Aggregation intelligent control method and system for user flexible resources

Abstract

The invention relates to the technical field of information, and discloses an aggregation intelligent control method and system for user flexible resources. The method comprises the following steps: acquiring equipment real-time data from a resource pool and processing the equipment real-time data to generate a resource deformation sequence; determining deformation trend description based on the sequence; k-means clustering is adopted to generate contour feature parameters; if the nonlinear fluctuation exists, using the LSTM network to generate a future deformation trend prediction value; comparing the current state data to adjust aggregation opportunity parameters; calculating a capacity distribution uniformity index in combination with the speed parameter; and if the index is lower than the threshold value, iteratively optimizing the speed parameter and generating an aggregation control instruction to execute adjustment. Through multi-step data processing and intelligent optimization, the problem that flexible resource aggregation lacks dynamic accurate control is solved, and the aggregation stability and efficiency are improved.

Description

Technical Field

[0001] This application relates to the field of information technology, and in particular to a method and system for aggregated intelligent control of user flexible resources. Background Technology

[0002] As digital businesses increase their demands for resource flexibility, users' flexible resources (such as equipment performance and capacity) need to be adjusted in real time according to load fluctuations to achieve efficient aggregation and stable operation. The core requirement in this field is to solve the problem of precise control under dynamic resource changes, ensure the balance of resource allocation and real-time response, and support the reliable execution of high-concurrency, multi-type tasks.

[0003] Existing technologies have significant shortcomings in the intelligent control of flexible user resource aggregation: First, traditional control strategies rely heavily on static parameter settings, which cannot adapt to the nonlinear fluctuations in the performance and capacity of resource pool devices. When faced with sudden loads or hardware aging, imbalances can easily occur, with some devices overloaded and others idle, leading to low resource utilization efficiency. Second, existing methods lack multi-dimensional data collaborative processing capabilities. They capture resource changes only through single time-series analysis, making it difficult to extract typical trend characteristics and providing insufficient accuracy in predicting future resource deformation trends, thus failing to provide a reliable basis for adjusting aggregation timing. Third, the resource scheduling process lacks a dynamic optimization mechanism and cannot update control parameters in real time based on the uniformity of capacity distribution. This often results in delayed aggregation commands or over-adjustments, especially in large-scale resource pool scenarios, which can easily cause system response delays and affect service quality. These problems severely restrict the stability and intelligence level of flexible user resource aggregation.

[0004] To address the aforementioned shortcomings, this application proposes to generate resource deformation sequences by real-time data acquisition, combine time series analysis and K-means clustering to extract features, use LSTM networks to predict trends, and then dynamically adjust parameters based on deviation and uniformity to generate aggregated control commands, thereby achieving precise and intelligent control of flexible resources. Summary of the Invention

[0005] This application proposes an intelligent control method and system for user flexible resources, aiming to solve the problems of lack of dynamic and precise control, insufficient prediction accuracy, and lack of scheduling optimization in the existing technology of flexible resource aggregation. Through multi-step data processing, intelligent analysis and dynamic parameter adjustment, it realizes stable and efficient control of flexible resource aggregation, and improves resource utilization efficiency and system service quality.

[0006] Firstly, this application provides a method for aggregated intelligent control of user flexible resources, the method comprising: Step S101: Obtain real-time data on equipment performance and capacity from the resource pool, perform preprocessing, time-series analysis and fluctuation feature extraction on the real-time data, and generate a resource deformation sequence; Step S102: Based on the resource deformation sequence, determine the deformation trend description through smoothing, time-series feature extraction, and trend fitting; Step S103: Use the K-means clustering algorithm to group the deformation trend descriptions, determine the cluster center points, extract the feature combination of the cluster center points into a multi-dimensional vector, and form contour feature parameters; Step S104: Determine whether there is nonlinear fluctuation in the contour feature parameters. If so, use an LSTM network to train and generate a predicted value for future deformation trends. Step S105: Obtain the current status data of the resource pool, calculate the deviation value between the predicted value of the future deformation trend and the current status data, and adjust the aggregation timing parameters based on the comparison result between the deviation value and the preset deviation threshold. Step S106: Based on the aggregation timing parameters and the initial setting of the speed parameters, obtain the current capacity distribution of the equipment, perform statistical analysis on the capacity distribution, and calculate the uniformity index. Step S107: Determine whether the uniformity index is lower than the preset uniformity threshold. If so, update the speed parameter through iterative optimization and generate an aggregation control command in combination with the aggregation timing parameter, and output it to the resource pool for dynamic adjustment.

[0007] Optionally, step S101 includes: By deploying a sensor network in the resource pool, the performance and capacity fluctuations of the devices in the resource pool are monitored, real-time data is collected, and noise values ​​are removed from the real-time data using a mean filtering method. The real-time data, after noise filtering, is normalized and then sorted by timestamp to generate time series data. Identify the scale of the resource pool scenario. If it is a regular-scale scenario, use an autoregressive moving average model to capture the autocorrelation and moving average components of the time series data, and extract the fluctuation amplitude and periodic features as fluctuation features. If it is a large-scale resource pool scenario, use the Fourier transform method to convert the time series data from the time domain to the frequency domain, and extract the dominant frequency as fluctuation features. The fluctuation characteristics are combined into a multi-dimensional vector sequence to generate a resource deformation sequence that characterizes the dynamic change trend of equipment performance and capacity.

[0008] Optionally, step S102 includes: The resource deformation sequence is smoothed by applying a moving average method to generate a smoothed sequence. The smoothed sequence is subjected to feature extraction using time series analysis methods. An autoregressive moving average model is used to capture time series dependencies, and trend slope and periodic fluctuations are extracted as trend features. The slope and fluctuation pattern in the trend features are input into a linear regression model to fit a long-term trend line and generate a numerical index that represents the long-term direction of change. Then, based on the numerical index, a deformation trend description that represents the long-term direction of change of the resource deformation sequence is constructed.

[0009] Optionally, step S103 includes: The K-means clustering algorithm was used to group the deformation trend descriptions, generating multiple stable clusters. Analyze multiple stable clusters, extract the average value of all data points in each cluster, and use the average value as the center point of the corresponding cluster. The center point represents the concentrated characteristics of the deformation trend within the cluster. Calculate the fluctuation amplitude value of the deformation trend description corresponding to the center point of each cluster, where the fluctuation amplitude value is the difference between the trend peak and the trough value; The average number of fluctuations in the deformation trend description corresponding to each cluster center point per unit time is used as the fluctuation frequency value. The deformation trend at each cluster center point is measured to describe the average time from the start to the stabilization point, which is taken as the duration of the fluctuation. The fluctuation amplitude value, fluctuation frequency value, and fluctuation duration are combined into a multi-dimensional vector to generate the contour feature parameters, wherein the contour feature parameters are used to characterize typical features of deformation trends.

[0010] Optionally, step S104 includes: Calculate the second difference of the contour feature parameter sequence. If the absolute value of the second difference exceeds a preset fluctuation threshold for three or more consecutive time points, the contour feature parameter is determined to be nonlinear fluctuation. The contour feature parameters that are determined to be nonlinear fluctuations within a historical time period are used as feature inputs. An LSTM network is used to train the network, and the number of hidden layer units and the number of training iterations are set. The mean squared error is used as the loss function for backpropagation optimization. Based on the trained LSTM network, the current contour feature parameters are input to generate a prediction sequence for multiple future time points, thereby obtaining the predicted value of the future deformation trend, wherein the predicted value of the future deformation trend characterizes the future performance and capacity change trend of the resource pool.

[0011] Optionally, step S105 includes: Obtain the current status data of the resource pool, wherein the current status data is the real-time performance and capacity data of the devices in the resource pool; Based on the predicted future deformation trend, calculate the performance and capacity data of the resource pool at future points in time. Calculate the deviation between the performance and capacity data at the future time point and the current state data; Determine whether the deviation value is greater than a preset deviation threshold. If so, the aggregation timing parameter needs to be adjusted to maintain the stability and balance of the resource pool. If not, the aggregation timing parameter does not need to be adjusted.

[0012] Optionally, adjusting the aggregation timing parameters includes: The deviation adjustment coefficient is calculated based on the deviation value. The deviation adjustment coefficient is obtained by first calculating the ratio of the absolute value of the deviation to the preset deviation threshold, and then comparing the ratio with the natural number 1, taking the minimum value between the two. The new aggregation timing parameter is calculated based on the deviation adjustment coefficient. The new aggregation timing parameter = current aggregation timing parameter × (1 - adjustment coefficient × deviation adjustment coefficient), wherein the adjustment coefficient is preset to be within the range of 0 to 1.

[0013] Optionally, step S106 includes: Based on the aggregation timing parameters and the initial setting of the speed parameters, the current capacity distribution of devices in the resource pool is obtained; Statistical analysis is performed on the current capacity distribution to calculate the mean, standard deviation, and maximum and minimum difference, generating distribution characteristic data; The uniformity index characterizing the balance of capacity distribution is calculated based on the distribution characteristic data. The uniformity index is obtained by subtracting the standard deviation from the mean value from the natural number 1.

[0014] Optionally, step S107 includes: Determine whether the uniformity index is lower than the preset uniformity threshold. If it is lower, update the speed parameter value through a step-size adaptive iterative optimization method to reduce the gap between the uniformity index and the preset uniformity threshold. Based on the updated speed parameters and aggregation timing parameters, an aggregation control command containing the values ​​of both is generated; The aggregation control command is output to the resource pool, instructing the resource pool to perform resource aggregation operations according to the rate defined by the speed parameter value and the timing defined by the aggregation timing parameter, so as to achieve stable aggregation.

[0015] Secondly, this application provides an aggregated intelligent control system for user flexible resources, the aggregated intelligent control system for user flexible resources comprising: The data acquisition module is used to obtain real-time data on equipment performance and capacity from the resource pool, perform preprocessing, time series analysis and fluctuation feature extraction, and generate resource deformation sequences. The trend extraction module is used to smooth the resource deformation sequence, extract time-series features and fit trends to generate a deformation trend description that represents the long-term direction of change. The feature clustering module is used to generate stable clusters based on deformation trend description using the K-means clustering algorithm, extract center point features, and construct multi-dimensional contour feature parameters; The trend prediction module is used to train an LSTM network and generate future deformation trend prediction values ​​when there are non-linear fluctuations in the contour feature parameters. The deviation calculation module is used to compare the predicted value of future deformation trend with the current state data, calculate the deviation value, and adjust the aggregation timing parameters accordingly. The uniformity calculation module is used to statistically analyze the current capacity distribution and calculate the uniformity index based on aggregation timing parameters and speed parameters. The instruction output module is used to generate and output aggregation control instructions by iteratively optimizing and updating the speed parameters when the uniformity index is lower than the preset uniformity threshold, combined with the aggregation timing parameters.

[0016] This application proposes an aggregated intelligent control method and system for user flexible resources, applicable to the aggregated control of user flexible resources (such as equipment performance and capacity) in digital business scenarios. It can solve the problems of lack of dynamic and precise control, insufficient prediction accuracy, and lack of scheduling optimization in the existing technology for flexible resource aggregation. Compared with the existing technology, the beneficial effects of the technical solution of this application are at least as follows: First, it can adapt to the non-linear fluctuations in the performance and capacity of resource pool devices, avoiding the imbalance problem of some devices being overloaded and others being idle when facing sudden loads or hardware aging, thus improving resource utilization efficiency.

[0017] Secondly, it possesses multi-dimensional data collaborative processing capabilities, which can effectively extract typical characteristics of resource change trends, improve the accuracy of predicting future resource deformation trends, and provide a reliable basis for adjusting aggregation timing.

[0018] Third, a dynamic optimization mechanism is established in the resource scheduling process. Control parameters can be updated in real time based on the uniformity of capacity distribution to avoid lag or over-adjustment of aggregation instructions. Especially in large-scale resource pool scenarios, this can reduce system response latency and ensure service quality. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating a method for aggregated intelligent control of user flexible resources in this application; Figure 2 This is a schematic diagram of an aggregated intelligent control process for user flexible resources in this application; Figure 3 This is a schematic diagram of the structure of an intelligent control system for user flexible resources in this application. Detailed Implementation

[0021] This application provides a method and system for aggregated intelligent control of user flexible resources. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0022] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the aggregation and intelligent control method for user flexible resources in this application includes: Step S101: Obtain real-time data on device performance and capacity from the resource pool, perform preprocessing, time-series analysis and fluctuation feature extraction on the real-time data, and generate a resource deformation sequence.

[0023] In one specific embodiment, the process of performing step S101 may specifically include the following steps: By deploying a sensor network in the resource pool, the performance and capacity fluctuations of the devices in the resource pool are monitored, real-time data is collected, and noise values ​​are removed from the real-time data using a mean filtering method. The real-time data, after noise filtering, is normalized and then sorted by timestamp to generate time series data. Identify the scale of the resource pool scenario. If it is a regular-scale scenario, use an autoregressive moving average model to capture the autocorrelation and moving average components of the time series data, and extract the fluctuation amplitude and periodic features as fluctuation features. If it is a large-scale resource pool scenario, use the Fourier transform method to convert the time series data from the time domain to the frequency domain, and extract the dominant frequency as fluctuation features. The fluctuation characteristics are combined into a multi-dimensional vector sequence to generate a resource deformation sequence that characterizes the dynamic change trend of equipment performance and capacity.

[0024] Specifically, in a scenario of dynamic resource pool adjustment, a sensor network deployed within the resource pool continuously monitors fluctuations in device performance (such as CPU utilization and memory usage) and capacity (such as remaining storage space and bandwidth usage). It acquires real-time data at a preset collection frequency (e.g., once per second). For example, in a server cluster, sensors collect data on each server's CPU utilization (range 0%–100%), memory usage (range 0%–100%), and remaining storage space (range 0GB–1000GB) every second. For the collected real-time data, a median filtering method is applied to remove noise. Specifically, a window is formed by taking a preset number (e.g., 5) of adjacent data points before and after a given data point, and the median value within the window replaces that data point. For example, if a server's CPU utilization values ​​for five consecutive data points are 25%, 28%, 150%, 26%, and 27%, the 150% value represents noise caused by network interference. After median filtering, this data point is replaced with 27%. This operation solves the problem of data distortion caused by noise interference in real-time data, ensuring the reliability of subsequent data processing.

[0025] The real-time data after noise filtering is normalized using the min-max normalization formula. (in x This is the original data. This is the minimum value of the indicator. This is the maximum value of the indicator. (This is for normalized data), unifying the performance and capacity indicators of different devices to a range of 0-1. For example, the original CPU utilization data range is 0% to 100%, and a server's CPU utilization is 50%, which is normalized to 0.5; the original storage space data range is 0GB to 1000GB, and a server's storage space is 500GB, which is normalized to 0.5. After normalization, the data is sorted according to the timestamp of data collection to generate time-series data, such as data arranged by timestamp. (in , The timestamp is used as the data source, and the subsequent values ​​are the normalized CPU utilization, memory usage, and remaining storage space, respectively. This process solves the problem that different devices have large differences in the range of values ​​of different indicators, making it impossible to perform collaborative analysis directly, and provides a standardized data format for subsequent time series analysis.

[0026] When processing time series data using time series analysis methods, it is necessary to first identify the scale of the resource pool scenario, and then select the corresponding method based on the scale of the resource pool. In typical scenarios (such as resource pools with fewer than 50 devices), an autoregressive moving average model is used. ,in p Let the order be the autoregressive order. q The model processes data (for the moving average order), and uses the formula... (in for t Time series data at any given moment These are the autoregressive coefficients. The moving average coefficient is... To capture the autocorrelation and moving average components of time series data (using white noise), after model fitting, the fluctuation amplitude (such as the standard deviation of the model residuals) and periodic characteristics (such as the interval of peaks in the autocorrelation function graph) of the data are extracted. For example, after processing the time series data of a regular resource pool with the ARMA(2,1) model, the fluctuation amplitude is 0.05, and the periodic characteristic is that a peak occurs every 30 minutes. In large-scale resource pool scenarios (such as resource pools with more than or equal to 50 devices), the Fourier transform method is used. (in For time-domain time series data, For frequency domain data, For frequency, t This process converts time-domain time-series data into frequency-domain data. By analyzing the power spectral density of the frequency-domain data, the dominant frequency (i.e., the frequency corresponding to the maximum power spectral density) is extracted as a fluctuation feature. For example, after Fourier transform, the dominant frequency of a large-scale resource pool is 0.0005Hz (corresponding to a period of approximately 33 minutes). This process solves the problem that the data characteristics of resource pools of different sizes are very different, and a single analysis method cannot effectively extract fluctuation features, ensuring that key information on data fluctuations can be accurately obtained in different scenarios.

[0027] Based on the extracted fluctuation features, the fluctuation features at the same time point are combined into a multi-dimensional vector. For example, in a typical scenario, if the fluctuation amplitude at a certain time point is 0.05 and the periodic feature corresponds to a period of 30 minutes (converted to a value of 0.00056Hz), then the multi-dimensional vector at that time point would be: In large-scale scenarios, if the dominant frequency at a certain point in time is 0.0005Hz, then the multidimensional vector is (0.0005), which is the resource deformation sequence that characterizes the dynamic change trend of equipment performance and capacity. This process solves the problem that the fluctuation characteristic sequence has short-term fluctuations and cannot clearly reflect the long-term dynamic change trend, providing a smooth and continuous data foundation for subsequent determination of deformation trend description.

[0028] Step S102: Based on the resource deformation sequence, determine the deformation trend description through smoothing, temporal feature extraction and trend fitting.

[0029] In one specific embodiment, the process of performing step S102 may specifically include the following steps: The resource deformation sequence is smoothed by applying a moving average method to generate a smoothed sequence. The smoothed sequence is subjected to feature extraction using time series analysis methods. An autoregressive moving average model is used to capture time series dependencies, and trend slope and periodic fluctuations are extracted as trend features. The slope and fluctuation pattern in the trend features are input into a linear regression model to fit a long-term trend line and generate a numerical index that represents the long-term direction of change. Then, based on the numerical index, a deformation trend description that represents the long-term direction of change of the resource deformation sequence is constructed.

[0030] Specifically, for the generated resource deformation sequences (such as a multi-dimensional vector sequence of a server cluster that occurs every 5 minutes and includes normalized fluctuations in CPU utilization and memory usage; example data is...), , , , , The moving average method is used for smoothing. The moving average method employs a sliding window mechanism, with the window size set according to the sampling frequency of the resource deformation sequence. If the sampling frequency is 5 minutes / time, the window size is set to 3, meaning the arithmetic mean of the resource deformation sequence data at three consecutive time points is calculated. The formula is as follows: (in for Moving average over time, They are respectively (Resource deformation sequence data at any given time). Calculated using example data. The moving average over time is , Time is This process generates a smooth sequence in the same way. This treatment solves the problem of short-term random noise (such as abnormal jumps in CPU utilization at a certain moment caused by instantaneous network fluctuations) interfering with trend analysis in resource deformation sequences, making the sequence more closely reflect the actual changes in equipment performance and capacity.

[0031] After smoothing, features are extracted from the smoothed sequence using time series analysis methods, with an autoregressive moving average model selected. Capture temporal dependencies. Model order parameter. p (Autoregressive order) and q(Moving average order) is adjusted based on the length of the smoothed series. If the smoothed series contains 100 time points, different moving average orders are calculated using the AIC information criterion. Choose the combination with the lowest AIC value. p =2、 q =1, the model formula is (in for Time-smoothed sequence data, These are the autoregressive coefficients. The moving average coefficient is... for (Time-based white noise). Smooth sequence data is input into the model for training, and coefficients are estimated using the least squares method, as shown in the following figure. Based on the trained model, trend slope and periodic fluctuations are extracted as trend features. The trend slope is calculated using the first derivative of the fitted curve. If the first derivative of the fitted curve is 0.002 / 5 minutes over a certain period, it indicates that the smoothed sequence data increases by an average of 0.002 every 5 minutes. Periodic fluctuations are analyzed using the autocorrelation function (ACF) of the model residuals. If the ACF peaks at a lag of 12 time points (corresponding to 60 minutes), it indicates the existence of 60-minute periodic fluctuations. This process solves the problem that a single data point cannot reflect temporal correlation and is difficult to extract long-term change features, providing key feature support for subsequent trend fitting.

[0032] Based on the extracted trend features, the trend slope and fluctuation pattern are input into a linear regression model for trend fitting. The linear regression model uses timestamps as independent variables. t To smooth sequence data The dependent variable is [variable name], and the model formula is [formula]. (in k These are regression coefficients, corresponding to the average level of the trend slope. b (This is the intercept). The trend slope (e.g., 0.002 / 5 minutes) in the trend characteristics is converted into a unit time slope (0.0004 / minute). This slope, combined with the time node data corresponding to the fluctuation pattern (e.g., peak data every 60 minutes), is input into the model, and the least squares method is used to solve for the slope. k and b If calculated k =0.0004、 b =0.4, then the fitted long-term trend line is Based on this trend line, numerical indicators representing the long-term direction of change are generated, such as predicted values ​​for the next 120 minutes. , and the current value (e.g. t time Comparing this with (=0.43), the numerical indicator obtained is "0.024 per hour". Based on this numerical indicator, a deformation trend description is constructed, such as "the smooth sequence of CPU utilization and memory occupancy of resource pool devices shows a stable growth trend of 0.024 per hour, accompanied by periodic fluctuations over 60 minutes". This description directly represents the long-term change direction of the resource deformation sequence, solving the problem of lacking a clear trend description and being unable to provide directional guidance for subsequent resource scheduling, and providing a clear trend input for the feature clustering module.

[0033] Step S103: Use the K-means clustering algorithm to group the deformation trend descriptions, determine the cluster center points, extract the feature combination of the cluster center points into a multi-dimensional vector, and form contour feature parameters.

[0034] In one specific embodiment, the process of performing step S103 may specifically include the following steps: The K-means clustering algorithm was used to group the deformation trend descriptions, generating multiple stable clusters. Analyze multiple stable clusters, extract the average value of all data points in each cluster, and use the average value as the center point of the corresponding cluster. The center point represents the concentrated characteristics of the deformation trend within the cluster. Calculate the fluctuation amplitude value of the deformation trend description corresponding to the center point of each cluster, where the fluctuation amplitude value is the difference between the trend peak and the trough value; The average number of fluctuations in the deformation trend description corresponding to each cluster center point per unit time is used as the fluctuation frequency value. The deformation trend at each cluster center point is measured to describe the average time from the start to the stabilization point, which is taken as the duration of the fluctuation. The fluctuation amplitude value, fluctuation frequency value, and fluctuation duration are combined into a multi-dimensional vector to generate the contour feature parameters, wherein the contour feature parameters are used to characterize typical features of deformation trends.

[0035] Specifically, the deformation trend description is presented in the form of multi-dimensional data. For example, the deformation trend description of different device clusters in a cloud computing resource pool includes the rate of change per unit time and the duration of periodic fluctuations. This data constitutes the input dataset, and each data point corresponds to a vector composed of the above dimensions. When applying the K-means clustering algorithm, the number of clusters is first preset according to the data dimensions of the deformation trend description and the scale of the resource pool devices. Then, a corresponding number of data points are randomly selected from the input dataset as initial centroids. Subsequently, the Euclidean distance from each data point in the input dataset to each initial centroid is calculated iteratively, and each data point is assigned to the cluster containing the nearest centroid based on the distance. After all data points are assigned, the average value of each dimension of all data points in each cluster is calculated, and the cluster centroid is updated accordingly. The process of distance calculation, data point assignment, and centroid update is repeated until the difference between the centroid coordinates of two iterations is less than a preset threshold, generating multiple stable clusters. This process solves the problems of lack of multi-dimensional data collaborative processing capabilities and difficulty in extracting typical trend features in existing technologies. By clustering, similar deformation trends are classified, laying the foundation for subsequent feature extraction.

[0036] After generating stable clusters, each cluster is analyzed, and the average value of each dimension of all data points within the cluster is extracted. This average value is used as the center point of the corresponding cluster. The values ​​of each dimension of the center point directly represent the concentrated characteristics of the deformation trend within the cluster. For example, if the unit time change rate of a cluster center point is 0.023 and the periodic fluctuation duration is 59 minutes, it means that the overall deformation trend of the equipment in the cluster exhibits a change rate of 0.023 per unit time and a periodic fluctuation every 59 minutes. Based on the deformation trend description corresponding to the center point, the fluctuation amplitude value is calculated. Specifically, the trend peak and trough values ​​are extracted from the deformation trend data corresponding to the center point, and the difference between the two is the fluctuation amplitude value. For example, if the peak value of the deformation trend corresponding to a center point is 0.08 and the trough value is 0.03, the fluctuation amplitude value is 0.05. When calculating the fluctuation frequency value, the unit time is used as the statistical period. The number of fluctuations of the deformation trend description corresponding to the center point within a unit time is counted and the average value is taken. If the unit time is 60 minutes, and the trend fluctuates 12 times within 60 minutes, the fluctuation frequency value is 12 times / hour. When measuring the duration of fluctuation, record the time from the start of the deformation trend at the center point to the time it takes to enter a steady state. After multiple measurements, take the average value as the duration of fluctuation. For example, if the multiple measurement results are 175 minutes, 180 minutes, and 185 minutes, the duration of fluctuation is 180 minutes.

[0037] The fluctuation amplitude, fluctuation frequency, and fluctuation duration are combined in a fixed order to form a multidimensional vector, which is the contour feature parameter. For example, if a cluster has a fluctuation amplitude of 0.05, a fluctuation frequency of 12 times / hour, and a fluctuation duration of 180 minutes, the resulting multidimensional vector (0.05, 12, 180) is the contour feature parameter of the cluster's deformation trend. This contour feature parameter integrates key typical features of the deformation trend, solving the problem of insufficient accuracy in predicting future resource deformation trends by relying solely on single time-series analysis to capture resource changes. It provides standardized feature inputs for subsequent judgment of nonlinear fluctuations and generation of future deformation trend predictions, enabling subsequent prediction processes to be based on more comprehensive and typical trend features, improving prediction accuracy, and thus providing a reliable basis for adjusting aggregation timing. This avoids the imbalance problem that occurs when traditional static control strategies face nonlinear resource fluctuations.

[0038] Step S104: Determine whether the contour feature parameters have nonlinear fluctuations. If so, use an LSTM network to train and generate a predicted value for future deformation trends.

[0039] In one specific embodiment, the process of executing step S104 may specifically include the following steps: Calculate the second difference of the contour feature parameter sequence. If the absolute value of the second difference exceeds a preset fluctuation threshold for three or more consecutive time points, the contour feature parameter is determined to be nonlinear fluctuation. The contour feature parameters that are determined to be nonlinear fluctuations within a historical time period are used as feature inputs. An LSTM network is used to train the network, and the number of hidden layer units and the number of training iterations are set. The mean squared error is used as the loss function for backpropagation optimization. Based on the trained LSTM network, the current contour feature parameters are input to generate a prediction sequence for multiple future time points, thereby obtaining the predicted value of the future deformation trend, wherein the predicted value of the future deformation trend characterizes the future performance and capacity change trend of the resource pool.

[0040] Specifically, the contour feature parameters include fluctuation amplitude, fluctuation frequency, and fluctuation duration. These parameters are arranged in timestamp order to form a contour feature parameter sequence. For example, a resource pool collects data every 10 minutes, forming a sequence containing multiple time points (e.g., ...). The parameter sequence of the sequence, where each time point corresponds to a vector consisting of fluctuation amplitude, fluctuation frequency, and fluctuation duration. To determine whether the sequence exhibits nonlinear fluctuations, the second-order difference needs to be calculated for each dimension. Taking the fluctuation amplitude dimension as an example, the first-order difference between the fluctuation amplitude values ​​of two adjacent time points is first calculated, i.e. (in to (These represent the fluctuation amplitude values ​​at different time points), and then the second difference is calculated based on the first-order difference, i.e. The second-order difference is calculated using the same logic for both the fluctuation frequency and fluctuation duration dimensions. Preset fluctuation thresholds are set for each dimension, such as a second-order difference threshold of 0.02 for fluctuation amplitude, 1 time / hour for fluctuation frequency, and 10 minutes for fluctuation duration. If the absolute value of the second-order difference for any dimension exceeds the corresponding threshold for three or more consecutive time points, the profile feature parameter is determined to exhibit nonlinear fluctuations. This process solves the problem that traditional control strategies in existing technologies cannot adapt to the nonlinear fluctuations in the performance and capacity of resource pool equipment. Through multi-dimensional second-order difference analysis, it accurately identifies the nonlinear characteristics of resource changes, avoiding misjudgments of fluctuations due to single-dimensional or first-order difference analysis, and providing accurate triggering conditions for subsequent predictions.

[0041] If nonlinear fluctuations are identified, multiple sets of contour feature parameters identified as nonlinear fluctuations within a historical time period are used as feature inputs to the LSTM network. The time step size of the input sequence is consistent with the parameter acquisition period; for example, if the acquisition period is 10 minutes, the time step size is set to 10. Each time step corresponds to three feature dimensions (fluctuation amplitude, fluctuation frequency, and fluctuation duration). The number of hidden layer units in the LSTM network is set according to the data scale; if the historical data contains 1000 time steps, the number of hidden layer units is set to 64, and the number of training iterations is set to 100. When constructing training samples, the contour feature parameter sequence of N consecutive time steps is used as the input sample, and the contour feature parameter of the (N+1)th time step is used as the label, for example, using... The parameter sequence is the input. The parameter is the label; The parameter sequence is the input. The parameters are the labels, which are used to generate batch training samples. The mean squared error is used as the loss function, calculated as follows: (in k For the sample size, The actual value of the label. The network weights and biases are adjusted using the backpropagation algorithm to gradually converge the loss function value to a preset range (e.g., less than 0.001). This training process addresses the problem of insufficient accuracy in predicting future resource deformation trends in existing methods. By leveraging the ability of LSTM networks to capture long-sequence dependencies, the network fully learns the changing patterns of contour feature parameters under nonlinear fluctuations, providing reliable model support for subsequent predictions.

[0042] After training, the contour feature parameters of the current time step and previous consecutive time steps are input into the LSTM network. For example, the current time step is... ,enter The network processes the input sequence through its internal memory units, taking a parameter sequence of 10 time steps, and outputs multiple future time points (e.g., ...). The prediction sequence is generated, with each prediction time point corresponding to a set of prediction vectors containing fluctuation amplitude, fluctuation frequency, and fluctuation duration. These prediction vectors together constitute the predicted value of future deformation trend. Since each dimension of the contour feature parameters is directly related to changes in resource pool performance and capacity—for example, fluctuation amplitude corresponds to the performance fluctuation range, fluctuation frequency corresponds to the capacity fluctuation frequency, and fluctuation duration corresponds to the duration of resource instability—the predicted value of future deformation trend can directly characterize the future performance and capacity change trends of the resource pool. This prediction process provides a reliable basis for adjusting aggregation timing, avoiding the problem of delayed aggregation instructions due to a lack of accurate trend references in existing technologies. It ensures that resource aggregation can adapt to future resource changes in advance, maintaining the stable operation of the resource pool.

[0043] Step S105: Obtain the current status data of the resource pool, calculate the deviation value between the predicted value of the future deformation trend and the current status data, and adjust the aggregation timing parameters based on the comparison result between the deviation value and the preset deviation threshold.

[0044] In one specific embodiment, the process of executing step S105 may specifically include the following steps: Obtain the current status data of the resource pool, wherein the current status data is the real-time performance and capacity data of the devices in the resource pool; Based on the predicted future deformation trend, calculate the performance and capacity data of the resource pool at future points in time. Calculate the deviation between the performance and capacity data at the future time point and the current state data; Determine whether the deviation value is greater than a preset deviation threshold. If so, the aggregation timing parameter needs to be adjusted to maintain the stability and balance of the resource pool. If not, the aggregation timing parameter does not need to be adjusted.

[0045] Specifically, obtaining current status data of the resource pool relies on a sensor network deployed within the resource pool to continuously collect real-time performance and capacity data from the devices. Performance data includes CPU utilization and memory usage for each device, while capacity data covers remaining storage space and bandwidth usage. The collection frequency is adapted to the dynamic change rate of the resource pool, for example, once per second, to ensure that the data reflects the current operating status of the devices in real time. When calculating future performance and capacity data based on predicted future deformation trends, a mapping relationship between the predicted values ​​and actual business indicators needs to be established. The predicted future deformation trends include a multi-dimensional vector containing fluctuation amplitude, fluctuation frequency, and fluctuation duration. The fluctuation amplitude corresponds to the range of performance and capacity changes, the fluctuation frequency corresponds to the number of changes per unit time, and the fluctuation duration corresponds to the time from the start of the change to stabilization. Taking CPU utilization as an example, if the current CPU utilization is 50%, and the fluctuation range of the predicted future deformation trend at a certain point in time is 0.05 (after normalization), combined with the original CPU utilization data range of 0% to 100%, we can deduce that the predicted CPU utilization at that point in time is 50% + 0.05 × 100% = 55%. The current value of the remaining storage space is 500GB, and the predicted fluctuation range is 0.1 (after normalization, the original range is 0GB-1000GB). Then, the predicted value of the remaining storage space at that point in time is 500GB + 0.1 × 1000GB = 600GB. Through similar logic, we can calculate the specific data of all performance and capacity indicators at various future points in time.

[0046] When calculating the deviation between future time point data and current state data, it is necessary to calculate separately for each type of performance and capacity indicator, and then obtain the comprehensive deviation value through weighted summation or Euclidean distance. Taking CPU utilization and remaining storage space as two core indicators as examples, let the current value of CPU utilization be A1 and the future predicted value be B1, and the current value of remaining storage space be A2 and the future predicted value be B2. First, calculate the deviation of each indicator: the deviation of CPU utilization is |B1-A1|, and the deviation of remaining storage space is |B2-A2|. If the weights of the two indicators are ω1 and ω2 respectively (set according to the importance of business, such as ω1=0.6, ω2=0.4), then the comprehensive deviation value is ω1×|B1-A1|+ω2×|B2-A2|. For example, if A1=50%, B1=55%, A2=500GB, B2=600GB, substituting these values, the CPU utilization deviation is 5%, the remaining storage space deviation is 100GB, and the overall deviation is 0.6×5%+0.4×(100GB / 1000GB)=3%+4%=7% (the remaining storage space deviation is normalized first).

[0047] When determining the relationship between the deviation value and the preset deviation threshold, the preset deviation threshold needs to be determined in conjunction with the historical stable operation data of the resource pool and the service quality requirements of the business. For example, according to historical data statistics, when the comprehensive deviation value exceeds 8%, the resource pool is prone to equipment overload or idleness issues, so the preset deviation threshold is set to 8%. If the calculated comprehensive deviation value is 7%, which is less than 8%, it means that future resource changes are within the adaptability range of the current aggregation strategy, and there is no need to adjust the aggregation timing parameters; if the comprehensive deviation value is 9%, which is greater than 8%, then the aggregation timing parameters need to be adjusted. The adjustment process needs to dynamically adjust the aggregation interval based on the ratio of the deviation value to the threshold. The aggregation timing parameters are usually expressed as the aggregation interval time. Let the current aggregation interval be T, the deviation value be D, the preset threshold be D0, and the adjustment coefficient be k (0 < T). k <1, such as k =0.2), then the new aggregation interval For example, if the current aggregation interval T=10 minutes, D=9%, and D0=8%, substituting these values ​​into the calculation, we get T'=10×(1-0.2×(9%-8%) / 8%)=10×(1-0.0025)=9.975 minutes. By shortening the aggregation interval, resource aggregation operations can be performed more frequently, allowing for proactive responses to future resource changes.

[0048] By acquiring current status data in real time and comparing it with future predictions, the aggregation timing parameters are dynamically adjusted to avoid resource scheduling delays or over-adjustments caused by fixed aggregation timing. Simultaneously, a mapping relationship between predicted values ​​and actual business indicators is established, along with a multi-indicator comprehensive deviation calculation logic. This overcomes the limitations of single-dimensional analysis in existing methods, ensuring more comprehensive and accurate deviation judgments and providing a reliable basis for adjusting aggregation timing. Ultimately, this improves the operational stability and resource utilization efficiency of the resource pool, alleviating the imbalance problem of some devices being overloaded and others idle.

[0049] In one specific embodiment, the process of adjusting the polymerization timing parameters may specifically include the following steps: The deviation adjustment coefficient is calculated based on the deviation value. The deviation adjustment coefficient is obtained by first calculating the ratio of the absolute value of the deviation to the preset deviation threshold, and then comparing the ratio with the natural number 1, taking the minimum value between the two. The new aggregation timing parameter is calculated based on the deviation adjustment coefficient. The new aggregation timing parameter = current aggregation timing parameter × (1 - adjustment coefficient × deviation adjustment coefficient), wherein the adjustment coefficient is preset to be within the range of 0 to 1.

[0050] Specifically, to adjust the aggregation timing parameters, a deviation adjustment coefficient must first be calculated based on the deviation value. This process requires clarifying the specific values ​​of the deviation value and the preset deviation threshold. The deviation value is derived from the difference between future performance and capacity data and current state data. The preset deviation threshold is set based on the historical stable operation data of the resource pool and the business's tolerance for resource fluctuations. For example, in a server cluster resource pool scenario, if the calculated comprehensive deviation value for CPU utilization and remaining storage space is 12%, and the preset deviation threshold is set to 8%, then the ratio of the absolute value of the deviation to the preset deviation threshold is first calculated. The absolute value of the deviation is 12%, and the ratio is 12% ÷ 8% = 1.5. This ratio is then compared with the natural number 1, and the minimum of the two is taken as the deviation adjustment coefficient. Since 1.2 is greater than 1, the deviation adjustment coefficient is 1. If the deviation value is 6%, and the preset deviation threshold is still 8%, then the ratio is 6% ÷ 8% = 0.75, which is less than 1, and the deviation adjustment coefficient is 0.75.

[0051] When calculating the new aggregation timing parameter, the current aggregation timing parameter and adjustment coefficient must first be determined. The current aggregation timing parameter is the interval at which the resource pool currently performs resource aggregation operations, for example, currently set to 30 minutes. The adjustment coefficient is preset in the range of 0 to 1, and its value needs to be set in conjunction with the sensitivity of the resource pool to parameter adjustments. If the number of devices in the resource pool is small and the load fluctuation is mild, the adjustment coefficient can be set to 0.3; if the number of devices is large and the load fluctuation is frequent, the adjustment coefficient can be set to 0.6. Here, we take an adjustment coefficient of 0.4 as an example. When the deviation adjustment coefficient is 0.75, substituting into the formula "new aggregation timing parameter = current aggregation timing parameter × (1 - adjustment coefficient × deviation adjustment coefficient)", we can obtain the new aggregation timing parameter as 30 minutes × (1 - 0.4 × 0.75) = 30 minutes × 0.7 = 21 minutes.

[0052] This process addresses the problem that traditional control strategies rely on static parameters and cannot adapt to nonlinear resource fluctuations. By calculating the deviation adjustment coefficient, it avoids excessive adjustments that could lead to system instability when the deviation value is too large. For example, when the deviation value is far above the preset threshold, the deviation adjustment coefficient is set to 1, limiting the upper limit of the adjustment range and preventing frequent resource scheduling caused by a sudden drop in aggregation timing parameters. When the deviation value is close to the preset threshold, the deviation adjustment coefficient is calculated according to the actual ratio to achieve fine-tuning and ensure that the parameter adjustment is in line with the actual resource fluctuation situation. At the same time, the introduction of the adjustment coefficient can be flexibly adapted according to the characteristics of the resource pool scenario, solving the problems of existing methods lacking dynamic optimization mechanisms and having a single adjustment strategy. For example, in large-scale resource pool scenarios, a larger adjustment coefficient is set to accelerate the response speed of aggregation timing parameters and cope with complex load fluctuations. In conventional-scale resource pool scenarios, a smaller adjustment coefficient is set to maintain parameter stability and avoid over-adjustment. Ultimately, dynamic and precise control of aggregation timing parameters is achieved, alleviating the imbalance problem of some equipment being overloaded and some equipment being idle, and improving resource utilization efficiency and system service quality.

[0053] Step S106: Based on the aggregation timing parameters and the initial setting of the speed parameters, obtain the current capacity distribution of the equipment, perform statistical analysis on the capacity distribution, and calculate the uniformity index.

[0054] In one specific embodiment, the process of executing step S106 may specifically include the following steps: Based on the aggregation timing parameters and the initial setting of the speed parameters, the current capacity distribution of devices in the resource pool is obtained; Statistical analysis is performed on the current capacity distribution to calculate the mean, standard deviation, and maximum and minimum difference, generating distribution characteristic data; The uniformity index characterizing the balance of capacity distribution is calculated based on the distribution characteristic data. The uniformity index is obtained by subtracting the standard deviation from the mean value from the natural number 1.

[0055] Specifically, when obtaining the current capacity distribution of the devices based on the initial settings of the aggregation timing parameter and the speed parameter, the aggregation timing parameter determines the time node for triggering capacity data collection. For example, if the aggregation timing parameter is set to perform data collection once every 20 minutes, the system will automatically start the data acquisition process after reaching this time node. The initial settings of the speed parameter limit the rate of data transmission and processing, such as setting it to process 15 device capacity data per second, to ensure that the full amount of device data collection is completed within the time window specified by the aggregation timing parameter. In practice, the system sends data collection commands to all devices in the resource pool based on the aggregation timing parameters. After responding to the commands, the devices provide real-time capacity data, such as the remaining storage space of the servers and the available memory of the computing nodes. Taking a resource pool containing 20 servers as an example, the remaining storage space data provided by each server are 45GB, 52GB, 48GB, 55GB, 49GB, 51GB, 47GB, 53GB, 50GB, 46GB, 54GB, 48GB, 50GB, 52GB, 49GB, 51GB, 47GB, 53GB, 48GB, and 50GB, respectively. These data are stored according to device identifiers to form the current capacity distribution dataset.

[0056] When performing statistical analysis on the current capacity distribution to calculate the mean, standard deviation, and maximum / minimum difference, the collected capacity data is first cleaned to remove obvious outliers (such as data exceeding the device's rated storage capacity). Then, the mean is calculated based on the cleaned data using the arithmetic mean method, with the formula: Mean = Sum of all device capacity data ÷ Number of devices. The average of the remaining storage space data for the 20 servers is 49.8GB, and the standard deviation is approximately 2.87GB. When calculating the maximum / minimum difference, the maximum value of 55GB and the minimum value of 45GB are selected from the capacity data, with a difference of 10GB. The mean, standard deviation, and maximum / minimum difference together constitute the distribution characteristic data.

[0057] When calculating the evenness index based on distribution characteristic data, the formula " Substituting the calculated standard deviation of 2.87 GB and mean of 49.8 GB into the formula, we first calculate the ratio of the standard deviation to the mean, i.e. Subtracting this ratio from 1 yields a uniformity index of approximately 0.9424.

[0058] By aggregating timing and speed parameters, the timeliness and completeness of capacity data are ensured, solving the problem of data lag in traditional static acquisition methods. Statistical analysis generates multi-dimensional distribution characteristic data, avoiding the limitation that a single data dimension cannot reflect the dispersion of resource distribution. The uniformity index calculation transforms abstract distribution characteristics into intuitive quantitative indicators of balance, solving the problem that traditional methods struggle to determine whether resource allocation is reasonable. For example, when the uniformity index is too low, capacity distribution imbalances can be identified in a timely manner, providing a basis for subsequent parameter optimization, avoiding the problem of some equipment being overloaded and others being idle, and improving resource utilization efficiency.

[0059] Step S107: Determine whether the uniformity index is lower than the preset uniformity threshold. If so, update the speed parameter through iterative optimization and generate an aggregation control command in combination with the aggregation timing parameter, and output it to the resource pool for dynamic adjustment.

[0060] In one specific embodiment, the process of executing step S107 may specifically include the following steps: Determine whether the uniformity index is lower than the preset uniformity threshold. If it is lower, update the speed parameter value through a step-size adaptive iterative optimization method to reduce the gap between the uniformity index and the preset uniformity threshold. Based on the updated speed parameters and aggregation timing parameters, an aggregation control command containing the values ​​of both is generated; The aggregation control command is output to the resource pool, instructing the resource pool to perform resource aggregation operations according to the rate defined by the speed parameter value and the timing defined by the aggregation timing parameter, so as to achieve stable aggregation.

[0061] Specifically, the preset uniformity threshold needs to be determined based on the historical stable operation data of the resource pool and the business requirements for the balance of capacity distribution. For example, in a resource pool scenario containing 50 servers, if historical data shows that the uniformity index is maintained at 0.9 or above, the resource allocation is balanced and the equipment is operating stably, and the preset uniformity threshold can be set to 0.9. The uniformity index is calculated through step S106, and its value comes from the statistical analysis of the current capacity distribution of the equipment. Specifically, it is 1 minus the ratio of the standard deviation of the capacity distribution to the average value. If the average value of the current capacity distribution of a resource pool is 50GB and the standard deviation is 3GB, then the uniformity index is 1-(3÷50)=0.94. When the uniformity index (e.g., 0.85) is lower than the preset uniformity threshold (0.9), the iterative optimization and update process of the speed parameters needs to be initiated.

[0062] The iterative optimization of speed parameters employs an adaptive step-size iterative optimization method. The core of this method is to dynamically adjust the step size based on the deviation between the uniformity index and a preset uniformity threshold, thereby achieving precise updates to the speed parameters. First, the deviation value is calculated, which is the preset uniformity threshold minus the current uniformity index. If the preset threshold is 0.9 and the current uniformity index is 0.85, the deviation value is 0.05. Next, the initial step size is determined based on the deviation value; the larger the deviation value, the larger the initial step size. For example, when the deviation value is 0.05, the initial step size is set to 0.02 (the unit for speed parameter adjustment is GB / min). Then, the iterative process begins. In the first iteration, the current speed parameter (e.g., initially set to 2 GB / min) is added to the initial step size to obtain a new candidate speed parameter value of 2.02 GB / min. Based on this candidate value, the capacity distribution of the resource pool devices is re-acquired, and a new uniformity index is calculated. If the deviation of the new uniformity index (e.g., 0.87) from the preset threshold (0.03) is less than the previous deviation (0.05), it indicates that the step direction is correct. In the next iteration, the step direction can be maintained and the step size adjusted according to the deviation change, setting the step size to 0.015. If the deviation of the new uniformity index from the preset threshold increases, the step direction is adjusted in the opposite direction. This process is repeated until the calculated uniformity index reaches or approaches the preset uniformity threshold. The corresponding speed parameter at this point is the updated speed parameter. For example, after 3 iterations, the speed parameter is updated to 2.04 GB / min, and the corresponding uniformity index increases to 0.902, meeting the deviation requirement from the preset threshold.

[0063] After updating the speed parameters, an aggregation control command needs to be generated by combining the updated speed parameters with the aggregation timing parameters adjusted in step S105. The aggregation timing parameter is the interval between resource pool aggregation operations, for example, 15 minutes after deviation adjustment. The aggregation control command must include the specific values ​​of the speed parameters and aggregation timing parameters, forming a command such as "Execute a resource aggregation operation every 15 minutes at a rate of 2.04 GB / min". This command must conform to the command format specification of the resource pool control system to ensure that the equipment can accurately interpret the meaning of the parameters.

[0064] After the aggregation control command is output to the resource pool, the control module within the resource pool receives the command and parses the speed parameters and aggregation timing parameters. According to the rate defined by the speed parameters, the devices adjust the transmission rate of capacity allocation during resource aggregation, for example, transferring the capacity of idle devices to devices with higher loads at a rate of 2.04 GB / min. According to the timing defined by the aggregation timing parameters, a capacity reallocation operation is triggered every 15 minutes across the entire resource pool, achieving dynamic adjustment of device capacity distribution.

[0065] By initiating dynamic adjustments based on the relationship between the uniformity index and a preset threshold, the problem of traditional static parameters being unable to adapt to changes in capacity distribution is solved. The application of the step-size adaptive iterative optimization method avoids over- or under-adjustment during the speed parameter adjustment process, ensuring that parameter updates can accurately narrow the gap between the uniformity index and the threshold, thus solving the problem of insufficient flexibility in parameter adjustment in existing methods. The aggregation control command includes specific speed and timing parameters, giving resource aggregation operations a clear execution basis and avoiding resource imbalance caused by lag in aggregation commands. In large-scale resource pool scenarios, reasonable aggregation timing and rate control can reduce system response latency, ensure service quality, and alleviate the problem of some devices being overloaded and others being idle, thereby improving resource utilization efficiency.

[0066] The above describes an aggregated intelligent control method for user flexible resources in an embodiment of this application. Please refer to [link / reference]. Figure 2 In this application embodiment, an aggregated intelligent control process 200 for user flexible resources includes seven sub-processes, which are sequentially connected and logically coherent. The specific operations are as follows: The data acquisition and preprocessing process 201 begins with sensor network monitoring. Real-time performance and capacity data of each device in the resource pool are collected via the sensor network. After median filtering for noise reduction and min-max normalization, a time series at a uniform scale is formed. Different feature extraction methods are selected based on the scale of the devices: for fewer than 50 devices (a standard scenario), the ARMA model is used to analyze fluctuation amplitude and period; for more than 50 devices (a large-scale scenario), Fourier transform is applied to extract the dominant frequency in the frequency domain. Finally, these features are combined into a multi-dimensional vector sequence to construct a resource deformation sequence.

[0067] The deformation trend description generation process 202 uses the resource deformation sequence as input. First, the sliding window size is set according to the collection frequency. The moving average method is used to obtain the arithmetic mean of data from multiple consecutive time points to generate a smooth sequence to eliminate short-term noise. Then, the ARMA model is used to extract features. The model order is determined according to the length of the smooth sequence using the AIC criterion. After training with input data, the coefficients are estimated using the least squares method, and the trend slope and periodic fluctuations are extracted as trend features. Finally, the trend features are input into a linear regression model, with the timestamp as the independent variable and the smooth sequence data as the dependent variable. The least squares method is used to calculate the regression coefficients and intercept, fit the trend line, generate a numerical index representing the long-term direction of change, and construct the deformation trend description.

[0068] The process of forming contour feature parameters 203 is based on deformation trend description, using K-means clustering for grouping: The number of clusters is preset according to data dimensions and device scale; initial center points are selected; the Euclidean distance from data points to the center points is iteratively calculated and clusters are assigned; the center points are updated (taking the average value of each dimension of data within the cluster) until the center point difference is less than a threshold, generating stable clusters. The average value of each dimension of data within each cluster is extracted as the center point, representing the cluster deformation trend concentration characteristics. The fluctuation amplitude value (peak-valley value) of the deformation trend of the center point is calculated; the average number of fluctuations per unit time is used as the fluctuation frequency value; and the average time from the start to stabilization is measured as the fluctuation duration. These three are combined in a fixed order into a multi-dimensional vector to form the contour feature parameters.

[0069] The LSTM network predicts future deformation trends by using contour feature parameters as the object of analysis. It calculates the second-order differences for each dimension of the time series parameters. If the difference value at three or more consecutive time points in any dimension exceeds a threshold, nonlinear fluctuations are identified. If present, historical nonlinear fluctuation contour features are used as input. The LSTM network structure is configured according to the data scale (e.g., the time step size matches the acquisition period, and the number of hidden layer units is configured according to the step size). Samples are constructed by predicting the N+1th step using consecutive N-step inputs. Mean squared error is used as the loss function for training until convergence. Finally, the current consecutive time step parameters are input into the trained LSTM, which outputs a multi-step prediction sequence to obtain the deformation trend.

[0070] The process of adjusting aggregation timing parameters 205 first acquires real-time performance and capacity data of the equipment through the sensor network as the current status data. Then, it combines the mapping relationship between the predicted value of future deformation trends and actual business indicators (such as using normalized fluctuation amplitude to back-calculate the original indicators) to calculate the performance and capacity data at future time points. Next, it calculates the deviation value: first calculates the deviation of individual indicators, then sets weights according to business importance, and sums them to obtain the comprehensive deviation value. It determines whether the deviation value exceeds the preset threshold (combining historical stable data and business requirements): if it does not exceed the threshold, the current aggregation timing parameters are maintained; if it does, it is adjusted. First, the deviation adjustment coefficient is calculated (the ratio of the absolute value of the deviation to the threshold and 1, taking the smaller value), and then the new parameters are calculated according to the formula "New aggregation timing parameter = Current parameter × (1 - Adjustment coefficient × Deviation adjustment coefficient)". The adjustment coefficient is preset to 0-1 (set according to resource pool sensitivity) to obtain the new parameters.

[0071] The process 206 for calculating the uniformity index, based on the adjusted aggregation timing parameters (determining the acquisition time) and initial speed parameter settings (limiting the data transmission processing rate), acquires the current capacity distribution data of the devices (such as remaining storage space), and forms a dataset by classifying it according to device identification. After cleaning the data to remove outliers, the mean, standard deviation, maximum and minimum differences are calculated to generate distribution characteristic data. Finally, the uniformity index, characterizing the balance of capacity distribution, is calculated using the formula: uniformity index = 1 - (standard deviation / mean).

[0072] The aggregation control command generation process 207 is based on the uniformity index. If the current uniformity index value is lower than the preset uniformity index threshold, adaptive iterative optimization with a step size is adopted: an initial step size is set according to the deviation, and the speed parameters are adjusted iteratively (generating candidate values ​​and recalculating the uniformity index to update the step size) until the index approaches the threshold; otherwise, the current speed parameters are maintained. Subsequently, the final speed parameters and aggregation timing parameters are assembled into a command according to the control system requirements (such as "aggregate at XX rate every XX minutes"). After the command is sent to the resource pool control module, it is parsed and the capacity allocation rate and aggregation trigger timing are adjusted according to the parameters to achieve dynamic resource regulation.

[0073] The above describes an aggregated intelligent control process for user flexible resources in an embodiment of this application. Please refer to [link / reference]. Figure 3 One embodiment of the aggregated intelligent control system 300 for user flexible resources in this application includes: The data acquisition module 301 is used to acquire real-time data on equipment performance and capacity from the resource pool, perform preprocessing, time series analysis and fluctuation feature extraction, and generate a resource deformation sequence. The trend extraction module 302 is used to smooth the resource deformation sequence, extract time-series features and fit trends to generate a deformation trend description that represents the long-term change direction. The feature clustering module 303 is used to generate stable clusters based on deformation trend description using the K-means clustering algorithm, extract center point features, and construct multi-dimensional contour feature parameters; The trend prediction module 304 is used to train an LSTM network and generate future deformation trend prediction values ​​when there are nonlinear fluctuations in the contour feature parameters. The deviation calculation module 305 is used to compare the predicted value of future deformation trend with the current state data, calculate the deviation value, and adjust the aggregation timing parameters accordingly. The uniformity calculation module 306 is used to statistically analyze the current capacity distribution and calculate the uniformity index based on the aggregation timing parameters and speed parameters. The instruction output module 307 is used to generate and output aggregation control instructions by iteratively optimizing and updating the speed parameters and combining them with the aggregation timing parameters when the uniformity index is lower than the preset uniformity threshold.

[0074] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for aggregated intelligent control of user flexible resources, characterized in that, The method includes: Step S101: Obtain real-time data on equipment performance and capacity from the resource pool, perform preprocessing, time-series analysis and fluctuation feature extraction on the real-time data, and generate a resource deformation sequence; Step S102: Based on the resource deformation sequence, determine the deformation trend description through smoothing, time-series feature extraction, and trend fitting; Step S103: Use the K-means clustering algorithm to group the deformation trend descriptions, determine the cluster center points, extract the feature combination of the cluster center points into a multi-dimensional vector, and form contour feature parameters; Step S104: Determine whether there is nonlinear fluctuation in the contour feature parameters. If so, use an LSTM network to train and generate a predicted value for future deformation trends. Step S105: Obtain the current status data of the resource pool, calculate the deviation value between the predicted value of the future deformation trend and the current status data, and adjust the aggregation timing parameters based on the comparison result between the deviation value and the preset deviation threshold. Step S106: Based on the aggregation timing parameters and the initial setting of the speed parameters, obtain the current capacity distribution of the equipment, perform statistical analysis on the capacity distribution, and calculate the uniformity index. Step S107: Determine whether the uniformity index is lower than the preset uniformity threshold. If so, update the speed parameter through iterative optimization and generate an aggregation control command in combination with the aggregation timing parameter, and output it to the resource pool for dynamic adjustment.

2. The method for aggregated intelligent control of user flexible resources according to claim 1, characterized in that, Step S101 includes: By deploying a sensor network in the resource pool, the performance and capacity fluctuations of the devices in the resource pool are monitored, real-time data is collected, and noise values ​​are removed from the real-time data using a mean filtering method. The real-time data, after noise filtering, is normalized and then sorted by timestamp to generate time series data. Identify the scale of the resource pool scenario. If it is a regular-scale scenario, use an autoregressive moving average model to capture the autocorrelation and moving average components of the time series data, and extract the fluctuation amplitude and periodic features as fluctuation features. If it is a large-scale resource pool scenario, use the Fourier transform method to convert the time series data from the time domain to the frequency domain, and extract the dominant frequency as fluctuation features. The fluctuation characteristics are combined into a multi-dimensional vector sequence to generate a resource deformation sequence that characterizes the dynamic change trend of equipment performance and capacity.

3. The method for aggregated intelligent control of user flexible resources according to claim 1, characterized in that, Step S102 includes: The resource deformation sequence is smoothed by applying a moving average method to generate a smoothed sequence. The smoothed sequence is subjected to feature extraction using time series analysis methods. An autoregressive moving average model is used to capture time series dependencies, and trend slope and periodic fluctuations are extracted as trend features. The slope and fluctuation pattern in the trend features are input into a linear regression model to fit a long-term trend line and generate a numerical index that represents the long-term direction of change. Then, based on the numerical index, a deformation trend description that represents the long-term direction of change of the resource deformation sequence is constructed.

4. The method for aggregated intelligent control of user flexible resources according to claim 1, characterized in that, Step S103 includes: The K-means clustering algorithm was used to group the deformation trend descriptions, generating multiple stable clusters. Analyze multiple stable clusters, extract the average value of all data points in each cluster, and use the average value as the center point of the corresponding cluster. The center point represents the concentrated characteristics of the deformation trend within the cluster. Calculate the fluctuation amplitude value of the deformation trend description corresponding to the center point of each cluster, where the fluctuation amplitude value is the difference between the trend peak and the trough value; The average number of fluctuations in the deformation trend description corresponding to each cluster center point per unit time is used as the fluctuation frequency value. The deformation trend at each cluster center point is measured to describe the average time from the start to the stabilization point, which is taken as the duration of the fluctuation. The fluctuation amplitude value, fluctuation frequency value, and fluctuation duration are combined into a multi-dimensional vector to generate the contour feature parameters, wherein the contour feature parameters are used to characterize typical features of deformation trends.

5. The method for aggregated intelligent control of user flexible resources according to claim 1, characterized in that, Step S104 includes: Calculate the second difference of the contour feature parameter sequence. If the absolute value of the second difference exceeds a preset fluctuation threshold for three or more consecutive time points, the contour feature parameter is determined to be nonlinear fluctuation. The contour feature parameters that are determined to be nonlinear fluctuations within a historical time period are used as feature inputs. An LSTM network is used to train the network, and the number of hidden layer units and the number of training iterations are set. The mean squared error is used as the loss function for backpropagation optimization. Based on the trained LSTM network, the current contour feature parameters are input to generate a prediction sequence for multiple future time points, thereby obtaining the predicted value of the future deformation trend, wherein the predicted value of the future deformation trend characterizes the future performance and capacity change trend of the resource pool.

6. The method for aggregated intelligent control of user flexible resources according to claim 1, characterized in that, Step S105 includes: Obtain the current status data of the resource pool, wherein the current status data is the real-time performance and capacity data of the devices in the resource pool; Based on the predicted future deformation trend, calculate the performance and capacity data of the resource pool at future points in time. Calculate the deviation between the performance and capacity data at the future time point and the current state data; Determine whether the deviation value is greater than a preset deviation threshold. If so, the aggregation timing parameter needs to be adjusted to maintain the stability and balance of the resource pool. If not, the aggregation timing parameter does not need to be adjusted.

7. The method for aggregated intelligent control of user flexible resources according to claim 6, characterized in that, The adjustment of the aggregation timing parameters includes: The deviation adjustment coefficient is calculated based on the deviation value. The deviation adjustment coefficient is obtained by first calculating the ratio of the absolute value of the deviation to the preset deviation threshold, and then comparing the ratio with the natural number 1, taking the minimum value between the two. The new aggregation timing parameter is calculated based on the deviation adjustment coefficient. The new aggregation timing parameter = current aggregation timing parameter × (1 - adjustment coefficient × deviation adjustment coefficient), wherein the adjustment coefficient is preset to be within the range of 0 to 1.

8. The method for aggregated intelligent control of user flexible resources according to claim 1, characterized in that, Step S106 includes: Based on the aggregation timing parameters and the initial setting of the speed parameters, the current capacity distribution of devices in the resource pool is obtained; Statistical analysis is performed on the current capacity distribution to calculate the mean, standard deviation, and maximum and minimum difference, generating distribution characteristic data; The uniformity index characterizing the balance of capacity distribution is calculated based on the distribution characteristic data. The uniformity index is obtained by subtracting the standard deviation from the mean value from the natural number 1.

9. The method for aggregated intelligent control of user flexible resources according to claim 1, characterized in that, Step S107 includes: Determine whether the uniformity index is lower than the preset uniformity threshold. If it is lower, update the speed parameter value through a step-size adaptive iterative optimization method to reduce the gap between the uniformity index and the preset uniformity threshold. Based on the updated speed parameters and aggregation timing parameters, an aggregation control command containing the values ​​of both is generated; The aggregation control command is output to the resource pool, instructing the resource pool to perform resource aggregation operations according to the rate defined by the speed parameter value and the timing defined by the aggregation timing parameter, so as to achieve stable aggregation.

10. A converged intelligent control system for user flexible resources, characterized in that, For implementing the aggregated intelligent control method for user flexible resources as described in any one of claims 1 to 9, the aggregated intelligent control system for user flexible resources comprises: The data acquisition module is used to obtain real-time data on equipment performance and capacity from the resource pool, perform preprocessing, time series analysis and fluctuation feature extraction, and generate resource deformation sequences. The trend extraction module is used to smooth the resource deformation sequence, extract time-series features and fit trends to generate a deformation trend description that represents the long-term direction of change. The feature clustering module is used to generate stable clusters based on deformation trend description using the K-means clustering algorithm, extract center point features, and construct multi-dimensional contour feature parameters; The trend prediction module is used to train an LSTM network and generate future deformation trend prediction values ​​when there are non-linear fluctuations in the contour feature parameters. The deviation calculation module is used to compare the predicted value of future deformation trend with the current state data, calculate the deviation value, and adjust the aggregation timing parameters accordingly. The uniformity calculation module is used to statistically analyze the current capacity distribution and calculate the uniformity index based on aggregation timing parameters and speed parameters. The instruction output module is used to generate and output aggregation control instructions by iteratively optimizing and updating the speed parameters when the uniformity index is lower than the preset uniformity threshold, combined with the aggregation timing parameters.

Images