CN116470485A - Time-series processing system, method and computer-readable storage medium - Google Patents
Time-series processing system, method and computer-readable storage medium Download PDFInfo
- Publication number
- CN116470485A CN116470485A CN202310355943.5A CN202310355943A CN116470485A CN 116470485 A CN116470485 A CN 116470485A CN 202310355943 A CN202310355943 A CN 202310355943A CN 116470485 A CN116470485 A CN 116470485A
- Authority
- CN
- China
- Prior art keywords
- time sequence
- target
- sequence
- original
- data
- 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
- 238000012545 processing Methods 0.000 title claims abstract description 147
- 238000000034 method Methods 0.000 title claims abstract description 61
- 238000003860 storage Methods 0.000 title claims abstract description 32
- 238000007781 pre-processing Methods 0.000 claims abstract description 47
- 238000010276 construction Methods 0.000 claims abstract description 10
- 238000001514 detection method Methods 0.000 claims description 62
- 238000004422 calculation algorithm Methods 0.000 claims description 41
- 230000008569 process Effects 0.000 claims description 34
- 230000002159 abnormal effect Effects 0.000 claims description 27
- 238000000354 decomposition reaction Methods 0.000 claims description 26
- 238000010606 normalization Methods 0.000 claims description 24
- 238000012805 post-processing Methods 0.000 claims description 18
- 238000001914 filtration Methods 0.000 claims description 13
- 238000012217 deletion Methods 0.000 claims description 12
- 230000037430 deletion Effects 0.000 claims description 12
- 238000000605 extraction Methods 0.000 claims description 11
- 238000006243 chemical reaction Methods 0.000 claims description 8
- 230000003993 interaction Effects 0.000 claims description 7
- 238000012544 monitoring process Methods 0.000 claims description 5
- 230000015654 memory Effects 0.000 description 34
- 230000008859 change Effects 0.000 description 22
- 238000010586 diagram Methods 0.000 description 19
- 230000000737 periodic effect Effects 0.000 description 17
- 238000003672 processing method Methods 0.000 description 15
- 230000005540 biological transmission Effects 0.000 description 11
- 238000004891 communication Methods 0.000 description 8
- 230000005856 abnormality Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 5
- 239000000654 additive Substances 0.000 description 4
- 230000000996 additive effect Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 3
- 238000010248 power generation Methods 0.000 description 3
- 230000000306 recurrent effect Effects 0.000 description 3
- 239000000470 constituent Substances 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 238000010295 mobile communication Methods 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 208000025174 PANDAS Diseases 0.000 description 1
- 208000021155 Paediatric autoimmune neuropsychiatric disorders associated with streptococcal infection Diseases 0.000 description 1
- 240000000220 Panda oleosa Species 0.000 description 1
- 235000016496 Panda oleosa Nutrition 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000013475 authorization Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 230000001427 coherent effect Effects 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000013079 data visualisation Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000008021 deposition Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000004372 laser cladding Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 238000013450 outlier detection Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/003—Load forecast, e.g. methods or systems for forecasting future load demand
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Power Engineering (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A time-series processing system, method, and computer-readable storage medium are disclosed. Wherein, this system includes: the data input module is used for acquiring an original time sequence; the data preprocessing module is used for preprocessing the data of the original time sequence to obtain a target time sequence; the feature engineering module is used for carrying out feature construction on the target time sequence to obtain target features of the target time sequence; and the data prediction module is used for carrying out data prediction on the target time sequence based on the target characteristics to obtain a target prediction result of the original time sequence. The method and the device solve the technical problem that in the related art, the power prediction accuracy is low because the power time sequence is affected by sensing errors, acquisition faults and the like.
Description
Technical Field
The present application relates to the field of data processing, and in particular, to a time-series processing system, method, and computer-readable storage medium.
Background
The accurate electric power prediction is the basis of operation and analysis of an electric power system, has important significance for unit combination, economic dispatch, safety check and the like, and good electric power data prediction can effectively help the safe and stable operation of a power grid, so that electric power can be subjected to peak-welcome summer and peak-welcome winter.
However, at present, the power data is affected by sensing errors, acquisition faults and the like, abnormal values and missing values usually occur, and because different types of power loads are affected by different factors, the prediction result is inaccurate when the power data of a future time period is predicted according to the power data of a historical time period.
In view of the above problems, no effective solution has been proposed at present.
Disclosure of Invention
The embodiment of the application provides a processing system, a processing method and a computer readable storage medium of a time sequence, which are used for at least solving the technical problem that the accuracy of power prediction is lower because the power time sequence is affected by sensing errors, acquisition faults and the like in the related technology.
According to an aspect of an embodiment of the present application, there is provided a time-series processing system including: the data input module is used for acquiring an original time sequence, wherein the original time sequence is used for representing a time sequence generated by the operation of the target system in a historical time period; the data preprocessing module is used for preprocessing data of the original time sequence to obtain a target time sequence, wherein the target time sequence fills a sequence value with a deletion in the original time sequence to obtain a sequence; the feature engineering module is used for carrying out feature construction on the target time sequence to obtain target features of the target time sequence; the data prediction module is used for carrying out data prediction on the target time sequence based on the target characteristics to obtain a target prediction result of the original time sequence, wherein the target prediction result is used for representing the time sequence in a preset time period after the current moment.
According to an aspect of the embodiments of the present application, there is provided a time-series processing method, including: acquiring an original time sequence, wherein the original time sequence is used for representing a time sequence generated by the operation of a target system in a historical time period; performing data preprocessing on the original time sequence to obtain a target time sequence, wherein the target time sequence fills a sequence value with a deletion in the original time sequence to obtain a sequence; performing feature construction on the target time sequence to obtain target features of the target time sequence; and carrying out data prediction on the target time sequence based on the target characteristics to obtain a target prediction result of the original time sequence, wherein the target prediction result is used for representing the time sequence in a preset time period after the current moment.
According to another aspect of the embodiments of the present application, there is also provided a time-series processing method, including: monitoring the operation process of the power system and displaying an original time sequence on an operation interface, wherein the original time sequence is used for representing the time sequence generated by the power system in a historical time period; and responding to a prediction instruction acted on the operation interface, and displaying a target prediction result of the original time sequence on the operation interface, wherein the target prediction result is used for representing the time sequence in a preset time period after the current moment, the target prediction result is obtained by carrying out data prediction on the target time sequence based on target characteristics of the target time sequence, the target time sequence is obtained by carrying out data preprocessing on the basis of the original time sequence, and the target characteristics are obtained by carrying out characteristic extraction on the target time sequence.
According to another aspect of the embodiments of the present application, there is also provided a time-series processing method, including: the method comprises the steps of obtaining an original time sequence by calling a first interface, wherein the first interface comprises a first parameter, the parameter value of the first parameter is the original time sequence, and the original time sequence is used for representing a time sequence generated by the operation of a target system in a historical time period; performing data preprocessing on the original time sequence to obtain a target time sequence, wherein the target time sequence fills a sequence value with a deletion in the original time sequence to obtain a sequence; performing feature construction on the target time sequence to obtain target features of the target time sequence; performing data prediction on the target time sequence based on the target characteristics to obtain a target prediction result of the original time sequence; and outputting a target prediction result by calling a second interface, wherein the second interface comprises a second parameter, the parameter value of the second parameter is the target prediction result, and the target prediction result is used for representing a time sequence in a preset time period after the current moment.
According to another aspect of the embodiments of the present application, there is also provided a computer readable storage medium including a stored executable program, where the executable program when executed controls a processing system and method for executing the time sequence of any one of the above steps by a device in which the computer readable storage medium is located.
In the embodiment of the application, a processing system comprising a data input module, a data preprocessing module, a characteristic engineering module and a data prediction module is adopted, and the data input module is utilized to acquire an original time sequence; performing data preprocessing on the original time sequence by using a data preprocessing module to obtain a target time sequence; extracting features of the target time sequence by utilizing feature engineering to obtain target features of the target time sequence; the method comprises the steps of carrying out data prediction on a target time sequence based on target characteristics by using a data prediction module to obtain a target prediction result of an original time sequence, carrying out data preprocessing on the original time sequence, filling sequence values which are missing in the original time sequence to obtain the target time sequence which is close to an actual value, improving the processing capacity of a processing system of the time sequence on the original time sequence, extracting target characteristics of the target time sequence, carrying out data prediction processing on the target time sequence based on the target characteristics to obtain the target prediction result of the original time sequence, further ensuring the rationality of the target prediction result, further improving the accuracy of the predicted power data, and further solving the technical problem that the power prediction accuracy is lower due to the influence of sensing errors, acquisition faults and the like on the power time sequence in the related technology.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the application.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application and do not constitute an undue limitation to the application. In the drawings:
FIG. 1 is a block diagram of the hardware architecture of a computer terminal (or mobile device) for implementing a time-series processing system according to embodiment 1 of the present application;
FIG. 2 is a schematic diagram of a computer terminal interaction according to embodiment 1 of the present application;
FIG. 3 is a block diagram of a time-series processing system according to embodiment 1 of the present application;
FIG. 4 is a schematic illustration of a time series process according to embodiment 1 of the present application;
FIGS. 5 (a) and 5 (b) are schematic diagrams of a time-series processing system framework according to embodiment 1 of the present application;
FIG. 6 is a flow chart of a time-series processing method according to embodiment 2 of the present application;
FIG. 7 is a flow chart of a time-series processing method according to embodiment 3 of the present application;
FIG. 8 is a flow chart of a time-series processing method according to embodiment 4 of the present application;
fig. 9 is a block diagram of a processing apparatus of generated power according to embodiment 5 of the present application;
fig. 10 is a block diagram of a structure of a processing device of generated power according to embodiment 6 of the present application;
fig. 11 is a block diagram of an electronic device according to embodiment 8 of the present application.
Detailed Description
In order to make the present application solution better understood by those skilled in the art, the following description will be made in detail and with reference to the accompanying drawings in the embodiments of the present application, it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, shall fall within the scope of the present application.
It should be noted that the terms "first," "second," and the like in the description and 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 embodiments of the present 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 but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
First, partial terms or terminology appearing in describing embodiments of the present application are applicable to the following explanation:
time series (time series) is a group of data points arranged according to time sequence, and is widely applied to the fields of signal processing, statistics, weather, electric power and the like.
Power prediction (electricity forecasting), an application of time series prediction, by modeling power timing and other relevant metrics that give a historical time, power timing predictions for a future time, a typical scenario includes: and predicting bus load, system load and new energy power.
Automatic machine learning (automatic machine learning), a technique that automates the process of machine learning to solve problems, focuses mainly on feature engineering, model selection, and super-parametric optimization.
And in the process of converting the original data into better characteristics for expressing the nature of the problem, the characteristic engineering extracts and screens the data characteristics in the power time sequence to determine target characteristics capable of clearly representing the change rule of the data.
Heterogeneous refers to different information such as constituent elements, element distribution and the like among different objects, and in the application, refers to different application scenes, different data generated by a system under application conditions, and different corresponding information such as structures, constituent elements and the like among different original time sequences.
The robust decomposition can decompose the time sequence into different components, can be suitable for decomposing the time sequence with high complexity and large data volume, and has better recognition capability and high adaptability compared with the common time sequence decomposition method for outliers and noise of the time sequence.
The robust trend filtering can extract the sequence part with the variation trend from the time sequence, and compared with the common time sequence filtering method, the robust trend filtering has better extraction capability and very high adaptability to outliers and noise mutation in a small range. It should be noted that, the robust trend filtering corresponds to a filter to extract a specified sequence portion, that is, a trend portion, from the time sequence, and the robust decomposition corresponds to a splitter to split the time sequence according to differences between different components in the time sequence, and both components in the time sequence can be obtained, but specific execution processes are different.
The prediction interpretation can be used for marking and interpreting the prediction result, so that a user can quickly determine the influence degree of the used data on the prediction result in the process of generating the prediction result, for example, the corresponding weight, contribution value and other parameters of each data in the prediction process are determined, and the user can conveniently understand the prediction result.
Example 1
In accordance with the embodiments of the present application, there is also provided a time-series processing system embodiment, it being noted that the steps illustrated in the flowcharts of the figures may be performed in a computer system, such as a set of computer-executable instructions, and that, although a logical order is illustrated in the flowcharts, in some cases, the steps illustrated or described may be performed in an order other than that illustrated herein.
The system embodiment provided in the first embodiment of the present application may be executed in a mobile terminal, a computer terminal, or a similar computing device. Fig. 1 is a block diagram of a hardware architecture of a computer terminal (or mobile device) for implementing a time-series processing system according to embodiment 1 of the present application. As shown in fig. 1, the computer terminal 10 (or mobile device) may include one or more processors 102 (shown as 102a,102b, … …,102 n) which may include, but are not limited to, a microprocessor (Microcontroller Unit, MCU) or a programmable logic device (Field Programmable Gate Array, FPGA) or the like, a memory 104 for storing data, and a transmission 106 for communication functions. In addition, the method may further include: a display, an input/output interface (I/O interface), a universal serial BUS (Universal Serial Bus, USB) port (which may be included as one of the ports of the BUS), a network interface, a power supply, and/or a camera. It will be appreciated by those of ordinary skill in the art that the configuration shown in fig. 1 is merely illustrative and is not intended to limit the configuration of the electronic device described above. For example, the computer terminal 10 may also include more or fewer components than shown in FIG. 1, or have a different configuration than shown in FIG. 1.
It should be noted that the one or more processors 102 and/or other data processing circuits described above may be referred to generally herein as "data processing circuits. The data processing circuit may be embodied in whole or in part in software, hardware, firmware, or any other combination. Furthermore, the data processing circuitry may be a single stand-alone processing module, or incorporated, in whole or in part, into any of the other elements in the computer terminal 10 (or mobile device). As referred to in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of the path of the variable resistor termination to interface).
The memory 104 may be used to store software programs and modules of application software, such as program instructions/data storage devices of a time-series processing system in the embodiments of the present application, and the processor 102 executes the software programs and modules stored in the memory 104 to perform various functional applications and data processing, i.e., a processing system implementing the time-series processing system described above. Memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include memory located remotely from the processor 102, which may be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The transmission means 106 is arranged to receive or transmit data via a network. The specific examples of the network described above may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC) that can connect to other network devices through a base station to communicate with the internet. In one example, the transmission device 106 may be a Radio Frequency (RF) module for communicating with the internet wirelessly.
The display may be, for example, a touch screen type liquid crystal display (Laser Cladding Deposition, LCD) that may enable a user to interact with a user interface of the computer terminal 10 (or mobile device).
Fig. 1 shows a block diagram of a hardware structure, which may be used not only as an exemplary block diagram of the computer terminal 10 (or mobile device) but also as an exemplary block diagram of the server, and in an alternative embodiment, fig. 2 is a schematic diagram of a computer terminal interaction according to embodiment 1 of the present application, and fig. 2 shows, in a block diagram, an embodiment using the computer terminal 10 (or mobile device) shown in fig. 1 as a receiving end. As shown in fig. 2, the computer terminal 10 (or mobile device) may be connected or electronically connected to one or more clients via a data network connection. In an alternative embodiment, the computer terminal 10 (or mobile device) may be a cloud server. The data network connection may be a local area network connection, a wide area network connection, an internet connection, or other type of data network connection. The computer terminal 10 (or mobile device) may provide network services to a connected client or group of clients 20. The web server is a web-based user service such as power prediction.
In the above-described operating environment, the present application provides a time-series processing system as shown in FIG. 3. Fig. 3 is a block diagram of a time-series processing system according to embodiment 1 of the present application, and as shown in fig. 3, the system 300 includes: a data input module 302, a data preprocessing module 304, a feature engineering module 306, and a data prediction module 308.
The data input module 302 is configured to obtain an original time sequence.
Wherein the original time series is used to characterize the time series produced by the target system operating over a historical period of time.
The target system may refer to a system capable of generating and recording historical data and predicting future data, and may include, but is not limited to: power systems, wind systems, etc.
The historical time period may be a time period between the start of the target system and the current time, or may be a time period before the current time.
The original time sequence may be a time sequence constructed by time sequence of original data, i.e. historical data, generated by the target system operating within a preset historical period.
In an alternative of this embodiment, taking the target system as an example of the power prediction system, in order to improve accuracy of a power time sequence in a future period predicted by the power prediction system, a time sequence processing system (hereinafter, simply referred to as a processing system, it should be noted that the processing system may be a server or a client installed on a mobile terminal, specific selection may be determined according to practical situations, and specific limitation is not herein made) may first obtain historical power data generated by the power prediction system in a preset historical period, and construct a corresponding time sequence, that is, the above-mentioned original time sequence according to the historical power data.
The data preprocessing module 304 is configured to perform data preprocessing on the original time sequence to obtain a target time sequence.
And filling the sequence value with the deletion in the original time sequence by the target time sequence formula to obtain the sequence.
The above-described target time series may refer to a time series capable of reflecting an actual value or a target value of data generated when the target system operates in a history period, and the target data in the target time series is more accurate than the original data in the original time series.
In an alternative scheme of this embodiment, since errors may exist in the process of generating and recording the historical data, for example, the generated historical data is abnormal, the recorded historical data is wrong in time, and the data with errors or abnormalities usually affect the accuracy of the prediction result of the processing system, generally, the processing system may remove or manually delete the detected errors, that is, the data with errors or abnormalities, at the time point corresponding to the data with errors or abnormalities in the original time sequence, the corresponding data value is lacking, that is, the missing value also affects the accuracy of the prediction result of the processing system, so, in order to avoid the influence of the missing value in the original data on the process of predicting the data, for example, the efficiency of data prediction, the accuracy of the prediction result, and the like, after the original time sequence is obtained, the data preprocessing may be further performed on the original time sequence, that is, to fill the sequence value with the missing in the original time sequence, so as to obtain the complete time sequence, that is, the target time sequence.
And the feature engineering module 306 is used for carrying out feature construction on the target time sequence to obtain target features of the target time sequence.
The above target features may refer to features capable of clearly representing time sequence elements of the target time sequence, and may include, but are not limited to: the alignment feature may be a feature obtained by performing alignment processing on an original time sequence and a similar time sequence according to time points in the sequence, the similar time sequence may be a time sequence generated by other systems in the same application environment as the target system in a historical time period, the sample normalization feature may be a feature obtained by performing normalization processing on a past value and a future value to be predicted by using a preset value, and a specific alignment feature and a sample normalization feature determining process are shown below.
In an alternative of this embodiment, considering that the above-mentioned target feature can clearly represent the time sequence element of the target time sequence, in order to improve the accuracy of the data predicted by the target time sequence, the target system may perform feature construction on the target time sequence by using feature engineering, obtain the target feature of the target time sequence, and apply the target feature to the data prediction process performed on the target time sequence.
In general, the above feature engineering can be performed using automated learning.
The data prediction module 308 is configured to perform data prediction on the target time sequence based on the target feature, so as to obtain a target prediction result of the original time sequence.
The target prediction result is used for representing a time sequence in a preset time period after the current moment.
The preset time period may refer to the foregoing future time period, may be a time period automatically generated by the processing system according to the operation condition of the target system, for example, the operation time, the data generating efficiency, etc., or may be a time period set by the user, specifically may be according to the actual condition, and is not limited herein specifically.
In an alternative of this embodiment, after obtaining the target time sequence and the corresponding target feature, the data prediction may be performed on the target time sequence according to the target feature, so as to obtain a future time sequence that may be generated by the target system in a preset time period after the current time, that is, the target prediction result.
In an alternative scheme of this embodiment, in order to improve the universality of the processing system, to facilitate the user to understand the processing system, a conventional prediction model and a common deep learning model may be adopted to perform data prediction on the target time sequence, so as to obtain the target prediction result.
Among other things, a traditional machine learning predictive model, such as a tree model, may include: gradient lifting tree models (such as extreme gradient boosting extreme gradient lifting, abbreviated as XGB, light gradient boosting machine lightweight gradient lifting, abbreviated as LGBM), support vector regression models (Support Vector Regression, abbreviated as SVR), categorised gradient lifting models (Catboost), random Forest models (Random Forest), and the like; the conventional model may include: differential integration moving average autoregressive model (Autoregressive Integrated Moving Average model, abbreviated as ARIMA), prophet, generalized additive model (Generalized Additive Model, abbreviated as GAM), and the like. Common deep learning models, such as models built based on transformations, may include: a fourier prediction model (e.g., fedFormer (Frequency Enhanced Decomposed Transformer, frequency-enhanced decomposition transducer)), quaternion Transformer (QuatFormer), a long-time sequence prediction model (e.g., a long-sequence prediction model of an infomer transconverter), a prediction model of a deep decomposition architecture (e.g., a long-time sequence prediction model of an AutoFormer based on a deep decomposition structure and an autocorrelation mechanism), etc.; a model constructed based on a recurrent neural network (Recurrent Neural Network, abbreviated RNN) may include: a Long Short-Term Memory artificial neural network (LSTM), a portal cyclic neural network (Gate Recurrent Unit, GRU), a time convolution network (Temporal Convolutional Network TCN), etc.
In an alternative scheme of this embodiment, in order to improve the accuracy of the processing system and improve the accuracy of the predicted target prediction result, when the data of the target data sequence is predicted according to the target feature, a specific prediction model may be generated by using the deep neural network according to the target feature, for example, the timestamp feature, the hysteresis feature, the difference feature, the binning feature, the alignment feature, and the like, and then the data of the target time sequence is predicted by using the specific preset model, so as to obtain the target prediction result.
In the embodiment of the application, a processing system comprising a data input module, a data preprocessing module, a characteristic engineering module and a data prediction module is adopted, and the data input module is utilized to acquire an original time sequence; performing data preprocessing on the original time sequence by using a data preprocessing module to obtain a target time sequence; extracting features of the target time sequence by utilizing feature engineering to obtain target features of the target time sequence; the method comprises the steps of carrying out data prediction on a target time sequence based on target characteristics by using a data prediction module to obtain a target prediction result of an original time sequence, carrying out data preprocessing on the original time sequence, filling sequence values which are missing in the original time sequence to obtain the target time sequence which is close to an actual value, improving the processing capacity of a processing system of the time sequence on the original time sequence, extracting target characteristics of the target time sequence, carrying out data prediction processing on the target time sequence based on the target characteristics to obtain the target prediction result of the original time sequence, further ensuring the rationality of the target prediction result, further improving the accuracy of the predicted power data, and further solving the technical problem that the power prediction accuracy is lower due to the influence of sensing errors, acquisition faults and the like on the power time sequence in the related technology.
In the above embodiments of the present application, the data processing module includes: the period detection unit is used for carrying out period detection on the original time sequence to obtain a period detection result of the original time sequence, wherein the period detection result is used for representing whether the original time sequence is a period sequence or not and the period of the original time sequence; and the data filling unit is used for filling the missing value in the original time sequence based on the period detection result to obtain a target time sequence, wherein the missing value is used for representing the sequence value with the missing in the original time sequence.
Generally, the structures of the original time sequences under different application scenes and different application conditions are different, and the power prediction system is taken as a target system for example, in different power scenes, for example, different new energy sources for generating power, different bus loads, different system loads and the like, the historical power data generated by the power prediction system are different, and the corresponding original time sequences are also different; even though the power scenario is the same, the historical power data generated by the power prediction system is different under different power application conditions, such as different geographic locations, different purposes of use, etc.
For example, in the same geographic location, the power generation curves generated by photovoltaic power generation and wind power generation are different, and the corresponding generated power data is also different.
Therefore, in order to improve the accuracy of predicting future data, which may be generated by the processing system in a future period by using the original time sequence, the original time sequence may be firstly analyzed and cleaned to improve the robustness of the process of predicting the future data by using the original time sequence, so as to improve the accuracy of the predicted future data, where the analysis of the exploratory data may refer to the analysis of the original data corresponding to the original time sequence by using the data visualization and the data value statistics, so as to determine whether the original time sequence has information such as a period, a number of missing values, and whether the original data has a trend of change, so as to facilitate the subsequent efficiency of cleaning the original time sequence.
Specifically, the processing system can utilize the period detection unit to determine sequence properties such as sequence period, data missing value, data change trend and the like of different original time sequences, decompose and correct the original time sequences by utilizing the sequence properties, and perform structure normalization processing on the original time sequences with different structures, so that the efficiency of the processing system for predicting data by utilizing the original time sequences is improved.
And the original time sequence can be detected periodically, whether the original time sequence is a periodic sequence is judged, and if the original time sequence is the periodic sequence, the corresponding period is of a specific size, so that the period detection result of the original time sequence is obtained.
In an alternative of this embodiment, a preset period detection frame (for example, a RobustPeriod robust time sequence multi-period detection frame) may be used to perform period detection on an original time sequence, where an input of the RobustPeriod frame is the original time sequence, an output is a period value of the original time sequence, and if the original time sequence has no period, the period value may be 0; if the original time series has a period, the period value may be the period size of the original time series, for example, the period is 10, 12, 18, etc.
In an alternative scheme of this embodiment, if no sequence period exists in the original time sequence, the possible occurrence of this situation is that the period of the original time sequence is longer, the current original time sequence is shorter, and the processing system cannot effectively identify the period of the original time sequence, then it may be determined whether the original time sequence can continue to be extended, for example, whether the data amount of the original data in the current original time sequence is smaller, the data type is smaller, etc., if the original time sequence can continue to be extended, then the historical time period may be increased, so that the original time sequence is extended longer, and then the period detection is performed again on the original time sequence after the original time sequence is lengthened; if the expansion cannot be continued, it can be determined that the original time series has no period at this time.
In an alternative of this embodiment, as shown in the foregoing, since there may be an error in the process of generating and recording the historical data in the target system, there may be a missing value in the original time sequence corresponding to the error, and in order to avoid the missing value from affecting the data prediction process, the processing system may perform periodic detection on the original time sequence to obtain a periodic detection result, and further may further use a data filling unit to fill the missing value in combination with the periodic detection result, so as to improve that the target time sequence obtained by filling is closer to the real value.
In the above embodiments of the present application, the data filling unit is further configured to: under the condition that the period detection result is used for representing that the original time sequence is the period sequence, filling the missing value based on the position of the missing value in the original time sequence to obtain a target time sequence; and under the condition that the period detection result is used for representing that the original time sequence is not the period sequence, carrying out interpolation processing on the missing value based on the original time sequence to obtain a target time sequence.
In an alternative scheme of the embodiment, when the original time sequence is considered to be a periodic sequence, the change trend of the corresponding original data is regular, and the missing values are corrected and filled by using the rule, so that the filling efficiency can be improved to a great extent; when the original time sequence is not the periodic sequence, the corresponding change trend of the original data is random, so that the missing values can be directly corrected and filled when the missing values are corrected and filled, and the filling efficiency is improved.
In an alternative of this embodiment, in order to improve the efficiency of filling the missing values according to the variation trend, the processing system may perform trend detection on the original time sequence while performing period detection on the original time sequence, so as to determine whether the original data in the original time sequence has a variation trend, for example, monotonically increasing, monotonically decreasing, and fluctuating along a fixed value.
In an alternative scheme of this embodiment, in order to further improve the filling efficiency of the missing value, when the processing system determines that the original time sequence is the periodic sequence according to the period detection result, the position of the missing value in the original time sequence may be determined first, and the missing value may be filled according to the position and other original data in the period where the missing value is located and combined with the trend of the original data in the previous period and/or the subsequent period. For example, if the original time sequence has 3 periods, the current position of the missing value is the middle position of the 2 nd period, then the change trend of the 1 st period and/or the change trend of the 3 rd period can be determined, for example, the change trend is monotonically increased, then the processing system can fill the missing value according to other original data in the 2 nd period and the monotonically increased change trend, so as to obtain the complete target time sequence of the filled data.
When the processing system determines that the original time sequence is not the periodic sequence according to the periodic detection result, a preset interpolation algorithm can be directly utilized, other original data in the original time sequence are combined, and missing values are filled in an interpolation processing mode, so that the target time sequence is obtained.
In an alternative of this embodiment, the preset interpolation algorithm may include, but is not limited to: linear interpolation algorithms (e.g., linear interpolation algorithms), quadratic interpolation algorithms (e.g., quadratic interpolation algorithms). The specific interpolation process may refer to the relevant literature and will not be described in detail herein. The filling efficiency can be improved to a great extent by directly filling missing values in the original time sequence which is not the periodic sequence by using an interpolation algorithm, so that the efficiency of determining the target time sequence is improved.
In the above embodiments of the present application, the data filling unit is further configured to: determining whether the missing value is a target missing value based on the position of the missing value in the original time sequence, wherein the target missing value is used for representing a plurality of missing values adjacent to the position; under the condition that the missing value is a target missing value, filling the missing value based on a sequence value of a target period in the original time sequence to obtain a target time sequence, wherein the target period is used for representing a period before the period where the missing value is located; and under the condition that the missing value is not the target missing value, carrying out interpolation processing on the missing value based on the original time sequence to obtain a target time sequence.
The target missing value may be a missing value with a plurality of consecutive positions, for example, if there are 3 missing values currently, and the positions of the 3 missing values in the original time series are 4, 5, and 6, respectively, then it is indicated that the 3 missing values are the target missing values.
In an alternative scheme of this embodiment, since the complexity of filling a segment of continuous missing values and filling a single independent missing value is different, for example, when the original time sequence in which the missing values are located is a periodic sequence, the processing system needs to consider the rationality of the filling result of filling the missing values, and when the original time sequence in which the missing values are located is not a periodic sequence, the processing system can directly fill the missing values by using an interpolation algorithm, and does not need to consider factors such as a change trend, a change relation of the missing values with other original data, and the filling constraint is less.
Therefore, in order to improve the accuracy of the determined target time sequence, when the original time sequence is a periodic sequence, the processing system may first determine the type of the missing value, and determine whether the missing value is the target missing value, that is, whether positions corresponding to the missing values are continuous when the original time sequence has the missing values.
If the missing value is the target missing value, at this time, a periodic interpolation (periodic imputation) manner may be adopted, and according to the variation trend of the period preceding the period in which the missing value is located, in combination with other original data of the period in which the missing value is located, the missing value is filled, so as to ensure consistency of the filling result, and meanwhile, reduce the probability of occurrence of an outlier, which may be according to the foregoing example: and filling the missing value according to the position and other original data in the period of the missing value and combining the change trend of the original data in the previous period and/or the next period. Thereby improving the consistency, rationality and accuracy of the target time sequence obtained by filling.
In addition to filling the missing value according to the trend of the change of the other period, if the missing value is the target missing value, the missing value may be directly filled with the data identical to the missing value in the position of the current period in the previous period or the next period, or the missing value may be filled with an average value of the data of a plurality of other periods in the position, for example, if the time corresponding to the original time sequence is 3 days, the period is 1 day, and the current day 3 is 10:00-12:00, the data segment a is a missing value, then day 1, 10, can be directly utilized at this time: 00-12: data B of 00, or day 10, 2: 00-12:00, the missing value can be directly filled, or the average value of the data B and the data C can be utilized to fill the missing value, so that the rationality of the filled data is ensured.
If the missing value is not the target missing value, in order to improve the filling efficiency, the original time sequence can be directly processed by interpolation processing, for example, the missing value can be filled by an average value of values of two original data before and after the missing value, so that the consistency, the rationality and the accuracy of the determined target time sequence are ensured.
In the above embodiments of the present application, the data processing module further includes: the time sequence decomposition unit is used for splitting the original time sequence based on the period detection result to obtain time sequence elements of the original time sequence, wherein the time sequence elements at least comprise one of the following components: trend, period and residual terms of the original time sequence, the residual terms being used to characterize the time sequences of the original time sequence except the trend and period; the data conversion unit is used for detecting abnormal values of the residual error items, determining abnormal values in the original time sequence and replacing the abnormal values with missing values, wherein the abnormal values are used for representing abnormal sequence values in the original time sequence.
The above-mentioned timing elements may refer to data elements corresponding to the target system when generating the original data, and may include, but are not limited to: the original time sequence includes elements such as change period, change trend, residual error item and the like of original data.
The change period may be a period corresponding to original data having regular changes in the original time sequence, for example, a regular oscillation frequency generated in the power prediction system, for example, a time of day, week, month, etc.; the change trend may refer to the trend of the original data in the original time sequence, such as increase, decrease, etc., and may refer to the trend of the original data corresponding to different periods when the original time sequence is a period sequence, or may refer to the data which has no change period but is more coherent in the original data; the residual term may refer to relatively discrete data, i.e., outliers, among the original data of the original time series, except for the data corresponding to the period of variation and the trend of variation.
In an alternative scheme of this embodiment, an original time sequence may have a sequence formed by multiple elements of period, trend and residual, if the foregoing period detection algorithm, for example, robustPeriod framework, is directly used, the period detection result may be inaccurate, and thus the determined missing value may be inaccurate.
In an alternative aspect of the present embodiment, if the period detection result is that the original time sequence is a period sequence, the description at least includes: the period part with the period of change, the trend part with the trend of change and the residual term part with relatively discrete data, the time sequence elements obtained by splitting the original time sequence at this time can at least comprise: period, trend, and residual; if the period detection result is that the original time sequence is not the period sequence, the original time sequence at least comprises: the trend part with the trend of variation and the residual term part with the data relatively offline, but no period part with the period of variation, the time sequence obtained by differentiating the original time sequence at this time may initially at least include: trend and residual, without periodicity.
In an alternative scheme of this embodiment, since the original data in the original time sequence having a trend or a period of change has higher rationality and accuracy, no further missing value confirmation is required for the two parts, and since the change rule of the original data corresponding to the residual term cannot be accurately determined, after determining the residual term of the original time sequence, the processing system may use the data conversion unit to perform outlier detection on the residual term, and determine whether the original data corresponding to the residual term, that is, the sequence value, meets the condition of the original data generated by the target system, for example, whether the difference value between the original data and the front and rear data is within the preset range is reasonable. And converting the detected abnormal data into a missing value, while taking the abnormal data as the abnormal value.
In an alternative of this embodiment, a filter (e.g., hampel filter) may be used to detect an outlier of the residual term, that is, a rolling time sequence window may be established, and whether the residual term has an outlier may be determined by determining a statistical value, such as a median, of the residual term obtained by using the time sequence window, where the statistical value is reasonable
In the above embodiment of the present application, the timing decomposition unit is further configured to: under the condition that the period detection result is used for representing that the original time sequence is the period sequence, carrying out robust decomposition on the original time sequence to obtain a time sequence element, wherein the time sequence element comprises: trend, period, and residual terms; and under the condition that the period detection result is used for representing that the original time sequence is not the period sequence, carrying out robust trend filtering on the original time sequence value to obtain a time sequence element, wherein the time sequence element comprises: trend and residual terms.
In an alternative of this embodiment, in order to improve the effect of splitting the original time sequence, for example, improve the splitting efficiency and the accuracy of the splitting result, the original time sequence may be split in combination with whether the original time sequence is a periodic sequence. If the period detection result is a period sequence, it is indicated that multiple time sequence elements may exist in the original time sequence, and at this time, the processing system may perform robust decomposition on the original time sequence to improve the accuracy of the decomposed result, for example, a decomposition algorithm (for example, a RobustSTL high-robustness decomposition algorithm) is used to split the multiple time sequence elements of the original time sequence, that is, the variation trend, the variation period and the residual term of the original time sequence; if the period detection result is not the period sequence, it indicates that there are fewer time periods Xu Yuansu in the original time sequence, and at this time, the processing system may directly perform robust trend filtering on the original time sequence by using a filtering manner to improve the decomposition efficiency, for example, a trend filtering algorithm (for example, robustTrend high-robustness trend filtering algorithm) is used to filter the variation trend and residual term of the original data from the original time sequence.
In order to clearly show the above process of filling the missing values to obtain the target time sequence, fig. 4 is a schematic diagram of a time sequence processing process according to embodiment 1 of the present application, as shown in fig. 4, the processing system may first perform period detection on the original time sequence, then perform robust decomposition on the portion of the original time sequence where the period exists, to obtain time sequence elements such as trend, period, residual error, and the like, and then further select different filling manners to fill the missing values according to the type of the missing values. For example, if the missing value type is segment missing, that is, the positions of the missing values are continuous, then the missing values of the portion may be filled in the first filling manner at this time; if the missing value type is point missing, namely the position of the missing value is discrete, filling the missing value of the part according to a second filling mode; and carrying out robust trend filtering on the original time sequence part without the period to obtain trend, residual error and other time elements, and filling the missing value of the part by using a second filling mode to obtain the target time sequence.
In the above embodiments of the present application, the feature engineering module is further configured to: aligning the target time sequence with time points of the similar time sequence, determining a first time window in the similar time sequence, and carrying out statistical processing on sequence values in the first time window to obtain alignment features of the target time sequence, wherein geographic conditions and weather conditions of the similar time sequence are identical to those of the target time sequence, and time points of sequence values in the first time window are identical to time points of sequence values in the target time sequence; carrying out statistical processing on the sequence value of the second time window in the target time sequence to obtain a statistical result, and normalizing the sequence value of the target time sequence based on the statistical result to obtain a sample normalization feature of the target time sequence; the target feature is derived based at least on the alignment feature and the sample normalization feature.
The similar time sequence may refer to a time sequence generated by other systems having the same application scenario as the target system in a historical time period, for example, the other systems may refer to other same systems having the same city as the target system but different buses.
The geographical conditions and weather conditions may be, but are not limited to, conditions such as terrain, altitude, wind direction, and barometric pressure.
In an alternative of this embodiment, since there may be multiple alignment features and sample normalization features in one target time sequence, taking feature extraction of the target sequence by using multiple feature extraction algorithms to obtain multiple alignment features and sample normalization features as an example, a specific extraction process may include, but is not limited to:
for the alignment feature, the processing system may first align the target time sequence with the similar time sequence according to different time points of the historical time period, and determine the alignment feature of the target time sequence from the target time sequence and the similar time sequence by using a preset first time window, for example, may use the maximum value, the minimum value, the average value, and the like of the target data in the target time sequence and the similar data in the similar time sequence as the above alignment feature, thereby improving the correlation between the target time sequence and other similar time sequences, and further improving the accuracy of the original prediction result predicted by using the alignment feature.
For the sample normalization feature, a shorter second time window can be set, and the second time window is utilized to carry out statistics and normalization processing on sequence values in the target time sequence, for example, the average value, variance and other numerical values of a plurality of sequence values obtained by utilizing the second time window can be obtained, and the numerical values are used as the sample normalization feature, so that when the target prediction model predicts by utilizing the sample normalization feature, the predicted result is more stable, and the problem of distribution deviation of the predicted result is avoided.
In an alternative of this embodiment, the alignment feature and the sample normalization feature may be at least used as the target feature of the target time series, considering that the accuracy and the stability of the predicted result predicted from the alignment feature and the sample normalization feature are high.
In the above embodiment of the present application, the data prediction module includes: the algorithm module is used for carrying out data prediction on the target time sequence based on the target characteristics by utilizing the target prediction model to obtain an original prediction result of the original time sequence; and the post-processing module is used for carrying out post-processing on the original predicted result by utilizing a target post-processing algorithm to obtain a target predicted result.
The target prediction model may refer to a model that primarily predicts future data generated by the target system in a future period of time according to target characteristics of the target time sequence; the above-described original prediction result may refer to future data preliminarily predicted by the target prediction model, the prediction result corresponding to the original time series; the target post-processing algorithm may be an algorithm for adjusting an original prediction result to obtain future data with high precision, and may include, but is not limited to: gaussian Blur algorithm (Gaussian blue), visualization algorithm (Visualized Computing), etc.; the target prediction result may refer to future data with higher accuracy.
In an alternative scheme of this embodiment, when the processing system uses the data prediction module to predict data of the target time sequence according to the target feature, the processing system may first use the target prediction model to predict data of the target time sequence according to the target feature, so as to obtain the original prediction result. And then, performing post-processing on the original preset predicted result by using the target post-processing algorithm to obtain the target predicted result, thereby improving the accuracy of the predicted result.
In the above embodiment of the present application, the system further includes: and the interpretation module is used for interpreting the target prediction result by using a target prediction interpretation algorithm to obtain a prediction interpretation result of the original time sequence, wherein the prediction interpretation result is used for representing the influence degree of the target feature on the target prediction result.
The target prediction interpretation algorithm may be an algorithm for interpreting a target prediction result, and the algorithm may output contribution values corresponding to different target features in the target prediction result, that is, the target prediction interpretation result.
In an alternative scheme of this embodiment, the target prediction result may be interpreted by using the target prediction interpretation algorithm, so as to obtain the above prediction interpretation result, that is, the influence degree of the target feature on the target prediction result, so that the user can understand the process and the prediction result of the processing system for predicting the data of the target time sequence based on the prediction interpretation result.
In an alternative scheme of this embodiment, a feature-based algorithm (e.g., feature Attribution feature-based algorithm) may be used to interpret the target prediction result, that is, a reference time point may be selected from the target time sequence, and then the target feature in the reference time period approximates the predicted value in the target prediction result by means of a global random tree model (e.g., totally randomized tree global random tree) to determine the influence degree of the target feature on the target prediction result, so that the target prediction result has interpretability, and the user can understand the target processing result conveniently.
In the above embodiment of the present application, the system further includes: the human-computer interaction module is used for outputting a target prediction result and a prediction interpretation result and receiving feedback information corresponding to the target prediction result, wherein the feedback information is obtained by modifying the target prediction result based on the prediction interpretation result; and the model updating module is used for updating the target prediction model based on the feedback information.
In an alternative scheme of this embodiment, after the target prediction result and the prediction interpretation result corresponding to the target prediction result are obtained, the processing system may further use a man-machine interaction module to output the target prediction result and the corresponding prediction interpretation result to a display device preset in value, so as to facilitate the user to view. The user can adaptively modify the target prediction result according to the prediction interpretation result, for example, the target prediction result is displayed according to the prediction interpretation result, the target time sequence is mainly obtained by carrying out data prediction on the target time sequence according to the normalization feature, the relevance between the time sequence actually generated by the target system and the normalization feature is not high, the corresponding process of predicting the target prediction result does not take the normalization feature as the main part, therefore, the user can manually adjust the target prediction result according to the prediction interpretation result, and input feedback information generated when the target prediction result is adjusted to the processing system, and the processing system can receive the feedback information based on the man-machine interaction module.
After receiving the feedback information, the processing system may further utilize a model update module to update the target prediction model with the feedback coefficient, for example, to reduce the weight of the normalized feature in the prediction process, thereby improving the accuracy of the target prediction model.
In the above embodiments of the present application, the data input module is further configured to: acquiring original data generated by the operation of a target system in a historical time period; and carrying out data structure conversion on the original data to obtain an original time sequence.
The raw data may refer to time-related data generated by the target system, such as weather raw data, load raw data, maintenance data of the power equipment, and the like. It may also refer to static data of the target system, such as static topology information of the bus, installed capacity of the wind power plant, etc. Specific raw data may be set according to practical situations, and are not limited herein.
In an alternative of this embodiment, when the processing system acquires the above-mentioned original data, the processing system may acquire the data generated by the target system in the historical period, and then may perform data structure conversion on the original data to obtain the above-mentioned original time sequence.
In an alternative of this embodiment, when converting the structure of the original data, the above-mentioned original time sequence may be obtained by converting the structure of the original data into a data structure (for example, a DataFrame) by a preset database, for example, a database (for example, a pandas database) in a programming language (for example, a python programming language), and then converting the data from the DataFrame data structure into another data structure (for example, an efore database) in another database (for example, an energy dataset load decomposition data set), so as to improve the efficiency of the processing system for processing the original time sequence.
In order to clearly show the operation framework of the above-mentioned time-series processing system, fig. 5 (a) and 5 (b) are schematic diagrams of the time-series processing system framework according to embodiment 1 of the present application, and as shown in fig. 5 (a) and 5 (b), the framework of the entire system may be divided into six parts, specifically:
the first portion is an original time series acquisition portion for acquiring original data generated by the target system over a historical period of time and generating an original time series from the original data, wherein the original data may include, but is not limited to: power data, weather data, etc., are different for raw data acquired by different target systems.
The second part is a robust processing part, which is used for performing operations such as period detection, sequence decomposition, missing value filling and the like on the obtained original time sequence to obtain a target time sequence with complete data, and the specific operations can include but are not limited to: outlier correction, regularization, time sequence decomposition, period detection, automatic data filling, data conversion and the like.
The third part is a feature engineering part for extracting target features of a target time series, wherein the target features may include, but are not limited to: alignment features, sample normalization features, time stamp features, hysteresis features, difference features, binning features, etc.
The fourth part is a data prediction part, which is used for performing data prediction on the target time sequence by using the target prediction model and the target characteristics to obtain an original prediction result, wherein, as shown in the foregoing, the statistical machine learning algorithm can include but is not limited to: LGBM, XGB, catBoost, SVR, GAM, etc., the deep learning algorithm may include, but is not limited to: fedFormer, quatFormer, LSTM, GRU, TCN, etc.
The fifth part is a post-processing part for post-processing predicted future data of future time, i.e. original predicted results, by using a target post-processing algorithm in combination with the operation condition of the target system to improve the accuracy of the target predicted results, wherein the operation condition that can be considered may include, but is not limited to: extreme weather, holidays, etc.
The sixth section is a prediction interpretation section for determining a degree of influence of the target feature on the target prediction result, and in determining the degree of influence, the operations performed may include, but are not limited to: feature attribution, dependency partitioning, feature interaction, interpretation with additive interpretation algorithms (Shapley Additive exPlanations, abbreviated as bskap), gradient integration, etc.
The entire processing system may output applications in different scenarios, which may include, but are not limited to: bus load prediction, system load prediction, new energy power prediction and the like.
It should be noted that, the data (including, but not limited to, data for analysis, stored data, displayed data, etc.) referred to in the present application are all information and data authorized by the user or fully authorized by each party, and the collection, use and processing of the related data need to comply with the related laws and regulations and standards of the related country and region, and are provided with corresponding operation entries for the user to select authorization or rejection.
It should be noted that, for simplicity of description, the foregoing system embodiments are all illustrated as a series of acts, but it should be understood by those skilled in the art that the present application is not limited by the order of acts described, as some steps may be performed in other orders or concurrently in accordance with the present application. Further, those skilled in the art will also appreciate that the embodiments described in the specification are all preferred embodiments, and that the acts and modules referred to are not necessarily required in the present application.
From the description of the above embodiments, it will be clear to a person skilled in the art that the system according to the above embodiments may be implemented by means of software plus a necessary general hardware platform, but may also be implemented by means of hardware. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk), comprising several instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to execute the system of the embodiments of the present application.
Example 2
There is also provided in accordance with an embodiment of the present application a time-series processing method, it being noted that the steps shown in the flowcharts of the figures may be performed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowcharts, in some cases the steps shown or described may be performed in an order other than that shown or described herein.
Fig. 6 is a flowchart of a time-series processing method according to embodiment 2 of the present application, and as shown in fig. 6, the method may include the steps of:
Step S602, an original time sequence is acquired.
Wherein the original time series is used to characterize the time series produced by the target system operating over a historical period of time.
The target system may refer to a system capable of generating and recording historical data and predicting future data, and may include, but is not limited to: power systems, wind systems, etc. The historical time period may be a time period between the start of the target system and the current time, or may be a time period before the current time. The original time sequence may be a time sequence that can be obtained and is constructed from original data generated by the power system in a preset historical time period, i.e., the historical data.
In an alternative of this embodiment, in order to improve the accuracy of the predicted power time sequence of the target system in a future period, the processing system may first obtain historical system data generated by the power prediction system in a preset historical period, and construct a corresponding time sequence, that is, the original time sequence, according to the historical power data.
Step S604, data preprocessing is performed on the original time sequence to obtain a target time sequence.
And filling the sequence value with the deletion in the original time sequence by the target time sequence formula to obtain the sequence.
The above-described target time series may refer to a time series capable of reflecting an actual value or a target value of data generated when the target system operates in a history period, and the target data in the target time series is more accurate than the original data in the original time series.
In an alternative scheme of this embodiment, since errors may exist in the process of generating and recording the historical data by the target system, for example, the generated historical data is abnormal, the recorded historical data is time-erroneous, and the data corresponding to these errors is usually deleted by the system or the user, at this time, corresponding missing values will be generated in the original time sequence, and since the missing values will also affect the accuracy of the prediction result of the processing system, in order to avoid the influence of the missing values in the original data on the process of predicting the data, for example, the efficiency of data prediction, the accuracy of the prediction result, etc., after the original time sequence is obtained, the data preprocessing may be further performed on the original time sequence, that is, the missing sequence values in the original time sequence are filled, so as to obtain the complete time sequence, that is, the target time sequence.
Step S606, extracting features of the target time sequence to obtain target features of the target time sequence.
The target feature may be a feature capable of clearly expressing a time series element of the target time series.
In an alternative of this embodiment, considering that the above-mentioned target feature can clearly represent the time sequence element of the target time sequence, in order to improve the accuracy of the data predicted by the target time sequence, the target system may perform feature extraction on the target time sequence by using feature engineering, obtain the target feature of the target time sequence, and apply the target feature to the data prediction process performed on the target time sequence.
Step S608, performing data prediction on the target time sequence based on the target feature to obtain a target prediction result of the original time sequence.
The target prediction result is used for representing a time sequence in a preset time period after the current moment.
The preset time period may refer to the foregoing future time period, may be a time period automatically generated by the processing system according to the operation condition of the target system, for example, the operation time, the data generating efficiency, etc., or may be a time period set by the user, specifically may be according to the actual condition, and is not limited herein specifically.
In an alternative of this embodiment, after obtaining the target time sequence and the corresponding target feature, the data prediction may be performed on the target time sequence according to the target feature, so as to obtain a future time sequence that may be generated by the target system in a preset time period after the current time, that is, the target prediction result.
It should be noted that, the preferred embodiments in the foregoing examples of the present application are the same as the embodiments provided in example 1, the application scenario and the implementation process, but are not limited to the embodiments provided in example 1.
Example 3
There is also provided in accordance with an embodiment of the present application a time-series processing method, it being noted that the steps shown in the flowcharts of the figures may be performed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowcharts, in some cases the steps shown or described may be performed in an order other than that shown or described herein.
Fig. 7 is a flowchart of a time-series processing method according to embodiment 3 of the present application, and as shown in fig. 7, the method may include the steps of:
step S702, the operation process of the power system is monitored, and the original time sequence is displayed on the operation interface.
Wherein the raw time series is used to characterize the time series produced by the power system over a historical period of time.
The original time sequence may be a time sequence that can be obtained and is constructed from original data generated by the power system in a preset historical time period, i.e., the historical data.
In an alternative of this embodiment, the processing system may monitor the operation of the power system, for example, the power data generated by the power system during a preset historical period, and generate the corresponding original time sequence according to the power data.
In an alternative of this embodiment, to facilitate the user's viewing of the original time series, the processing system may display the original time series in a preset operation interface.
In step S704, in response to the prediction instruction acting on the operation interface, the target prediction result of the original time series is displayed on the operation interface.
The target prediction result is used for representing a time sequence in a preset time period after the current moment, the target prediction result is obtained by carrying out data prediction on the target time sequence based on target characteristics of the target time sequence, the target time sequence is obtained by carrying out data preprocessing on the basis of the original time sequence, and the target characteristics are obtained by carrying out characteristic extraction on the target time sequence.
The prediction instruction may be an instruction sent by a user to the processing system to predict data generated by the power system in a future period of time.
After receiving the prediction instruction and the original time sequence, the operating system may first perform data preprocessing on the original time sequence, and fill the sequence value with the missing in the original time sequence, so as to obtain the target time sequence. And then extracting the characteristics of the target time sequence, determining the target characteristics of the target time sequence, and predicting the data of the target time sequence by utilizing the target characteristics to obtain a target prediction result in a preset time period in the future.
It should be noted that, the preferred embodiments in the foregoing examples of the present application are the same as the embodiments provided in example 1, the application scenario and the implementation process, but are not limited to the embodiments provided in example 1.
Example 4
There is also provided in accordance with an embodiment of the present application a time-series processing method, it being noted that the steps shown in the flowcharts of the figures may be performed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowcharts, in some cases the steps shown or described may be performed in an order other than that shown or described herein.
Fig. 8 is a flowchart of a time-series processing method according to embodiment 4 of the present application, and as shown in fig. 8, the method may include the steps of:
step S802, the original time sequence is obtained by calling the first interface.
The first interface comprises a first parameter, wherein the parameter value of the first parameter is an original time sequence, and the original time sequence is used for representing a time sequence generated by the operation of the target system in a historical time period.
The first interface may be referred to as sending an instruction to the target system to acquire the original time sequence.
In an alternative of this embodiment, the processing system may send an instruction to the target system to acquire the original time sequence by calling a preset first interface.
Step S804, data preprocessing is carried out on the original time sequence to obtain a target time sequence.
And filling the sequence value with the deletion in the original time sequence by the target time sequence formula to obtain the sequence.
In an alternative scheme of this embodiment, since errors may exist in the process of generating and recording the historical data, for example, the generated historical data is abnormal, the recorded historical data is wrong in time, and the data with errors or abnormalities usually affect the accuracy of the prediction result of the processing system, generally, the processing system may remove or manually delete the detected errors, that is, the data with errors or abnormalities, at the time point corresponding to the data with errors or abnormalities in the original time sequence, the corresponding data value is lacking, that is, the missing value also affects the accuracy of the prediction result of the processing system, so, in order to avoid the influence of the missing value in the original data on the process of predicting the data, for example, the efficiency of data prediction, the accuracy of the prediction result, and the like, after the original time sequence is obtained, the data preprocessing may be further performed on the original time sequence, that is, to fill the sequence value with the missing in the original time sequence, so as to obtain the complete time sequence, that is, the target time sequence.
Step S806, extracting features of the target time sequence to obtain target features of the target time sequence.
Wherein the missing values are used to characterize the sequence values in the original time sequence where the missing occurred.
In an alternative of this embodiment, considering that the above-mentioned target feature can clearly represent the time sequence element of the target time sequence, in order to improve the accuracy of the data predicted by the target time sequence, the target system may perform feature extraction on the target time sequence by using feature engineering, obtain the target feature of the target time sequence, and apply the target feature to the data prediction process performed on the target time sequence.
Step S808, data prediction is carried out on the target time sequence based on the target characteristics, and a target prediction result of the original time sequence is obtained.
After the target time sequence is obtained, the processing system may perform data prediction on the target time sequence according to the target feature, so as to obtain the target prediction result.
Step S810, outputting a target prediction result by calling a second interface.
The second interface comprises a second parameter, wherein a parameter value of the second parameter is a target prediction result, and the target prediction result is used for representing a time sequence within a preset time period after the current moment.
The second interface may be an interface for sending, by the processing system, the target prediction result and the corresponding display instruction to the preset display device.
In order to facilitate the user to check the target prediction result, the processing system can output the target prediction result to the display device with preset output value for the user to check after determining the target prediction result by calling the second interface.
It should be noted that, the preferred embodiments in the foregoing examples of the present application are the same as the embodiments provided in example 1, the application scenario and the implementation process, but are not limited to the embodiments provided in example 1.
Example 5
According to an embodiment of the present application, there is also provided a time-series processing apparatus for implementing the above-mentioned time-series processing method, where the apparatus may be deployed in a target client. Fig. 9 is a block diagram of a processing apparatus for generated power according to embodiment 5 of the present application, and as shown in fig. 9, the apparatus 900 includes: the sequence monitoring module 902 and the result display module 904.
The sequence monitoring module 1002 is configured to monitor an operation process of the power system, and display an original time sequence on an operation interface, where the original time sequence is used to characterize a time sequence generated by the power system in a historical time period; the result display module 1004 is configured to display, on the operation interface, a target prediction result of the original time sequence in response to a prediction instruction acting on the operation interface, where the target prediction result is used to characterize a time sequence within a preset time period after a current time, the target prediction result is obtained by performing data prediction on the target time sequence based on a target feature of the target time sequence, the target time sequence is obtained by performing data preprocessing on the original time sequence, and the target feature is obtained by performing feature extraction on the target time sequence.
It should be noted that, the sequence monitoring module 902 and the result displaying module 904 correspond to steps S702 to S704 in embodiment 2, and the two modules are the same as the examples and application scenarios implemented by the corresponding steps, but are not limited to those disclosed in embodiment 1. It should be noted that the above-mentioned modules or units may be hardware components or software components stored in a memory (e.g., the memory 104) and processed by one or more processors (e.g., the processors 102a,102b, … …,102 n), or the above-mentioned modules may also be executed as part of the apparatus in the computer terminal 12 provided in embodiment 1.
Example 6
According to an embodiment of the present application, there is also provided a time-series processing apparatus for implementing the above-mentioned time-series processing method, where the apparatus may be deployed in a target client. Fig. 10 is a block diagram of a processing apparatus for generated power according to embodiment 6 of the present application, and as shown in fig. 10, the apparatus 1000 includes: the system comprises a sequence acquisition module 1002, a data preprocessing module 1004, a target feature acquisition module 1006, a data prediction module 1008 and a result output module 1010.
The sequence obtaining module 1002 is configured to obtain an original time sequence by calling a first interface, where the first interface includes a first parameter, and a parameter value of the first parameter is the original time sequence, and the original time sequence is used to characterize a time sequence generated by the target system running in a historical time period; the data preprocessing model 1004 is configured to perform data preprocessing on an original time sequence to obtain a target time sequence, where the target time sequence fills a sequence value with a deletion in the original time sequence to obtain a sequence; a target feature obtaining module 1006, configured to perform feature extraction on the target time sequence to obtain a target feature of the target time sequence; the data prediction module 1008 is configured to perform data prediction on the target time sequence based on the target feature, so as to obtain a target prediction result of the original time sequence; the result output module 1010 is configured to output a target prediction result by calling a second interface, where the second interface includes a second parameter, and a parameter value of the second parameter is the target prediction result, and the target prediction result is used to characterize a time sequence within a preset time period after the current time.
Here, the sequence obtaining module 1002, the data preprocessing module 1004, the target feature obtaining module 1006, the data predicting module 1008, and the result outputting module 1010 correspond to steps S802 to S810 in embodiment 4, and the five modules are the same as the examples and application scenarios implemented by the corresponding steps, but are not limited to those disclosed in embodiment 1. It should be noted that the above-mentioned modules or units may be hardware components or software components stored in a memory (e.g., the memory 104) and processed by one or more processors (e.g., the processors 102a,102b, … …,102 n), or the above-mentioned modules may also be executed as part of the apparatus in the computer terminal 12 provided in embodiment 1.
Example 7
Embodiments of the present application also provide a computer-readable storage medium. In the above-described embodiment of the present application, in this embodiment, the above-described computer-readable storage medium may be used to store the program code executed by the processing system of the time series provided in the above-described embodiment 1.
In the foregoing embodiments of the present application, in this embodiment, the computer-readable storage medium may be located in any one of a group of computer terminals in a computer network, or in any one of a group of mobile terminals.
In the above-described embodiments of the present application, in the present embodiment, the storage medium is configured to store program code for performing the steps of: the data input module is used for acquiring an original time sequence, wherein the original time sequence is used for representing a time sequence generated by the operation of the target system in a historical time period; the data preprocessing module is used for preprocessing data of the original time sequence to obtain a target time sequence, wherein the target time sequence fills a sequence value with a deletion in the original time sequence to obtain a sequence; the feature engineering module is used for carrying out feature construction on the target time sequence to obtain target features of the target time sequence; the data prediction module is used for carrying out data prediction on the target time sequence based on the target characteristics to obtain a target prediction result of the original time sequence, wherein the target prediction result is used for representing the time sequence in a preset time period after the current moment.
In the above embodiment of the present application, the above storage medium may further execute program codes of the following steps: the data processing module comprises: the period detection unit is used for carrying out period detection on the original time sequence to obtain a period detection result of the original time sequence, wherein the period detection result is used for representing whether the original time sequence is a period sequence or not and the period of the original time sequence; and the data filling unit is used for filling the missing value in the original time sequence based on the period detection result to obtain a target time sequence, wherein the missing value is used for representing the sequence value with the missing in the original time sequence.
In the above embodiment of the present application, the above storage medium may further execute program codes of the following steps: the data stuffing unit is also used for: under the condition that the period detection result is used for representing that the original time sequence is the period sequence, filling the missing value based on the position of the missing value in the original time sequence to obtain a target time sequence; and under the condition that the period detection result is used for representing that the original time sequence is not the period sequence, carrying out interpolation processing on the missing value based on the original time sequence to obtain a target time sequence. In the above embodiment of the present application, the above storage medium may further execute program codes of the following steps: the data stuffing unit is also used for: determining whether the missing value is a target missing value based on the position of the missing value in the original time sequence, wherein the target missing value is used for representing a plurality of missing values adjacent to the position; under the condition that the missing value is a target missing value, filling the missing value based on a sequence value of a target period in the original time sequence to obtain a target time sequence, wherein the target period is used for representing a period before the period where the missing value is located; and under the condition that the missing value is not the target missing value, carrying out interpolation processing on the missing value based on the original time sequence to obtain a target time sequence.
In the above embodiment of the present application, the above storage medium may further execute program codes of the following steps: the data processing module further includes: the time sequence decomposition unit is used for splitting the original time sequence based on the period detection result to obtain time sequence elements of the original time sequence, wherein the time sequence elements at least comprise one of the following components: trend, period and residual terms of the original time sequence, the residual terms being used to characterize the time sequences of the original time sequence except the trend and period; the data conversion unit is used for detecting abnormal values of the residual error items, determining abnormal values in the original time sequence and replacing the abnormal values with missing values, wherein the abnormal values are used for representing abnormal sequence values in the original time sequence.
In the above embodiment of the present application, the above storage medium may further execute program codes of the following steps: the time sequence decomposition unit is also used for: under the condition that the period detection result is used for representing that the original time sequence is the period sequence, carrying out robust decomposition on the original time sequence to obtain a time sequence element, wherein the time sequence element comprises: trend, period, and residual terms; and under the condition that the period detection result is used for representing that the original time sequence is not the period sequence, carrying out robust trend filtering on the original time sequence value to obtain a time sequence element, wherein the time sequence element comprises: trend and residual terms.
In the above embodiment of the present application, the above storage medium may further execute program codes of the following steps: the feature engineering module is also for: aligning the target time sequence with time points of the similar time sequence, determining a first time window in the similar time sequence, and carrying out statistical processing on sequence values in the first time window to obtain alignment features of the target time sequence, wherein geographic conditions and weather conditions of the similar time sequence are identical to those of the target time sequence, and time points of sequence values in the first time window are identical to time points of sequence values in the target time sequence; carrying out statistical processing on the sequence value of the second time window in the target time sequence to obtain a statistical result, and normalizing the sequence value of the target time sequence based on the statistical result to obtain a sample normalization feature of the target time sequence; the target feature is derived based at least on the alignment feature and the sample normalization feature.
In the above embodiment of the present application, the above storage medium may further execute program codes of the following steps: the data prediction module comprises: the algorithm module is used for carrying out data prediction on the target time sequence based on the target characteristics by utilizing the target prediction model to obtain an original prediction result of the original time sequence; and the post-processing module is used for carrying out post-processing on the original predicted result by utilizing a target post-processing algorithm to obtain a target predicted result.
In the above embodiment of the present application, the above storage medium may further execute program codes of the following steps: the system further comprises: and the interpretation module is used for interpreting the target prediction result by using a target prediction interpretation algorithm to obtain a prediction interpretation result of the original time sequence, wherein the prediction interpretation result is used for representing the influence degree of the target feature on the target prediction result.
In the embodiment of the application, a processing system comprising a data input module, a data preprocessing module, a characteristic engineering module and a data prediction module is adopted, and the data input module is utilized to acquire an original time sequence; performing data preprocessing on the original time sequence by using a data preprocessing module to obtain a target time sequence; extracting features of the target time sequence by utilizing feature engineering to obtain target features of the target time sequence; the method comprises the steps of carrying out data prediction on a target time sequence based on target characteristics by using a data prediction module to obtain a target prediction result of an original time sequence, carrying out data preprocessing on the original time sequence, filling sequence values which are missing in the original time sequence to obtain the target time sequence which is close to an actual value, improving the processing capacity of a processing system of the time sequence on the original time sequence, extracting target characteristics of the target time sequence, carrying out data prediction processing on the target time sequence based on the target characteristics to obtain the target prediction result of the original time sequence, further ensuring the rationality of the target prediction result, further improving the accuracy of the predicted power data, and further solving the technical problem that the power prediction accuracy is lower due to the influence of sensing errors, acquisition faults and the like on the power time sequence in the related technology.
Example 8
Embodiments of the present application may provide an electronic device, which may be any one of a group of electronic devices. In the above embodiment of the present application, in this embodiment, the electronic device may be replaced by a terminal device such as a mobile terminal.
In the above embodiment of the present application, in this embodiment, the electronic device may be located in at least one network device of a plurality of network devices of the computer network.
In this embodiment, the electronic device may execute the program code of the following steps in the time-series processing system: the data input module is used for acquiring an original time sequence, wherein the original time sequence is used for representing a time sequence generated by the operation of the target system in a historical time period; the data preprocessing module is used for preprocessing data of the original time sequence to obtain a target time sequence, wherein the target time sequence fills a sequence value with a deletion in the original time sequence to obtain a sequence; the feature engineering module is used for carrying out feature construction on the target time sequence to obtain target features of the target time sequence; the data prediction module is used for carrying out data prediction on the target time sequence based on the target characteristics to obtain a target prediction result of the original time sequence, wherein the target prediction result is used for representing the time sequence in a preset time period after the current moment.
In the above-described embodiments of the present application, fig. 11 is a block diagram of the structure of an electronic device according to embodiment 8 of the present application. As shown, the electronic device a may include: the processor 1102, the memory 1104, the communication interface and the communication bus, wherein the processor 1102, the memory 1104 and the communication interface complete communication with each other through the communication bus; the memory is used to store at least one executable instruction that causes the processor to execute the time-series processing system shown in embodiment 1.
In the above embodiment of the present application, as shown in fig. 11, the electronic device a may further include: the system comprises a storage controller and a peripheral interface, wherein the peripheral interface is connected with a radio frequency module, an audio frequency module and a display.
The memory may be used to store software programs and modules, such as program instructions/modules corresponding to the time-series processing system and method in the embodiments of the present application, and the processor executes the software programs and modules stored in the memory, thereby performing various functional applications and data processing, that is, implementing the time-series processing system described above. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory may further include memory remotely located with respect to the processor, which may be connected to terminal a through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
In the above embodiment of the present application, the processor may call the information and the application program stored in the memory through the transmission device to execute the following steps: the data processing module comprises: the period detection unit is used for carrying out period detection on the original time sequence to obtain a period detection result of the original time sequence, wherein the period detection result is used for representing whether the original time sequence is a period sequence or not and the period of the original time sequence; and the data filling unit is used for filling the missing value in the original time sequence based on the period detection result to obtain a target time sequence, wherein the missing value is used for representing the sequence value with the missing in the original time sequence.
In the above embodiment of the present application, the processor may call the information and the application program stored in the memory through the transmission device to execute the following steps: the data stuffing unit is also used for: under the condition that the period detection result is used for representing that the original time sequence is the period sequence, filling the missing value based on the position of the missing value in the original time sequence to obtain a target time sequence; and under the condition that the period detection result is used for representing that the original time sequence is not the period sequence, carrying out interpolation processing on the missing value based on the original time sequence to obtain a target time sequence. In the above embodiment of the present application, the above storage medium may further execute program codes of the following steps: the data stuffing unit is also used for: determining whether the missing value is a target missing value based on the position of the missing value in the original time sequence, wherein the target missing value is used for representing a plurality of missing values adjacent to the position; under the condition that the missing value is a target missing value, filling the missing value based on a sequence value of a target period in the original time sequence to obtain a target time sequence, wherein the target period is used for representing a period before the period where the missing value is located; and under the condition that the missing value is not the target missing value, carrying out interpolation processing on the missing value based on the original time sequence to obtain a target time sequence.
In the above embodiment of the present application, the processor may call the information and the application program stored in the memory through the transmission device to execute the following steps: the data processing module comprises: the time sequence decomposition unit is used for splitting the original time sequence based on the period detection result to obtain time sequence elements of the original time sequence, wherein the time sequence elements at least comprise one of the following components: trend, period and residual terms of the original time sequence, the residual terms being used to characterize the time sequences of the original time sequence except the trend and period; the data conversion unit is used for detecting abnormal values of the residual error items, determining abnormal values in the original time sequence and replacing the abnormal values with missing values, wherein the abnormal values are used for representing abnormal sequence values in the original time sequence.
In the above embodiment of the present application, the processor may call the information and the application program stored in the memory through the transmission device to execute the following steps: the time sequence decomposition unit is also used for: under the condition that the period detection result is used for representing that the original time sequence is the period sequence, carrying out robust decomposition on the original time sequence to obtain a time sequence element, wherein the time sequence element comprises: trend, period, and residual terms; and under the condition that the period detection result is used for representing that the original time sequence is not the period sequence, carrying out robust trend filtering on the original time sequence value to obtain a time sequence element, wherein the time sequence element comprises: trend and residual terms.
In the above embodiment of the present application, the processor may call the information and the application program stored in the memory through the transmission device to execute the following steps: the feature engineering module is also for: aligning the target time sequence with time points of the similar time sequence, determining a first time window in the similar time sequence, and carrying out statistical processing on sequence values in the first time window to obtain alignment features of the target time sequence, wherein geographic conditions and weather conditions of the similar time sequence are identical to those of the target time sequence, and time points of sequence values in the first time window are identical to time points of sequence values in the target time sequence; carrying out statistical processing on the sequence value of the second time window in the target time sequence to obtain a statistical result, and normalizing the sequence value of the target time sequence based on the statistical result to obtain a sample normalization feature of the target time sequence; the target feature is derived based at least on the alignment feature and the sample normalization feature.
In the above embodiment of the present application, the processor may call the information and the application program stored in the memory through the transmission device to execute the following steps: the data prediction module comprises: the algorithm module is used for carrying out data prediction on the target time sequence based on the target characteristics by utilizing the target prediction model to obtain an original prediction result of the original time sequence; and the post-processing module is used for carrying out post-processing on the original predicted result by utilizing a target post-processing algorithm to obtain a target predicted result.
In the above embodiment of the present application, the processor may call the information and the application program stored in the memory through the transmission device to execute the following steps: the system further comprises: and the interpretation module is used for interpreting the target prediction result by using a target prediction interpretation algorithm to obtain a prediction interpretation result of the original time sequence, wherein the prediction interpretation result is used for representing the influence degree of the target feature on the target prediction result.
In the embodiment of the application, a processing system comprising a data input module, a data preprocessing module, a characteristic engineering module and a data prediction module is adopted, and the data input module is utilized to acquire an original time sequence; performing data preprocessing on the original time sequence by using a data preprocessing module to obtain a target time sequence; extracting features of the target time sequence by utilizing feature engineering to obtain target features of the target time sequence; the method comprises the steps of carrying out data prediction on a target time sequence based on target characteristics by using a data prediction module to obtain a target prediction result of an original time sequence, carrying out data preprocessing on the original time sequence, filling sequence values which are missing in the original time sequence to obtain the target time sequence which is close to an actual value, improving the processing capacity of a processing system of the time sequence on the original time sequence, extracting target characteristics of the target time sequence, carrying out data prediction processing on the target time sequence based on the target characteristics to obtain the target prediction result of the original time sequence, further ensuring the rationality of the target prediction result, further improving the accuracy of the predicted power data, and further solving the technical problem that the power prediction accuracy is lower due to the influence of sensing errors, acquisition faults and the like on the power time sequence in the related technology.
It will be appreciated by those skilled in the art that the structure shown in fig. 11 is only illustrative, and the electronic device may be a terminal device such as a smart phone (e.g. an Android phone, an iOS phone, etc.), a tablet computer, a palm computer, a mobile internet device (Mobile Internet Devices, MID), a PAD, etc. Fig. 11 does not limit the structure of the electronic apparatus a. For example, electronic device A may also include more or fewer components (e.g., network interfaces, display devices, etc.) than shown in FIG. 11, or have a different configuration than shown in FIG. 11.
Those of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be implemented by a program for instructing a terminal device to execute in association with hardware, the program may be stored in a computer readable storage medium, and the storage medium may include: flash disk, read-Only Memory (ROM), random-access Memory (Random Access Memory, RAM), magnetic or optical disk, and the like.
In the foregoing embodiments of the present application, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.
In the several embodiments provided in the present application, it should be understood that the disclosed technology content may be implemented in other manners. The above-described embodiments of the apparatus are merely exemplary, and are merely a logical functional division, and there may be other manners of dividing the apparatus in actual implementation, for example, multiple units or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interfaces, units or modules, or may be in electrical or other forms.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in each embodiment of the present application may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be embodied in essence or a part contributing to the prior art or all or part of the technical solution, in the form of a software product stored in a storage medium, including several instructions to cause a computer device (which may be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the methods of the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a removable hard disk, a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The foregoing is merely a preferred embodiment of the present application and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present application and are intended to be comprehended within the scope of the present application.
Claims (14)
1. A time-sequential processing system, comprising:
the data input module is used for acquiring an original time sequence, wherein the original time sequence is used for representing a time sequence generated by the operation of the target system in a historical time period;
the data preprocessing module is used for preprocessing the data of the original time sequence to obtain a target time sequence, wherein the target time sequence is a sequence obtained by filling a sequence value with a deletion in the original time sequence;
the feature engineering module is used for carrying out feature construction on the target time sequence to obtain target features of the target time sequence;
and the data prediction module is used for carrying out data prediction on the target time sequence based on the target characteristics to obtain a target prediction result of the original time sequence, wherein the target prediction result is used for representing the time sequence in a preset time period after the current moment.
2. The system of claim 1, wherein the data processing module comprises:
the period detection unit is used for carrying out period detection on the original time sequence to obtain a period detection result of the original time sequence, wherein the period detection result is used for representing whether the original time sequence is a period sequence or not and the period of the original time sequence;
and the data filling unit is used for filling the missing value in the original time sequence based on the period detection result to obtain a target time sequence, wherein the missing value is used for representing the sequence value with the missing in the original time sequence.
3. The system of claim 2, wherein the data populating unit is further configured to:
filling the missing value based on the position of the missing value in the original time sequence to obtain the target time sequence under the condition that the period detection result is used for representing that the original time sequence is a period sequence;
and under the condition that the period detection result is used for representing that the original time sequence is not a period sequence, carrying out interpolation processing on the missing value based on the original time sequence to obtain the target time sequence.
4. A system according to claim 3, wherein the data populating unit is further configured to:
determining whether the missing value is a target missing value based on the position of the missing value in the original time sequence, wherein the target missing value is used for representing a plurality of missing values with adjacent positions;
filling the missing value based on a sequence value of a target period in the original time sequence to obtain the target time sequence under the condition that the missing value is the target missing value, wherein the target period is used for representing a period before the period where the missing value is located;
and under the condition that the missing value is not the target missing value, carrying out interpolation processing on the missing value based on the original time sequence to obtain the target time sequence.
5. The system of claim 2, wherein the data processing module further comprises:
the time sequence decomposition unit is used for splitting the original time sequence based on the period detection result to obtain time sequence elements of the original time sequence, wherein the time sequence elements at least comprise one of the following: a trend, a period, and a residual term of the original time series, the residual term being used to characterize a time series of the original time series other than the trend and the period;
And the data conversion unit is used for detecting the abnormal value of the residual error item, determining the abnormal value in the original time sequence and replacing the abnormal value with the missing value, wherein the abnormal value is used for representing the sequence value with the abnormal occurrence in the original time sequence.
6. The system of claim 5, wherein the timing decomposition unit is further configured to:
and under the condition that the period detection result is used for representing that the original time sequence is a period sequence, performing robust decomposition on the original time sequence to obtain the time sequence element, wherein the time sequence element comprises: the trend, the period, and the residual term;
and under the condition that the period detection result is used for representing that the original time sequence is not a period sequence, performing robust trend filtering on the original time sequence value to obtain the time sequence element, wherein the time sequence element comprises: the trend and the residual term.
7. The system of claim 1, wherein the feature engineering module is further configured to:
aligning the target time sequence with time points of a similar time sequence, determining a first time window in the similar time sequence, and carrying out statistical processing on sequence values in the first time window to obtain alignment features of the target time sequence, wherein geographic conditions and weather conditions of the similar time sequence are identical to those of the target time sequence, and time points of sequence values in the first time window are identical to time points of sequence values in the target time sequence;
Carrying out statistical processing on the sequence value of a second time window in the target time sequence to obtain a statistical result, and normalizing the sequence value of the target time sequence based on the statistical result to obtain a sample normalization feature of the target time sequence;
the target feature is derived based at least on the alignment feature and the sample normalization feature.
8. The system of claim 1, wherein the data prediction module comprises:
the algorithm module is used for carrying out data prediction on the target time sequence based on the target characteristics by utilizing a target prediction model to obtain an original prediction result of the original time sequence;
and the post-processing module is used for carrying out post-processing on the original predicted result by utilizing a target post-processing algorithm to obtain the target predicted result.
9. The system of claim 8, wherein the system further comprises:
and the interpretation module is used for interpreting the target prediction result by using a target prediction interpretation algorithm to obtain a prediction interpretation result of the original time sequence, wherein the prediction interpretation result is used for representing the influence degree of target characteristics on the target prediction result.
10. The system of claim 9, wherein the system further comprises:
the human-computer interaction module is used for outputting the target prediction result and the prediction interpretation result and receiving feedback information corresponding to the target prediction result, wherein the feedback information is obtained by modifying the target prediction result based on the prediction interpretation result;
and the model updating module is used for updating the target prediction model based on the feedback information.
11. A method of processing a time series, comprising:
acquiring an original time sequence, wherein the original time sequence is used for representing a time sequence generated by the operation of a target system in a historical time period;
performing data preprocessing on the original time sequence to obtain a target time sequence, wherein the target time sequence fills a sequence value with a deletion in the original time sequence to obtain a sequence;
performing feature construction on the target time sequence to obtain target features of the target time sequence;
and carrying out data prediction on the target time sequence based on the target characteristics to obtain a target prediction result of the original time sequence, wherein the target prediction result is used for representing the time sequence in a preset time period after the current moment.
12. A method of processing a time series, comprising:
monitoring the operation process of the power system and displaying an original time sequence on an operation interface, wherein the original time sequence is used for representing the time sequence generated by the power system in a historical time period;
and responding to a prediction instruction acted on the operation interface, and displaying a target prediction result of the original time sequence on the operation interface, wherein the target prediction result is used for representing the time sequence within a preset time period after the current moment, the target prediction result is obtained by carrying out data prediction on the target time sequence based on target characteristics of the target time sequence, the target time sequence is obtained by carrying out data preprocessing on the original time sequence, and the target characteristics are obtained by carrying out characteristic extraction on the target time sequence.
13. A method of processing a time series, comprising:
acquiring an original time sequence by calling a first interface, wherein the first interface comprises a first parameter, the parameter value of the first parameter is the original time sequence, and the original time sequence is used for representing a time sequence generated by the operation of a target system in a historical time period;
Performing data preprocessing on the original time sequence to obtain a target time sequence, wherein the target time sequence fills a sequence value with a deletion in the original time sequence to obtain a sequence;
extracting features of the target time sequence to obtain target features of the target time sequence;
performing data prediction on the target time sequence based on the target characteristics to obtain a target prediction result of the original time sequence;
and outputting the target prediction result by calling a second interface, wherein the second interface comprises a second parameter, the parameter value of the second parameter is the target prediction result, and the target prediction result is used for representing a time sequence within a preset time period after the current moment.
14. A computer readable storage medium, characterized in that the computer readable storage medium comprises a stored executable program, wherein the executable program when run controls a device in which the computer readable storage medium is located to perform the system and method of any one of claims 1 to 13.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310355943.5A CN116470485A (en) | 2023-03-31 | 2023-03-31 | Time-series processing system, method and computer-readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310355943.5A CN116470485A (en) | 2023-03-31 | 2023-03-31 | Time-series processing system, method and computer-readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116470485A true CN116470485A (en) | 2023-07-21 |
Family
ID=87184935
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310355943.5A Pending CN116470485A (en) | 2023-03-31 | 2023-03-31 | Time-series processing system, method and computer-readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116470485A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117745472A (en) * | 2023-12-21 | 2024-03-22 | 江苏省工程勘测研究院有限责任公司 | River management method and system based on lightweight sensing model |
-
2023
- 2023-03-31 CN CN202310355943.5A patent/CN116470485A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117745472A (en) * | 2023-12-21 | 2024-03-22 | 江苏省工程勘测研究院有限责任公司 | River management method and system based on lightweight sensing model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113626502B (en) | Power grid data anomaly detection method and device based on ensemble learning | |
| KR102361423B1 (en) | Artificial intelligence system and method for predicting maintenance demand | |
| CN116470485A (en) | Time-series processing system, method and computer-readable storage medium | |
| CN111932292A (en) | Cigarette product sales prediction method and device based on deep neural network | |
| CN115617606A (en) | Equipment monitoring method and system, electronic equipment and storage medium | |
| CN109643087B (en) | System and method for selecting controllable parameters for operational safety of a device | |
| CN113094284A (en) | Application fault detection method and device | |
| CN115577851A (en) | Energy consumption prediction method, device, equipment and storage medium | |
| CN114595113A (en) | Anomaly detection method and device in application system and anomaly detection function setting method | |
| CN113807951A (en) | Transaction data trend prediction method and system based on deep learning | |
| CN116435995A (en) | Time series processing method, computer readable storage medium and electronic device | |
| CN111614520B (en) | IDC flow data prediction method and device based on machine learning algorithm | |
| Javed et al. | Cloud-based collaborative learning (ccl) for the automated condition monitoring of wind farms | |
| CN115713044B (en) | Method and device for analyzing residual life of electromechanical equipment under multi-condition switching | |
| US20230376398A1 (en) | System and method for predicting remaining useful life of a machine component | |
| CN115118580B (en) | Alarm analysis method and device | |
| KR102557151B1 (en) | Method and device for improving quality of time series data | |
| CN115936185A (en) | Short-term power load and carbon emission prediction method and system based on DCNN-LSTM-AE-AM | |
| US20190080259A1 (en) | Method of learning robust regression models from limited training data | |
| CN111931798B (en) | Method for classifying and detecting cold head state and predicting service life | |
| CN114095390A (en) | Method and device for predicting object flow in area, electronic equipment and storage medium | |
| CN114546754A (en) | Automatic intelligent monitoring method and system and map data cloud platform | |
| Sharma et al. | The fundamentals and strategies of maintenance, repair, and overhaul (MRO) in Industry 4.0 | |
| CN113837358B (en) | System strategy prediction method based on Grangel causal relationship and related equipment | |
| CN117081666B (en) | Fault prediction method, device, electronic equipment, storage medium and program product |
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 |