Cloud platform resource dynamic calling automation operation method and system based on AI prediction model

Abstract

The application relates to the technical field of platform operation and maintenance, and discloses a cloud platform resource dynamic calling automatic operation and maintenance method and system based on an AI prediction model, which comprises a resource data acquisition module, an AI prediction analysis module, a resource dynamic scheduling module, an operation and maintenance monitoring module and an operation and maintenance execution module. The AI prediction analysis module adopts a hybrid neural network model fusing LSTM and CNN, and through online incremental learning optimization, can accurately predict resource demand, and the prediction accuracy is quantitatively evaluated by average absolute error. The resource dynamic scheduling module realizes elastic scheduling according to a priority strategy based on the prediction result and real-time resource state. The modules are cooperatively linked to form a closed loop of "data acquisition-prediction analysis-resource scheduling-monitoring operation and maintenance-data feedback", realize full-process automatic operation and maintenance, effectively improve cloud platform resource utilization and business stability, reduce manual operation and maintenance cost, adapt to various enterprise-level cloud platform complex operation and maintenance scenes, and are excellent in practicability and expansibility.

Description

Technical Field

[0001] This invention belongs to the field of platform operation and maintenance technology, specifically referring to an automated operation and maintenance method and system for dynamic resource retrieval on cloud platforms based on AI prediction models. Background Technology

[0002] Cloud computing technology has become a core infrastructure for enterprise digital transformation, supporting the operation and data storage needs of various business systems. The scheduling and operation efficiency of cloud platform resources (including computing, storage, and network resources) directly determines the stability, reliability, and resource utilization of business operations.

[0003] Existing cloud platform resource operation and maintenance methods mostly adopt static configuration or threshold-triggered semi-automated scheduling modes, which have obvious limitations: On the one hand, static configuration cannot adapt to the dynamic changes in business load, easily leading to insufficient resources during peak hours causing business outages, or idle resources during off-peak hours causing waste; on the other hand, threshold-triggered scheduling is a "post-event response," lacking the ability to predict future resource needs, with high scheduling latency, and is difficult to cope with sudden fluctuations in business load. At the same time, traditional operation and maintenance relies on manual intervention, which is not only costly and inefficient, but also prone to business failures due to human error, failing to meet the refined operation and maintenance needs of large-scale cloud platforms.

[0004] Although some solutions introduce simple predictive models to assist scheduling, existing models mostly use a single algorithm, which makes it difficult to capture both the temporal and burst characteristics of business load at the same time, resulting in limited prediction accuracy. Furthermore, each operation and maintenance module is independent of the others, lacking data closure and collaborative linkage, which leads to insufficient continuity and accuracy of resource scheduling and operation and maintenance actions, and makes it impossible to achieve a fully automated operation and maintenance closure loop.

[0005] Therefore, there is an urgent need for a technical solution that integrates AI prediction capabilities, enables dynamic resource allocation and fully automated operation and maintenance, and solves the problems of low resource utilization, slow response and high operation and maintenance costs in the operation and maintenance of existing cloud platforms. Summary of the Invention

[0006] In response to the above situation, the present invention aims to overcome the shortcomings of existing cloud platform resource operation and maintenance technologies, and provide an automated operation and maintenance system and method for dynamic resource scheduling of cloud platforms based on AI prediction models. By accurately predicting resource demand through AI prediction models, the system links multiple modules to achieve dynamic resource scheduling and fully automated operation and maintenance, thereby improving resource utilization, business stability and operation and maintenance efficiency, and reducing the cost of manual intervention.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: an automated operation and maintenance system for dynamic resource allocation on a cloud platform based on an AI prediction model, comprising a resource data acquisition module, an AI prediction and analysis module, a resource dynamic scheduling module, an operation and maintenance monitoring module, and an operation and maintenance execution module; The resource data acquisition module is used to collect various resource operation data, business load data and historical operation and maintenance data of the cloud platform. After cleaning and standardizing the collected data, it is synchronized to the AI ​​prediction and analysis module and the operation and maintenance monitoring module. The AI ​​prediction and analysis module includes a prediction model training unit and a resource demand prediction unit. The prediction model training unit trains and optimizes the AI ​​prediction model based on historical data synchronized by the resource data acquisition module. The resource demand prediction unit uses the trained AI prediction model to predict the changes in business load and resource demand of the cloud platform within a preset time period in the future, and outputs the resource demand prediction results to the resource dynamic scheduling module. The resource dynamic scheduling module receives the resource demand prediction results output by the AI ​​prediction and analysis module, and formulates a dynamic resource allocation strategy based on the current resource occupancy status of the cloud platform. It performs cross-node scheduling, elastic scaling and shrinking, and resource allocation adjustment of the computing resources, storage resources and network resources of the cloud platform, and feeds back the resource scheduling results to the operation and maintenance monitoring module. The operation and maintenance monitoring module monitors the cloud platform's resource operation status, resource scheduling execution, and business operation stability in real time. It generates abnormal alarm information by comparing with preset thresholds and synchronizes the monitoring data and alarm information to the AI ​​prediction and analysis module and the operation and maintenance execution module, respectively, to provide a basis for AI prediction model optimization and operation and maintenance action execution. The operation and maintenance execution module receives alarm information from the operation and maintenance monitoring module and scheduling assistance instructions from the resource dynamic scheduling module. It automatically executes operation and maintenance actions such as fault repair, configuration optimization, log archiving, and operation and maintenance report generation. The execution results are back-synchronized to the resource data acquisition module to form a data closed loop.

[0008] As a further description of the above technical solution: The resource data acquisition module is used to comprehensively collect various types of data during the operation of the cloud platform, providing data support for AI prediction, resource scheduling, and operation and maintenance monitoring. The collection scope includes cloud platform resource operation data (CPU utilization, memory usage, disk I / O rate, network bandwidth usage, etc.), business load data (business request volume, response latency, concurrent users, etc.), and historical operation and maintenance data (historical scaling records, fault handling logs, configuration change records, etc.).

[0009] During the data collection process, this module cleans (removes outliers and missing values) and standardizes (unifies data format and magnitude) the raw data to prevent dirty data from affecting the accuracy of subsequent analysis. It also supports dynamically adjusting the collection frequency based on business needs, using high-frequency collection (e.g., once per second) for core business scenarios and low-frequency collection (e.g., once per minute) for non-core scenarios, thus reducing resource consumption while ensuring data timeliness. The processed data is then synchronized to the AI ​​prediction and analysis module and the operation and maintenance monitoring module.

[0010] As a further description of the above technical solution: The AI ​​predictive analysis module is the core predictive unit of the system, including a predictive model training unit and a resource demand prediction unit. It achieves accurate prediction of the future resource demand of the cloud platform through a hybrid neural network model.

[0011] The prediction model training unit employs a hybrid neural network model that integrates LSTM and CNN networks. The LSTM network captures the temporal characteristics of business load (such as daily and weekly load variation patterns), while the CNN network extracts bursty characteristics of load (such as sudden business requests and abnormal load fluctuations). Together, they enhance the model's adaptability to complex load changes. Model optimization utilizes an online incremental learning approach, continuously iterating and updating parameters based on real-time new data synchronized from the resource data acquisition module. This prevents model aging and ensures prediction accuracy.

[0012] The resource demand forecasting unit uses a trained hybrid neural network model to predict the resource demand of the cloud platform within a preset time period (which can be customized to 1 hour, 6 hours, 24 hours, etc.), outputting prediction results such as peak demand, off-peak demand, and sudden demand. To quantify the model's prediction accuracy, the mean absolute error (MAE) is used for evaluation, calculated using the following formula:

[0013] Where n is the number of prediction samples. Let i be the actual resource requirement value for the i-th sample. The model predicts the resource demand value for the i-th sample. The prediction error is quantified by this formula, providing a quantitative basis for model iterative optimization. When the error exceeds the preset range, the model emergency optimization process is triggered to ensure the reliability of the prediction results. The prediction results are synchronized to the resource dynamic scheduling module.

[0014] As a further description of the above technical solution: The resource dynamic scheduling module receives the resource demand prediction results output by the AI ​​prediction and analysis module, combines them with the current resource occupancy status of the cloud platform synchronized by the operation and maintenance monitoring module, formulates and executes dynamic resource allocation strategies, and achieves precise allocation and elastic scheduling of resources.

[0015] The strategy formulation process follows priority adaptation rules, allocating the highest resource scheduling priority to core businesses (such as transaction systems and core databases) to ensure their resource supply and avoid impacting business operations due to insufficient resources. Simultaneously, it aims to maximize resource utilization by allocating idle resources across nodes to those with higher loads, preventing resource waste. Scheduling operations include elastic scaling of computing, storage, and network resources, cross-node migration, and resource quota adjustments. Elastic scaling employs a smooth scaling strategy, gradually increasing or decreasing resource amounts to avoid resource fluctuations affecting business operations; the scheduling process is completely transparent to the business. After scheduling is complete, this module feeds back the resource scheduling results to the operations and maintenance monitoring module for subsequent monitoring and verification.

[0016] As a further description of the above technical solution: The operations and maintenance monitoring module is used to monitor the entire cloud platform status in real time, providing real-time feedback for resource scheduling and operations and maintenance execution, while also generating monitoring data to support the optimization of AI prediction models. The monitoring scope includes resource operating status (real-time indicators such as CPU, memory, disk, and network), resource scheduling execution status (scaling progress, resource migration results, quota adjustment effects, etc.), and business operation stability (response latency, error rate, concurrency, etc.).

[0017] This module sets up multi-level alarm thresholds, including early warning thresholds, emergency thresholds, and critical thresholds: Early warning thresholds alert potential resource or business risks and are synchronized to the AI ​​predictive analytics module for optimizing model prediction parameters; Emergency thresholds trigger automated operation and maintenance response processes and are synchronized to the operation and maintenance execution module for troubleshooting and repair; Critical thresholds, while triggering automated operation and maintenance, push emergency alarms to operation and maintenance personnel to ensure business security in extreme scenarios. Monitoring data and alarm information are synchronized in real time to the AI ​​predictive analytics module, resource dynamic scheduling module, and operation and maintenance execution module, enabling collaborative operation among these modules.

[0018] As a further description of the above technical solution: The operations and maintenance (O&M) execution module receives alarm information from the O&M monitoring module and scheduling assistance instructions from the resource dynamic scheduling module, automatically executing the entire O&M process to achieve automation and standardization of O&M operations. Core O&M actions include fault repair (such as abnormal resource restarts, service self-healing, data recovery, etc.), configuration optimization (such as resource quota adjustments, service parameter optimization, network configuration changes, etc.), log archiving (classified archiving and backup of O&M operation logs, fault logs, and monitoring logs), and O&M report generation (regularly generating reports on resource utilization, O&M fault statistics, and prediction accuracy analysis, etc.).

[0019] This module features customizable operation and maintenance (O&M) strategies, allowing users to preset different automated O&M rules based on business scenarios (such as repair processes for specific faults, configuration templates for core businesses, etc.). For configuration changes after resource scheduling, it automatically executes configuration verification and rollback mechanisms to verify the validity and compliance of configuration changes. If configuration errors or business anomalies occur, a rollback operation is immediately triggered to restore the system to its pre-change state, ensuring the security of O&M actions. After the O&M actions are completed, this module synchronizes the execution results back to the resource data acquisition module, incorporating them into the data loop as historical O&M data, providing data support for AI prediction model training and optimization.

[0020] As a further description of the above technical solution: This method is applied to the above system to achieve dynamic resource allocation and fully automated operation and maintenance of the cloud platform. The specific steps are as follows: S1: Data Acquisition and Preprocessing. The resource data acquisition module collects resource operation data, business load data, and historical operation and maintenance data from the cloud platform. After cleaning and standardization, the data is synchronized to the AI ​​predictive analysis module and the operation and maintenance monitoring module to ensure data quality and timeliness.

[0021] S2: AI Model Training and Resource Demand Prediction. The AI ​​predictive analysis module trains and optimizes a hybrid neural network model using an online incremental learning method based on synchronous historical and real-time data through the predictive model training unit. Then, through the resource demand prediction unit, it uses the trained model to predict resource demand within a preset time period in the future, outputting the prediction results of resource peak, trough, and sudden demand. The prediction accuracy is evaluated using the MAE formula, and if the error exceeds the range, emergency model optimization is triggered.

[0022] S3: Dynamic Resource Scheduling. Based on the prediction results and the current resource status synchronized by the operation and maintenance monitoring module, the dynamic resource scheduling module formulates resource allocation strategies according to priority adaptation rules, performs operations such as elastic scaling and cross-node scheduling, adopts a smooth scaling strategy to avoid business impact, and feeds back the results to the operation and maintenance monitoring module after the scheduling is completed.

[0023] S4: Real-time end-to-end monitoring. The operations and maintenance monitoring module monitors resource operation, scheduling execution, and business stability in real time, compares multi-level alarm thresholds to generate monitoring data and alarm information, and synchronizes them to the corresponding modules to provide a basis for subsequent actions.

[0024] S5: Automated Operations and Maintenance Execution. Based on alarm information and scheduling auxiliary instructions, the operations and maintenance execution module automatically performs operations and maintenance actions such as fault repair, configuration optimization, and log archiving. After execution, the results are synchronized to the resource data acquisition module, completing one operations and maintenance loop.

[0025] S6: Iterative optimization. Repeat steps S1-S5 to achieve fully automated operation and maintenance of dynamic cloud platform resource allocation; the AI ​​predictive analysis module continuously optimizes model parameters based on real-time collected operation and maintenance data and monitoring data, gradually improving prediction accuracy and operation and maintenance precision.

[0026] The beneficial effects achieved by the present invention using the above structure are as follows: (1) In this invention, by integrating the hybrid neural network model of LSTM and CNN, the temporal and burst features of the business load are captured at the same time. The model is continuously optimized by combining online incremental learning, and the prediction error is controllable. Based on accurate prediction, resources are scheduled in advance to avoid insufficient resources during peak periods and idle resources during off-peak periods, thereby improving resource utilization compared to traditional methods.

[0027] (2) In this invention, through the collaborative linkage of multiple modules, the entire process from data collection, demand forecasting, resource scheduling to operation and maintenance execution is automated, which greatly reduces manual intervention, improves operation and maintenance efficiency, and avoids human error, thereby reducing the business failure rate. (3) It adopts a smooth scaling up and down, priority scheduling and multi-level alarm mechanism to ensure that the resource scheduling process is unaware of the business and the supply of core business resources is given priority. At the same time, it supports the customization of operation and maintenance policies and configuration rollback, adapts to cloud platform business scenarios of different industries and different scales, and has excellent compatibility and security. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the structure of an automated operation and maintenance system for dynamic resource allocation on a cloud platform based on an AI prediction model, as proposed in this invention. Figure 2 This is a flowchart illustrating an automated operation and maintenance method for dynamic resource allocation on a cloud platform based on an AI prediction model, as proposed in this invention. Detailed Implementation

[0029] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0030] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0031] This application presents an automated operation and maintenance method and system for dynamic resource allocation on a cloud platform based on an AI prediction model. This embodiment is applied to an enterprise-level cloud platform that hosts the enterprise's core transaction system, data analysis system, and office system. It needs to ensure stable 24 / 7 operation while controlling resource maintenance costs. The specific operation flow after deployment of the system and method based on this invention is as follows: Data Acquisition Phase: The resource data acquisition module collects platform resource operation data (CPU, memory, disk I / O, network bandwidth, etc.), business load data (transaction request volume, response latency, concurrent users, etc.), and historical operation and maintenance data (past expansion and contraction records, fault handling logs, etc.) at preset frequencies. A high-frequency acquisition mode is used for core business scenarios, while a low-frequency acquisition mode is used for non-core business scenarios. After cleaning and removing abnormal data, the data is standardized into a unified format and synchronized to the AI ​​prediction and analysis module and the operation and maintenance monitoring module.

[0032] Model Training and Prediction Phase: The AI ​​Predictive Analysis module's prediction model training unit trains a hybrid neural network model based on historical data. The LSTM layer captures the load time-series characteristics of the trading system during peak hours, while the CNN layer captures sudden load characteristics caused by promotional activities and other scenarios. Through online incremental learning, model parameters are continuously optimized based on real-time data. The resource demand prediction unit predicts resource demand within a preset time period, outputting prediction results such as peak-hour resource demand, off-peak-hour resource demand, and sudden load demand. The prediction error is evaluated using the MAE formula. If the error is below a preset threshold, the prediction results are synchronized to the resource dynamic scheduling module.

[0033] Resource scheduling phase: The dynamic resource scheduling module, based on the current platform resource occupancy status, formulates scheduling strategies: Before peak hours, it initiates elastic scaling up to increase resource quotas for the core transaction system; during off-peak hours, it initiates scaling down to reduce resource quotas; simultaneously, it allocates idle resources to the data analysis system with higher load. The scheduling process employs a smooth scaling up / down strategy, gradually adjusting resource amounts to avoid business fluctuations. After scheduling is complete, the results are fed back to the operations and maintenance monitoring module.

[0034] Monitoring and Operation Phase: The operation and maintenance monitoring module monitors the resource scheduling progress and business operation status in real time. During peak hours, the resource utilization rate is maintained within a reasonable range, and no alarms are triggered. When the sudden load exceeds the predicted value in the afternoon and the resource utilization rate reaches the warning threshold, the relevant information is synchronized to the AI ​​prediction model for parameter optimization. At night, a disk I / O anomaly occurs on a certain node, triggering the emergency threshold. The operation and maintenance execution module automatically performs disk detection and fault repair. After the repair is completed, the results are synchronized to the resource data acquisition module and included in historical data.

[0035] Iterative optimization phase: The system executes the above process in a loop, generates operation and maintenance reports regularly, and continuously optimizes the AI ​​prediction model based on the actual daily load and prediction data. The prediction error gradually decreases, the resource utilization rate is significantly improved compared with the traditional operation and maintenance method, the workload of manual operation and maintenance is greatly reduced, and the business response latency is significantly reduced, achieving the dual goals of efficient resource utilization and stable business operation.

[0036] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0037] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

[0038] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A cloud platform resource dynamic allocation automated operation and maintenance system based on an AI prediction model, characterized in that, It includes a resource data acquisition module, an AI predictive analysis module, a resource dynamic scheduling module, an operation and maintenance monitoring module, and an operation and maintenance execution module; The resource data acquisition module is used to collect various resource operation data, business load data and historical operation and maintenance data of the cloud platform. After cleaning and standardizing the collected data, it is synchronized to the AI ​​prediction and analysis module and the operation and maintenance monitoring module. The AI ​​prediction and analysis module includes a prediction model training unit and a resource demand prediction unit. The prediction model training unit trains and optimizes the AI ​​prediction model based on historical data synchronized by the resource data acquisition module. The resource demand prediction unit uses the trained AI prediction model to predict the changes in business load and resource demand of the cloud platform within a preset time period in the future, and outputs the resource demand prediction results to the resource dynamic scheduling module. The resource dynamic scheduling module receives the resource demand prediction results output by the AI ​​prediction and analysis module, and formulates a dynamic resource allocation strategy based on the current resource occupancy status of the cloud platform. It performs cross-node scheduling, elastic scaling and shrinking, and resource allocation adjustment of the computing resources, storage resources and network resources of the cloud platform, and feeds back the resource scheduling results to the operation and maintenance monitoring module. The operation and maintenance monitoring module monitors the cloud platform's resource operation status, resource scheduling execution, and business operation stability in real time. It generates abnormal alarm information by comparing with preset thresholds and synchronizes the monitoring data and alarm information to the AI ​​prediction and analysis module and the operation and maintenance execution module, respectively, to provide a basis for AI prediction model optimization and operation and maintenance action execution. The operation and maintenance execution module receives alarm information from the operation and maintenance monitoring module and scheduling assistance instructions from the resource dynamic scheduling module. It automatically executes operation and maintenance actions such as fault repair, configuration optimization, log archiving, and operation and maintenance report generation. The execution results are back-synchronized to the resource data acquisition module to form a data closed loop.

2. The automated operation and maintenance system for dynamic resource allocation on cloud platforms based on an AI prediction model as described in claim 1, characterized in that, The resource data acquisition module collects data including: CPU utilization, memory usage, disk I / O rate, network bandwidth usage, service request volume, response latency, historical scaling records, and fault handling logs. The data acquisition frequency is dynamically adjusted based on business needs.

3. The automated operation and maintenance system for dynamic resource allocation on a cloud platform based on an AI prediction model as described in claim 2, characterized in that, The AI ​​predictive analysis module's prediction model training unit employs a hybrid neural network model, integrating LSTM and CNN networks. This model can capture the temporal and burst features of business load. Model optimization utilizes online incremental learning, continuously iterating and updating based on real-time new data to improve prediction accuracy. Prediction accuracy is evaluated using the mean absolute error (MAE), calculated as follows:

4. Among them, n is the number of prediction samples. Let i be the actual resource requirement value for the i-th sample. The model predicts the resource demand value for the i-th sample. This formula quantifies the model prediction error, providing a quantitative basis for model iterative optimization.

5. The automated operation and maintenance system for dynamic resource allocation on a cloud platform based on an AI prediction model as described in claim 3, characterized in that, When formulating resource allocation strategies, the resource dynamic scheduling module follows priority adaptation rules, assigns resource scheduling priorities to core businesses, prioritizes the supply of resources to core businesses, and also takes into account maximizing resource utilization.

6. The automated operation and maintenance system for dynamic resource allocation on a cloud platform based on an AI prediction model as described in claim 4, characterized in that, The operation and maintenance monitoring module sets up multiple alarm thresholds, including early warning thresholds, emergency thresholds, and fatal thresholds. Different operation and maintenance response mechanisms are triggered for different levels of alarm information. The early warning threshold information is synchronized to the AI ​​prediction and analysis module to optimize model prediction parameters.

7. The automated operation and maintenance system for dynamic resource allocation on a cloud platform based on an AI prediction model as described in claim 5, characterized in that, The operation and maintenance execution module has a custom operation and maintenance strategy function, which supports users to preset automated operation and maintenance rules for different scenarios. For configuration changes after resource scheduling, it automatically executes configuration verification and rollback mechanisms to ensure the security of operation and maintenance actions.

8. A method for automated operation and maintenance of cloud platform resources based on AI prediction models, characterized in that, Applied to the system according to any one of claims 1-6, the method includes the following steps: S1: The resource data acquisition module collects cloud platform resource operation data, business load data, and historical operation and maintenance data. After cleaning and standardization, the data is synchronized to the AI ​​prediction and analysis module and the operation and maintenance monitoring module. S2: The AI ​​prediction analysis module trains and optimizes the AI ​​prediction model based on historical data through the prediction model training unit, and then uses the resource demand prediction unit to predict future resource demand using the trained model and outputs the prediction results. S3: The resource dynamic scheduling module formulates resource allocation strategies based on prediction results and current resource status, executes resource scheduling operations, and feeds back the scheduling results to the operation and maintenance monitoring module. S4: The operation and maintenance monitoring module monitors resource operation, scheduling execution and business stability in real time, generates monitoring data and alarm information, and synchronizes them to the corresponding modules. S5: The operation and maintenance execution module automatically executes operation and maintenance actions based on alarm information and scheduling auxiliary instructions, and synchronizes the execution results to the resource data acquisition module to complete one operation and maintenance loop; S6: Repeat steps S1-S5 to achieve fully automated operation and maintenance of dynamic cloud platform resource calls. The AI ​​predictive analysis module continuously optimizes the model based on real-time data to improve the accuracy of operation and maintenance.

9. The automated operation and maintenance method and system for dynamic resource allocation on cloud platforms based on AI prediction models according to claim 7, characterized in that, In step S2, the resource demand forecast results include peak demand, trough demand and sudden demand for resources within a preset time period in the future. The forecast error is controlled within a preset range. When the error exceeds the range, the model emergency optimization process is triggered.

10. The automated operation and maintenance method and system for dynamic resource allocation on cloud platforms based on AI prediction models according to claim 7, characterized in that, In step S3, when the resource dynamic scheduling module performs elastic scaling up and down operations, it adopts a smooth scaling up and down strategy to avoid the impact of resource fluctuations on business operations, and the scheduling process is completely seamless.

Images