Method, device and method for generating a prediction model of a reduced pool repair duration

Abstract

The application discloses a prediction model generation method, device and prediction method for shortening pool repair duration, and relates to the technical field of server repair. The prediction model generation method comprises the following steps: obtaining training set data and test set data of the repair duration; establishing a support vector machine model related to the number of categories of environment parameters, determining model parameters of the support vector machine model according to the training set data, and taking the support vector machine model with the determined model parameters as a prediction model for the shortening pool repair duration. Through implementation of the prediction model generation method, device and prediction method for shortening pool repair duration disclosed in the application, accurate shortening pool repair duration expectation can be obtained, the convergence speed of the model is improved, the training error is reduced, and the proportion of extreme value samples in the classification model is improved in terms of model parameter optimization.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of server repair, and in particular to a method and device for generating a prediction model for predicting the duration of thin pool repair and a prediction method. BACKGROUND

[0002] When a server cluster experiences a failure such as power failure and multiple node battery failure resulting in cache loss, manual commands are usually implemented by engineers to perform T3 recovery on the server cluster. Thin pool repair is an important part of T3 recovery. Through thin pool repair, the data of the server cluster can be recovered to the greatest extent. The server cluster after thin pool repair can be online and further recover its read-write function. It can be seen that thin pool repair plays an important role in the whole process of T3 recovery of the server cluster. However, the duration of thin pool repair is difficult to estimate, because the duration of thin pool repair is affected by many factors such as the number of processor cores, the frequency of the processor, the size of the memory, and the frequency of the memory. Too many influencing factors make it difficult to intuitively estimate the duration of thin pool repair. This uncertainty of time is not conducive to the planning of repair work for engineers performing manual operations, affects repair efficiency, and even interferes with the progress of repair work, and seriously affects the service quality of the enterprise. SUMMARY

[0003] To solve the problem that the duration of thin pool repair is difficult to estimate in the prior art, the present application provides a method and device for generating a prediction model for predicting the duration of thin pool repair and a prediction method, which accurately predicts the duration of thin pool repair, and facilitates engineers to plan the recovery work of the server cluster according to the predicted duration of thin pool repair.

[0004] To solve one or more of the above technical problems, the technical solution adopted by the present application is as follows:

[0005] In a first aspect, a method for generating a prediction model for predicting the duration of thin pool repair is provided, comprising:

[0006] obtaining training set data and test set data for the duration of repair;

[0007] establishing a support vector machine (SVM) model related to the number of categories of environmental parameters, determining the model parameters of the support vector machine model according to the training set data, and determining the support vector machine model after the model parameters as the prediction model for the duration of thin pool repair; wherein the model parameters include kernel function parameters, weight values and bias values.

[0008] Further, obtaining training set data and test set data for the duration of repair comprises:

[0009] Collect N groups of repair duration original sample data as an original data set; wherein, N is a positive integer, and N≥2; each group of original sample data includes: M environment parameter data and 1 time data; the environment parameters at least include: the number of processor cores, the processor frequency, the memory size, the memory frequency, the storage pool capacity, the write data volume, the deduplication rate, the write mode, and the repair time;

[0010] Normalize the original data set to obtain a corresponding normalized data set;

[0011] Select X groups of data in the normalized data set as training set data, and the remaining N-X groups of data as test set data; wherein, X is a positive integer, and X

[0012] Further, a support vector machine model related to the number of categories of environment parameters is established, model parameters of the support vector machine model are determined according to the training set data, and the support vector machine model with the determined model parameters is used as a simplified pool repair duration prediction model, which includes:

[0013] The kernel function parameters of the kernel function are determined using the K-fold cross-validation method; wherein, the kernel function is a radial basis function (RBF) kernel, and K is a positive integer;

[0014] The weight value and the bias value are obtained according to the radial basis function, the kernel function parameters, and the training set data.

[0015] Further, the kernel function parameters of the kernel function are determined using the K-fold cross-validation method, which includes:

[0016] The inverse of the initial value of the kernel function parameters is associated with the number of categories of environment parameters;

[0017] The number of folds of the K-fold cross-validation method is associated with the number of categories of environment parameters;

[0018] The training data set is randomly divided into K mutually exclusive sub-training sets, the support vector machine model is trained K times, and K kernel function parameter process values are obtained;

[0019] The average of the K kernel function parameter process values is obtained to obtain the kernel function parameters.

[0020] Further, the weight value and the bias value are obtained according to the radial basis function, the kernel function parameters, and the training set data, which includes:

[0021] The kernel function parameters are substituted into the support vector machine model, and the support vector machine model is trained using the training set data to obtain a first weight and a first bias;

[0022] Obtaining second training set data by eliminating data of the group where the extreme value sample is located from the training set data, and training the support vector machine model with the training set data to obtain second weight and second bias;

[0023] Taking the average of the first weight and the second weight as the weight value, and taking the average of the first bias and the second bias as the bias value.

[0024] Further, the support vector machine model after the model parameters are determined is used as the simplified pool repair duration prediction model, and the simplified pool repair duration prediction model further comprises:

[0025] Performing prediction on the simplified pool repair duration prediction model, and in response to the prediction relative error value being greater than a preset range, adjusting the kernel function parameter and updating the simplified pool repair duration prediction model.

[0026] Further, the prediction on the simplified pool repair duration prediction model comprises:

[0027] Inputting at least one group of environmental parameter data in the test set data into the simplified pool repair duration prediction model to obtain a prediction result corresponding to the environmental parameter data;

[0028] Performing inverse normalization on the prediction result to obtain a corresponding prediction duration;

[0029] Obtaining a prediction relative error value between the prediction duration corresponding to the environmental parameter data and time data, and correcting the simplified pool repair duration prediction model according to the prediction relative error value.

[0030] Further, the adjustment of the kernel function parameter and the updating of the simplified pool repair duration prediction model comprise:

[0031] Starting from the kernel function parameter of the kernel function determined by using the K-fold cross-validation method, the simplified pool repair duration prediction model is obtained again.

[0032] In a second aspect, a prediction model generation device for a simplified pool repair duration is provided, comprising:

[0033] A data set acquisition module is configured to acquire training set data and test set data for a repair duration;

[0034] A model establishment module is configured to establish a support vector machine model related to the number of categories of environmental parameters, determine model parameters of the support vector machine model according to the training set data, and use the support vector machine model after the model parameters are determined as a simplified pool repair duration prediction model; wherein the model parameters comprise a kernel function parameter, a weight value and a bias value.

[0035] Further, the prediction model generation device further comprises:

[0036] The model optimization module is used to predict the simplified pool repair time prediction model. In response to the prediction relative error value being greater than the preset range, the kernel function parameters are adjusted and the simplified pool repair time prediction model is updated.

[0037] Thirdly, it provides methods for predicting the repair time of the simplified pool, including:

[0038] Obtain the environmental parameters to be predicted;

[0039] Normalize the environmental parameters to be predicted;

[0040] The normalized environmental parameters to be predicted are input into the simplified pool repair time prediction model to obtain the model prediction result. The simplified pool repair time prediction model is generated according to the simplified pool repair time prediction model generation method described in the first aspect.

[0041] The model prediction results are inversely normalized to obtain the simplified pool repair prediction time.

[0042] The beneficial effects of the technical solution provided by the embodiments of the present invention are as follows:

[0043] 1. By implementing the prediction model generation method, apparatus and prediction method for simplified pool repair time disclosed in the embodiments of this application, an accurate prediction of the simplified pool repair time can be obtained, which is convenient for scheduling the workflow based on the time.

[0044] 2. By associating the initial values ​​of the kernel function parameters with the number of categories of environmental parameters, the convergence speed of the model can be improved;

[0045] 3. Improve model accuracy by correlating the number of folds in cross-validation with the number of categories of environmental parameters;

[0046] 4. In terms of parameter optimization, the model is trained with two sets of training data, one containing extreme value groups and the other not, to obtain the corresponding weight and bias values. The average value of each set is then taken as the final weight and bias, which increases the proportion of extreme value samples in the classification model and enables accurate prediction of repair time in special cases. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a schematic diagram of the method for generating a prediction model for the simplified pool repair time provided in an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram of another method for generating a prediction model for the simplified pool repair time provided in an embodiment of the present invention;

[0050] Figure 3 This is a schematic diagram of the device for generating a prediction model for the simplified pool repair time provided in an embodiment of the present invention;

[0051] Figure 4 This is a schematic diagram of the method for predicting the repair time of the simplified pool provided in an embodiment of the present invention. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0053] Unless otherwise defined, the technical or scientific terms used in this disclosure shall have the ordinary meaning understood by one of ordinary skill in the art to which this disclosure pertains. The terms “first,” “second,” and similar terms used in this disclosure do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, the terms “an,” “a,” or “the,” and similar terms do not indicate a quantity limitation, but rather indicate the presence of at least one. The numbers in the accompanying drawings are only used to distinguish individual functional parts or modules and do not indicate logical relationships between parts or modules. The terms “comprising,” “including,” or “including,” and similar terms mean that the element or object preceding the term encompasses the element or object listed following the term and its equivalents, without excluding other elements or objects. The terms “connected,” “linked,” and similar terms are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. “Above,” “below,” “left,” “right,” etc., are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0054] The various embodiments according to this disclosure will now be described in detail with reference to the accompanying drawings. It should be noted that in the drawings, the same reference numerals are assigned to components having substantially the same or similar structure and function, and repeated descriptions of them will be omitted.

[0055] In one embodiment, such as Figure 1 As shown, the method for generating a prediction model for the simplified pool repair time includes:

[0056] S100: Obtain the training set data and test set data for the repair duration;

[0057] S200: Establish a support vector machine model related to the number of categories of environmental parameters, determine the model parameters of the support vector machine model based on the training set data, and use the support vector machine model with determined model parameters as the simplified pool repair duration prediction model; wherein, the model parameters include: kernel function parameters, weight values ​​and bias values.

[0058] Specifically, the training set data and test set data for the repair duration are obtained, including:

[0059] S110: Collect N sets of original sample data on repair time as the original dataset; where N is a positive integer and N≥2; each set of original sample data includes: M environmental parameter data and 1 time data; the environmental parameters include: number of processor cores, processor frequency, memory size, memory frequency, storage pool capacity, amount of data written, deduplication rate, write mode, and repair time.

[0060] S120: Normalize the original dataset to obtain the corresponding normalized dataset;

[0061] S130: Select X groups of data from the normalized dataset as the training set data, and use the remaining NX groups of data as the test set data; where X is a positive integer, and X... <N。

[0062] Thin pool repair time is directly related to data recovery speed. The environmental parameters mentioned above include data recovery speed and the amount of data written. Data recovery speed is affected by the storage device's processor and memory; the amount of data written includes: storage pool size, deduplication rate, and data write mode (sequential or random); in addition, it is also affected by other running processes on the device. Generally, the following eight factors have a critical impact on thin pool repair time: number of processor cores, processor frequency, memory size, memory frequency, storage pool capacity, amount of data written, deduplication rate, and write mode.

[0063] T3 recovery was performed on the server cluster under different storage configurations, and the repair time of different thin pools was recorded as sample data. For example, sample [s1, s2, s3, s4, s5, s6, s7, s8, r], where s1 is the number of processor cores, s2 is the processor frequency, s3 is the memory size, s4 is the memory frequency, s5 is the storage pool capacity, s6 is the amount of data written, s7 is the deduplication rate, s8 is the write mode (sequential write is denoted as 1, random write is denoted as 2), and r is the thin pool repair time. In this case, M = 8. Taking the original sample data shown in Table 1 as an example, N = 7.

[0064] Since the units and value ranges of each feature in the sample are different. For example, the unit of the storage pool can be GB or TB; the unit of the CPU main frequency can be MHz or GHz. At the same time, there may also be some outliers (such as the data write volume being 0 due to device failure). This requires reasonable normalization of the sample so that the preprocessed data is limited within a certain range, thereby eliminating the adverse effects caused by singular sample data and improving the convergence speed of the prediction model.

[0065] The original sample data will be normalized to the range [0, 1]. The original sample data is normalized by the following formula.

[0066]

[0067] Among them, [a, b] represents the normalization range, and Max and Min respectively represent the maximum and minimum values of the original sample data.

[0068] Table 1 Original sample data table

[0069]

[0070] Collect N groups of samples (N is a positive integer greater than 2), and normalize all the collected data to obtain a normalized data set, as follows matrix:

[0071]

[0072] In one embodiment, select the 1st to the Xth groups of samples as the training set data, and the (X + 1)th to the 7th groups as the test set data. At this time, 1 < X < 7 and X is an integer.

[0073] Establish a support vector machine model related to the number of categories of environmental parameters, determine the model parameters of the support vector machine model according to the training set data, and use the support vector machine model after determining the model parameters as the prediction model for the refined pool repair duration.

[0074] Specifically include:

[0075] S210: Use the K-fold cross-validation method to determine the kernel function parameters of the kernel function; among them, the kernel function is the radial basis kernel function, and K is a positive integer;

[0076] Specifically include:

[0077] S211: Associate the reciprocal of the initial value of the kernel function parameters with the number of categories of environmental parameters;

[0078] S212: Associate the number of folds of the K-fold cross-validation method with the number of categories of environmental parameters;

[0079] S213: Randomly divide the training dataset into K mutually exclusive sub-training sets, train the support vector machine model K times, and obtain the process values ​​of K kernel function parameters;

[0080] S214: Calculate the average value of the K kernel function parameter process values ​​to obtain the kernel function parameters.

[0081] For model training, you can use pre-built support vector machine (SVM) model libraries, such as libSVM, which has implementations in Python, Java, and Matlab. When using libSVM, you need to ensure that the data conforms to its required format. After training, you will obtain the prediction model.

[0082] Choosing the right kernel function is crucial for building a support vector machine (SVM) model. The kernel function implicitly maps data from low to high dimensions, thus avoiding the direct computation of inner products in high dimensions. Mapping sample data to a high-dimensional space requires the use of kernel functions. Radial basis function (RBF) kernels are suitable for solving problems involving sample space mapping and inner product calculations.

[0083] Using the radial basis function kernel, the support vector machine model can be represented as: Here, ω represents the weights, b represents the biases, and the input to x is a feature vector of the form [s1, s2, s3, s4, s5, s6, s7, s8, r]. The feature vectors are separable in high dimension after being mapped by the kernel function. A precise value of f(x) is obtained based on the trained weight vectors and biases. The first step is to determine the kernel function parameters. Based on these parameters, the optimal solutions for the weights and biases can be obtained through model learning and training.

[0084] To obtain the kernel function parameters, the initial values ​​of the kernel function parameters first need to be correlated with the number of categories of environmental parameters. This application uses the correlation method: g0 = 1 / M, where g0 is the initial value of the kernel function parameters. This strongly correlates the initial values ​​of the kernel function parameters with the horizontal dimension of the training set data matrix, improving the model convergence speed. For K-fold cross-validation, K = M is chosen. For the case of eight environmental parameter categories, M = 8, g0 = 1 / 8, and K = 8. Correlating the number of folds in K-fold cross-validation with the number of categories of environmental parameters improves training accuracy. The training set data is randomly divided into eight mutually exclusive subsets; during training, seven subsets are randomly selected as training samples, and the remaining one is used as a test sample. After each training iteration, the kernel function parameters g are adjusted based on the training results, and a penalty factor c is obtained. The average of the eight model results is used as the kernel function parameters.

[0085] During training, the error is set to e = 0.0000000001. After each training iteration, a training error is generated, such as p0 for the first iteration, and a penalty factor c0 is obtained, c0 = p0 - e. The kernel parameter g is adjusted according to the penalty factor, so g1 = g0 + c0. After training, p1 is obtained, and c1 = p1 - e; and so on. When the kernel function parameters are adjusted incorrectly, the calculated penalty factor increases (with both positive and negative values), thus affecting the kernel function parameter values ​​in the next training iteration. When the kernel function parameter values ​​are adjusted correctly, the training error decreases, the penalty factor decreases, and the adjustment range of the kernel function parameters decreases in the next training iteration.

[0086] S220: Obtain weight and bias values ​​based on the radial basis function kernel function, kernel function parameters, and training set data.

[0087] The kernel function parameters are determined using K-fold cross-validation, specifically including:

[0088] The weight and bias values ​​are obtained based on the radial basis function kernel function, kernel function parameters, and training set data, specifically including:

[0089] S221: Substitute the kernel function parameters into the support vector machine model, and train the support vector machine model with the training set data to obtain the first weight and the first bias;

[0090] S222: Remove the data of the group containing the extreme value samples from the training set data to obtain the second training set data, and use the training set data to train the support vector machine model to obtain the second weight and the second bias;

[0091] S223: Take the average of the first weight and the second weight as the weight value, and take the average of the first deviation and the second deviation as the deviation value.

[0092] Learning from all data in the training set yields the first weight ω and the first bias b. However, for samples with extreme values, although the proportion of extreme samples is reduced during normalization, their high dispersion and low reliability can interfere with the convergence speed of the training process. Therefore, samples containing extreme values ​​are removed, and the model is retrained to obtain the second weight ω′ and the second bias b′. Since the removed extreme samples actually exist in the actual testing process (possibly as faults), they still have training value. Finally, ω*=(ω+ω′) / 2 and b*=(b+b′) / 2 are taken as the weight and bias values, respectively. In this way, extreme samples are retained for model training while increasing the proportion of other samples in the model.

[0093] At this point, the prediction model for the simplified pool repair time has been trained.

[0094] In another embodiment, such as Figure 2 As shown, after determining the model parameters, the support vector machine model is used as the simplified pool repair time prediction model, which also includes:

[0095] S300: Make predictions on the simplified pool repair time prediction model. In response to the prediction relative error value being greater than the preset range, adjust the kernel function parameters and update the simplified pool repair time prediction model.

[0096] The prediction model for the simplified pool repair time includes:

[0097] S310: Input at least one set of environmental parameter data from the test set into the simplified pool repair time prediction model to obtain the prediction results corresponding to the environmental parameter data;

[0098] S320: Perform inverse normalization on the prediction results to obtain the corresponding prediction duration;

[0099] Inputting the prepared test samples into the function model determined above will yield the corresponding predicted value of the thin pool repair time. However, this value is not the true time prediction value. Since the original sample data was normalized during the previous execution of the technical solution, the predicted time output must be denormalized to obtain the predicted values ​​of the number of processor cores, processor frequency, memory size, memory frequency, storage pool capacity, written data volume, deduplication rate, and thin pool repair time in write mode corresponding to the test samples.

[0100] S330: Obtain the prediction relative error value between the prediction duration and time data corresponding to the environmental parameter data, and correct the simplified pool repair duration prediction model based on the prediction relative error value.

[0101] Adjust kernel function parameters and update the simplified pool repair duration prediction model, including:

[0102] S340: Starting from determining the kernel function parameters using K-fold cross-validation, obtain the simplified pool repair duration prediction model again.

[0103] The repair time of the simplified pool is affected by a variety of factors, some of which are even random. Predictive models cannot account for all factors, so there will be some error between the predicted and actual values. If the error is too large, the implementation of the predictive model becomes meaningless and fails to achieve its initial intended purpose. Therefore, it is necessary to verify the prediction results.

[0104] In one embodiment, the Forecasting Relative Error (FRE) is used to test the prediction accuracy of the simplified pool repair duration prediction model. FRE is calculated as follows: FRE = [(y i′-y i ) / y i ]×100%, where y i ′ and y i , i = 1, 2, ..., representing the predicted and actual values, respectively, and i = 1, 2, ..., representing the data samples of each test set. A positive FRE value indicates that the predicted value is higher than the actual simplified pool repair time; a negative FRE value indicates that the predicted value is lower than the actual simplified pool repair time; the smaller the absolute value of FRE, the higher the accuracy of the prediction and the more accurate the prediction of the simplified pool repair time.

[0105] In one embodiment, if 90% of the FRE values ​​are within the error range (e.g., the predicted value differs from the actual repair time of the simplified pool by 1 minute, with the predicted value in minutes), then the model's prediction accuracy is considered to meet the requirements; otherwise, the reasons need to be analyzed, the kernel function parameters adjusted, and the model relearned.

[0106] During the correction process, we return to the K-fold validation stage, incrementing the value of K by 1. The larger K is, the more data is input into the training set each time, and the smaller the model bias. However, the larger K is, the greater the correlation between the selected training set and the previous ones, which may lead to overfitting and slow convergence.

[0107] By implementing the above model optimization method, the prediction accuracy of the simplified pool repair time prediction model is further improved.

[0108] In another embodiment, such as Figure 3 As shown, the device for generating a prediction model for the simplified pool repair time includes:

[0109] The dataset acquisition module is used to acquire training set data and test set data for the repair time.

[0110] The model building module is used to build a support vector machine model related to the number of categories of environmental parameters. The model parameters of the support vector machine model are determined based on the training set data, and the support vector machine model with determined model parameters is used as the simplified pool repair time prediction model. The model parameters include: kernel function parameters, weight values ​​and bias values.

[0111] In another embodiment, the prediction model generation apparatus further includes:

[0112] The model optimization module is used to predict the simplified pool repair time prediction model. In response to the prediction relative error value being greater than the preset range, the kernel function parameters are adjusted and the simplified pool repair time prediction model is updated.

[0113] In another embodiment, such as Figure 4 As shown, the method for predicting the repair time of the simplified pool includes:

[0114] S1: Obtain the environmental parameters to be predicted;

[0115] S2: Normalize the environmental parameters to be predicted;

[0116] S3: Input the normalized environmental parameters to be predicted into the simplified pool repair time prediction model to obtain the model prediction result. The simplified pool repair time prediction model is generated according to the above-mentioned simplified pool repair time prediction model generation method.

[0117] S4: Perform inverse normalization on the model prediction results to obtain the simplified pool repair prediction duration.

[0118] By implementing the simplified pool repair time prediction model generation method, apparatus and prediction method disclosed in the embodiments of this application, accurate simplified pool repair time prediction can be obtained; the convergence speed of the model is improved by associating the kernel function with the number of samples, and the training error is reduced by associating the number of cross-validation times with the number of samples; in terms of parameter optimization, the model is trained with two sets of training data, one containing extreme value group data and the other not containing extreme value group data, to obtain the corresponding weight values ​​and bias values, and the average value is taken as the final weight and bias, thereby increasing the proportion of extreme value samples in the classification model.

[0119] Specific limitations regarding the device for generating the prediction model for the reduced pool repair time can be found in the limitations on the method for generating the prediction model for the reduced pool repair time described above, and will not be repeated here. Each module in the aforementioned device for generating the prediction model for the reduced pool repair time can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0120] All of the above-mentioned optional technical solutions can be combined in any way to form optional embodiments of the present invention, and will not be described in detail here.

[0121] Example 1

[0122] Methods for generating prediction models for simplified pool repair time include:

[0123] S100: Obtain the training set data and test set data for the repair duration;

[0124] S110: Collect N sets of original sample data on repair time as the original dataset; where N is a positive integer and N≥2; each set of original sample data includes: M environmental parameter data and 1 time data; the environmental parameters include: number of processor cores, processor frequency, memory size, memory frequency, storage pool capacity, amount of data written, deduplication rate, write mode, and repair time;

[0125] S120: Normalize the original dataset to obtain the corresponding normalized dataset;

[0126] S130: Select X groups of data from the normalized dataset as the training set data, and use the remaining NX groups of data as the test set data; where X is a positive integer, and X... <N。

[0127] S200: Establish a support vector machine model related to the number of categories of environmental parameters, determine the model parameters of the support vector machine model based on the training set data, and use the support vector machine model with determined model parameters as the simplified pool repair duration prediction model; wherein, the model parameters include: kernel function parameters, weight values ​​and bias values.

[0128] S210: Use K-fold cross-validation to determine the kernel function parameters; where the kernel function is a radial basis function and K is a positive integer;

[0129] S220: Obtain weight and bias values ​​based on the radial basis function kernel function, kernel function parameters, and training set data.

[0130] The kernel function parameters are determined using K-fold cross-validation, specifically including:

[0131] S211: Associate the reciprocal of the initial value of the kernel function parameter with the number of categories of the environment parameter;

[0132] S212: Correlate the number of folds in the K-fold cross-validation method with the number of categories of environmental parameters;

[0133] S213: Randomly divide the training dataset into K mutually exclusive sub-training sets, train the support vector machine model K times, and obtain the process values ​​of K kernel function parameters;

[0134] S214: Calculate the average value of the K kernel function parameter process values ​​to obtain the kernel function parameters.

[0135] The weight and bias values ​​are obtained based on the radial basis function kernel function, kernel function parameters, and training set data, specifically including:

[0136] S221: Substitute the kernel function parameters into the support vector machine model, and train the support vector machine model with the training set data to obtain the first weight and the first bias;

[0137] S222: Remove the data of the group containing the extreme value samples from the training set data to obtain the second training set data, and use the training set data to train the support vector machine model to obtain the second weight and the second bias;

[0138] S223: Take the average of the first weight and the second weight as the weight value, and take the average of the first deviation and the second deviation as the deviation value.

[0139] By implementing the simplified pool repair time prediction model generation method, apparatus and prediction method disclosed in the embodiments of this application, an accurate prediction of the simplified pool repair time can be obtained.

[0140] Example 2

[0141] Methods for generating prediction models for simplified pool repair time include:

[0142] S100: Obtain the training set data and test set data for the repair duration;

[0143] S110: Collect N sets of original sample data on repair time as the original dataset; where N is a positive integer and N≥2; each set of original sample data includes: M environmental parameter data and 1 time data; the environmental parameters include: number of processor cores, processor frequency, memory size, memory frequency, storage pool capacity, amount of data written, deduplication rate, write mode, and repair time;

[0144] S120: Normalize the original dataset to obtain the corresponding normalized dataset;

[0145] S130: Select X groups of data from the normalized dataset as the training set data, and use the remaining NX groups of data as the test set data; where X is a positive integer, and X... <N。

[0146] S200: Establish a support vector machine model related to the number of categories of environmental parameters, determine the model parameters of the support vector machine model based on the training set data, and use the support vector machine model with determined model parameters as the simplified pool repair duration prediction model; wherein, the model parameters include: kernel function parameters, weight values ​​and bias values.

[0147] S210: Use K-fold cross-validation to determine the kernel function parameters; where the kernel function is a radial basis function and K is a positive integer;

[0148] S220: Obtain weight and bias values ​​based on the radial basis function kernel function, kernel function parameters, and training set data.

[0149] The kernel function parameters are determined using K-fold cross-validation, specifically including:

[0150] S211: Associate the reciprocal of the initial value of the kernel function parameter with the number of categories of the environment parameter;

[0151] S212: Correlate the number of folds in the K-fold cross-validation method with the number of categories of environmental parameters;

[0152] S213: Randomly divide the training dataset into K mutually exclusive sub-training sets, train the support vector machine model K times, and obtain the process values ​​of K kernel function parameters;

[0153] S214: Calculate the average value of the K kernel function parameter process values ​​to obtain the kernel function parameters.

[0154] The weight and bias values ​​are obtained based on the radial basis function kernel function, kernel function parameters, and training set data, specifically including:

[0155] S221: Substitute the kernel function parameters into the support vector machine model, and train the support vector machine model with the training set data to obtain the first weight and the first bias;

[0156] S222: Remove the data of the group containing the extreme value samples from the training set data to obtain the second training set data, and use the training set data to train the support vector machine model to obtain the second weight and the second bias;

[0157] S223: Take the average of the first weight and the second weight as the weight value, and take the average of the first deviation and the second deviation as the deviation value.

[0158] S300: Make predictions on the simplified pool repair time prediction model. In response to the prediction relative error value being greater than the preset range, adjust the kernel function parameters and update the simplified pool repair time prediction model.

[0159] S310: Input at least one set of environmental parameter data from the test set into the simplified pool repair time prediction model to obtain the prediction results corresponding to the environmental parameter data;

[0160] S320: Perform inverse normalization on the prediction results to obtain the corresponding prediction duration;

[0161] S330: Obtain the prediction relative error value between the prediction duration and time data corresponding to the environmental parameter data, and correct the simplified pool repair duration prediction model based on the prediction relative error value.

[0162] S340: Starting from determining the kernel function parameters using K-fold cross-validation, obtain the simplified pool repair duration prediction model again.

[0163] By linking the kernel function with the number of samples, the convergence speed of the model is improved, and by linking the number of cross-validations with the number of samples, the training error is reduced. In terms of parameter optimization, the model is trained with two sets of training data, one containing extreme value groups and the other not, to obtain the corresponding weight and bias values. The average value of each set is then taken as the final weight and bias, which increases the proportion of extreme value samples in the classification model.

[0164] Example 3

[0165] like Figure 3 As shown, the device for generating a prediction model for the simplified pool repair time includes:

[0166] The dataset acquisition module is used to acquire training set data and test set data for the repair time.

[0167] The model building module is used to build a support vector machine model related to the number of categories of environmental parameters. The model parameters of the support vector machine model are determined based on the training set data, and the support vector machine model with determined model parameters is used as the simplified pool repair time prediction model. The model parameters include: kernel function parameters, weight values ​​and bias values.

[0168] Example 4

[0169] Based on Embodiment 3, the device for generating a prediction model for the simplified pool repair time further includes:

[0170] The model optimization module is used to predict the simplified pool repair time prediction model. In response to the prediction relative error value being greater than the preset range, the kernel function parameters are adjusted and the simplified pool repair time prediction model is updated.

[0171] Example 5

[0172] like Figure 4 As shown, the methods for predicting the repair time of the simplified pool include:

[0173] S1: Obtain the environmental parameters to be predicted;

[0174] S2: Normalize the environmental parameters to be predicted;

[0175] S3: Input the normalized environmental parameters to be predicted into the simplified pool repair time prediction model to obtain the model prediction result. The simplified pool repair time prediction model is generated according to the above-mentioned simplified pool repair time prediction model generation method.

[0176] S4: Perform inverse normalization on the model prediction results to obtain the simplified pool repair prediction duration.

[0177] The prediction model for the simplified pool repair time is generated based on the prediction model generation method for the simplified pool repair time, specifically including:

[0178] S100: Obtain the training set data and test set data for the repair duration;

[0179] S110: Collect N sets of original sample data on repair time as the original dataset; where N is a positive integer and N≥2; each set of original sample data includes: M environmental parameter data and 1 time data; the environmental parameters include: number of processor cores, processor frequency, memory size, memory frequency, storage pool capacity, amount of data written, deduplication rate, write mode, and repair time;

[0180] S120: Normalize the original dataset to obtain the corresponding normalized dataset;

[0181] S130: Select X groups of data from the normalized dataset as the training set data, and use the remaining NX groups of data as the test set data; where X is a positive integer, and X... <N。

[0182] S200: Establish a support vector machine model related to the number of categories of environmental parameters, determine the model parameters of the support vector machine model based on the training set data, and use the support vector machine model with determined model parameters as the simplified pool repair duration prediction model; wherein, the model parameters include: kernel function parameters, weight values ​​and bias values.

[0183] S210: Use K-fold cross-validation to determine the kernel function parameters; where the kernel function is a radial basis function and K is a positive integer;

[0184] S220: Obtain weight and bias values ​​based on the radial basis function kernel function, kernel function parameters, and training set data.

[0185] The kernel function parameters are determined using K-fold cross-validation, specifically including:

[0186] S211: Associate the reciprocal of the initial value of the kernel function parameter with the number of categories of the environment parameter;

[0187] S212: Correlate the number of folds in the K-fold cross-validation method with the number of categories of environmental parameters;

[0188] S213: Randomly divide the training dataset into K mutually exclusive sub-training sets, train the support vector machine model K times, and obtain the process values ​​of K kernel function parameters;

[0189] S214: Calculate the average value of the K kernel function parameter process values ​​to obtain the kernel function parameters.

[0190] The weight and bias values ​​are obtained based on the radial basis function kernel function, kernel function parameters, and training set data, specifically including:

[0191] S221: Substitute the kernel function parameters into the support vector machine model, and train the support vector machine model with the training set data to obtain the first weight and the first bias;

[0192] S222: Remove the data of the group containing the extreme value samples from the training set data to obtain the second training set data, and use the training set data to train the support vector machine model to obtain the second weight and the second bias;

[0193] S223: Take the average of the first weight and the second weight as the weight value, and take the average of the first deviation and the second deviation as the deviation value.

[0194] S300: Make predictions on the simplified pool repair time prediction model. In response to the prediction relative error value being greater than the preset range, adjust the kernel function parameters and update the simplified pool repair time prediction model.

[0195] S310: Input at least one set of environmental parameter data from the test set into the simplified pool repair time prediction model to obtain the prediction results corresponding to the environmental parameter data;

[0196] S320: Perform inverse normalization on the prediction results to obtain the corresponding prediction duration;

[0197] S330: Obtain the prediction relative error value between the prediction duration and time data corresponding to the environmental parameter data, and correct the simplified pool repair duration prediction model based on the prediction relative error value.

[0198] S340: Starting from determining the kernel function parameters using K-fold cross-validation, obtain the simplified pool repair duration prediction model again.

[0199] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program loaded on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from memory, or installed from ROM. When the computer program is executed by an external processor, it performs the functions defined in the methods of embodiments of this application.

[0200] It should be noted that the computer-readable medium in the embodiments of this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In the embodiments of this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the embodiments of this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (Radio Frequency), etc., or any suitable combination thereof.

[0201] The aforementioned computer-readable medium may be included in the aforementioned server; or it may exist independently and not assembled into the server. The aforementioned computer-readable medium carries one or more programs that, when executed by the server, cause the server to: in response to detecting that the peripheral mode of the terminal is not activated, acquire the frame rate of the application on the terminal; when the frame rate meets the screen-off condition, determine whether the user is acquiring the terminal's screen information; and in response to the determination that the user is not acquiring the terminal's screen information, control the screen to enter an immediate dimming mode.

[0202] Computer program code for performing the operations of the embodiments of this application can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0203] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and relevant parts can be referred to the descriptions in the method embodiments. The systems and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0204] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

[0205] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting the repair time of a simplified pool, characterized in that, The method includes: Obtain the environmental parameters to be predicted; The environmental parameters to be predicted are normalized; The normalized environmental parameters to be predicted are input into the simplified pool repair time prediction model to obtain the model prediction results. The model prediction results are inversely normalized to obtain the simplified pool repair prediction time; The prediction model for the simplified pool repair time is generated according to the following process: Obtain the training set data and test set data for the repair duration; A support vector machine (SVM) model related to the number of categories of environmental parameters is established. The model parameters of the SVM model are determined based on the training set data, and the SVM model with the determined model parameters is used as the prediction model for the repair time of the simplified storage pool. The model parameters include: kernel function parameters, weight values, and bias values. The environmental parameters include at least: number of processor cores, processor clock speed, memory size, memory clock speed, storage pool capacity, amount of data written, deduplication rate, write mode, and repair time. The simplified pool repair time prediction model is used to make predictions. If the relative error value of the prediction is greater than a preset range, the kernel function parameters are adjusted and the simplified pool repair time prediction model is updated. The step of establishing a support vector machine (SVM) model related to the number of categories of environmental parameters, determining the model parameters of the SVM model based on the training set data, and using the SVM model with the determined model parameters as the prediction model for the simplified pool repair time includes: The kernel function parameters are determined using K-fold cross-validation; wherein the kernel function is a radial basis function and K is a positive integer; The weight values ​​and bias values ​​are obtained based on the radial basis function kernel function, the kernel function parameters, and the training set data. The step of determining the kernel function parameters using K-fold cross-validation includes: The reciprocal of the initial value of the kernel function parameter is correlated with the number of categories of the environment parameter; The number of folds in the K-fold cross-validation method is correlated with the number of categories of the environmental parameter; The training dataset is randomly divided into K mutually exclusive sub-training sets, and the support vector machine model is trained K times to obtain the process values ​​of K kernel function parameters. The kernel function parameters are obtained by averaging the process values ​​of the K kernel function parameters.

2. The method for predicting the repair time of a simplified pool according to claim 1, characterized in that, The training set data and test set data for obtaining the repair time include: N sets of original sample data on repair time are collected as the original dataset; where N is a positive integer and N≥2; each set of original sample data includes: M environmental parameter data and 1 time data; the original dataset is normalized to obtain the corresponding normalized dataset; X groups of data are selected from the normalized dataset as the training set, and the remaining NX groups of data are used as the test set; where X is a positive integer, and X... <N。 3. The method for predicting the repair time of a simplified pool according to claim 1, characterized in that, The step of obtaining the weight values ​​and bias values ​​based on the radial basis function kernel function, the kernel function parameters, and the training set data includes: Substitute the kernel function parameters into the support vector machine model, and train the support vector machine model with the training set data to obtain the first weight and the first bias. The data of the group containing the extreme value samples are removed from the training set data to obtain the second training set data, and the support vector machine model is trained with the training set data to obtain the second weight and the second bias. The average of the first weight and the second weight is taken as the weight value, and the average of the first deviation and the second deviation is taken as the deviation value.

4. The method for predicting the repair time of the simplified pool according to claim 1, characterized in that, The prediction of the simplified pool repair time prediction model includes: Input at least one set of environmental parameter data from the test set data into the simplified pool repair time prediction model to obtain the prediction result corresponding to the environmental parameter data; The prediction results are inversely normalized to obtain the corresponding prediction duration; Obtain the error value between the predicted duration corresponding to the environmental parameter data and a time data point, and correct the simplified pool repair duration prediction model based on the error value.

5. The method for predicting the repair time of a simplified pool according to claim 1, characterized in that, Adjusting the kernel function parameters and updating the simplified pool repair duration prediction model includes: Starting from the kernel function parameters determined by the K-fold cross-validation method, the simplified pool repair time prediction model is obtained again.

Images