Multi-source time series data sum prediction method and device, computer device and medium

Abstract

This disclosure relates to the field of energy data processing technology, and provides a method, apparatus, computer equipment, and medium for summative prediction of multi-source time series data. The method includes: generating a joint test set based on at least two acquired original test sets; processing the joint test set according to a preset dataset decomposition method to obtain a joint trend subset and a joint period subset; training an initial trend model using the joint trend subset to generate a target trend model; training an initial period model using the joint period subset to generate a target period model; and generating target prediction data based on a preset calculation strategy, the joint trend subset, the joint period subset, the target trend model, and the target period model. This disclosure can significantly reduce prediction errors.

Description

Technical Field

[0001] This disclosure relates to the field of energy data processing technology, and in particular to methods, apparatus, computer equipment and media for summative prediction of multi-source time series data. Background Technology

[0002] With the rapid development of data processing technology, the energy sector is generating increasing data processing demands. In forecasting, it is common to encounter situations where forecasts from multiple energy companies need to be summed. Because forecasts from individual companies contain errors, summing the data from multiple companies leads to an accumulation of errors, resulting in excessively large forecast errors. Summary of the Invention

[0003] In view of this, embodiments of this disclosure provide a method, apparatus, computer equipment, and medium for predicting the sum of multi-source time series data, in order to solve the problem in the prior art where single-source prediction has errors and the summation of data from multiple companies leads to the accumulation of errors, resulting in excessive prediction errors.

[0004] A first aspect of this disclosure provides a method for summing and predicting multi-source time series data, comprising: generating a joint test set based on at least two acquired original test sets; processing the joint test set according to a preset dataset decomposition method to obtain a joint trend subset and a joint period subset; training an initial trend model using the joint trend subset to generate a target trend model; training an initial period model using the joint period subset to generate a target period model; and generating target prediction data based on a preset calculation strategy, the joint trend subset, the joint period subset, the target trend model, and the target period model.

[0005] A second aspect of this disclosure provides a multi-source time-series data summation prediction apparatus, comprising: a generation module configured to generate a joint test set based on at least two acquired original test sets; a processing module configured to process the joint test set based on a preset dataset decomposition method to obtain a joint trend subset and a joint period subset; a trend generation module configured to train an initial trend model using the joint trend subset to generate a target trend model; a period generation module configured to train an initial period model using the joint period subset to generate a target period model; and a target generation module configured to generate target prediction data based on a preset calculation strategy, the joint trend subset, the joint period subset, the target trend model, and the target period model.

[0006] A third aspect of this disclosure provides a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described above.

[0007] A fourth aspect of this disclosure provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described method.

[0008] The beneficial effects of this disclosure embodiment compared with the prior art include at least the following: generating a joint test set based on at least two obtained original test sets; processing the joint test set according to a preset dataset decomposition method to obtain a joint trend subset and a joint period subset; training an initial trend model using the joint trend subset to generate a target trend model; training an initial period model using the joint period subset to generate a target period model; and generating target prediction data based on a preset calculation strategy, the joint trend subset, the joint period subset, the target trend model, and the target period model, which can greatly reduce prediction errors. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this disclosure, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of a joint learning architecture provided in an embodiment of this disclosure;

[0011] Figure 2 This is a flowchart of a multi-source time series data summation prediction method provided in an embodiment of this disclosure;

[0012] Figure 3 This is a flowchart of a specific embodiment of a multi-source time series data summation prediction method provided in this disclosure;

[0013] Figure 4 This is a block diagram of a multi-source time-series data summation prediction device provided in an embodiment of this disclosure;

[0014] Figure 5 This is a schematic diagram of a computer device provided in an embodiment of this disclosure. Detailed Implementation

[0015] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, so as to provide a thorough understanding of the embodiments of this disclosure. However, those skilled in the art will understand that this disclosure may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this disclosure with unnecessary detail.

[0016] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0017] Federation learning refers to the comprehensive utilization of multiple AI (Artificial Intelligence) technologies, under the premise of ensuring data security and user privacy, to collaboratively explore the value of data and foster new intelligent business forms and models based on joint modeling. Federation learning has at least the following characteristics:

[0018] (1) Participating nodes control their own data in a weakly centralized joint training mode to ensure data privacy and security in the process of co-creating intelligence.

[0019] (2) In different application scenarios, various model aggregation optimization strategies are established by using screening and / or combination of AI algorithms and privacy-preserving computing to obtain high-level and high-quality models.

[0020] (3) Under the premise of ensuring data security and user privacy, based on multiple model aggregation optimization strategies, obtain methods to improve the performance of the federated learning engine. The performance methods can be improved by solving problems such as parallel computing architecture, information interaction under large-scale cross-domain networks, intelligent perception, and anomaly handling mechanisms.

[0021] (4) Obtain the needs of multiple users in various scenarios, determine the true contribution of each joint participant through a mutual trust mechanism, and allocate incentives accordingly.

[0022] Based on the above approach, an AI technology ecosystem based on collaborative learning can be established, fully leveraging the value of industry data and promoting the implementation of scenarios in vertical fields.

[0023] This disclosure will now be described in detail with reference to the accompanying drawings.

[0024] Figure 1 This is a schematic diagram of a joint learning architecture according to an embodiment of this disclosure. Figure 1 As shown, the architecture of joint learning may include a server (central node) 101 and participants 102, 103 and 104.

[0025] In the joint learning process, a basic model can be established through server 101, which then sends this model to participants 102, 103, and 104 with whom it has established a communication connection. Alternatively, any participant can establish the basic model and upload it to server 101, which then sends it to other participants with whom it has established a communication connection. Participants 102, 103, and 104 construct models based on the downloaded basic structure and model parameters, train the models using local data, obtain updated model parameters, and encrypt and upload these updated model parameters to server 101. Server 101 aggregates the model parameters sent by participants 102, 103, and 104 to obtain global model parameters, which are then transmitted back to participants 102, 103, and 104. Participants 102, 103, and 104 iterate on their respective models based on the received global model parameters until the models converge, thus achieving model training. During the collaborative learning process, the data uploaded by participants 102, 103, and 104 are model parameters. Local data is not uploaded to server 101, and all participants can share the final model parameters. Therefore, collaborative modeling can be achieved while ensuring data privacy. It should be noted that the number of participants is not limited to the three mentioned above, but can be set as needed. This embodiment of the disclosure does not impose any restrictions on this.

[0026] Figure 2 This is a flowchart of a multi-source time series data summation prediction method provided in an embodiment of this disclosure. Figure 2 Multi-source time series data summation prediction methods can be derived from Figure 1 The server or participating party executes this. For example... Figure 2 As shown, this multi-source time series data summation prediction method includes:

[0027] S201, Generate a joint test set based on at least two original test sets obtained.

[0028] The original test set and the joint test set can refer to a data structure composed of multiple metadata. Metadata can refer to a data structure composed of at least one piece of data. Metadata can refer to a data structure composed of one or more feature data. Feature data can refer to the basic numerical unit. As an example, feature data can be "average temperature: 30.2", "daily gas consumption: 35.86", or "first-order difference of daily gas consumption: 15.326", etc., where the first-order difference of daily gas consumption can refer to the difference between two consecutive adjacent daily gas consumption values. Metadata can be a data structure composed of the above three types of feature data, or a data structure composed of other feature data, as needed, without specific restrictions here. The joint test set can refer to the data set obtained by processing and merging the at least two original test sets.

[0029] S202, based on the preset dataset decomposition method, processes the joint test set to obtain a joint trend subset and a joint period subset.

[0030] Dataset decomposition can refer to the method of dividing a dataset consisting of at least one set of metadata into multiple subsets with the same structure as the original dataset. Identical structure means that at least two datasets have the same types and quantities of metadata features, as well as the same number of metadata items. A trend subset can refer to a subset of data decomposed based on a trend phenomenon. A periodic subset can refer to a subset of data decomposed based on a periodic phenomenon. A trend phenomenon refers to a tendency or state that continues to develop and change over a relatively long period. For example, when a predicted value shows a clear upward or downward trend over a period of time, we consider it to have a trend. A periodic phenomenon can refer to a continuous, irregular, periodic change caused by seasonal variations within a certain period of time. For example, when a predicted value shows a clear, continuous, periodic fluctuation within a specific time period, we consider it to have a periodicity.

[0031] S203, the initial trend model is trained by combining trend subsets to generate the target trend model.

[0032] The initial trend model can refer to an existing or user-defined mathematical formula related to the trend phenomenon, with its parameters set to initial default values. These parameters can be constants, arrays, vectors, etc. Model training refers to the process of determining the model parameters based on existing data through a series of steps or methods. The target trend model can refer to a trained mathematical formula with its parameters already determined.

[0033] S204, the initial periodic model is trained by combining periodic subsets to generate the target periodic model.

[0034] The initial periodic model can refer to an existing or user-defined mathematical formula related to periodic phenomena, with its parameters set to initial default values. These parameters can be constants, arrays, vectors, etc. Model training refers to the process of determining the model parameters based on existing data through a series of steps or methods. The target periodic model can refer to a trained mathematical formula with its parameters determined.

[0035] S205 generates target prediction data based on a preset calculation strategy, a joint trend subset, a joint period subset, a target trend model, and a target period model.

[0036] The computational strategy can refer to the methods or steps for generating target prediction data based on a baseline trend subset, a baseline period subset, a target trend model, and a target period model.

[0037] According to the technical solution provided in this disclosure, a joint test set is generated based on at least two obtained original test sets; the joint test set is processed according to a preset dataset decomposition method to obtain a joint trend subset and a joint period subset; an initial trend model is trained using the joint trend subset to generate a target trend model; an initial period model is trained using the joint period subset to generate a target period model; and target prediction data is generated based on a preset calculation strategy, the joint trend subset, the joint period subset, the target trend model, and the target period model, which can greatly reduce prediction errors.

[0038] In some embodiments, generating a joint test set based on at least two acquired original test sets includes: acquiring at least two original test sets, wherein each original test set includes at least one metadata, and each metadata includes a kernel data; performing isomorphic processing on each of the at least two original test sets to generate at least two isomorphic test sets; summing the at least two isomorphic test sets to obtain a summed test set; and processing each metadata in the summed test set based on a basic data processing strategy to generate basic processed metadata to obtain the joint test set.

[0039] Core data can refer to key data within metadata. For example, when predicting daily gas consumption, the "daily gas consumption" in the metadata is core data. Homogeneous processing can refer to the steps or methods of processing each of at least two datasets to have the same timestamps and data structure. For example, dataset A contains 180 data points, and one of its metadata points could be: {"Daily gas consumption: 35.2", "Time: September 2, 2021", "Collection location: City A"}. Dataset B contains 230 data points, and one of its metadata points could be: {"Daily gas consumption: 38.2", "Time: May 1, 2021"}.

[0040] In this dataset, 168 metadata entries share the same timestamp as dataset A and dataset B. After isomorphic processing, datasets A and B will retain these 168 metadata entries with the same timestamp, while deleting the rest. Taking the feature data of dataset A as an example, the feature data of the metadata in the isomorphic dataset could be: {"Daily gas consumption: 35.2", "Time: September 2, 2021"}. It should be noted that isomorphic data can generally be achieved by deleting feature data or by adding differences, depending on the needs; no specific restrictions are imposed here.

[0041] Addition processing refers to the method of adding the kernel data from multiple datasets and retaining the rest. For example, kernel data A can be {"Daily gas consumption: 35.2", "Time: September 2, 2021"}, kernel data B can be {"Daily gas consumption: 75.2", "Time: September 2, 2021"}, then after addition processing, kernel data C can be obtained as {"Daily gas consumption: 110.4", "Time: September 2, 2021"}.

[0042] In some embodiments, the basic data processing strategy includes: performing anomaly processing on at least one acquired metadata to obtain at least one metadata after anomaly processing; performing smoothing processing on at least one metadata after anomaly processing to obtain at least one metadata after smoothing processing, thereby obtaining a smoothed dataset; and processing the smoothed dataset based on a splitting processing strategy to obtain at least one basic processed metadata.

[0043] Anomaly handling can refer to deleting or replacing anomalous data in at least one metadata element. Anomalous data refers to one or more data points in at least one metadata element that differ significantly from the majority of the data. Anomaly handling can include checking data consistency or handling invalid and missing values. Data consistency refers to data having the same basic characteristics or properties, and similar other characteristics or properties. Invalid values ​​can refer to null values, values ​​that do not meet data type requirements, or other anomalous values. Missing values ​​refer to the incompleteness of one or more attributes in the existing dataset. The target test set can refer to the dataset consisting of time-series data obtained after anomaly handling.

[0044] Smoothing refers to reducing the magnitude of change in at least one piece of metadata. Smoothing can make data trends more apparent. As an example, smoothing can be expressed using the following mathematical formula:

[0045] F(t+n) / 2=(F(t+1)+F(t+1)+F(t+1)+...+F(t+n)) / n,

[0046] Where t can refer to the data number, and F(t+n) can refer to the data with the (t+n)th number, where t and n are integers.

[0047] A splitting processing strategy can refer to the steps or methods of further splitting at least one feature data based on the timestamp data. As an example, a timestamp data can be "September 2, 2021", and the timestamp data can be further split into data including but not limited to any of the following: "day of the week: 4", "day of the year: 243", "week of the year: 35", "annual information: 2021" or "monthly information: 09".

[0048] By splitting the timestamp into additional time feature data, we can increase the dimensionality and complexity of the data, making the training results more accurate.

[0049] In some embodiments, the smoothed dataset is processed based on a splitting processing strategy to obtain at least one basic processed metadata, including: obtaining at least one splitting metric; generating at least one intermediate timestamp dataset based on the at least one splitting metric and the timestamp data of each metadata in the smoothed dataset; and updating each timestamp data in the smoothed dataset to each metadata in the smoothed dataset to obtain at least one basic processed metadata.

[0050] A splitting metric can refer to a splitting identifier that extracts additional feature data. For example, if a timestamp is "September 2, 2021" and the splitting metric is "Annual Information", then the feature data extracted based on this splitting metric will be "Annual Information: 2021".

[0051] In some embodiments, generating at least one intermediate timestamp dataset based on the timestamp data of each metadata in the smoothing dataset and at least one splitting metric includes: Step 1: Obtaining one of the at least one splitting metrics that has not been marked as split, to obtain an intermediate metric; Step 2: Processing the timestamp data of each metadata in the smoothing dataset based on the intermediate metric to generate at least one intermediate timestamp data, to obtain an intermediate timestamp dataset; Step 3: Marking the intermediate metric as split; Repeating steps 1 to 3 until each of the at least one splitting metrics is marked as split, to obtain at least one intermediate timestamp dataset.

[0052] In some embodiments, the preset dataset decomposition method is the STL multiplicative dataset decomposition method.

[0053] STL's multiplicative dataset decomposition method divides the dataset into three subsets: a trend subset, a period subset, and a residual subset. The terms "trend subset," "period subset," and "residual subset" are explained above and will not be repeated here. Multiplying the corresponding metadata in each subset yields the original dataset. It's important to note that the residual subset typically accounts for a very small percentage, usually less than 1%, and is generally ignored. For example, with an initial dataset of 100, after STL multiplicative decomposition, the trend value is 20, and the period value is 5 (residual value ignored). Multiplying the trend value and the period value gives the original dataset of 100 before splitting.

[0054] It should be noted that the dataset can be decomposed in other ways as needed, and no specific restrictions are imposed here.

[0055] In some embodiments, the calculation strategy includes: importing a joint trend subset into a target trend model to obtain trend target data; importing a joint period subset into a target period model to obtain period target data; and obtaining target prediction data based on the trend target data and the period target data.

[0056] Trend target data refers to the predicted trend value. Period target data refers to the predicted period value. If the dataset is decomposed using the STL multiplicative dataset decomposition method, the target predicted data can be obtained by multiplying the trend value and the target value.

[0057] Figure 3 This is a flowchart of the method for predicting the total daily gas volume of Company A and Company B in 2021, provided in an embodiment of this disclosure. Figure 3 The method for predicting the total daily gas consumption of Company A and Company B in 2021 can be derived from... Figure 1 Server execution. For example... Figure 3 As shown, the prediction method includes:

[0058] S301, obtain the original test set A of company A and the original test set B of company B respectively. Both the original test set A and the original test set B include at least one metadata, and each metadata includes a kernel data.

[0059] As an example, the original test set A can contain 190 metadata entries. The data format of the metadata in the original test set A can be referenced from one of the metadata entries in the original test set A: {"Daily gas consumption: 65.2", "Time: September 2, 2020", "Sampling location: City A"}. The core data of the original test set A is "Daily gas consumption". The original test set B can contain 223 metadata entries. The data format of its metadata can be referenced from one of the metadata entries in the original test set B: {"Daily gas consumption: 55.62", "Time: May 1, 2020"}.

[0060] S302, perform isomorphic processing on the original test set A and the original test set B to generate isomorphic test set A and isomorphic test set B.

[0061] As an example, the original test set A and the original test set B have 190 metadata entries with the same timestamp. After isomorphic processing, the number of metadata entries in both the original test set A and the original test set B is 190. Before processing, a metadata entry in the original test set A could be {"Daily gas consumption: 65.2", "Time: September 2, 2020", "Collection location: City A"}, and after processing, it can be {"Daily gas consumption: 65.2", "Time: September 2, 2020"}.

[0062] S303, sum the isomorphic test set A and isomorphic test set B to obtain the summed test set.

[0063] The addition process can be referred to above, and will not be repeated here.

[0064] S304. Based on the basic data processing strategy, each metadata in the summation test set is processed to generate basic processed metadata, thus obtaining the joint test set.

[0065] S305, based on the STL multiplicative time series decomposition method, processes the joint test set to obtain a joint trend subset and a joint period subset.

[0066] S306, the target trend model is generated by training the initial trend model with a joint trend subset.

[0067] S307 trains the initial periodic model by combining periodic subsets to generate the target periodic model.

[0068] S308 imports the joint trend subset into the target trend model to obtain the trend target data.

[0069] S309. Import the joint periodic subset into the target periodic model to obtain the periodic target data.

[0070] S310, multiply the trend target data and the cycle target data to obtain the predicted total daily gas consumption of Company A and Company B in 2021.

[0071] All of the above-mentioned optional technical solutions can be combined in any way to form the optional embodiments of this application, and will not be described in detail here.

[0072] The following are embodiments of the apparatus disclosed herein, which can be used to execute embodiments of the method disclosed herein. For details not disclosed in the apparatus embodiments of this disclosure, please refer to the embodiments of the method disclosed herein.

[0073] Figure 4 This is a schematic diagram of a multi-source time-series data summation prediction device provided in an embodiment of this disclosure. Figure 4 As shown, the multi-source time-series data summation prediction device includes:

[0074] The generation module 401 is configured to generate a joint test set based on at least two obtained original test sets;

[0075] The processing module 402 is configured to process the joint test set based on a preset dataset decomposition method to obtain a joint trend subset and a joint period subset;

[0076] The trend generation module 403 is configured to train the initial trend model by using a joint trend subset to generate a target trend model;

[0077] The period generation module 404 is configured to train the initial period model by using a joint period subset to generate the target period model;

[0078] The target generation module 405 is configured to generate target prediction data based on a preset calculation strategy, a joint trend subset, a joint period subset, a target trend model, and a target period model.

[0079] According to the technical solution provided in this disclosure, a joint test set is generated based on at least two obtained original test sets; the joint test set is processed according to a preset dataset decomposition method to obtain a joint trend subset and a joint period subset; an initial trend model is trained using the joint trend subset to generate a target trend model; an initial period model is trained using the joint period subset to generate a target period model; and target prediction data is generated based on a preset calculation strategy, the joint trend subset, the joint period subset, the target trend model, and the target period model, which can greatly reduce prediction errors.

[0080] In some embodiments, the generation module 401 of the multi-source time series data summation prediction device is further configured to: acquire at least two original test sets, wherein each original test set includes at least one metadata, and each metadata includes a kernel data; perform isomorphic processing on each of the at least two original test sets to generate at least two isomorphic test sets; perform summation processing on the at least two isomorphic test sets to obtain a summed test set; and process each metadata in the summed test set based on the basic data processing strategy to generate basic processed metadata to obtain a joint test set.

[0081] In some embodiments, the basic data processing strategy includes: performing anomaly processing on at least one acquired metadata to obtain at least one metadata after anomaly processing; performing smoothing processing on at least one metadata after anomaly processing to obtain at least one metadata after smoothing processing, thereby obtaining a smoothed dataset; and processing the smoothed dataset based on a splitting processing strategy to obtain at least one basic processed metadata.

[0082] In some embodiments, the smoothed dataset is processed based on a splitting processing strategy to obtain at least one basic processed metadata, including: obtaining at least one splitting metric; generating at least one intermediate timestamp dataset based on the at least one splitting metric and the timestamp data of each metadata in the smoothed dataset; and updating each timestamp data in the smoothed dataset to each metadata in the smoothed dataset to obtain at least one basic processed metadata.

[0083] In some embodiments, generating at least one intermediate timestamp dataset based on timestamp data of each metadata in at least one splitting metric and a smoothed dataset includes: Step 1: Obtaining one of the at least one splitting metrics that has not been marked as split, to obtain an intermediate metric; Step 2: Processing each timestamp data in the smoothed dataset based on the intermediate metric to generate at least one intermediate timestamp data, to obtain an intermediate timestamp dataset; Step 3: Marking the intermediate metric as split; Repeating steps 1 to 3 until each of the at least one splitting metrics is marked as split, to obtain at least one intermediate timestamp dataset.

[0084] In some embodiments, the preset weighting method is the STL multiplicative time series decomposition method.

[0085] In some embodiments, the calculation strategy includes: importing a joint trend subset into a target trend model to obtain trend target data; importing a joint period subset into a target period model to obtain period target data; and obtaining target prediction data based on the trend target data and the period target data.

[0086] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this disclosure.

[0087] Figure 5 This is a schematic diagram of a computer device 500 provided in an embodiment of this disclosure. Figure 5 As shown, the computer device 500 of this embodiment includes a processor 501, a memory 502, and a computer program 503 stored in the memory 502 and executable on the processor 501. When the processor 501 executes the computer program 503, it implements the steps in the various method embodiments described above. Alternatively, when the processor 501 executes the computer program 503, it implements the functions of each module / unit in the various device embodiments described above.

[0088] For example, computer program 503 may be divided into one or more modules / units, which are stored in memory 502 and executed by processor 501 to perform the present disclosure. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 503 in computer device 500.

[0089] Computer device 500 can be a desktop computer, laptop, handheld computer, cloud server, or other similar computer device. Computer device 500 may include, but is not limited to, a processor 501 and a memory 502. Those skilled in the art will understand that... Figure 5This is merely an example of computer device 500 and does not constitute a limitation on computer device 500. It may include more or fewer components than shown, or combine certain components, or different components. For example, computer device may also include input / output devices, network access devices, buses, etc.

[0090] Processor 501 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. A general-purpose processor can be a microprocessor or any conventional processor.

[0091] The memory 502 can be an internal storage unit of the computer device 500, such as a hard disk or RAM of the computer device 500. The memory 502 can also be an external storage device of the computer device 500, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computer device 500. Furthermore, the memory 502 can include both internal and external storage units of the computer device 500. The memory 502 is used to store computer programs and other programs and data required by the computer device. The memory 502 can also be used to temporarily store data that has been output or will be output.

[0092] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, 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. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0093] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0094] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this disclosure.

[0095] In the embodiments provided in this disclosure, it should be understood that the disclosed apparatus / computer devices and methods can be implemented in other ways. For example, the apparatus / computer device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. Multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0096] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0097] Furthermore, the functional units in the various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0098] If an integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program may include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. A computer-readable medium may include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in a computer-readable medium may be appropriately added to or subtracted according to the requirements of legislation and patent practice in a jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media may not include electrical carrier signals and telecommunication signals.

[0099] The above embodiments are only used to illustrate the technical solutions of this disclosure, and are not intended to limit it. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure, and should all be included within the protection scope of this disclosure.

Claims

1. A method for summative prediction of multi-source time series data, characterized in that, include: A joint test set is generated based on at least two original test sets obtained; The original test set is a collection of time-series data from different energy companies. The time-series data set refers to a collection of data sequences with continuous time identifiers, and the data sequences include daily gas consumption, time, and collection location. The joint test set is processed based on a preset dataset decomposition method to obtain a joint trend subset and a joint period subset; The initial trend model is trained using the aforementioned joint trend subset to generate the target trend model; The initial periodic model is trained using the aforementioned joint periodic subset to generate the target periodic model; Based on the preset calculation strategy, the joint trend subset, the joint period subset, the target trend model, and the target period model, target prediction data is generated. The generation of a joint test set based on at least two obtained original test sets includes: At least two original test sets are obtained, wherein each original test set includes at least one metadata, and each metadata includes a kernel data; the metadata is a data structure composed of at least one feature data, the data structure including daily gas consumption, time, and collection location; the kernel data is the key numerical data in the metadata used for prediction, the key numerical data being daily gas consumption; Perform isomorphic processing on each of the at least two original test sets to generate at least two isomorphic test sets; The at least two isomorphic test sets are summed to obtain a summed test set. Based on the basic data processing strategy, each metadata in the summed test set is processed to generate basic processed metadata, thus obtaining the joint test set; The basic data processing strategy includes: Perform exception handling on at least one of the acquired metadata to obtain at least one metadata after exception handling; At least one metadata after anomaly processing is smoothed to obtain at least one metadata after smoothing, and a smoothed dataset is obtained. The smoothed dataset is processed based on a splitting processing strategy to obtain at least one basic processed metadata. The process of processing the smoothed dataset based on the splitting strategy to obtain at least one basic processed metadata includes: Obtain at least one splitting metric; Based on the timestamp data of each metadata in the at least one splitting metric and the smoothed dataset, at least one intermediate timestamp dataset is generated; Update each timestamp data in the smoothed dataset to each metadata in the smoothed dataset to obtain the at least one basic processed metadata; The generation of at least one intermediate timestamp dataset based on the timestamp data of each metadata in the at least one splitting metric and the smoothed dataset includes: Obtain at least one of the split metrics that is not marked as split, and obtain the intermediate metric; Based on the intermediate index, each timestamp data in the smoothed dataset is processed to generate at least one intermediate timestamp data, resulting in an intermediate timestamp dataset. Mark intermediate metrics as split; Until each split metric is marked as split, at least one intermediate timestamp dataset is obtained.

2. The method according to claim 1, characterized in that, The preset dataset decomposition method is the STL multiplicative time series decomposition method.

3. The method according to any one of claims 1 to 2, characterized in that, The calculation strategy includes: The combined trend subset is imported into the target trend model to obtain the trend target data; The joint periodic subset is imported into the target periodic model to obtain the periodic target data; Based on the trend target data and the periodic target data, target prediction data is obtained.

4. A multi-source time-series data summation prediction device, wherein the device employs the method described in any one of claims 1-3, characterized in that, The device includes: The generation module is configured to generate a joint test set based on at least two original test sets obtained; The processing module is configured to process the joint test set based on a preset dataset decomposition method to obtain a joint trend subset and a joint period subset; The trend generation module is configured to train the initial trend model using the joint trend subset to generate a target trend model; The period generation module is configured to train the initial period model using the joint period subset to generate the target period model; The target generation module is configured to generate target prediction data based on a preset calculation strategy, the joint trend subset, the joint period subset, the target trend model, and the target period model.

5. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 3.

6. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 3.

Images