function classification label corresponding to each activity vector into the corresponding element value in chronological order. P2 is the second proportion threshold.

[0018] Obtain the classification label vectors corresponding to other multi-functional servers belonging to the same organization as the newly connected server respectively.

[0019] If the similarity between the classification label vector corresponding to the newly connected server and the classification label vector corresponding to any other multi-functional server is greater than the preset similarity threshold, then the classification label corresponding to the multi-functional server is determined as the function attribute data of the newly connected server.

[0020] Furthermore, after generating the classification result confidence based on the function classification labels and classification confidences corresponding to all activity vectors, the multi-functional type judgment process also includes:

[0021] If C < P2, then generate the first classification confidence corresponding to each type of function classification label whose classification confidence in all activity vectors is greater than the third confidence threshold Y3. Y3 < Y2, where the first classification confidence Dn corresponding to the n-th type of function classification label whose classification confidence is greater than Y3 satisfies the following condition:

[0022] Dn = (P n ×K1) / P Y3 sum +M1.

[0023] P n is the number of n-th functional classification labels whose classification confidence is greater than Y3, P Y3 sum is the total number of functional classification labels whose classification confidence is greater than Y3, K1 is the first confidence adjustment amplitude, M1 is the first confidence base, M1∈[0.8, 0.9].

[0024] Each type of function classification label is sorted in descending order of the first classification confidence, and a first function classification recommendation list of the new access server is generated.

[0025] Furthermore, after generating the first classification confidence corresponding to each type of function classification label whose classification confidence is greater than the third confidence threshold Y3 in all activity vectors, the multi-function type determination process further includes:

[0026] Obtain each activity vector whose classification confidence is less than Y3 in all activity vectors, and obtain a classification sequence in the target classification model. The classification sequence includes the confidence value corresponding to each preset functional classification.

[0027] According to the number of confidence values ​​of classification confidence greater than Y4 in all classification sequences, the second classification confidence corresponding to each type of functional classification label is generated. Among them, the second classification confidence Em corresponding to the mth type of functional classification label whose classification confidence is greater than the confidence value of Y4 satisfies the following conditions:

[0028] Em = (P m ×K1) / P Y4 sum +M2.

[0029] P m is the number of confidence values ​​corresponding to the mth functional classification whose classification confidence is greater than Y4 in all classification sequences, P Y4 sum It is the total number of confidence values ​​corresponding to all functional classifications in all classification sequences whose classification confidence is greater than Y4, and M2 is the second confidence base. M2∈[0.7, 0.8).

[0030] Each type of function classification label is sorted in descending order of the second classification confidence, and a second function classification recommendation list of the new access server is generated.

[0031] Furthermore, after generating the second classification confidence corresponding to each type of functional classification label according to the number of confidence values ​​of the classification confidence greater than Y4 in all classification sequences, the multifunctional type determination process further includes:

[0032] If there is a function classification with a classification confidence greater than Y4 in all classification sequences that is the same as the function classification in the first function classification recommendation list, then the confidence adjustment coefficient corresponding to the same function classification is generated. Among them, the confidence adjustment coefficient Fu corresponding to the u-th same function classification satisfies the following conditions:

[0033] Fu=P u ×L / P Y4 sum .

[0034] Among them, P u is the number of confidence values ​​corresponding to the u-th classification with the same function in all classification sequences whose classification confidence is greater than Y4. L is the adjustment coefficient.

[0035] According to the confidence adjustment coefficient, the first classification confidence of the corresponding function classification in the first function classification recommendation list is adjusted and updated. Among them, the updated first classification confidence Gu corresponding to the u-th same function classification in the first function classification recommendation list satisfies the following conditions:

[0036] Gu=(1+Fu)×Dn.

[0037] According to the updated first classification confidence, the original order in the first function classification recommendation list is readjusted in a descending order according to the updated first classification confidence.

[0038] Furthermore, after generating the confidence adjustment coefficient corresponding to the same functional classification, the multi-functional type determination process further includes:

[0039] Delete all function category labels corresponding to the same function category in the second function category recommendation list.

[0040] Furthermore, the first function classification recommendation list and the second function classification recommendation list are sequentially joined end to end to generate a function classification recommendation list for a new access server.

[0041] Furthermore, M1=0.9, M2=0.75, K1=0.1, L=0.05.

[0042] Furthermore, the multiple historical time periods are multiple continuous time periods within a historical working cycle, and the target classification model is a logistic regression classification model or a deep learning classification model.

[0043] According to a second aspect of the present invention, a non-transitory computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned method for generating digital twin data based on multi-source heterogeneous data.

[0044] According to a third aspect of the present invention, there is provided an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for generating digital twin data based on multi-source heterogeneous data is implemented.

[0045] The present invention has at least the following beneficial effects:

[0046] In the digital twin metadata generation system of the present invention, by setting up a data standardization platform, the attribute data in different formats obtained by multiple initial metadata acquisition platforms can be uniformly standardized to improve the availability of metadata provided to other digital twin construction platforms in the later stage.

[0047] In addition, when collecting the functional attribute data of the newly connected server in the data center, the type and quantity characteristics of the executed operations reflected in the log information of the server are used to form multiple server activity vectors corresponding to multiple working periods, and then each server activity vector is classified and processed. Finally, the classification results corresponding to all server activity vectors can be counted, and the functional attribute data of the newly connected server can be more accurately obtained according to the statistical results. Therefore, the functional attribute data of the newly connected server can be more accurately determined according to the characteristics of the specific operations performed by the newly connected server, thereby reducing the problem of low accuracy of the functional attribute results existing in human judgment. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0049] Figure 1 A flowchart of a method for generating digital twin data based on multi-source heterogeneous data provided by an embodiment of the present invention;

[0050] Figure 2 A structural block diagram of a digital twin metadata generation system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0052] As a possible embodiment of the present invention, Figure 1 As shown, a digital twin data generation method based on multi-source heterogeneous data is provided, and the method is applied to a digital twin data generation system, such as Figure 2 As shown, the digital twin metadata generation system includes a data standardization platform and multiple initial metadata collection platforms. The multiple initial metadata collection platforms are all connected to the data standardization platform in communication, and the multiple initial metadata collection platforms are used to collect attribute data corresponding to different entity objects. There are differences in the formats of attribute values ​​collected by multiple initial metadata collection platforms for the same attribute. The data standardization platform is used to perform data standardization processing on the acquired attribute data of different entities to generate metadata in a preset format.

[0053] The data standardization platform sets a format conversion mapping table according to the required formats of various types of data required in the corresponding digital twin scene, and the formats of the corresponding data collected by the multiple initial metadata collection platforms connected, so as to unify the data provided by different data collection platforms. Because, when generating a digital twin of some entities, a large amount of information about the entity will need to be collected. Taking buildings as an example, this information includes but is not limited to geographic location, house specifications, basic frame structure, main materials, house use information, house rent, selling price and house age, etc., and this information is usually mastered in different organizational platforms. In addition, different platforms use different standards for data formats and representation methods of certain attributes, so the data standardization platform in the digital twin metadata generation system of this embodiment can standardize the above data to improve the availability of later data.

[0054] The multiple initial metadata collection platforms include a server metadata collection platform for collecting server attribute data, and the server attribute data includes server function attribute data. The server metadata collection platform obtains the function attribute data of the newly connected server according to the following steps:

[0055] S100: Generate a server activity vector corresponding to each historical period based on the log information of multiple historical periods of the newly connected server. a =(A a 1 , A a 2 , …, Aa i , …, A a z ), A a is the server activity vector corresponding to the a-th historical period, A a i is the total number of times the server executes the i-th type of operation in the a-th historical period, z is the total number of types of operations that can be executed during server operation, i=1, 2…z.

[0056] Specifically, multiple historical time periods are multiple continuous time periods within a historical working cycle. The working cycle can be determined based on the main service items of the data center. Generally speaking, a natural day can be a working cycle, and the corresponding historical time period can also be one hour. Of course, in order to make the subsequent classification results more statistically characteristic, the data of the historical time period can be increased, such as shortening the time interval or lengthening the working cycle.

[0057] Services with different functions usually frequently perform different types of operations. For example, for storage-type servers, there will be a large number of data addition, deletion, modification and query operations and regular data backup operations in their logs; for servers with data collection tasks, there will be a large number of data requests, data reception, raw data cleaning and delayed query management operations in their logs. Based on this, the type and quantity characteristics of the executed operations reflected in the log information of the server can roughly describe the functional attributes of the corresponding server. The total number of types of executable operations included in the server activity vector in this embodiment can be the collection of operation types executed by all types of servers in the corresponding data center, or the collection of main execution operations used to distinguish different types of services.

[0058] S200: Input each activity vector into the target classification model, and input the functional classification label and classification confidence corresponding to each activity vector. The target classification model uses the classification label corresponding to the highest classification confidence as the output classification result. The target classification model can be a logistic regression classification model or a deep learning classification model.

[0059] The logistic regression classification model is one of the most typical probability models. It maps the linear combination to between 0 and 1 through the Sigmoid function to obtain the probability of belonging to each category. The output probability can be used for binary or multi-classification problems.

[0060] Deep learning models, such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs), usually use a softmax layer as the last layer to output category probabilities. When these models output probabilities, they usually go through some form of probability calibration to ensure that the output probabilities are closer to the probabilities in the real world. For example, methods such as Platt scaling and isotonic regression are often used for probability calibration.

[0061] In this step, the classification confidence of each corresponding preset function type can be classified for the server activity vector through the above classification model. The type of preset function type is related to the type of task undertaken by the data center. Usually, the preset function type is a collection of the task types undertaken by all servers in the corresponding data center.

[0062] S300: Generate a matching value corresponding to each type of function according to the function classification labels and classification confidences corresponding to all activity vectors. b The following conditions must be met:

[0063] B b =P b Y1 / P sum , where P b Y1 is the total number of classification labels corresponding to the b-th function in the function classification labels whose classification confidence is greater than the first confidence threshold Y1. sum is the total number of functional classification labels.

[0064] S400: If B b >P1, the classification label corresponding to the b-th function in the function classification label is determined as the function attribute data of the new access server. P1 is the first proportion threshold.

[0065] Specifically, Y1 may be 0.9, and P1 may be 0.8.

[0066] If a server is only responsible for one type of task, then a large number of classification results corresponding to all server activity vectors will be consistent, and the confidence corresponding to the high-proportion classification results will also be high. Therefore, the functional attributes of the server responsible for a single task type can be accurately determined through S400.

[0067] S500: If B b ≤P1, then perform multi-function type determination processing to generate at least two classification labels corresponding to the new access server.

[0068] When B bIf ≤ P1, then it is very likely that the server undertakes at least two types of functions. As a result, there may be a mixture of different types of operations in the server activity vector, which will in turn lead to a decrease in the classification confidence corresponding to the server activity vector and / or the proportion of the same classification result. In this case, it is necessary to perform a function type judgment process to determine at least two function attributes corresponding to the server.

[0069] Specifically, the multi-function type judgment process includes:

[0070] S500: Generate a classification result confidence C based on the function classification labels and classification confidences corresponding to all activity vectors. C satisfies the following condition: C = P Y2 / P sum 。P Y2 is the total number of classification labels with a classification confidence greater than the second confidence threshold Y2 in the function classification labels.

[0071] S510: If C > P2, obtain the classification label vector corresponding to the newly connected server. The classification label vector is formed by converting the function classification labels corresponding to each activity vector into corresponding element values in chronological order. P2 is the second proportion threshold.

[0072] Specifically, P2 < P1, and P2 can be 0.4; Y2 < Y1, and Y2 can be 0.7.

[0073] S511: Obtain the classification label vectors corresponding to other multi-function servers belonging to the same organization as the newly connected server respectively.

[0074] S512: If the similarity between the classification label vector corresponding to the newly connected server and the classification label vector corresponding to any other multi-function server is greater than the preset similarity threshold, then determine the classification label corresponding to the multi-function server as the function attribute data of the newly connected server. The method for obtaining the classification label vector corresponding to other multi-function servers is the same as the method for obtaining the classification label vector of the newly connected server.

[0075] Since there may be a mixture of different types of operations in the server activity vector, it will lead to a decrease in the classification confidence corresponding to the server activity vector and / or the proportion of the same classification result. Therefore, appropriately lower the values corresponding to P2 and Y2 during the determination. In this embodiment, when S510 is satisfied, it indicates that the credibility of the classification result is still relatively high. Thus, the classification results corresponding to all server activity vectors can be converted into a time series vector. For example, if the classification results are a1, a2, a3, a2, a1,... they can be converted into the corresponding time series vector (1, 2, 3, 2, 1,...).

[0076] Then, since there are also servers performing multiple tasks in the original data center, and the functions of the newly connected servers are roughly the same as some of them. Then for the corresponding original multi-task servers, the server activity vectors obtained during the same time period are roughly the same as the server activity vectors of the newly connected servers. Thus, the similarity of the time series vectors corresponding to the classification results can be calculated to determine the multiple functional attributes corresponding to the multi-task servers. This embodiment is applicable to determining the functional attributes when the function of the newly connected server is the same as that of a certain original server before.

[0077] In the actual use process, sometimes in order to share the server pressure, the function of the newly connected server may be the same as that of a certain original server before. Of course, sometimes it is also possible to fine-tune the function of the newly connected server according to the existing requirements, so that the function of the newly connected server has certain differences from the existing servers. In this case, the following method can be used to form a functional classification recommendation list for the newly connected server to provide an effective reference when determining the server function.

[0078] In S500: After generating the classification result confidence based on the function classification labels and classification confidence corresponding to all activity vectors, the multi-functional type judgment process further includes:

[0079] S520: If C < P2, then generate the first classification confidence corresponding to each type of function classification label with a classification confidence greater than the third confidence threshold Y3 among all activity vectors. Y3 < Y2, where the first classification confidence Dn corresponding to the nth type of function classification label with a classification confidence greater than Y3 satisfies the following conditions:

[0080] Dn = (P n × K1) / P Y3 sum + M1.

[0081] P n is the number of the nth type of function classification label with a classification confidence greater than Y3, P Y3 sum is the total number of function classification labels with a classification confidence greater than Y3, K1 is the first confidence adjustment amplitude. M1 is the first confidence base number. M1 ∈ [0.8, 0.9].

[0082] Specifically, in this embodiment, the respective defined coefficients can be the following situations: Y3 = 0.6, M1 = 0.9, M2 = 0.75, K1 = 0.1, L = 0.05.

[0083] S521: Sort each type of function classification label in descending order of the first classification confidence to generate the first functional classification recommendation list for the newly connected server.

[0084] In this embodiment, all classification results with a classification confidence greater than Y3 are taken out, and then the number of classification results of each category accounts for the total number P Y3 sum The first classification confidence is generated according to the ratio of . For example, if there are 3 categories in all classification results greater than Y3, and the number of each category is a=30, b=50, and c=80, then their corresponding first classification confidences are a: (30×0.1) / 160+0.9=0.919, b: (50×0.1) / 160+0.9=0.931, and c: (80×0.1) / 160+0.9=0.950.

[0085] Therefore, the results with higher confidence levels are taken out from the classification results, and then a higher first classification confidence level is assigned based on the proportion of quantity, thereby forming a first functional classification recommendation list.

[0086] After generating the first classification confidence corresponding to each type of function classification label whose classification confidence is greater than the third confidence threshold Y3 in all activity vectors at S520, the multi-function type determination process further includes:

[0087] S522: Obtain a classification sequence in the target classification model for each activity vector whose classification confidence is less than Y3 among all activity vectors. The classification sequence includes the confidence value corresponding to each preset functional classification.

[0088] S523: Generate a second classification confidence corresponding to each type of functional classification label according to the number of classification confidence values ​​greater than Y4 in all classification sequences. The second classification confidence Em corresponding to the mth type of functional classification label whose classification confidence value is greater than Y4 satisfies the following conditions:

[0089] Em = (P m ×K1) / P Y4 sum +M2.

[0090] P m is the number of confidence values ​​corresponding to the mth functional classification whose classification confidence is greater than Y4 in all classification sequences, P Y4 sum is the total number of confidence values ​​corresponding to all functional classifications in all classification sequences whose classification confidence is greater than Y4, and M2 is the second confidence base. M2∈[0.7, 0.8). Y4<Y3, in this embodiment, Y4 can be 0.45.

[0091] S524: Sort each type of function classification label in descending order of the second classification confidence, and generate a second function classification recommendation list for the new access server.

[0092] In this embodiment, since the confidences of the remaining classification results are all less than Y3, it means that the classification results outputted in the end may not be accurate. There may be multiple classification results with similar confidences in the multiple originally preset classifications. For example, after a certain activity vector is classified, there may be three preset classifications with confidences of 0.51, 0.48 and 0.5 respectively. However, due to the operation mechanism of the classification model, the classification result corresponding to the highest 0.51 is finally selected for output. However, the classification results corresponding to the other 0.48 and 0.5 are also very likely to be correct results.

[0093] Therefore, in this embodiment, for the remaining activity vectors with low classification confidence (that is, the classification confidence is less than Y3), the confidence of the activity vector for each preset function classification, that is, the classification sequence, will be obtained from the target classification model. Then, the classification results with higher (that is, the classification confidence is greater than Y4) in the classification sequence corresponding to all the remaining activity vectors are statistically analyzed to generate the second classification confidence. For example, there are 3 categories for all classification results greater than Y4, and the corresponding numbers for each category are d=20, e=70, and f=40, respectively. Then, the first classification confidences corresponding to them are d: (20×0.1) / 130+0.75=0.765, e: (70×0.1) / 130+0.75=0.804, and f: (40×0.1) / 130+0.75=0.780.

[0094] Therefore, after processing the results with lower confidence in the classification results, the corresponding second classification confidence can be generated according to the quantity proportion within a lower confidence range, thereby forming a second functional classification recommendation list.

[0095] S525: The first function classification recommendation list and the second function classification recommendation list are sequentially combined head to tail to generate a function classification recommendation list for a new access server.

[0096] As another embodiment of the present invention, after S523: generating the second classification confidence corresponding to each type of functional classification label according to the number of confidence values ​​of the classification confidence greater than Y4 in all classification sequences, the multi-function type determination process further includes:

[0097] S530: If there is a function category in all classification sequences whose classification confidence is greater than Y4 and is the same as the function category in the first function category recommendation list, generate a confidence adjustment coefficient corresponding to the same function category. The confidence adjustment coefficient Fu corresponding to the u-th same function category satisfies the following conditions:

[0098] Fu=P u ×L / P Y4 sum .

[0099] Among them, P u is the number of confidence values ​​corresponding to the u-th classification with the same function in all classification sequences whose classification confidence is greater than Y4. L is the adjustment coefficient.

[0100] S531: According to the confidence adjustment coefficient, the first classification confidence of the corresponding function classification in the first function classification recommendation list is adjusted and updated. Among them, the updated first classification confidence Gu corresponding to the u-th same function classification in the first function classification recommendation list satisfies the following conditions:

[0101] Gu=(1+Fu)×Dn.

[0102] This embodiment is to process the same classification items in the first function classification recommendation list and the second function classification recommendation list. Take the same classification item a as an example. The first classification confidence of a in the first function classification recommendation list is: (30×0.1) / 160+0.9=0.919. The second classification confidence of a in the second function classification recommendation list is: (40×0.1) / 150+0.75=0.777. Among them, in all classification sequences, the number of confidence values ​​corresponding to the a function classification with a classification confidence greater than Y4 is 40, P Y4 sum=150. In this case, the confidence adjustment coefficient corresponding to function category a is 40×0.05 / 150=0.013. The updated first category confidence corresponding to function category a in the first function category recommendation list is (1+0.013)×0.919=0.931. It can be seen that the original first category confidence can be further improved based on the same function category in the first function category recommendation list.

[0103] S532: According to the updated first classification confidence, the original order in the first function classification recommendation list is readjusted in a descending order according to the updated first classification confidence.

[0104] S533: Delete all function category labels corresponding to the same function category in the second function category recommendation list.

[0105] S534: The first function classification recommendation list and the second function classification recommendation list are sequentially combined head to tail to generate a function classification recommendation list for a new access server.

[0106] Therefore, after the processing in this embodiment, there will be no duplicate function classifications in the first function classification recommendation list and the second function classification recommendation list, so that relevant personnel can refer to them.

[0107] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0108] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by combining software with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0109] In an exemplary embodiment of the present disclosure, an electronic device capable of implementing the above method is also provided.

[0110] It will be appreciated by those skilled in the art that various aspects of the present invention may be implemented as a system, method or program product. Therefore, various aspects of the present invention may be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to herein as a "circuit", "module" or "system".

[0111] The electronic device according to this embodiment of the present invention is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0112] The electronic device is presented in the form of a general-purpose computing device. The components of the electronic device may include, but are not limited to: the at least one processor mentioned above, the at least one storage device mentioned above, and a bus connecting different system components (including storage devices and processors).

[0113] The storage stores program codes, which can be executed by the processor, so that the processor executes the steps according to various exemplary embodiments of the present invention described in the above “Exemplary Method” section of this specification.

[0114] The memory may include readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory, and may further include read only memory (ROM).

[0115] The storage may also include a program / utility having a set (at least one) of program modules, such program modules including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0116] The bus may represent one or more of several types of bus structures including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures.

[0117] The electronic device may also communicate with one or more external devices (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may communicate with one or more devices that enable a user to interact with the electronic device, and / or may communicate with any device that enables the electronic device to communicate with one or more other computing devices (e.g., routers, modems, etc.). This communication may be performed through an input / output (I / O) interface. In addition, the electronic device may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) through a network adapter. The network adapter communicates with other modules of the electronic device through a bus. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0118] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes a number of instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0119] In an exemplary embodiment of the present disclosure, a computer-readable storage medium is also provided, on which a program product capable of implementing the above method of the present specification is stored. In some possible implementations, various aspects of the present invention may also be implemented in the form of a program product, which includes a program code, and when the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps according to various exemplary embodiments of the present invention described in the above "Exemplary Method" section of the present specification.

[0120] The program product may adopt any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, a system, device or component of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0121] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0122] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.

[0123] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0124] In addition, the above-mentioned figures are only schematic illustrations of the processes included in the method according to an exemplary embodiment of the present invention, and are not intended to be limiting. It is easy to understand that the processes shown in the above-mentioned figures do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.

[0125] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0126] The above are only specific embodiments 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 shall be subject to the protection scope of the claims.

Claims

1. A digital twin data generation method based on multi-source heterogeneous data, characterized in that: The method is applied to a digital twin metadata generation system, which includes a data standardization platform and multiple initial metadata collection platforms. The multiple initial metadata collection platforms are all communicatively connected to the data standardization platform, and the multiple initial metadata collection platforms are respectively used to collect attribute data corresponding to different entity objects; There are differences in the formats of the attribute values collected by the multiple initial metadata collection platforms for the same attribute; The data standardization platform is used to perform data standardization processing on the attribute data of different entities obtained to generate metadata in a preset format; The multiple initial metadata collection platforms include a server metadata collection platform for collecting server attribute data, and the server attribute data includes server function attribute data; the server metadata collection platform obtains the newly accessed server function attribute data according to the following steps: According to the log information of multiple historical periods of the newly accessed server, a server activity vector corresponding to each historical period is generated respectively; wherein A a =(A a 1 , A a 2 , …, A a i , …, A a z ), A a is the server activity vector corresponding to the a-th historical period, A a i is the total number of times the server executes the i-th type of operation in the a-th historical period, z is the total number of types of operations that can be executed during server operation, i=1, 2…z; Input each activity vector into the target classification model, and input the function classification label and classification confidence corresponding to each activity vector; the target classification model uses the classification label corresponding to the highest classification confidence as the output classification result; According to the functional classification labels and classification confidences corresponding to all activity vectors, the matching value corresponding to each type of function is generated; among them, the matching value B corresponding to the b-th type of function in the functional classification label is b The following conditions must be met: B b =P b Y1 / P sum , where P b Y1 is the total number of classification labels corresponding to the b-th function in the function classification labels whose classification confidence is greater than the first confidence threshold Y1; sum is the total number of functional classification labels; If B b >P1, then the classification label corresponding to the b-th function in the function classification label is determined as the function attribute data of the new access server; P1 is the first proportion threshold.

2. The method according to claim 1, characterized in that The steps for the server metadata collection platform to obtain the newly accessed server function attribute data further include: If B b ≤P1, then perform multifunctional type determination processing to generate at least two classification labels corresponding to the new access server; The multi-functional type judgment processing includes: According to the functional classification labels and classification confidences corresponding to all activity vectors, the classification result confidence C is generated; C satisfies the following conditions: C=P Y2 / P sum ;P Y2 Y2 is the total number of classification labels in the functional classification labels whose classification confidence is greater than the second confidence threshold Y2; <Y1; If C > P2, obtain the classification label vector corresponding to the newly accessed server; the classification label vector is formed by converting the function classification label corresponding to each activity vector into a corresponding element value in chronological order; P2 is the second ratio threshold; P2 < P1; Obtain the classification label vectors corresponding to other multi-functional servers belonging to the same organization as the newly accessed server respectively; If the similarity between the classification label vector corresponding to the newly accessed server and the classification label vector corresponding to any other multi-functional server is greater than the preset similarity threshold, determine the classification label corresponding to the multi-functional server as the function attribute data of the newly accessed server.

3. The method according to claim 2, characterized in that After generating the classification result confidence according to the function classification labels and classification confidences corresponding to all activity vectors, the multi-functional type judgment processing further includes: If C < P2, generate the first classification confidence corresponding to each type of function classification label with a classification confidence greater than the third confidence threshold Y3 among all activity vectors; Y3 < Y2, where the first classification confidence Dn corresponding to the nth type of function classification label with a classification confidence greater than Y3 satisfies the following conditions: Dn=(P n ×K1) / P Y3 sum +M1; P n is the number of n-th functional classification labels whose classification confidence is greater than Y3, P Y3 sum is the total number of functional classification labels whose classification confidence is greater than Y3, K1 is the first confidence adjustment amplitude; M1 is the first confidence base; M1∈[0.8, 0.9]; Sort each type of function classification label in descending order of the first classification confidence to generate the first function classification recommendation list of the newly accessed server.

4. The method according to claim 3, characterized in that After generating the first classification confidence corresponding to each type of function classification label with a classification confidence greater than the third confidence threshold Y3 among all activity vectors, the multi-functional type judgment processing further includes: Obtain the classification sequence obtained by each activity vector with a classification confidence less than Y3 in the target classification model; the classification sequence includes the confidence values corresponding to each preset function classification; According to the number of confidence values ​​of classification confidence greater than Y4 in all classification sequences, the second classification confidence corresponding to each type of functional classification label is generated; wherein, the second classification confidence Em corresponding to the mth type of functional classification label whose classification confidence is greater than the confidence value of Y4 satisfies the following conditions: I=(P m ×K1) / P Y4 sum +M2; P m is the number of confidence values ​​corresponding to the mth functional classification whose classification confidence is greater than Y4 in all classification sequences, P Y4 sum is the total number of confidence values ​​corresponding to all functional classifications in all classification sequences whose classification confidence is greater than Y4, M2 is the second confidence base; M2∈[0.7, 0.8); Each type of function classification label is sorted in descending order of the second classification confidence, and a second function classification recommendation list of the new access server is generated.

5. The method according to claim 4, characterized in that After generating the second classification confidence corresponding to each type of functional classification label according to the number of confidence values ​​of the classification confidence greater than Y4 in all classification sequences, the multifunctional type determination process further includes: If there is a functional classification with a classification confidence greater than Y4 in all classification sequences that is the same as the functional classification in the first functional classification recommendation list, then the confidence adjustment coefficient corresponding to the same functional classification is generated; wherein the confidence adjustment coefficient Fu corresponding to the u-th same functional classification satisfies the following conditions: Fu=P u ×L / P Y4 sum ; Among them, P u is the number of confidence values ​​corresponding to the u-th classification with the same function whose classification confidence is greater than Y4 in all classification sequences; L is the adjustment coefficient; Y4<Y3; According to the confidence adjustment coefficient, the first classification confidence of the corresponding function classification in the first function classification recommendation list is adjusted and updated; wherein the updated first classification confidence Gu corresponding to the u-th same function classification in the first function classification recommendation list satisfies the following conditions: Gu = (1 + Fu) × Dn; According to the updated first classification confidence, the original order in the first function classification recommendation list is readjusted in a descending order according to the updated first classification confidence.

6. The method according to claim 5, characterized in that After generating the confidence adjustment coefficient corresponding to the same functional classification, the multi-functional type determination process further includes: Delete all function category labels corresponding to the same function category in the second function category recommendation list.

7. The method according to claim 6, characterized in that The function classification recommendation list of the new access server is generated by sequentially joining the first function classification recommendation list and the second function classification recommendation list head to tail.

8. The method according to claim 7, characterized in that M1=0.9, M2=0.75, K1=0.1, L=0.

05.

9. The method according to claim 1, characterized in that: The multiple historical time periods are multiple continuous time periods within a historical working cycle, and the target classification model is a logistic regression classification model or a deep learning classification model.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, it implements a method for generating digital twin data based on multi-source heterogeneous data as described in any one of claims 1 to 8.

Images