A computing resource allocation and control method for an open hardware platform in an industrial Internet environment

Abstract

The present invention discloses a method for allocating and controlling computing resources of an open hardware platform under an industrial Internet environment, and specifically relates to the technical field of computing resource allocation and control. The present invention obtains hardware component status data of an edge computing node through a node monitoring system, utilizes an ARIMA and LSTM hybrid model in combination with the task requirements of the edge computing node to accurately predict the maintenance cycle of the edge computing node, analyzes the dynamic expansion capability of the edge computing node to calculate the dynamic expansion response efficiency coefficient, analyzes the load scheduling situation to calculate the load scheduling bottleneck coefficient, calculates the load tolerance change index, analyzes the high load tolerance change capability of the edge computing node, provides an accurate basis for resource scheduling and maintenance cycle regulation of the edge computing node, effectively identifies potential bottlenecks, and optimizes and regulates the maintenance cycle of the edge computing node according to the classification results.

Description

Technical Field

[0001] The present invention relates to the field of computing resource allocation and control technology, and more specifically, to a computing resource allocation and control method for an open hardware platform in an industrial Internet environment. Background Art

[0002] With the rapid development of the Industrial Internet, more and more companies and manufacturers have begun to adopt edge computing technology to distribute data processing capabilities from traditional centralized cloud platforms to edge nodes close to data sources. This shift not only reduces network transmission latency and improves real-time response capabilities, but also effectively reduces the computing burden on the cloud. In the application scenarios of the Industrial Internet, edge computing nodes are usually responsible for processing real-time data streams from a large number of industrial devices, such as sensor data, machine status monitoring, and production process optimization. These edge nodes often have relatively limited computing resources and need to process a large number of real-time computing tasks. With the increase in the number of devices and changes in data processing requirements, the computing load bearing capacity of edge computing nodes has gradually exposed its limitations. When the computing power of edge nodes reaches a bottleneck, the system may face problems such as performance degradation, data loss, and response delays.

[0003] In order to deal with these problems, the maintenance of edge computing nodes is particularly important. Usually, maintenance work is triggered based on the real-time fault characterization of the node, such as abnormal temperature, hardware failure, excessive load, etc. However, traditional maintenance methods often only intervene after the node fails, and cannot predict the changes in the node's computing load bearing capacity in advance, nor can they reasonably regulate the maintenance cycle. Therefore, how to accurately capture the changes in the load bearing capacity of edge nodes and optimize the maintenance cycle to ensure the efficient allocation of computing resources and continuous and stable operation has become a difficult problem that needs to be solved in the industrial Internet system. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a method for allocating and controlling computing resources of an open hardware platform in an industrial Internet environment to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for allocating and controlling computing resources of an open hardware platform in an industrial Internet environment comprises the following steps:

[0007] Step S1, obtaining the hardware component status data of the edge computing node through the node monitoring system, and using the ARIMA and LSTM hybrid model combined with the task requirements of the edge computing node to predict the maintenance cycle of the edge computing node;

[0008] Step S2, building a high load tolerance change model according to the dynamic expansion and load scheduling of the edge computing node under high load, generating a load tolerance change index, and analyzing the high load tolerance change capability of the edge computing node;

[0009] Step S3, comparing the load tolerance variation index with a preset load tolerance variation index threshold, and classifying the high load tolerance variation capability of the edge computing node;

[0010] Step S4: Regulate and optimize the maintenance cycle of the predicted edge computing nodes according to the classification results.

[0011] In a preferred embodiment, the hardware component status data includes CPU usage percentage, remaining memory, disk read and write rate and response time, network bandwidth, and hardware temperature; the task requirements include task volume and task type.

[0012] In a preferred embodiment, the dynamic expansion of the edge computing node under high load is dynamically expanded to respond to the efficiency coefficient To quantify;

[0013] Load scheduling of edge computing nodes under high load conditions and load scheduling bottleneck coefficient To quantify.

[0014] In a preferred embodiment, the logic for obtaining the dynamic expansion response efficiency coefficient is as follows:

[0015] The load of the edge node is marked as , the load includes computing load and network load, according to the trigger function ,when , triggering the expansion mechanism, where Indicates the preset load threshold;

[0016] Get the time when the load of the edge node first reaches the load threshold And the time when the system actually detects that the load exceeds the limit , calculate the load detection delay , the expression is as follows ;

[0017] When the expansion mechanism is triggered, the time required for the edge computing node to request resources from the cloud for resource scheduling and allocation is obtained and the time to start new computing resources , calculate the extended response time , the expression is as follows ;

[0018] Get the newly added computing resources and the increase in load , computing resource growth ratio , the expression is as follows ;

[0019] Calculate the dynamic expansion response efficiency coefficient , the expression is as follows .

[0020] In a preferred implementation, the logic for obtaining the load scheduling bottleneck coefficient is as follows:

[0021] The load of the edge node is marked as , the load includes computing load and network load, according to the trigger function ,when , triggering the load scheduling mechanism, where Indicates the preset load threshold;

[0022] When the load scheduling mechanism is triggered, the computing resource value scheduled by the edge computing node is obtained, and each computing resource is standardized. The standardized formula is as follows ,in, represents the value of the i-th computing resource, represents the standardized value of the i-th computing resource, represents the maximum value of the i-th computing resource in historical scheduling, represents the minimum value of the i-th computing resource in the historical scheduling;

[0023] Computing and Scheduling Resource Fluctuation , the expression is as follows ,in, represents the specific value of the i-th computing resource at time t; Indicates the sampling time window The mean value of the i-th computing resource in , Indicates the sampling time window size, , is a positive integer;

[0024] Calculate the scheduling resource balance index , the expression is as follows ,in , is a positive integer;

[0025] Calculate the load scheduling bottleneck coefficient , the expression is as follows , Respectively , , The preset scaling factor of Both are greater than 0.

[0026] In a preferred embodiment, a high load tolerance variation model is constructed based on the dynamic expansion response efficiency coefficient and the load scheduling bottleneck coefficient to generate a load tolerance variation index. The model is based on the following formula , where represents the dynamic expansion response efficiency coefficient, represents the load scheduling bottleneck coefficient, They represent the preset proportional coefficients of the dynamic expansion response efficiency coefficient and the load scheduling bottleneck coefficient, respectively, and are greater than 0, is a minimum constant to avoid the denominator being 0.

[0027] In a preferred embodiment, the load tolerance variation index is compared with a preset load tolerance variation index threshold, and the high load tolerance variation capability of the edge computing node is classified as follows:

[0028] If the load tolerance index is greater than the load tolerance index threshold, the edge computing node's tolerance for changes under high load is marked as a low tolerance state. When the edge computing node is in a low tolerance state, the administrator needs to be notified immediately for maintenance.

[0029] If the load bearing variation index is less than or equal to the load bearing variation index threshold, the load bearing variation index generated by the subsequent high load bearing variation model of multiple monitoring points is obtained to establish a potential bottleneck risk set. ,in represents the load bearing variation index generated by the high load bearing variation model of the jth monitoring point, , is a positive integer, and the mean value of the load bearing variation index is calculated , the expression is as follows , calculate the standard deviation of the load bearing variation index , the expression is as follows , according to the maintenance cycle control trigger function, determine whether to control and optimize the maintenance cycle of the predicted edge computing node. The maintenance cycle control trigger function is as follows: ,in It represents the preset standard deviation threshold. 1 means that the edge computing node's ability to withstand changes under high load has a downward trend and the predicted maintenance cycle of the edge computing node needs to be regulated and optimized. 0 means that there is no need to regulate and optimize the predicted maintenance cycle of the edge computing node.

[0030] In a preferred embodiment, when the output value of the maintenance cycle control trigger function is 1, the predicted maintenance cycle of the edge computing node is controlled and optimized according to the control function, and the control function is specifically as follows: ,in To predict the maintenance period of edge computing nodes, is the maintenance cycle of the edge computing nodes after regulation. Represents a regulatory factor, which is used to control the degree of shortening of the maintenance cycle.

[0031] Technical effects and advantages of the present invention:

[0032] 1. The present invention obtains the hardware component status data of the edge computing node through the node monitoring system, uses the ARIMA and LSTM hybrid model combined with the task requirements of the edge computing node to accurately predict the maintenance cycle of the edge computing node, analyzes the dynamic expansion capability of the edge computing node to calculate the dynamic expansion response efficiency coefficient, analyzes the load scheduling situation to calculate the load scheduling bottleneck coefficient, constructs a high load tolerance change model, generates a load tolerance change index, analyzes the high load tolerance change capability of the edge computing node, and provides an accurate basis for the resource scheduling and maintenance cycle regulation of the edge computing node, avoiding performance bottlenecks or excessive resource consumption when the load is too high, flexibly responding to changes under different load states, effectively identifying potential bottlenecks, and optimizing the maintenance cycle of the edge computing node based on the high load tolerance change capability of the edge computing node and the classification results. Through intelligent adjustment of the maintenance cycle, the maintenance strategy can be quickly adjusted when high load or potential bottlenecks occur, the resource utilization of the edge computing node is improved, and the continuous and stable operation of the industrial Internet system in complex scenarios is guaranteed. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings;

[0034] Figure 1 The figure is a flow chart of a method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] Example: Figure 1 The present invention provides a method for allocating and controlling computing resources of an open hardware platform in an industrial Internet environment, comprising the following steps:

[0037] Step S1, obtaining the hardware component status data of the edge computing node through the node monitoring system, and using the ARIMA and LSTM hybrid model combined with the task requirements of the edge computing node to predict the maintenance cycle of the edge computing node;

[0038] Step S2, building a high load tolerance change model according to the dynamic expansion and load scheduling of the edge computing node under high load, generating a load tolerance change index, and analyzing the high load tolerance change capability of the edge computing node;

[0039] Step S3, comparing the load tolerance variation index with a preset load tolerance variation index threshold, and classifying the high load tolerance variation capability of the edge computing node;

[0040] Step S4, regulating and optimizing the maintenance cycle of the predicted edge computing node according to the classification result;

[0041] Step S1, obtaining the hardware component status data of the edge computing node through the node monitoring system, and using the ARIMA and LSTM hybrid model combined with the task requirements of the edge computing node to predict the maintenance cycle of the edge computing node;

[0042] The hardware component status data includes the CPU usage percentage, remaining memory, disk read and write rate and response time, network bandwidth, and hardware temperature; the task requirements include task volume and task type;

[0043] Data preprocessing is performed on the acquired hardware component status data and task requirements, including missing value processing (processing missing values ​​through interpolation, mean filling or deleting incomplete data records), data cleaning (removing outliers), and data standardization (common methods include Min-Max standardization or Z-Score standardization); the collected real-time data is converted into a time series format, and the data is arranged in chronological order; the autoregressive order and moving average order are preliminarily determined based on the tailing and truncation characteristics of the autocorrelation graph ACF and the partial autocorrelation graph PACF, and the time series data is differentially processed. The number of differentials starts from 1 and increases gradually until the time series data is stable. The ARIMA model is trained based on the selected autoregressive order, moving average order, and number of differentials. The optimal combination of autoregressive order, moving average order, and number of differentials is selected using information criteria (such as the AIC minimum information criterion and the BIC Bayesian information criterion), and the stable time series data is output based on the trained ARIMA model, and the trend and seasonal components are removed. The output of the ARIMA model is used as the input data of the LSTM model;

[0044] The LSTM model is a deep learning method based on neural networks, which is particularly suitable for time series prediction tasks and can capture long-term dependencies in data;

[0045] The LSTM model includes an input layer, an LSTM layer, and a fully connected layer. The input layer is used to receive time series data including hardware component status data and task requirement data. In the LSTM layer, the network learns long-term dependencies based on the input time series data to capture the relationship between the edge computing node hardware component status and task requirements and maintenance cycles. The output of the LSTM model is mapped to the predicted value of the maintenance cycle through the fully connected layer, and the maintenance cycle of the edge computing node is output. The LSTM model is trained using time series data, and the Adam optimizer is selected to update the weight parameters in the LSTM network. The maintenance cycle of the edge computing node is predicted based on the trained LSTM model.

[0046] Step S2, building a high load tolerance change model according to the dynamic expansion and load scheduling of the edge computing node under high load, generating a load tolerance change index, and analyzing the high load tolerance change capability of the edge computing node;

[0047] Dynamic expansion of edge computing nodes under high load conditions and dynamic expansion response efficiency coefficient To quantify;

[0048] Load scheduling of edge computing nodes under high load conditions and load scheduling bottleneck coefficient To quantify;

[0049] The dynamic expansion response efficiency coefficient is used to measure the dynamic expansion of edge computing nodes under high load conditions and to evaluate their responsiveness and efficiency under high load conditions. The larger the dynamic expansion response efficiency coefficient, the faster the system can respond under high load and effectively increase resources, thereby ensuring the efficient operation of the system. On the contrary, a lower dynamic expansion response efficiency coefficient indicates that the system expansion response is slow, which may lead to system performance degradation or bottlenecks. By calculating the dynamic expansion response efficiency coefficient, the stability, response speed and resource utilization efficiency of the system can be significantly improved. When the system faces load changes, the dynamic expansion response efficiency coefficient can quantitatively evaluate the efficiency of resource allocation, response delay and load increase during the expansion process, helping edge nodes to expand resources faster and more efficiently. This not only reduces the possible service interruption and performance degradation when the load is too high, but also ensures that resources are fully and reasonably utilized, avoiding excessive or insufficient resource allocation. By optimizing the dynamic expansion response efficiency coefficient, the system can respond to high-load environments more flexibly and quickly expand computing power, thereby ensuring that edge nodes can withstand load fluctuations and operate stably, improving the overall system performance and user experience. In addition, the dynamic expansion response efficiency coefficient can also promote the intelligent scheduling and preprocessing of resources, reduce energy consumption and costs, improve the system's adaptability, and provide strong support for future scale expansion and optimization of edge node maintenance cycle regulation;

[0050] The logic for obtaining the dynamic expansion response efficiency coefficient is as follows:

[0051] The load of the edge node is marked as , the load includes computing load and network load, according to the trigger function ,when , triggering the expansion mechanism, where Indicates the preset load threshold;

[0052] Get the time when the load of the edge node first reaches the load threshold And the time when the system actually detects that the load exceeds the limit , calculate the load detection delay , the expression is as follows ;

[0053] When the expansion mechanism is triggered, the time required for the edge computing node to request resources from the cloud for resource scheduling and allocation is obtained and the time to start new computing resources , calculate the extended response time , the expression is as follows ;

[0054] Get the newly added computing resources and the increase in load , computing resource growth ratio , the expression is as follows ;

[0055] Calculate the dynamic expansion response efficiency coefficient , the expression is as follows ;

[0056] It should be noted that before calculating the dynamic expansion response efficiency coefficient, it is necessary to ensure that the load detection delay, expansion response time, and resource growth ratio are all normalized. Common normalization methods include Min-Max normalization and Z-Score normalization.

[0057] The load scheduling bottleneck coefficient is used to measure the load scheduling of edge computing nodes under high load conditions to evaluate their load scheduling efficiency, adaptability and bottleneck identification capabilities under high load conditions. When the node is under high load, the load scheduling bottleneck coefficient can reflect how the system reasonably distributes the load to multiple cores or subsystems to avoid certain nodes or subsystems becoming performance bottlenecks, thereby reducing the risk of single-point overload. Specifically, when the load scheduling bottleneck coefficient is high, it means that the system load is unevenly distributed, and some cores may be overloaded due to limited task processing capabilities, resulting in longer system response time and reduced processing capabilities, which may eventually affect the overall computing performance. When the load scheduling bottleneck coefficient is low, it means that the load is more uniform, and the node can better adapt to high load fluctuations, avoiding system performance bottlenecks caused by local overload. Therefore, by calculating and analyzing the load scheduling bottleneck coefficient, the ability of edge computing nodes to withstand changes under high load can be intuitively evaluated. Further, combined with different load scheduling strategies (such as polling, minimum number of connections, weighted scheduling, etc.), the load scheduling bottleneck coefficient not only helps identify potential bottlenecks, but also provides data support for optimizing load distribution and improving scheduling strategies. For example, if the bottleneck coefficient of the system is too high under high load, it may be necessary to adjust the scheduling strategy and use a more appropriate weighted scheduling or dynamic load balancing algorithm to improve the carrying capacity and elasticity of edge nodes, ensure that tasks can still be processed stably when the load increases sharply, and avoid system crashes or response delays;

[0058] The logic for obtaining the load scheduling bottleneck coefficient is as follows:

[0059] The load of the edge node is marked as , the load includes computing load and network load, according to the trigger function ,when , triggering the load scheduling mechanism, where Indicates the preset load threshold;

[0060] When the load scheduling mechanism is triggered, the computing resource value scheduled by the edge computing node is obtained, and each computing resource is standardized. The standardized formula is as follows ,in, represents the value of the i-th computing resource, represents the standardized value of the i-th computing resource, represents the maximum value of the i-th computing resource in historical scheduling, Represents the minimum value of the i-th computing resource in the historical scheduling;

[0061] Computing and Scheduling Resource Fluctuation , the expression is as follows ,in, represents the specific value of the i-th computing resource at time t; Indicates the sampling time window The mean value of the i-th computing resource in , Indicates the sampling time window size, , is a positive integer;

[0062] It should be noted that the computing resources of the edge computing node can be the number of CPU cores, memory, bandwidth resources, etc., and its specific scheduling value is collected according to the actual situation;

[0063] Calculate the scheduling resource balance index , the expression is as follows ,in , is a positive integer;

[0064] Calculate the load scheduling bottleneck coefficient , the expression is as follows , Respectively , , The preset scaling factor of All are greater than 0;

[0065] It should be noted that Set it according to the actual situation. For example, adopt the expert empowerment method, that is, invite experts in relevant fields to determine the preset proportion coefficients of various indicators through professional opinion surveys and comprehensive evaluations;

[0066] Build a high load tolerance change model based on the dynamic expansion response efficiency coefficient and load scheduling bottleneck coefficient, and generate a load tolerance change index The model is based on the following formula , where represents the dynamic expansion response efficiency coefficient, represents the load scheduling bottleneck coefficient, They represent the preset proportional coefficients of the dynamic expansion response efficiency coefficient and the load scheduling bottleneck coefficient, respectively, and are greater than 0, is a minimum constant to avoid the denominator being 0;

[0067] It should be noted that before building a high load tolerance change model, it is necessary to ensure that the dynamic expansion response efficiency coefficient and the load scheduling bottleneck coefficient are normalized; Set it according to the actual situation. For example, adopt the expert empowerment method, that is, invite experts in relevant fields to determine the preset proportion coefficients of various indicators through professional opinion surveys and comprehensive evaluations;

[0068] From the above calculation expression, it can be seen that the smaller the dynamic expansion response efficiency coefficient and the larger the load scheduling bottleneck coefficient, the larger the load tolerance change index, which means that the edge computing node is prone to bottlenecks or cannot be effectively expanded under high load, and the carrying capacity is weak. On the contrary, the larger the dynamic expansion response efficiency coefficient and the smaller the load scheduling bottleneck coefficient, the smaller the load tolerance change index, which means that the edge computing node has a strong tolerance under high load and can effectively use the expansion mechanism and scheduling mechanism to cope with the load;

[0069] Step S3, compare the load tolerance change index with a preset load tolerance change index threshold, and classify the edge computing node's high load tolerance change capability, as follows:

[0070] If the load tolerance index is greater than the load tolerance index threshold, it means that the edge computing node has poor bottleneck performance under high load, which may cause system response time delay, computing task loss or node overload. In severe cases, it may affect the stability of the entire system. The edge computing node's tolerance for changes under high load is marked as low tolerance for changes. When the edge computing node is in the low tolerance for changes state, it is necessary to immediately notify the administrator for maintenance;

[0071] If the load tolerance variation index is less than or equal to the load tolerance variation index threshold, the edge computing node shows a certain adaptability under high load conditions, but there may still be potential bottleneck risks. Although the edge computing node can still maintain operation, it may show a trend of performance degradation when the load is high. Obtain the load tolerance variation index generated by the high load tolerance variation model of multiple subsequent monitoring points to establish a potential bottleneck risk set ,in represents the load bearing variation index generated by the high load bearing variation model of the jth monitoring point, , is a positive integer, and the mean value of the load bearing variation index is calculated , the expression is as follows , calculate the standard deviation of the load bearing variation index , the expression is as follows , according to the maintenance cycle control trigger function, determine whether to control and optimize the maintenance cycle of the predicted edge computing node. The maintenance cycle control trigger function is as follows: ,in Indicates the preset standard deviation threshold. 1 indicates that the edge computing node's ability to withstand changes under high load has a downward trend and the predicted maintenance cycle of the edge computing node needs to be regulated and optimized. 0 indicates that the predicted maintenance cycle of the edge computing node does not need to be regulated and optimized.

[0072] Step S4, regulating and optimizing the maintenance cycle of the predicted edge computing node according to the classification result;

[0073] When the output value of the maintenance cycle control trigger function is 1, the predicted maintenance cycle of the edge computing node is regulated and optimized according to the control function. The control function is as follows: ,in To predict the maintenance period of edge computing nodes, is the maintenance cycle of the edge computing nodes after regulation. It represents the regulating factor, which is used to control the shortening degree of the maintenance cycle;

[0074] The present invention obtains hardware component status data of edge computing nodes through a node monitoring system, uses an ARIMA and LSTM hybrid model in combination with the task requirements of the edge computing nodes to accurately predict the maintenance cycle of the edge computing nodes, analyzes the dynamic expansion capability of the edge computing nodes to calculate the dynamic expansion response efficiency coefficient, analyzes the load scheduling situation to calculate the load scheduling bottleneck coefficient, constructs a high-load bearing change model, generates a load bearing change index, analyzes the high-load bearing change capability of the edge computing nodes, provides an accurate basis for the resource scheduling and maintenance cycle regulation of the edge computing nodes, avoids performance bottlenecks or excessive resource consumption when the load is too high, flexibly responds to changes under different load states, effectively identifies potential bottlenecks, and optimizes and regulates the maintenance cycle of the edge computing nodes based on the high-load bearing change capability of the edge computing nodes and the classification results. Through intelligent adjustment of the maintenance cycle, the maintenance strategy can be quickly adjusted when high load or potential bottlenecks appear, the resource utilization of the edge computing nodes is improved, and the continuous and stable operation of the industrial Internet system in complex scenarios is guaranteed.

[0075] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0076] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0077] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for allocating and controlling computing resources on an open hardware platform in an industrial Internet environment, characterized in that: The steps include: Step S1, obtaining the hardware component status data of the edge computing node through the node monitoring system, and using the ARIMA and LSTM hybrid model combined with the task requirements of the edge computing node to predict the maintenance cycle of the edge computing node; Step S2, building a high load tolerance change model according to the dynamic expansion and load scheduling of the edge computing node under high load, generating a load tolerance change index, and analyzing the high load tolerance change capability of the edge computing node; Step S3, comparing the load tolerance variation index with a preset load tolerance variation index threshold, and classifying the high load tolerance variation capability of the edge computing node; Step S4, regulating and optimizing the maintenance cycle of the predicted edge computing node according to the classification result; Dynamic expansion of edge computing nodes under high load conditions and dynamic expansion response efficiency coefficient To quantify; Load scheduling of edge computing nodes under high load conditions and load scheduling bottleneck coefficient To quantify; Build a high load tolerance change model based on the dynamic expansion response efficiency coefficient and load scheduling bottleneck coefficient, and generate a load tolerance change index The model is based on the following formula , where represents the dynamic expansion response efficiency coefficient, represents the load scheduling bottleneck coefficient, They represent the preset proportional coefficients of the dynamic expansion response efficiency coefficient and the load scheduling bottleneck coefficient, respectively, and are greater than 0, is a minimum constant to avoid the denominator being 0; The load tolerance variation index is compared with the preset load tolerance variation index threshold, and the edge computing node's high load tolerance variation capability is classified as follows: If the load tolerance index is greater than the load tolerance index threshold, the edge computing node's tolerance for changes under high load is marked as a low tolerance state. When the edge computing node is in a low tolerance state, the administrator needs to be notified immediately for maintenance. If the load bearing variation index is less than or equal to the load bearing variation index threshold, the load bearing variation index generated by the subsequent high load bearing variation model of multiple monitoring points is obtained to establish a potential bottleneck risk set. ,in represents the load bearing variation index generated by the high load bearing variation model of the jth monitoring point, , is a positive integer, and the mean value of the load bearing variation index is calculated , the expression is as follows , calculate the standard deviation of the load bearing variation index , the expression is as follows , according to the maintenance cycle control trigger function, determine whether to control and optimize the maintenance cycle of the predicted edge computing node. The maintenance cycle control trigger function is as follows: ,in It represents the preset standard deviation threshold. 1 means that the edge computing node's ability to withstand changes under high load has a downward trend and the predicted maintenance cycle of the edge computing node needs to be regulated and optimized. 0 means that there is no need to regulate and optimize the predicted maintenance cycle of the edge computing node.

2. According to the method for allocating and controlling computing resources of an open hardware platform in an industrial Internet environment in claim 1, it is characterized by: The hardware component status data includes the CPU usage percentage, remaining memory, disk read and write rate and response time, network bandwidth, and hardware temperature; the task requirements include task volume and task type.

3. The method for allocating and controlling computing resources of an open hardware platform in an industrial Internet environment according to claim 1 is characterized in that: The logic for obtaining the dynamic expansion response efficiency coefficient is as follows: The load of the edge node is marked as , the load includes computing load and network load, according to the trigger function ,when , triggering the expansion mechanism, where Indicates the preset load threshold; Get the time when the load of the edge node first reaches the load threshold And the time when the system actually detects that the load exceeds the limit , calculate the load detection delay , the expression is as follows ; When the expansion mechanism is triggered, the time required for the edge computing node to request resources from the cloud for resource scheduling and allocation is obtained and the time to start new computing resources , calculate the extended response time , the expression is as follows ; Get the newly added computing resources and the increase in load , computing resource growth ratio , the expression is as follows ; Calculate the dynamic expansion response efficiency coefficient , the expression is as follows .

4. The method for allocating and controlling computing resources of an open hardware platform in an industrial Internet environment according to claim 1, characterized in that: The logic for obtaining the load scheduling bottleneck coefficient is as follows: The load of the edge node is marked as , the load includes computing load and network load, according to the trigger function ,when , triggering the load scheduling mechanism, where Indicates the preset load threshold; When the load scheduling mechanism is triggered, the computing resource value scheduled by the edge computing node is obtained, and each computing resource is standardized. The standardized formula is as follows ,in, represents the value of the i-th computing resource, represents the standardized value of the i-th computing resource, represents the maximum value of the i-th computing resource in historical scheduling, represents the minimum value of the i-th computing resource in the historical scheduling; Computing and Scheduling Resource Fluctuation , the expression is as follows ,in, represents the specific value of the i-th computing resource at time t; Indicates the sampling time window The mean value of the i-th computing resource in , Indicates the sampling time window size, , is a positive integer; Calculate the scheduling resource balance index , the expression is as follows ,in , is a positive integer; Calculate the load scheduling bottleneck coefficient , the expression is as follows , Respectively , , The preset scaling factor of Both are greater than 0.

5. The method for allocating and controlling computing resources of an open hardware platform in an industrial Internet environment according to claim 1, characterized in that: When the output value of the maintenance cycle control trigger function is 1, the predicted maintenance cycle of the edge computing node is regulated and optimized according to the control function. The control function is as follows: ,in To predict the maintenance period of edge computing nodes, is the maintenance cycle of the edge computing nodes after regulation. Represents a regulatory factor, which is used to control the degree of shortening of the maintenance cycle.

Images