Cloud resource data prediction method, system, device, medium and product
By using conditional generation adversarial network and bidirectional gated loop unit combination model in cloud resource prediction, the problem of low accuracy of cloud resource prediction in the existing technology is solved, and more accurate cloud resource data prediction and container cloud resource management are achieved.
Patent Information
- Application Number
- CN202411139430.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-19
- Publication Date
- 2024-12-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing cloud resource prediction methods have low prediction accuracy under complex actual conditions, making it difficult to accurately deal with fluctuations and uncertainties in cloud resource data.
The combined model of conditional generation adversarial network (CGAN) and bidirectional gating cycle unit (BiGRU) is used to preprocess and train historical cloud resource data with conditional variables to build a cloud resource prediction model to improve prediction accuracy.
By introducing CGAN and BiGRU networks with conditional tags, data timing relationships and multivariate interactions can be captured more accurately, improving the prediction accuracy of cloud resource data and achieving more accurate container cloud resource allocation and management.
Smart Images

Figure CN119089305A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of cloud resource data prediction, and in particular to a cloud resource data prediction method, system, device, medium and product. Background Art
[0002] With the rapid progress and widespread application of network technology and computer technology, cloud computing technology has become the main tool for data transmission, which enables numerous computing and network resources to be shared and allocated and predicted according to demand through a virtualized capital pool. In the field of cloud resource computing, cloud resource prediction plays a key role. By predicting the data in the cloud database for a long period and a short period, the purpose is to control the jitter amplitude of cloud resource data, so as to operate the platform safely, reliably and efficiently. Although experts and scholars from various countries have always attached importance to and studied the technology of cloud resource prediction, the complex actual situation leads to low prediction accuracy, so it is particularly important to improve the prediction accuracy.
[0003] Cloud resource data is classified as time series samples, and traditional prediction methods include physical models and statistical analysis methods. However, due to the actual fluctuations and uncertainties of cloud resource data, traditional physical models are susceptible to multiple interferences, and traditional statistical methods are still limited in their prediction accuracy when dealing with these complex factors. Therefore, developing an innovative cloud resource prediction model has great research value and practical application potential, which can not only improve the utilization efficiency of cloud resources and optimize user experience, but also promote the further development of related technologies, enhance the market competitiveness of cloud service providers, and promote the sustainable development of the cloud computing industry. Summary of the invention
[0004] The purpose of this application is to provide a cloud resource data prediction method, system, device, medium and product to improve the accuracy of cloud resource prediction.
[0005] To achieve the above objectives, this application provides the following solutions:
[0006] In a first aspect, the present application provides a cloud resource data prediction method, comprising:
[0007] Obtain cloud resource data for a historical time period; the cloud resource data may be CPU utilization, memory utilization, inbound network traffic, outbound network traffic, or disk utilization;
[0008] Preprocessing the cloud resource data of the historical time period to obtain processed cloud resource data of the historical time period;
[0009] According to the processed cloud resource data of the historical time period, the cloud resource data at the current moment is determined using a cloud resource prediction model; wherein the cloud resource prediction model is obtained by training a combination model using a training data set in combination with conditional variables; the combination model includes a conditional generative adversarial network and a bidirectional gated recurrent unit connected in sequence; the training data set includes cloud resource data for the historical time period for training and cloud resource data at the corresponding prediction moment; the conditional variables are date and time.
[0010] Optionally, preprocessing the cloud resource data of the historical time period to obtain processed cloud resource data of the historical time period specifically includes:
[0011] Based on usage, request volume, response time and resource types, the cloud resource data of the historical time period is screened and classified to obtain the classified cloud resource data of the historical time period; the resource types include computing resources, storage resources and network resources;
[0012] Performing data cleaning, filling missing values and removing outliers on the classified cloud resource data of the historical time period to obtain the filled and removed cloud resource data of the historical time period;
[0013] The cloud resource data after padding and removal of the historical time period is normalized using the MinMaxScaler function to obtain the processed cloud resource data of the historical time period.
[0014] Optionally, based on usage, request volume, response time, and resource type, the cloud resource data for the historical time period is screened and classified to obtain the classified cloud resource data for the historical time period, specifically including:
[0015] Filtering the usage, request volume, and response time of the cloud resource data in the historical time period to obtain filtered cloud resource data in the historical time period;
[0016] The filtered cloud resource data of the historical time period is classified based on resource types to obtain the classified cloud resource data of the historical time period.
[0017] Optionally, the combined model is trained using a training data set, specifically including:
[0018] Build training datasets and ensemble models;
[0019] Preprocessing the cloud resource data of the training historical time period to obtain processed cloud resource data of the training historical time period;
[0020] The combined model is trained with the processed cloud resource data and conditional variables of the training historical time period as input and the cloud resource data corresponding to the prediction moment as output to obtain the cloud resource prediction model.
[0021] In a second aspect, the present application provides a cloud resource data prediction system, comprising:
[0022] A data acquisition module is used to acquire cloud resource data of a historical time period; the cloud resource data is CPU utilization, memory utilization, inbound network traffic, outbound network traffic or disk utilization;
[0023] A data processing module, used to pre-process the cloud resource data of the historical time period to obtain processed cloud resource data of the historical time period;
[0024] A prediction module is used to determine the cloud resource data at the current moment based on the processed cloud resource data of the historical time period using a cloud resource prediction model; wherein the cloud resource prediction model is obtained by training a combination model using a training data set in combination with conditional variables; the combination model includes a conditional generative adversarial network and a bidirectional gated recurrent unit connected in sequence; the training data set includes cloud resource data for the historical time period for training and cloud resource data at the corresponding prediction moment; the conditional variables are date and time.
[0025] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the cloud resource data prediction methods described above.
[0026] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the cloud resource data prediction methods described above.
[0027] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements any of the cloud resource data prediction methods described above.
[0028] According to the specific embodiments provided in this application, this application discloses the following technical effects:
[0029] The present application provides a cloud resource data prediction method, system, device, medium and product, firstly obtaining cloud resource data of a historical time period; the cloud resource data is CPU utilization, memory utilization, inlet network traffic, outlet network traffic or disk utilization; preprocessing the cloud resource data of the historical time period to obtain processed cloud resource data of the historical time period; according to the processed cloud resource data of the historical time period, using a cloud resource prediction model, determining the cloud resource data at the current moment; wherein the cloud resource prediction model is obtained by training a combined model using a training data set combined with conditional variables; the combined model includes a conditional generative adversarial network and a bidirectional gated recurrent unit connected in sequence. In the present application, the prediction accuracy of cloud resource data is improved by introducing a conditional generative adversarial network with conditional labels and a bidirectional gated recurrent unit network. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0031] Figure 1 A flowchart of a cloud resource data prediction method provided in one embodiment of the present application;
[0032] Figure 2 A training flowchart for combining models in practical applications;
[0033] Figure 3 This is the architecture diagram of the combined model;
[0034] Figure 4 This is an application environment diagram of a cloud resource data prediction method in one embodiment of the present application;
[0035] Figure 5 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION
[0036] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0037] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.
[0038] In order to solve the problems of low efficiency, instability and inability to make targeted predictions in traditional cloud resource data prediction methods, an optimized combined model was invented. This combined model uses the advanced technologies of conditional generative adversarial networks (CGAN) and bidirectional gate recurrent units (BiGRU), which can accurately capture the temporal relationship of data and the interaction between multiple variables in a limited time sample sequence. By preprocessing the original data and building a reasonable and efficient prediction model, and filtering according to the conditional tags specified by the user, it can make more accurate predictions on the allocation and management of container cloud resources based on the collected historical cloud resource data.
[0039] In an exemplary embodiment, Figure 1 As shown, a cloud resource data prediction method is provided, comprising:
[0040] S1: Obtain cloud resource data for a historical time period; the cloud resource data includes CPU utilization, memory utilization, inbound network traffic, outbound network traffic, or disk utilization.
[0041] S2: Preprocessing the cloud resource data of the historical time period to obtain processed cloud resource data of the historical time period.
[0042] As an optional implementation, S2 specifically includes:
[0043] S21: Based on usage, request volume, response time and resource type, the cloud resource data of the historical time period is screened and classified to obtain the classified cloud resource data of the historical time period; the resource types include computing resources, storage resources and network resources. The resource types are computing resources, storage resources, network resources, etc., specifically CPU usage, memory usage, ingress / egress traffic, and disk usage.
[0044] S21, specifically including:
[0045] The cloud resource data in the historical time period is filtered for usage, request volume, and response time to obtain filtered cloud resource data in the historical time period.
[0046] The filtered cloud resource data of the historical time period is classified based on resource types to obtain the classified cloud resource data of the historical time period.
[0047] S22: performing data cleaning, filling in missing values and removing outliers on the classified cloud resource data of the historical time period to obtain the filled and removed cloud resource data of the historical time period.
[0048] S23: using the MinMaxScaler function to normalize the cloud resource data after padding and removal in the historical time period to obtain processed cloud resource data in the historical time period.
[0049] S3: Determine the cloud resource data at the current moment according to the processed cloud resource data of the historical time period using a cloud resource prediction model; wherein the cloud resource prediction model is obtained by training a combination model using a training data set in combination with conditional variables; the combination model includes a conditional generative adversarial network and a bidirectional gated recurrent unit connected in sequence; the training data set includes cloud resource data for the historical time period for training and cloud resource data at the corresponding prediction moment; the conditional variables are date and time.
[0050] As an optional implementation method, the combined model is trained using a training data set, specifically including:
[0051] Build training datasets and ensemble models.
[0052] The cloud resource data of the training historical time period is preprocessed to obtain processed cloud resource data of the training historical time period.
[0053] The combined model is trained with the processed cloud resource data and conditional variables of the training historical time period as input and the cloud resource data corresponding to the prediction moment as output to obtain the cloud resource prediction model.
[0054] In practical applications, such as Figure 2 As shown, the training process of the combined model is as follows:
[0055] Step 1: Collect and prepare historical cloud resource utilization data for screening and classification, including the usage of various resources, request volume, response time, etc.
[0056] The relationship between CPU utilization and other characteristic factors is fully considered, and according to the clock series change law of CPU utilization, each sampling point has four characteristics, including memory utilization, inlet network traffic, outlet network traffic and disk utilization.
[0057] Step 2: Preprocess the cloud resource data that meets the user's prediction needs after screening and classification.
[0058] In practical applications, methods of data cleaning, filling missing values and removing outliers were adopted, and the data were normalized to ensure that they all fell in the same numerical range, thereby reducing the interference of different dimensions on the model.
[0059] Use MinMaxScaler to normalize the data and map the data to the interval [0, 1]. The calculation formula of MinMaxScaler is:
[0060]
[0061] X scaled =X std ×(X max -X min )+X min .
[0062] In the above formula, X max and X min are the maximum and minimum values in the data, respectively, std is the standardized result, X scaled is the final normalized data.
[0063] Step 3: Divide the processed data into training set, validation set and test set, using the first part of the time series data as the training set, the middle part as the validation set, and the last part as the test set. In practical applications, the ratio of 7:2:1 is used to leave an appropriate amount of data for subsequent model training.
[0064] Step 4: Data is extracted by using the conditional generative adversarial network (CGAN). The input of the generator includes both past data and conditional variables, and the final output is the generated data. In order to enhance the stability of training, a gradient penalty function is added to the discriminator, where the input data is real and the generated data is used as the discriminant output.
[0065] like Figure 3 As shown in Figure 1, in the conditional generative adversarial network CGAN, the generator and discriminator are represented by differentiable functions G and D, respectively. The inputs of these two functions are random variables z and actual data x, respectively. For each generator, there is a set of random vectors used to describe its generation process. The sample data generated by the generator is marked as G(z). The generated sample data and the actual data form a synthetic sample y, and the probability distribution of the actual data is called P data; The discriminator is used to determine whether the sample data generated by the generator is greater than or equal to the true value. When the discriminator inputs actual data, the output result of the discriminator will be marked as 1; if the input data is the G(z) value generated by the generator, then the output value of the discriminator will be marked as 0; if the input is not the data type generated by the generator, the output label is not zero; if the input is a virtual sample, the output of the discriminator will not appear; the main purpose of the generator is to simulate the complex distribution pattern of real sample data, so as to achieve the purpose of "deceiving" the discriminator function; the interaction between the generator and the discriminator actually constitutes a constantly changing game process. Under the most ideal conditions, the entire CGAN framework will converge, and the generator will continue to generate fake information that is highly similar to the actual data until the discriminator cannot accurately identify the source of the data. In this process, the network training has been completed and the generator has mastered the distribution pattern of the real data. The recognition performance of the discriminator when generating virtual samples will become more stable, so that the adversarial system can reach a local optimal solution.
[0066] In the iterative process, when the training of the generator G reaches a certain level, its parameters need to be fixed, and then the discriminator D is optimized; this process is carried out by minimizing the cross entropy, and the loss function is:
[0067]
[0068]
[0069] Among them, θ D is the parameter set of the discriminator D; θ G is the parameter set of the generator G; is the objective function of the discriminator, used to adjust θ D To maximize its accuracy; is the game goal between the generator G and the discriminator D, where the discriminator classifies the real and generated data as accurately as possible, while the generator tries to make the discriminator misclassify the generated data as much as possible; x comes from the real data distribution P data (x), z comes from the prior distribution P z (z), For the real data distribution P data The expected value of the data x in; D(x) is the probability that the real data x is identified as real data by the discriminator D; is the expected value of a random vector z in the noise distribution Pz; D(G(z)) is the probability that the data G(z) generated by the generator G from the random vector z is identified as false data by the discriminator D.
[0070] Step 5: Integrate conditional information in the generator and discriminator to ensure that the bidirectional gated recurrent unit BiGRU can use this information to make more accurate predictions. Secondly, introduce a gradient regularization term in the loss function to alleviate the mode collapse problem in model training.
[0071] like Figure 3 As shown, BiGRU has the ability to process both forward and reverse information of the input sequence at the same time. It exhibits stronger generalization performance when processing sequence data of different lengths and structures, which helps to more accurately identify the various factors that cause the current load changes, thereby improving the accuracy of feature vector extraction. The output of the BiGRU neural network is shown below:
[0072]
[0073]
[0074]
[0075] in, represents the state of the forward hidden layer at time t, shows the state of the forward hidden layer at time t-1, represents the state of the reverse hidden layer at time t, represents the state of the reverse hidden layer at time t-1; h t represents the prediction result of the model input at time t; w t represents the weight of the forward hidden layer state at time t, v t represents the weight of the reverse hidden layer state at time t; b t Represents the bias of the hidden layer state at time t.
[0076] Step 6: Use historical data for training and monitor the performance of the model. Use the validation set to evaluate the model and make adjustments as needed.
[0077] First, we preprocess the data by collecting and preparing historical cloud resource utilization data. Then, we use the CGAN model to extract it. The input to the generator includes historical data and conditional variables, and the output is generated data. At the same time, we introduce gradient penalty GP in the discriminator to improve the stability of training. Secondly, we need to integrate conditional information in the generator and discriminator to ensure that BiGRU can use this information to make more accurate predictions. Finally, we use historical data for training and monitor the performance of the model. We use the validation set to evaluate the model and make adjustments as needed.
[0078] When evaluating the pros and cons of resource prediction strategies, we choose to start from five dimensions: CPU utilization, memory utilization, inbound network traffic, outbound network traffic, and disk utilization. The optimization goal is to achieve shorter time, lower cost, and more balanced load. Therefore, a fitness formula is formulated, and the global optimization characteristics of the BiGRU algorithm are used to find a prediction strategy that better meets the needs. As an indicator for evaluating the comprehensive load rate, factors such as the node's CPU utilization, memory utilization, inbound network traffic, outbound network traffic, and disk utilization are considered.
[0079] This application has the following advantages:
[0080] (1) Compared with traditional cloud resource prediction methods, this application introduces a conditional generative adversarial network with conditional labels, so that the generator can generate samples that meet the conditions more specifically, allowing the generator to screen cloud resource data and limit the type of cloud resource data, thereby achieving the effect of improving prediction accuracy.
[0081] (2) Compared with the traditional gated recurrent unit in data extraction and processing, the present application adds a bidirectional gated recurrent unit network. The information acquisition and processing level of the conventional gate controller recurrent unit is the same. By adding a bidirectional gate controller recurrent unit network, the present application can directly link two hidden layers with completely opposite input directions to an output layer; in this way, the data layer can obtain detailed information about past and future conditions in order to estimate the data more accurately.
[0082] The cloud resource data prediction method provided in the embodiment of the present application can be applied to Figure 4 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the cloud resource data to be processed to the server 104. After the server 104 receives the cloud resource data to be processed, the server 104 pre-processes the cloud resource data to be processed based on the cloud resource data of the historical time period to obtain the processed cloud resource data of the historical time period; according to the processed cloud resource data of the historical time period, the cloud resource data at the current moment is determined by using the cloud resource prediction model. The server 104 can feed back the obtained cloud resource data at the current moment to the terminal 102. In addition, in some embodiments, the cloud resource data prediction method can also be implemented by the server 104 or the terminal 102 alone, such as the terminal 102 can directly process the cloud resource data to be processed, or the server 104 can obtain the cloud resource data to be processed from the data storage system and process the cloud resource data to be processed.
[0083] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart vehicle-mounted devices, etc. The portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers, or may be a cloud server.
[0084] In an exemplary embodiment, a cloud resource data prediction system is provided, comprising:
[0085] The data acquisition module is used to acquire cloud resource data for a historical time period; the cloud resource data is CPU utilization, memory utilization, inbound network traffic, outbound network traffic or disk utilization.
[0086] The data processing module is used to pre-process the cloud resource data of the historical time period to obtain the processed cloud resource data of the historical time period.
[0087] A prediction module is used to determine the cloud resource data at the current moment based on the processed cloud resource data of the historical time period using a cloud resource prediction model; wherein the cloud resource prediction model is obtained by training a combination model using a training data set in combination with conditional variables; the combination model includes a conditional generative adversarial network and a bidirectional gated recurrent unit connected in sequence; the training data set includes cloud resource data for the historical time period for training and cloud resource data at the corresponding prediction moment; the conditional variables are date and time.
[0088] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store cloud resource data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a cloud resource data prediction method is implemented.
[0089] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the above-mentioned cloud resource data prediction method when executing the computer program.
[0090] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the above-mentioned cloud resource data prediction method when executed by a processor.
[0091] In an exemplary embodiment, a computer program product is provided, including a computer program, which implements the above-mentioned cloud resource data prediction method when executed by a processor.
[0092] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.
[0093] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).
[0094] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.
[0095] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0096] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.
Claims
1. A cloud resource data prediction method, It is characterized in that include: Obtain cloud resource data for a historical time period; the cloud resource data may be CPU utilization, memory utilization, inbound network traffic, outbound network traffic, or disk utilization; Preprocessing the cloud resource data of the historical time period to obtain processed cloud resource data of the historical time period; According to the processed cloud resource data of the historical time period, the cloud resource data at the current moment is determined using a cloud resource prediction model; wherein the cloud resource prediction model is obtained by training a combination model using a training data set in combination with conditional variables; the combination model includes a conditional generative adversarial network and a bidirectional gated recurrent unit connected in sequence; the training data set includes cloud resource data for the historical time period for training and cloud resource data at the corresponding prediction moment; the conditional variables are date and time.
2. The cloud resource data prediction method according to claim 1, It is characterized in that Preprocessing the cloud resource data of the historical time period to obtain processed cloud resource data of the historical time period specifically includes: Based on usage, request volume, response time and resource types, the cloud resource data of the historical time period is screened and classified to obtain the classified cloud resource data of the historical time period; the resource types include computing resources, storage resources and network resources; Performing data cleaning, filling missing values and removing outliers on the classified cloud resource data of the historical time period to obtain the filled and removed cloud resource data of the historical time period; The cloud resource data after padding and removal of the historical time period is normalized using the MinMaxScaler function to obtain the processed cloud resource data of the historical time period.
3. The cloud resource data prediction method according to claim 2, It is characterized in that Based on usage, request volume, response time, and resource type, the cloud resource data for the historical time period is screened and classified to obtain the classified cloud resource data for the historical time period, specifically including: Filtering the usage, request volume, and response time of the cloud resource data in the historical time period to obtain filtered cloud resource data in the historical time period; The filtered cloud resource data of the historical time period is classified based on resource types to obtain the classified cloud resource data of the historical time period.
4. The cloud resource data prediction method according to claim 1, It is characterized in that The combined model is trained using the training data set, including: Build training datasets and ensemble models; Preprocessing the cloud resource data of the training historical time period to obtain processed cloud resource data of the training historical time period; The combined model is trained with the processed cloud resource data and conditional variables of the training historical time period as input and the cloud resource data corresponding to the prediction moment as output to obtain the cloud resource prediction model.
5. A cloud resource data prediction system, It is characterized in that include: A data acquisition module is used to acquire cloud resource data of a historical time period; the cloud resource data is CPU utilization, memory utilization, inbound network traffic, outbound network traffic or disk utilization; A data processing module, used to pre-process the cloud resource data of the historical time period to obtain processed cloud resource data of the historical time period; A prediction module is used to determine the cloud resource data at the current moment based on the processed cloud resource data of the historical time period using a cloud resource prediction model; wherein the cloud resource prediction model is obtained by training a combination model using a training data set in combination with conditional variables; the combination model includes a conditional generative adversarial network and a bidirectional gated recurrent unit connected in sequence; the training data set includes cloud resource data for the historical time period for training and cloud resource data at the corresponding prediction moment; the conditional variables are date and time.
6. A computer device, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the cloud resource data prediction method according to any one of claims 1 to 4.
7. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the cloud resource data prediction method according to any one of claims 1 to 4 is implemented.
8. A computer program product comprising a computer program, It is characterized in that When the computer program is executed by a processor, the cloud resource data prediction method according to any one of claims 1 to 4 is implemented.
Citation Information
Cited By
Private cloud resource demand prediction and configuration method and system based on space-time sequence
CN119814590A
