CN109783528A - A kind of electricity consumption schema extraction method and system - Google Patents

A kind of electricity consumption schema extraction method and system Download PDF

Info

Publication number
CN109783528A
CN109783528A CN201811402866.XA CN201811402866A CN109783528A CN 109783528 A CN109783528 A CN 109783528A CN 201811402866 A CN201811402866 A CN 201811402866A CN 109783528 A CN109783528 A CN 109783528A
Authority
CN
China
Prior art keywords
tuple
sequence
electricity consumption
month
inflection point
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.)
Granted
Application number
CN201811402866.XA
Other languages
Chinese (zh)
Other versions
CN109783528B (en
Inventor
徐超
邓君华
邹云峰
赵磊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Grid Corp of China SGCC
State Grid Jiangsu Electric Power Co Ltd
Electric Power Research Institute of State Grid Jiangsu Electric Power Co Ltd
Original Assignee
State Grid Corp of China SGCC
State Grid Jiangsu Electric Power Co Ltd
Electric Power Research Institute of State Grid Jiangsu Electric Power Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by State Grid Corp of China SGCC, State Grid Jiangsu Electric Power Co Ltd, Electric Power Research Institute of State Grid Jiangsu Electric Power Co Ltd filed Critical State Grid Corp of China SGCC
Priority to CN201811402866.XA priority Critical patent/CN109783528B/en
Publication of CN109783528A publication Critical patent/CN109783528A/en
Application granted granted Critical
Publication of CN109783528B publication Critical patent/CN109783528B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • YGENERAL 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
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02DCLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00Energy efficient computing, e.g. low power processors, power management or thermal management

Landscapes

  • Supply And Distribution Of Alternating Current (AREA)

Abstract

The invention discloses a kind of electricity consumption schema extraction method and systems, and wherein method includes establishing the feature tuple sequence of electricity consumption enterprise and calculating the distance between two feature tuples based on the maximum common subsequence of feature tuple sequence;Power mode is used for what business electrical amount sequence carried out that the cluster based on distance obtains target industry.Present invention employs the design philosophys of maximum common subsequence, the method of the present invention is set to be more suitable the more similar situation of business electrical amount of the same trade, to provide more correct input for the extraction of user power utilization mode, lay the foundation to construct significantly more efficient schema extraction model.

Description

A kind of electricity consumption schema extraction method and system
Technical field
The present invention relates to a kind of electricity consumption schema extraction methods, and in particular to be to use power mode in a kind of power industry Extracting method.
Background technique
In power industry, the variation of business electrical mode often reflects enterprise management condition variation.Business circumstance compared with Good enterprise, steadily increasing situation is presented with power mode;The poor enterprise of business circumstance, it is rapid that often there is electricity consumptions Subtract or even the case where recurrent fluctuations.In addition, being influenced by factors such as state's laws regulations, the electricity consumption of an enterprise inside the circle Often there is certain common variation tendencies for variation.
Therefore, the differentiation of trade power consumption mode reflects management state in industry.And the manual analysis that Utilities Electric Co. is traditional Mode, there is inefficiency, analyze the limited defect of data volume.Therefore a kind of conscientiously efficient electricity consumption mould is badly in need of in Utilities Electric Co. The extracting method of formula.
Summary of the invention
The technical problem to be solved by the present invention is to overcome the deficiencies of existing technologies, provide it is a kind of it is more efficient, be more suitable for The method for extracting target industry business electrical mode in the case where business electrical amount similarity-rough set is high.
In order to solve the above technical problems, the invention adopts the following technical scheme:
In one aspect, the present invention provides a kind of electricity consumption schema extraction methods, comprising the following steps:
Step 1: establishing target data set: according to the electricity consumption enterprise of target industry in particular year use corresponding with month Electricity data establishes business electrical amount sequence rec, rec={ < month1,power1>,<month2,power2>,……
<monthn,powern>, 1≤n≤12, monthnN-th month is represented, powernRepresent n-th month electricity;
Step 2: the slope for calculating two neighboring moon electricity consumption to the electricity consumption sequence of each enterprise obtains slope characteristics sequence Arrange recKt, recKt={ k1,k2... ..., kn }, 1≤n≤11;
If Step 3: the adjacent slope k of judgementiAnd ki+1Meet such as ki*ki+1< 0 or Then by month monthi+1It is defined as inflection point, wherein ε is parameter preset and 0≤ε≤1;Finally obtain inflection point sequence rec;If inflection point sequence rec In be not present inflection point, then by the business electrical amount sequence from target data concentrate delete;
Step 4: the feature tuple sequence RecTuple of electricity consumption enterprise is established according to the inflection point sequence rec of acquisition, RecTuple={ tuple1,tuple2,……,tuplen, wherein 2≤n≤11;Feature tuple is wherein used between inflection point two-by-two tupleiIt indicates, tuplei={ k, t }, wherein k indicates that the slope between two inflection points, t indicate current inflection point and previous inflection point Between month it is poor;
Step 5: calculating the distance between two feature tuples based on maximum common subsequence;
Step 6: carrying out the cluster based on distance for business electrical amount sequence and extracting the mainstream distributed area each to cluster Between obtain target industry use power mode.
In above technical scheme, the slope k of two neighboring moon electricity consumption is calculatediExpression formula it is as follows:
In above technical scheme, the distance between two feature tuples are calculated based on maximum common subsequence in step 5 Method it is as follows:
Step 5.1: being directed to the feature tuple sequence RecTuple of Liang Ge electricity consumption enterprise respectivelyA={ tuple1, tuple2,……,tuplenAnd RecTupleB={ tuple1,tuple2····tuplemDetermine both public sub- sequence Column:
The common subsequence is expressed as follows:
ComSeqAB={ A: < tuple1,A,tuple2,A... ..., tuplek,A>,
B:<tuple1,B,tuple2,B... ..., tuplek,B>}。
Step 5.2: calculating separately feature tuple RecTupleAAnd RecTupleBDuty ratio, the duty ratio is characterized The sum of inflection point duration of the corresponding common subsequence of tuple sequence is than the sum of inflection point duration in upper feature tuple sequence;
Step 5.3: according to the duty ratio of the feature tuple sequence of two electricity consumption business electrical amounts A and B, feature tuple sequence And maximum common subsequence calculates the distance dis (A, B) of electricity consumption A and B, expression formula is as follows:
Wherein, α and β is two-part weight proportion, meets alpha+beta=1 and α >=0, β >=0;len(RecTupleA) represent The length of feature tuple.
On the other hand, the present invention provides a kind of electricity consumption schema extraction systems, comprising:
Target data set module is established, for the electricity consumption enterprise according to target industry in particular year use corresponding with month Electricity data establishes business electrical amount sequence rec, rec={ < month1,power1>,<month2,power2>,……
<monthn,powern>, 1≤n≤12, monthnN-th month is represented, powernRepresent n-th month electricity;
Slope characteristics sequence generating module calculates two neighboring moon electricity consumption for the electricity consumption sequence to each enterprise Slope obtains slope characteristics sequence recKt, recKt={ k1,k2... ..., kn }, 1≤n≤11;
Inflection point sequence generating module, if for judging adjacent slope kiAnd ki+1Meet such as ki*ki+1< 0 orThen by month monthi+1It is defined as inflection point, wherein ε is parameter preset and 0≤ε≤1;It finally obtains Inflection point sequence rec;If inflection point is not present in inflection point sequence rec, which is concentrated from target data and is deleted;
Feature tuple sequence generating module establishes the feature tuple of electricity consumption enterprise for the inflection point sequence rec according to acquisition Sequence RecTuple, RecTuple={ tuple1,tuple2,……,tuplen, wherein 2≤n≤11;Wherein two-by-two inflection point it Between use feature tuple tupleiIt indicates, tuplei={ k, t }, wherein k indicates that the slope between two inflection points, t indicate current inflection point Month between previous inflection point is poor;
Distance calculation module, for calculating the distance between two feature tuples based on maximum common subsequence;
Pattern clustering module, for carrying out the cluster based on distance and extracting each to cluster for business electrical amount sequence The electricity consumption that mainstream distributed area obtains target industry is more close.
Advantageous effects of the invention:
Present invention employs the design philosophy of maximum common subsequence, so that the method for the present invention is more suitable enterprise of the same trade and use The similar situation of comparision of quantity of electricity, so that the extraction for user power utilization mode provides more correct input, it is more efficient to construct Schema extraction model lay the foundation;
The present invention calculates the distance between business electrical amount by introducing based on the duty ratio of maximum common subsequence, The similarity of direct solution business electrical amount, and consider holding for maximum common subsequence duration and entire feature tuple sequence The ratio between continuous time, method is more succinct intuitive, high-efficient and does not lose correctness, has very strong adaptability;
The present invention is based on the distance functions provided to carry out electricity consumption schema extraction using clustering method, has higher efficiency.
Detailed description of the invention
Fig. 1 is two feature tuples of the invention apart from calculation flow chart;
Fig. 2 is electricity consumption schema extraction flow diagram of the present invention.
Specific embodiment
The invention will be further described below in conjunction with the accompanying drawings.Following embodiment is only used for clearly illustrating the present invention Technical solution, and not intended to limit the protection scope of the present invention.
Embodiment:
Electricity consumption schema extraction method, comprising the following steps:
Step 1: establishing target data set: according to the electricity consumption enterprise of target industry in particular year use corresponding with month Electricity data establishes business electrical amount sequence rec, rec={ < month1,power1>,<month2,power2>,……
<monthn,powern>, 1≤n≤12, monthnN-th month is represented, powernRepresent n-th month electricity;
Raw data set S is as follows, wherein each [...] represents 12 months data of an enterprise, each (... ...) Represent the electricity consumption of some month of enterprise:
First enterprise:
[(1,4153),(2,2954),(3,527),(4,2603),(5,4419),(6,1113),(7,3264),(8, 3220),(9,4941),(10,4779),(11,462),(12,3959)];
Second enterprise:
[(1,2134),(2,1918),(3,700),(4,69),(5,3120),(6,1061),(7,417),(8,2070), (9,1669),(10,1014),(11,2845),(12,3357)];
Third enterprise:
[(1,1458),(2,4360),(3,1088),(4,2423),(5,3306),(6,1521),(7,538),(8, 1594),(9,66),(10,438),(11,3746),(12,657)];
4th enterprise:
[(1,3196),(2,3364),(3,790),(4,3189),(5,4498),(6,714),(7,891),(8,688), (9,3041),(10,3826),(11,1362),(12,1227)];
5th enterprise:
[(1,544),(2,3413),(3,4959),(4,2440),(5,851),(6,4667),(7,2935),(8, 2609),(9,88),(10,2220),(11,1942),(12,3215)];
6th enterprise:
[(1,3413),(2,4688),(3,3531),(4,1287),(5,4508),(6,210),(7,3376),(8, 2855),(9,1477),(10,300),(11,4089),(12,3438)];
7th enterprise:
[(1,2718),(2,4052),(3,2568),(4,4905),(5,3113),(6,948),(7,1819),(8, 4609),(9,1850),(10,3688),(11,1475),(12,1986)];
8th enterprise:
[(1,3212),(2,4912),(3,761),(4,3175),(5,431),(6,2716),(7,367),(8, 4065),(9,4623),(10,591),(11,2177),(12,1831)];
9th enterprise:
[(1,4220),(2,1477),(3,3718),(4,1846),(5,1682),(6,3054),(7,4983),(8, 3289),(9,1613),(10,2612),(11,3658),(12,1213)];
Tenth enterprise:
[(1,2050),(2,3177),(3,363),(4,4603),(5,4588),(6,170),(7,1976),(8, 2833),(9,98),(10,3456),(11,2512),(12,3666)];
11st enterprise:
[(1,4538),(2,3637),(3,1615),(4,2712),(5,220),(6,3339),(7,3272),(8, 3868),(9,2053),(10,3952),(11,1568),(12,4229)];
12nd enterprise:
[(1,3719),(2,206),(3,3539),(4,1156),(5,1473),(6,4719),(7,1661),(8, 680),(9,4696),(10,2677),(11,4719),(12,4285)];
13rd enterprise:
[(1,3970),(2,3575),(3,1875),(4,1456),(5,1026),(6,4320),(7,4926),(8, 429),(9,3861),(10,1122),(11,2674),(12,1745)];
14th enterprise:
[(1,4480),(2,4136),(3,4829),(4,620),(5,3100),(6,3305),(7,3193),(8, 942),(9,255),(10,1327),(11,3657),(12,378)];
15th enterprise:
[(1,2216),(2,359),(3,4556),(4,1559),(5,2663),(6,1148),(7,3641),(8, 2395),(9,447),(10,4900),(11,1000),(12,1323)]。
Step 2: the slope for calculating two neighboring moon electricity consumption to the electricity consumption sequence of each enterprise obtains slope characteristics sequence Arrange recKt, recKt={ k1,k2... ..., kn }, 1≤n≤11;
Calculate the slope of two neighboring moon electricityRec then is recorded for any onet, can To obtain its corresponding slope characteristics recKt={ k1,k2... ..., kn},1≤n≤11;
If Step 3: the adjacent slope k of judgementiAnd ki+1Meet such as ki*ki+1< 0 or Then by the moon Part monthi+1It is defined as inflection point, wherein ε is parameter preset and 0≤ε≤1;Finally obtain inflection point sequence rec;If inflection point sequence Inflection point is not present in rec, then the business electrical amount sequence is concentrated from target data and is deleted;
When judging inflection point in view of there is no data before starting January in month, terminates and do not counted behind December in month According to;Partial mode can be lost by directly erasing data in January and data in December from inflection point, therefore by January and December It is defined as inflection point.
The inflection point sequence for calculating each enterprise according to this step is as follows, in the present embodiment ε=0.6:
First enterprise:
[(1,4153.0),(3,527.0),(5,4419.0),(6,1113.0),(7,3264.0),(8,3220.0),(9, 4941.0),(10,4779.0),(11,462.0),(12,3959.0)];
Second enterprise:
[(1,2134.0),(2,1918.0),(3,700.0),(4,69.0),(5,3120.0),(6,1061.0),(7, 417.0),(8,2070.0),(10,1014.0),(11,2845.0),(12,3357.0)];
Third enterprise:
[(1,1458.0),(2,4360.0),(3,1088.0),(5,3306.0),(6,1521.0),(7,538.0),(8, 1594.0),(9,66.0),(10,438.0),(11,3746.0),(12,657.0)];
4th enterprise:
[(1,3196.0),(2,3364.0),(3,790.0),(4,3189.0),(5,4498.0),(6,714.0),(7, 891.0),(8,688.0),(9,3041.0),(10,3826.0),(11,1362.0),(12,1227.0)];
5th enterprise:
[(1,544.0),(2,3413.0),(3,4959.0),(5,851.0),(6,4667.0),(7,2935.0),(8, 2609.0),(9,88.0),(10,2220.0),(11,1942.0),(12,3215.0)];
6th enterprise:
[(1,3413.0),(2,4688.0),(4,1287.0),(5,4508.0),(6,210.0),(7,3376.0),(8, 2855.0),(10,300.0),(11,4089.0),(12,3438.0)];
7th enterprise:
[(1,2718.0),(2,4052.0),(3,2568.0),(4,4905.0),(6,948.0),(7,1819.0),(8, 4609.0),(9,1850.0),(10,3688.0),(11,1475.0),(12,1986.0)];
8th enterprise:
[(1,3212.0),(2,4912.0),(3,761.0),(4,3175.0),(5,431.0),(6,2716.0),(7, 367.0),(8,4065.0),(9,4623.0),(10,591.0),(11,2177.0),(12,1831.0)];
9th enterprise:
[(1,4220.0),(2,1477.0),(3,3718.0),(4,1846.0),(5,1682.0),(7,4983.0), (9,1613.0),(11,3658.0),(12,1213.0)];
Tenth enterprise:
[(1,2050.0),(2,3177.0),(3,363.0),(4,4603.0),(5,4588.0),(6,170.0),(7, 1976.0),(8,2833.0),(9,98.0),(10,3456.0),(11,2512.0),(12,3666.0)];
11st enterprise:
[(1,4538.0),(3,1615.0),(4,2712.0),(5,220.0),(6,3339.0),(7,3272.0),(8, 3868.0),(9,2053.0),(10,3952.0),(11,1568.0),(12,4229.0)];
12nd enterprise:
[(1,3719.0),(2,206.0),(3,3539.0),(4,1156.0),(5,1473.0),(6,4719.0),(7, 1661.0),(8,680.0),(9,4696.0),(10,2677.0),(11,4719.0),(12,4285.0)];
13rd enterprise:
[(1,3970.0),(2,3575.0),(3,1875.0),(5,1026.0),(6,4320.0),(7,4926.0), (8,429.0),(9,3861.0),(10,1122.0),(11,2674.0),(12,1745.0)];
14th enterprise:
[(1,4480.0),(2,4136.0),(3,4829.0),(4,620.0),(5,3100.0),(6,3305.0),(7, 3193.0),(8,942.0),(9,255.0),(11,3657.0),(12,378.0)];
15th enterprise:
[(1,2216.0),(2,359.0),(3,4556.0),(4,1559.0),(5,2663.0),(6,1148.0),(7, 3641.0),(9,447.0),(10,4900.0),(11,1000.0),(12,1323.0)]。
Step 4: the feature tuple sequence RecTuple of electricity consumption enterprise is established according to the inflection point sequence rec of acquisition, RecTuple={ tuple1,tuple2,……,tuplen, wherein 2≤n≤11;Feature tuple is wherein used between inflection point two-by-two Tuplei indicates that tuplei={ k, t }, wherein k indicates that the slope between two inflection points, t indicate that current inflection point is turned with previous Month between point is poor;
For the slope characteristics recK of any one recordt={ k1,k2... ..., kn, 1≤n≤11, if adjacent slope ki And ki+1Meet such as ki*ki+1< 0 orThen by month monthi+1It is defined as inflection point, wherein 0≤ε≤ 1;It above can be in the hope of inflection point sequence the recSeq={ < month of rec1,power1>,<month2,power2>... ...,< monthn,powern>, wherein 1≤n≤11;If inflection point is not present in rec, by the business electrical amount sequence from target data It concentrates and deletes;
It is as follows that the inflection point sequence obtained according to step 3 can calculate feature tuple sequence:
First enterprise:
[(-1813.0,2),(1946.0,2),(-3306.0,1),(2151.0,1),(-44.0,1),(1721.0,1), (-162.0,1),(-4317.0,1),(3497.0,1)];
Second enterprise:
[(-216.0,1),(-1218.0,1),(-631.0,1),(3051.0,1),(-2059.0,1),(-644.0,1), (1653.0,1),(-528.0,2),(1831.0,1),(512.0,1)];
Third enterprise:
[(2902.0,1),(-3272.0,1),(1109.0,2),(-1785.0,1),(-983.0,1),(1056.0,1), (-1528.0,1),(372.0,1),(3308.0,1),(-3089.0,1)];
4th enterprise:
[(168.0,1),(-2574.0,1),(2399.0,1),(1309.0,1),(-3784.0,1),(177.0,1),(- 203.0,1),(2353.0,1),(785.0,1),(-2464.0,1),(-135.0,1)];
5th enterprise:
[(2869.0,1),(1546.0,1),(-2054.0,2),(3816.0,1),(-1732.0,1),(-326.0,1), (-2521.0,1),(2132.0,1),(-278.0,1),(1273.0,1)];
6th enterprise:
[(1275.0,1),(-1700.5,2),(3221.0,1),(-4298.0,1),(3166.0,1),(-521.0,1), (-1277.5,2),(3789.0,1),(-651.0,1)];
7th enterprise:
[(1334.0,1),(-1484.0,1),(2337.0,1),(-1978.5,2),(871.0,1),(2790.0,1), (-2759.0,1),(1838.0,1),(-2213.0,1),(511.0,1)];
8th enterprise:
[(1700.0,1),(-4151.0,1),(2414.0,1),(-2744.0,1),(2285.0,1),(-2349.0, 1),(3698.0,1),(558.0,1),(-4032.0,1),(1586.0,1),(-346.0,1)];
9th enterprise:
[(-2743.0,1),(2241.0,1),(-1872.0,1),(-164.0,1),(1650.5,2),(-1685.0, 2),(1022.5,2),(-2445.0,1)];
Tenth enterprise:
[(1127.0,1),(-2814.0,1),(4240.0,1),(-15.0,1),(-4418.0,1),(1806.0,1), (857.0,1),(-2735.0,1),(3358.0,1),(-944.0,1),(1154.0,1)];
11st enterprise:
[(-1461.5,2),(1097.0,1),(-2492.0,1),(3119.0,1),(-67.0,1),(596.0,1),(- 1815.0,1),(1899.0,1),(-2384.0,1),(2661.0,1)];
12nd enterprise:
[(-3513.0,1),(3333.0,1),(-2383.0,1),(317.0,1),(3246.0,1),(-3058.0,1), (-981.0,1),(4016.0,1),(-2019.0,1),(2042.0,1),(-434.0,1)];
13rd enterprise:
[(-395.0,1),(-1700.0,1),(-424.5,2),(3294.0,1),(606.0,1),(-4497.0,1), (3432.0,1),(-2739.0,1),(1552.0,1),(-929.0,1)];
14th enterprise:
[(-344.0,1),(693.0,1),(-4209.0,1),(2480.0,1),(205.0,1),(-112.0,1),(- 2251.0,1),(-687.0,1),(1701.0,2),(-3279.0,1)];
15th enterprise:
[(-1857.0,1),(4197.0,1),(-2997.0,1),(1104.0,1),(-1515.0,1),(2493.0, 1),(-1597.0,2),(4453.0,1),(-3900.0,1),(323.0,1)]。
Step 5: calculating the distance between two feature tuples based on maximum common subsequence;
The thought of maximum common subsequence refers to: for two sequence A={ x1,x2... ..., xk,……xm... ..., xp... ... and B={ y1,y2... ..., yt,……yv... ..., yu... ... yq, meet x in two sequences of A and Bk=yt, xm,=yv, xp=yuAnd remaining element is unequal, then the common subsequence of A and B is { xk,xm,xpOr { yt,yv,yu}。
If two tuple tuplepAnd tupleq, slope kp, kqMeet kp/kq∈ [1- λ, 1+ λ], and duration tp, tqIt is full Foot | tp-tq|/max(tp,tq) ∈ [0, a], then two tuples are referred to as " public " tuple, wherein 0≤λ≤1.
For feature tuple sequence RecTupleA={ tuple1,tuple2,……,tuplenAnd RecTupleB= {tuple1,tuple2····tuplem, the common subsequence of the two is ComSeqAB={ A: < tuple1,A,tuple2, A ... ..., tuplek, A>, B:<tuple1,B,tuple2, B ... ..., tuplek,B>}。
Only with the feature tuple of first enterprise and second enterprise in the present embodiment for example:
Calculating distance between the two according to following two feature tuple, steps are as follows:
RecTupleA:
(-1813.0,2),(1946.0,2),(-3306.0,1),(2151.0,1),(-44.0,1),(1721.0,1),(- 162.0,1),(-4317.0,1),(3497.0,1)]
RecTupleB:
(-216.0,1),(-1218.0,1),(-631.0,1),(3051.0,1),(-2059.0,1),(-644.0,1), (1653.0,1),(-528.0,2),(1831.0,1),(512.0,1)]
(1) the public tuple of maximum of two feature tuples is sought, the method is as follows:
Public tuple sequence is solved using the method for maximum public substring, i.e. public substring is only required
Relative ranks are consistent, continuous without requiring, and can obtain public tuple sequence are as follows:
ComSeqAB=A:<(2151.0,1), (1721.0,1), (3497.0,1)>,
B:<(3051.0,1),(1653.0,1),(1831.0,1)>}。
Judgment method are as follows:
According to this step wherein λ=0.6, a=0.3;Judging characteristic tuple (2151,1) and (3051,
1) whether be public tuple process it is as follows:
2151/3051=0.7 ∈ [1-0.6,1+0.6]
(1-1)/max (1,1)=0 ∈ [0,0.3]
Therefore, feature tuple (2151,1) and (3051,1) are public tuples.Remaining with
This analogizes, and can obtain public tuple sequence ComSeqAB={ A: < (2151.0,1), (1721.0,1), (3497.0,1) >, B:<(3051.0,1), (1653.0,1), (1831.0,1)>}.
(2) duty ratio of two feature tuples is sought:
According to the meaning of duty ratio: the ratio between the duration of common subsequence duration and entire feature tuple sequence, Pass through formulaCalculate the duty ratio of feature tuple A are as follows:
SpaceTA=(1+1+1)/(2+2+1+1+1+1+1+1)=0.33
SpaceTB=(1+1+1)/(1+1+1+1+1+1+1+2+1+1)=0.3
(3) distance of two feature tuples is sought? this step can be according to deriving above.
The distance between electricity consumption curve Ah and B can be calculated according to the following formula:
Wherein α=0.5, β=0.5;
Wherein the meaning of len (RecTupleB) is the length of RecTupleB.
Step 6: carrying out the cluster based on distance for business electrical amount sequence and extracting the mainstream distributed area each to cluster Between obtain target industry use power mode.
It can quantify the cluster between two electricity consumption curves using the distance defining method based on maximum common subsequence.It adopts With the unsupervised learning clustering method based on density, business electrical data are clustered.It generates several to cluster: C1, C2... ..., CkThe mainstream distributed area for extracting electricity consumption every month that each clusters just constitutes the use power mode to cluster:
Pattern(Ci)={ < month1,minK1, maxK1>,<month2,minK2,maxK2>... ...,<monthn, minKn,maxKn>},1≤n≤12.
Thus, it is possible to obtain the use power mode Pattern of the sector yearindustry{Pattern(C1),……, Pattern(Ck)}。
Above data is divided into three in the present embodiment to cluster.
First clusters are as follows:
[(-1813.0,2),(1946.0,2),(-3306.0,1),(2151.0,1),(-44.0,1),(1721.0,1), (- 162.0,1), (- 4317.0,1), (3497.0,1)],
[(1275.0,1),(-1700.5,2),(3221.0,1),(-4298.0,1),(3166.0,1),(-521.0,1), (- 1277.5,2), (3789.0,1), (- 651.0,1)],
[(-2743.0,1),(2241.0,1),(-1872.0,1),(-164.0,1),(1650.5,2),(-1685.0, 2), (1022.5,2), (- 2445.0,1)],
[(-344.0,1),(693.0,1),(-4209.0,1),(2480.0,1),(205.0,1),(-112.0,1),(- 2251.0,1), (- 687.0,1), (1701.0,2), (- 3279.0,1)],
[(-395.0,1),(-1700.0,1),(-424.5,2),(3294.0,1),(606.0,1),(-4497.0,1), (3432.0,1),(-2739.0,1),(1552.0,1),(-929.0,1)]。
Second clusters are as follows:
[(-216.0,1),(-1218.0,1),(-631.0,1),(3051.0,1),(-2059.0,1),(-644.0,1), (1653.0,1), (- 528.0,2), (1831.0,1), (512.0,1)],
[(2902.0,1),(-3272.0,1),(1109.0,2),(-1785.0,1),(-983.0,1),(1056.0,1), (- 1528.0,1), (372.0,1), (3308.0,1), (- 3089.0,1)],
[(168.0,1),(-2574.0,1),(2399.0,1),(1309.0,1),(-3784.0,1),(177.0,1),(- 203.0,1),(2353.0,1),(785.0,1),(-2464.0,1),(-135.0,1)]。
Third clusters are as follows:
[(1127.0,1),(-2814.0,1),(4240.0,1),(-15.0,1),(-4418.0,1),(1806.0,1), (857.0,1), (- 2735.0,1), (3358.0,1), (- 944.0,1), (1154.0,1)],
[(-1461.5,2),(1097.0,1),(-2492.0,1),(3119.0,1),(-67.0,1),(596.0,1),(- 1815.0,1), (1899.0,1), (- 2384.0,1), (2661.0,1)],
[(-3513.0,1),(3333.0,1),(-2383.0,1),(317.0,1),(3246.0,1),(-3058.0,1), (-981.0,1),(4016.0,1),(-2019.0,1),(2042.0,1),(-434.0,1)]。
The mainstream power mode for extracting every month can obtain the use electrical feature to cluster:
First use electrical feature to cluster are as follows:
[(1,<-3512,-688>),(2,<-1831,3333>),(3,<-688,3051>),(4,<-2051,3051>), (5,<147,382>),(6,<147,382>),(7,<147,382>),(8,<147,382>),(9,<-431,382>),(10,<- 431,382>)];
Second use electrical feature to cluster are as follows:
[(1,<168,1127>),(2,<-2814,-2574>),(3,<1854,4240),(4,<-15,1854>),(5,<- 4418,-3780>),(6,<177,1860>),(7,<-203,857>),(8,<-2735,2353>),(9,<785,3358>), (10,-2464,-944),(11,<-135,1154>)];
What third clustered uses electrical feature are as follows:
[(1,<-2743,-344>),(2,<-1461,4197>),(3,<-4209,1097>),(4,<-2492,2480>), (5,<-277,3119>),(6,<-1881,1650>),(7,<-1685,596>),(8,<-1685,-277>),(9,<-766, 1701>),(10,<-766,1701>),(11,<-3279,2661>)];
The 4th use electrical feature to cluster are as follows:
[(1,<1275,2902>),(2,<-3272,2207>,(3,<-1700332>),(4,<-286,400>),(5,<- 286,400>),(6,<-286,400>),(7,<-286,400>),(8,<-286,400>),(9,<-286,400>),(10,<- 286,400>),(11,<-3089,511>)]。
Present invention employs the design philosophy of maximum common subsequence, so that the method for the present invention is more suitable enterprise of the same trade and use The similar situation of comparision of quantity of electricity, so that the extraction for user power utilization mode provides more correct input, it is more efficient to construct Schema extraction model lay the foundation.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art For member, without departing from the technical principles of the invention, several improvement and deformations can also be made, these improvement and deformations Also it should be regarded as protection scope of the present invention.

Claims (9)

1. a kind of electricity consumption schema extraction method, characterized in that the following steps are included:
Step 1: establishing target data set: according to the electricity consumption enterprise of target industry in particular year electricity consumption corresponding with month Data establish business electrical amount sequence rec, rec={ < month1,power1>,<month2,power2>,……
<monthn,powern>, 1≤n≤12, monthnN-th month is represented, powernRepresent n-th month electricity;
Step 2: the slope for calculating two neighboring moon electricity consumption to the electricity consumption sequence of each enterprise obtains slope characteristics sequence recKt, recKt={ k1,k2... ..., kn }, 1≤n≤11;
If Step 3: the adjacent slope k of judgementiAnd ki+1Meet such as ki*ki+1< 0 or Then by month monthi+1It is defined as inflection point, wherein ε is parameter preset and 0≤ε≤1;Finally obtain inflection point sequence rec;If inflection point sequence rec In be not present inflection point, then by the business electrical amount sequence from target data concentrate delete;
Step 4: establishing feature the tuple sequence RecTuple, RecTuple=of electricity consumption enterprise according to the inflection point sequence rec of acquisition {tuple1,tuple2,……,tuplen, wherein 2≤n≤11;Feature tuple tuple is wherein used between inflection point two-by-twoiIt indicates, tuplei={ k, t }, wherein k indicates that the slope between two inflection points, t indicate the month between current inflection point and previous inflection point Difference;
Step 5: calculating the distance between two feature tuples based on maximum common subsequence;
Step 6: carrying out the cluster based on distance for business electrical amount sequence and extracting the mainstream distributed area that each clusters obtaining Power mode is used to target industry.
2. electricity consumption schema extraction method according to claim 1, characterized in that calculate the slope of two neighboring moon electricity consumption kiExpression formula it is as follows:
3. electricity consumption schema extraction method according to claim 1, characterized in that based on maximum common subsequence in step 5 The method for calculating the distance between two feature tuples is as follows:
Step 5.1: being directed to the feature tuple sequence RecTuple of Liang Ge electricity consumption enterprise respectivelyA={ tuple1,tuple2,……, tuplenAnd RecTupleB={ tuple1,tuple2....tuplemDetermine both common subsequence:
The common subsequence is expressed as follows:
ComSeqAB={ A: < tuple1,A,tuple2,A... ..., tuplek,A>,
B:<tuple1,B,tuple2,B... ..., tuplek,B>};
Step 5.2: calculating separately feature tuple RecTupleAAnd RecTupleBDuty ratio, the duty ratio is characterized tuple The sum of inflection point duration of the corresponding common subsequence of sequence is than the sum of inflection point duration in upper feature tuple sequence;
Step 5.3: according to the feature tuple sequence of two electricity consumption business electrical amounts A and B, feature tuple sequence duty ratio and Maximum common subsequence calculates the distance dis (A, B) of electricity consumption A and B, and expression formula is as follows:
Wherein, α and β is two-part weight proportion, meets alpha+beta=1 and α >=0, β >=0;len(RecTupleA) represent characteristic element The length of group.
4. electricity consumption schema extraction method according to claim 3, characterized in that the method for determining the common subsequence of the two It is as follows: if two tuple tuplepAnd tupleq, slope kp, kqMeet kp/kq∈ [1- λ, 1+ λ], and duration tp, tqMeet | tp-tq|/max(tp,tq) ∈ [0, a], then two tuples are referred to as " public " tuple, wherein 0≤λ≤1, λ and a are preset Parameter.
5. electricity consumption schema extraction method according to claim 1, characterized in that ε value 0.6.
6. electricity consumption schema extraction method according to claim 1, characterized in that step 6 includes:
It generates several to cluster: C1, C2... ..., CkThe mainstream distributed area for extracting electricity consumption every month that each clusters just constitutes What is clustered uses power mode:
Pattern(Ci)={ < month1,minK1, maxK1>,<month2,minK2,maxK2>... ...,<monthn,minKn, maxKn>, 1≤n≤12,
Obtain certain industry year uses power mode Patternindustry{Pattern(C1),……,Pattern(Ck)}。
7. a kind of electricity consumption schema extraction system, characterized in that include:
Target data set module is established, for the electricity consumption enterprise according to target industry in particular year electricity consumption corresponding with month Data establish business electrical amount sequence rec, rec={ < month1,power1>,<month2,power2>,……
<monthn,powern>, 1≤n≤12, monthnN-th month is represented, powernRepresent n-th month electricity;
Slope characteristics sequence generating module calculates the slope of two neighboring moon electricity consumption for the electricity consumption sequence to each enterprise Obtain slope characteristics sequence recKt, recKt={ k1,k2... ..., kn }, 1≤n≤11;
Inflection point sequence generating module, if for judging adjacent slope kiAnd ki+1Meet such as ki*ki+1< 0 orThen by month monthi+1It is defined as inflection point, wherein ε is parameter preset and 0≤ε≤1;It finally obtains Inflection point sequence rec;If inflection point is not present in inflection point sequence rec, which is concentrated from target data and is deleted;
Feature tuple sequence generating module establishes the feature tuple sequence of electricity consumption enterprise for the inflection point sequence rec according to acquisition RecTuple, RecTuple={ tuple1,tuple2,……,tuplen, wherein 2≤n≤11;Wherein used between inflection point two-by-two Feature tuple tupleiIt indicates, tuplei={ k, t }, wherein k indicates that the slope between two inflection points, t indicate current inflection point with before Month between one inflection point is poor;
Distance calculation module, for calculating the distance between two feature tuples based on maximum common subsequence;
Pattern clustering module, for carrying out the cluster based on distance for business electrical amount sequence and extracting the mainstream each to cluster What distributed area obtained target industry uses power mode.
8. electricity consumption schema extraction system according to claim 7, characterized in that the distance calculation module further includes maximum Common subsequence generation module, for being directed to the feature tuple sequence RecTuple of Liang Ge electricity consumption enterprise respectivelyA={ tuple1, tuple2,……,tuplenAnd RecTupleB={ tuple1,tuple2....tuplemDetermine both common subsequence:
The common subsequence is expressed as follows:
ComSeqAB={ A: < tuple1,A,tuple2,A... ..., tuplek,A>,
B:<tuple1,B,tuple2,B... ..., tuplek,B>}。
9. electricity consumption schema extraction system according to claim 8, characterized in that
The distance calculation module, also execution following steps:
Calculate separately feature tuple RecTupleAAnd RecTupleBDuty ratio, it is corresponding that the duty ratio is characterized tuple sequence The sum of inflection point duration of common subsequence than the sum of inflection point duration in upper feature tuple sequence;
Step 5.3: according to the feature tuple sequence of two electricity consumption business electrical amounts A and B, feature tuple sequence duty ratio and Maximum common subsequence calculates the distance dis (A, B) of electricity consumption A and B, and expression formula is as follows:
Wherein, α and β is two-part weight proportion, meets alpha+beta=1 and α >=0, β >=0;Len represents the length of feature tuple.
CN201811402866.XA 2018-11-23 2018-11-23 A kind of electricity consumption schema extraction method and system Active CN109783528B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811402866.XA CN109783528B (en) 2018-11-23 2018-11-23 A kind of electricity consumption schema extraction method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811402866.XA CN109783528B (en) 2018-11-23 2018-11-23 A kind of electricity consumption schema extraction method and system

Publications (2)

Publication Number Publication Date
CN109783528A true CN109783528A (en) 2019-05-21
CN109783528B CN109783528B (en) 2019-09-10

Family

ID=66496477

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811402866.XA Active CN109783528B (en) 2018-11-23 2018-11-23 A kind of electricity consumption schema extraction method and system

Country Status (1)

Country Link
CN (1) CN109783528B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104158175B (en) * 2014-04-28 2016-08-24 广东电网公司佛山供电局 A kind of computational methods of power system distribution transformer terminals real-time electricity consumption classed load
CN105893385A (en) * 2015-01-04 2016-08-24 伊姆西公司 Method and device for analyzing user behavior
CN106709822A (en) * 2017-03-14 2017-05-24 国家电网公司 Industry power consumption data correlation mining method and device
US20170147733A1 (en) * 2015-11-25 2017-05-25 International Business Machines Corporation Tool to provide integrated circuit masks with accurate dimensional compensation of patterns
CN107896160A (en) * 2017-10-27 2018-04-10 中国科学技术大学 A kind of data center network flowmeter factor method based on distributed system
CN108460410A (en) * 2018-02-08 2018-08-28 合肥工业大学 Electricity consumption mode identification method and system, the storage medium of citizen requirement side
CN108460789A (en) * 2018-03-19 2018-08-28 国家基础地理信息中心 A kind of artificial earth's surface timing variations on-line detecting system and method

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104158175B (en) * 2014-04-28 2016-08-24 广东电网公司佛山供电局 A kind of computational methods of power system distribution transformer terminals real-time electricity consumption classed load
CN105893385A (en) * 2015-01-04 2016-08-24 伊姆西公司 Method and device for analyzing user behavior
US20170147733A1 (en) * 2015-11-25 2017-05-25 International Business Machines Corporation Tool to provide integrated circuit masks with accurate dimensional compensation of patterns
CN106709822A (en) * 2017-03-14 2017-05-24 国家电网公司 Industry power consumption data correlation mining method and device
CN107896160A (en) * 2017-10-27 2018-04-10 中国科学技术大学 A kind of data center network flowmeter factor method based on distributed system
CN108460410A (en) * 2018-02-08 2018-08-28 合肥工业大学 Electricity consumption mode identification method and system, the storage medium of citizen requirement side
CN108460789A (en) * 2018-03-19 2018-08-28 国家基础地理信息中心 A kind of artificial earth's surface timing variations on-line detecting system and method

Also Published As

Publication number Publication date
CN109783528B (en) 2019-09-10

Similar Documents

Publication Publication Date Title
CN103812872B (en) A kind of network navy behavioral value method and system based on mixing Di Li Cray process
CN104063480B (en) A kind of load curve parallel clustering method of electrically-based big data
CN111507412B (en) Voltage missing value filling method based on historical data auxiliary scene analysis
CN110781332A (en) Electric power resident user daily load curve clustering method based on composite clustering algorithm
CN106709035A (en) Preprocessing system for electric power multi-dimensional panoramic data
CN111210170B (en) Environment-friendly management and control monitoring and evaluation method based on 90% electricity distribution characteristic index
CN109190890A (en) A kind of user behavior analysis method based on custom power consumption data
CN111461189B (en) Resident rhythm detection method based on non-invasive measurement technology
CN109766950A (en) A kind of industrial user&#39;s short-term load forecasting method based on form cluster and LightGBM
CN107704888B (en) Data identification method based on combined clustering deep learning neural network
CN104036360A (en) User data processing system and processing method based on magcard attendance behaviors
CN101980211A (en) Machine learning model and establishing method thereof
CN109634940A (en) A kind of typical low pressure platform area&#39;s electricity consumption model building method based on magnanimity low-voltage platform area electricity consumption data
CN105868183A (en) Method and device for predicting staff demission
CN105956125A (en) Patent monitoring system and method
CN110244099A (en) Stealing detection method based on user&#39;s voltage
CN105608105B (en) It is a kind of that method is recommended based on the music for listening to context
CN102799627A (en) Data association method based on first-order logic and nerve network
CN115759393A (en) Cumulative load baseline prediction method based on ensemble learning
CN108596227A (en) A kind of leading influence factor method for digging of user power utilization behavior
CN111461197A (en) Spatial load distribution rule research method based on feature extraction
CN109783528B (en) A kind of electricity consumption schema extraction method and system
CN106022640B (en) Electric quantity index checking system and method
CN105447767A (en) Power consumer subdivision method based on combined matrix decomposition model
CN103246814A (en) Personal electric device state identification method based on K-means modeling

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