CN114944831A - Multi-cycle time series data decomposition method, device, equipment and storage medium - Google Patents
Multi-cycle time series data decomposition method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN114944831A CN114944831A CN202210514494.XA CN202210514494A CN114944831A CN 114944831 A CN114944831 A CN 114944831A CN 202210514494 A CN202210514494 A CN 202210514494A CN 114944831 A CN114944831 A CN 114944831A
- Authority
- CN
- China
- Prior art keywords
- series data
- cycle time
- cycle
- period
- filter
- 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
- 238000000354 decomposition reaction Methods 0.000 title claims abstract description 91
- 238000000034 method Methods 0.000 title claims abstract description 51
- 230000000737 periodic effect Effects 0.000 claims description 77
- 230000002159 abnormal effect Effects 0.000 claims description 50
- 238000004422 calculation algorithm Methods 0.000 claims description 20
- 238000001228 spectrum Methods 0.000 claims description 17
- 230000015654 memory Effects 0.000 claims description 15
- 238000001914 filtration Methods 0.000 claims description 12
- 238000012216 screening Methods 0.000 claims description 12
- 230000008439 repair process Effects 0.000 claims description 7
- 230000005856 abnormality Effects 0.000 claims description 3
- 230000005611 electricity Effects 0.000 description 17
- 238000001514 detection method Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 238000007405 data analysis Methods 0.000 description 6
- 238000004891 communication Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000007774 longterm Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 125000004122 cyclic group Chemical group 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 238000013515 script Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H17/00—Networks using digital techniques
- H03H17/02—Frequency selective networks
- H03H17/0283—Filters characterised by the filter structure
- H03H17/0286—Combinations of filter structures
-
- 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
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- Mathematical Physics (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention discloses a method, a device, equipment and a storage medium for decomposing multi-cycle time series data, and belongs to the technical field of signal processing. The method comprises the following steps: acquiring original multi-cycle time sequence data and a plurality of cycle parameters thereof; obtaining a plurality of corresponding high-pass filters, a low-pass filter and an HP filter according to the plurality of period parameters; decomposing the original multi-cycle time sequence data by using a plurality of high-pass filters to obtain a plurality of cycle components based on the cut-off frequencies of the high-pass filters; resolving the trend component from the residual multi-cycle time series data by using an HP filter and a low-pass filter; and obtaining a multi-cycle time series data decomposition result based on the original multi-cycle time series data, the plurality of cycle components and the trend component. The method is different from the STL decomposition method, utilizes a plurality of different filters to sequentially decompose the original sequence in a linear mode without repeated cycle iteration, thereby improving the sequence decomposition efficiency.
Description
Technical Field
The present invention relates to the field of signal processing technologies, and in particular, to a method, an apparatus, a device, and a storage medium for decomposing multi-cycle time series data.
Background
In the related art, for the multicycle time series data, the multicycle time series data can be decomposed by using various methods such as a classical timing decomposition algorithm, a Holt-winter algorithm, and an STL (Standard and Trend decomposition use) algorithm, wherein the STL algorithm is the most common method at present.
However, when the STL algorithm decomposes multicycle time series data, it needs to repeatedly perform various operations such as trend removal, cycle removal, filtering, smoothing, etc. until the finally obtained result converges, and the calculation process is complicated and complex, and the decomposition efficiency is low. In daily life, the time series of many data have multi-periodicity qualities, such as household electricity consumption and office electricity consumption in the power data; network traffic and the like in network data and traffic volume and passenger flow and the like in traffic data, that is, in practical applications, a large amount of multi-cycle time series data may need to be decomposed for subsequent processing, so that if only the STL algorithm is used, the decomposition efficiency and the decomposition accuracy are low.
Disclosure of Invention
The invention mainly aims to provide a method, a device, equipment and a storage medium for decomposing multi-cycle time series data, and aims to solve the technical problem that the decomposition efficiency and the decomposition precision of the multi-cycle time series data in the prior art are low.
According to a first aspect of the present invention, there is provided a multi-cycle time series data decomposition method, the method comprising:
acquiring original multi-cycle time sequence data and a plurality of cycle parameters of the original multi-cycle time sequence data;
obtaining a plurality of corresponding high-pass filters, a low-pass filter and an HP filter according to the plurality of period parameters;
decomposing a plurality of periodic components from the original multi-period time series data by using a plurality of high-pass filters based on cut-off frequencies of the high-pass filters;
decomposing a trend component from the remaining multi-cycle time-series data after decomposing the plurality of cycle components by using the HP filter and the low-pass filter;
and obtaining a multi-cycle time series data decomposition result based on the original multi-cycle time series data, the plurality of cycle components and the trend component.
Optionally, the obtaining, according to the plurality of period parameters, a plurality of corresponding high-pass filters, a low-pass filter, and an HP filter includes:
determining a plurality of corresponding cut-off frequencies according to the plurality of period parameters;
obtaining a plurality of high-pass filters according to a plurality of cut-off frequencies;
screening out a target period parameter with the largest period from the plurality of period parameters;
and obtaining the low-pass filter and the HP filter according to the target period parameter.
Optionally, decomposing the original multi-cycle time series data into a plurality of cycle components by using a plurality of high-pass filters based on cut-off frequencies of the plurality of high-pass filters includes:
screening a high-pass filter with the maximum cut-off frequency from the plurality of high-pass filters as a target filter;
performing filter decomposition on the original multi-cycle time series data by using the target filter to separate out corresponding cycle components;
screening out a high-pass filter with the maximum cut-off frequency from the rest high-pass filters, and updating the target filter;
decomposing the residual sequence of the original multi-period time sequence data by using the updated target filter, and separating out corresponding period components;
and returning to execute the high-pass filter with the maximum cut-off frequency screened from the rest high-pass filters, and updating the target filter until the original multi-cycle time sequence data passes through all the high-pass filters to obtain a plurality of corresponding cycle components.
Optionally, after decomposing a plurality of periodic components from the original multi-period time series data by using a plurality of high-pass filters based on cut-off frequencies of the plurality of high-pass filters, the method includes:
screening out at least one to-be-tested periodic component with the period smaller than a preset periodic threshold value from the plurality of periodic components;
aiming at any one periodic component to be detected, obtaining a corresponding frequency spectrum component;
and screening abnormal values with the significance degree not within a preset significance degree range by utilizing a spectrum residual error algorithm according to the frequency spectrum components, and determining abnormal point positions corresponding to the abnormal values.
Optionally, the method further includes, after the abnormal value whose significance is not within a preset significance range is screened out by using a spectrum residual algorithm according to the spectrum component and the abnormal point position corresponding to the abnormal value is determined, the method further includes:
aiming at any abnormal value, a plurality of sequence values in a preset range are obtained by taking a corresponding abnormal point as a center;
replacing the abnormal value by using the average value of the sequence values to obtain a periodic component for repairing the abnormal value;
and updating the corresponding periodic component according to the periodic component for repairing the abnormality.
Optionally, the decomposing with the HP filter and the low-pass filter a trend component from the remaining multi-cycle time-series data after decomposing the plurality of cycle components includes:
denoising the residual multi-period time sequence data after decomposing the plurality of periodic components by using the low-pass filter;
and resolving the trend component from the residual multi-cycle time sequence data subjected to the denoising processing by using the HP filter.
Optionally, the obtaining a multi-cycle time series data decomposition result based on the original multi-cycle time series data, the plurality of cycle components and the trend component includes:
obtaining a residual error component according to the original multi-cycle time sequence data, the cycle component after abnormal repair and the trend component;
and obtaining the multi-cycle time series data decomposition result according to the cycle component after abnormal restoration, the trend component and the residual error component.
According to a second aspect of the present invention, there is provided a multicycle time series data decomposition apparatus comprising:
the system comprises a sequence acquisition module, a sequence acquisition module and a data processing module, wherein the sequence acquisition module is used for acquiring original multi-cycle time sequence data and a plurality of cycle parameters of the original multi-cycle time sequence data;
the filter generation module is used for obtaining a plurality of corresponding high-pass filters, a low-pass filter and an HP filter according to the plurality of period parameters;
the period decomposition module is used for decomposing the original multi-period time sequence data by using a plurality of high-pass filters to obtain a plurality of period components based on the cut-off frequencies of the high-pass filters;
a trend decomposition module for decomposing the residual multi-period time sequence data after decomposing the plurality of period components by using the HP filter and the low-pass filter to obtain a trend component;
and the result output module is used for obtaining a multi-cycle time series data decomposition result based on the original multi-cycle time series data, the plurality of cycle components and the trend component.
According to a third aspect of the present invention, there is provided a multicycle time series data decomposition device comprising: a memory, a processor and a multi-cycle time series data decomposition program stored on the memory and executable on the processor, the multi-cycle time series data decomposition program when executed by the processor implementing the steps set forth in any one of the possible implementations of the first aspect.
According to a fourth aspect of the present invention, there is provided a computer-readable storage medium having stored thereon a multi-cycle time-series data decomposition program which, when executed by a processor, implements the various steps set forth in any one of the possible implementations of the first aspect.
The embodiment of the invention provides a method, a device, equipment and a storage medium for decomposing multi-cycle time series data, wherein original multi-cycle time series data and a plurality of cycle parameters of the original multi-cycle time series data are obtained through multi-cycle time series data decomposing equipment; obtaining a plurality of corresponding high-pass filters, a low-pass filter and an HP filter according to the plurality of period parameters; decomposing a plurality of periodic components from the original multi-period time series data by using a plurality of high-pass filters based on cut-off frequencies of the high-pass filters; decomposing a trend component from the remaining multi-cycle time-series data after decomposing the plurality of cycle components by using the HP filter and the low-pass filter; and obtaining a multi-cycle time series data decomposition result based on the original multi-cycle time series data, the plurality of cycle components and the trend component.
The method is different from the condition that the decomposition efficiency and the decomposition precision of the multi-period time sequence data are lower in the prior art, the original multi-period time sequence data are sequentially decomposed in a linear mode by using various different filters, and the periodic components and the trend components of different periods can be decomposed without carrying out multiple cyclic iterations, so that the decomposition efficiency of the multi-period time sequence data is improved; meanwhile, various filters related to the invention are selected according to a plurality of sequence periods of the original multi-period time sequence data, so that different filters strictly correspond to multi-period time sequence data components with different periods or different frequencies, different period components in the original multi-period time sequence data can be decomposed strictly according to different cut-off frequencies by utilizing the filters, and the decomposition precision of the multi-period time sequence data is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a schematic structural diagram of a multi-cycle time series data decomposition device of a hardware operating environment according to an embodiment of the present invention;
FIG. 2 is a schematic flowchart of a method for decomposing multi-cycle time-series data according to a first embodiment of the present invention;
FIG. 3 is a detailed flowchart of the step S202 in FIG. 2 according to the present invention;
FIG. 4 is a detailed flowchart of the step S203 in FIG. 2 according to the present invention;
FIG. 5 is a schematic flow chart of the present invention after the step S203 in FIG. 2;
FIG. 6 is a detailed flowchart of the step S204 in FIG. 2 according to the present invention;
FIG. 7 is a detailed flowchart of the step S205 in FIG. 2 according to the present invention;
fig. 8 is a functional block diagram of a multi-cycle time-series data decomposition device according to an embodiment of the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The main solution of the embodiment of the invention is as follows: acquiring original multi-cycle time sequence data and a plurality of cycle parameters of the original multi-cycle time sequence data; obtaining a plurality of corresponding high-pass filters, a low-pass filter and an HP filter according to the plurality of period parameters; decomposing a plurality of periodic components from the original multi-period time series data by using a plurality of high-pass filters based on cut-off frequencies of the high-pass filters; decomposing a trend component from the remaining multi-cycle time-series data after decomposing the plurality of cycle components by using the HP filter and the low-pass filter; and obtaining a power data decomposition result based on the original multi-cycle time series data, the plurality of cycle components and the trend component.
In the related art, for the multi-cycle time sequence data, various methods such as a classical time sequence decomposition algorithm, a Holt-windows algorithm, and an STL (selective and Trend decomposition using loess) algorithm can be used to decompose the multi-cycle time sequence data, wherein the STL algorithm is the most common method at present. However, when the STL algorithm decomposes multicycle time series data, it needs to repeatedly perform various operations such as trend removal, cycle removal, filtering, smoothing, etc. until the finally obtained result converges, and the calculation process is complicated and complex, and the decomposition efficiency is low. In daily life, the time series of many data have multi-periodicity qualities, such as household electricity consumption and office electricity consumption in the power data; network traffic and the like in network data and traffic volume and passenger flow and the like in traffic data, that is, in practical applications, a large amount of multi-cycle time series data may need to be decomposed for subsequent processing, so that if only the STL algorithm is used, the decomposition efficiency and the decomposition accuracy are low.
The invention provides a solution, which is used for multi-period time sequence data decomposition equipment, original multi-period time sequence data are sequentially decomposed in a linear mode by utilizing a plurality of different filters, and periodic components and trend components of different periods can be decomposed without carrying out multiple cyclic iterations, so that the decomposition efficiency of the multi-period time sequence data is improved; meanwhile, various filters related to the invention are selected according to a plurality of sequence periods of the original multi-period time sequence data, so that different filters strictly correspond to multi-period time sequence data components with different periods or different frequencies, different period components in the original multi-period time sequence data can be decomposed strictly according to different cut-off frequencies by utilizing the filters, and the decomposition precision of the multi-period time sequence data is improved.
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Where "first" and "second" are used in the description and claims of embodiments of the invention to distinguish between similar elements and not necessarily for describing a particular sequential or chronological order, it is to be understood that such data may be interchanged where appropriate so that embodiments described herein may be implemented in other sequences than those illustrated or described herein.
Referring to fig. 1, fig. 1 is a schematic structural diagram of a multi-cycle time series data decomposition apparatus of a hardware operating environment according to an embodiment of the present invention.
As shown in fig. 1, the multi-cycle time-series data decomposition device may include: a processor 1001, such as a Central Processing Unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Wherein a communication bus 1002 is used to enable connective communication between these components. The user interface 1003 may include a Display screen (Display), an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a WIreless interface (e.g., a WIreless-FIdelity (WI-FI) interface). The Memory 1005 may be a high-speed Random Access Memory (RAM) or a Non-Volatile Memory (NVM), such as a disk Memory. The memory 1005 may alternatively be a storage device separate from the processor 1001.
Those skilled in the art will appreciate that the configuration shown in fig. 1 does not constitute a limitation of the multi-cycle time-series data decomposition device, and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.
As shown in fig. 1, the memory 1005, which is a storage medium, may include an operating system, a sequence acquisition module, a sequence processing module, a result output module, and a multi-cycle time series data decomposition program, wherein the sequence decomposition module may be further refined into a filter generation module, a cycle decomposition module, and a trend decomposition module.
In the multi-cycle time-series data decomposition device shown in fig. 1, the network interface 1004 is mainly used for data communication with a network server; the user interface 1003 is mainly used for data interaction with a user; the processor 1001 and the memory 1005 in the multicycle time series data decomposition device of the present invention may be provided in the multicycle time series data decomposition device, and the multicycle time series data decomposition device calls the multicycle time series data decomposition program stored in the memory 1005 through the processor 1001 and executes the multicycle time series data decomposition method provided by the embodiment of the present invention.
Based on the above hardware structure but not limited to the above hardware structure, the present invention provides a first embodiment of a method for decomposing multicycle time series data. Referring to fig. 2, fig. 2 is a schematic flow chart of a method for decomposing multi-cycle time series data according to a first embodiment of the present invention.
In this embodiment, the method includes:
step S201, acquiring original multi-cycle time sequence data and a plurality of cycle parameters of the original multi-cycle time sequence data;
in this embodiment, the execution subject is a multi-cycle time series data decomposition device, which may be a pc computer or a dedicated electronic device with a multi-cycle time series decomposition function, and this embodiment does not limit this. The original multi-cycle time series data can be a signal obtained by actively requesting a background database by the multi-cycle time series data decomposition device, or a signal obtained by passively receiving user input. It should be noted that the time sequence referred to in this embodiment is a time sequence with a multi-periodicity property, that is, the time sequence includes a plurality of period components with different periods, and corresponding period parameters are known and can be obtained by any period detection method, which is not limited in this embodiment; in practical applications, the original multi-cycle time series data can be various types of multi-cycle time series data, and can include multi-cycle power time series data such as household power consumption, office power consumption and the like; the multi-cycle network time series data such as network traffic and the like, and the multi-cycle traffic time series data such as vehicle traffic, passenger traffic and the like are explained below by taking the household electricity as an example, for the household electricity utilization situation, the electricity utilization situation in a period of time (such as a month, a quarter or a year and the like) can be observed according to the decomposition result, and further, the electricity utilization situation can be analyzed, abnormal values therein can be detected and repaired, or electricity data in a future period of time or at a certain time point can be predicted, and the like. It should be understood that, the present invention is only an example of the household electricity in the power data, and does not represent that the present invention can only be used for this purpose, as long as various time series related to multi-period components and long-term trend components (such as the network traffic, the vehicle traffic, the passenger traffic, and other time series with multi-periodicity), the present invention can be decomposed by using this method for subsequent analysis, and therefore, any time series with multi-periodicity is within the protection scope of the present invention.
Step S202, obtaining a plurality of corresponding high-pass filters, a low-pass filter and an HP filter according to the plurality of period parameters;
in this embodiment, for different periodic components and trend components contained in the original multi-period time series data, because the periods of the different components are different, that is, the frequencies are different, the components can be sequentially decomposed according to the frequency by using filters with different cut-off frequencies. Then based on this principle and thought, the corresponding filter needs to be obtained first.
Continuing to take the household electricity as an example, it can be understood that the use condition of the electricity in real life shows a certain periodicity in both small scale and large scale, for example, for the household electricity in one day, because most people are not at home for work, school, etc. in the daytime, the household electricity in the evening is often larger than the household electricity in the daytime, and after most people have a rest in the morning, the data is reduced; for example, for household electricity consumption in a week, similarly, most people are not at home for work, school, and the like in a weekday, so that the household electricity consumption on weekends is often larger than that on the weekday; for example, for household power utilization in one year, because electrical appliances such as air conditioners and fans are used more in summer and winter, the household power utilization in summer and winter is often larger than that in spring and autumn; it can be understood that it has similar properties for other multicycle time series data, such as more network traffic in the daytime, less traffic in the evening, more traffic in the home network on weekends, and more traffic in the office network on weekdays; and more traffic flow working days, less weekends and more holidays. Since the trend component is generally obtained based on a long-term trend and has a long period, the frequency of the periodic component is generally higher than that of the trend component. Therefore, each periodic component and each trend component can be decomposed based on the residual components, and the residual components are residual components which mainly consist of unstable factors such as noise and the like.
In a specific embodiment, referring to fig. 3, fig. 3 is a detailed flowchart of the step S202 in fig. 2, where the obtaining a plurality of corresponding high-pass filters, a low-pass filter and an HP filter according to a plurality of the period parameters includes:
step A10, determining a plurality of corresponding cut-off frequencies according to a plurality of the period parameters;
a step a20 of obtaining a plurality of high-pass filters according to a plurality of cut-off frequencies;
first, as for the periodic component, as described above, since the frequency of the periodic component is generally higher than that of the trend component, the decomposition of the periodic component may be performed using a high-pass filter in the present embodiment. Specifically, since the above-mentioned each period parameter in the original multi-period time series data has been acquired, the corresponding each cut-off frequency can be determined accordingly, in this embodiment, the cut-off frequency can be set to be the inverse of twice the corresponding period, for example, if a certain period is a, the corresponding cut-off frequency can be set to be 1/2a, and the cut-off frequency of the corresponding high-pass filter is 1/2a, that is, only the period component with the frequency higher than 1/2a can pass through the high-pass filter. It should be emphasized that the filter has a cutoff frequency of twice the reciprocal of the period, which is a recommended value, and in practical applications, a user may modify the filter according to actual needs, which is generally 1.5 to 3 times the reciprocal of the period.
Step A30, screening out a target period parameter with the largest period from the period parameters;
and A40, obtaining the low-pass filter and the HP filter according to the target period parameter.
Secondly, as for the trend component, as described above, since the frequency of the periodic component is generally higher than that of the trend component, in the embodiment, the corresponding low-pass filter and HP filter may be set to resolve the trend component, with the maximum period of the periodic component as a standard. Specifically, firstly, the reciprocal of two times of the maximum period is set as a low-pass cutoff frequency to obtain a corresponding low-pass filter, and then, in order to avoid that the component signal obtained through the low-pass filter is not smooth enough, the embodiment may further use the maximum period as a target period parameter to obtain a corresponding HP filter for performing secondary filtering to obtain a smooth trend component. The HP filter can remove the low-frequency components in the signal, and can separate out the long-term trend term, namely the trend component.
Step S203, decomposing the original multi-period time sequence data by using a plurality of high-pass filters to obtain a plurality of period components based on the cut-off frequencies of the high-pass filters;
after the plurality of high-pass filters are obtained, the high-pass filters can be used for decomposing different periodic components according to the frequency because the cut-off frequencies of the different high-pass filters are different.
In a specific embodiment, referring to fig. 3, fig. 3 is a schematic diagram illustrating a detailed flow of the step S203 in fig. 2, where the decomposing by using a plurality of high-pass filters to obtain a plurality of periodic components from the original multi-period time-series data based on the cut-off frequencies of the plurality of high-pass filters includes:
step B10, selecting the high-pass filter with the maximum cut-off frequency from the plurality of high-pass filters as a target filter;
step B20, using the target filter to filter and decompose the original multi-period time sequence data, and separating out corresponding period components;
in this embodiment, each periodic component may be sequentially filtered according to the order of the period from small to large, that is, the order of the frequency from large to small. Continuing with the above example of the home power data, for convenience of explanation, it is assumed that the original multicycle time-series data, i.e., the original power data series, includes two cycle components: if the period is a one-day component and a one-week component, and the power consumption is recorded every half hour, then the periods corresponding to the two components are 24 × 2 — 48 and 48 × 7 — 336, respectively, then based on this case, a high-pass filter with the maximum cutoff frequency among all the obtained high-pass filters may be determined as a target filter, that is, a high-pass filter with a cutoff frequency of 1/(2 × 48), and then the original power data sequence may be passed through the target filter, so that the component with the period of 48 may be decomposed, while the other components may remain in the time sequence because the frequency is lower than the cutoff frequency.
Step B30, selecting the high-pass filter with the maximum cut-off frequency from the rest high-pass filters, and updating the target filter;
and step B40, decomposing the residual sequence of the original multi-period time series data by using the updated target filter, and separating out the corresponding period component.
After the periodic component with the highest frequency is separated, the periodic component with the second smallest period and the second highest frequency can be decomposed. Specifically, the high-pass filter with the largest cutoff frequency, i.e., the highest overall cutoff frequency, is selected from the remaining high-pass filters as a new target filter, i.e., the high-pass filter with the cutoff frequency of 1/(2 × 336), and then the remaining sequence with the largest frequency cycle component filtered out is passed through the high-pass filter, so that the cycle component with the cycle of 336 can be decomposed, and the other components are retained in the time sequence because the frequency is lower than the cutoff frequency.
It can be understood that, if a sequence includes a plurality of different periodic components, the filtering step may be repeated, and the periodic component with the largest frequency in the remaining sequences is decomposed each time until the sequence passes through all the high-pass filters to obtain all the periodic components. Of course, in another feasible manner, the periodic components may be sequentially filtered in the order from large period to small period, i.e., from small frequency to large frequency, that is, the sequence may be sequentially passed through the filters in the order of the frequency of the periodic components, so as to perform the linear decomposition operation.
Further, after decomposing the above-mentioned multiple periodic components, referring to fig. 5, fig. 5 is a schematic flowchart after the step of S203 in fig. 2, where after decomposing the multiple periodic components from the original multiple periodic time series data by using multiple high-pass filters based on the cut-off frequencies of the multiple high-pass filters, the method includes:
step S501, screening out at least one to-be-tested periodic component with the period smaller than a preset periodic threshold value from a plurality of periodic components;
as described above, since the plurality of periodic components are obtained by filtering with the high-pass filter, a high-frequency portion of the original multi-period time series data is obtained, and it can be understood that interference factors such as noise are generally concentrated in the high-frequency portion, and therefore, for the periodic components obtained by filtering, particularly, for the small periodic components with high frequency, interference factors such as noise are easily mixed in the periodic components, and some abnormal points appear in the corresponding components, so that abnormality detection and repair are required for the small periodic components in order to further improve the resolution accuracy. Therefore, at least one to-be-detected periodic component with a period smaller than a preset periodic threshold needs to be screened out, and then subsequent processing is performed on the to-be-detected periodic components with the small periods, wherein the preset periodic threshold can be selected and adjusted according to actual needs and historical experience.
Step S502, aiming at any one periodic component to be detected, obtaining a corresponding frequency spectrum component;
step S503, according to the frequency spectrum components, utilizing a spectrum residual error algorithm to screen out abnormal values with the significance degree not within a preset significance degree range, and determining abnormal point positions corresponding to the abnormal values;
in this embodiment, the anomaly detection and the subsequent processing can be performed on the small-period periodic component to be detected in the frequency domain angle. Specifically, the periodic component to be measured is first converted into a corresponding frequency spectrum component, and the common conversion method may be fourier transform, laplace transform, Z transform, and the like; then, for the obtained spectrum component, a spectrum residual algorithm can be used to perform significance detection on the spectrum component, specifically, the amplitude of the frequency domain component is determined first, in order to facilitate calculation, the logarithm of the amplitude can be converted into a linear line close to linear arrangement, then the logarithm amplitude is differentiated from the logarithm average value, the significance is determined according to the difference, if the difference is too large, namely the significance is not within a preset significance range, the point is abnormal, and since the time domain signal and the frequency domain signal are corresponding, the time domain point corresponding to the point is also abnormal, so that the position of the abnormal point in the time domain can be determined.
Step S504, aiming at any abnormal value, a plurality of sequence values in a preset range are obtained by taking a corresponding abnormal point as a center;
step S505, replacing the abnormal value by using the average value of the plurality of sequence values to obtain a periodic component for repairing the abnormal value;
step S506, according to the abnormal periodic component of the repair, the corresponding periodic component is updated.
If the abnormal point corresponding to at least one abnormal value is found through the steps, in order to obtain a decomposition result with higher precision, the abnormal values need to be repaired. Specifically, in this embodiment, a plurality of normal values, for example, 10 normal values, around the abnormal point may be found first, and it can be understood that, in a normal condition, a time domain waveform of a signal should be relatively smooth and continuous, and no particularly sharp peak and valley occurs, therefore, an average value of the 10 normal values around the abnormal point may be used as a replacement value to perform interpolation replacement on the abnormal point, so that a difference between a sequence value corresponding to the abnormal point and a normal sequence value around the abnormal point is not large, so as to obtain a smooth and continuous time domain periodic component, and finally, the corresponding periodic component is updated by using the smooth and continuous time domain periodic component, so that the repair can be completed in a time domain angle. Therefore, the influence caused by abnormal points in the original sequence can be effectively eliminated, and the decomposition precision can be improved.
Step S204, decomposing the residual multi-period time sequence data after decomposing the plurality of period components by using the HP filter and the low-pass filter to obtain a trend component;
after the periodic component is decomposed, the high-frequency part in the original multi-period time sequence data is filtered, and the long-term trend term, i.e. the trend component is generally long in period and low in frequency, so that the trend component can be decomposed from the remaining low-frequency component.
In one embodiment, referring to fig. 6, fig. 6 is a detailed flowchart of the step S204 in fig. 2, where the decomposing the remaining multi-cycle time series data after decomposing the plurality of cycle components by using the HP filter and the low pass filter to obtain the trend component includes:
step C10, performing denoising processing on the residual multi-period time series data after the decomposition of the plurality of periodic components by using the low-pass filter;
as described above, the cutoff frequency of the low-pass filter is set based on the maximum period of the original sequence, that is, all periodic components in the original multicycle time-series data and high-frequency components such as noise included therein cannot pass through the low-pass filter. Therefore, even if there are some residual high frequency components such as noise in the decomposition process of the periodic component, the residual high frequency components will be completely filtered out in this step to obtain the trend component with low frequency.
And C20, resolving the trend component from the residual multi-cycle time sequence data subjected to the denoising processing by using the HP filter.
It can be understood that, since the frequency of the trend component is very low, it is very easy to be interfered by other components, and thus a situation of glitch and non-smoothness occurs, so in order to make the finally obtained trend component have better quality and be smoother, the sequence component passing through the low-pass filter is further subjected to secondary filtering processing by using the HP filter in this embodiment. Therefore, the low-frequency part is filtered by the low-pass filter, the high-frequency part mixed in the obtained low-frequency part is further filtered by the HP filter, and therefore the trend component which is smoother and better in quality can be obtained, and the accuracy of the whole decomposition is improved.
Step S205, obtaining a multi-cycle time series data decomposition result based on the original multi-cycle time series data, the plurality of cycle components, and the trend component.
Finally, after the periodic components and the trend components are decomposed, the residual components in the original multicycle time series data are residual components, and for a multicycle time series, the residual components are generally composed of the periodic components, the trend components and the residual components, so that the decomposition of the original multicycle time series data is completed, and a multicycle time series data decomposition result is obtained. The residual component is mainly composed of some unstable factors such as noise, and is generally an interference component in the original sequence, which can significantly affect the accuracy of subsequent data analysis and processing, so that in this embodiment, a periodic component and a trend component with higher accuracy and better quality can be obtained, which is also beneficial to reducing the influence of the residual component, and further improving the accuracy of subsequent processing.
In a specific embodiment, referring to fig. 7, fig. 7 is a schematic diagram illustrating a detailed flow of the step S205 in fig. 2, where the obtaining a decomposition result of the multicycle time series data based on the original multicycle time series data, the plurality of cyclic components and the trend component includes:
step D10, obtaining residual error components according to the original multi-cycle time sequence data, the cycle components after abnormal repair and the trend components;
and D20, obtaining the multi-cycle time series data decomposition result according to the cycle component after abnormal repair, the trend component and the residual component.
First, as described above, for a multi-cycle time series, it is generally composed of a cycle component, a trend component and a residual component, and therefore, after the cycle component and the trend component are obtained through the above various filtering decompositions, the remaining component is the residual component, and of course, in this embodiment, since the small cycle component and the trend component are further processed twice, i.e., the abnormal repairing and the secondary filtering, in order to ensure the accuracy of the decomposition result of the obtained multi-cycle time series data, the cycle component after repairing the abnormal and the trend component after the secondary filtering may be subtracted from the original multi-cycle time series data to obtain a more accurate residual component, thereby ensuring the accuracy of the whole decomposition process.
It should be noted that, after the whole decomposition process is completed, various subsequent data analyses and practical applications can be performed according to the decomposition result. For example, in a power data detection scene, the abnormal points in the power usage detection data can be located and repaired through the decomposition, and further, as the periodic components and the trend components with higher precision and better quality can be obtained through the steps, the power usage in a future period can be more accurately predicted by utilizing the rules of the periodic components and the trend components, and the method can also be used for other related data analysis processing processes, and can also be used for related processing of data with multi-periodicity qualities such as network traffic, traffic flow, passenger flow and the like, and the accuracy of analysis processing is favorably improved; meanwhile, as the iterative loop repeated for many times like the STL algorithm is not needed in the embodiment, the decomposition efficiency of the original sequence of the embodiment is higher, and the improvement of the efficiency of the subsequent data analysis and processing is facilitated.
In this embodiment, on one hand, it is not necessary to perform repeated iterative loop as in the STL algorithm, but a linear manner is adopted, and the original sequence is passed through each filter in sequence, so that the operation is simpler, the decomposition efficiency is higher, and the efficiency of subsequent data analysis and processing is further improved; on the other hand, in the embodiment, the small periodic component can be subjected to anomaly detection and restoration, and the trend component is subjected to polarity secondary filtering to be smoother, so that the obtained decomposition result is higher in precision, and the accuracy of subsequent data analysis and processing is further improved.
Based on the same inventive concept, an embodiment of the present invention further provides a multi-cycle time series data decomposition device, which is shown in fig. 7 and includes:
the system comprises a sequence acquisition module, a sequence acquisition module and a data processing module, wherein the sequence acquisition module is used for acquiring original multi-cycle time sequence data and a plurality of cycle parameters of the original multi-cycle time sequence data;
the filter generation module is used for obtaining a plurality of corresponding high-pass filters, a low-pass filter and an HP filter according to the plurality of period parameters;
the period decomposition module is used for decomposing the original multi-period time sequence data by using a plurality of high-pass filters to obtain a plurality of period components based on the cut-off frequencies of the high-pass filters;
a trend decomposition module for decomposing the residual multi-period time sequence data after decomposing the plurality of period components by using the HP filter and the low-pass filter to obtain a trend component;
and the result output module is used for obtaining a multi-cycle time series data decomposition result based on the original multi-cycle time series data, the plurality of cycle components and the trend component.
It should be noted that, in the present embodiment, for each embodiment of the multi-cycle time-series data decomposition device and the technical effects achieved by the embodiment, reference may be made to various embodiments of the multi-cycle time-series data decomposition method in the foregoing embodiments, and details are not repeated here.
Furthermore, in an embodiment, the present application further provides a computer storage medium having a computer program stored thereon, where the computer program is executed by a processor to implement the steps of the method in the foregoing method embodiments.
In some embodiments, the computer-readable storage medium may be memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash, magnetic surface memory, optical disk, or CD-ROM; or may be various devices including one or any combination of the above memories. The computer may be a variety of computing devices including intelligent terminals and servers.
In some embodiments, executable instructions may be written in any form of programming language (including compiled or interpreted languages), in the form of programs, software modules, scripts or code, and may be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.
By way of example, executable instructions may correspond, but do not necessarily have to correspond, to files in a file system, may be stored in a portion of a file that holds other programs or data, e.g., in one or more scripts in a hypertext Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files that store one or more modules, sub-programs, or portions of code).
As an example, executable instructions may be deployed to be executed on one computing device or on multiple computing devices located at one site or distributed across multiple sites and interconnected by a communication network.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention, and all equivalent structures or equivalent processes performed by the present invention or directly or indirectly applied to other related technical fields are also included in the scope of the present invention.
Claims (10)
1. A method of multi-cycle time series data decomposition, the method comprising:
acquiring original multi-cycle time sequence data and a plurality of cycle parameters of the original multi-cycle time sequence data;
obtaining a plurality of corresponding high-pass filters, a low-pass filter and an HP filter according to the plurality of period parameters;
decomposing a plurality of periodic components from the original multi-period time series data by using a plurality of high-pass filters based on cut-off frequencies of the high-pass filters;
decomposing the residual multi-cycle time series data after decomposing the plurality of cycle components by using the HP filter and the low-pass filter to obtain a trend component;
and obtaining a multi-cycle time series data decomposition result based on the original multi-cycle time series data, the plurality of cycle components and the trend component.
2. The method of claim 1, wherein deriving a corresponding plurality of high pass filters, a low pass filter, and an HP filter based on a plurality of said period parameters comprises:
determining a plurality of corresponding cut-off frequencies according to the plurality of period parameters;
obtaining a plurality of high-pass filters according to a plurality of cut-off frequencies;
screening out a target period parameter with the largest period from the plurality of period parameters;
and obtaining the low-pass filter and the HP filter according to the target period parameter.
3. The method of claim 1, wherein decomposing the plurality of periodic components from the raw multicycle time series data using the plurality of high pass filters based on cutoff frequencies of the plurality of high pass filters comprises:
screening a high-pass filter with the maximum cut-off frequency from the plurality of high-pass filters as a target filter;
performing filtering decomposition on the original multi-period time sequence data by using the target filter to separate out corresponding period components;
screening out a high-pass filter with the maximum cut-off frequency from the rest high-pass filters, and updating the target filter;
decomposing the residual sequence of the original multi-cycle time sequence data by using the updated target filter, and separating out a corresponding cycle component;
and returning to execute the high-pass filter with the maximum cut-off frequency screened from the rest high-pass filters, and updating the target filter until the original multi-cycle time sequence data passes through all the high-pass filters to obtain a plurality of corresponding cycle components.
4. The method of claim 1, wherein after decomposing a plurality of periodic components from the raw multicycle time series data using a plurality of the high pass filters based on cutoff frequencies of the plurality of high pass filters, the method comprises:
screening out at least one to-be-tested periodic component with the period smaller than a preset periodic threshold value from the plurality of periodic components;
aiming at any one periodic component to be detected, obtaining a corresponding frequency spectrum component;
and screening abnormal values with the significance degree not within a preset significance degree range by utilizing a spectrum residual error algorithm according to the frequency spectrum components, and determining abnormal point positions corresponding to the abnormal values.
5. The method according to claim 4, wherein the method further comprises, after the abnormal value with the degree of significance not within a preset degree of significance is screened out by using a spectrum residual algorithm according to the spectrum components and the abnormal point position corresponding to the abnormal value is determined, the method further comprises:
aiming at any abnormal value, a plurality of sequence values in a preset range are obtained by taking a corresponding abnormal point as a center;
replacing the abnormal value by using the average value of the sequence values to obtain a periodic component for repairing the abnormal value;
and updating the corresponding periodic component according to the periodic component for repairing the abnormality.
6. The method according to claim 5, wherein said decomposing with said HP filter and said low pass filter a trend component from the remaining multicycle time series data after said decomposing a plurality of said cycle components comprises:
denoising the residual multi-period time sequence data after decomposing the plurality of periodic components by using the low-pass filter;
and resolving the trend component from the residual multi-cycle time series data subjected to the denoising processing by using the HP filter.
7. The method according to claim 6, wherein obtaining a decomposition of the multicycle time series data based on the raw multicycle time series data, a plurality of the cycle components and the trend component comprises:
obtaining a residual error component according to the original multi-cycle time sequence data, the cycle component after abnormal repair and the trend component;
and obtaining the multi-cycle time series data decomposition result according to the cycle component, the trend component and the residual component after the abnormity is repaired.
8. A multi-cycle time series data decomposition apparatus, comprising:
the system comprises a sequence acquisition module, a sequence acquisition module and a data processing module, wherein the sequence acquisition module is used for acquiring original multi-cycle time sequence data and a plurality of cycle parameters of the original multi-cycle time sequence data;
the filter generation module is used for obtaining a plurality of corresponding high-pass filters, a low-pass filter and an HP filter according to the plurality of period parameters;
the period decomposition module is used for decomposing the original multi-period time sequence data by using a plurality of high-pass filters to obtain a plurality of period components based on the cut-off frequencies of the high-pass filters;
a trend decomposition module for decomposing the residual multi-period time sequence data after decomposing the plurality of period components by using the HP filter and the low-pass filter to obtain a trend component;
and the result output module is used for obtaining a multi-cycle time series data decomposition result based on the original multi-cycle time series data, the plurality of cycle components and the trend component.
9. A multi-cycle time-series data decomposition device comprising a memory, a processor, and a multi-cycle time-series data decomposition program stored on the memory and executable on the processor, the multi-cycle time-series data decomposition program when executed by the processor implementing the steps of the multi-cycle time-series data decomposition method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that a multi-cycle time-series data disassembly program is stored thereon, which when executed by a processor implements the steps of the multi-cycle time-series data disassembly method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210514494.XA CN114944831A (en) | 2022-05-12 | 2022-05-12 | Multi-cycle time series data decomposition method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210514494.XA CN114944831A (en) | 2022-05-12 | 2022-05-12 | Multi-cycle time series data decomposition method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114944831A true CN114944831A (en) | 2022-08-26 |
Family
ID=82907257
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210514494.XA Pending CN114944831A (en) | 2022-05-12 | 2022-05-12 | Multi-cycle time series data decomposition method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114944831A (en) |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020186895A1 (en) * | 1996-08-12 | 2002-12-12 | National Aeronautics And Space Administration | Three dimensional empirical mode decomposition analysis apparatus, method and article manufacture |
| US6772185B1 (en) * | 1999-06-02 | 2004-08-03 | Japan Science And Technology Corporation | Time-series predicting method using wavelet number series and device thereof |
| US20090106839A1 (en) * | 2007-10-23 | 2009-04-23 | Myeong-Seok Cha | Method for detecting network attack based on time series model using the trend filtering |
| CN102568205A (en) * | 2012-01-10 | 2012-07-11 | 吉林大学 | Traffic parameter short-time prediction method based on empirical mode decomposition and classification combination prediction in abnormal state |
| US9471544B1 (en) * | 2012-05-24 | 2016-10-18 | Google Inc. | Anomaly detection in a signal |
| CN108346099A (en) * | 2018-02-08 | 2018-07-31 | 上海宽全智能科技有限公司 | Hedging transaction analysis method, equipment and storage medium based on wavelet analysis |
| CN111431507A (en) * | 2020-04-10 | 2020-07-17 | 自然资源部第一海洋研究所 | Self-adaptive signal decomposition and filtering method for constructing envelope curve by half-cycle simple harmonic function |
| CN111735381A (en) * | 2020-07-21 | 2020-10-02 | 湖南联智科技股份有限公司 | Beidou monitoring result error elimination method |
| CN112000830A (en) * | 2020-08-26 | 2020-11-27 | 中国科学技术大学 | Time sequence data detection method and device |
| CN112989271A (en) * | 2019-12-02 | 2021-06-18 | 阿里巴巴集团控股有限公司 | Time series decomposition |
| US20210224677A1 (en) * | 2020-01-21 | 2021-07-22 | International Business Machines Corporation | Forecasting model generation for time series data with change point and seasonality |
| CN113849374A (en) * | 2021-09-28 | 2021-12-28 | 平安科技（深圳）有限公司 | CPU occupancy rate prediction method, system, electronic device and storage medium |
| CN114328662A (en) * | 2021-12-27 | 2022-04-12 | 中国电信股份有限公司 | Abnormal data positioning method and device, electronic equipment and storage medium |
| CN114358634A (en) * | 2022-01-11 | 2022-04-15 | 平安科技（深圳）有限公司 | Data analysis method, device and equipment based on data driving and storage medium |
-
2022
- 2022-05-12 CN CN202210514494.XA patent/CN114944831A/en active Pending
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020186895A1 (en) * | 1996-08-12 | 2002-12-12 | National Aeronautics And Space Administration | Three dimensional empirical mode decomposition analysis apparatus, method and article manufacture |
| US6772185B1 (en) * | 1999-06-02 | 2004-08-03 | Japan Science And Technology Corporation | Time-series predicting method using wavelet number series and device thereof |
| US20090106839A1 (en) * | 2007-10-23 | 2009-04-23 | Myeong-Seok Cha | Method for detecting network attack based on time series model using the trend filtering |
| CN102568205A (en) * | 2012-01-10 | 2012-07-11 | 吉林大学 | Traffic parameter short-time prediction method based on empirical mode decomposition and classification combination prediction in abnormal state |
| US9471544B1 (en) * | 2012-05-24 | 2016-10-18 | Google Inc. | Anomaly detection in a signal |
| CN108346099A (en) * | 2018-02-08 | 2018-07-31 | 上海宽全智能科技有限公司 | Hedging transaction analysis method, equipment and storage medium based on wavelet analysis |
| CN112989271A (en) * | 2019-12-02 | 2021-06-18 | 阿里巴巴集团控股有限公司 | Time series decomposition |
| US20210224677A1 (en) * | 2020-01-21 | 2021-07-22 | International Business Machines Corporation | Forecasting model generation for time series data with change point and seasonality |
| CN111431507A (en) * | 2020-04-10 | 2020-07-17 | 自然资源部第一海洋研究所 | Self-adaptive signal decomposition and filtering method for constructing envelope curve by half-cycle simple harmonic function |
| CN111735381A (en) * | 2020-07-21 | 2020-10-02 | 湖南联智科技股份有限公司 | Beidou monitoring result error elimination method |
| CN112000830A (en) * | 2020-08-26 | 2020-11-27 | 中国科学技术大学 | Time sequence data detection method and device |
| CN113849374A (en) * | 2021-09-28 | 2021-12-28 | 平安科技（深圳）有限公司 | CPU occupancy rate prediction method, system, electronic device and storage medium |
| CN114328662A (en) * | 2021-12-27 | 2022-04-12 | 中国电信股份有限公司 | Abnormal data positioning method and device, electronic equipment and storage medium |
| CN114358634A (en) * | 2022-01-11 | 2022-04-15 | 平安科技（深圳）有限公司 | Data analysis method, device and equipment based on data driving and storage medium |
Non-Patent Citations (4)
| Title |
|---|
| IN LIU, XUEFENG SANG, JIAXUAN CHANG & YANG ZHENG: "《Multi-Model Coupling Water Demand Prediction Optimization Method for Megacities Based on Time 》", 《WATER RESOURCES MANAGEMENT》 * |
| 孙晓涛: "《趋势周期分解理论与我国经济的周期》", 《知网》 * |
| 张冀青: "《预测理论与方法及其MATLAB实现》", 30 September 2020 * |
| 张沈立主编, 辽宁大学出版社 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11888316B2 (en) | Method and system of predicting electric system load based on wavelet noise reduction and EMD-ARIMA | |
| Carbone et al. | Detrending moving average algorithm: Frequency response and scaling performances | |
| Lv et al. | High-order synchroextracting transform for characterizing signals with strong AM-FM features and its application in mechanical fault diagnosis | |
| CN112326213B (en) | Abnormal data detection method and device and mechanical fault detection method and device | |
| CN111950805A (en) | Medium-and-long-term power load prediction method and device, computer equipment and storage medium | |
| CN111611152A (en) | Test case generation method and device, electronic equipment and readable storage medium | |
| CN114944831A (en) | Multi-cycle time series data decomposition method, device, equipment and storage medium | |
| CN114492528A (en) | Pulse signal denoising method and device, electronic equipment and storage medium | |
| CN112446389A (en) | Fault judgment method and device | |
| CN112283882B (en) | Method and device for acquiring sampling value and air conditioner | |
| CN115902396B (en) | Resonance detection method and device for large wind power grid-connected system | |
| CN115859054A (en) | Hydroelectric generating set tail water pipe pressure pulsation data filtering method based on MIC and CEEMDAN | |
| CN107527057B (en) | Wind speed and power abnormal data eliminating method and device | |
| CN113449933B (en) | Regional medium-term load prediction method and device based on clustering electric quantity curve decomposition | |
| CN113095170B (en) | Fault diagnosis method based on adjustable Q wavelet motor | |
| CN112613674B (en) | Medium-and-long-term wind power generation capacity prediction method and device, electronic equipment and storage medium | |
| CN112241813B (en) | Rainfall prediction method and system based on wavelet decomposition | |
| CN114281658A (en) | Cluster resource utilization rate prediction method, device, equipment and storage medium | |
| CN110659442A (en) | System, short-term data prediction method and device thereof, and storage medium | |
| CN114152454A (en) | Mechanical equipment fault diagnosis method based on CEEMDAN-CSE model and establishment method of model | |
| Nicolae et al. | Hybrid Algorithms for Denoising Electrical Waveforms Containing Steady Segments | |
| CN117613905B (en) | Power demand mid-term prediction method and system based on multidimensional component decomposition | |
| CN105335539A (en) | Data processing method and apparatus for power load | |
| Mönke et al. | Optimal Analysis for Rhythmic Time-Dependent Biological Data-pyBOAT | |
| CN113067891B (en) | Intelligent construction site data acquisition method and device, computer equipment and storage medium thereof |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220826 |
|
| RJ01 | Rejection of invention patent application after publication |