CN116909754A - Mobile terminal CPU short-term load prediction method based on combined model and related equipment - Google Patents
Mobile terminal CPU short-term load prediction method based on combined model and related equipment Download PDFInfo
- Publication number
- CN116909754A CN116909754A CN202311174944.6A CN202311174944A CN116909754A CN 116909754 A CN116909754 A CN 116909754A CN 202311174944 A CN202311174944 A CN 202311174944A CN 116909754 A CN116909754 A CN 116909754A
- Authority
- CN
- China
- Prior art keywords
- load
- time sequence
- cpu
- linear
- load time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 85
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 claims abstract description 38
- 238000012549 training Methods 0.000 claims abstract description 9
- 238000000354 decomposition reaction Methods 0.000 claims description 19
- 230000006870 function Effects 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 14
- 230000009467 reduction Effects 0.000 claims description 13
- 238000011156 evaluation Methods 0.000 claims description 4
- 238000002790 cross-validation Methods 0.000 claims description 3
- 238000012417 linear regression Methods 0.000 claims description 3
- 238000007476 Maximum Likelihood Methods 0.000 claims description 2
- 238000010276 construction Methods 0.000 claims description 2
- 230000010354 integration Effects 0.000 claims description 2
- 238000013507 mapping Methods 0.000 claims description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 14
- 230000008569 process Effects 0.000 description 14
- 238000010801 machine learning Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 5
- 230000008859 change Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 238000004590 computer program Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 238000007619 statistical method Methods 0.000 description 3
- 238000003491 array Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000001583 randomness test Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
- G06F9/505—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the load
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
- G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
Abstract
The application discloses a mobile terminal CPU short-term load prediction method and related equipment based on a combined model, which relate to the technical field of mobile terminal load prediction and mainly comprise the following core steps: constructing an ARIMA model, and inputting the linear load time sequence into the ARIMA model to obtain a linear predicted value of the load at the next moment of the linear load time sequence; obtaining a residual error of the load time sequence according to the CPU load time sequence and the linear load time sequence at the corresponding moment; establishing and training an SVR model, and inputting the residual error of the load time sequence into the SVR model to obtain a residual error predicted value of the load at the next moment; and obtaining a prediction result of the short-term load of the CPU of the mobile terminal according to the linear prediction value of the load at the next moment and the residual prediction value of the load at the next moment.
Description
Technical Field
The application relates to the technical field of mobile terminal load prediction, in particular to a mobile terminal CPU short-term load prediction method based on a combined model and related equipment.
Background
With the popularization and rapid development of mobile terminals, how to balance the CPU resource consumption of various applications on the terminals and schedule resources in different interaction scenarios in time becomes an important problem. The CPU load is a function of time, and the current common CPU load prediction method is to extract characteristic information of load in time and predict the load by using a time sequence method. Common methods are methods based on statistical analysis and methods based on machine learning. The method based on statistical analysis comprises a method based on exponential smoothing and moving average, and the model is used for trend prediction by combining historical data aiming at a stable linear sequence, so that the algorithm has strong interpretation, simple realization and small required calculation force and storage space. However, only the linear part of the time sequence is considered, the nonlinear part is not predicted, and part of the model precision is lost. The machine learning-based method mainly abstracts the time sequence prediction problem into a regression problem, constructs a sample data set, adopts a mode from time features to sample values, and learns association relations between the features and labels through supervised learning, thereby realizing the time sequence prediction. In the past decades, research proves that a machine learning method can well process nonlinear data, but most researches based on a machine learning algorithm are improved aiming at a specific single model, extracted features are incomplete, an updating mechanism is not considered, a time sequence with frequent fluctuation cannot be well adapted, meanwhile, prediction accuracy is pursued, and dependence of the model on resources is ignored.
Disclosure of Invention
Aiming at the defects in the prior art, the application combines the traditional method with the machine learning algorithm, adopts a lightweight ARIMA model and a machine learning model SVR method to respectively predict the linear and nonlinear components of the CPU load, can effectively improve the prediction accuracy, and simultaneously sets an updating mechanism to adapt to the terminal load change.
In order to achieve the above purpose, the present application provides the following technical solutions:
in a first aspect, the present application provides a method for predicting a short-term load of a CPU of a mobile terminal based on a combined model, including the steps of:
acquiring CPU load data of the mobile terminal, wherein the CPU load data at least comprises a CPU load time sequence;
performing wavelet noise reduction operation on the CPU load time sequence, and decomposing to obtain a linear load time sequence;
constructing an ARIMA model, and inputting the linear load time sequence into the ARIMA model to obtain a linear predicted value of the load at the next moment of the linear load time sequence;
obtaining a residual error of the load time sequence according to the CPU load time sequence and the linear load time sequence at the corresponding moment;
establishing and training an SVR model, and inputting the residual error of the load time sequence into the SVR model to obtain a residual error predicted value of the load at the next moment;
and obtaining a prediction result of the short-term load of the CPU of the mobile terminal according to the linear prediction value of the load at the next moment and the residual prediction value of the load at the next moment.
In a second aspect, the present application provides a mobile terminal CPU short-term load prediction system based on a combined model, which includes:
a data acquisition unit for acquiring CPU load data of the mobile terminal, wherein the CPU load data at least comprises a CPU load time sequence;
a data processing unit for performing the steps of:
performing wavelet noise reduction operation on the CPU load time sequence, and decomposing to obtain a linear load time sequence;
constructing an ARIMA model, and inputting the linear load time sequence into the ARIMA model to obtain a linear predicted value of the load at the next moment of the linear load time sequence;
obtaining a residual error of the load time sequence according to the CPU load time sequence and the linear load time sequence at the corresponding moment;
establishing and training an SVR model, and inputting the residual error of the load time sequence into the SVR model to obtain a residual error predicted value of the load at the next moment;
and obtaining a prediction result of the short-term load of the CPU of the mobile terminal according to the linear prediction value of the load at the next moment and the residual prediction value of the load at the next moment.
In a third aspect, the present application also provides an electronic device, including a processor and a memory;
the memory is used for storing programs;
the processor executes the program to implement the method as described above.
In a fourth aspect, the present application also provides a computer-readable storage medium storing a program for execution by a processor to implement a method as described above.
In a fifth aspect, the present application also provides a computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The computer instructions may be read from a computer-readable storage medium by a processor of a computer device, and executed by the processor, to cause the computer device to perform the foregoing method.
Compared with the prior art, the application has the beneficial effects that:
1. according to the application, the CPU load change of the mobile terminal is researched and predicted, and a dispatcher is informed in advance to dispatch the task to a proper CPU for operation, so that the method has important significance in improving the resource utilization efficiency, improving the user experience and enhancing the product competitiveness.
2. According to the application, the characteristics of non-stability and non-linearity of the CPU load data of the mobile terminal are considered, and the linear and non-linear factors in the time sequence are respectively predicted by adopting a mode of combining a classical ARIMA model and a SVR-based machine learning model, so that the effective prediction of the load condition of a period of time in the future through the historical data of the CPU load is realized, and the CPU resource calling strategy is improved.
3. According to the actual load time sequence, the order of the ARIMA model is automatically updated. According to the method, the ARIMA model is arranged, the model order can be adjusted in time according to a new load sequence sample, and the change of the load sequence is learned.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following description will briefly explain the drawings needed in the embodiments, and it is obvious that the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of steps of a method for predicting a short-term load of a CPU of a mobile terminal based on a combined model in an embodiment of the application;
FIG. 2 is a schematic diagram of a wavelet threshold denoising process in an embodiment of the present application;
FIG. 3 is a schematic diagram of a construction flow of an ARIMA model in an embodiment of the present application;
FIG. 4 is a schematic diagram of the linear predictor sequence generation of CPU load in an embodiment of the present application;
FIG. 5 is a schematic diagram of a system architecture in an embodiment of the application;
fig. 6 is a schematic diagram of an electronic device in an embodiment of the application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Examples:
it should be noted that the terms "first," "second," and the like in the description and the claims of the present application and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the application described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.
Example 1
The embodiment of the application combines the traditional method with the machine learning algorithm, adopts a lightweight ARIMA model and a machine learning model SVR method to respectively predict the linear and nonlinear components of the CPU load, can effectively improve the prediction accuracy, and simultaneously sets an updating mechanism to adapt to the load change of the terminal.
Referring to fig. 1, a method for predicting a short-term load of a CPU of a mobile terminal based on a combined model may include the steps of:
step 1: CPU load data of the mobile terminal is obtained, wherein the CPU load data at least comprises a CPU load time sequence.
In this step, the CPU load data of the mobile terminal may be obtained according to a preset time sliding window size, and specifically, the previous sliding CPU load data of the time to be predicted is taken. The sliding is a preset sliding window size, which indicates that the load value of the next moment is to be predicted according to the previous historical load values of the sliding. The value of sliding in the embodiment of the application is obtained through multiple tests on the CPU load data, and specific data can be specifically analyzed according to experience without special limitation.
Step 2: and carrying out wavelet noise reduction operation on the CPU load time sequence, and decomposing to obtain a linear load time sequence.
In this step, the useful signal usually shows a low frequency signal or is relatively stable, while the noise signal usually shows a high frequency signal, after wavelet transformation, the useful signal is concentrated on a few wavelet coefficients with a larger amplitude, and the white noise is dispersed on a large number of wavelet coefficients in the wavelet transformation domain with a smaller amplitude. Based on the above features, from the amplitude of the spectrum, the useful signal and noise can be separated. The embodiment of the application adopts a wavelet threshold denoising method to denoise the CPU load time sequence.
In some embodiments, the process of wavelet denoising operation is shown in fig. 2, and the steps for obtaining the denoised linear load time sequence are as follows:
step 21: and setting a wavelet basis function and a decomposition layer number j, and performing wavelet decomposition operation on the CPU load time sequence to obtain j high-frequency detail components and 1 low-frequency approximate component.
In the substep, proper wavelet functions and the number j of decomposition layers are selected to carry out wavelet decomposition on the CPU load time sequence, so as to obtain j decomposed high-frequency detail components and 1 low-frequency approximate component;
in particular, there are a plurality of wavelet basis functions, but none of them can achieve optimal decomposition effects for different types of signals. In the case of wavelet decomposition, the higher the decomposition level is, the more low-frequency components are removed, the more obvious the denoising effect is, but the degree of distortion is increased, so that the decomposition level is not too high for conservation. For more fluctuating CPU load time series, typically no more than 4 layers. To select a suitable number of decomposition levels to solve the contradiction between the signal-to-noise separation effect and the distortion of the reconstructed signal, a number of experiments can be performed according to the actual load time sequence, the denoising effect is compared by mean square error MSE (Mean Square Error, MSE) and signal-to-noise ratio (Signal Noise Ratio, SNR), and the optimal wavelet basis and decomposition level are determined. The formula for MSE and SNR is as follows:
wherein, the liquid crystal display device comprises a liquid crystal display device,representing the original load +.>Representing the estimated load after denoising. The smaller the MSE, the better, and the larger the SNR.
Step 22: and (5) denoising the j high-frequency detail components by using a hard threshold or soft threshold method.
In this sub-step, the wavelet coefficients obtained by the decomposition are thresholded. Selecting a hard threshold value or a soft threshold value method for processing the high-frequency detail components from the first layer to the j-th layer;
the hard thresholding method is to make the absolute value of the wavelet coefficient equal to or larger than a given threshold value, and to make the absolute value of the wavelet coefficient equal to or smaller than the given threshold value equal to or smaller than 0. Namely:
wherein, the liquid crystal display device comprises a liquid crystal display device,is wavelet coefficient +.>For threshold value->Is the wavelet coefficient after thresholding.
Specifically, in the soft threshold method, when the absolute value of the wavelet coefficient is greater than or equal to a given threshold value, the threshold value is subtracted from the wavelet coefficient; and when the value is smaller than the threshold value, the wavelet coefficient is set to zero. Namely:
wherein the sgn function represents a step function,is wavelet coefficient +.>For threshold value->Is the wavelet coefficient after thresholding.
In the embodiment of the application, the threshold selection algorithm is as follows:
wherein the Median function represents a return-to-intermediate function,for the detail coefficients of the first layer decomposition, +.>For data length +.>To decompose the layer number.
Step 23: for the part after the decomposition operation and the noise reduction operationAnd (3) carrying out reconstruction operation on the components, and obtaining the denoised linear load time sequence after integration operation.
In this sub-step, reconstruction is performed based on the j-th layer low frequency coefficient of wavelet decomposition and the high frequency coefficients of the first layer to the j-th layer, to obtain a denoised linear load time sequence.
Step 3: and constructing an ARIMA model, and inputting the linear load time sequence into the ARIMA model to obtain a linear predicted value of the load at the next moment of the linear load time sequence.
In this step, the linear load time series is inputThe method comprises the steps of inputting into an ARIMA model, automatically calculating the model order according to the AIC value minimum principle, selecting the optimal ARIMA model for prediction, obtaining the linear prediction value of the load at the next moment of the linear load time sequence, and recording as。
In some embodiments, the process of constructing the ARIMA model is shown in fig. 3, and the process of constructing the ARIMA model includes:
step 31: CPU historical load data of the mobile terminal is obtained, wherein the CPU load data at least comprises a CPU historical load time sequence.
Step 32: and performing decomposition operation, noise reduction operation and reconstruction operation on the historical load time sequence to obtain a denoised linear historical load time sequence.
Step 33: and performing stationarity check operation on the linear historical load time sequence, wherein if the linear historical load time sequence is a non-stationary sequence, performing differential operation, and if the linear historical load time sequence is a stationary sequence, performing randomness check operation to obtain a part with close correlation between sequence values of the linear historical load time sequence.
Step 34: and constructing an ARIMA model by using the linear historical load time sequence of the part, wherein the steps p and q of the model are determined by adopting a grid search method, all permutation and combination of p and q in a set range are obtained, the linear historical load time sequence is respectively fitted by using ARIMA (p and q) models with different steps, the parameter estimation is carried out by using a maximum likelihood estimation or a least square method, and the steps (p and q) when the minimum value is taken in a red pool information quantity criterion are selected, so that the optimal ARIMA model is obtained.
Specifically, as shown in fig. 3, in the stability check link, it is checked whether the linear load time series is stable. A graph inspection method and a method for constructing test statistics are adopted. And the graph inspection method is used for performing visual judgment according to the time sequence diagram and the autocorrelation diagram. The statistics may be selected according to actual needs, and embodiments of the present application are not specifically limited. If the sequence is a non-stationary sequence, the sequence needs to be stationary. The present embodiment adopts differential processing, and the formula thereof is as follows:
wherein, the liquid crystal display device comprises a liquid crystal display device,represents the original load value at time t, +.>Representing the data after the difference. This is a formula of "first order difference", and for the data after the first order difference, the next time data minus the previous time data is used again, called "second order difference", and so on.
In the randomness test link, the stable sequence is tested to ensure that the sequence has analyzability. For a purely random sequence, also known as a white noise sequence, there is no correlation between the various items of the sequence, and it is not meaningful to analyze such a sequence when the sequence is subject to random fluctuations that are completely unordered. Only sequences with certain influence on future development by the historical data due to the close correlation among the sequence values can the effective information in the historical data be mined to predict the future development of the sequences. For purely random sequences, it is generally checked by constructing statistics. The statistics may be selected according to actual needs, and embodiments of the present application are not specifically limited.
In the model identification link, the order p, q, d of the ARIMA model is determined. d is determined by the number of differences, and a first order difference is made, d=1, a second order difference, d=2, and so on. p is the number of autoregressive terms and q is the number of moving average terms. And (5) adopting grid search to obtain all permutation and combination of p and q in a given range. The ranges of p and q in the embodiments of the present application are obtained through multiple experiments and experiences, and are not specifically limited herein.
In the parameter estimation link, the order p, q, d of the ARIMA model is determined. d is determined by the number of differences, and a first order difference is made, d=1, a second order difference, d=2, and so on. p is the number of autoregressive terms and q is the number of moving average terms. And (5) adopting grid search to obtain all permutation and combination of p and q in a given range. The ranges of p and q in the embodiments of the present application are obtained through multiple experiments and experiences, and are not specifically limited herein.
In the AIC calculation step, according to the information quantity criterion (Akaike information criterion, AIC) of the red pool, the order (p, q) of the AIC when taking the minimum value is selected to obtain an optimal model, and the optimal model with the minimum AIC is selected in real time according to the load sequence by the model to adapt to the load change.
Specifically, the mathematical expression of the AIC criterion is:
wherein, the liquid crystal display device comprises a liquid crystal display device,likelihood function expressed as model, +.>Is the number of unknown parameters in the model.
Inputting the linear load time sequence into the optimal ARIMA model for prediction, and performing anti-differential processing on the output result to obtain a linear prediction value of the load at the next moment of the linear load time sequence, and recording the linear prediction value as。
The linear prediction value generation process of the load at the next moment is shown in fig. 4, and the specific process is as follows:
(1) acquiring the real values of the sliding historical CPU loads according to the preset sliding window size to obtain a historical load time sequence, and marking the historical load time sequence as X;
(2) processing X by using a wavelet noise reduction method to obtain a linear historical load time sequence after noise reduction, and marking the linear historical load time sequence as X';
(3) constructing an ARIMA model according to the steps 31 to 34;
(4) inputting the linear historical load time sequence X' into an ARIMA model for prediction, and performing inverse differential processing on an output result to obtain a linear prediction value of a load at the next moment;
(5) sliding the sliding window backwards according to the step length of 1, continuously updating the historical load time sequence, repeating the steps (1) to (4) to obtain a linear predictive value sequence of the load, and recording as。
Step 4: and obtaining a residual error of the load time sequence according to the CPU load time sequence and the linear load time sequence at the corresponding moment.
In this step, the predicted value of the linear load time series at the corresponding time is subtracted from the actual value of the CPU load time series to obtain the residual error of the load time series. As shown in the formula:
wherein, the liquid crystal display device comprises a liquid crystal display device,for CPU history load time series,/I>For loading a linear predictor sequence,/->Time series are loaded for the residual.
Step 5: and establishing and training an SVR model, and inputting the residual error of the load time sequence into the SVR model to obtain a residual error predicted value of the load at the next moment.
In the step, a CPU load residual error time sequence in a designated time range before a predicted time is selected, processed into a model input feature, input into a pre-trained SVR model to obtain a residual error predicted value of a load at the next time, and recorded as；
In some embodiments, residual prediction is performed using an SVR model, mainly as follows:
step 51: and carrying out normalization operation on the residual error of the load time sequence, converting the residual error into a plurality of samples with input and output components, establishing a data set according to the plurality of samples, mapping the data of the samples into a high-dimensional feature space by using a nonlinear function, and carrying out linear regression operation.
In this sub-step, a dataset is acquired. An existing load residual time sequence is obtained, normalized, and converted into a plurality of samples with input and output components by using a sliding window with a step length of 1 and a preset size. The input characteristics of the sample are a group of historical load residual values, and the output label is the load residual value at the next moment; the data range after normalization can be selected according to actual needs, and the embodiment of the application is not particularly limited.
Then, the data is processed. The sample data is mapped to a high-dimensional feature space using a nonlinear function, and linear regression is performed. The embodiment of the application adopts a radial basis function, and the formula is as follows:
wherein, the liquid crystal display device comprises a liquid crystal display device,is the width coefficient of the radial basis function.
Step 52: determining parameter range, traversing all combined parameters and determining penalty factor of SVR modelAnd->And establishing SVR models, evaluating each SVR model by using a cross-validation method, using average absolute errors as evaluation indexes to obtain optimal parameter combinations which enable the average absolute errors to be minimum, and constructing the SVR models by using the parameters.
In this sub-step, an optimal model is obtained. Penalty factorAnd->The method and the device have great influence on the prediction result, and in order to select proper parameters, the embodiment of the application adopts grid search to traverse all combined parameters, establishes models, evaluates each model by using 5-fold cross validation, and uses average absolute error MAE as an evaluation index to obtain the optimal parameter combination which enables the MAE to be minimum. Wherein, the MAE formula is as follows:
wherein, the liquid crystal display device comprises a liquid crystal display device,for the number of samples, +.>Is true value +.>Is a predicted value.
Specifically, the data set is equally divided into 5 parts, 4 parts are used as training sets, 1 part is used as test set, each model is trained 5 times, the weighted average of MAE obtained by the 5 test sets is used as an evaluation index, and the parameter when the MAE is minimum is selected to obtain the optimal model.
Finally, converting the load residual time sequence into input characteristics of a model, calling the optimal SVR model obtained in the step 3 to predict, and performing inverse normalization processing on the output to obtain a residual predicted value of the load at the next moment of the residual load sequence。
Step 6: and obtaining a prediction result of the short-term load of the CPU of the mobile terminal according to the linear prediction value of the load at the next moment and the residual prediction value of the load at the next moment.
In this step, the linear prediction value Ya of the load at the next time and the residual prediction value Ys are added to obtain a final prediction result of the load at the next time.
According to the method, through the steps 1 to 6 in the embodiment of the application, the combined use of the ARIMA model and the SVR-based machine learning model is realized, linear and nonlinear factors in a time sequence are fully captured, the problems of incomplete extraction time sequence characteristics and low prediction accuracy of a single model are solved, and the CPU resource calling strategy is assisted to be improved.
Example 2
Referring to fig. 5, based on the same inventive concept, an embodiment of the present application further provides a mobile terminal CPU short-term load prediction system based on a combined model, which includes:
a data acquisition unit for acquiring CPU load data of the mobile terminal, wherein the CPU load data at least comprises a CPU load time sequence;
a data processing unit for performing the steps of:
performing wavelet noise reduction operation on the CPU load time sequence, and decomposing to obtain a linear load time sequence;
constructing an ARIMA model, and inputting the linear load time sequence into the ARIMA model to obtain a linear predicted value of the load at the next moment of the linear load time sequence;
obtaining a residual error of the load time sequence according to the CPU load time sequence and the linear load time sequence at the corresponding moment;
establishing and training an SVR model, and inputting the residual error of the load time sequence into the SVR model to obtain a residual error predicted value of the load at the next moment;
and obtaining a prediction result of the short-term load of the CPU of the mobile terminal according to the linear prediction value of the load at the next moment and the residual prediction value of the load at the next moment.
Because the system is a system corresponding to the mobile terminal CPU short-term load prediction method based on the combined model according to the embodiment of the present application, and the principle of solving the problem of the system is similar to that of the method, the implementation of the system can refer to the implementation process of the above method embodiment, and the repetition is omitted.
Example 3
Referring to fig. 6, based on the same inventive concept, an embodiment of the present application further provides an electronic device including a processor and a memory, where at least one instruction, at least one program, a code set, or an instruction set is stored in the memory, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the combined model-based mobile terminal CPU short-term load prediction method as described above.
It is understood that the Memory may include random access Memory (Random Access Memory, RAM) or Read-Only Memory (RAM). Optionally, the memory includes a non-transitory computer readable medium (non-transitory computer-readable storage medium). The memory may be used to store instructions, programs, code sets, or instruction sets. The memory may include a stored program area and a stored data area, wherein the stored program area may store instructions for implementing an operating system, instructions for at least one function, instructions for implementing the various method embodiments described above, and the like; the storage data area may store data created according to the use of the server, etc.
The processor may include one or more processing cores. The processor uses various interfaces and lines to connect various portions of the overall server, perform various functions of the server, and process data by executing or executing instructions, programs, code sets, or instruction sets stored in memory, and invoking data stored in memory. Alternatively, the processor may be implemented in hardware in at least one of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), programmable logic array (Programmable Logic Array, PLA). The processor may integrate one or a combination of several of a central processing unit (Central Processing Unit, CPU) and a modem etc. Wherein, the CPU mainly processes an operating system, application programs and the like; the modem is used to handle wireless communications. It will be appreciated that the modem may not be integrated into the processor and may be implemented by a single chip.
Because the electronic device is the electronic device corresponding to the method for predicting the short-term load of the mobile terminal CPU based on the combined model according to the embodiment of the present application, and the principle of solving the problem of the electronic device is similar to that of the method, the implementation of the electronic device can refer to the implementation process of the embodiment of the method, and the repetition is omitted.
Example 4
Based on the same inventive concept, the embodiments of the present application also provide a computer-readable storage medium having stored therein at least one instruction, at least one program, a code set, or an instruction set, which is loaded and executed by a processor to implement the mobile terminal CPU short-term load prediction method based on a combined model as described above.
Those of ordinary skill in the art will appreciate that all or part of the steps of the various methods of the above embodiments may be implemented by a program that instructs associated hardware, the program may be stored in a computer readable storage medium including Read-Only Memory (ROM), random access Memory (Random Access Memory, RAM), programmable Read-Only Memory (Programmable Read-Only Memory, PROM), erasable programmable Read-Only Memory (Erasable Programmable Read Only Memory, EPROM), one-time programmable Read-Only Memory (OTPROM), electrically erasable programmable Read-Only Memory (EEPROM), compact disc Read-Only Memory (Compact Disc Read-Only Memory, CD-ROM) or other optical disk Memory, magnetic disk Memory, tape Memory, or any other medium that can be used for carrying or storing data that is readable by a computer.
Because the storage medium is a storage medium corresponding to the mobile terminal CPU short-term load prediction method based on the combined model according to the embodiment of the present application, and the principle of solving the problem by the storage medium is similar to that of the method, the implementation of the storage medium can refer to the implementation process of the above method embodiment, and the repetition is omitted.
Example 5
In some possible implementations, the aspects of the method of the embodiments of the present application may also be implemented in the form of a program product comprising program code for causing a computer device to carry out the steps of the method for short-term load prediction of a CPU of a mobile terminal based on a combined model according to the various exemplary embodiments of the present application as described in the present specification, when the program product is run on a computer device. Wherein executable computer program code or "code" for performing the various embodiments may be written in a high-level programming language such as C, C ++, c#, smalltalk, java, javaScript, visual Basic, structured query language (e.g., act-SQL), perl, or in a variety of other programming languages.
It is to be understood that portions of the present application may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, may be implemented using any one or combination of the following techniques, as is well known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present application. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, the different embodiments or examples described in this specification and the features of the different embodiments or examples may be combined and combined by those skilled in the art without contradiction.
The above embodiments are only for illustrating the technical concept and features of the present application, and are intended to enable those skilled in the art to understand the content of the present application and implement the same, and are not intended to limit the scope of the present application. All equivalent changes or modifications made in accordance with the essence of the present application are intended to be included within the scope of the present application.
Claims (10)
1. A mobile terminal CPU short-term load prediction method based on a combination model is characterized by comprising the following steps:
acquiring CPU load data of the mobile terminal, wherein the CPU load data at least comprises a CPU load time sequence;
performing wavelet noise reduction operation on the CPU load time sequence, and decomposing to obtain a linear load time sequence;
constructing an ARIMA model, and inputting the linear load time sequence into the ARIMA model to obtain a linear predicted value of the load at the next moment of the linear load time sequence;
obtaining a residual error of the load time sequence according to the CPU load time sequence and the linear load time sequence at the corresponding moment;
establishing and training an SVR model, and inputting the residual error of the load time sequence into the SVR model to obtain a residual error predicted value of the load at the next moment;
and obtaining a prediction result of the short-term load of the CPU of the mobile terminal according to the linear prediction value of the load at the next moment and the residual prediction value of the load at the next moment.
2. The method for predicting the short-term load of the CPU of the mobile terminal based on the combined model as claimed in claim 1, wherein the CPU load data of the mobile terminal is obtained according to a preset time sliding window size.
3. The method for predicting the short-term load of the mobile terminal CPU based on the combined model according to claim 1, wherein the wavelet denoising operation specifically comprises the steps of:
setting a wavelet basis function and a decomposition layer number j, and performing wavelet decomposition operation on the CPU load time sequence to obtain j high-frequency detail components and 1 low-frequency approximate component;
using a hard threshold value or a soft threshold value method to perform noise reduction operation on j high-frequency detail components;
and (3) carrying out reconstruction operation on the j+1 components after the decomposition operation and the noise reduction operation, and obtaining the denoised linear load time sequence after the integration operation.
4. The method for predicting the short-term load of the mobile terminal CPU based on the combined model according to claim 1, wherein the construction of the ARIMA model specifically comprises the following steps:
acquiring CPU historical load data of the mobile terminal, wherein the CPU historical load data at least comprises a CPU historical load time sequence; performing decomposition operation, noise reduction operation and reconstruction operation on the CPU historical load time sequence to obtain a denoised linear historical load time sequence;
performing stationarity check operation on the linear historical load time sequence, wherein if the linear historical load time sequence is a non-stationary sequence, performing differential operation, and if the linear historical load time sequence is a stationary sequence, performing randomness check operation to obtain a part with close correlation between sequence values of the linear historical load time sequence;
and constructing an ARIMA model by using the linear historical load time sequence of the part, wherein the steps p and q of the ARIMA model are determined by adopting a grid search method, all permutation and combination of p and q in a set range are acquired, the ARIMA models with different steps (p and q) are respectively fitted with the linear historical load time sequence, the parameter estimation is carried out by using the maximum likelihood estimation or the least square method, and the steps (p and q) taking the minimum value in the red pool information quantity criterion are selected to obtain the optimal ARIMA model.
5. The method for predicting the short-term load of a CPU of a mobile terminal based on a combined model according to claim 1, wherein,
the SVR model building and training method specifically comprises the following steps:
normalizing the residual error of the load time sequence, converting the residual error into a plurality of samples with input and output components, establishing a data set according to the plurality of samples, mapping the data of the samples into a high-dimensional feature space by using a nonlinear function, and performing linear regression operation;
determining parameter range, traversing all combined parameters and determining penalty factor of SVR modelAnd->And establishing SVR models, evaluating each SVR model by using a cross-validation method, using average absolute errors as evaluation indexes to obtain an optimal parameter combination which enables the average absolute errors to be minimum, and constructing the SVR models by using the parameter combination.
6. The method for predicting the short-term load of a CPU of a mobile terminal based on a combined model according to claim 1, wherein,
and the residual error of the obtained load time sequence is specifically obtained by subtracting the predicted value of the linear load time sequence at the corresponding moment from the real value of the CPU load time sequence.
7. The method for predicting the short-term load of a CPU of a mobile terminal based on a combined model according to claim 1, wherein,
and the obtained prediction result of the short-term load of the CPU of the mobile terminal is specifically that the linear prediction value of the load at the next moment and the residual prediction value of the load at the next moment are added.
8. A mobile terminal CPU short-term load prediction system based on a combined model, comprising:
a data acquisition unit for acquiring CPU load data of the mobile terminal, wherein the CPU load data at least comprises a CPU load time sequence;
a data processing unit for performing the steps of:
performing wavelet noise reduction operation on the CPU load time sequence, and decomposing to obtain a linear load time sequence;
constructing an ARIMA model, and inputting the linear load time sequence into the ARIMA model to obtain a linear predicted value of the load at the next moment of the linear load time sequence;
obtaining a residual error of the load time sequence according to the CPU load time sequence and the linear load time sequence at the corresponding moment;
establishing and training an SVR model, and inputting the residual error of the load time sequence into the SVR model to obtain a residual error predicted value of the load at the next moment;
and obtaining a prediction result of the short-term load of the CPU of the mobile terminal according to the linear prediction value of the load at the next moment and the residual prediction value of the load at the next moment.
9. An electronic device comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, a set of codes, or a set of instructions, the at least one instruction, the at least one program, the set of codes, or the set of instructions being loaded and executed by the processor to implement the combined model-based mobile terminal CPU short-term load prediction method of any one of claims 1 to 7.
10. A computer readable storage medium having stored therein at least one instruction, at least one program, code set, or instruction set, loaded and executed by a processor to implement a combined model based mobile terminal CPU short term load prediction method as claimed in any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311174944.6A CN116909754A (en) | 2023-09-13 | 2023-09-13 | Mobile terminal CPU short-term load prediction method based on combined model and related equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311174944.6A CN116909754A (en) | 2023-09-13 | 2023-09-13 | Mobile terminal CPU short-term load prediction method based on combined model and related equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116909754A true CN116909754A (en) | 2023-10-20 |
Family
ID=88358782
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311174944.6A Pending CN116909754A (en) | 2023-09-13 | 2023-09-13 | Mobile terminal CPU short-term load prediction method based on combined model and related equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116909754A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107908891A (en) * | 2017-11-28 | 2018-04-13 | 河海大学 | A kind of Hydrological Time Series rejecting outliers method based on ARIMA SVR |
| CN110149237A (en) * | 2019-06-13 | 2019-08-20 | 东北大学 | A kind of Hadoop platform calculate node load predicting method |
| US20220045509A1 (en) * | 2020-08-05 | 2022-02-10 | Wuhan University | Method and system of predicting electric system load based on wavelet noise reduction and emd-arima |
| CN114221877A (en) * | 2021-12-14 | 2022-03-22 | 中国联合网络通信集团有限公司 | Load prediction method, device, equipment and computer readable medium |
-
2023
- 2023-09-13 CN CN202311174944.6A patent/CN116909754A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107908891A (en) * | 2017-11-28 | 2018-04-13 | 河海大学 | A kind of Hydrological Time Series rejecting outliers method based on ARIMA SVR |
| CN110149237A (en) * | 2019-06-13 | 2019-08-20 | 东北大学 | A kind of Hadoop platform calculate node load predicting method |
| US20220045509A1 (en) * | 2020-08-05 | 2022-02-10 | Wuhan University | Method and system of predicting electric system load based on wavelet noise reduction and emd-arima |
| CN114221877A (en) * | 2021-12-14 | 2022-03-22 | 中国联合网络通信集团有限公司 | Load prediction method, device, equipment and computer readable medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112785016B (en) | New energy automobile maintenance and fault monitoring and diagnosis method based on machine learning | |
| CN112288021B (en) | Medical wastewater monitoring data quality control method, device and system | |
| CN113052317B (en) | Quantum state information acquisition method and device, quantum measurement and control system and computer | |
| CN112926644A (en) | Method and system for predicting residual service life of rolling bearing | |
| CN113326590A (en) | Rolling bearing service life prediction method and device based on deep learning model | |
| CN115409252A (en) | Plant growth data multi-step prediction method, device, medium and equipment based on LSTM | |
| CN114386454B (en) | Medical time sequence signal data processing method based on signal mixing strategy | |
| CN116647405B (en) | Multipath QUIC abnormal flow detection method, system and equipment | |
| Wang et al. | A novel rolling bearing fault diagnosis method based on adaptive denoising convolutional neural network under noise background | |
| CN113221382A (en) | Method, system and equipment for predicting residual life of industrial equipment | |
| CN117435979A (en) | Geological soil layer test evaluation method and system based on classification algorithm | |
| CN116909754A (en) | Mobile terminal CPU short-term load prediction method based on combined model and related equipment | |
| CN111259926A (en) | Meat freshness detection method and device, computing equipment and storage medium | |
| CN116826739A (en) | Ultra-short-term agricultural power load prediction method and device based on univariate time sequence | |
| CN116046396A (en) | Bearing fault diagnosis method, device, equipment and medium | |
| CN115130584A (en) | Time series prediction method, device, equipment and storage medium | |
| CN111027680B (en) | Monitoring quantity uncertainty prediction method and system based on variational self-encoder | |
| CN114511064A (en) | Neural network model interpretation method and device, electronic equipment and storage medium | |
| CN114399101A (en) | TCN-BIGRU-based gas load prediction method and device | |
| CN117407666B (en) | Intelligent garbage can parameter analysis and control method and device based on artificial intelligence | |
| Xu et al. | Bearing fault diagnosis based on hermitian wavelet and one-dimensional convolutional neural network | |
| CN117711505A (en) | Enzyme kinetic parameter prediction method, device, equipment and medium | |
| Pang et al. | Recurrent Variational Information Bottleneck for Time Series Trend Prediction | |
| CN117932287A (en) | Feature screening method and device, electronic equipment and storage medium | |
| Zhou et al. | A Novel Method for Signal Sequence Classification Based on Markov Reward Models |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |