Time Series Prediction Method, Device, Computer Equipment and Readable Storage Medium

Abstract

Embodiments of the present disclosure provide a time series prediction method, apparatus, computer device, and readable storage medium. The method includes: decomposing at least one original time series of the operation of a target device to obtain a trend item time data group and a seasonal item time data group of the operation of the target device; performing joint learning and training using the trend item time data group and the seasonal item time data group to obtain a set of trend items of the device operation; based on the target time series and the set of trend items, selecting an original time series from the at least one original time series as the target original time series to obtain a set of target original time series; determining a predicted value for a target date based on the trend items and seasonal items of the target time series and each target original time series in the set of target original time series. The present disclosure determines the predicted value for the target date based on the target original time series, which can integrate multiple data sources and improve the prediction accuracy.

Description

Technical Field

[0001] The present disclosure relates to the technical field of energy equipment, and particularly to a time series prediction method, apparatus, computer device, and readable storage medium. Background Art

[0002] A complete long time series usually requires the producer to accumulate data for several years or even decades to perform effective time series prediction, which is extremely costly. If multiple parties' data can be combined and the prediction accuracy can be improved by leveraging similar data, a large amount of manpower and material resources will be saved. However, directly performing joint learning on time series cannot obtain effective results due to factors such as feature input and noise redundancy. Summary of the Invention

[0003] In view of this, embodiments of the present disclosure provide a time series prediction method, apparatus, computer device, and readable storage medium to solve the technical problem in the prior art that the predicted value of time series cannot be accurately predicted.

[0004] In a first aspect of the embodiments of the present disclosure, a time series prediction method is provided, including: decomposing at least one original time series of the operation of a target device to obtain a trend item time data group and a seasonal item time data group of the operation of the target device; performing joint learning training using the trend item time data group and the seasonal item time data group to obtain a set of trend items of device operation; based on a target time series and the set of trend items, selecting an original time series from the at least one original time series as a target original time series to obtain a set of target original time series; and determining a predicted value for a target date based on the target time series and the trend items and seasonal items of each target original time series in the set of target original time series.

[0005] In a second aspect of the embodiments of the present disclosure, a time series prediction apparatus is provided, including: a decomposition unit configured to decompose at least one original time series of the operation of a target device to obtain a trend item time data group and a seasonal item time data group of the operation of the target device; a trend item set acquisition unit configured to perform joint learning training using the trend item time data group and the seasonal item time data group to obtain a set of trend items of device operation; a selection unit configured to select an original time series from the at least one original time series as a target original time series based on a target time series and the set of trend items to obtain a set of target original time series; and a prediction unit configured to determine a predicted value for a target date based on the target time series and the trend items and seasonal items of each target original time series in the set of target original time series.

[0006] In a third aspect of the embodiments of the present disclosure, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above method are implemented.

[0007] In a fourth aspect of the embodiments of the present disclosure, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0008] One embodiment among the above various embodiments of the present disclosure has the following beneficial effects: First, at least one original time series is decomposed to obtain a set of trend terms and a set of seasonal terms; then, based on the target time series and the set of trend terms obtained by decomposition, a set of target original time series is selected from the above at least one original time series; finally, based on the seasonal terms, the predicted value of the target date is determined. The method provided by the present disclosure can remove the original time series with low similarity to the target time series through the selection of the original time series, improve the prediction accuracy, and determine the predicted value of the target date based on the target original time series can integrate multiple data sources and enhance the prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In combination with the accompanying drawings and with reference to the following specific embodiments, the above and other features, advantages, and aspects of the various embodiments of the present disclosure will become more apparent. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic, and the elements and elements are not necessarily drawn to scale.

[0010] Figure 1 is a schematic diagram of an architecture for joint learning according to an embodiment of the present disclosure;

[0011] Figure 2 is a flowchart of an embodiment of a time series prediction method according to the present disclosure;

[0012] Figure 3 is a schematic structural diagram of an embodiment of a time series prediction device according to the present disclosure;

[0013] Figure 4 is a schematic structural diagram of an electronic device suitable for implementing an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0014] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although some embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. On the contrary, these embodiments are provided to more thoroughly and completely understand the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not used to limit the protection scope of the present disclosure.

[0015] In addition, it should be noted that for ease of description, only parts related to the relevant invention are shown in the drawings. Without conflict, the embodiments in the present disclosure and the features in the embodiments can be combined with each other.

[0016] It should be noted that concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0017] It should be noted that the modifications of "one" and "multiple" mentioned in the present disclosure are illustrative rather than restrictive. Those skilled in the art should understand that unless clearly specified otherwise in the context, it should be understood as "one or more".

[0018] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only for illustrative purposes and are not used to limit the scope of these messages or information.

[0019] The present disclosure will be described in detail below with reference to the drawings and in combination with embodiments.

[0020] Federated learning refers to the comprehensive utilization of various AI (Artificial Intelligence) technologies on the premise of ensuring data security and user privacy, jointly mining data value by multiple parties in cooperation, and giving birth to new intelligent business forms and models based on joint modeling. Federated learning has at least the following characteristics:

[0021] (1) A weakly centralized joint training mode in which participating nodes control their own data, ensuring data privacy and security in the process of co-creating intelligence.

[0022] (2) In different application scenarios, by using screening and / or combining AI algorithms and privacy-preserving computing, establishing various model aggregation and optimization strategies to obtain high-level and high-quality models.

[0023] (3) On the premise of ensuring data security and user privacy, based on a variety of model aggregation optimization strategies, a method for improving the performance of the federated learning engine is obtained. The performance method can be to improve the overall performance of the federated learning engine by solving problems including parallel computing architectures, information interaction under large-scale cross-domain networks, intelligent perception, and anomaly handling mechanisms.

[0024] (4) Obtain the requirements of multiple parties of users in each scenario, and through a mutual trust mechanism, determine a reasonable evaluation of the true contribution of each federated participant, and conduct distribution incentives.

[0025] Based on the above methods, an AI technology ecosystem based on federated learning can be established to give full play to the value of industry data and promote the implementation of scenarios in vertical fields.

[0026] Next, a time series prediction method and device based on federated learning according to an embodiment of the present disclosure will be described in detail with reference to the accompanying drawings.

[0027] Figure 1 is a schematic diagram of the architecture of a federated learning according to an embodiment of the present disclosure. As Figure 1 shown, the architecture of the federated learning may include a server (central node) 101 and participants 102, 103, and 104.

[0028] In the process of federated learning, the basic model can be established by the server 101, and the server 101 sends the model to the participants 102, 103, and 104 that have established a communication connection with it. The basic model can also be established by any participant and uploaded to the server 101, and the server 101 sends the model to other participants that have established a communication connection with it. Participants 102, 103, and 104 construct a model according to the downloaded basic structure and model parameters, use local data to train the model, obtain updated model parameters, and encrypt and upload the updated model parameters to the server 101. The server 101 aggregates the model parameters sent by the participants 102, 103, and 104 to obtain global model parameters, and sends the global model parameters back to the participants 102, 103, and 104. Participants 102, 103, and 104 iterate their respective models according to the received global model parameters until the model finally converges, thereby realizing the training of the model. In the process of federated learning, all participants such as participants 102, 103, and 104 can share the final model parameters. It should be noted that the number of participants is not limited to the three as described above, but can be set as needed, and the embodiments of the present disclosure do not limit this.

[0029] Figure 2 shows the flow 200 of an embodiment of the time series prediction method according to the present disclosure. The method can be performed byFigure 1 be executed by any of the participating parties or the central node therein. The time series prediction method includes the following steps:

[0030] Step S201: Decompose at least one original time series of the target device's operation to obtain a trend item time data group and a seasonal item time data group of the target device's operation.

[0031] In this embodiment, the execution subject of the time series prediction method can decompose at least one original time series to obtain a trend item time data group and a seasonal item time data group. Here, the above-mentioned execution subject can use the time series decomposition method (Seasonal and Trend decomposition using Loess, STL) with robust locally weighted regression as the smoothing method to decompose the above-mentioned original time series.

[0032] Step S202: Use the trend item time data group and the seasonal item time data group for joint learning and training to obtain a trend item set of the device's operation. During the training process, for the seasonal item time data group, each participating party can train separately and then combine the training results; for the trend item time data group, each participating party conducts joint learning and training; by aggregating the results of separate training and joint learning and training, a trend item set of the device's operation is obtained, and the trend item set can include multiple data groups.

[0033] Step S203: Based on the target time series and the above-mentioned trend item set, select an original time series from the above-mentioned at least one original time series as the target original time series to obtain a target original time series set.

[0034] In the embodiment, the above-mentioned target time series can be obtained through a wired connection method or a wireless connection method. The above-mentioned execution subject can select an original time series from the above-mentioned at least one original time series as the target original time series through the following steps to obtain a target original time series set:

[0035] The first step is to decompose the above-mentioned target time series to obtain the trend item and seasonal item of the above-mentioned target time series. As an example, the above-mentioned execution subject can use the time series decomposition method (Seasonal and Trend decomposition using Loess, STL) with robust locally weighted regression as the smoothing method to decompose the above-mentioned target time series.

[0036] In the second step, the distance between each trend item in the above-mentioned trend item set and the trend item of the above-mentioned target time series can be calculated to obtain a distance set. Here, the distance between trend items can be the Dynamic Time Warping (DTW) distance. Specifically, the DTW algorithm mainly solves the distance between two templates. Generally, in the case of the same dimension or the same number of sequences, the Euclidean distance or the Mahalanobis distance is used to obtain the similarity between two templates. However, when the dimension or the number of sequences is different and they cannot be corresponded one by one, the DTW algorithm needs to be used to expand or reduce to the same number of sequences before calculating the distance. Among them, the Euclidean distance can be a commonly used distance definition. It is the actual distance between two points in the m-dimensional space. The Euclidean distance in the two-dimensional space is the straight-line segment distance between two points. The Mahalanobis distance can be an effective method for calculating the similarity between two unknown sample sets.

[0037] In the third step, based on the above-mentioned distance set, a trend item can be selected from the above-mentioned trend item set as the target trend item to obtain a target trend item set. As an example, based on the above-mentioned distance set, the above-mentioned execution entity can select the trend item with a distance less than a preset threshold as the target trend item.

[0038] In the fourth step, based on the above-mentioned target trend item set, the original time series corresponding to the target trend item in the above-mentioned target trend item set can be selected from the above-mentioned at least one original time series as the target original time series to obtain the above-mentioned target original time series.

[0039] Step S204: Based on the above-mentioned target time series and the trend items and seasonal items of each target original time series in the above-mentioned target original time series set, determine the predicted value of the target date.

[0040] In the embodiment, based on the above-mentioned target time series and the trend items and seasonal items of each target original time series in the above-mentioned target original time series set, the above-mentioned execution entity can determine the predicted value of the target date through the following steps:

[0041] In the first step, feature extraction can be performed on the target trend items in the above-mentioned target trend item set to obtain a trend item feature set. Here, the feature extraction is mainly time-domain and frequency-domain features. Features can be extracted using a sliding window, such as the average value, variance, zero-crossing rate, etc., as well as the amplitude, frequency, mean, etc. after Fourier transform.

[0042] In the second step, the above-mentioned set of trend item features can be trained using a preset training method to obtain a trained set of trend item features. Here, the above-mentioned preset training method can be a training method that uses the LightGBM (Light Gradient Boosting Machine) algorithm for joint learning. Specifically, LightGBM can be a distributed gradient boosting framework based on the decision tree algorithm.

[0043] In the third step, the above-mentioned trained set of trend item features and the above-mentioned target sequence can be fitted to obtain a trend item fitting result. Here, the above-mentioned fitting can be a method of establishing a regression model of the sequence value changing with time, taking time as the independent variable and the corresponding sequence observation value as the dependent variable, where the fitting includes linear fitting and non-linear fitting. Specifically, the occasion for using linear fitting is the occasion where the long-term trend presents a linear feature. The parameter estimation method is the least squares estimation. Linear regression can be used to handle it. The occasion for using non-linear fitting is that the long-term trend presents a non-linear feature, and a curve fitting model can be used for fitting. When estimating the parameters of the curve model, those that can be converted into a linear model are converted into a linear model, and the linear least squares method is used for parameter estimation. Those that cannot be converted into a linear model can be estimated using the iterative method.

[0044] In the fourth step, the seasonal items of the above-mentioned target time series can be feature-extracted to obtain the seasonal item features of the above-mentioned target time series.

[0045] In the fifth step, the seasonal items of each target original time series in the above-mentioned set of target original time series can be feature-extracted to obtain a set of seasonal item features.

[0046] In the sixth step, the above-mentioned set of seasonal item features can be trained using the above-mentioned preset training method to obtain a trained set of seasonal item features. As an example, the above-mentioned preset training method can be a training method that uses the LightGBM algorithm for joint learning.

[0047] In the seventh step, the above-mentioned seasonal item features and the above-mentioned set of seasonal item features can be fitted to obtain a seasonal item fitting result.

[0048] In the eighth step, an initial model can be obtained. Here, the initial model can be a primary neural network model using the LightGBM algorithm.

[0049] In the ninth step, based on the above-mentioned trend item fitting result, the above-mentioned seasonal item fitting result, and the above-mentioned initial model, a time series trend change model is trained. As an example, the above-mentioned execution entity can fuse the above-mentioned trend item fitting result, the above-mentioned seasonal item fitting result, and the above-mentioned initial model to obtain a time series trend change model.

[0050] In the tenth step, the above target date can be input into the above time series trend change model to obtain a predicted value.

[0051] In the embodiment, using LightGBM can reduce the memory usage of data, ensuring that a single machine can use as much data as possible without sacrificing speed. It can reduce the communication cost, improve the efficiency in multi-machine parallelism, achieve linear acceleration in computing, and relieve the operating pressure of the computer at the same time.

[0052] In an alternative implementation of the embodiment, the above method further includes: transmitting the above predicted value to a target device with a display function, and controlling the target device to display the above predicted value.

[0053] One embodiment of the above various embodiments of the present disclosure has the following beneficial effects: First, decompose at least one original time series to obtain a trend item set and a seasonal item set; then, based on the target time series and the decomposed trend item set, select a target original time series set from the above at least one original time series; finally, determine the predicted value of the target date based on the seasonal items. The method provided by the present disclosure can remove the original time series with low similarity to the target time series through the selection of the original time series, improve the prediction accuracy, and determine the predicted value of the target date based on the target original time series can synthesize multi-party data through joint learning, seek common ground while reserving differences, and improve the prediction accuracy.

[0054] Any combination of the above all optional technical solutions can form an optional embodiment of the present application, which will not be elaborated here one by one.

[0055] Further referring to Figure 3 , as an implementation of the above methods in the above figures, the present disclosure provides some embodiments of a time series prediction device. These device embodiments correspond to Figure 2 the above method embodiments, and the device can be specifically applied to various electronic devices.

[0056] Such as Figure 3As shown, the time series prediction device 300 of the embodiment includes: a decomposition unit 301, a trend item set acquisition unit 302, a selection unit 303, and a prediction unit 304. Among them, the decomposition unit 301 is configured to decompose at least one original time series of the target device operation to obtain a trend item time data group and a seasonal item time data group of the target device operation; the trend item set acquisition unit 302 is configured to perform joint learning training using the trend item time data group and the seasonal item time data group to obtain a trend item set of the device operation; the selection unit 303 is configured to select an original time series from the at least one original time series as the target original time series based on the target time series and the above-mentioned trend item set to obtain a target original time series set; the prediction unit 304 is configured to determine a predicted value of the target date based on the above-mentioned target time series and the trend items and seasonal items of each target original time series in the above-mentioned target original time series set.

[0057] In an alternative implementation manner of the embodiment, the selection unit 303 of the time series prediction device 300 is further configured to: decompose the above-mentioned target time series to obtain the trend item and seasonal item of the above-mentioned target time series; calculate the distance between each trend item in the above-mentioned trend item set and the trend item of the above-mentioned target time series to obtain a distance set; select a trend item from the above-mentioned trend item set as the target trend item based on the above-mentioned distance set to obtain a target trend item set; select an original time series corresponding to the target trend item in the above-mentioned target trend item set from the above-mentioned at least one original time series as the target original time series to obtain the above-mentioned target original time series set.

[0058] In an alternative implementation manner of the embodiment, the prediction unit 304 of the time series prediction device 300 is further configured to: obtain an initial model; obtain a time series trend change model based on the obtained trend item fitting result, the obtained seasonal item fitting result, and the above-mentioned initial model; input the above-mentioned target date into the above-mentioned time series trend change model to obtain a predicted value.

[0059] In an alternative implementation manner of the embodiment, the prediction unit 304 of the time series prediction device 300 obtaining the trend item fitting result is specifically configured to: extract features from the target trend items in the above-mentioned target trend item set to obtain a trend item feature set; train the above-mentioned trend item feature set using a preset training method to obtain a trained trend item feature set; fit the above-mentioned trained trend item feature set and the above-mentioned target time series to obtain a trend item fitting result.

[0060] In an alternative implementation of the embodiment, the prediction unit 304 of the time series prediction device 300 is specifically configured to obtain the seasonal item fitting result as follows: extract features from the seasonal items of the target time series to obtain the seasonal item features of the target time series; extract features from the seasonal items of each target original time series in the target original time series set to obtain a set of seasonal item features; use the preset training method to train the set of seasonal item features to obtain a trained set of seasonal item features; fit the seasonal item features and the set of seasonal item features to obtain the seasonal item fitting result.

[0061] In an alternative implementation of the embodiment, the time series prediction device 300 is further configured to transmit the predicted value to a target device with a display function and control the target device to display the predicted value.

[0062] It can be understood that the various units described in the device 300 correspond to the respective steps in the method described with reference Figure 2 Therefore, the operations, features, and beneficial effects described above for the method also apply to the device 300 and the units included therein, and will not be repeated here.

[0063] It should be understood that the sequence numbers of the steps in the above embodiments do not indicate the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present disclosure.

[0064] Figure 4 is a schematic diagram of the computer device 4 provided by the embodiment of the present disclosure. As Figure 4 shown, the computer device 4 of this embodiment includes: a processor 401, a memory 402, and a computer program 403 stored in the memory 402 and executable on the processor 401. When the processor 401 executes the computer program 403, the steps in the above various method embodiments are implemented. Alternatively, when the processor 401 executes the computer program 403, the functions of the various modules / units in the above device embodiments are implemented.

[0065] Exemplarily, the computer program 403 can be divided into one or more modules / units. The one or more modules / units are stored in the memory 402 and executed by the processor 401 to complete the present disclosure. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program 403 in the computer device 4.

[0066] The computer device 4 can be a desktop computer, a notebook, a palm computer, a cloud server, or other computer devices. The computer device 4 may include, but is not limited to, a processor 401 and a memory 402. Those skilled in the art can understand that Figure 4 merely examples of the computer device 4 do not constitute a limitation on the computer device 4, and may include more or fewer components than shown in the figure, or combine some components, or different components. For example, the computer device may also include input / output devices, network access devices, a bus, etc.

[0067] The processor 401 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0068] The memory 402 can be an internal storage unit of the computer device 4. For example, the hard disk or memory of the computer device 4. The memory 402 can also be an external storage device of the computer device 4. For example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device 4. Further, the memory 402 can also include both the internal storage unit and the external storage device of the computer device 4. The memory 402 is used to store computer programs and other programs and data required by the computer device. The memory 402 can also be used to temporarily store data that has been output or will be output.

[0069] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiments can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction and do not limit the protection scope of this application. The specific working processes of the units and modules in the above system can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0070] In the above embodiments, the descriptions of the various embodiments have their own emphases. For the parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0071] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or by a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this disclosure.

[0072] In the embodiments provided by this disclosure, it should be understood that the disclosed device / computer device and method can be implemented in other ways. For example, the device / computer device embodiments described above are only illustrative. For example, the division of modules or units is only a logical functional division. In actual implementation, there may be other division methods. Multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of devices or units can be in electrical, mechanical or other forms.

[0073] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0074] In addition, in each embodiment of the present disclosure, each functional unit may be integrated into one processing unit, may exist independently as individual units physically, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of a software functional unit.

[0075] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it may be stored in a computer-readable storage medium. Based on such an understanding, all or part of the processes in the above-described embodiment methods of the present disclosure may also be completed by instructing relevant hardware through a computer program. The computer program may be stored in the computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-described method embodiments may be implemented. The computer program may include computer program code, and the computer program code may be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0076] The above embodiments are only used to illustrate the technical solutions of the present disclosure, rather than to limit them; although the present disclosure has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications 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 disclosure, and should all be included in the protection scope of the present disclosure.

Claims

1. A time series prediction method, characterized in that, it includes: Decompose at least one original time series of the operation of the target device to obtain a trend item time data group and a seasonal item time data group of the operation of the target device; Use the trend item time data group and the seasonal item time data group for joint learning and training to obtain a set of trend items of device operation; Based on the target time series and the set of trend items, select an original time series from the at least one original time series as the target original time series to obtain a set of target original time series; Based on the target time series and the trend items and seasonal items of each target original time series in the set of target original time series, determine the predicted value of the target date; The step of based on the target time series and the set of trend items, selecting an original time series from the at least one original time series as the target original time series to obtain a set of target original time series includes: Decompose the target time series to obtain the trend item and seasonal item of the target time series; Calculate the distance between each trend item in the set of trend items and the trend item of the target time series to obtain a distance set; Based on the distance set, select a trend item with a distance less than a preset threshold from the set of trend items as the target trend item to obtain a set of target trend items; Based on the set of target trend items, select the original time series corresponding to the target trend items in the set of target trend items from the at least one original time series as the target original time series to obtain the set of target original time series; The step of based on the target time series and the trend items and seasonal items of each target original time series in the set of target original time series, determining the predicted value of the target date includes: Obtain an initial model; Based on the obtained trend item fitting result, the obtained seasonal item fitting result and the initial model, obtain a time series trend change model; Input the target date into the time series trend change model to obtain a predicted value.

2. The time series prediction method according to claim 1, characterized in that, The process of obtaining the trend item fitting result includes: Extract features from the target trend items in the set of target trend items to obtain a set of trend item features; Use a preset training method to train the set of trend item features to obtain a trained set of trend item features; Fit the trained set of trend item features and the target time series to obtain a trend item fitting result.

3. The time series prediction method according to claim 1, characterized in that, The process of obtaining the seasonal item fitting result includes: Extract features from the seasonal items of the target time series to obtain the seasonal item features of the target time series; Extract features from the seasonal items of each target original time series in the set of target original time series to obtain a set of seasonal item features; Use a preset training method to train the set of seasonal item features to obtain a trained set of seasonal item features; Fit the seasonal item features and the set of seasonal item features to obtain a seasonal item fitting result.

4. The time series prediction method according to any one of claims 1 to 3, wherein, the method further includes: transmitting the predicted value to a target device having a display function, and controlling the target device to display the predicted value.

5. A time series prediction device, wherein, it includes: a decomposition unit configured to decompose at least one original time series of the operation of the target device to obtain a trend item time data group and a seasonal item time data group of the operation of the target device; a trend item set acquisition unit configured to perform joint learning and training using the trend item time data group and the seasonal item time data group to obtain a trend item set of the device operation; a selection unit configured to select an original time series from the at least one original time series as a target original time series based on the target time series and the trend item set to obtain a target original time series set; a prediction unit configured to determine a predicted value of a target date based on the target time series and the trend items and seasonal items of each target original time series in the target original time series set; the selection unit is further configured to: decompose the target time series to obtain a trend item and a seasonal item of the target time series; calculate the distance between each trend item in the trend item set and the trend item of the target time series to obtain a distance set; select, based on the distance set, a trend item with a distance less than a preset threshold from the trend item set as a target trend item to obtain a target trend item set; select, based on the target trend item set, an original time series corresponding to the target trend item in the target trend item set from the at least one original time series as a target original time series to obtain the target original time series set; the prediction unit is further configured to: obtain an initial model; obtain a time series trend change model based on the obtained trend item fitting result, the obtained seasonal item fitting result, and the initial model; input the target date into the time series trend change model to obtain a predicted value.

6. A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein, when the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium stores a computer program, wherein, when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images