CN112766590B - Method and system for extracting typical residential power consumption pattern - Google Patents
Method and system for extracting typical residential power consumption pattern Download PDFInfo
- Publication number
- CN112766590B CN112766590B CN202110108000.3A CN202110108000A CN112766590B CN 112766590 B CN112766590 B CN 112766590B CN 202110108000 A CN202110108000 A CN 202110108000A CN 112766590 B CN112766590 B CN 112766590B
- Authority
- CN
- China
- Prior art keywords
- power consumption
- time
- day
- resident
- working day
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 32
- 230000002159 abnormal effect Effects 0.000 claims abstract description 16
- 230000002776 aggregation Effects 0.000 claims abstract description 13
- 238000004220 aggregation Methods 0.000 claims abstract description 13
- 238000012545 processing Methods 0.000 claims abstract description 10
- 238000006243 chemical reaction Methods 0.000 claims description 13
- 230000005611 electricity Effects 0.000 claims description 12
- 230000011218 segmentation Effects 0.000 claims description 11
- 238000000605 extraction Methods 0.000 claims description 10
- 238000005070 sampling Methods 0.000 claims description 9
- 238000010606 normalization Methods 0.000 claims description 4
- 238000013144 data compression Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 230000002354 daily effect Effects 0.000 description 23
- 230000006399 behavior Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000012935 Averaging Methods 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 229910052799 carbon Inorganic materials 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000006116 polymerization reaction Methods 0.000 description 2
- 238000011425 standardization method Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/10—Office automation; Time management
- G06Q10/109—Time management, e.g. calendars, reminders, meetings or time accounting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Abstract
The invention discloses a method and a system for extracting typical power consumption modes of residents, and belongs to the field of application of intelligent electric meters. The method comprises the following steps: acquiring M-day power consumption data of an object to obtain M daily load curves; carrying out standardization processing on each daily load curve data; for each standardized daily load curve, converting the daily load curve into a symbol time sequence by adopting symbolic aggregation approximation; classifying the M symbol time sequences by adopting a spatial clustering algorithm with noise based on density, and eliminating abnormal values; restoring each symbol time series to a power consumption level time series; and calculating the average value of the power consumption levels in the same period in all the power consumption level time series to obtain the typical power consumption pattern of the residents. According to the invention, the interference power characteristic is eliminated, so that the intra-cluster similarity and inter-cluster difference in the user classification result are higher; the abnormal value in the resident electric power consumption pattern is eliminated, so that the extracted resident typical electric power consumption pattern is more representative.
Description
Technical Field
The invention belongs to the application field of intelligent electric meters, and particularly relates to a method and a system for extracting typical residential power consumption patterns.
Background
With the continuous update of renewable energy and nuclear energy utilization technologies, most of the electric energy needs can be met by clean energy conversion. Then, in order to cope with the load demand during peak periods of electricity consumption, fossil energy still needs to be burned to obtain more electric energy, thus causing excessive carbon emission and serious environmental pollution. Residential power consumption is one of the main bodies of power consumption. More reasonable resident power consumption behaviors are modeled, and the effects of peak clipping and valley filling can be achieved, so that carbon emission is reduced. With the development and application of the smart grid technology, the established advanced measurement system and the user energy management system can realize the detailed measurement and storage of the user power consumption data, and provide a solid data base for the scientific analysis of the user power utilization behaviors.
Considering that the electricity consumption behaviors of residential users are variable, the electricity consumption patterns of the users extracted based on the data of the intelligent electric meters are an efficient mode for describing the electricity consumption behaviors. The daily load curve is formed by the data of the intelligent electric meters of the single user every day, and the extraction of the power consumption mode of the user is to perform data mining on the daily load curve of the user. The resident typical power consumption pattern extraction method needs to consider two important aspects. On the one hand, massive intelligent electric meter data can be efficiently processed, so that the real-time requirement of future intelligent electric meter data analysis is met. On the other hand, it is required to highlight the power consumption characteristics of the user as much as possible.
The typical power consumption patterns of the main users are extracted into two types at present. The first is an average value method, which directly calculates an average value of a user daily load curve as a typical power consumption pattern thereof. The method has the advantages of simplicity, quickness and small calculation amount, but has poor effect on users with abnormal values in daily load curves. The second method is that the abnormal value is filtered by adopting a spatial clustering algorithm with noise and based on density, and then the average value method is adopted for calculation. Compared with an average value method, the method solves the problem of influence of abnormal values, but the problem that the difference between partial categories is not obvious when users are classified based on the results of the method is solved. The reason for this is that some features with weak relevance to the power consumption mode still exist in the extracted load curve, so that the clustering algorithm is too much concerned about the weak relevance features when performing user classification, and further the clustering result is not significant enough.
Disclosure of Invention
Aiming at the defects of disturbance and improvement requirements in typical power consumption modes of residents in the prior art, the invention provides a method and a system for extracting typical power consumption modes of residents, aiming at carrying out segmented aggregation and symbolization processing on an original load curve by adopting symbol aggregation approximation to obtain a symbol time sequence, and then carrying out mode extraction on the symbol time sequence by adopting an improved noisy density-based spatial clustering algorithm to obtain a typical power consumption mode of residents with small disturbance, eliminating abnormal values and the characteristic of weak relevance with the typical power consumption mode, and having good mode characteristics. When the extracted typical power consumption patterns of residents are used for clustering analysis, the pattern characteristics among the same types are obvious, and the difference among different types is obvious. Therefore, the extraction method can be used to obtain the typical electricity consumption pattern of the residents with good pattern characteristics.
To achieve the above object, according to a first aspect of the present invention, there is provided a method for extracting a typical electricity consumption pattern of a resident, the method comprising the steps of:
s1, taking a single resident as an object, acquiring power consumption data of an object on N working days and non-working days in the working days, wherein the sampling interval of each day is N hours, and obtaining M working day-day load curves and M non-working day-day load curves of the object, wherein M is more than or equal to 30, N is less than or equal to 2, and the number H of sampling points is an integer of 24/N;
s2, carrying out standardization processing on data in each daily load curve to obtain a standardized daily load curve;
s3, converting each standardized daily load curve into a symbol time sequence by adopting symbolic aggregation approximation, wherein each symbol in the sequence reflects the power consumption level of a corresponding time period;
s4, dividing M symbol time sequences of the object into two types by adopting a spatial clustering algorithm with noise and based on density, wherein one type is a resident working day power consumption mode, and the other type is an abnormal value; dividing m symbol time sequences of the object into two types by adopting a spatial clustering algorithm with noise and based on density, wherein one type is a resident non-working day power consumption mode, the other type is an abnormal value, and removing all the abnormal values;
s5, restoring each symbol time sequence in the resident working day power consumption mode to a corresponding power consumption level time sequence, and restoring each symbol time sequence in the resident non-working day power consumption mode to a corresponding power consumption level time sequence;
s6, calculating an average value of the power consumption levels in the same time period in all the power consumption level time sequences in the resident working day power consumption mode to obtain a typical power consumption mode of the resident working day; and calculating an average value of the power consumption levels of the same period in all the power consumption level time series in the resident non-workday power consumption pattern to obtain the typical power consumption pattern of the resident on the non-workday.
Preferably, in step S2, the maximum and minimum value normalization method is used for processing.
Has the advantages that: the method adopts a maximum and minimum value standardization method for processing, can clean the baseline value in the power consumption data, and better keeps the power characteristics of users.
Preferably, step S3 includes the following sub-steps:
s31, time intervals are divided for 24 hours a day based on the standardized daily load curve, so that the ratio of the sum of squares of the residual errors of each time interval to the total time sequence freedom is minimized, and T time intervals are obtained;
s32, performing data compression on the daily load curve in T time periods by adopting a piecewise aggregation approximation to obtain a power consumption level time sequence;
and S33, converting each power consumption level time sequence into a symbol time sequence by adopting symbolic conversion.
Has the advantages that: the symbolic aggregation approximation method is improved, and the time intervals are divided for 24 hours a day, so that the ratio of the residual sum of squares of each time interval to the total time sequence freedom is minimized, and the obtained symbolic time sequence can better represent the power consumption level of a user.
Preferably, step S31 is specifically as follows:
s311, collecting the segmentation points B l Initializing to {1, H }, wherein H represents the number of sampling points in one day, and the time period number l is 1;
s312.l ═ l +1, minimize objective function f (B) l ) To obtain a new segmentation point b l The objective function is as follows:
f(B l )＝SS/df(P * )
wherein SS represents P * Sum of squared residuals of (d), df (P) * ) H-l represents P * Degree of freedom of (P) * Shows the normalized daily load curve, SS i Represents a period of time T i Inner residual sum of squares, p * (t) represents P * The value at the time point of the middle t,
represents a period of time T i Inner p * (t), the number of data points contained in each interval is more than 3;
s313. if f (B) l )≥f(B l-1 ) Then stop, resulting in B l ＝{b 0 ,b 1 ,…,b l The time interval is divided, and l at this time is assigned to T, otherwise, the process goes to step S312.
Has the advantages that: according to the invention, by setting an optimization target, a new segmentation point is obtained by solving the optimal solution of the optimization problem each time. Through continuous iteration, a segmentation point set is obtained to divide the time period, so that automatic division of the time period is realized, the obtained time period has good adaptation to curve characteristics, and the method can be applied to other scenes needing time period division.
Preferably, the dividing the M symbol time series of the object into two categories by using the noisy density-based spatial clustering algorithm includes the following sub-steps:
(1) the clustering radius epsilon and the step length a are initialized to the same value, and the threshold value minPts of the number of points in the epsilon neighborhood is set to be
Search accuracy epsilon end InitialTransforming, wherein each symbol time sequence is regarded as a point of the T-dimensional space;
(2) clustering the M points by adopting a density-based clustering algorithm with noise;
(3) counting the number N of core points p The core point is a point with the number of points in the neighborhood not less than minPts;
(4) if N is present p >0, entering the step (5), otherwise, a is a/2, and epsilon is epsilon + a, and entering the step (2);
(5) if a<ε end If yes, recording the current value of epsilon and the index of core point, otherwise, a ═ a/2 and epsilon-a, and entering step (2).
Has the beneficial effects that: the invention designs a quick search strategy for selecting parameters, which is used for carrying out the optimal selection of the parameters for each user. Because the daily load curves of different users fluctuate to different degrees, each user is required to independently select parameters. The quick parameter search strategy is selected, the dynamic step length is adopted, and compared with the static step length in the prior art, the running speed of the algorithm is improved on the premise of ensuring the precision, and a foundation is provided for the real-time application of the algorithm.
To achieve the above object, according to a second aspect of the present invention, there is provided an extraction system of a typical electricity consumption pattern of residents, comprising: a computer-readable storage medium and a processor;
the computer-readable storage medium is used for storing executable instructions;
the processor is used for reading the executable instructions stored in the computer readable storage medium and executing the resident typical electricity consumption pattern extraction method of the first aspect.
Generally, by the above technical solution conceived by the present invention, the following beneficial effects can be obtained:
the method adopts the symbol aggregation approximation to carry out the segmented aggregation and symbolization processing on the original load curve to obtain the symbol time sequence, eliminates the power characteristics of interference, and ensures that the result clusters of user classification have higher similarity and more obvious inter-cluster difference; and then, pattern extraction is carried out on the symbol time sequence by adopting a noisy density-based spatial clustering algorithm to obtain a typical residential power consumption pattern with small disturbance, and abnormal values in the typical residential power consumption pattern are eliminated, so that the extracted typical residential power consumption pattern is more representative.
Drawings
FIG. 1 is a flow chart of a typical residential power consumption pattern extraction method provided by the present invention;
FIG. 2 is a diagram illustrating exemplary symbolic aggregate approximation conversion results provided by the present invention;
FIG. 3 is a flow chart of the improved noisy density-based spatial clustering algorithm provided by the present invention;
fig. 4 is a schematic diagram of a clustering result obtained based on typical residential power consumption patterns, where (a) - (f) correspond to categories one to six, respectively.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
As shown in fig. 1, the present invention provides a method for extracting typical electricity consumption patterns of residents, the method comprising the steps of:
step S1, taking a single resident as an object, acquiring power consumption data of an M-day working day and a non-working day in the period of the working day of the object, wherein the sampling interval of each day is N hours, and obtaining M working day-day load curves and M non-working day-day load curves of the object, wherein M is more than or equal to 30, N is less than or equal to 2, and the number of sampling points H is an integer of 24/N.
Firstly, the data of the intelligent electric meters of residents needs to be collected, and the intelligent electric meters are classified according to users and days to form a daily load curve of the residents. The original time series of a single user's day is denoted as P ═ { P (1), P (2), …, P (h) }, i.e. the daily load curve. In this example 60 (one quarter), N0.5, and H48.
And S2, carrying out standardization processing on data in each daily load curve to obtain a standardized daily load curve.
Preferably, in step S2, the maximum and minimum normalization method is used to ensure that the daily load curve of the user is in the range of 0-1.
For the daily load curve of residents, the daily load curve is processed by adopting a maximum and minimum value standardization method, and the formula is as follows:
wherein p (t) and p * (t) respectively representing the actual active power at the moment t and the normalized active power; max of i p (i) and min i And p (i) respectively represents the maximum active power and the minimum active power of the user in one day.
And S3, converting each standardized daily load curve into a symbol time sequence by adopting symbolic aggregation approximation, wherein each symbol in the sequence reflects the power consumption level of a corresponding time period.
In the segmented polymerization approximate conversion, the invention designs a time interval division method of numerical iteration, improves the generalization of the algorithm and reduces the manual workload. According to the time interval division method, the numerical solving algorithm is established, the load similarity in each time interval is guaranteed, the load difference among different time intervals is obvious, the optimal segmentation point of the time interval division is automatically calculated, the time of one day is divided into a plurality of time intervals, and the effect of the segmentation aggregation approximate conversion is improved.
In symbolic aggregate approximation conversion, the load curve after segmented aggregate approximation conversion is converted into a symbol time sequence. In the invention, the symbol and the specific numerical value have corresponding relation, and the symbol and the specific numerical value can be mutually converted, thereby being beneficial to algorithm calculation.
Preferably, step S3 includes the following sub-steps:
and S31, carrying out time interval division on 24 hours a day based on the standardized daily load curve, so that the ratio of the sum of the squares of the residuals of each time interval to the total time sequence freedom is minimized, and obtaining T time intervals.
Preferably, as shown in fig. 3, step S31 is specifically as follows:
s311, collecting the segmentation points B l Initializing to {1, H }, wherein H represents the number of sampling points in one day, and the time period number l is 1;
s312.l ═ l +1, minimize objective function f (B) l ) To obtain a new segmentation point b l The objective function is as follows:
f(B l )＝SS/df(P * )
wherein SS represents P * Sum of squared residuals of (d), df (P) * ) H-l represents P * Degree of freedom of (P) * Shows the normalized daily load curve, SS i Represents a period of time T i Inner residual sum of squares, p * (t) represents P * The value at the time point of the middle t,
represents a period of time T i Inner p * (t) the number of data points contained in each interval is greater than 3;
s313. if f (B) l )≥f(B l-1 ) Then stop, resulting in B l ＝{b 0 ，b 1 ，…，b l The time interval is divided, and l at this time is assigned to T, otherwise, the process goes to step S312.
And S32, respectively carrying out data compression on the daily load curve in T time periods by adopting a piecewise aggregation approximation to obtain a power consumption level time sequence.
In the piecewise aggregate approximation conversion, the load curve P is obtained for the normalization method * ＝{p * (1)，p * (2)，…，p * (H) Is converted to obtain
Wherein the content of the first and second substances,
the calculation formula of (c) is as follows:
wherein, T i Denotes the ith time interval, m i Indicating the number of data values in the ith time interval.
And S33, converting each power consumption level time sequence into a symbol time sequence by adopting symbolization conversion.
Obtained by a segmented polymerization approximation conversion
Later, in the symbolic aggregate approximation conversion, a series of segmentation points { z ] need to be set 0 ，z 1 ,…，z Q-1 ，z Q Divide the vertical axis into Q segments. Each segment corresponds to a character, and thus a string of character sequences a will be set 1 ，…，α Q-1 ，α Q }. Let z 0 0 and z Q 1. Then, if
Then will be
Mapping to alpha j . Then, the compressed time series
Converted into a time sequence of symbols. A typical symbolic aggregate approximation conversion result is shown in fig. 2, where the corresponding symbol time sequence is 'babbcefe'.
S4, dividing M symbol time sequences of the object into two types by adopting a spatial clustering algorithm with noise and based on density, wherein one type is a resident working day power consumption mode, and the other type is an abnormal value; and dividing m symbol time sequences of the object into two types by adopting a spatial clustering algorithm with noise and based on density, wherein one type is a resident non-working day power consumption mode, the other type is an abnormal value, and removing all the abnormal values.
Preferably, the dividing the M symbol time series of the object into two categories by using the noisy density-based spatial clustering algorithm includes the following sub-steps:
(1) the clustering radius delta and the step length a are initialized to the same value, and the threshold value minPts of the number of points in delta neighborhood is set to be
Search accuracy epsilon end Initializing, wherein each symbol time sequence is regarded as one point of a T-dimensional space;
(2) clustering the M points by adopting a density-based clustering algorithm with noise;
(3) counting the number N of core points p The core point is a point with the number of points in the neighborhood not less than minPts;
(4) if N is present p >0, entering the step (5), otherwise, a is a/2, and epsilon is epsilon + a, and entering the step (2);
(5) if a<ε end If not, a is equal to a/2, and if not, is equal to epsilon-a, and the step (2) is entered.
The processing mode of the resident non-workday power consumption mode is similar to that of the resident workday power consumption mode, and is not described herein again.
In this embodiment, minPts is 20 days, and because the daily load curves of different users fluctuate to different degrees, it is necessary to select epsilon for each user. Epsilon end Determining the precision with a search range of [ epsilon ] end ,2ε ini -ε end ]. And due to epsilon end Close to 0, and therefore, the initial setting of ε ini Should be greater than half the maximum radius of the user, epsilon end Should be less than the user minimum radius. In this embodiment, the cluster radius is initialized to 4, and the search precision ∈ is set end The initialization was 0.01.
And S5, restoring each symbol time sequence in the resident working day power consumption mode into a corresponding power consumption level time sequence, and restoring each symbol time sequence in the resident non-working day power consumption mode into a corresponding power consumption level time sequence.
For example, the symbol time sequence is 'babcefe', and the restored power consumption level time sequences are '1/4, 1/12,1/4,5/12,5/12,3/4,11/12, 3/4'.
S6, calculating an average value of the power consumption levels in the same time period in all the power consumption level time sequences in the resident working day power consumption mode to obtain a typical power consumption mode of the resident working day; and calculating the average value of the power consumption levels of the same time period in all the power consumption level time series in the non-working day power consumption pattern of the residents to obtain the non-working day typical power consumption pattern of the residents.
The load curves screened by the improved noise-carrying density-based spatial clustering algorithm correspond to core points in the algorithm. And finally, averaging the core points of the specific period of the single user by adopting an averaging method, so as to obtain the typical power consumption mode of the residents.
Further, after the typical power consumption patterns of the residents are obtained, user clustering is performed based on the typical power consumption patterns of the residents, and the obtained user clustering results are shown in (a) - (f) of fig. 4. (a) - (f) corresponding to categories one to six, respectively, of the results produced by the user clusters. As shown in (a), light gray thin lines represent typical power consumption patterns of respective users in the same category, black thick lines represent category means of the category, and typical power consumption patterns of users in the same category are similar. By comparing the category means in (a) - (f), it can be seen that typical power consumption patterns between different categories have obvious differences, i.e. the clustering effect of users obtained based on typical power consumption patterns of residents is good.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (5)
1. A method for extracting typical electricity consumption patterns of residents, comprising the steps of:
s1, taking a single resident as an object, acquiring power consumption data of an M-day working day and a non-working day in the period of the working day of the object, wherein the sampling interval of each day is N hours, and obtaining M working day-day load curves and M non-working day-day load curves of the object, wherein M is more than or equal to 30, N is less than or equal to 2, and the number H of sampling points is an integer of 24/N;
s2, carrying out standardization processing on data in each daily load curve to obtain a standardized daily load curve;
s3, converting each standardized daily load curve into a symbol time sequence by adopting symbolic aggregation approximation, wherein each symbol in the sequence reflects the power consumption level of a corresponding time period;
s4, dividing M symbol time sequences of the object into two types by adopting a spatial clustering algorithm with noise and based on density, wherein one type is a resident working day power consumption mode, and the other type is an abnormal value; dividing m symbol time sequences of the object into two types by adopting a spatial clustering algorithm with noise and based on density, wherein one type is a resident non-working day power consumption mode, the other type is an abnormal value, and removing all the abnormal values; the method for dividing the M symbol time sequences of the object into two types by adopting a noisy density-based spatial clustering algorithm comprises the following substeps:
(1) the clustering radius epsilon and the step length a are initialized to the same value, and the threshold value minPts of the number of points in the epsilon neighborhood is set to be
Search accuracy epsilon end Initializing, wherein each symbol time sequence is regarded as one point of a T-dimensional space;
(2) clustering the M points by adopting a density-based clustering algorithm with noise;
(3) counting the number N of core points p The core point is a point with the number of points in the neighborhood not less than minPts;
(4) if N is present p >0, entering the step (5), otherwise, a is a/2, and epsilon is epsilon + a, and entering the step (2);
(5) if a<ε end If yes, recording the current value of epsilon and the index of the core point, otherwise, recording a to a/2 and epsilon to epsilon-a, and entering the step (2);
s5, restoring each symbol time sequence in the resident working day power consumption mode into a corresponding power consumption level time sequence, and restoring each symbol time sequence in the resident non-working day power consumption mode into a corresponding power consumption level time sequence;
s6, calculating an average value of the power consumption levels in the same time period in all the power consumption level time sequences in the resident working day power consumption mode to obtain a typical power consumption mode of the resident working day; and calculating the average value of the power consumption levels of the same time period in all the power consumption level time series in the non-working day power consumption pattern of the residents to obtain the non-working day typical power consumption pattern of the residents.
2. The method of claim 1, wherein in step S2, the maximum and minimum normalization method is used for processing.
3. The method of claim 1, wherein step S3 includes the sub-steps of:
s31, time intervals are divided for 24 hours a day based on the standardized daily load curve, so that the ratio of the sum of squares of the residual errors of each time interval to the total time sequence freedom is minimized, and T time intervals are obtained;
s32, performing data compression on the daily load curve in T time periods by adopting a piecewise aggregation approximation to obtain a power consumption level time sequence;
and S33, converting each power consumption level time sequence into a symbol time sequence by adopting symbolic conversion.
4. The method according to claim 3, wherein step S31 is specifically as follows:
s311, collecting the segmentation points B l Initializing to {1, H }, wherein H represents the number of sampling points in one day, and the time period number l is 1;
s312.l ═ l +1, minimize objective function f (B) l ) To obtain a new segmentation point b l The objective function is as follows:
f(B l )＝SS/df(P * )
wherein SS represents P * Sum of squared residuals of (d), df (P) * ) H-l represents P * Degree of freedom of (P) * Shows the normalized daily load curve, SS i Represents a period of time T i Inner residual sum of squares, p * (t) represents P * The value at the time point of the middle t,
represents a period of time T i Inner p * (t) the number of data points contained in each interval is greater than 3;
s313. if f (B) l )≥f(B l-1 ) Then stop, resulting in B l ＝{b 0 ,b 1 ,…,b l The time interval is divided, and l at this time is assigned to T, otherwise, the process goes to step S312.
5. An extraction system of typical electricity consumption patterns of residents, comprising: a computer-readable storage medium and a processor;
the computer-readable storage medium is used for storing executable instructions;
the processor is used for reading executable instructions stored in the computer readable storage medium and executing the resident typical electricity consumption pattern extraction method according to any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110108000.3A CN112766590B (en) | 2021-01-27 | 2021-01-27 | Method and system for extracting typical residential power consumption pattern |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110108000.3A CN112766590B (en) | 2021-01-27 | 2021-01-27 | Method and system for extracting typical residential power consumption pattern |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112766590A CN112766590A (en) | 2021-05-07 |
| CN112766590B true CN112766590B (en) | 2022-09-20 |
Family
ID=75707391
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110108000.3A Expired - Fee Related CN112766590B (en) | 2021-01-27 | 2021-01-27 | Method and system for extracting typical residential power consumption pattern |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112766590B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113723481A (en) * | 2021-08-19 | 2021-11-30 | 吉林大学 | Multivariate load classification method and system based on symbolic aggregation approximate representation |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108009938B (en) * | 2016-11-02 | 2021-12-03 | 中国电力科学研究院 | System load clustering and load period pattern recognition method based on shape |
| CN111177216B (en) * | 2019-12-23 | 2024-01-05 | 国网天津市电力公司电力科学研究院 | Association rule generation method and device for comprehensive energy consumer behavior characteristics |
| CN111967919A (en) * | 2020-09-03 | 2020-11-20 | 上海积成能源科技有限公司 | System and method for analyzing electricity consumption behavior of residents based on autoregressive and adaptive boosting algorithm |
-
2021
- 2021-01-27 CN CN202110108000.3A patent/CN112766590B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN112766590A (en) | 2021-05-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107800140B (en) | Large user power supply access decision method considering load characteristics | |
| CN111414717A (en) | XGboost-L ightGBM-based unit power prediction method | |
| CN110503256A (en) | Short-term load forecasting method and system based on big data technology | |
| CN110766200A (en) | Method for predicting generating power of wind turbine generator based on K-means mean clustering | |
| CN112186761B (en) | Wind power scene generation method and system based on probability distribution | |
| CN106570790B (en) | Wind power plant output data restoration method considering wind speed data segmentation characteristics | |
| CN103955521A (en) | Cluster classification method for wind power plant | |
| CN112766590B (en) | Method and system for extracting typical residential power consumption pattern | |
| CN116796403A (en) | Building energy saving method based on comprehensive energy consumption prediction of commercial building | |
| CN116148753A (en) | Intelligent electric energy meter operation error monitoring system | |
| CN112001441A (en) | Power distribution network line loss anomaly detection method based on Kmeans-AHC hybrid clustering algorithm | |
| CN113505465B (en) | Fully unsupervised non-invasive electrical appliance state model self-adaptive construction method | |
| CN110909786A (en) | New user load identification method based on characteristic index and decision tree model | |
| CN117113126A (en) | Industry electricity utilization characteristic analysis method based on improved clustering algorithm | |
| CN116089777A (en) | Intelligent new energy settlement method and system based on intelligent information matching | |
| CN114036845A (en) | Wind power plant group model construction method | |
| CN115329899A (en) | Clustering equivalent model construction method, system, equipment and storage medium | |
| CN114722098A (en) | Typical load curve identification method based on normal cloud model and density clustering algorithm | |
| Liu et al. | Hierarchical Identification Method of Household-Transformer Relationship based on Adaptive Piecewise Aggregation Approximation | |
| Wang et al. | Analysis of user’s power consumption behavior based on k-means | |
| CN117096877B (en) | Multi-view-based regional dynamic electricity-carbon data electricity analysis method and system | |
| Shen et al. | A Novel AI-based Method for EV Charging Load Profile Clustering | |
| CN117291299B (en) | Moon electricity quantity prediction method considering various influence factors | |
| CN111695599B (en) | Elastic identification method for user electricity load time | |
| CN110932265B (en) | Static voltage stability evaluation method and system |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20220920 |