A method and system for detecting computing power anomaly based on data analysis

Abstract

The present invention proposes a computing power anomaly detection method and system based on data analysis, which relates to the technical field of computing power anomaly detection, extracts computing task data, and obtains computing power monitoring nodes; calculates computing power resource coefficients, performs computing power resource allocation, obtains allocated reserve computing power resources, obtains compensation resource data based on the reserve computing power resources, and compensates the reserve computing power resources based on the compensation resource data; calculates computing power anomaly coefficients, obtains abnormal computing power nodes and normal computing power nodes based on the computing power anomaly coefficients, and then obtains abnormal computing power node information of the abnormal computing power nodes; calculates resource compensation data of abnormal computing power nodes, and dispatches the resource compensation data to abnormal computing power nodes until the abnormal computing power nodes are determined to be normal computing power nodes. The present invention can significantly improve the utilization efficiency of computing power resources and the overall performance of the system, and provide strong support for the development of fields such as big data processing, cloud computing, and artificial intelligence.

Description

Technical Field

[0001] The present invention proposes a computing power anomaly detection method and system based on data analysis, which relate to the technical field of computing power anomaly detection, and specifically to the technical field of computing power anomaly detection based on data analysis. Background Art

[0002] With the rapid development of information technology, especially in the fields of big data processing, cloud computing and artificial intelligence, the management and scheduling of computing resources have become the key to ensure efficient and stable operation of the system. However, in practical applications, the allocation and utilization of computing resources are often restricted by many factors, such as the diversity of computing tasks, performance differences of computing nodes, dynamic changes in resources, etc. These factors make the management of computing resources complex and challenging.

[0003] In existing computing resource management and scheduling systems, static or simple rule-based scheduling strategies are usually adopted. These strategies are often difficult to accurately match and efficiently utilize when faced with complex computing needs. For example, some systems simply allocate resources based on the priority of the task or the load of the node, ignoring the specific characteristics of the task and the actual performance differences of the node. In addition, when the system has insufficient computing resources or abnormal node performance, there is often a lack of effective compensation and recovery mechanisms, resulting in overall performance degradation or even system crashes. Summary of the invention

[0004] The present invention provides a computing power anomaly detection method and system based on data analysis to solve the problem that the existing technology is difficult to achieve accurate matching and efficient utilization. For example, some systems only perform simple resource allocation based on the priority of the task or the load of the node, ignoring the specific characteristics of the task and the actual performance differences of the node. In addition, when the system has insufficient computing power resources or abnormal node performance, there is often a lack of effective compensation and recovery mechanisms, resulting in overall performance degradation or even system crashes.

[0005] The present invention proposes a method and system for detecting computing power anomalies based on data analysis, the method comprising:

[0006] S1. Obtain computing demand information, extract computing task data according to the computing demand information, obtain computing progress bars of computing task types and computing power monitoring nodes of each computing progress bar;

[0007] S2. Calculate the computing power resource coefficient, allocate computing power resources, obtain the allocated reserve computing power resources, obtain compensation resource data according to the reserve computing power resources, and compensate the reserve computing power resources according to the compensation resource data;

[0008] S3. Calculate the computing power abnormality coefficient, obtain the abnormal computing power node and the normal computing power node according to the computing power abnormality coefficient, and then obtain the abnormal computing power node information of the abnormal computing power node;

[0009] S4. Calculate the resource compensation data of the abnormal computing power node, and dispatch the resource compensation data to the abnormal computing power node until the abnormal computing power node is determined to be a normal computing power node.

[0010] Furthermore, the S1 includes:

[0011] Obtain computing demand information, extract computing task data according to a plurality of preset different computing task types based on the computing demand information, and obtain a plurality of computing target data of different computing task types;

[0012] The calculation target data of each calculation task type are sorted according to the calculation target task order to obtain a calculation progress bar corresponding to the calculation task type;

[0013] Each calculation target data in each calculation progress bar is a computing power monitoring node.

[0014] Further, the S2 includes:

[0015] Obtain the computing target data of each computing power monitoring node, and calculate the computing power resource coefficient of the computing power monitoring node according to each computing target data;

[0016] Allocate computing power resources to each computing power monitoring node according to the computing power resource coefficient to obtain computing power resource allocation data;

[0017] Obtaining remaining computing resources according to the computing resource allocation data, where the remaining computing resources are the reserve computing resources;

[0018] Compare the reserve computing power resources with a preset resource threshold to obtain a resource comparison result;

[0019] Acquire compensation resource data according to the resource comparison result;

[0020] Collecting compensation resources according to the compensation resource data to obtain compensation resource collection data;

[0021] The reserve computing power resources are compensated according to the compensation resource collection data until the reserve computing power resources are greater than a preset resource threshold.

[0022] Further, the S3 includes:

[0023] Obtaining actual calculation data of each computing power monitoring node, and calculating a computing power anomaly coefficient according to the actual calculation data;

[0024] Compare the computing power abnormality coefficient with a preset computing power abnormality threshold to obtain an abnormal comparison result;

[0025] When the computing power abnormality coefficient is greater than the preset computing power abnormality threshold, the corresponding computing power monitoring node is marked as an abnormal computing power node;

[0026] When the computing power abnormality coefficient is less than or equal to the preset computing power abnormality threshold, the corresponding computing power monitoring node is marked as a normal computing power node;

[0027] Identify and obtain abnormal computing power node information.

[0028] Further, the S4 includes:

[0029] Obtain computing power resource allocation data of the abnormal computing power node, and calculate resource compensation data of the abnormal computing power node according to the computing power resource allocation data combined with the abnormal computing power node information;

[0030] The resource compensation data is dispatched to the abnormal computing power node by using the reserved computing power resources to obtain the dispatch completion data;

[0031] After the scheduling is completed, the abnormal computing power coefficient of the abnormal computing power node is recalculated to obtain the recalculation result until the abnormal computing power node is determined to be a normal computing power node.

[0032] Furthermore, the system comprises:

[0033] A node distribution module is used to obtain computing demand information, extract computing task data according to the computing demand information, obtain a computing progress bar of the computing task type and a computing power monitoring node of each computing progress bar;

[0034] A resource allocation module, used to calculate the computing power resource coefficient, perform computing power resource allocation, obtain the allocated reserve computing power resources, obtain compensation resource data according to the reserve computing power resources, and compensate the reserve computing power resources according to the compensation resource data;

[0035] An abnormality identification module is used to calculate a computing power abnormality coefficient, obtain abnormal computing power nodes and normal computing power nodes according to the computing power abnormality coefficient, and then obtain abnormal computing power node information of the abnormal computing power node;

[0036] The abnormal allocation module is used to calculate the resource compensation data of the abnormal computing power node and dispatch the resource compensation data to the abnormal computing power node until the abnormal computing power node is determined to be a normal computing power node.

[0037] Furthermore, the node distribution module includes:

[0038] Obtain computing demand information, extract computing task data according to a plurality of preset different computing task types based on the computing demand information, and obtain a plurality of computing target data of different computing task types;

[0039] The calculation target data of each calculation task type are sorted according to the calculation target task order to obtain a calculation progress bar corresponding to the calculation task type;

[0040] Each calculation target data in each calculation progress bar is a computing power monitoring node.

[0041] Furthermore, the resource allocation module includes:

[0042] Obtain the computing target data of each computing power monitoring node, and calculate the computing power resource coefficient of the computing power monitoring node according to each computing target data;

[0043] Allocate computing power resources to each computing power monitoring node according to the computing power resource coefficient to obtain computing power resource allocation data;

[0044] Obtaining remaining computing resources according to the computing resource allocation data, where the remaining computing resources are the reserve computing resources;

[0045] Compare the reserve computing power resources with a preset resource threshold to obtain a resource comparison result;

[0046] Acquire compensation resource data according to the resource comparison result;

[0047] Collecting compensation resources according to the compensation resource data to obtain compensation resource collection data;

[0048] The reserve computing power resources are compensated according to the compensation resource collection data until the reserve computing power resources are greater than a preset resource threshold.

[0049] Furthermore, the abnormality identification module includes:

[0050] Obtaining actual calculation data of each computing power monitoring node, and calculating a computing power anomaly coefficient according to the actual calculation data;

[0051] Compare the computing power abnormality coefficient with a preset computing power abnormality threshold to obtain an abnormal comparison result;

[0052] When the computing power abnormality coefficient is greater than the preset computing power abnormality threshold, the corresponding computing power monitoring node is marked as an abnormal computing power node;

[0053] When the computing power abnormality coefficient is less than or equal to the preset computing power abnormality threshold, the corresponding computing power monitoring node is marked as a normal computing power node;

[0054] Identify and obtain abnormal computing power node information.

[0055] Furthermore, the abnormal distribution module includes:

[0056] Obtain computing power resource allocation data of the abnormal computing power node, and calculate resource compensation data of the abnormal computing power node according to the computing power resource allocation data combined with the abnormal computing power node information;

[0057] The resource compensation data is dispatched to the abnormal computing power node by using the reserved computing power resources to obtain the dispatch completion data;

[0058] After the scheduling is completed, the abnormal computing power coefficient of the abnormal computing power node is recalculated to obtain the recalculation result until the abnormal computing power node is determined to be a normal computing power node.

[0059] Beneficial effects of the present invention: By dynamically allocating computing resources and timely compensating for reserve computing power, it is ensured that computing resources are efficiently utilized and resource waste is reduced. Real-time monitoring of computing node status, timely discovery and processing of abnormal nodes, effective prevention of the spread of single point failures, and improved overall system stability. By visually displaying the calculation progress and dynamically adjusting task priorities, users can more intuitively understand the task execution status and improve user satisfaction. The automated monitoring, allocation, and compensation mechanism reduces the need for manual intervention and reduces operation and maintenance costs and time costs. The system can flexibly respond to the needs of large-scale computing tasks, and ensure that tasks are completed on time and with quality through dynamic expansion and resource optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a schematic diagram of a computing power anomaly detection method based on data analysis;

[0061] Figure 2 This is a schematic diagram of the calculation progress bar. DETAILED DESCRIPTION

[0062] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0063] In one embodiment of the present invention, a method and system for detecting computing power anomalies based on data analysis are proposed by the present invention, and the method includes:

[0064] S1. Obtain computing demand information, extract computing task data according to the computing demand information, obtain computing progress bars of computing task types and computing power monitoring nodes of each computing progress bar;

[0065] S2. Calculate the computing power resource coefficient, allocate computing power resources, obtain the allocated reserve computing power resources, obtain compensation resource data according to the reserve computing power resources, and compensate the reserve computing power resources according to the compensation resource data;

[0066] S3. Calculate the computing power abnormality coefficient, obtain the abnormal computing power node and the normal computing power node according to the computing power abnormality coefficient, and then obtain the abnormal computing power node information of the abnormal computing power node;

[0067] S4. Calculate the resource compensation data of the abnormal computing power node, and dispatch the resource compensation data to the abnormal computing power node until the abnormal computing power node is determined to be a normal computing power node. Figure 1 shown.

[0068] The working principle of the above technical solution is as follows: the system first receives computing demand information from users or applications, which may include the type, scale, priority, etc. of computing tasks. Based on this information, the system extracts the corresponding computing task data from the database and prepares for subsequent processing and allocation. For each computing task, the system generates a corresponding computing progress bar to visualize the completion progress of the task. At the same time, the system monitors the computing power monitoring node corresponding to each computing progress bar, and obtains information such as the computing power usage and load status of the node in real time. By analyzing the current usage of computing power resources, the priority and urgency of the task, the system calculates a computing power resource coefficient. Based on this coefficient, the system dynamically allocates computing power resources to ensure that high-priority or urgent tasks can obtain sufficient computing power resources. After allocation, the system evaluates the remaining reserve computing power resources to prepare for possible computing power shortages. When it is found that the reserve computing power resources are insufficient, the system compensates the computing power resources according to the compensation resource data to ensure that the system can operate continuously and stably. The system calculates the computing power abnormality coefficient and identifies abnormal computing power nodes by comparing the difference between the actual computing power and the expected computing power. Nodes are divided into abnormal computing nodes and normal computing nodes, and detailed information of abnormal nodes is further obtained, such as fault type, impact range, etc. For abnormal computing nodes, the system calculates the required resource compensation data, including computing power, storage, network and other resources. These resource compensation data are dispatched to abnormal nodes for fault recovery or performance optimization until the nodes return to normal and are re-identified as normal computing nodes by the system.

[0069] The technical effects of the above technical solutions are: by dynamically allocating computing resources and timely compensating for reserve computing power, computing resources are ensured to be efficiently utilized and resource waste is reduced. Real-time monitoring of computing node status, timely discovery and processing of abnormal nodes, effective prevention of single point failure spread, and improvement of overall system stability. By visualizing the computing progress and dynamically adjusting task priorities, users can more intuitively understand the task execution status and improve user satisfaction. The automated monitoring, allocation, and compensation mechanisms reduce the need for manual intervention and reduce operation and maintenance costs and time costs. The system can flexibly respond to the needs of large-scale computing tasks, and ensure that tasks are completed on time and with quality through dynamic expansion and resource optimization.

[0070] In one embodiment of the present invention, the S1 includes:

[0071] Obtain computing demand information, extract computing task data according to the computing demand information according to a plurality of preset different computing task types, and obtain multiple computing target data of different computing task types; the computing task types include serial computing and parallel computing, real-time computing and non-real-time computing, or small-scale tasks and large-scale tasks, etc.

[0072] The calculation target data of each calculation task type are sorted according to the calculation target task order to obtain a calculation progress bar corresponding to the calculation task type;

[0073] Each calculation target data in each calculation progress bar is a computing power monitoring node, such as Figure 2 shown.

[0074] The working principle of the above technical solution is as follows: the system first receives computing demand information from users, applications or other system components. This information describes in detail the specific requirements of the computing task, such as task type, scale, priority, deadline, etc. According to the received computing demand information, the system extracts the corresponding computing task data according to a preset variety of different computing task types (such as serial computing and parallel computing, real-time computing and non-real-time computing, small-scale tasks and large-scale tasks, etc.). Under each computing task type, multiple computing target data will be extracted, which represent the specific goals or subtasks that need to be completed under this task type. For each computing task type, the system will sort the extracted computing target data according to the order of the computing target tasks (which may be based on priority, dependency, execution order, etc.). The purpose of sorting is to determine the order of task execution, so as to ensure that the tasks can be carried out according to the predetermined plan. Based on the sorted computing target data, the system generates a corresponding computing progress bar for each computing task type. The computing progress bar is used to visually display the completion progress of the task, helping users or system administrators to understand the execution of the task in real time. In each computing progress bar, each computing target data is regarded as a computing power monitoring node. These nodes are used to monitor and record the computing power usage, execution time, whether anomalies occur, etc. of the target data during execution. Through these nodes, the system can grasp the execution status of the task in real time and make corresponding adjustments or optimizations in a timely manner.

[0075] The technical effect of the above technical solution is: by classifying, extracting and sorting the computing task data, the system can more clearly understand the specific goals and execution order of each task type, thereby improving the efficiency and accuracy of task management. The setting of the computing progress bar and the computing power monitoring node makes the execution process of the task more transparent, and the user or system administrator can understand the progress and status of the task in real time, and promptly discover and handle problems. Based on the information of the computing progress bar and the computing power monitoring node, the system can more reasonably allocate and schedule computing resources to ensure that critical tasks or urgent tasks can obtain sufficient computing power support first, thereby improving resource utilization and task execution efficiency. By real-time monitoring of the status of the computing power monitoring node, the system can promptly discover and handle abnormal situations, such as insufficient computing power, task timeout, etc., thereby avoiding the spread of single point failures to the entire system and improving the stability and reliability of the system. The system can flexibly respond to a variety of different types and requirements of computing tasks, including serial computing and parallel computing, real-time computing and non-real-time computing, small-scale tasks and large-scale tasks, etc., to provide personalized computing services for different users.

[0076] In one embodiment of the present invention, S2 includes:

[0077] Obtain the computing target data of each computing power monitoring node, and calculate the computing power resource coefficient of the computing power monitoring node according to each computing target data;

[0078] The calculation formula of the computing power resource coefficient is:

[0079]

[0080] Among them, Z sl is the computing resource coefficient, S zy is the total computing power resource data, D ms is the target data volume of the current computing power monitoring node, j is the total number of computing power monitoring nodes other than the current computing power monitoring node, L si is the calculation target data volume of the i-th computing power monitoring node;

[0081] Allocate computing power resources to each computing power monitoring node according to the computing power resource coefficient to obtain computing power resource allocation data;

[0082] Obtaining remaining computing resources according to the computing resource allocation data, where the remaining computing resources are the reserve computing resources;

[0083] Compare the reserve computing power resources with a preset resource threshold to obtain a resource comparison result;

[0084] Acquire compensation resource data according to the resource comparison result;

[0085] When the reserve computing power resource is greater than a preset resource threshold, obtaining compensation resource data;

[0086] When the reserve computing power resource is less than or equal to the preset resource threshold, no compensation resource data is obtained;

[0087] The compensation resource data is the difference between the reserve computing power resources and the preset resource threshold;

[0088] Collecting compensation resources according to the compensation resource data to obtain compensation resource collection data;

[0089] The compensation resource collection data is the idle resource data of other computing power monitoring nodes;

[0090] The reserve computing power resources are compensated according to the compensation resource collection data until the reserve computing power resources are greater than a preset resource threshold.

[0091] The working principle of the above technical solution is as follows: the system first obtains the computing target data of each computing power monitoring node, which includes key information such as task type, scale, and execution time. The system calculates the computing power resource coefficient of each computing power monitoring node. This coefficient reflects the amount of computing power resources required by the node under the current task. Based on the computing power resource coefficient, the system allocates computing power resources to each computing power monitoring node and generates computing power resource allocation data. When allocating, the system will dynamically adjust according to the availability of overall computing power resources and the priority of tasks to ensure that critical tasks or urgent tasks can obtain sufficient computing power resources first. After allocating computing power resources, the system calculates the remaining computing power resources, which are the reserve computing power resources. Reserve computing power resources are used to respond to sudden tasks or changes in computing power requirements. The system compares the reserve computing power resources with the preset resource threshold to evaluate whether the current reserves are sufficient. The resource threshold is a preset standard value set according to the overall operation and needs of the system. If the reserve computing power resources are greater than the preset resource threshold, it means that the current reserve is sufficient and no compensation is required; if it is less than or equal to the preset resource threshold, the system calculates the amount of resources that need to be compensated, that is, the difference between the reserve computing power resources and the preset resource threshold. The system begins to collect compensation resources, which mainly come from the idle resources of other computing power monitoring nodes. Through the monitoring and scheduling mechanism, the system collects the resources of nodes with low current load and underutilized computing power resources and uses them as compensation resources. After collecting enough compensation resources, the system allocates these resources to the reserve computing power resources until the reserve computing power resources are greater than the preset resource threshold again. In this way, the system will have sufficient computing power resources to cope with sudden tasks or increased computing power demand.

[0092] The technical effect of the above technical solution is: by calculating the computing power resource coefficient in real time and dynamically allocating computing power resources, the system can ensure that critical tasks or emergency tasks obtain sufficient computing power support and improve the execution efficiency and accuracy of tasks. By collecting and using the idle resources of other computing power monitoring nodes as compensation resources, the system can maximize the use of overall computing power resources, reduce resource waste, and improve resource utilization. Through the preset resource threshold and timely compensation mechanism, the system can quickly replenish when the reserve computing power resources are insufficient, ensuring that the system can operate stably in the face of emergencies, and avoiding task failures or system crashes caused by insufficient computing power. The automated resource allocation and compensation mechanism reduces the need for manual intervention and reduces operation and maintenance costs and time costs. At the same time, by optimizing resource utilization, the system can reduce dependence on and investment in hardware resources while ensuring task execution efficiency. The system can flexibly adjust resource allocation and compensation strategies according to actual computing power requirements and resource conditions, and support flexible expansion and changes of computing tasks. This provides users with more flexible and efficient computing service support.

[0093] In one embodiment of the present invention, S3 includes:

[0094] Obtaining actual calculation data of each computing power monitoring node, and calculating a computing power anomaly coefficient according to the actual calculation data;

[0095] The calculation formula of the computing power abnormality coefficient is:

[0096]

[0097] Among them, S yc is the computing power abnormality coefficient, T d is the current calculation duration, T y Preset the duration for the total computation, W cl is the current calculation error, Z sl is the total amount of calculation, Z dx is the current computing resource consumption, Z sy is the total computing resources;

[0098] Compare the computing power abnormality coefficient with a preset computing power abnormality threshold to obtain an abnormal comparison result;

[0099] When the computing power abnormality coefficient is greater than the preset computing power abnormality threshold, the corresponding computing power monitoring node is marked as an abnormal computing power node;

[0100] When the computing power abnormality coefficient is less than or equal to the preset computing power abnormality threshold, the corresponding computing power monitoring node is marked as a normal computing power node;

[0101] Identify and obtain abnormal computing power node information.

[0102] The working principle of the above technical solution is as follows: the system first collects the actual computing data of each computing power monitoring node, which includes but is not limited to key indicators such as the execution time of the computing task, resource utilization, and task completion. Based on the collected actual computing data, the system evaluates the performance of each computing power monitoring node. This formula usually compares the actual computing data with the expected or standard value to obtain a coefficient that reflects the degree of abnormality of the node computing power. The system compares the calculated computing power abnormality coefficient with the preset computing power abnormality threshold. This threshold is used to determine whether the node is in an abnormal state. If the computing power abnormality coefficient is greater than the preset threshold, it means that the performance of the computing power monitoring node is abnormal, and the system will mark it as an abnormal computing power node. This means that the node may require further investigation, optimization or repair. If the computing power abnormality coefficient is less than or equal to the preset threshold, it means that the node is performing normally, and the system will mark it as a normal computing power node. For monitoring points marked as abnormal computing power nodes, the system will further identify and obtain their detailed information, such as node ID, abnormality type, abnormality occurrence time, possible causes, etc. This information is crucial for subsequent troubleshooting, performance optimization, and resource scheduling.

[0103] The technical effect of the above technical solution is: by real-time monitoring and calculating the computing power abnormality coefficient, the system can timely discover and mark the computing power monitoring nodes with abnormal performance, so as to avoid the expansion of abnormal problems and affect the execution of the overall computing tasks. Timely discovery and handling of computing power anomalies helps reduce the risk of system crashes due to single point failures and improve the stability and reliability of the system. After identifying abnormal computing power nodes, the system can reschedule resources according to actual conditions and transfer the computing power resources originally allocated to abnormal nodes to normal nodes or tasks with higher priority, thereby improving resource utilization efficiency. The automated anomaly detection and labeling mechanism reduces the need for manual intervention and reduces operation and maintenance costs and time costs. At the same time, by optimizing resource utilization and reducing the occurrence of failures, the overall operating cost of the system will also be reduced. Obtaining detailed abnormal computing power node information helps operation and maintenance personnel to more accurately locate the root cause of the problem, develop solutions and optimize performance. This refined operation and maintenance method can further improve the overall performance and user experience of the system.

[0104] In one embodiment of the present invention, the S4 includes:

[0105] Obtain computing power resource allocation data of the abnormal computing power node, and calculate resource compensation data of the abnormal computing power node according to the computing power resource allocation data combined with the abnormal computing power node information;

[0106] The calculation formula of the resource compensation data is:

[0107]

[0108] Among them, ΔBL is the resource compensation data, F zy Allocate computing resources to abnormal computing nodes, S yd is the abnormal computing power coefficient of the abnormal computing power node, D is the total number of abnormal computing power nodes, S yci is the computing power abnormality coefficient of the i-th abnormal computing power node;

[0109] The resource compensation data is dispatched to the abnormal computing power node by using the reserved computing power resources to obtain the dispatch completion data;

[0110] After the scheduling is completed, the abnormal computing power coefficient of the abnormal computing power node is recalculated to obtain the recalculation result until the abnormal computing power node is determined to be a normal computing power node.

[0111] The working principle of the above technical solution is as follows: when a computing power monitoring node is marked as abnormal, the system will first obtain the current computing power resource allocation data of the node. These data include the amount of resources such as CPU, memory, storage, etc. allocated to the node and the actual use of these resources. The system will calculate the required resource compensation amount using the preset resource compensation data calculation formula based on the acquired computing power resource allocation data and the specific information of the abnormal computing power node (such as the type of abnormality, the degree of abnormality, etc.). Once the resource compensation data is determined, the system will dispatch the corresponding amount of resources from the reserve computing power resources to the abnormal computing power node. This scheduling process may be real-time or may be carried out at a suitable time to ensure the overall stability and performance of the system. After the scheduling is completed, the system will record the scheduling completion data, including the scheduling time, resource amount, target node and other information. Then, the system will recalculate the computing power abnormality coefficient of the abnormal computing power node to evaluate the effect after scheduling. If the recalculation result shows that the computing power abnormality coefficient of the abnormal computing power node has dropped below the preset threshold, it means that the node has returned to normal, and the system will mark it as a normal computing power node. If it is still above the threshold, the system may continue to perform resource scheduling or take other measures to further resolve the problem.

[0112] The technical effect of the above technical solution is: by timely obtaining the resource allocation data of abnormal computing power nodes and calculating resource compensation data, the system can quickly respond to computing power anomalies and reduce the impact of anomalies on the overall performance of the system. The formula takes into account the specific computing power anomaly coefficient (S yci ), can calculate the precise resource compensation amount (ΔBL) for each node. The larger the abnormal coefficient of the computing power node in the formula, the more resources are allocated to the node. This accuracy helps to more effectively solve the computing power abnormality problem and ensure that resources are reasonably allocated to where they are most needed. yciIt is updated in real time or regularly, so it can reflect the current abnormal computing power status of the system. This dynamic adaptability enables the resource compensation strategy to be flexibly adjusted as the system status changes, so as to better cope with emergencies. D (the total number of abnormal computing power nodes) in the formula is one of the denominators, ensuring the fair distribution of resource compensation among multiple abnormal nodes. Even if there are multiple abnormal computing power nodes in the system, each node can obtain a corresponding proportion of resource compensation according to the degree of its abnormal computing power. By timely compensating resources for abnormal computing power nodes, this formula helps to reduce task delays, failures or performance degradation caused by insufficient computing power. This can not only improve the completion rate and quality of tasks, but also optimize the overall resource utilization efficiency and reduce operating costs. Resource compensation is an important part of system stability management. The resource compensation data calculated by this formula can solve the problem of abnormal computing power in a targeted manner, thereby enhancing the stability and reliability of the system. When computing power resources are tight or there is a sudden failure, the resource compensation strategy provided by this formula can respond quickly and restore the normal operation of the system. By dispatching reserve computing power resources to abnormal nodes, the system can quickly restore the computing power performance of nodes, prevent the spread of abnormalities, and improve the stability and reliability of the system. The calculation of resource compensation data takes into account a variety of factors to ensure the rationality and efficiency of resource scheduling. This helps to optimize the overall resource utilization of the system and avoid resource waste. The automated resource compensation and scheduling mechanism reduces the need for manual intervention and reduces operation and maintenance costs and time costs. At the same time, by quickly resolving computing power anomalies, the system can reduce downtime and losses caused by failures. Recalculating the computing power anomaly coefficient of the abnormal computing power node after scheduling is completed helps the system continuously monitor the performance of the node and make further optimizations and adjustments as needed. This continuous monitoring and optimization mechanism can continuously improve the overall performance of the system and user experience.

[0113] In one embodiment of the present invention, the system comprises:

[0114] A node distribution module is used to obtain computing demand information, extract computing task data according to the computing demand information, obtain a computing progress bar of the computing task type and a computing power monitoring node of each computing progress bar;

[0115] A resource allocation module, used to calculate the computing power resource coefficient, perform computing power resource allocation, obtain the allocated reserve computing power resources, obtain compensation resource data according to the reserve computing power resources, and compensate the reserve computing power resources according to the compensation resource data;

[0116] An abnormality identification module is used to calculate a computing power abnormality coefficient, obtain abnormal computing power nodes and normal computing power nodes according to the computing power abnormality coefficient, and then obtain abnormal computing power node information of the abnormal computing power node;

[0117] The abnormal allocation module is used to calculate the resource compensation data of the abnormal computing power node and dispatch the resource compensation data to the abnormal computing power node until the abnormal computing power node is determined to be a normal computing power node.

[0118] The working principle of the above technical solution is as follows: the system first receives computing demand information from users or applications, which may include the type, scale, priority, etc. of computing tasks. Based on this information, the system extracts the corresponding computing task data from the database and prepares for subsequent processing and allocation. For each computing task, the system generates a corresponding computing progress bar to visualize the completion progress of the task. At the same time, the system monitors the computing power monitoring node corresponding to each computing progress bar, and obtains information such as the computing power usage and load status of the node in real time. By analyzing the current usage of computing power resources, the priority and urgency of the task, the system calculates a computing power resource coefficient. Based on this coefficient, the system dynamically allocates computing power resources to ensure that high-priority or urgent tasks can obtain sufficient computing power resources. After allocation, the system evaluates the remaining reserve computing power resources to prepare for possible computing power shortages. When it is found that the reserve computing power resources are insufficient, the system compensates the computing power resources according to the compensation resource data to ensure that the system can operate continuously and stably. The system calculates the computing power abnormality coefficient and identifies abnormal computing power nodes by comparing the difference between the actual computing power and the expected computing power. Nodes are divided into abnormal computing nodes and normal computing nodes, and detailed information of abnormal nodes is further obtained, such as fault type, impact range, etc. For abnormal computing nodes, the system calculates the required resource compensation data, including computing power, storage, network and other resources. These resource compensation data are dispatched to abnormal nodes for fault recovery or performance optimization until the nodes return to normal and are re-identified as normal computing nodes by the system.

[0119] The technical effects of the above technical solutions are: by dynamically allocating computing resources and timely compensating for reserve computing power, computing resources are ensured to be efficiently utilized and resource waste is reduced. Real-time monitoring of computing node status, timely discovery and processing of abnormal nodes, effective prevention of single point failure spread, and improvement of overall system stability. By visualizing the computing progress and dynamically adjusting task priorities, users can more intuitively understand the task execution status and improve user satisfaction. The automated monitoring, allocation, and compensation mechanisms reduce the need for manual intervention and reduce operation and maintenance costs and time costs. The system can flexibly respond to the needs of large-scale computing tasks, and ensure that tasks are completed on time and with quality through dynamic expansion and resource optimization.

[0120] In one embodiment of the present invention, the node distribution module includes:

[0121] Obtain computing demand information, extract computing task data according to the computing demand information according to a plurality of preset different computing task types, and obtain multiple computing target data of different computing task types; the computing task types include serial computing and parallel computing, real-time computing and non-real-time computing, or small-scale tasks and large-scale tasks, etc.

[0122] The calculation target data of each calculation task type are sorted according to the calculation target task order to obtain a calculation progress bar corresponding to the calculation task type;

[0123] Each calculation target data in each calculation progress bar is a computing power monitoring node.

[0124] The working principle of the above technical solution is as follows: the system first receives computing demand information from users, applications or other system components. This information describes in detail the specific requirements of the computing task, such as task type, scale, priority, deadline, etc. According to the received computing demand information, the system extracts the corresponding computing task data according to a preset variety of different computing task types (such as serial computing and parallel computing, real-time computing and non-real-time computing, small-scale tasks and large-scale tasks, etc.). Under each computing task type, multiple computing target data will be extracted, which represent the specific goals or subtasks that need to be completed under this task type. For each computing task type, the system will sort the extracted computing target data according to the order of the computing target tasks (which may be based on priority, dependency, execution order, etc.). The purpose of sorting is to determine the order of task execution, so as to ensure that the tasks can be carried out according to the predetermined plan. Based on the sorted computing target data, the system generates a corresponding computing progress bar for each computing task type. The computing progress bar is used to visually display the completion progress of the task, helping users or system administrators to understand the execution of the task in real time. In each computing progress bar, each computing target data is regarded as a computing power monitoring node. These nodes are used to monitor and record the computing power usage, execution time, whether anomalies occur, etc. of the target data during execution. Through these nodes, the system can grasp the execution status of the task in real time and make corresponding adjustments or optimizations in a timely manner.

[0125] The technical effect of the above technical solution is: by classifying, extracting and sorting the computing task data, the system can more clearly understand the specific goals and execution order of each task type, thereby improving the efficiency and accuracy of task management. The setting of the computing progress bar and the computing power monitoring node makes the execution process of the task more transparent, and the user or system administrator can understand the progress and status of the task in real time, and promptly discover and handle problems. Based on the information of the computing progress bar and the computing power monitoring node, the system can more reasonably allocate and schedule computing resources to ensure that critical tasks or urgent tasks can obtain sufficient computing power support first, thereby improving resource utilization and task execution efficiency. By real-time monitoring of the status of the computing power monitoring node, the system can promptly discover and handle abnormal situations, such as insufficient computing power, task timeout, etc., thereby avoiding the spread of single point failures to the entire system and improving the stability and reliability of the system. The system can flexibly respond to a variety of different types and requirements of computing tasks, including serial computing and parallel computing, real-time computing and non-real-time computing, small-scale tasks and large-scale tasks, etc., to provide personalized computing services for different users.

[0126] In one embodiment of the present invention, the resource allocation module includes:

[0127] Obtain the computing target data of each computing power monitoring node, and calculate the computing power resource coefficient of the computing power monitoring node according to each computing target data;

[0128] The calculation formula of the computing power resource coefficient is:

[0129]

[0130] Among them, Z sl is the computing resource coefficient, S zy is the total computing power resource data, D ms is the target data volume of the current computing power monitoring node, j is the total number of computing power monitoring nodes other than the current computing power monitoring node, L si is the calculation target data volume of the i-th computing power monitoring node;

[0131] Allocate computing power resources to each computing power monitoring node according to the computing power resource coefficient to obtain computing power resource allocation data;

[0132] Obtaining remaining computing resources according to the computing resource allocation data, where the remaining computing resources are the reserve computing resources;

[0133] Compare the reserve computing power resources with a preset resource threshold to obtain a resource comparison result;

[0134] Acquire compensation resource data according to the resource comparison result;

[0135] When the reserve computing power resource is greater than a preset resource threshold, obtaining compensation resource data;

[0136] When the reserve computing power resource is less than or equal to the preset resource threshold, no compensation resource data is obtained;

[0137] The compensation resource data is the difference between the reserve computing power resources and the preset resource threshold;

[0138] Collecting compensation resources according to the compensation resource data to obtain compensation resource collection data;

[0139] The compensation resource collection data is the idle resource data of other computing power monitoring nodes;

[0140] The reserve computing power resources are compensated according to the compensation resource collection data until the reserve computing power resources are greater than a preset resource threshold.

[0141] The working principle of the above technical solution is as follows: the system first obtains the computing target data of each computing power monitoring node, which includes key information such as task type, scale, and execution time. The system calculates the computing power resource coefficient of each computing power monitoring node. This coefficient reflects the amount of computing power resources required by the node under the current task. Based on the computing power resource coefficient, the system allocates computing power resources to each computing power monitoring node and generates computing power resource allocation data. When allocating, the system will dynamically adjust according to the availability of overall computing power resources and the priority of tasks to ensure that critical tasks or urgent tasks can obtain sufficient computing power resources first. After allocating computing power resources, the system calculates the remaining computing power resources, which are the reserve computing power resources. Reserve computing power resources are used to respond to sudden tasks or changes in computing power requirements. The system compares the reserve computing power resources with the preset resource threshold to evaluate whether the current reserves are sufficient. The resource threshold is a preset standard value set according to the overall operation and needs of the system. If the reserve computing power resources are greater than the preset resource threshold, it means that the current reserve is sufficient and no compensation is required; if it is less than or equal to the preset resource threshold, the system calculates the amount of resources that need to be compensated, that is, the difference between the reserve computing power resources and the preset resource threshold. The system begins to collect compensation resources, which mainly come from the idle resources of other computing power monitoring nodes. Through the monitoring and scheduling mechanism, the system collects the resources of nodes with low current load and underutilized computing power resources and uses them as compensation resources. After collecting enough compensation resources, the system allocates these resources to the reserve computing power resources until the reserve computing power resources are greater than the preset resource threshold again. In this way, the system will have sufficient computing power resources to cope with sudden tasks or increased computing power demand.

[0142] The technical effect of the above technical solution is: by calculating the computing power resource coefficient in real time and dynamically allocating computing power resources, the system can ensure that critical tasks or emergency tasks obtain sufficient computing power support and improve the execution efficiency and accuracy of tasks. By collecting and using the idle resources of other computing power monitoring nodes as compensation resources, the system can maximize the use of overall computing power resources, reduce resource waste, and improve resource utilization. Through the preset resource threshold and timely compensation mechanism, the system can quickly replenish when the reserve computing power resources are insufficient, ensuring that the system can operate stably in the face of emergencies, and avoiding task failures or system crashes caused by insufficient computing power. The automated resource allocation and compensation mechanism reduces the need for manual intervention and reduces operation and maintenance costs and time costs. At the same time, by optimizing resource utilization, the system can reduce dependence on and investment in hardware resources while ensuring task execution efficiency. The system can flexibly adjust resource allocation and compensation strategies according to actual computing power requirements and resource conditions, and support flexible expansion and changes of computing tasks. This provides users with more flexible and efficient computing service support.

[0143] In one embodiment of the present invention, the abnormality identification module includes:

[0144] Obtaining actual calculation data of each computing power monitoring node, and calculating a computing power anomaly coefficient according to the actual calculation data;

[0145] The calculation formula of the computing power abnormality coefficient is:

[0146]

[0147] Among them, S yc is the computing power abnormality coefficient, T d is the current calculation duration, T y Preset the duration for the total computation, W cl is the current calculation error, Z sl is the total amount of calculation, Z dx is the current computing resource consumption, Z sy is the total computing resources;

[0148] Compare the computing power abnormality coefficient with a preset computing power abnormality threshold to obtain an abnormal comparison result;

[0149] When the computing power abnormality coefficient is greater than the preset computing power abnormality threshold, the corresponding computing power monitoring node is marked as an abnormal computing power node;

[0150] When the computing power abnormality coefficient is less than or equal to the preset computing power abnormality threshold, the corresponding computing power monitoring node is marked as a normal computing power node;

[0151] Identify and obtain abnormal computing power node information.

[0152] The working principle of the above technical solution is as follows: the system first collects the actual computing data of each computing power monitoring node, which includes but is not limited to key indicators such as the execution time of the computing task, resource utilization, and task completion. Based on the collected actual computing data, the system evaluates the performance of each computing power monitoring node. This formula usually compares the actual computing data with the expected or standard value to obtain a coefficient that reflects the degree of abnormality of the node computing power. The system compares the calculated computing power abnormality coefficient with the preset computing power abnormality threshold. This threshold is used to determine whether the node is in an abnormal state. If the computing power abnormality coefficient is greater than the preset threshold, it means that the performance of the computing power monitoring node is abnormal, and the system will mark it as an abnormal computing power node. This means that the node may require further investigation, optimization or repair. If the computing power abnormality coefficient is less than or equal to the preset threshold, it means that the node is performing normally, and the system will mark it as a normal computing power node. For monitoring points marked as abnormal computing power nodes, the system will further identify and obtain their detailed information, such as node ID, abnormality type, abnormality occurrence time, possible causes, etc. This information is crucial for subsequent troubleshooting, performance optimization, and resource scheduling.

[0153] The technical effect of the above technical solution is: by real-time monitoring and calculating the computing power abnormality coefficient, the system can timely discover and mark the computing power monitoring nodes with abnormal performance, so as to avoid the expansion of abnormal problems and affect the execution of the overall computing tasks. Timely discovery and handling of computing power anomalies helps reduce the risk of system crashes due to single point failures and improve the stability and reliability of the system. After identifying abnormal computing power nodes, the system can reschedule resources according to actual conditions and transfer the computing power resources originally allocated to abnormal nodes to normal nodes or tasks with higher priority, thereby improving resource utilization efficiency. The automated anomaly detection and labeling mechanism reduces the need for manual intervention and reduces operation and maintenance costs and time costs. At the same time, by optimizing resource utilization and reducing the occurrence of failures, the overall operating cost of the system will also be reduced. Obtaining detailed abnormal computing power node information helps operation and maintenance personnel to more accurately locate the root cause of the problem, develop solutions and optimize performance. This refined operation and maintenance method can further improve the overall performance and user experience of the system.

[0154] In one embodiment of the present invention, the abnormal distribution module includes:

[0155] Obtain computing power resource allocation data of the abnormal computing power node, and calculate resource compensation data of the abnormal computing power node according to the computing power resource allocation data combined with the abnormal computing power node information;

[0156] The calculation formula of the resource compensation data is:

[0157]

[0158] Among them, ΔBL is the resource compensation data, F zy Allocate computing resources to abnormal computing nodes, S yd is the abnormal computing power coefficient of the abnormal computing power node, D is the total number of abnormal computing power nodes, S yci is the computing power abnormality coefficient of the i-th abnormal computing power node;

[0159] The resource compensation data is dispatched to the abnormal computing power node by using the reserved computing power resources to obtain the dispatch completion data;

[0160] After the scheduling is completed, the abnormal computing power coefficient of the abnormal computing power node is recalculated to obtain the recalculation result until the abnormal computing power node is determined to be a normal computing power node.

[0161] The working principle of the above technical solution is as follows: when a computing power monitoring node is marked as abnormal, the system will first obtain the current computing power resource allocation data of the node. These data include the amount of resources such as CPU, memory, storage, etc. allocated to the node and the actual use of these resources. The system will calculate the required resource compensation amount using the preset resource compensation data calculation formula based on the acquired computing power resource allocation data and the specific information of the abnormal computing power node (such as the type of abnormality, the degree of abnormality, etc.). Once the resource compensation data is determined, the system will dispatch the corresponding amount of resources from the reserve computing power resources to the abnormal computing power node. This scheduling process may be real-time or may be carried out at a suitable time to ensure the overall stability and performance of the system. After the scheduling is completed, the system will record the scheduling completion data, including the scheduling time, resource amount, target node and other information. Then, the system will recalculate the computing power abnormality coefficient of the abnormal computing power node to evaluate the effect after scheduling. If the recalculation result shows that the computing power abnormality coefficient of the abnormal computing power node has dropped below the preset threshold, it means that the node has returned to normal, and the system will mark it as a normal computing power node. If it is still above the threshold, the system may continue to perform resource scheduling or take other measures to further resolve the problem.

[0162] The technical effect of the above technical solution is: by timely obtaining the resource allocation data of the abnormal computing power node and calculating the resource compensation data, the system can quickly respond to the computing power anomaly and reduce the impact of the anomaly on the overall performance of the system. By dispatching the reserve computing power resources to the abnormal node, the system can quickly restore the computing power performance of the node, prevent the spread of anomalies, and improve the stability and reliability of the system. The calculation of resource compensation data takes into account a variety of factors to ensure the rationality and efficiency of resource scheduling. This helps to optimize the overall resource utilization of the system and avoid resource waste. The automated resource compensation and scheduling mechanism reduces the need for manual intervention and reduces the operation and maintenance costs and time costs. At the same time, by quickly resolving computing power anomalies, the system can reduce downtime and losses caused by failures. Recalculating the computing power anomaly coefficient of the abnormal computing power node after the scheduling is completed helps the system to continuously monitor the performance of the node and make further optimizations and adjustments as needed. This continuous monitoring and optimization mechanism can continuously improve the overall performance and user experience of the system.

[0163] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A computing power anomaly detection method based on data analysis, characterized in that: The method comprises: S1. Obtain computing demand information, extract computing task data according to the computing demand information, obtain computing progress bars of computing task types and computing power monitoring nodes of each computing progress bar; Wherein, the S1 includes: Obtain computing demand information, extract computing task data according to a plurality of preset different computing task types based on the computing demand information, and obtain a plurality of computing target data of different computing task types; The calculation target data of each calculation task type are sorted according to the calculation target task order to obtain a calculation progress bar corresponding to the calculation task type; Each calculation target data in each calculation progress bar is a computing power monitoring node; S2. Calculate the computing power resource coefficient, allocate computing power resources, obtain the allocated reserve computing power resources, obtain compensation resource data according to the reserve computing power resources, and compensate the reserve computing power resources according to the compensation resource data; Wherein, the S2 includes: Obtain the computing target data of each computing power monitoring node, and calculate the computing power resource coefficient of the computing power monitoring node according to each computing target data; The calculation formula of the computing power resource coefficient is: Among them, Z sl is the computing resource coefficient, S zy is the total computing power resource data, D ms is the target data volume of the current computing power monitoring node, j is the total number of computing power monitoring nodes other than the current computing power monitoring node, L si is the calculation target data volume of the i-th computing power monitoring node; Allocate computing power resources to each computing power monitoring node according to the computing power resource coefficient to obtain computing power resource allocation data; Obtaining remaining computing resources according to the computing resource allocation data, where the remaining computing resources are the reserve computing resources; Compare the reserve computing power resources with a preset resource threshold to obtain a resource comparison result; Acquire compensation resource data according to the resource comparison result; Collecting compensation resources according to the compensation resource data to obtain compensation resource collection data; Compensating the reserve computing power resources according to the compensation resource collection data until the reserve computing power resources are greater than a preset resource threshold; S3. Calculate the computing power abnormality coefficient, obtain the abnormal computing power node and the normal computing power node according to the computing power abnormality coefficient, and then obtain the abnormal computing power node information of the abnormal computing power node; Wherein, the S3 includes: Obtaining actual calculation data of each computing power monitoring node, and calculating a computing power anomaly coefficient according to the actual calculation data; The calculation formula of the computing power abnormality coefficient is: Among them, S yc is the computing power abnormality coefficient, T d is the current calculation duration, T y Preset the duration for the total computation, W cl is the current calculation error, Z sl is the total amount of calculation, Z dx is the current computing resource consumption, Z sy is the total computing resources; Compare the computing power abnormality coefficient with a preset computing power abnormality threshold to obtain an abnormal comparison result; When the computing power abnormality coefficient is greater than the preset computing power abnormality threshold, the corresponding computing power monitoring node is marked as an abnormal computing power node; When the computing power abnormality coefficient is less than or equal to the preset computing power abnormality threshold, the corresponding computing power monitoring node is marked as a normal computing power node; Identify and obtain abnormal computing power node information; S4. Calculate resource compensation data of the abnormal computing power node, and dispatch the resource compensation data to the abnormal computing power node until the abnormal computing power node is determined to be a normal computing power node; Wherein, the S4 includes: Obtain computing power resource allocation data of the abnormal computing power node, and calculate resource compensation data of the abnormal computing power node according to the computing power resource allocation data combined with the abnormal computing power node information; The calculation formula of the resource compensation data is: Among them, ΔBL is the resource compensation data, F zy Allocate computing resources to abnormal computing nodes, S yd is the abnormal computing power coefficient of the abnormal computing power node, D is the total number of abnormal computing power nodes, S yci is the computing power abnormality coefficient of the i-th abnormal computing power node; The resource compensation data is dispatched to the abnormal computing power node by using the reserved computing power resources to obtain the dispatch completion data; After the scheduling is completed, the abnormal computing power coefficient of the abnormal computing power node is recalculated to obtain the recalculation result until the abnormal computing power node is determined to be a normal computing power node.

2. A computing power anomaly detection system based on data analysis, characterized in that: The system comprises: A node distribution module is used to obtain computing demand information, extract computing task data according to the computing demand information, obtain a computing progress bar of the computing task type and a computing power monitoring node of each computing progress bar; Wherein, the node distribution module includes: Obtain computing demand information, extract computing task data according to a plurality of preset different computing task types based on the computing demand information, and obtain a plurality of computing target data of different computing task types; The calculation target data of each calculation task type are sorted according to the calculation target task order to obtain a calculation progress bar corresponding to the calculation task type; Each calculation target data in each calculation progress bar is a computing power monitoring node; A resource allocation module, used to calculate the computing power resource coefficient, perform computing power resource allocation, obtain the allocated reserve computing power resources, obtain compensation resource data according to the reserve computing power resources, and compensate the reserve computing power resources according to the compensation resource data; Wherein, the resource allocation module includes: Obtain the computing target data of each computing power monitoring node, and calculate the computing power resource coefficient of the computing power monitoring node according to each computing target data; The calculation formula of the computing power resource coefficient is: Among them, Z sl is the computing resource coefficient, S zy is the total computing power resource data, D ms is the target data volume of the current computing power monitoring node, j is the total number of computing power monitoring nodes other than the current computing power monitoring node, L si is the calculation target data volume of the i-th computing power monitoring node; Allocate computing power resources to each computing power monitoring node according to the computing power resource coefficient to obtain computing power resource allocation data; Obtaining remaining computing resources according to the computing resource allocation data, where the remaining computing resources are the reserve computing resources; Compare the reserve computing power resources with a preset resource threshold to obtain a resource comparison result; Acquire compensation resource data according to the resource comparison result; Collecting compensation resources according to the compensation resource data to obtain compensation resource collection data; Compensating the reserve computing power resources according to the compensation resource collection data until the reserve computing power resources are greater than a preset resource threshold; An abnormality identification module is used to calculate a computing power abnormality coefficient, obtain abnormal computing power nodes and normal computing power nodes according to the computing power abnormality coefficient, and then obtain abnormal computing power node information of the abnormal computing power node; Wherein, the abnormality identification module includes: Obtaining actual calculation data of each computing power monitoring node, and calculating a computing power anomaly coefficient according to the actual calculation data; The calculation formula of the computing power abnormality coefficient is: Among them, S yc is the computing power abnormality coefficient, T d is the current calculation duration, T y Preset the duration for the total computation, W cl is the current calculation error, Z sl is the total amount of calculation, Z dx is the current computing resource consumption, Z sy is the total computing resources; Compare the computing power abnormality coefficient with a preset computing power abnormality threshold to obtain an abnormal comparison result; When the computing power abnormality coefficient is greater than the preset computing power abnormality threshold, the corresponding computing power monitoring node is marked as an abnormal computing power node; When the computing power abnormality coefficient is less than or equal to the preset computing power abnormality threshold, the corresponding computing power monitoring node is marked as a normal computing power node; Identify and obtain abnormal computing power node information; An abnormal allocation module is used to calculate the resource compensation data of the abnormal computing power node and dispatch the resource compensation data to the abnormal computing power node until the abnormal computing power node is determined to be a normal computing power node; Wherein, the abnormal allocation module includes: Obtain computing power resource allocation data of the abnormal computing power node, and calculate resource compensation data of the abnormal computing power node according to the computing power resource allocation data combined with the abnormal computing power node information; The calculation formula of the resource compensation data is: Among them, ΔBL is the resource compensation data, F zy Allocate computing resources to abnormal computing nodes, S yd is the abnormal computing power coefficient of the abnormal computing power node, D is the total number of abnormal computing power nodes, S yci is the computing power abnormality coefficient of the i-th abnormal computing power node; The resource compensation data is dispatched to the abnormal computing power node by using the reserved computing power resources to obtain the dispatch completion data; After the scheduling is completed, the abnormal computing power coefficient of the abnormal computing power node is recalculated to obtain the recalculation result until the abnormal computing power node is determined to be a normal computing power node.

Images