A storage performance prediction calculation method, device, electronic device and storage medium for rural road slice files

Abstract

The present invention provides a method, apparatus, electronic device and storage medium for predicting and calculating the storage performance of rural road slice files, including: obtaining a historical file storage feature sequence before the time point to be predicted; calculating a stationary actual storage sequence and a differencing order based on a time series stationary processing method; determining a target smoothing coefficient corresponding to the time point to be predicted based on a smoothing coefficient calculation method; obtaining an outlier-free actual linear storage sequence corresponding to the historical actual linear storage sequence based on an outlier detection method; calculating a predicted linear storage value based on a predicted linear storage value calculation method; calculating a predicted non-linear storage value based on a predicted non-linear storage value calculation method; adding the predicted linear storage value and the predicted non-linear storage value to obtain a predicted storage value. In this way, the accuracy of predicting the storage requirements of rural road slice files can be improved, thereby significantly reducing the CPU occupancy rate and I / O reading pressure on the server side.

Description

Technical Field

[0001] The present invention relates to the technical field of slice file storage, and in particular to a method, device, electronic device and storage medium for predicting and calculating the storage performance of rural road slice files. Background Art

[0002] When browsing and replacing and updating rural road slice data files through the browser side, the user requests slice files of adjacent positions and index targets through moving or zooming operations, so that the server side needs to frequently process a large number of storage and access requests for slice files. The existing processing methods usually directly store and load slice files based on actual access requirements, and fail to effectively predict the storage and loading requirements of files, resulting in a high CPU occupancy rate and low I / O reading efficiency on the server side in a distributed storage environment. For this reason, the present invention adopts a dynamic double exponential smoothing prediction model based on time series to realize the storage prediction of slice files and optimize the performance of the distributed storage system. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide a method, device, electronic device and storage medium for predicting and calculating the storage performance of rural road slice files, which can improve the accuracy of predicting the storage requirements of slice files, thus significantly reducing the CPU occupancy rate and I / O reading pressure on the server side, and further optimizing the overall performance of the distributed storage system to meet the requirements for the storage efficiency of slice files in high-frequency access scenarios.

[0004] In a first aspect, an embodiment of the present invention provides a method for predicting and calculating the storage performance of rural road slice files, including: obtaining a historical file storage feature sequence before a time point to be predicted; the historical file storage feature sequence includes a historical actual storage sequence and a historical actual linear storage sequence; calculating a stationary actual storage sequence and a difference order corresponding to the historical actual storage sequence based on a preset time series stationary processing method; determining a target smoothing coefficient corresponding to the time point to be predicted based on the stationary actual storage sequence, the difference order and a preset smoothing coefficient calculation method; performing outlier detection on the historical actual linear storage sequence based on a preset outlier detection method to obtain an outlier-free actual linear storage sequence; calculating a predicted linear storage value corresponding to the time point to be predicted based on the outlier-free actual linear storage sequence, the target smoothing coefficient and a preset predicted linear storage value calculation method; calculating a predicted non-linear storage value corresponding to the time point to be predicted based on the historical file storage feature sequence, the target smoothing coefficient and a preset predicted non-linear storage value calculation method; adding the predicted linear storage value and the predicted non-linear storage value to obtain a predicted storage value corresponding to the time point to be predicted; the predicted storage value includes a predicted slice file size and a predicted slice file storage time.

[0005] Further, after the step of adding the predicted linear storage value and the predicted non-linear storage value to obtain the predicted storage value corresponding to the time point to be predicted, the method further includes: saving the predicted storage value, the predicted linear storage value, the predicted non-linear storage value, and the target smoothing coefficient to the historical file storage feature sequence; obtaining the target slice file from the storage node; wherein, the size of the target slice file is the size of the predicted slice file; and storing the target slice file to the server based on the storage time of the predicted slice file.

[0006] Further, the step of calculating the stationary actual storage sequence and the difference order corresponding to the historical actual storage sequence based on the pre-set time series stationary processing method includes: S1: determining whether the historical actual storage sequence meets the pre-set stationary judgment criterion; S2: if the historical actual storage sequence does not meet the stationary judgment criterion, performing difference processing on the historical actual storage sequence based on the pre-set difference detection method to obtain the updated historical actual storage sequence, and incrementing the current difference order by 1; wherein, the initial value of the current difference order is 0; S3: repeating steps S1 - S2 until the historical actual storage sequence meets the stationary judgment criterion; S4: determining the historical actual storage sequence that meets the stationary judgment criterion as the stationary actual storage sequence, and determining the current difference order as the difference order.

[0007] Further, the step of determining the target smoothing coefficient corresponding to the time point to be predicted based on the stationary actual storage sequence, the difference order, and the pre-set smoothing coefficient calculation method includes: calculating the autocorrelation function and the partial autocorrelation function corresponding to the historical actual storage sequence based on the historical actual storage sequence and the pre-set time series modeling method; determining the range of values of the autoregressive term order and the range of values of the moving average term order based on the autocorrelation function and the partial autocorrelation function; obtaining the sample number and variance corresponding to the historical actual storage sequence; screening the range of values of the autoregressive term order and the range of values of the moving average term order based on the sample number, the variance, and the pre-set model order optimization criterion to determine the target autoregressive term order and the target moving average term order; and calculating the target smoothing coefficient corresponding to the time point to be predicted based on the difference order, the target autoregressive term order, the target moving average term order, and the smoothing coefficient calculation method.

[0008] Further, the steps of detecting outliers in the historical actual linear storage sequence based on a preset outlier detection method to obtain an outlier-free actual linear storage sequence include: dividing the historical actual linear storage sequence into sliding windows based on a preset window size to obtain at least one load interval; calculating the interval eigenvalue corresponding to each load interval, and determining the confidence interval corresponding to each load interval based on the interval eigenvalue; the interval eigenvalue includes the mean and the mean square deviation; calculating the distance radius corresponding to each confidence interval; calculating the adjacent interval correlation between the distance radius of the confidence interval and the distance radius of the previous confidence interval; determining whether the adjacent interval correlation is greater than a preset critical value; if so, determining the load interval corresponding to the confidence interval as an abnormal load interval; taking the data points not within the confidence interval corresponding to the abnormal load interval in the abnormal load interval as outliers; and deleting the outliers in each abnormal load interval to obtain an outlier-free actual linear storage sequence.

[0009] Further, the method for calculating the predicted linear storage value is as follows: ; where is the predicted linear storage value corresponding to the time point to be predicted, is the target smoothing coefficient, is the actual storage value corresponding to the previous time point of the time point to be predicted, is the predicted linear storage value corresponding to the previous time point of the time point to be predicted.

[0010] Further, the steps of calculating the predicted non-linear storage value corresponding to the time point to be predicted based on the historical file storage feature sequence, the target smoothing coefficient, and a preset predicted non-linear storage value calculation method include: calculating the residual value corresponding to the time point to be predicted based on the historical file storage feature sequence and the following residual value calculation formula: ; where is the residual value, is the actual storage value corresponding to the previous time point of the time point to be predicted, is the predicted linear storage value corresponding to the previous time point; calculating the predicted non-linear storage value corresponding to the time point to be predicted based on the residual value, the target smoothing coefficient, and the following predicted non-linear storage value calculation method: , where is the predicted non-linear storage value, is the random error value, represents weighted smoothing calculation of the residual value sequence corresponding to the residual value, is the smoothing coefficient.

[0011] Second aspect, an embodiment of the present invention provides a storage performance prediction calculation device for rural road slice files, including: an acquisition module, configured to acquire a historical file storage feature sequence before a time point to be predicted; the historical file storage feature sequence includes a historical actual storage sequence and a historical actual linear storage sequence; a stationary processing module, configured to calculate a stationary actual storage sequence corresponding to the historical actual storage sequence and a difference order based on a preset time series stationary processing method; a smoothing coefficient calculation module, configured to determine a target smoothing coefficient corresponding to the time point to be predicted based on the stationary actual storage sequence, the difference order, and a preset smoothing coefficient calculation method; an outlier processing module, configured to perform outlier detection on the historical actual linear storage sequence based on a preset outlier detection method to obtain an outlier-free actual linear storage sequence; a predicted linear storage value calculation module, configured to calculate a predicted linear storage value corresponding to the time point to be predicted based on the outlier-free actual linear storage sequence, the target smoothing coefficient, and a preset predicted linear storage value calculation method; a predicted non-linear storage value calculation module, configured to calculate a predicted non-linear storage value corresponding to the time point to be predicted based on the historical file storage feature sequence, the target smoothing coefficient, and a preset predicted non-linear storage value calculation method; a predicted storage value calculation module, configured to add the predicted linear storage value and the predicted non-linear storage value to obtain a predicted storage value corresponding to the time point to be predicted; the predicted storage value includes a predicted slice file size and a predicted slice file storage time.

[0012] Third aspect, an embodiment of the present invention provides an electronic device, including a memory and a processor, where a computer program is stored on the memory and can run on the processor, and when the processor executes the computer program, the method described above is implemented.

[0013] Fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, and the program code causes the processor to execute the method described above.

[0014] An embodiment of the present invention provides a method, device, electronic device, and storage medium for predicting the storage performance of rural road slice files, including: obtaining a historical file storage feature sequence before a time point to be predicted; the historical file storage feature sequence includes a historical actual storage sequence and a historical actual linear storage sequence; based on a preset time series stationary processing method, calculating a stationary actual storage sequence and a difference order corresponding to the historical actual storage sequence; based on the stationary actual storage sequence, the difference order, and a preset smoothing coefficient calculation method, determining a target smoothing coefficient corresponding to the time point to be predicted; based on a preset outlier detection method, performing outlier detection on the historical actual linear storage sequence to obtain an outlier-free actual linear storage sequence; based on the outlier-free actual linear storage sequence, the target smoothing coefficient, and a preset predicted linear storage value calculation method, calculating a predicted linear storage value corresponding to the time point to be predicted; based on the historical file storage feature sequence, the target smoothing coefficient, and a preset predicted non-linear storage value calculation method, calculating a predicted non-linear storage value corresponding to the time point to be predicted; adding the predicted linear storage value and the predicted non-linear storage value to obtain a predicted storage value corresponding to the time point to be predicted; the predicted storage value includes a predicted slice file size and a predicted slice file storage time. In this way, by performing stationarity processing, outlier detection, and prediction of linear and non-linear storage values on the historical actual storage sequence, the rules of storage data can be mined from different levels, and future storage requirements can be accurately estimated, thereby avoiding prediction errors caused by outlier interference. Through sliding window division, confidence interval calculation, and interval correlation analysis, data in abnormal load intervals can be effectively removed, reducing the impact of abnormal data on model prediction, and thus making the prediction results more accurate. Through in-depth analysis and precise processing of the storage sequence, the prediction efficiency of storage performance can be improved on the premise of ensuring accuracy, and storage resource waste and network load in practical applications can be reduced. Through in-depth analysis and precise processing of the storage sequence, the prediction efficiency of storage performance can be improved on the premise of ensuring accuracy, and storage resource waste and network load in practical applications can be reduced.

[0015] Other features and advantages of the present invention will be described in the following specification, and some of them will become obvious from the specification, or can be understood by implementing the present invention. The objectives and other advantages of the present invention are achieved and obtained by the structures specifically pointed out in the specification, claims, and drawings.

[0016] To make the above objectives, features, and advantages of the present invention more obvious and understandable, the following specific preferred embodiments are given in conjunction with the accompanying drawings and described in detail as follows. Brief Description of the Drawings

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1 Flowchart of the storage performance prediction calculation method for rural road slice files provided by the embodiments of the present invention;

[0019] Figure 2 Schematic diagram of the sliding window detection method provided by the embodiments of the present invention;

[0020] Figure 3 Schematic diagram of the storage performance prediction calculation device for rural road slice files provided by the embodiments of the present invention;

[0021] Figure 4 Schematic diagram of the structure of an electronic device provided by the embodiments of the present invention.

[0022] Icons: 1 - Acquisition module; 2 - Smooth processing module; 3 - Smoothing coefficient calculation module; 4 - Outlier processing module; 5 - Prediction linear storage value calculation module; 6 - Prediction non - linear storage value calculation module; 7 - Prediction storage value calculation module; 301 - Processor; 302 - Memory; 303 - Bus; 304 - Communication interface. Specific embodiments

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions of the present invention with reference to the accompanying drawings. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0024] In the storage and management of rural road slice files, since these slice files carry spatial geographical coordinate information, when users perform moving or zooming operations on the browser side, they need to frequently request slice files at adjacent positions from the server. With the continuous increase of rural road data, the server side needs to process a large number of slice file requests, and these files are usually stored in a distributed manner on different server nodes. With the increase in access frequency, how to effectively manage and store these massive slice files has become an urgent problem to be solved.

[0025] Currently, common file storage methods usually rely on the single exponential smoothing prediction model for storage performance prediction. This model can predict future storage requirements based on historical storage data. However, the single exponential smoothing prediction model has a serious drawback. That is, for time series with a linear change trend, the prediction results often have large deviations and there is a certain time lag, usually about 1000 - 1500 milliseconds. This lag causes the system to be unable to respond promptly to frequent requests from the browser side, affecting the user experience, and resulting in high load on the server CPU and low I / O reading efficiency.

[0026] In order to meet the high-frequency access requirements of rural road slice files on the browser side, while reducing the CPU occupancy rate of each server in the distributed storage environment and improving the I / O reading efficiency, this application provides a storage performance prediction calculation method, device, electronic device, and storage medium for rural road slice files, which can effectively overcome the lag and deviation problems of the single exponential smoothing prediction model, more accurately predict storage requirements, and thus optimize system performance, reduce server load, and improve the overall file storage and access efficiency while meeting high-frequency user access.

[0027] For the convenience of understanding this embodiment, the embodiments of the present invention will be introduced in detail below.

[0028] Embodiment 1:

[0029] Figure 1 It is a flowchart of the storage performance prediction calculation method for rural road slice files provided by the embodiments of the present invention.

[0030] Referring to Figure 1 , the storage performance prediction calculation method for rural road slice files includes:

[0031] Step S101, obtaining the historical file storage feature sequence before the time point to be predicted; the historical file storage feature sequence includes the historical actual storage sequence and the historical actual linear storage sequence.

[0032] Here, the historical file storage feature sequence is the actually recorded file storage data, reflecting the specific characteristics of past file storage.

[0033] The historical file storage feature sequence includes the historical actual storage sequence, the historical actual linear sequence, the historical actual non-linear sequence, the historical predicted storage sequence, the historical predicted linear storage sequence, the historical predicted non-linear storage sequence, and the historical smoothing coefficient sequence, etc.

[0034] The time point to be predicted is t, the historical actual storage sequence before the time point to be predicted is , the historical actual linear sequence , the historical actual non-linear sequence , historical prediction storage sequence , historical prediction linear storage sequence , historical prediction non - linear storage sequence 。

[0035] The actual storage sequence is the sum of the actual linear storage sequence and the actual non - linear storage sequence, , and the predicted storage sequence is the sum of the predicted linear storage sequence and the predicted non - linear storage sequence, 。

[0036] The historical prediction storage sequence includes at least one predicted storage value. The predicted storage value includes the predicted slice file size and the predicted slice file storage time. The predicted slice file size and the predicted slice file storage time are predicted separately, and the prediction methods are the same. The prediction of the slice file size can optimize the allocation of storage space, improve the utilization efficiency of storage resources, and avoid overloading of storage devices or waste of resources. The prediction of the slice file storage time helps to improve the storage operation efficiency, optimize the I / O performance, and ensure the efficient and rapid completion of data storage operations.

[0037] The historical actual storage sequence includes the historical actual slice file size sequence and the historical actual slice file storage time sequence. The historical actual linear sequence includes the historical actual linear slice file size sequence and the historical actual linear slice file storage time sequence. The historical actual non - linear sequence includes the historical actual non - linear slice file size sequence and the historical actual non - linear slice file storage time sequence. The historical prediction storage sequence includes the historical prediction slice file size sequence and the historical prediction slice file storage time sequence. The historical prediction linear storage sequence includes the historical prediction linear slice file size sequence and the historical prediction linear slice file storage time sequence. The historical prediction non - linear storage sequence includes the historical prediction non - linear slice file size sequence and the historical prediction non - linear slice file storage time sequence.

[0038] Step S102: Calculate the stationary actual storage sequence and the difference order corresponding to the historical actual storage sequence based on a pre - set time - series stationary processing method.

[0039] Here, if the historical actual storage sequence does not meet the stationarity requirement, it is transformed into a stationary sequence through differencing operations. The difference order d represents how many differencing operations are required to make the data stationary. Through the stationary processing, the unnecessary fluctuations in the historical actual storage sequence are removed, and the long - term trend of the stored data can be identified more accurately.

[0040] In one embodiment, the steps of step S102 include the following steps S1 - S4.

[0041] Step S1: Determine whether the historical actual storage sequence meets the preset stationarity judgment criterion.

[0042] Here, the method of calculating the mean and variance of the historical actual storage sequence can be used to determine whether the historical actual storage sequence is stationary. If the mean and variance of the historical actual storage sequence remain unchanged over time, it is determined that the historical actual storage sequence meets the preset stationarity judgment criteria, that is, the historical actual storage sequence is stationary.

[0043] It is also possible to calculate the autocorrelation function (ACF) and partial autocorrelation function (PACF) of the time series and plot their graphs to observe whether there is an obvious trend or seasonality in the time series. If the ACF plot shows significant autocorrelation at long time lags or the PACF plot does not decay rapidly, it indicates that the historical actual storage sequence time series does not meet the preset stationarity judgment criteria, that is, the historical actual storage sequence is not stationary.

[0044] Step S2: If the historical actual storage sequence does not meet the stationarity judgment criteria, perform differencing on the historical actual storage sequence based on a preset differencing detection method to obtain an updated historical actual storage sequence, and increment the current differencing order by 1; where the initial value of the current differencing order is 0.

[0045] Specifically, assume a data sequence regarding the storage of rural road slice files, representing the file storage size within a certain period of time. Table 1 below is the original historical actual storage sequence.

[0046] Table 1 Original historical actual storage sequence

[0047]

[0048] Based on Table 1 above, it is determined that the mean and variance of the historical actual storage sequence gradually increase over time, showing an obvious trend change and not meeting the stationarity judgment criteria.

[0049] Perform first-order differencing, that is, calculate the difference between adjacent time points: . Table 2 below is the result table after first-order differencing.

[0050] Table 2 Result table after first-order differencing

[0051]

[0052] After first-order differencing, the new sequence (i.e., the differenced sequence) becomes: {2, 2, 4, 4, 5, 5}.

[0053] Determine that the current differencing order is 1.

[0054] Step S3: Repeat Steps S1 - S2 until the historical actual storage sequence meets the stationarity judgment criteria.

[0055] Here, the stationarity of the differenced sequence is judged again. If the mean and variance of the differenced sequence are stable, and the ACF graph shows no obvious trend, it indicates that the sequence is stationary. If the sequence is still non-stationary, second-order differencing needs to be continued.

[0056] Specifically, the differenced sequence {2, 2, 4, 4, 5, 5} has become stationary enough.

[0057] Step S4: Determine the historical actual storage sequence that meets the stationarity judgment criterion as the stationary actual storage sequence, and determine the current differencing order as the differencing order.

[0058] Specifically, determine the differenced sequence {2, 2, 4, 4, 5, 5} as the stationary actual storage sequence, and the differencing order d is the current differencing order, that is, d = 1.

[0059] Step 103, based on the stationary actual storage sequence, the differencing order, and the preset smoothing coefficient calculation method, determine the target smoothing coefficient corresponding to the time point to be predicted.

[0060] In one embodiment, the steps of step S103 include the following steps S201 - S205.

[0061] Step S201, based on the historical actual storage sequence and the preset time series modeling method, calculate the autocorrelation function and partial autocorrelation function corresponding to the historical actual storage sequence.

[0062] Here, the autocorrelation function (ACF) measures the correlation between the current value and the first n lagged values, and is used to understand the continuity and trend of the time series. If the ACF graph shows a sharp drop to zero after a certain lag period, it indicates that the sequence has short-term correlation, and the order p of the autoregressive term should be limited before this point.

[0063] The partial autocorrelation function (PACF) measures the direct correlation between the current value and the first n lagged values, excluding the influence of all intermediate moments. The PACF can help determine the order p of the autoregressive model. If the PACF graph shows a sharp drop to zero after a certain lag period, it indicates that the data points before this lag period have the greatest influence, and the order q of the moving average term may be limited before this point.

[0064] Step S202, based on the autocorrelation function and the partial autocorrelation function, determine the value range of the autoregressive term order and the value range of the moving average term order.

[0065] Here, the truncation of the ACF / PACF plot means that if the ACF / PACF plot rapidly approaches zero after a certain lag period, it indicates that there is no significant correlation after that lag period. The tailing of the ACF / PACF plot means that if the ACF / PACF plot shows that as the lag period increases, the autocorrelation gradually weakens but does not rapidly approach zero. If the ACF / PACF plot is in a tailing state, higher-order autoregressive terms p or moving average terms q are required to capture this dependence. If both the ACF and PACF are in a truncated state, the values corresponding to the start of the truncation are taken as the maximum values of p and q.

[0066] The range of values for the autoregressive term order p is determined by analyzing the characteristics of the PACF plot. Usually, the number of lag periods after which the PACF plot rapidly drops to zero at the truncation position is the candidate value for the autoregressive term order. Among them, p always has a non-zero value.

[0067] The range of values for the moving average term order q is determined by analyzing the characteristics of the ACF plot. If the ACF plot rapidly approaches zero after a certain lag period, the number of lags before that lag period can be used as the candidate value for the moving average term order.

[0068] For example, if the ACF plot approaches zero after lag 5 and the PACF plot approaches zero after lag 3, it can be considered that the range of values for the autoregressive term order p is from 1 to 3, and the range of values for the moving average term order q is from 1 to 5.

[0069] Step S203, obtain the sample size and variance corresponding to the historical actual storage sequence.

[0070] Here, the sample size n is the number of small file load data corresponding to the historical actual storage sequence obtained in advance.

[0071] The variance σ2 measures the degree of data fluctuation. A sequence with a larger variance may contain more fluctuation information.

[0072] Step S204, based on the sample size, variance, and the pre-set model order optimization criterion, screen the range of values for the autoregressive term order and the range of values for the moving average term order to determine the target autoregressive term order and the target moving average term order.

[0073] Here, the pre-set model order optimization criterion is shown in the following formula (1):

[0074] (1)

[0075] where n is the sample size, σ2 is the variance, p is the autoregressive term order, q is the moving average term order, is the model scoring function.

[0076] Evaluate the effects of different combinations of p and q based on a pre-set model order optimization criterion, and select the optimal order combination.

[0077] By calculating and comparing the f(p,q) values of different order combinations, select the smallest value as the final model order to avoid overfitting or underfitting. If the f(p,q) value is too large, it indicates that the model complexity is too high, and p and q may need to be reduced. If the f(p,q) value is small, it indicates that this combination has a good fitting effect.

[0078] Determine the p that satisfies the minimum f(p,q) value as the target autoregressive term order, and determine the q that satisfies the minimum f(p,q) value as the target moving average term order.

[0079] Step S205, calculate the target smoothing coefficient corresponding to the time point to be predicted based on the difference order, target autoregressive term order, target moving average term order, and smoothing coefficient calculation method.

[0080] Here, the smoothing coefficient calculation method is α = f(p,d,q), where α is the smoothing coefficient.

[0081] The target coefficient is calculated through the following formula (2):

[0082] (2)

[0083] Among them, the larger n is, the higher the accuracy of the data, so the smoothing coefficient should also be increased accordingly. σ 2 The larger it is, the stronger the data fluctuation is, and the smoothing coefficient is correspondingly reduced to reduce the influence brought by the fluctuation. The larger P is, it means that the historical data has a strong influence on the current value, and the smoothing coefficient should be correspondingly reduced. The larger D is, it indicates that the data is more complex, and the smoothing coefficient should be correspondingly reduced. The larger P is, it means that the historical information of the error has a greater influence, and the smoothing coefficient should be correspondingly reduced.

[0084] Step S104, based on a pre-set outlier detection method, detect outliers in the historical actual linear storage sequence to obtain a historical actual linear storage sequence without outliers.

[0085] Here, in the historical actual linear storage sequence, there may be some outliers caused by various reasons (such as equipment failures, data acquisition errors, etc.). These outliers are removed through the preset outlier detection method to obtain a historical actual linear storage sequence without outliers. The existence of outliers will interfere with the prediction results, so they need to be removed in advance to ensure the accuracy of the prediction.

[0086] The pre-set outlier detection method is the sliding window detection method. Refer to Figure 2, divide the rural road slice file sequence into multiple intervals, then extract the eigenvalue of each interval for the first determination, and then make the second determination through the correlation between adjacent intervals, and finally process the outliers.

[0087] In one embodiment, the steps of step S104 include the following steps S301 - S308.

[0088] Step S301, based on a preset window size, perform a sliding window division on the historical actual linear storage sequence to obtain at least one load interval.

[0089] Here, the preset window size m is set in advance according to the actual situation. The preset window size m determines the time length within each load interval. For example, if m = 5, then each load interval contains data of 5 consecutive time points.

[0090] Based on the preset window size, divide the historical actual storage sequence into multiple load intervals W. .

[0091] Step S302, calculate the interval eigenvalue corresponding to each load interval, and determine the confidence interval corresponding to each load interval based on the interval eigenvalue; the interval eigenvalue includes the mean and the mean square deviation.

[0092] Here, for each load interval, calculate its interval eigenvalue, and then use these eigenvalues to determine the confidence interval of each load interval.

[0093] The calculation method of the interval eigenvalue is shown in the following formula (3):

[0094] (3)

[0095] Where, is the mean in the interval eigenvalue, is the variance in the interval eigenvalue.

[0096] Based on the mean and the standard deviation, combined with the normal distribution, construct the confidence interval as shown in the following formula (4):

[0097] (4)

[0098] Where, is the upper bound of the confidence interval, is the lower bound of the confidence interval, Z represents a random variable subject to the 𝑁(0,1) distribution, represents the confidence level, = 0.05.

[0099] Step S303, calculate the distance radius corresponding to each confidence interval.

[0100] Here, the distance radius of the confidence interval is half of the difference between the upper and lower bounds of the confidence interval, as shown in the following formula (5):

[0101] (5)

[0102] where the distance radius of the confidence interval represents the uncertainty of the load interval, that is, the range of data fluctuations.

[0103] Step S304, calculate the correlation of adjacent intervals between the distance radius of the confidence interval and the distance radius of the previous confidence interval.

[0104] Here, the correlation of adjacent intervals can be quantified by calculating the difference in the distance radius of two adjacent intervals through the following formula (6):

[0105] (6)

[0106] where, is the distance radius of the current confidence interval, is the distance radius of the previous confidence interval. A larger difference indicates a greater change in the characteristics of adjacent intervals, which may be caused by outliers.

[0107] Step S305, determine whether the correlation of adjacent intervals is greater than a preset critical value.

[0108] Here, the preset critical value is set in advance according to the actual situation and can be set to 0.3.

[0109] If , it indicates that there is an outlier in the j-th interval.

[0110] Step S306, if so, determine that the load interval corresponding to the confidence interval is an abnormal load interval.

[0111] Here, if the correlation of adjacent intervals exceeds the preset critical value, then the load interval is considered an abnormal load interval.

[0112] Step S307, for the abnormal load interval, the data points that are not within the confidence interval corresponding to the abnormal load interval are outliers.

[0113] Here, after determining the abnormal load interval, the abnormal data points within this interval are removed. Abnormal data points refer to those points that do not fall within the confidence interval of this interval.

[0114] For each load interval, if some data points fall outside the confidence interval (i.e., not within and If it is between them, it is regarded as an outlier and deleted from the data.

[0115] Step S308: Delete the outliers in each abnormal load interval to obtain an actual linear storage sequence without outliers.

[0116] Here, all the data points marked as abnormal are deleted, so as to obtain an actual linear storage sequence without outliers that does not contain outliers.

[0117] Step S105: Calculate the predicted linear storage value corresponding to the time point to be predicted based on the actual linear storage sequence without outliers, the target smoothing coefficient, and the preset calculation method of the predicted linear storage value.

[0118] Here, the historical actual storage sequence without outliers is used, combined with the historical predicted linear storage sequence and the target smoothing coefficient, and the storage demand at the time point to be predicted is predicted based on the preset calculation method of the linear storage value.

[0119] In one embodiment, in step S105, the calculation method of the predicted linear storage value is as shown in the following formula (7):

[0120] (7)

[0121] Wherein, is the predicted linear storage value corresponding to the time point to be predicted, is the target smoothing coefficient, is the actual storage value corresponding to the previous time point of the time point to be predicted, is the predicted linear storage value corresponding to the previous time point of the time point to be predicted.

[0122] Among them, the actual storage value corresponding to the previous time point of the time point to be predicted is obtained through the stationary actual storage sequence, and the predicted linear storage value corresponding to the previous time point of the time point to be predicted is obtained through the actual linear storage sequence without outliers.

[0123] Step S106: Calculate the predicted non-linear storage value corresponding to the time point to be predicted based on the historical file storage feature sequence, the target smoothing coefficient, and the preset calculation method of the predicted non-linear storage value.

[0124] In one embodiment, the steps of step S106 include:

[0125] Calculate the residual value corresponding to the time point to be predicted based on the historical actual storage sequence, the historical predicted linear storage sequence, and the following residual value calculation formula (8):

[0126] (8)

[0127] Wherein, is the residual value, is the actual storage value corresponding to the previous time point of the time point to be predicted, is the predicted non-linear storage value corresponding to the previous time point.

[0128] Here, the residual value is the difference between the actual storage value and the predicted storage value at the current moment. By calculating the residual value, we can obtain the error between the prediction and the actual storage at the previous moment.

[0129] Based on the residual value, the target smoothing coefficient, and the following method for calculating the predicted non-linear storage value (9), calculate the predicted non-linear storage value corresponding to the time point to be predicted:

[0130] (9)

[0131] Among them, is the predicted non-linear storage value, is the random error value, represents performing weighted smoothing calculation on the residual value sequence corresponding to the residual value, is the smoothing coefficient.

[0132] The residual value sequence includes at least one historical residual value, and the smoothing coefficient is used to adjust the influence of each historical residual value on the current predicted value. is used to perform weighted smoothing calculation on the historical residual value sequence to optimize the prediction accuracy.

[0133] Here, the random error value is as shown in formula (10):

[0134] d (10)

[0135] Among them, is the pre-set small file load sequence, which is used to represent the load data at each time point. Δ d represents the difference operation, which is used to calculate the change of time series data.

[0136] Step S107, add the predicted linear storage value and the predicted non-linear storage value to obtain the predicted storage value corresponding to the time point to be predicted; the predicted storage value includes the predicted slice file size and the predicted slice file storage time.

[0137] Here, add the results of linear prediction and non-linear prediction to obtain the final storage prediction value.

[0138] In one embodiment, after the steps of step S107, the method further includes the following steps S401 - S403.

[0139] Step S401, save the predicted storage value, the predicted linear storage value, the predicted non-linear storage value, and the target smoothing coefficient to the historical file storage feature sequence.

[0140] Step S402: Obtain the target slice file from the storage node; wherein, the size of the target slice file is the predicted slice file size.

[0141] Here, the storage node refers to the device or system in the data storage system used to store and manage slice files.

[0142] The target slice file refers to the specific slice file that needs to be obtained when requested at the browser end. The size of the target slice file is determined based on the predicted storage value. According to the predicted storage value, the system can dynamically adjust the storage capacity and size of the file to better adapt to the actual storage requirements.

[0143] Step S403: Store the target slice file to the server based on the predicted slice file storage time.

[0144] Here, based on the predicted storage time of the slice file, that is, considering factors such as the size of the file and the timeliness of storage, the target slice file is stored in the server. During storage, the file storage policy will be adjusted according to the dynamically predicted results to ensure the efficiency of the storage process and the high efficiency of file access. After being stored in the server, the load of the storage node can be optimized, thereby reducing the latency of I / O operations and effectively reducing the CPU occupancy rate on the server side.

[0145] An embodiment of the present invention provides a method for predicting and calculating the storage performance of rural road slice files, including: obtaining a historical file storage feature sequence before the time point to be predicted; the historical file storage feature sequence includes a historical actual storage sequence and a historical actual linear storage sequence; calculating a stationary actual storage sequence and a difference order corresponding to the historical actual storage sequence based on a preset time series stationary processing method; determining a target smoothing coefficient corresponding to the time point to be predicted based on the stationary actual storage sequence, the difference order, and a preset smoothing coefficient calculation method; performing outlier detection on the historical actual linear storage sequence based on a preset outlier detection method to obtain an outlier-free actual linear storage sequence; calculating a predicted linear storage value corresponding to the time point to be predicted based on the outlier-free actual linear storage sequence, the target smoothing coefficient, and a preset predicted linear storage value calculation method; calculating a predicted non-linear storage value corresponding to the time point to be predicted based on the historical file storage feature sequence, the target smoothing coefficient, and a preset predicted non-linear storage value calculation method; adding the predicted linear storage value and the predicted non-linear storage value to obtain a predicted storage value corresponding to the time point to be predicted; the predicted storage value includes a predicted slice file size and a predicted slice file storage time. In this way, by performing stationarity processing, outlier detection, and prediction of linear and non-linear storage values on the historical actual storage sequence, the rules of storage data can be mined from different levels, and the future storage requirements can be accurately estimated, thus avoiding prediction errors caused by outlier interference. By dividing the sliding window, calculating the confidence interval, and analyzing the interval correlation, the data in the abnormal load interval can be effectively removed, reducing the impact of abnormal data on the model prediction, and thus making the prediction result more accurate. Through in-depth analysis and precise processing of the storage sequence, the prediction efficiency of the storage performance can be improved on the premise of ensuring accuracy, and the waste of storage resources and network load in practical applications can be reduced. Through in-depth analysis and precise processing of the storage sequence, the prediction efficiency of the storage performance can be improved on the premise of ensuring accuracy, and the waste of storage resources and network load in practical applications can be reduced.

[0146] Embodiment 2:

[0147] Figure 3 It is a schematic diagram of a device for predicting and calculating the storage performance of rural road slice files provided by an embodiment of the present invention.

[0148] Referring to Figure 3 , the device for predicting and calculating the storage performance of rural road slice files includes:

[0149] An acquisition module 1, configured to obtain a historical file storage feature sequence before the time point to be predicted; the historical file storage feature sequence includes a historical actual storage sequence and a historical actual linear storage sequence;

[0150] The stationary processing module 2 is used to calculate the stationary actual storage sequence and the difference order corresponding to the historical actual storage sequence based on a preset time series stationary processing method;

[0151] The smoothing coefficient calculation module 3 is used to determine the target smoothing coefficient corresponding to the time point to be predicted based on the stationary actual storage sequence, the difference order, and a preset smoothing coefficient calculation method;

[0152] The outlier processing module 4 is used to detect outliers in the historical actual linear storage sequence based on a preset outlier detection method to obtain an actual linear storage sequence without outliers;

[0153] The predicted linear storage value calculation module 5 is used to calculate the predicted linear storage value corresponding to the time point to be predicted based on the actual linear storage sequence without outliers, the target smoothing coefficient, and a preset predicted linear storage value calculation method;

[0154] The predicted non-linear storage value calculation module 6 is used to calculate the predicted non-linear storage value corresponding to the time point to be predicted based on the historical file storage feature sequence, the target smoothing coefficient, and a preset predicted non-linear storage value calculation method;

[0155] The predicted storage value calculation module 7 is used to add the predicted linear storage value and the predicted non-linear storage value to obtain the predicted storage value corresponding to the time point to be predicted; the predicted storage value includes the predicted slice file size and the predicted slice file storage time.

[0156] In one embodiment, the predicted storage value calculation module 7 is further used for:

[0157] Save the predicted storage value, the predicted linear storage value, the predicted non-linear storage value, and the target smoothing coefficient to the historical file storage feature sequence.

[0158] Obtain the target slice file from the storage node; wherein, the size of the target slice file is the predicted slice file size.

[0159] Store the target slice file to the server based on the predicted slice file storage time.

[0160] In one embodiment, the stationary processing module 2 is further used for:

[0161] S1: Determine whether the historical actual storage sequence meets a preset stationarity judgment criterion.

[0162] S2: If the historical actual storage sequence does not meet the stationarity judgment criterion, perform differencing processing on the historical actual storage sequence based on a preset differencing detection method to obtain an updated historical actual storage sequence, and increment the current difference order by 1; wherein, the initial value of the current difference order is 0.

[0163] S3: Repeat steps S1 - S2 until the historical actual storage sequence meets the stationarity judgment criterion.

[0164] S4: Determine the historical actual storage sequence that meets the stationarity judgment criterion as the stationary actual storage sequence, and determine the current differencing order as the differencing order.

[0165] In one embodiment, the smoothing coefficient calculation module 3 is further configured to:

[0166] Based on the historical actual storage sequence and a preset time series modeling method, calculate the autocorrelation function and partial autocorrelation function corresponding to the historical actual storage sequence.

[0167] Based on the autocorrelation function and partial autocorrelation function, determine the range of autoregressive term orders and the range of moving average term orders.

[0168] Obtain the sample size and variance corresponding to the historical actual storage sequence.

[0169] Based on the sample size, variance, and a preset model order optimization criterion, screen the range of autoregressive term orders and the range of moving average term orders to determine the target autoregressive term order and the target moving average term order.

[0170] Based on the differencing order, target autoregressive term order, target moving average term order, and smoothing coefficient calculation method, calculate the target smoothing coefficient corresponding to the time point to be predicted.

[0171] In one embodiment, the outlier processing module 4 is further configured to:

[0172] Based on a preset window size, perform a sliding window partition on the historical actual linear storage sequence to obtain at least one load interval.

[0173] Calculate the interval eigenvalue corresponding to each load interval, and determine the confidence interval corresponding to each load interval based on the interval eigenvalue; the interval eigenvalue includes the mean and the mean square deviation.

[0174] Calculate the distance radius corresponding to each confidence interval.

[0175] Calculate the adjacent interval correlation between the distance radius of the confidence interval and the distance radius of the previous confidence interval.

[0176] Judge whether the adjacent interval correlation is greater than a preset critical value.

[0177] If so, determine the load interval corresponding to the confidence interval as an abnormal load interval.

[0178] For the abnormal load interval, the data points that are not within the confidence interval corresponding to the abnormal load interval are outliers.

[0179] Delete the outliers in each abnormal load interval to obtain an actual linear storage sequence without outliers.

[0180] In one embodiment, the predicted linear storage value calculation module 5 is further configured to:

[0181] The method for calculating the predicted linear storage value is as follows:

[0182]

[0183] Wherein, is the predicted linear storage value corresponding to the time point to be predicted, is the target smoothing coefficient, is the actual storage value corresponding to the previous time point of the time point to be predicted, is the predicted linear storage value corresponding to the previous time point of the time point to be predicted.

[0184] In one embodiment, the predicted non-linear storage value calculation module 6 is further configured to:

[0185] Calculate the residual value corresponding to the time point to be predicted based on the historical file storage feature sequence and the following residual value calculation formula:

[0186]

[0187] Wherein, is the residual value, is the actual storage value corresponding to the previous time point of the time point to be predicted, is the predicted linear storage value corresponding to the previous time point.

[0188] Calculate the predicted non-linear storage value corresponding to the time point to be predicted based on the residual value, the target smoothing coefficient and the following predicted non-linear storage value calculation method:

[0189]

[0190] Wherein, is the predicted non-linear storage value, is the random error value, represents weighted smoothing calculation of the residual value sequence corresponding to the residual value, is the smoothing coefficient.

[0191] The embodiment of the present invention provides a storage performance prediction calculation device for rural road slice files, which can provide accurate predicted storage values through means such as stationary processing, outlier detection, and dynamic calculation of smoothing coefficients, avoid deviations and lags in traditional prediction models, significantly improve the accuracy and efficiency of storage prediction, reduce server load, and optimize the utilization of storage resources.

[0192] An embodiment of the present application further provides an electronic device, such as Figure 4 shown, which is a schematic structural diagram of the electronic device. The electronic device includes a processor 301 and a memory 302. The memory 302 stores computer-executable instructions that can be executed by the processor 301. The processor 301 executes the computer-executable instructions to implement the above-mentioned rural road connectivity path calculation method.

[0193] In Figure 4 the illustrated embodiment, the electronic device further includes a bus 303 and a communication interface 304. Among them, the processor 301, the communication interface 304, and the memory 302 are connected through the bus 303.

[0194] Among them, the memory 302 may include a high-speed random access memory (RAM, Random Access Memory), and may also include a non-volatile memory, such as at least one disk memory. Through at least one communication interface 304 (which can be wired or wireless), a communication connection is established between the system network element and at least one other network element. The Internet, wide area network, local area network, metropolitan area network, etc. can be used. The bus 303 can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 303 can be divided into an address bus, a data bus, a control bus, etc. For the sake of representation, Figure 4 only a bidirectional arrow is used in

[0195] The processor 301 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in the processor 301 or the instructions in the form of software. The above-mentioned processor 301 may be a general-purpose processor, including a central processing unit (CPU for short), a network processor (NP for short), etc.; it may also be a digital signal processor (DSP for short), an application specific integrated circuit (ASIC for short), a field-programmable gate array (FPGA for short), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in the embodiments of the present application can be directly embodied as being executed and completed by the hardware decoding processor, or executed and completed by the combination of the hardware and software modules in the decoding processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. This storage medium is located in the memory 302, and the processor 301 reads the information in the memory 302 and combines its hardware to complete the steps of the method for predicting the storage performance of the rural road slice file in the foregoing embodiments.

[0196] In the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. In addition, the terms "first", "second", "third" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0197] Finally, it should be noted that the above-described embodiments are only specific embodiments of the present invention, used to illustrate the technical solutions of the present invention, rather than limiting it. The protection scope of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that any technician familiar with the technical field of the present invention can still modify the technical solutions recorded in the foregoing embodiments, or can easily think of changes, or perform equivalent replacements on some of the technical features; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims described herein.

Claims

1. A storage performance prediction calculation method for rural road slice files, characterized in that: include: Obtain the historical file storage feature sequence before the time point to be predicted; The historical file storage feature sequence includes a historical actual storage sequence and a historical actual linear storage sequence; Based on a preset time series stationary processing method, a stationary actual storage sequence and a difference order corresponding to the historical actual storage sequence are calculated; Determine the target smoothing coefficient corresponding to the time point to be predicted based on the stationary actual storage sequence, the difference order and a preset smoothing coefficient calculation method; Based on a preset outlier detection method, outlier detection is performed on the historical actual linear storage sequence to obtain an actual linear storage sequence without outliers; Calculate the predicted linear storage value corresponding to the time point to be predicted based on the actual linear storage sequence without abnormal values, the target smoothing coefficient and a preset predicted linear storage value calculation method; Calculate the predicted nonlinear storage value corresponding to the time point to be predicted based on the historical file storage feature sequence, the target smoothing coefficient and a preset predicted nonlinear storage value calculation method; The predicted linear storage value and the predicted nonlinear storage value are added to obtain the predicted storage value corresponding to the time point to be predicted; the predicted storage value includes the predicted slice file size and the predicted slice file storage time.

2. According to claim 1, a storage performance prediction calculation method for rural road slice files is characterized in that: After the step of adding the predicted linear storage value and the predicted nonlinear storage value to obtain the predicted storage value corresponding to the time point to be predicted, the method further includes: Saving the predicted storage value, the predicted linear storage value, the predicted nonlinear storage value and the target smoothing coefficient to the history file storage feature sequence; Obtain a target slice file from a storage node; wherein the size of the target slice file is the predicted slice file size; Based on the predicted slice file storage time, the target slice file is stored in a server.

3. According to claim 1, a storage performance prediction calculation method for rural road slice files is characterized in that: The step of calculating the stationary actual storage sequence and the difference order corresponding to the historical actual storage sequence based on the preset time series stationary processing method includes: S1: Determine whether the historical actual storage sequence meets the preset stability judgment standard; S2: If the historical actual storage sequence does not meet the stationarity judgment standard, perform differential processing on the historical actual storage sequence based on a preset differential detection method to obtain an updated historical actual storage sequence, and add 1 to the current differential order; wherein the initial value of the current differential order is 0; S3: repeat steps S1-S2 until the historical actual storage sequence meets the stability judgment standard; S4: Determine the historical actual storage sequence that meets the stationarity judgment standard as the stationary actual storage sequence, and determine the current differential order as the differential order.

4. The storage performance prediction calculation method of a rural road slice file according to claim 3 is characterized in that: The step of determining the target smoothing coefficient corresponding to the time point to be predicted based on the stationary actual storage sequence, the difference order and the preset smoothing coefficient calculation method includes: Based on the historical actual storage sequence and a preset time series modeling method, an autocorrelation function and a partial autocorrelation function corresponding to the historical actual storage sequence are calculated; Based on the autocorrelation function and the partial autocorrelation function, determining the value range of the autoregressive term order and the value range of the moving average term order; Obtaining the number of samples and variance corresponding to the historical actual storage sequence; Based on the sample size, the variance and a preset model order optimization criterion, the autoregressive term order value range and the moving average term order value range are screened to determine a target autoregressive term order and a target moving average term order; Based on the difference order, the target autoregressive term order, the target moving average term order and the smoothing coefficient calculation method, the target smoothing coefficient corresponding to the time point to be predicted is calculated.

5. The storage performance prediction calculation method of a rural road slice file according to claim 1 is characterized in that: The step of performing outlier detection on the historical actual linear storage sequence based on a preset outlier detection method to obtain an actual linear storage sequence without outliers includes: Based on a preset window size, the historical actual linear storage sequence is divided into sliding windows to obtain at least one load interval; Calculating an interval characteristic value corresponding to each of the load intervals, and determining a confidence interval corresponding to each of the load intervals based on the interval characteristic value; the interval characteristic value includes a mean and a mean square error; Calculate the distance radius corresponding to each of the confidence intervals; Calculating the adjacent interval correlation between the distance radius of the confidence interval and the distance radius of the previous confidence interval; Determining whether the adjacent interval correlation is greater than a preset critical value; If yes, determining that the load interval corresponding to the confidence interval is an abnormal load interval; In the abnormal load interval, data points that are not within the confidence interval corresponding to the abnormal load interval are regarded as abnormal values; The abnormal value in each abnormal load interval is deleted to obtain the actual linear storage sequence without abnormal values.

6. The storage performance prediction calculation method of a rural road slice file according to claim 1 is characterized in that: The predicted linear storage value calculation method is as follows: in, is the predicted linear storage value corresponding to the time point to be predicted, is the target smoothing coefficient, is the actual storage value corresponding to the previous time point of the time point to be predicted, The predicted linear storage value corresponding to the previous time point of the time point to be predicted.

7. The storage performance prediction calculation method of a rural road slice file according to claim 1 is characterized in that: The step of calculating the predicted nonlinear storage value corresponding to the time point to be predicted based on the historical file storage feature sequence, the target smoothing coefficient and a preset predicted nonlinear storage value calculation method comprises: Based on the characteristic sequence stored in the historical file and the following residual value calculation formula, the residual value corresponding to the time point to be predicted is calculated: in, is the residual value, is the actual storage value corresponding to the previous time point of the time point to be predicted, The predicted linear storage value corresponding to the previous time point; Based on the residual value, the target smoothing coefficient and the following predicted nonlinear storage value calculation method, the predicted nonlinear storage value corresponding to the time point to be predicted is calculated: in, storing a value for the predicted nonlinearity, is the random error value, Indicates that weighted smoothing calculation is performed on the residual value sequence corresponding to the residual value. is the smoothing coefficient.

8. A storage performance prediction and calculation device for rural road slice files, characterized in that: include: An acquisition module is used to acquire a feature sequence stored in historical files before the time point to be predicted; The historical file storage feature sequence includes a historical actual storage sequence and a historical actual linear storage sequence; A smooth processing module, used for calculating the smooth actual storage sequence and the difference order corresponding to the historical actual storage sequence based on a preset time series smooth processing method; A smoothing coefficient calculation module, used to determine the target smoothing coefficient corresponding to the time point to be predicted based on the stationary actual storage sequence, the difference order and a preset smoothing coefficient calculation method; An outlier processing module, used for performing outlier detection on the historical actual linear storage sequence based on a preset outlier detection method to obtain an actual linear storage sequence without outliers; A predicted linear storage value calculation module, used to calculate the predicted linear storage value corresponding to the time point to be predicted based on the actual linear storage sequence without abnormal values, the target smoothing coefficient and a preset predicted linear storage value calculation method; A predicted nonlinear storage value calculation module, used to calculate the predicted nonlinear storage value corresponding to the time point to be predicted based on the historical file storage feature sequence, the target smoothing coefficient and a preset predicted nonlinear storage value calculation method; The predicted storage value calculation module is used to add the predicted linear storage value and the predicted nonlinear storage value to obtain the predicted storage value corresponding to the time point to be predicted; the predicted storage value includes the predicted slice file size and the predicted slice file storage time.

9. An electronic device, characterized in that: It includes a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and is characterized in that when the processor executes the computer program, it implements the storage performance prediction calculation method of the rural road slice file described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, which, when executed by a processor, implement the storage performance prediction calculation method for rural road slice files as described in any one of claims 1-7.

Images