Resource demand quantity prediction method, device, equipment, medium and product

Abstract

The invention provides a resource demand quantity prediction method and device, equipment, a medium and a product, which can be applied to the technical field of artificial intelligence and the field of financial science and technology. The method comprises the steps of determining a historical resource use index, a historical operation log and historical code change information for a target business item; performing feature fusion processing on the determined historical resource use index, historical operation log and historical code change information to obtain a historical feature time sequence; different elements in the historical feature time sequence respectively correspond to different time windows; any element in the historical feature time sequence comprises a historical resource use index feature, a historical operation log feature and a historical code change information feature in a corresponding time window; and on the basis of a pre-trained first large model, according to the obtained historical feature time sequence, predicting the resource demand of the target business item in a future time window.

Description

Technical Field

[0001] This application relates to the fields of artificial intelligence technology and financial technology, specifically to a method, apparatus, equipment, medium, and product for predicting resource demand. Background Technology

[0002] During the operation of business projects such as applications or processes, it is usually necessary to utilize pre-allocated resources to perform operations. For example, applications deployed on servers can utilize pre-allocated bandwidth resources to perform corresponding business operations. When allocating resources for business projects, it is typically necessary to predict the future resource requirements of the business projects in advance in order to determine the amount of resources to be allocated.

[0003] However, currently, when forecasting resource demand, predictions are usually based solely on the historical resource usage of business projects, resulting in low accuracy in forecasting resource demand. Summary of the Invention

[0004] In view of the above problems, this application provides a method, apparatus, equipment, medium and product for improving the accuracy of resource demand forecasting.

[0005] According to the first aspect of this application, a method for predicting resource demand is provided, comprising: determining historical resource usage indicators, historical operation logs, and historical code change information for a target business project; performing feature fusion processing on the determined historical resource usage indicators, historical operation logs, and historical code change information to obtain a historical feature time series sequence; different elements in the historical feature time series sequence correspond to different time windows, and the order of different elements is arranged according to the time order of the corresponding time windows; any element in the historical feature time series sequence includes: historical resource usage indicator features, historical operation log features, and historical code change information features within the corresponding time window; and predicting the resource demand of the target business project within a future time window based on a pre-trained first model and the obtained historical feature time series sequence.

[0006] According to an embodiment of this application, the method further includes: predicting potential faults in the target business project based on a pre-trained first large model and the obtained historical feature time series; and performing automatic fault handling operations for the target business project when the risk level of the predicted potential fault is greater than a preset risk level threshold.

[0007] According to an embodiment of this application, predicting potential faults in the target business project includes: predicting fault information of potential faults in the target business project; the fault information includes at least one of the following: fault type, fault location, fault probability, and fault impact range; the method further includes: comprehensively analyzing the fault information of the predicted potential faults to determine the risk level of the predicted potential faults.

[0008] According to an embodiment of this application, the method further includes: based on a pre-trained first large model, predicting the degree of impact of code changes in the target business project according to the obtained historical feature time series; determining the amount of reserve resources for the target business project in the future time window according to the predicted degree of impact of code changes; wherein the predicted degree of impact of code changes is positively correlated with the determined amount of reserve resources.

[0009] According to an embodiment of this application, determining the amount of reserve resources for the target business project in the future time window based on the predicted impact of the code change includes: if the predicted impact of the code change is greater than a preset impact threshold, determining the amount of reserve resources for the target business project in the future time window based on the predicted impact of the code change and the current amount of reserve resources for the target business project; the determined amount of reserve resources is greater than the current amount of reserve resources for the target business project.

[0010] According to an embodiment of this application, the method further includes: determining the range of resource quantity adjustment set for the target business project in the future time window based on the predicted resource demand and the determined reserve resource quantity.

[0011] According to an embodiment of this application, the step of performing feature fusion processing on the determined historical resource usage indicators, historical operation logs, and historical code change information to obtain a historical feature time series sequence includes: determining the historical resource usage indicators, historical operation logs, and historical code change information belonging to the same time window; performing feature extraction on the historical resource usage indicators, historical operation logs, and historical code change information belonging to the same time window to obtain historical resource usage indicator features, historical operation log features, and historical code change information features belonging to the same time window; and sorting the historical resource usage indicator features, historical operation log features, and historical code change information features belonging to different time windows according to the time order between different time windows to obtain a historical feature time series sequence.

[0012] According to an embodiment of this application, the method for extracting the historical code change information features includes: extracting historical code change information features based on a pre-trained second model; the extracted historical code change information features include at least one of the following: code semantic features, code change content features, current code complexity features, and initial value of the degree of impact of code changes.

[0013] According to an embodiment of this application, the method for extracting the features of the historical operation log includes: extracting historical operation log features based on a pre-trained third model; the extracted historical operation log features include at least one of the following: log parsing features, historical fault features, and log semantic features.

[0014] A second aspect of this application provides a resource demand prediction device, comprising: a determination module, configured to determine historical resource usage indicators, historical operation logs, and historical code change information for a target business project; a fusion module, configured to perform feature fusion processing on the determined historical resource usage indicators, historical operation logs, and historical code change information to obtain a historical feature time series sequence; different elements in the historical feature time series sequence correspond to different time windows, and the order of different elements is arranged according to the time order of the corresponding time windows; any element in the historical feature time series sequence includes: historical resource usage indicator features, historical operation log features, and historical code change information features within the corresponding time window; and a prediction module, configured to predict the resource demand of the target business project within a future time window based on a pre-trained first model and the obtained historical feature time series sequence.

[0015] A third aspect of this application provides an electronic device comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method described above.

[0016] A fourth aspect of this application also provides a computer-readable storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implement the steps of the above-described method.

[0017] The fifth aspect of this application also provides a computer program product, including a computer program or instructions that, when executed by a processor, implement the steps of the above-described method. Attached Figure Description

[0018] The above-mentioned contents, other objects, features and advantages of this application will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0019] Figure 1 This illustration schematically depicts an application scenario of a resource demand prediction method according to an embodiment of this application.

[0020] Figure 2 A flowchart illustrating a resource demand forecasting method according to an embodiment of this application is shown schematically.

[0021] Figure 3 This schematic diagram illustrates a structural block diagram of a resource demand prediction device according to an embodiment of the present application;

[0022] Figure 4 A block diagram schematically illustrates an electronic device suitable for implementing a resource demand prediction method according to an embodiment of this application. Detailed Implementation

[0023] The embodiments of this application will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of this application. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of this application for ease of explanation. However, it will be apparent that one or more embodiments may be implemented without these specific details. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0024] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0025] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0026] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0027] During the operation of business projects such as applications or processes, it is usually necessary to perform operations using pre-allocated resources. For example, applications deployed on servers can utilize pre-allocated bandwidth resources to perform corresponding business operations. When allocating resources for business projects, it is usually necessary to predict the future resource requirements of the business project in order to determine the amount of resources to be allocated. However, currently, resource demand predictions are typically based solely on the historical resource usage of business projects, resulting in low accuracy in the predicted resource demand.

[0028] To address the aforementioned technical problems, embodiments of this application provide a method for predicting resource demand. This method combines historical resource usage metrics, historical operation logs, and historical code change information of a business project to comprehensively predict its future resource demand.

[0029] Historical resource usage metrics can include the amount of memory, processors, traffic, and bandwidth used by a business project in the past, which can help predict the future resource requirements of the business project. Historical operation logs can contain information such as the execution status of the business, and based on these logs, information such as business traffic or processing volume can be analyzed to help predict the future resource requirements of the business project. Historical code change information can include code version change information for the business project, as well as the current version of the business project code. Based on this information, changes in code complexity or changes in resource consumption can be analyzed to help predict the future resource requirements of the business project.

[0030] Understandably, if analyzing historical operation logs reveals a downward trend in business traffic, it can be predicted that the future resource requirements of the business project will decrease; similarly, if analyzing historical code change information reveals a gradual increase in code complexity, it can be predicted that the future resource requirements of the business project will increase.

[0031] Therefore, the above method can improve the accuracy of resource demand forecasting by introducing more information dimensions and information volume to predict the future resource demand of business projects.

[0032] Furthermore, the methods described above can leverage large language models to comprehensively analyze historical resource usage metrics, historical operation logs, and historical code change information for business projects, thereby predicting future resource requirements. The model analysis capabilities provided by large language models can improve the accuracy of resource requirement predictions.

[0033] In the above method, in order to improve the correlation between historical resource usage indicators, historical operation logs, and historical code change information, feature fusion processing can be performed on these three types of information to analyze the correlation between them. This information can then be used to predict the future resource demand of business projects and improve the accuracy of resource demand prediction.

[0034] Specifically, historical resource usage metrics, historical operation logs, and historical code change information can be aligned along a time dimension. This can be done by dividing the data into time windows and combining them into a time-series sequence. In the above method, historical resource usage metrics, historical operation logs, and historical code change information belonging to the same time window can be merged and sorted according to the chronological order between different time windows, resulting in a set of time-series sequences.

[0035] It's understandable that historical resource usage metrics, historical operation logs, and historical code change information belonging to the same time window can have temporal correlations. For example, code changes may cause changes in resource usage metrics; or changes in peak business traffic represented in the operation logs may cause changes in resource usage metrics, and so on. By aligning these within time windows, it's easier to mine and analyze the correlations between historical resource usage metrics, historical operation logs, and historical code change information. This information can then be used in subsequent model predictions of business project resource requirements, improving the accuracy of resource requirement forecasts.

[0036] It should be noted that the resource demand prediction method and apparatus provided in the embodiments of this application can be applied to the fields of artificial intelligence technology and fintech. For example, for business projects such as banking applications, applications deployed on bank servers, or containers deployed by banks in the cloud, the resource demand prediction method provided in the embodiments of this application can be used to predict resource demand, facilitating the advance allocation of resources for business projects. The resource demand prediction method and apparatus provided in the embodiments of this application can also be applied to any field other than fintech, and the application field of the resource demand prediction method and apparatus provided in the embodiments of this application is not limited.

[0037] In the technical solution of this application, the user information (including but not limited to user personal information, user image information, user device information, such as location information) and data (including but not limited to data used for analysis, stored data, and displayed data) involved are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of related data all comply with relevant laws, regulations, and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entry points for users to choose to authorize or refuse.

[0038] Figure 1 The illustration shows an application scenario diagram of a resource demand prediction method according to an embodiment of this application.

[0039] like Figure 1 As shown, application scenario 100 according to this embodiment may include: a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 serves as a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0040] Users can interact with server 105 via network 104 using first terminal device 101, second terminal device 102, or third terminal device 103 to receive or send messages, etc. Various communication client applications can be installed on first terminal device 101, second terminal device 102, and third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social media platform software, etc. (for example only).

[0041] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be various electronic devices with displays and support web browsing, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0042] Server 105 can be a server that provides various services, such as a backend management server that supports websites browsed by users using the first terminal device 101, the second terminal device 102, or the third terminal device 103 (this is just an example). The backend management server can analyze and process data such as received user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0043] It should be noted that the resource demand prediction method provided in this application embodiment can generally be executed by server 105. Correspondingly, the resource demand prediction device provided in this application embodiment can generally be located in server 105. The resource demand prediction method provided in this application embodiment can also be executed by a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105. Correspondingly, the resource demand prediction device provided in this application embodiment can also be located in a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105.

[0044] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0045] Figure 2 A flowchart illustrating a resource demand prediction method according to an embodiment of this application is shown schematically.

[0046] like Figure 2 As shown, the resource demand forecasting method provided in this embodiment may include operations S210 to S230. The embodiments of this application do not limit the executing entity of the resource demand forecasting method; it can be applied to any electronic device or any software application, and optionally, it can be applied to a server, user terminal, or client.

[0047] In operation S210, for the target business project, historical resource usage indicators, historical operation logs, and historical code change information are determined.

[0048] In operation S220, feature fusion processing is performed on the determined historical resource usage indicators, historical operation logs, and historical code change information to obtain a historical feature time series sequence. Different elements in the historical feature time series sequence correspond to different time windows, and the order of different elements is arranged according to the time order of the corresponding time window. Any element in the historical feature time series sequence includes: historical resource usage indicator features, historical operation log features, and historical code change information features within the corresponding time window.

[0049] In operation S230, based on the first pre-trained model, the resource requirements of the target business project within a future time window are predicted according to the obtained historical feature time series.

[0050] This method can perform feature fusion processing on the historical resource usage indicators, historical operation logs, and historical code change information of the target business project. It is used to predict the future resource demand of the target business project through a large model. The accuracy of resource demand prediction can be improved by increasing the information dimension and amount of information on the basis of prediction.

[0051] The embodiments of this application do not limit the target business project. Optionally, the target business project can be any business project, and any business project that requires prediction of resource demand can be referred to as the target business project. The target business project can be a runnable software project that utilizes resources to perform business, and its specific form can be an application, process, container, or system, etc. The embodiments of this application do not limit the form of the target business project.

[0052] The embodiments of this application do not limit the resources used by the target business project. Optionally, the resources used by the target business project may include: storage resources, computing resources, and communication resources, specifically, for example, the amount of memory, the number of processors, and the bandwidth. It is understood that the predicted resource requirements may include at least one of the following: the predicted memory requirements, the predicted number of processors required, and the predicted bandwidth requirements.

[0053] The embodiments of this application do not limit historical resource usage metrics, historical operation logs, and historical code change information. Optionally, historical resource usage metrics can be the historical resource usage monitored for a target business project, specifically including historical storage resources, historical computing resources, and historical communication resources used by the target business project, which can be monitored and recorded in the form of time series. Optionally, historical operation logs can include operation logs recorded by the target business project during its historical operation. Optionally, historical code change information can include version change information of the target business project, specifically including code differences between each version of the target business project and the previous version, and may also include information such as code additions, code updates, and code deletions of the target business project, as well as information such as code complexity and updates to dependent files.

[0054] Among them, historical resource usage indicators, historical operation logs, and historical code change information can have corresponding timestamps to facilitate subsequent feature fusion processing.

[0055] The embodiments of this application do not limit the specific method of feature fusion processing. Optionally, historical resource usage indicators, historical operation logs, and historical code change information can be aligned in terms of time dimension, so as to fuse historical resource usage indicators, historical operation logs, and historical code change information that are close in time. Of course, a pre-trained feature fusion model can also be used to extract fused features based on historical resource usage indicators, historical operation logs, and historical code change information.

[0056] Optionally, feature fusion processing is performed on the determined historical resource usage indicators, historical operation logs, and historical code change information to obtain a historical feature time series sequence. Specifically, this may include: identifying historical resource usage indicators, historical operation logs, and historical code change information belonging to the same time window; extracting features from these historical resource usage indicators, historical operation logs, and historical code change information belonging to the same time window; and sorting the historical resource usage indicator features, historical operation log features, and historical code change information features belonging to different time windows according to the chronological order of the different time windows to obtain a historical feature time series sequence. This embodiment can align historical resource usage indicators, historical operation logs, and historical code change information according to time windows to achieve feature fusion, improve the correlation of features within the same time window, and thus improve the accuracy of resource demand prediction.

[0057] It is understandable that historical resource usage metrics, historical operation logs, and historical code change information belonging to the same time window can be aligned using timestamps.

[0058] The embodiments of this application do not limit the size of the time window; it can be 1 day or 1 hour.

[0059] The embodiments of this application do not limit the specific method of feature extraction for historical resource usage indicators, historical operation logs, and historical code change information belonging to the same time window. Optionally, features can be extracted independently for historical resource usage indicators, historical operation logs, and historical code change information belonging to the same time window to obtain corresponding historical resource usage indicator features, historical operation log features, and historical code change information features; alternatively, a pre-trained machine learning model or feature fusion model can be used to output historical resource usage indicator features, historical operation log features, and historical code change information features for the input historical resource usage indicators, historical operation logs, and historical code change information belonging to the same time window, where each feature can be extracted by comprehensively extracting the three types of input information.

[0060] In one optional embodiment, a large model can be used to extract features from historical code change information. This involves performing semantic analysis on the historical code change information using the large model to determine the content and extent of the code changes, as well as the code complexity and the impact of code execution on resources. For example, the large model can analyze historical code change information to determine that the current version of the code has added more loop steps, thus indicating that more computing resources are required.

[0061] Furthermore, considering resource demand forecasting, a large-scale model can be used to analyze the impact of code changes. Specifically, this could be the impact of code changes on the resource demands of the target business project, or the impact of code changes on the resource usage of the target business project. For example, if significant changes occur to the code of a target business project, it may lead to substantial fluctuations in its future resource usage. Accordingly, more reserve resources can be prepared in advance to better address potential resource needs during the project's future operation.

[0062] Therefore, a large model can be used to analyze the impact of historical code change information to determine the degree of impact of code changes. Specifically, based on historical code change information, an initial value of the degree of impact of code changes can be determined. Subsequently, other information can be combined to further determine the degree of impact of code changes, thereby improving the accuracy of the determined degree of impact of code changes.

[0063] Optionally, the method for extracting historical code change information features includes: extracting historical code change information features based on a pre-trained second large model; the extracted historical code change information features include at least one of the following: code semantic features, code change content features, current code complexity features, and an initial value of the impact degree of the code change. This embodiment can extract multiple code change information features through a large model, thereby improving the comprehensiveness of code change information features and the accuracy of resource demand prediction.

[0064] The embodiments of this application do not limit the complexity characteristics of the current code, but may specifically include the cyclomatic complexity of the current code. The embodiments of this application do not limit the degree of impact of code changes, but may specifically include the degree of impact of code changes on the resource usage of the target business project. The greater the degree of impact, the more likely the code change is to cause fluctuations in the resources used by the target business project, thus allowing for the preparation of more spare resources.

[0065] For historical operation log features, feature extraction can also be performed using a large-scale model. Optionally, the extraction method for historical operation log features includes: extracting historical operation log features based on a pre-trained third-scale model; the extracted historical operation log features include at least one of the following: log parsing features, historical fault features, and log semantic features. This embodiment can extract multiple historical operation log features through a large-scale model, improving the comprehensiveness of operation log features and the accuracy of resource demand prediction.

[0066] For the characteristics of historical resource usage indicators, feature extraction can also be performed by combining large models. Specifically, it can be based on a pre-trained fourth model to extract the characteristics of historical resource usage indicators.

[0067] The above embodiments explain the extraction methods of historical resource usage index features, historical operation log features, and historical code change information features. Subsequently, they can be sorted according to the time order between their respective time windows to obtain a historical feature time series sequence.

[0068] Understandably, by combining historical feature time series, it is also convenient to analyze the temporal correlation of historical resource usage indicators, historical operation logs and historical code change information, which can improve the accuracy of resource demand prediction.

[0069] After determining the historical feature time series, we can further predict the resource demand of the target business project in the future time window based on the obtained historical feature time series, using the pre-trained first model.

[0070] The embodiments of this application are not limited to a first major model. Specifically, it can be a general major model or a major model that has been fine-tuned and trained, used to predict the future resource requirements of a business project based on the historical time series characteristics of the business project. The first, second, third, and fourth major models can be the same major model or different major models. The embodiments of this application are not limited.

[0071] The embodiments of this application are not limited to future time windows. Optionally, a future time window can be n future time windows, where n can be a positive integer. Optionally, a future time window can be the time window for the release of a new version of the target business project.

[0072] The embodiments of this application do not limit the subsequent operations after predicting the resource demand of the target business project within a future time window. Optionally, based on the predicted resource demand, the amount of resources allocated to the target business project within the future time window can be further determined, which can improve the accuracy of the allocated resource amount. Alternatively, the range of resource adjustments allowed for allocation to the target business project within the future time window can be determined. Based on the actual future operation of the target business project, resource adjustments or expansion / contraction can be carried out within the resource adjustment range, which may include the predicted resource demand.

[0073] In addition, in an optional embodiment, besides predicting the future resource requirements of the target business project, it is also possible to predict the potential failures of the target business project, thereby helping to detect potential failures in the future operation of the target business project in advance and improving the stability of the target business project.

[0074] Therefore, optionally, the above method flow may further include: predicting potential faults in the target business project based on a pre-trained first large model and the obtained historical feature time series; and performing automatic fault handling operations for the target business project if the predicted risk level of the potential fault is greater than a preset risk level threshold. This embodiment can improve the prediction accuracy of potential faults by using a large model to predict potential faults in the target business project.

[0075] It is understandable that with code changes, potential faults may exist in the target business project. Therefore, by leveraging historical feature time series and combining them with large models, we can help analyze and predict potential faults in the target business project, and perform fault analysis and handling in advance to improve the stability of the target business project in future operation.

[0076] The embodiments of this application do not limit the form of information on the predicted potential faults. Optionally, it may include information such as the fault type, fault location, and fault probability of the potential fault. The embodiments of this application do not limit the specific method of determining the risk level of the potential fault. Optionally, the risk level of the potential fault may be determined by comprehensively considering the information on the predicted potential faults.

[0077] Therefore, optionally, predicting potential faults in the target business project includes: predicting fault information of potential faults in the target business project; the fault information includes at least one of the following: fault type, fault location, fault probability, and fault impact range; the above method flow may further include: comprehensively analyzing the fault information of the predicted potential faults to determine the risk level of the predicted potential faults. This embodiment can determine the risk level of potential faults based on the fault information of the predicted potential faults, which can improve the accuracy of the risk level of potential faults.

[0078] The embodiments of this application do not limit the automatic fault handling operation. Optionally, the automatic fault handling operation may include at least one of the following: automatic operation such as fault location, fault analysis, fault isolation, and fault repair. In a specific example, when there are many potential faults and the risk is high, operations such as version rollback or traffic migration can be performed to improve the security of the target business project.

[0079] Furthermore, in an optional embodiment, in addition to predicting the future resource requirements of the target business project, the impact of code changes on the target business project can also be predicted. This helps to determine the stability of resource requirements during the future operation of the target business project in advance, facilitating the early determination of future reserve resources. If the resource usage of the target business project fluctuates significantly during future operation due to code changes, a larger reserve of resources can also improve operational stability.

[0080] Therefore, optionally, the above method flow may further include: based on a pre-trained first large model, predicting the impact of code changes in the target business project according to the obtained historical feature time series; determining the amount of reserve resources for the target business project in a future time window based on the predicted impact of code changes; wherein the predicted impact of code changes is positively correlated with the determined amount of reserve resources. This embodiment can predict the impact of code changes to determine the amount of reserve resources for the target business project in the future, improving the accuracy of the impact of code changes and the amount of reserve resources. For an explanation of the impact of code changes, please refer to the explanations of other embodiments.

[0081] Understandably, the first major model can determine the degree of impact of code changes based on the initial value of the impact of code changes in the historical code change information features, combined with other features in the historical feature time series. The greater the degree of impact of code changes, the more likely the code changes are to cause fluctuations in the resources used by the target business project, thus allowing for the preparation of more reserve resources. Therefore, the predicted degree of impact of code changes can be positively correlated with the determined amount of reserve resources.

[0082] Optionally, a comprehensive judgment can be made by combining thresholds. Based on the predicted impact of the code change, the amount of reserve resources for the target business project in the future time window can be determined. Specifically, this can include: if the predicted impact of the code change is greater than a preset impact threshold, determining the amount of reserve resources for the target business project in the future time window based on the predicted impact of the code change and the current reserve resources of the target business project; the determined amount of reserve resources is greater than the current reserve resources of the target business project. In this embodiment, if the impact of the code change is determined to be significant, the amount of reserve resources can be further increased based on the current reserve resources of the target business project, which can improve the stability of the target business project after the code change.

[0083] The embodiments of this application do not limit the amount of reserve resources. Optionally, the amount of reserve resources may be an additional amount of resources reserved for the target business project, in addition to the amount of resources allocated for the target business project, so that resources can be requested from the additional amount of resources when additional resources are needed during the operation of the target business project. The amount of reserve resources may also be the maximum amount of resources allowed to be allocated for the target business project, and resources can be allocated from the reserve amount for the target business project.

[0084] If the amount of reserve resources for the target business project in the future time window is determined, the range of resources that the target business project is allowed to apply for in the future time window can be determined by combining the predicted resource demand of the target business project in the future time window. This range can be called the resource adjustment range, which allows the target business project to flexibly apply for resources within the resource adjustment range according to the actual operational needs.

[0085] Therefore, optionally, the above method may further include: determining the range of resource adjustments for the target business project within a future time window based on the predicted resource demand and the determined reserve resource quantity. This embodiment can comprehensively consider the predicted resource demand and reserve resource quantity to determine the future adjustable resource range, thereby improving the resource stability of the target business project.

[0086] The embodiments of this application do not limit the specific method of determining the resource quantity adjustment range. Optionally, determining the resource quantity adjustment range can specifically be using the sum of the predicted resource demand and the determined reserve resource quantity as the upper limit of the resource quantity adjustment range; or it can be based on the predicted resource demand to determine a standard value, which can be greater than the resource demand, and further using the sum of the standard value and the determined reserve resource quantity as the upper limit of the resource quantity adjustment range.

[0087] Understandably, based on the determined range of resource adjustments, the target business project can dynamically adjust the requested resource amount during the operation of the future time window, and can expand or shrink its capacity.

[0088] Furthermore, in an alternative embodiment, if the potential faults of the target business item are predicted, repair suggestions for the potential faults can be generated based on a large model, which can help repair the target business item and improve its security and stability.

[0089] In one optional embodiment, if the predicted impact of a code change is significant, a larger model can be used to analyze the parts of the code that cause the significant impact and generate corresponding adjustment suggestions. This can help improve the stability of the target business project. Specifically, by analyzing historical code change information, historical operation logs, and historical resource usage metrics, code segments that consume a lot of resources can be identified. This allows for pinpointing the code segments that need adjustment, facilitating adjustments by business personnel and reducing the impact of the code change.

[0090] For ease of understanding, this application also provides an application embodiment that can schedule and manage cloud resources.

[0091] This embodiment provides a solution to address the following technical problems: (1) How to break down data silos between resource monitoring indicators (corresponding to resource usage indicators in the above method embodiment), log text (corresponding to operation logs in the above method embodiment), and code changes (corresponding to code change situations in the above method embodiment) to achieve cross-modal deep correlation analysis. (2) How to improve the foresight and accuracy of resource prediction so that it can not only reflect historical patterns but also perceive future resource changes caused by code changes and potential faults. (3) How to achieve the transformation from "passive response to alarms" to "proactive prediction and insight" to reduce the lag in resource management and improve system stability and resource utilization.

[0092] To address the aforementioned technical problems, this embodiment proposes a multimodal data fusion and decision-making scheme based on a large language model. Its key innovations include at least the following three points.

[0093] (1) Deep feature extraction of multimodal data: Using a pre-trained large language model to perform semantic understanding and fault entity recognition on unstructured log text and code changes, text information is transformed into structured feature vectors, which can convert various types of information into structured features, making it convenient for subsequent large models to make predictions.

[0094] (2) Feature fusion for resource scheduling: The temporal and spatial alignment and feature-level fusion of resource monitoring indicators, log semantic features and code change impact features are carried out to construct a multi-dimensional view that comprehensively describes the system status.

[0095] (3) Collaborative decision-making based on fusion features: Based on the fusion features, a comprehensive decision is generated through a machine learning model that includes resource demand prediction, potential fault assessment and change impact assessment, and drives precise resource scheduling actions (such as expansion, rollback and resource pool switching).

[0096] The system architecture provided in this embodiment can be divided into three layers: data fusion layer, model analysis layer, and dynamic execution layer.

[0097] 1. Data fusion layer.

[0098] This layer is responsible for preprocessing and feature extraction of multi-source heterogeneous data, providing high-quality fused features for subsequent analysis.

[0099] Resource monitoring index processing: Missing values ​​are filled and standardized for resource monitoring index data such as processor utilization and memory usage. Then, a time-series feature tensor is constructed using the sliding window method to extract features such as mean, trend, and periodicity.

[0100] Log data processing: Input log text stream. Core processing: Fine-tuning a language model pre-trained on general corpora and operation and maintenance logs to enable it to have semantic understanding capabilities in the operation and maintenance domain. The model performs the following specific tasks: (1) Log parsing and template: Summarize and parse the original logs according to the template. (2) Fault entity identification: Identify key entities from the template and original information, such as fault type, affected services, dependent resources and numerical parameters. (3) Semantic vectorization: Encode the semantic information of the entire log event (including templates and entities) into a fixed-length, high-dimensional log semantic feature vector.

[0101] Code submission data processing: Input the submission records of the version control system, including code differences, submission information, list of modified files, number of lines added or deleted, code complexity, etc. Core processing: (1) Extract key features through feature engineering, such as: whether the modified file is located in the core business module, whether the number of lines of code in this submission has increased or decreased, whether the semantics of the submission information contain keywords such as "refactoring", "performance optimization", "emergency fix", etc., and the historical change impact score of the submitter, etc. (2) Preliminary classification of impact level: Input the above features into a classifier, which outputs the resource impact level (high, medium, low) of this change. The classifier is trained under supervision using historical code submissions and subsequent actual resource change data. (3) Vector encoding: Encode the classification result (such as "high") and the main numerical features (such as the number of lines modified) together into a code impact feature vector.

[0102] Spatiotemporal alignment: Using a uniform time window (e.g., 5 minutes) as the granularity, the feature vectors extracted from the above three types of data are timestamped and spliced ​​together to form a multimodal fusion feature vector.

[0103] 2. Model Analysis Layer.

[0104] This layer is the intelligent core of the system, responsible for learning the fused features and using the capabilities of a large language model to output comprehensive decisions.

[0105] Specifically, this can include a dynamic resource optimization model and a decision engine.

[0106] For the resource dynamic optimization model, the model input includes: a multimodal fusion feature vector generated by the data fusion layer and organized according to time series.

[0107] Model Structure and Training: Training Data: The input is a sequence of historical multimodal fusion feature vectors, and the labels are the validated resource allocation states after the corresponding time period. Training Objective: Through supervised learning, minimize the gap (e.g., mean squared error) between the model's predicted resource allocation and the actual resource allocation.

[0108] Model output: The model output is a comprehensive decision vector, including: (1) Resource demand forecast: The specific demand for resources such as processors and memory for each service in the future. (2) Potential fault assessment results: The probability of one or more specific fault modes (such as "database connection pool exhaustion" or "cache breakdown") occurring. (3) Change impact assessment: The percentage change in core resource utilization that may be caused by this code change.

[0109] For the decision engine, the output of the dynamic resource optimization model can be used to determine whether to execute subsequent automation strategies or how to execute them, based on the pre-set upper and lower thresholds.

[0110] A threshold can be set for the predicted resource demand. If the actual future demand exceeds the lower limit threshold, an automatic scaling decision will be made. If it exceeds the upper limit threshold, the system will automatically scale up to the upper limit threshold and notify the operations and maintenance personnel, providing suggestions (such as code changes, peak business traffic, etc. that caused the resource demand to reach the upper limit).

[0111] Based on the results of potential fault assessment, the system can output assessment results according to the current model. If the results exceed the set fault threshold, it can perform operations such as rolling back the version or increasing resources. It can also promptly alert operations and development personnel and provide suggestions for resolving potential faults.

[0112] The system can assess the impact of changes and, if the model output exceeds the set lower threshold, automatically decide to roll back the version or expand resources. If the required resource expansion exceeds the upper threshold, it will automatically expand to the upper threshold, alert operations and development personnel, and provide modification suggestions.

[0113] 3. Dynamic execution layer.

[0114] This layer is responsible for translating the decisions made by the analytics layer into concrete, automated cloud resource operations.

[0115] Resource forecasting and pre-allocation: The model predicts resource demand at future points in time. Before that future point in time, the system can initiate a call to the cloud resource management platform to perform expansion or contraction operations, preparing resources to the target level.

[0116] Fault self-healing and avoidance process: When the probability of potential fault judgment results in the model output exceeds the threshold and is strongly correlated with the impact of code changes, an automated process is triggered: version rollback can be performed to roll back to a stable version, and traffic scheduling can be performed to divert some or all traffic from suspected abnormal service instances, or a degradation strategy for non-core functions can be enabled.

[0117] Elastic preset for change association: For releases with a "high" impact assessment of changes, the system can automatically and temporarily increase the upper limit of the elastic scaling policy threshold of the service during the release window, reserving more expansion space to cope with unforeseen performance overhead.

[0118] The technical effects achieved in this embodiment include at least the following:

[0119] (1) Improved accuracy and foresight of resource scheduling: By integrating code change information, this embodiment can predict the resource changes that may be caused before the new version is launched.

[0120] (2) Enhanced the intelligent location and self-healing capabilities of system faults: By combining log semantic features extracted from large language models, the cause of faults can be quickly located and corresponding fault handling operations can be triggered.

[0121] (3) It realizes collaborative optimization across data modalities, breaks down the "islands" between resource monitoring indicators, logs and code data, and improves the analysis effect through large models.

[0122] (4) Optimized resource utilization efficiency and system stability: proactively allocates resources based on multi-dimensional insights, reducing cloud resource costs. At the same time, proactive fault prediction and avoidance significantly reduce the risk of system downtime.

[0123] Based on the above method embodiments, embodiments of this application also provide a resource demand prediction device. The following will be combined with... Figure 3 The device is described in detail.

[0124] Figure 3 The diagram illustrates a structural block diagram of a resource demand prediction device according to an embodiment of this application.

[0125] like Figure 3 As shown, the resource demand prediction device 300 provided in this embodiment includes: a determination module 310, a fusion module 320 and a prediction module 330.

[0126] The determination module 310 is used to determine historical resource usage indicators, historical operation logs, and historical code change information for a target business project. In one embodiment, the determination module 310 can be used to perform the operation S210 described above and related operations, which will not be repeated here.

[0127] The fusion module 320 is used to perform feature fusion processing on the determined historical resource usage indicators, historical operation logs, and historical code change information to obtain a historical feature time series sequence. Different elements in the historical feature time series sequence correspond to different time windows, and the order of different elements is arranged according to the time order of the corresponding time window. Any element in the historical feature time series sequence includes: historical resource usage indicator features, historical operation log features, and historical code change information features within the corresponding time window. In one embodiment, the fusion module 320 can be used to execute the operation S220 described above and related operations, which will not be repeated here.

[0128] The prediction module 330 is used to predict the resource demand of a target business project within a future time window based on a pre-trained first model and the obtained historical feature time series. In one embodiment, the prediction module 330 can be used to perform the operation S230 described above and related operations, which will not be repeated here.

[0129] According to an embodiment of this application, the prediction module 330 is further configured to: predict potential faults in the target business project based on a pre-trained first large model and the obtained historical feature time series; and perform automatic fault handling operations for the target business project if the risk level of the predicted potential fault is greater than a preset risk level threshold.

[0130] According to an embodiment of this application, the prediction module 330 is used to: predict fault information of potential faults in the target business project; the fault information includes at least one of the following: fault type, fault location, fault probability and fault impact range; the prediction module 330 is also used to: comprehensively analyze the fault information of the predicted potential faults to determine the risk level of the predicted potential faults.

[0131] According to an embodiment of this application, the prediction module 330 is further configured to: predict the degree of impact of code changes in a target business project based on a pre-trained first large model and the obtained historical feature time series; determine the amount of reserve resources for the target business project in a future time window based on the predicted degree of impact of code changes; wherein the predicted degree of impact of code changes is positively correlated with the determined amount of reserve resources.

[0132] According to an embodiment of this application, the prediction module 330 is used to: determine the amount of reserve resources for the target business project in a future time window based on the predicted impact of the code change and the current reserve resource amount of the target business project when the predicted impact of the code change is greater than a preset impact threshold; the determined reserve resource amount is greater than the current reserve resource amount of the target business project.

[0133] According to an embodiment of this application, the prediction module 330 is further configured to: determine the range of resource quantity adjustment set for the target business project in a future time window based on the predicted resource demand and the determined reserve resource quantity.

[0134] According to an embodiment of this application, the fusion module 320 is used to: determine the historical resource usage indicators, historical operation logs, and historical code change information belonging to the same time window for the determined historical resource usage indicators, historical operation logs, and historical code change information; perform feature extraction on the historical resource usage indicators, historical operation logs, and historical code change information belonging to the same time window to obtain the features of historical resource usage indicators, historical operation logs, and historical code change information belonging to the same time window; and sort the features of historical resource usage indicators, historical operation logs, and historical code change information belonging to different time windows according to the time order between different time windows to obtain a historical feature time sequence.

[0135] According to an embodiment of this application, the fusion module 320 is used to: extract historical code change information features based on a pre-trained second model; the extracted historical code change information features include at least one of the following: code semantic features, code change content features, current code complexity features, and initial value of the degree of impact of code changes.

[0136] According to an embodiment of this application, the fusion module 320 is used to: extract historical operation log features based on a pre-trained third model; the extracted historical operation log features include at least one of the following: log parsing features, historical fault features, and log semantic features.

[0137] According to embodiments of this application, any plurality of modules among the determining module 310, fusion module 320, and prediction module 330 can be merged into one module, or any one of these modules can be split into multiple modules. Alternatively, at least a portion of the functionality of one or more of these modules can be combined with at least a portion of the functionality of other modules and implemented in one module. According to embodiments of this application, at least one of the determining module 310, fusion module 320, and prediction module 330 can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging circuitry, or implemented in any one of software, hardware, and firmware methods, or in a suitable combination of any of these. Alternatively, at least one of the determining module 310, fusion module 320, and prediction module 330 can be at least partially implemented as a computer program module, which, when run, can perform corresponding functions.

[0138] The explanation of the above device embodiments can be found in the explanation of other embodiments, and the various operations in the above method embodiments can be executed by the corresponding modules in the above device embodiments.

[0139] Figure 4 A block diagram schematically illustrates an electronic device suitable for implementing a resource demand prediction method according to an embodiment of this application.

[0140] like Figure 4As shown, an electronic device 900 according to an embodiment of this application includes a processor 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage portion 908 into a random access memory (RAM) 903. The processor 901 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 901 may also include onboard memory for caching purposes. The processor 901 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of this application.

[0141] RAM 903 stores various programs and data required for the operation of electronic device 900. Processor 901, ROM 902, and RAM 903 are interconnected via bus 904. Processor 901 executes various operations of the method flow according to embodiments of this application by executing programs in ROM 902 and / or RAM 903. It should be noted that the programs may also be stored in one or more memories other than ROM 902 and RAM 903. Processor 901 may also execute various operations of the method flow according to embodiments of this application by executing programs stored in said one or more memories.

[0142] According to embodiments of this application, the electronic device 900 may further include an input / output (I / O) interface 905, which is also connected to a bus 904. The electronic device 900 may also include one or more of the following components connected to the input / output (I / O) interface 905: an input section 906 including a keyboard, mouse, etc.; an output section 907 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 908 including a hard disk, etc.; and a communication section 909 including a network interface card such as a LAN card, modem, etc. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the input / output (I / O) interface 905 as needed. A removable medium 911, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 910 as needed so that computer programs read from it can be installed into the storage section 908 as needed.

[0143] This application also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of this application.

[0144] According to embodiments of this application, the computer-readable storage medium can be a non-volatile computer-readable storage medium, such as including but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of this application, the computer-readable storage medium may include ROM 902 and / or RAM 903 and / or one or more memories other than ROM 902 and RAM 903 described above.

[0145] Embodiments of this application also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code is used to cause the computer system to implement any of the method embodiments provided in the embodiments of this application.

[0146] When the computer program is executed by the processor 901, it performs the functions defined in the system / apparatus of this application embodiment. According to the embodiments of this application, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0147] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and downloaded and installed via the communication section 909, and / or installed from a removable medium 911. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0148] In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 909, and / or installed from the removable medium 911. When the computer program is executed by the processor 901, it performs the functions defined in the system of this application embodiment. According to the embodiments of this application, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0149] According to embodiments of this application, program code for executing the computer programs provided in the embodiments of this application can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0150] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0151] Those skilled in the art will understand that the features described in the various embodiments of this application can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this application. In particular, the features described in the various embodiments of this application can be combined and / or combined in various ways without departing from the spirit and teachings of this application. All such combinations and / or combinations fall within the scope of this application.

Claims

1. A method for predicting resource demand, characterized in that, The method includes: For the target business project, identify historical resource usage metrics, historical operation logs, and historical code change information; For the determined historical resource usage indicators, historical operation logs, and historical code change information, feature fusion processing is performed to obtain a historical feature time series sequence. Different elements in the historical feature time series sequence correspond to different time windows, and the order of different elements is arranged according to the time order of the corresponding time window. Any element in the historical feature time series sequence includes: historical resource usage indicator features, historical operation log features, and historical code change information features within the corresponding time window. Based on the first pre-trained model, the resource requirements of the target business project within a future time window are predicted according to the obtained historical feature time series.

2. The method according to claim 1, characterized in that, The method further includes: Based on the first pre-trained model, and according to the obtained historical feature time series, potential failures in the target business project are predicted; If the predicted risk level of a potential failure exceeds a preset risk level threshold, an automatic failure handling operation will be performed on the target business item.

3. The method according to claim 2, characterized in that, The prediction of potential faults in the target business project includes: predicting fault information of potential faults in the target business project; the fault information includes at least one of the following: fault type, fault location, fault probability, and fault impact range; The method further includes: integrating the fault information of the predicted potential faults to determine the risk level of the predicted potential faults.

4. The method according to claim 1, characterized in that, The method further includes: Based on the first pre-trained model, the impact of code changes in the target business project is predicted according to the obtained historical feature time series. Based on the predicted impact of the code change, the amount of reserve resources for the target business project in the future time window is determined; wherein the predicted impact of the code change is positively correlated with the determined amount of reserve resources.

5. The method according to claim 4, characterized in that, The step of determining the amount of reserve resources for the target business project in the future time window based on the predicted impact of the code change includes: If the predicted impact of the code change is greater than a preset impact threshold, the reserve resource quantity of the target business project in the future time window is determined based on the predicted impact of the code change and the current reserve resource quantity of the target business project; the determined reserve resource quantity is greater than the current reserve resource quantity of the target business project.

6. The method according to claim 4, characterized in that, The method further includes: Based on the predicted resource demand and the determined reserve resources, determine the range of resource adjustments set for the target business project within the future time window.

7. The method according to claim 1, characterized in that, The process involves performing feature fusion processing on the determined historical resource usage indicators, historical operation logs, and historical code change information to obtain a historical feature time series sequence, including: For the identified historical resource usage indicators, historical operation logs, and historical code change information, determine the historical resource usage indicators, historical operation logs, and historical code change information belonging to the same time window respectively. Feature extraction is performed on historical resource usage indicators, historical operation logs, and historical code change information belonging to the same time window to obtain the features of historical resource usage indicators, historical operation logs, and historical code change information belonging to the same time window. The historical resource usage indicators, historical operation logs, and historical code change information characteristics belonging to different time windows are sorted according to the time order between different time windows to obtain a historical feature time series sequence.

8. The method according to claim 1 or 7, characterized in that, The methods for extracting the features of historical code change information include: Based on the pre-trained second model, historical code change information features are extracted; the extracted historical code change information features include at least one of the following: code semantic features, code change content features, current code complexity features, and initial value of the impact of code changes.

9. The method according to claim 1 or 7, characterized in that, The methods for extracting the features of the historical operation log include: Based on the pre-trained third model, historical operation log features are extracted from historical operation logs; the extracted historical operation log features include at least one of the following: log parsing features, historical fault features, and log semantic features.

10. A resource demand prediction device, characterized in that, The device includes: The determination module is used to determine historical resource usage metrics, historical operation logs, and historical code change information for target business projects. The fusion module is used to perform feature fusion processing on the determined historical resource usage indicators, historical operation logs, and historical code change information to obtain a historical feature time series sequence. Different elements in the historical feature time series sequence correspond to different time windows, and the order of different elements is arranged according to the time order of the corresponding time window. Any element in the historical feature time series sequence includes: historical resource usage indicator features, historical operation log features, and historical code change information features within the corresponding time window. The prediction module is used to predict the resource requirements of the target business project within a future time window based on a pre-trained first model and the obtained historical feature time series.

11. An electronic device, comprising: One or more processors; Memory, used to store one or more computer programs. The characteristic feature is that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 9.

12. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 9.

13. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 9.

Images