Resource scheduling method, device and equipment based on digital twinning and storage medium

Abstract

The invention discloses a resource scheduling method and device based on digital twinning, equipment and a storage medium, and relates to the technical field of resource operation management, and the method comprises the steps: constructing a digital twinning environment synchronized with a physical cloud environment in real time, collecting environment state data of the physical cloud environment, and mapping the environment state data to the digital twinning environment; predicting a working load mode of the physical cloud environment based on the historical working data and the real-time trend of the physical cloud environment, and generating a candidate resource scheduling strategy set according to the working load mode; inputting the candidate resource scheduling strategy set into a digital twin environment for simulation execution so as to collect multi-dimensional index data; and calculating a comprehensive score of the candidate resource scheduling strategy according to the multi-dimensional index data, and generating a resource scheduling scheme according to the comprehensive score. The method has the technical effects of realizing pre-simulation optimization of the resource scheduling strategy, improving the scheduling decision accuracy, improving the resource utilization rate, reducing the scheduling execution risk and adapting to the dynamic workload demand of the bank marketing system.

Description

Technical Field

[0001] This application relates to the field of resource operation and management technology, and in particular to a resource scheduling method, apparatus, electronic device and storage medium based on digital twins. Background Technology

[0002] In the current wave of digital transformation in the banking industry, the marketing system, as a key platform directly facing customers and impacting business growth, is of paramount importance in terms of operational efficiency and stability. However, bank marketing activities are often accompanied by significant sudden and cyclical traffic fluctuations, such as holiday promotions and new product launches, which place extremely high demands on the scheduling capabilities of underlying cloud resources. Among related technologies, cloud resource scheduling technology, while supporting elastic scaling to some extent, still has significant shortcomings in addressing the unique needs of banking systems for high security, high performance, and low-cost collaborative optimization.

[0003] Currently, the cloud resource scheduling solutions commonly used in the industry mainly include threshold-based elastic scaling, heuristic algorithm-based scheduling methods, single-objective optimization strategies, and static resource partitioning schemes. While these methods each have their advantages, they all have certain limitations in practical applications: threshold-based schemes have slow response times and lack the ability to predict sudden traffic surges; heuristic algorithms have high computational complexity and are difficult to meet real-time scheduling requirements; single-objective optimization cannot balance performance, security, and cost; and while static resource partitioning ensures security, it leads to low resource utilization and is difficult to adapt to dynamic business loads.

[0004] Furthermore, banking systems have stringent requirements for data security, compliance, and business continuity. Traditional scheduling mechanisms often sacrifice resource flexibility due to rigid security policies, or neglect security risks in the pursuit of performance and cost optimization, making it difficult to further improve the overall efficiency and reliability of the system. Therefore, there is an urgent need for a new cloud resource scheduling method that can achieve second-level response, multi-objective collaborative optimization, dynamic adaptation of security policies, and support unified scheduling of heterogeneous resources, in order to support the efficient, secure, and economical operation of banking marketing systems in complex and ever-changing environments. Summary of the Invention

[0005] This application provides a resource scheduling method, apparatus, electronic device, and storage medium based on digital twins. It has the technical advantages of enabling pre-simulation optimization of resource scheduling strategies, improving the accuracy of scheduling decisions, shortening resource scheduling response time to the second level, increasing resource utilization, reducing scheduling execution risks, and adapting to the dynamic workload requirements of bank marketing systems.

[0006] According to a first aspect of this application, a resource scheduling method based on digital twins is provided, comprising: Construct a digital twin environment that is synchronized with the physical cloud environment in real time, collect environmental status data of the physical cloud environment, and map the environmental status data to the digital twin environment; Based on historical workload data and real-time trend prediction of the physical cloud environment, a set of candidate resource scheduling strategies is generated according to the workload pattern. The set of candidate resource scheduling strategies is input into the digital twin environment for simulation execution in order to collect multi-dimensional indicator data; The comprehensive score of candidate resource scheduling strategies is calculated based on multi-dimensional indicator data, and a resource scheduling scheme is generated based on the comprehensive score.

[0007] According to a second aspect of this application, a resource scheduling device based on digital twins is provided, comprising: The building module is configured to build a digital twin environment that is synchronized with the physical cloud environment in real time, collect environmental status data of the physical cloud environment, and map the environmental status data to the digital twin environment. The prediction module is configured to predict the workload mode of the physical cloud environment based on historical workload data and real-time trends, and generate a set of candidate resource scheduling strategies based on the workload mode. The collection module is configured to input a set of candidate resource scheduling strategies into a digital twin environment for simulated execution in order to collect multi-dimensional indicator data; The generation module is configured to calculate a comprehensive score for candidate resource scheduling strategies based on multi-dimensional indicator data, and generate a resource scheduling scheme based on the comprehensive score.

[0008] According to a third aspect of this application, an electronic device is provided, comprising: At least one processor; and memory that is communicatively connected to at least one processor; The memory stores instructions that can be executed by at least one processor, which enables the at least one processor to perform the digital twin-based resource scheduling method described in the first aspect above.

[0009] According to a fourth aspect of this application, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause a computer to execute the resource scheduling method based on digital twins described in the first aspect above.

[0010] According to a fifth aspect of this application, a computer program product is provided, including a computer program that, when executed by a processor, implements the resource scheduling method based on digital twins as described in the first aspect above.

[0011] This application provides a resource scheduling method, apparatus, device, and storage medium based on digital twins, comprising: constructing a digital twin environment synchronized in real time with a physical cloud environment; collecting environmental status data of the physical cloud environment; mapping the environmental status data to the digital twin environment; predicting the workload mode of the physical cloud environment based on historical work data and real-time trends; generating a set of candidate resource scheduling strategies based on the workload mode; inputting the set of candidate resource scheduling strategies into the digital twin environment for simulation execution to collect multi-dimensional indicator data; calculating a comprehensive score of the candidate resource scheduling strategies based on the multi-dimensional indicator data; and generating a resource scheduling scheme based on the comprehensive score. This application addresses several issues in related technologies. By constructing a digital twin environment synchronized in real-time with the physical cloud environment and mapping the physical cloud environment's status data to this digital twin, and by predicting the workload patterns based on the physical cloud environment's historical workload data and real-time trends to generate a set of candidate resource scheduling strategies, and by inputting these strategies into the digital twin environment for simulation execution to collect multi-dimensional indicator data, a resource scheduling scheme is generated after calculating a comprehensive score based on the multi-dimensional indicator data. This solves the problems of low resource utilization, delayed resource expansion response, and high scheduling decision risk caused by resource scheduling relying on static thresholds, historical experience, or single-objective optimization, lacking accurate synchronization of the physical environment's real-time status and pre-verification of scheduling strategies, failing to effectively cope with sudden and periodic traffic fluctuations in bank marketing systems, and struggling to comprehensively assess the adaptability of scheduling strategies across multiple dimensions. The goal is to achieve pre-simulation optimization of resource scheduling strategies, improve the accuracy of scheduling decisions, shorten resource scheduling response time to the second level, increase resource utilization, reduce scheduling execution risk, and adapt to the dynamic workload requirements of bank marketing systems.

[0012] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent from the following description. Attached Figure Description

[0013] To more clearly illustrate the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 A flowchart illustrating a resource scheduling method based on digital twins provided in an embodiment of this application; Figure 2 A flowchart illustrating another resource scheduling method based on digital twins provided in this application embodiment; Figure 3 A flowchart illustrating another resource scheduling method based on digital twins provided in this application embodiment; Figure 4 This is a schematic diagram of a resource scheduling device based on digital twins provided in an embodiment of this application. Detailed Implementation

[0015] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of this application, including various details to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0016] The resource scheduling method, apparatus, electronic device, and storage medium based on digital twins according to embodiments of this application are described below with reference to the accompanying drawings.

[0017] Figure 1 This is a flowchart illustrating a resource scheduling method based on digital twins provided in an embodiment of this application.

[0018] like Figure 1 As shown, the method includes the following steps: Step 101: Construct a digital twin environment that is synchronized with the physical cloud environment in real time, collect environmental status data of the physical cloud environment, and map the environmental status data to the digital twin environment.

[0019] In some embodiments, the physical cloud environment refers to the set of basic physical resources upon which the bank's marketing system relies for operation. This includes computing resource pools (such as server clusters), storage resource pools (such as disk arrays), and network resource pools (such as network links composed of switches and routers). It serves as the hardware resource carrier supporting user access, transaction processing, and other business processes during marketing activities. The digital twin environment is a virtualized mirror system of this physical cloud environment. Its core characteristic is real-time synchronization with the physical cloud environment. That is, any changes in the resource status of the physical environment are synchronously reflected in the virtual environment, ensuring a high degree of consistency between the virtual and physical environments in terms of resource configuration and operational status. When constructing this digital twin environment, a corresponding virtual instance must first be created for each independent resource unit within the physical cloud environment. Simultaneously, a virtual resource mapping table must be established to clarify the one-to-one correspondence between physical resources and virtual instances. A synchronization agent component must also be deployed, which continuously monitors the operational status of the physical cloud environment to prevent state discrepancies between the virtual and physical environments. Next, environmental status data of the physical cloud environment is collected. This environmental status data covers real-time load of computing resources (such as CPU utilization, memory usage, and number of running processes), storage resource usage (such as remaining storage space, data read / write speed, and storage device response latency), network resource operating parameters (such as bandwidth utilization, network latency, and packet loss rate), and real-time status related to bank marketing business (such as the number of currently online users and transaction request processing volume). Collection is achieved through monitoring units deployed on various nodes of the physical cloud environment. The monitoring units acquire data at a reasonable sampling frequency to ensure the accuracy and completeness of the collected data. Finally, the collected environmental status data is mapped to the digital twin environment. The mapping process uses a virtual resource mapping table to accurately match each status data item of the physical resource to the corresponding virtual instance parameters. For example, if the physical server's CPU utilization is 55%, the corresponding virtual server's CPU utilization parameter is updated synchronously. The mapped data is then verified to ensure that the virtual environment data is completely consistent with the actual physical environment data, avoiding the impact of data mapping errors on the simulation effect of subsequent scheduling strategies. By constructing a real-time synchronized digital twin environment and completing the mapping of environmental state data, a virtual scenario highly consistent with the physical environment is provided for the simulation of subsequent resource scheduling strategies. This solves the problem of inaccurate scheduling strategy evaluation caused by the lack of accurate environmental support in related technologies, and improves the reliability of subsequent scheduling decisions.

[0020] Step 102: Based on historical work data and real-time trend prediction of the physical cloud environment, predict the workload pattern of the physical cloud environment, and generate a set of candidate resource scheduling strategies according to the workload pattern.

[0021] In some embodiments, historical operational data of the physical cloud environment refers to various operational data accumulated by the environment in past banking marketing activities and daily operations, covering computing resource utilization, storage resource read / write volume, network traffic peaks and troughs at different times, as well as business-level data such as user access volume and transaction request processing volume. Real-time trends are the dynamic changes of various operational indicators of the physical cloud environment at the current moment, such as the real-time growth rate of current user access volume, instantaneous fluctuations in transaction requests, and real-time changes in network bandwidth usage. When predicting the workload pattern of the physical cloud environment, it is necessary to first sort out and analyze the historical operational data to identify the implicit regularities, such as the periodic peak periods of traffic and the slope of traffic growth during similar marketing activities in the past. Then, the historical regularities are dynamically corrected in combination with real-time trend data. For example, if the current real-time user access volume growth rate is higher than the same period in history, the predicted peak traffic height and duration are adjusted to form a workload pattern that fits the actual business scenario. This pattern can clearly reflect the changes in resource demand that the physical cloud environment may face in the future, such as when resource demand peaks occur, the scale of resource demand during peak periods, and the resource redundancy during trough periods.

[0022] Subsequently, a set of candidate resource scheduling strategies is generated based on the predicted workload patterns. The generation process requires designing corresponding strategies for different scenarios within the workload patterns: if a peak in resource demand is predicted, candidate strategies may include increasing the number of virtual machine instances, expanding the container cluster size, and increasing the network bandwidth of core business nodes; if a period of stable resource demand is predicted, candidate strategies may focus on balanced resource allocation to ensure comparable resource utilization efficiency across business modules; if a period of low resource demand is predicted, candidate strategies may include shutting down idle virtual machines, reducing the number of containers, and lowering the resource configuration of non-core nodes. Simultaneously, to ensure the comprehensiveness of candidate strategies, multiple strategies with different parameters need to be designed for the same workload scenario. By combining historical workload data with real-time trend predictions of workload patterns, the accuracy problem caused by relying on fixed rules or single historical experience to predict workloads is solved. Furthermore, generating diverse candidate strategies based on this pattern avoids the limitations of traditional single-strategy approaches, providing a rich foundation for subsequent selection of the optimal scheduling strategy and improving the foresight and adaptability of resource scheduling.

[0023] Step 103: Input the set of candidate resource scheduling strategies into the digital twin environment for simulation execution to collect multi-dimensional indicator data.

[0024] In some embodiments, the candidate resource scheduling strategy set is a variety of potential resource allocation schemes generated based on the predicted workload patterns of the physical cloud environment. These schemes cover resource adjustment approaches for different load scenarios, such as strategies involving different parameter combinations for scaling up the number of virtual machines, adjusting the size of container clusters, and optimizing network bandwidth allocation. The digital twin environment, a virtualized image system synchronized in real time with the physical cloud environment, can accurately reproduce the resource configuration, operating logic, and business load response characteristics of the physical environment, providing a virtual operating space highly consistent with the real-world scenario for simulating the execution of candidate strategies.

[0025] When inputting the set of candidate resource scheduling strategies into the digital twin environment for simulation execution, each candidate strategy must be imported one by one according to a preset process. Each strategy must be fully simulated in the digital twin environment under the corresponding workload mode, from the initial load rise phase, the peak maintenance phase to the decline and fall phase, to ensure that the simulation process covers various load fluctuation scenarios that the bank's marketing system may face. During the simulation execution, the indicator monitoring module built into the digital twin environment will track and record multi-dimensional indicator data in real time during the strategy operation. The "multi-dimensional" is specifically set around the core requirements of the bank's marketing system, including performance dimensions (such as business response latency, transaction throughput, CPU / memory resource utilization), cost dimensions (such as resource occupation cost, idle resource energy consumption cost), and adaptability dimensions (such as the strategy's ability to bear load peaks and response speed to sudden traffic). All indicator data are recorded by time series or load phase to ensure that the data can clearly reflect the actual performance of the strategy in different operating phases and avoid misjudgment of the strategy's effectiveness due to missing or incomplete data. By simulating the execution of candidate strategies and collecting multi-dimensional indicators in a digital twin environment, the effectiveness of strategies in real-world scenarios can be verified in advance. This avoids service interruptions or resource waste caused by trial and error in the physical cloud environment. At the same time, multi-dimensional indicators can comprehensively evaluate the merits of strategies, solving the limitations of evaluating strategies with a single indicator in related technologies. This provides accurate and comprehensive data support for the subsequent selection of the optimal resource scheduling scheme.

[0026] Step 104: Calculate the comprehensive score of the candidate resource scheduling strategy based on multi-dimensional indicator data, and generate a resource scheduling scheme based on the comprehensive score.

[0027] In some embodiments, multi-dimensional indicator data covers the core evaluation dimensions of resource scheduling in the bank marketing system, including performance, cost, and adaptability. When calculating the comprehensive score of candidate resource scheduling strategies, weights for each dimension are first set based on the characteristics of the bank marketing scenario. Then, the indicator data for each dimension are standardized. Subsequently, the scores for each dimension are weighted and summed. Abnormal data caused by instantaneous fluctuations are also removed during this process to ensure that the score objectively reflects the true effect of the strategy. When generating a resource scheduling plan based on the comprehensive score, the candidate strategy with the highest comprehensive score is selected first. If strategies with similar scores exist, further screening is conducted based on the actual constraints of the bank marketing system. The final determined plan specifies concrete resource adjustment operations, ensuring that the plan can be directly implemented. This multi-dimensional comprehensive scoring screening strategy solves the problem of resource allocation imbalance caused by single-objective optimization of related technologies, allowing the scheduling plan to balance performance, cost, and adaptability, better meeting the needs of the bank marketing scenario and improving the scientific and practical nature of resource scheduling.

[0028] Compared with related technologies, this embodiment constructs a digital twin environment synchronized with the physical cloud environment in real time, collects environmental status data of the physical cloud environment, and maps the environmental status data to the digital twin environment. Based on the historical working data and real-time trend prediction of the physical cloud environment, the workload pattern of the physical cloud environment is predicted, and a set of candidate resource scheduling strategies is generated according to the workload pattern. The set of candidate resource scheduling strategies is input into the digital twin environment for simulation execution to collect multi-dimensional indicator data. The comprehensive score of the candidate resource scheduling strategies is calculated based on the multi-dimensional indicator data, and a resource scheduling scheme is generated based on the comprehensive score. This solves the problems in related technologies where resource scheduling relies on static thresholds, historical experience, or single-objective optimization, lacks accurate synchronization of the real-time status of the physical environment and pre-verification of scheduling strategies, cannot effectively cope with sudden and periodic traffic fluctuations in the bank marketing system, and is difficult to comprehensively evaluate the adaptability of scheduling strategies in multiple dimensions, resulting in low resource utilization, delayed resource expansion response, and high scheduling decision risk. This achieves the technical effect of pre-simulation optimization of resource scheduling strategies, improving the accuracy of scheduling decisions, shortening resource scheduling response time to the second level, improving resource utilization, reducing scheduling execution risk, and adapting to the dynamic workload requirements of the bank marketing system.

[0029] Figure 2 A flowchart illustrating another resource scheduling method based on digital twins provided in this application embodiment includes the following steps: Step 201: Establish a virtual resource mapping table using a resource fingerprint matching algorithm. The resource fingerprint includes the processor architecture identifier of computing resources, the disk configuration identifier of storage resources, and the topology identifier of network resources.

[0030] In some embodiments, a resource fingerprint matching algorithm and parallel data acquisition technology are used to achieve accurate mapping and efficient data synchronization between physical resources and the virtual environment. The resource fingerprint matching algorithm is the core algorithm used to establish the association between physical resources and virtual instances. The resource fingerprints it relies on cover three key identifiers: processor architecture identifiers for computing resources (such as CPU model, number of cores, and instruction set version, which can accurately distinguish the computing power and compatibility characteristics of different processors), disk configuration identifiers for storage resources (including disk type such as SSD / HDD, capacity, and read / write speed parameters, directly reflecting the performance limit and usage status of storage resources), and topology identifiers for network resources (including node connection relationships, link bandwidth, and IP address ranges, clearly defining the layout of network resources and data transmission paths). This algorithm traverses all resources in the physical cloud environment, extracts the above fingerprint information, and binds it to the virtual instances created in the digital twin environment, ultimately generating a virtual resource mapping table to ensure a one-to-one correspondence between physical resources and virtual instances, avoiding misalignment in subsequent data mapping.

[0031] Step 202: Using parallel data acquisition technology, the physical environment status data is processed by segmenting according to the virtual resource mapping table and then synchronized to the digital twin environment.

[0032] In some embodiments, parallel data acquisition technology is employed, which can simultaneously launch multiple acquisition threads to collect environmental status data (such as CPU utilization, disk space remaining, and network bandwidth utilization) of different types (computing, storage, and network) or different nodes in the physical cloud environment. This significantly reduces data acquisition time compared to traditional serial acquisition. After acquisition, the data is fragmented according to resource category or corresponding virtual instance based on the virtual resource mapping table. Each data fragment contains only the status information of a single physical resource. The fragmented data is then synchronized to the corresponding virtual instance in the digital twin environment. During the synchronization process, data integrity is also verified to ensure that the status data received by the virtual instance is completely consistent with the physical resource.

[0033] Step 203: Based on historical workload data and real-time trends, predict the workload pattern of the physical cloud environment within a preset time period using a load forecaster.

[0034] In some embodiments, a load forecaster is used to link historical and real-time data to achieve accurate load forecasting, and a compliant and adaptable set of candidate strategies is generated in conjunction with business constraints. The load forecaster is a functional module that integrates data preprocessing, algorithm computation, and result output. It first processes historical work data from the physical cloud environment, filtering effective data from past bank marketing scenarios and daily operations, and removing abnormal interference data such as equipment failures and sudden network fluctuations. Then, through data normalization, it transforms historical data from different dimensions into a unified format. Subsequently, combined with real-time trends in the physical cloud environment, a time-series forecasting model is used to predict the workload pattern within a preset time period. This preset time period needs to be set according to the characteristics of the bank's marketing business. For example, for a single-day promotional activity, the preset time period can be set from 1 hour before the activity starts to 30 minutes after the activity ends, ensuring coverage of the complete cycle of load increase, peak, and decline. The forecast results must clearly specify key information such as the time of load peak occurrence, peak size, and load change slope in each time period, providing a basis for subsequent strategy generation.

[0035] Step 204: Based on the workload status and business constraints, generate a set of multiple candidate scheduling policies. The business constraints include service level agreement requirements and security compliance requirements.

[0036] In some embodiments, when generating a set of candidate resource scheduling strategies, the specific requirements of business constraints must first be clarified: Service Level Agreement (SLA) requirements include core metrics of the bank's marketing system, such as transaction response latency, system availability, and transaction success rate. These metrics directly determine the service baseline that the strategy must guarantee. Security and compliance requirements cover financial industry regulatory rules, such as customer transaction data processing must comply with local data storage requirements, core business resources must be physically isolated from ordinary business resources, and cross-border data transmission must comply with GDPR or domestic financial data regulatory standards to avoid strategies violating compliance red lines. Subsequently, based on the predicted workload pattern, candidate strategies are generated around the principle of "meeting constraints and adapting to load," ultimately forming a set of candidate resource scheduling strategies containing at least one effective strategy. Accurate load prediction through a load predictor solves the bias problem caused by traditional load predictions relying on experience. Combining business constraints with strategy generation avoids the shortcomings of traditional solutions that ignore compliance and service baselines, ensuring that candidate strategies both fit actual load needs and meet the compliance and service requirements of the bank's marketing system, thus improving the practicality and reliability of the candidate strategies.

[0037] Step 205: Input the set of candidate resource scheduling strategies into the digital twin environment for simulation execution to collect multi-dimensional indicator data.

[0038] For a description of step 205, please refer to the description of step 103 in the above embodiment. This embodiment will not repeat the details further.

[0039] Step 206: Based on the three-dimensional collaborative optimization model and dynamic weight adjustment algorithm, calculate the comprehensive score according to multi-dimensional index data, including performance data, security data and cost data.

[0040] In some embodiments, the three-dimensional collaborative optimization model is constructed around the core requirements of cloud resource scheduling in a bank marketing system, with performance, security, and cost as the three major optimization dimensions. The performance data, based on multi-dimensional indicators, includes system transaction response latency, CPU and memory resource utilization, and transaction throughput; security data includes data transmission encryption strength, access control accuracy, and real-time vulnerability detection rate; and cost data includes physical resource leasing fees, idle resource energy consumption costs, and software licensing and maintenance costs. A dynamic weight adjustment algorithm flexibly adjusts the weight ratios of the three dimensions according to the actual needs of the bank's marketing scenario. When calculating the comprehensive score, the multi-dimensional indicator data is first standardized, and then the standardized scores of each dimension are weighted and summed according to the adjusted weights to obtain the comprehensive score for each candidate resource scheduling strategy.

[0041] In this embodiment, in order to calculate the comprehensive score based on multi-dimensional indicator data, the execution logic of the three-dimensional collaborative optimization model can be:

[0042] in, This represents a multi-objective optimization function, where the objective is to simultaneously minimize a function of three dimensions. This represents the performance objective function. Represents the security objective function. Represents the cost objective function, Inequality constraints representing resource capacity limits and response time limits. This represents the equality constraints that govern resource allocation balance and mandatory security policy requirements. Indicates system response time. Indicates system throughput. Indicates system availability. Indicates the score for safety compliance. Indicates security protection capabilities. Indicates safe recovery capability. Indicates resource cost, Indicates energy cost, Indicates the cost of the software license. , , All represent performance weighting coefficients. , , Both represent safety weighting coefficients. , , All of these represent cost weighting coefficients.

[0043] In this embodiment, the weight coefficients are adaptively adjusted through a weight adaptive adjustment mechanism. This mechanism enables the system to intelligently adjust the importance allocation of optimization objectives based on external environment and internal state. Weight adjustment is based on a comprehensive analysis of multiple factors, including business scenario characteristics, system load, security threat level, and cost constraints. The system first identifies the current business scenario type (e.g., normal operation, marketing activities, system maintenance), and each scenario has a corresponding baseline weight configuration. Then, the system analyzes real-time operational data and calculates adjustment factors for each dimension. The security adjustment factor is calculated based on the threat level, the performance adjustment factor considers system load pressure, and the cost adjustment factor reflects resource utilization efficiency. These adjustment factors are combined with the baseline weights and normalized to obtain the final weight configuration. This dynamic adjustment mechanism ensures that the system can intelligently balance different objectives according to the actual situation, avoiding both excessive sacrifice of other aspects for a single objective and the adoption of a rigid fixed weight strategy.

[0044] Step 207: Calculate the optimal scheduling scheme based on the comprehensive score using the improved multi-objective particle swarm optimization algorithm, and generate a resource scheduling scheme based on the optimal scheduling scheme.

[0045] In some embodiments, the improved multi-objective particle swarm optimization algorithm treats each candidate strategy as a "particle," using the comprehensive score as the fitness index. After initializing the particle swarm, it continuously optimizes through an iterative process: first, it records the individual optimal score and the global optimal score of each particle; then, it adjusts the particle's "velocity" and "position" (corresponding to the optimization direction and magnitude of the strategy parameters) based on the difference between the two. Simultaneously, it incorporates business constraint verification, eliminating particles that do not meet the constraints. Finally, it selects the "optimal particle" with the highest comprehensive score that satisfies all constraints, and its corresponding candidate strategy becomes the optimal scheduling scheme. Based on this scheme, it clarifies specific resource adjustment operations and generates a directly executable resource scheduling scheme. Through three-dimensional collaborative optimization and dynamic weight adjustment, it ensures that the score fully matches the scenario requirements. Combined with the improved algorithm to select the optimal scheme, it solves the resource allocation imbalance problem caused by single-objective optimization or fixed weights in related technologies, allowing the scheduling scheme to balance performance, security, and cost, thus improving the scientific and practical nature of resource scheduling in the bank marketing system.

[0046] Figure 3 A flowchart illustrating another resource scheduling method based on digital twins provided in this application embodiment includes the following steps: Step 301: Construct a digital twin environment that is synchronized with the physical cloud environment in real time, collect environmental status data of the physical cloud environment, and map the environmental status data to the digital twin environment.

[0047] Step 302: Based on historical workload data and real-time trend prediction of the physical cloud environment, generate a set of candidate resource scheduling strategies according to the workload pattern.

[0048] Step 303: Input the set of candidate resource scheduling strategies into the digital twin environment for simulation execution to collect multi-dimensional indicator data.

[0049] Step 304: Calculate the comprehensive score of the candidate resource scheduling strategy based on the multi-dimensional indicator data, and generate a resource scheduling scheme based on the comprehensive score.

[0050] For a description of steps 301-304, please refer to the description of steps 101-104 in the above embodiment. This embodiment will not repeat them in detail.

[0051] Step 305: Construct a biomimetic immune dynamic security boundary based on the immune security engine, and perform self-learning and adaptive adjustment of security policies on the immune security engine.

[0052] In some embodiments, the immune security engine is a security management module designed by drawing on the core mechanisms of the biological immune system, namely "immune recognition, immune memory, and immune response." It is specifically designed for the cloud environment of the bank marketing system and can perceive threats in real time and dynamically adjust protection strategies. The biomimetic immune dynamic security boundary is a non-static protection system built around this engine. Unlike traditional fixed isolation boundaries, it can flexibly adjust the protection strength according to the changes in threats, adapting to the dynamic needs of business and security in the bank marketing scenario.

[0053] When constructing a biomimetic immune dynamic security boundary, it is necessary to first deploy distributed security agents of the immune security engine on each core node (compute node, storage node, network gateway) in the physical cloud environment. These agents act like "immune cells," continuously collecting node security status data, including abnormal access logs (such as unauthorized IP attempts to access marketing data), data transmission characteristics (such as transaction data transmission with abnormal encryption protocols), and process behavior (such as unregistered background data reading processes). Subsequently, the engine's core threat identification module compares the built-in financial industry-specific threat signature database (including known threat characteristics such as SQL injection, data leakage, and transaction hijacking) to complete "antigen identification." Then, based on preset threat classification standards (such as a single node suspicious access for a minor threat and a cross-node data theft attack for a severe threat), the severity of the threat is determined, and the security boundary is dynamically adjusted: for a minor threat, only the auditing of the node's access logs and real-time alerts are strengthened; for a severe threat, the end-to-end encryption level of data transmission is immediately upgraded, access from non-core business IPs is restricted, and the threatened node is temporarily isolated to prevent the threat from spreading, while not affecting the resource usage of other normal marketing businesses.

[0054] The implementation of self-learning and adaptive adjustment of security policies by the immune security engine needs to be carried out in two parts: At the self-learning level, the engine records the entire process data of each security event: threat characteristics, triggered protection policies, and policy execution effects. This data is analyzed using a gradient boosting tree algorithm to update the threat characteristic database and optimize policy parameters, forming "immune memory" that can shorten response time when encountering similar threats again. At the adaptive adjustment level, the engine correlates with the real-time business status of the physical cloud environment, dynamically balancing security and performance: During peak periods, while ensuring basic security such as transaction data encryption and core permission control, the frequency of security checks for non-core businesses is appropriately reduced to avoid excessive security resource consumption affecting service response; during off-peak periods, the intensity of security checks across all nodes is increased to identify potential vulnerabilities. Simultaneously, based on historical operational data, the types of threats that upcoming marketing activities may face are predicted, and corresponding protection policies are optimized in advance. This approach solves the problems of resource waste and inability to cope with dynamic threats associated with traditional static security isolation. The self-learning and adaptive capabilities reduce manual intervention costs, meeting the high compliance and security requirements of the bank's marketing system while avoiding excessive impact of security policies on business performance, thus improving system security stability and resource utilization efficiency.

[0055] Step 306: Based on the real-time threat assessment results, dynamically balance security and performance requirements.

[0056] In some embodiments, real-time threat assessment results are generated by the immune security engine of the biomimetic immune dynamic security boundary. These results are derived from real-time security monitoring data in the physical cloud environment and digital twin environment of the bank's marketing system, specifically covering threat type, threat impact scope, and threat severity. The assessment results are updated in real time to ensure they reflect the latest security situation. When dynamically balancing security and performance requirements, the real-time threat assessment results should be the core basis, combined with the performance requirements priority of the current business scenario of the bank's marketing system. When the threat assessment is mild, system performance is prioritized, with only lightweight security measures implemented, such as increasing the log auditing frequency of suspicious access nodes and enabling basic encryption for non-core data transmission. This avoids excessive additional computing or network resources, preventing impact on user access speed and transaction processing efficiency, especially during peak promotional periods, to prevent response delays due to excessive security measures. When the threat assessment is moderate, the security protection strength is appropriately increased, while performance is balanced through resource scheduling. For example, for abnormal data transmission modules, advanced encryption protocols are enabled, and additional CPU resources are temporarily allocated to these modules to prevent encryption operations from consuming too many existing resources and causing performance degradation. Furthermore, measures are taken only in the affected areas, without expanding to the entire system, minimizing performance interference with other normal business modules. When the threat assessment is severe, priority is given to ensuring security compliance and core data security, initiating high-strength security strategies, such as temporarily isolating threatened core nodes, enabling end-to-end encryption of all system transaction data, and restricting unnecessary external access. If security measures consume resources and cause performance degradation, redundant resources are quickly allocated to core business nodes through digital twin environment simulation to ensure uninterrupted core marketing operations. After the threat subsides, the security intensity is gradually reduced, releasing the resources consumed by security measures and restoring system performance to normal levels. Throughout the balancing process, the effectiveness of security measures and system performance indicators are monitored in real time, and the strength of measures and resource allocation are dynamically fine-tuned to avoid extreme situations of insufficient security or excessive performance loss. This approach solves the problem that traditional static security strategies cannot adapt to dynamic threats and performance requirements. It avoids performance waste caused by excessive security under mild threats and prevents the risk of neglecting security under severe threats, achieving precise adaptation of security and performance in the bank's marketing system, ensuring normal business operations while meeting compliance requirements.

[0057] Figure 4 This is a schematic diagram of the structure of a resource scheduling device based on digital twins provided in an embodiment of this application, as shown below. Figure 4 As shown, it includes: a construction module 401, a simulation module 402, a collection module 403, and a generation module 404.

[0058] Module 401 is configured to build a digital twin environment that is synchronized with the physical cloud environment in real time, collect environmental status data of the physical cloud environment, and map the environmental status data to the digital twin environment. Prediction module 402 is configured to predict the workload mode of the physical cloud environment based on historical workload data and real-time trends of the physical cloud environment, and generate a set of candidate resource scheduling strategies based on the workload mode. The collection module 403 is configured to input a set of candidate resource scheduling strategies into a digital twin environment for simulation execution in order to collect multi-dimensional indicator data; The generation module 404 is configured to calculate a comprehensive score of candidate resource scheduling strategies based on multi-dimensional indicator data, and generate a resource scheduling scheme based on the comprehensive score.

[0059] In some examples of this embodiment, the construction module 401 is specifically configured to establish a virtual resource mapping table through a resource fingerprint matching algorithm. The resource fingerprint includes the processor architecture identifier of the computing resource, the disk configuration identifier of the storage resource, and the topology identifier of the network resource. Parallel data acquisition technology is used to process physical environment status data into segments based on a virtual resource mapping table and then synchronize it to the digital twin environment.

[0060] In some examples of this embodiment, the prediction module 402 is specifically configured to predict the workload pattern of the physical cloud environment within a preset time period based on historical workload data and real-time trends using a load predictor; and to generate a set of multiple candidate scheduling strategies based on the workload status and business constraints, including service level agreement requirements and security compliance requirements.

[0061] In some examples of this embodiment, the generation module 404 is specifically configured to calculate a comprehensive score based on multi-dimensional index data, including performance data, security data, and cost data, based on a three-dimensional collaborative optimization model and a dynamic weight adjustment algorithm; calculate the optimal scheduling scheme based on the comprehensive score based on an improved multi-objective particle swarm optimization algorithm; and generate a resource scheduling scheme based on the optimal scheduling scheme.

[0062] It should be noted that other corresponding descriptions of the functional units involved in the resource scheduling device based on digital twins provided in this embodiment can be found in [reference needed]. Figure 1 , Figure 2 and Figure 3 The corresponding descriptions in [the document] will not be repeated here.

[0063] Based on the above, Figure 1 , Figure 2 and Figure 3 The embodiment illustrates a resource scheduling method based on digital twins. Correspondingly, this embodiment also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method. Figure 1 , Figure 2 and Figure 3This illustrates a resource scheduling method based on digital twins.

[0064] Based on the above, Figure 1 , Figure 2 and Figure 3 The embodiment illustrates a resource scheduling method based on digital twins. Correspondingly, this embodiment also provides a computer program product on which a computer program is stored. When executed by a processor, this computer program implements the above-described method. Figure 1 , Figure 2 and Figure 3 This illustrates a resource scheduling method based on digital twins.

[0065] Based on this understanding, the technical solution of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as CD-ROM, USB flash drive, mobile hard drive, etc.) and includes several instructions to cause a computer device (such as personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of this application.

[0066] Based on the above, Figure 1 , Figure 2 and Figure 3 The resource scheduling method based on digital twins is shown, and Figure 4 To achieve the above objectives, the present application also provides an electronic device, such as a personal computer or a server, in the illustrated virtual device embodiment. This device includes a storage medium and a processor; the storage medium stores a computer program; the processor executes the computer program to implement the above-described virtual device. Figure 1 , Figure 2 and Figure 3 This illustrates a resource scheduling method based on digital twins.

[0067] In some embodiments, the aforementioned physical device may further include a user interface, a network interface, a camera, radio frequency (RF) circuitry, sensors, audio circuitry, a Wi-Fi module, etc. The user interface may include a display screen, an input unit such as a keyboard, etc., and optionally, a USB interface, a card reader interface, etc. In some embodiments, the network interface may include a standard wired interface, a wireless interface (such as a Wi-Fi interface), etc.

[0068] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the aforementioned physical device, supporting the operation of information processing programs and other software and / or programs. The network communication module is used to enable communication between the various components within the storage medium, as well as communication with other hardware and software in the information processing physical device.

[0069] It should be noted that, in this document, relational terms such as "first" and "second" are used merely 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 a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0070] The above are merely specific embodiments of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to these embodiments, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A resource scheduling method based on digital twins, characterized in that, include: Construct a digital twin environment that is synchronized with the physical cloud environment in real time, collect environmental status data of the physical cloud environment, and map the environmental status data to the digital twin environment; Based on the historical working data and real-time trend prediction of the physical cloud environment, the workload mode of the physical cloud environment is predicted, and a set of candidate resource scheduling strategies is generated according to the workload mode. The candidate resource scheduling strategy set is input into the digital twin environment for simulation execution in order to collect multi-dimensional indicator data; The comprehensive score of the candidate resource scheduling strategy is calculated based on the multi-dimensional indicator data, and a resource scheduling scheme is generated based on the comprehensive score.

2. The resource scheduling method based on digital twins according to claim 1, characterized in that, The construction of a digital twin environment that is synchronized in real time with the physical cloud environment includes: A virtual resource mapping table is established by means of a resource fingerprint matching algorithm. The resource fingerprint includes the processor architecture identifier of computing resources, the disk configuration identifier of storage resources, and the topology identifier of network resources. Parallel data acquisition technology is used to process physical environment status data into segments based on the virtual resource mapping table and then synchronize it to the digital twin environment.

3. The resource scheduling method based on digital twins according to claim 1, characterized in that, The process involves predicting the workload patterns of the physical cloud environment based on historical workload data and real-time trends, and generating a set of candidate resource scheduling strategies based on these workload patterns, including: Based on the historical work data and the real-time trend, the workload pattern of the physical cloud environment within a future preset time period is predicted by the load predictor. Based on the workload status and business constraints, a set of multiple candidate scheduling strategies is generated. The business constraints include service level agreement requirements and security compliance requirements.

4. The resource scheduling method based on digital twins according to claim 1, characterized in that, The step of calculating a comprehensive score for candidate resource scheduling strategies based on the multi-dimensional data, and generating a resource scheduling scheme based on the comprehensive score, includes: Based on a three-dimensional collaborative optimization model and a dynamic weight adjustment algorithm, the comprehensive score is calculated according to the multi-dimensional index data, which includes performance data, security data, and cost data. The optimal scheduling scheme is calculated based on the comprehensive score using an improved multi-objective particle swarm optimization algorithm, and the resource scheduling scheme is generated based on the optimal scheduling scheme.

5. The resource scheduling method based on digital twins according to claim 4, characterized in that... The comprehensive score calculated based on the multi-dimensional data using the three-dimensional collaborative optimization model and dynamic weight adjustment algorithm includes: the execution logic of the three-dimensional collaborative optimization model is as follows: in, This represents a multi-objective optimization function, the objective of which is to simultaneously minimize a function of three dimensions. This represents the performance objective function. Represents the security objective function. Represents the cost objective function, Inequality constraints representing resource capacity limits and response time limits. This represents the equality constraints that govern resource allocation balance and mandatory security policy requirements. Indicates system response time. Indicates system throughput. Indicates system availability. Indicates the score for safety compliance. Indicates security protection capabilities. Indicates safe recovery capability. Indicates resource cost, Indicates energy cost, Indicates the cost of the software license. , , All represent performance weighting coefficients. , , Both represent safety weighting coefficients. , , All of these represent cost weighting coefficients.

6. The resource scheduling method based on digital twins according to claim 1, characterized in that, Also includes: A biomimetic immune dynamic security boundary is constructed based on the immune security engine, and the security policy is self-learned and adaptively adjusted on the immune security engine. Based on real-time threat assessment results, a dynamic balance is achieved between security and performance requirements.

7. A resource scheduling device based on digital twins, characterized in that, include: The building module is configured to build a digital twin environment that is synchronized with the physical cloud environment in real time, collect environmental status data of the physical cloud environment, and map the environmental status data to the digital twin environment; The prediction module is configured to predict the workload pattern of the physical cloud environment based on historical workload data and real-time trends, and generate a set of candidate resource scheduling strategies based on the workload pattern. The collection module is configured to input the set of candidate resource scheduling strategies into the digital twin environment for simulation execution in order to collect multi-dimensional indicator data; The generation module is configured to calculate a comprehensive score of candidate resource scheduling strategies based on the multi-dimensional indicator data, and generate a resource scheduling scheme based on the comprehensive score.

8. An electronic device, characterized in that, include: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the resource scheduling method based on digital twins as described in any one of claims 1-6.

9. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to execute the resource scheduling method based on digital twins according to any one of claims 1-6.

10. A computer program product, characterized in that, It includes a computer program that, when executed by a processor, implements the resource scheduling method based on digital twins according to any one of claims 1-6.

Images