CN105607475A - Electrical appliance operating state recognition method based on dual-sliding window - Google Patents

Abstract

The present invention discloses an electrical appliance operating state recognition method based on a dual-sliding window. The method comprises the steps of inputing a user electricity consumption overall load, identifying a steady-state load, detecting steady-state load change, correcting the steady-state load, identifying an electrical appliance operating state, and outputing electrical appliance operating operation sequence. According to the method, in the condition of known user total power consumption and electrical appliance power consumption information, the use condition information of a user electrical appliance is analyzed, the method has an important reference value for the power system data publication and has an important reference meaning for understanding the use condition of the electrical appliance and rationally using electrical energy by the user.

Description

A kind of Running Status of Electrical Appliances recognition methods based on two sliding windows

Technical field

The present invention relates to recognition technology field, particularly a kind of Running Status of Electrical Appliances recognition methods.

Background technology

In the field of identification electrical equipment, how accurately and effectively identification terminal equipment is a very important problem always. At presentThe identification of electrical equipment terminal is often needed to comparatively complicated polytype data, more complicated.

Summary of the invention

The object of the present invention is to provide a kind of Running Status of Electrical Appliances recognition methods based on two sliding windows, based on userElectricity consumption total power consumption information in long a period of time, the method for acquisition consumer electronics equipment running status information.

To achieve these goals, the present invention adopts following technical scheme:

A Running Status of Electrical Appliances recognition methods based on two sliding windows, comprises the steps:

S100, input: user power utilization total load;

S101, the identification of stable state load: adopt two modes that sliding window leapfrogs, load stable state is differentiated, will be accidentallyError point and transition point are eliminated for the impact of stable state load;

S102, stable state load variations detect: taking current data compared with the data jump degree of last data as foundation, carry out steady state dataThe differentiation of section turning point;

S103, stable state load are revised: the average center of being asked for taking each steady state data section and error radius, as foundation, are carried out stable state and bornThe correction of lotus;

S104, Running Status of Electrical Appliances identification: based on the method for integer programming, solve optimum Running Status of Electrical Appliances groupClose;

S105, output: the electric equipment sequence of operation.

Further, step S101 specifically comprises the following steps: initialize first sliding window with data initial starting point: setLength of window N and data variation threshold value Th%, pointwise is analyzed, when the absolute value of k point data and k-1 point data difference is less thanThe Th% of k-1 point data, counts k to refuse to take a passenger into sliding window; If be greater than,, taking k point as starting point, re-start slipThe initialization of window, and be zero by front k-1 assignment in total data, until window is written into N data; Calculation window data allValueNow the data in window are a steady-state process. If last data being written in window after initializing are kData, when k+1 data are less than window averageTh%, by window to unit of front slide; Otherwise, introduce secondIndividual window, same, length of window N and data variation threshold value Th%, re-start initialization taking these data as starting point; ?In the initialization procedure of two windows, in meter window, data amount check is j, establishes j data that are written in window and be the in total dataK data, are greater than threshold value Th% if there is data, i.e. the initialization of second window failure, now by first window forwardA slip j unit, and with average in windowReplace k-j+1 to the k data in total data; If initialized successfully,Second window is written into N data, is now judged as second stable state; Window exchanges afterwards, using second window as theA window is proceeded to slide, and first window empties, as second new follow-up use of window; By above step to negativeCarry steady state point and differentiate, have accidental error or transition point all to reject data origination; The idol that exceeds setting threshold weeding outSo error point and transition point original position, fills up by the average of sliding window; Wherein, N=4, Th=15; Now, k >=2,0≤j≤4.

Further, the mode that obtains stable state load dividing turning point in step S102 is after treatment that step S101 is obtainedTotal power consumption information pointwise input, later a bit before any jump degree be foundation, after judge, whether any is steady state data sectionTurning point, i.e. setting threshold Th%, in the time that rear data exceed the Th% of last some data, is trapezoidal segmentation turnover more afterwardsPoint.

Further, step S103 specifically comprises the following steps: obtained each steady state data section average is marked on number axis,And centered by average, make the circle of uncertainty, less that in the length by two circle of uncertainty intersection on number axis and two error radiusWhether the ratio of radius, as the standard of evaluating coincidence degree, can merge in these averages that judgement is tried to achieve; When coincidence journeyDegree while exceeding the CTh% of setting, is merged into one by two averages, using two means of mean as the average after merging, representative withOne stable state; Otherwise, do not carry out the correction at average center. Wherein, CTh%=60%.

Further, in step S104, solving optimum Running Status of Electrical Appliances combines the model adopting and is:

Ladder total load after treatment is L^t；

Constraints:

$\begin{array}{cc}{\Σ}_{j=1}^{K_m}{Z}_{m,j}^t\≤1& m=1,2,\mathrm{3...},M---\left(2\right)\end{array}$

$Z_{m,j}^t\∈\{0,1\}---\left(3\right)$

$|{\Σ}_{m=1}^M{\Σ}_{j=1}^{K_m}{P}_{m,j}\·Z_{m,j}^t-L^t|\≤\mathrm{\ϵ}---\left(4\right)$

Target:

minε(5)

Wherein, M represents electrical equipment number, and label is 1,2,3 respectively ...., m ..., M; K_mRepresent the state number of m electrical equipment,Represent the K of m electrical equipment_mThe corresponding power consumption number of individual state; K_mWithFor known, according to the actual electrical of familyDevice quantity and power consumption of electrical appliances situation obtain;

Finally solve according to model and constraints the optimum electrical appliance state combination that meets target, each electrical equipment is corresponding?Obtain required household electrical appliance running state information.

Further, step S101 specifically comprises the following steps:

Step1: initialize first sliding window. It is N (suggestion N=4) that sliding window width is set, and (Th builds setting threshold Th%View value is 15). Window initialization procedure is: from data initial starting point, data is written in sliding window one by one,Pointwise is analyzed, when the absolute value of m (m >=2) some data and m-1 point data difference is less than the Th% of m-1 point data,M is counted and refused to take a passenger into sliding window; If be greater than, taking m point as starting point, re-start the initialization of sliding window,And be zero by front m-1 assignment in total data.

Step2: judge whether sliding window initializes successful. When the data amount check being written into when window equals N, initialize successfully redirectStep3; Otherwise redirect Step1, re-starts window and initializes. Meter initialization is written into last in window when completeIndividual data are k data in total data.

Step3: calculate numerical value average in sliding window, be designated asAsk for k+1 numerical value and averageThe absolute value of difference, $\mathrm{\Δ}x=|x_{k+1}-\stackrel{\‾}{x}|,$ And do ratio with average, be designated as $r=\frac{\mathrm{\Δ}x}{\stackrel{\‾}{x}}.$

Step4: judge whether the ratio r of being tried to achieve by Step3 exceedes threshold value Th%. If continue Step5; Otherwise by sliding window toUnit of front slide, redirect Step3.

Step5: second sliding window initializes. Same, it is N (suggestion N=4) that sliding window width is set, setting threshold Th%(recommended value of Th is 15). K+1 the data x of new data of Th% will be exceeded_k+1Being written into a new width isIn the sliding window of N=5, be n=1 by data counts in sliding window.

Step6: ask for k+2 data x_k+2With x_k+1The absolute value of difference, i.e. Δ x=|x_k+2-x_k+1|, calculate and x_k+1RatioValue $ra=\frac{\mathrm{\Δ}x}{x_{k+1}}.$

Step7: whether the ratio r a that judgement is tried to achieve is less than threshold value Th%. Continue Step8 if be less than; Otherwise redirect Step13.

Step8: data amount check counting is added to 1, continue Step9.

Step9: judge whether new sliding window initializes successfully. Whether equal N according to data amount check m in window, if equal N,Initialize successfully, continue Step10; Otherwise, redirect Step12.

Step10: carry out window exchange, proceed to slide second window as first window, first window empties, asSecond new follow-up use of window, continues Step11.

Step11: judge whether data finish. If last data data processing finish, by the data output being disposed;Otherwise, redirect Step12.

Step12: sliding window continues to move forward a unit, redirect Step3.

Step13: the data in second sliding window (n, 0≤n≤3) are removed, by first window to a front slide n unit,And with average in windowReplace k-n+1 to the k data in total data, redirect Step12.

Further, step S103 specifically comprises the following steps:

Between adjacent two state switching points of asking for, the mean value of data, as the steady state power consumption typical value of this segment data, establishes adjacent twoIndividual state switching point k_mWith k_m+1Between have n data, be respectively x₁,x₂,x₃......,x_n, ask for averageWithAverageFor data center, ask for the error radius of this segment data, i.e. error radiusRepresent the worst error scope of this segment data. On number axis with averageFor the center of circle, radius is that r makes the circle of uncertainty, establishesWithRespectivelyFor the average center that two sections of stable states of needs comparison are loaded, its corresponding error radius is respectively r_mAnd r_n, andFor numerical value lessAverage, in the length using two circle of uncertainty intersection on number axis and two error radius, the ratio of that less radius is as evaluationThe standard of coincidence degree, with coincidence factor racAs criterion. When exceeding threshold value (the suggestion threshold of settingValue 60%) time, thinking that the state of two average representatives is same state, need to merge, when havingThisWhen sample m average center need to merge, with the mean value at this m average center,As revised shapeState represents power consumption number; Otherwise the mean data of the steady state power consumption typical value obtaining in step S102 is not revised.

With respect to prior art, the present invention has following beneficial effect:

1, the steady state characteristic of electric load is the key character that characterizes electric equipment operation process, and the differentiation of the stable state of electric load isDifficult point in Running Status of Electrical Appliances identification problem. The present invention innovates and adopts two sliding windows to analyze Power system load data,Effectively solve the stable state discrimination of electric load, and then determined turning point according to the variation of electric power stable state load, and with each steadyAverage center and error radius that state data segment is asked for are foundation, obtain the feature of stable state load;

2, the present invention is based on the steady state characteristic of electric load, utilize the running status of integer programming Optimization Solution electric equipment. RelativelyIn prior art, institute of the present invention extracting method has the following advantages: the first, and when multiple electric equipment operation, because its feature is mutually foldedAdding, are difficult points of prior art, and the present invention utilizes integer programming to solve electric equipment to the identification of multiple Running Status of Electrical AppliancesThe optimum combination of running status, solved the problem of multiple electric equipment operation state recognitions; The second, the inventive method is based on electricityThe steady-state process of power load, the Temporal Data that does not need the collection of a large amount of high frequencies to load, therefore the present invention is less demanding to sample frequency,Data processing is easy, is easy to realize.

Brief description of the drawings

Fig. 1 is method distance block diagram of the present invention.

Fig. 2 is the electrical load stable state method of discrimination flow chart based on two sliding windows in the present invention.

Detailed description of the invention

Below in conjunction with accompanying drawing and example, the present invention will be further described.

A kind of Running Status of Electrical Appliances recognition methods based on two sliding windows of the present invention, uses within longer a period of time based on userElectricity total power consumption information; Specifically comprise the following steps:

S100. input: user power utilization total load. Obtain electricity consumption total load data from the ammeter of the total power portal of user, willThese electricity consumption total load data inputs. The data of choosing in present case are active power data.

Give an example in case at this, the captured meritorious power consumption data of 17 minutes is studied as Given information. Except thisOutside, the known power consumption of electrical appliances information that also has laboratory to capture, as shown in table 1.

Table 1 power consumption of electrical appliances information

S101. stable state load identification.

Adopt two modes that sliding window leapfrogs, accidental error point and status transition point are rejected. Concrete sideMethod is shown in Fig. 2.

S201: it is N=4 that sliding window width is set, setting threshold 15%. Starting first window initializes.

S202: judge whether first window initializes successful. If initialized successfully, redirect S203; Otherwise, at the beginning of inciting somebody to actionThe data assignment that beginningization is failed is zero, taking first non-zero as starting point, restarts first windowInitialize. Last data that meter initializes are successfully k data.

S203: calculate numerical value average in sliding window, be designated asAsk for k+1 numerical value and averageDifference definitelyValue,And do ratio with average, be designated as

S204: judge whether the ratio r of being tried to achieve by S203 exceedes threshold value 15%. Continue S205 if exceed; Otherwise redirectS209。

S205: initialize second sliding window. Same, it is N=4 that sliding window width is set, setting threshold 15%.

S206: ask for k+2 data x_k+2With x_k+1The absolute value of difference, i.e. Δ x=|x_k+2-x_k+1|, calculate withx_k+1Ratio $ra=\frac{\mathrm{\Δ}x}{x_{k+1}}.$

S207: whether the ratio r a that judgement is tried to achieve is less than threshold value 15%. Continue S208 if be less than; Otherwise redirect S2013.

S208: data amount check counting is added to 1, continue S209.

S209: judge whether second sliding window initializes successfully. If the interior data amount check m of window equals 4 and is successfully,Continue S2010; Otherwise, redirect S2012.

S2010: carry out window exchange, proceed to slide second window as first window, first window empties,As second new follow-up use of window, continue S2011.

S2011: judge whether data finish, if finished, by the data output being disposed; Otherwise, redirect S2012.

S2012: sliding window is continued to unit of front slide, redirect S203.

S2013: replace and initialize failed data, redirect S2012 by required window average.

S102. stable state load variations detects.

The total power consumption information pointwise input after treatment that step S101 is obtained, setting threshold 15%, when k (k >=2)Point data value exceed k-1 point data value 15% time, oughtTime, k point is trapezoidal minuteSection turning point, by this kind of method, electrical appliance state switching point is chosen, and state switching point is carried out to record, result asShown in table 2.

Table 2 segmentation turning point

Note: initial time 2016s, stops moment 2745s

S103. stable state load is revised.

Ask for the mean value of data between adjacent two turning pointsSet it as the representative power consumption of this segment dataValue, calculates the error radius at each average center

On number axis with averageFor the center of circle, radius is that r makes the circle of uncertainty, establishesWithFor two average centers of needs comparison,AndFor the less average of numerical value, its corresponding error radius is respectively r_mAnd r_n, calculate coincidence factor racJudge whether rac exceedes threshold value 60%, if exceed, by status merging, when havingLike this in m averageWhen the heart need to merge, with the mean value at this m average center,As the final generation of state after mergingTable power consumption number; Otherwise these two states are two independent states.

A series of power consumption of electrical appliances information to gained further arrange, and obtain final state power consumption information table, as table 3Shown in.

Table 3 state power consumption information table

S104. Running Status of Electrical Appliances identification.

Ladder total load after treatment is L^t

Constraints:

$\begin{array}{cc}{\Σ}_{j=1}^{K_m}{Z}_{m,j}^t\≤1& m=1,\mathrm{2...},M---\left(2\right)\end{array}$

$Z_{m,j}^t\∈\{0,1\}---\left(3\right)$

$|{\Σ}_{m=1}^M{\Σ}_{j=1}^{K_m}{P}_{m,j}\·Z_{m,j}^t-L^t|\≤\mathrm{\ϵ}---\left(4\right)$

Target:

minε(5)

Wherein, electrical equipment number M in subscriber household; M electrical equipment K_mThe corresponding work power consumption of individual stateFor known,Obtain by first step data acquisition.

S105. output: the electric equipment sequence of operation.

Finally can go out optimum electrical appliance state combination according to model solution. The result obtaining is as follows:

1) in the time of initial time 2016, open state 1 pattern of computer, open state 1 pattern of TV, openPhone;

2) in 2028 moment, close computer and TV, open state 1 pattern of electric light and printer;

3) in 2033 moment, close electric light and printer, open state 1 pattern of fan and state 1 pattern of computer;

4) in 2043 moment, the state of fan 1 pattern switched to state 3 patterns, open the state 1 of electric light, printerState 1 pattern of pattern and TV;

5) in 2056 moment, the state of computer 1 pattern is switched to state 2 patterns, the state of TV 1 pattern is switched toState 2 patterns;

6), in 2153 moment, the state of printer 1 pattern is switched to state 3 patterns;

7), in 2270 moment, the state of printer 3 patterns are switched to state 1 pattern;

8) in 2310 moment, close electric light, phone, open heater, the state of fan 3 patterns are switched to state 1Pattern, switches to state 2 patterns by the state of printer 1 pattern;

9) in 2455 moment, closed heater, opened electric light, phone, the state of fan 1 pattern has been switched to state 3Pattern, switches to state 1 pattern by the state of printer 2 patterns;

10), in 2594 moment, electric light and computer have been closed;

11) in 2620 moment, open state 1 pattern of electric light and computer, the state of TV 2 patterns are switched to state 1Pattern;

12) in 2627 moment, close electric light and printer, the state of fan 3 patterns are switched to state 2 patterns;

13) in 2646 moment, close fan and phone, open state 1 pattern of printer;

14) in 2656 moment, closed computer;

15), in 2745 moment, printer and TV have been closed.

Claims (6)

1. the Running Status of Electrical Appliances recognition methods based on two sliding windows, is characterized in that, comprises the steps:

S100, input: user power utilization total load;

S101, the identification of stable state load: adopt two modes that sliding window leapfrogs, load stable state is differentiated, will be accidentallyError point and transition point are eliminated for the impact of stable state load;

S102, stable state load variations detect: taking current data compared with the data jump degree of last data as foundation, carry out steady state dataThe differentiation of section turning point;

S103, stable state load are revised: the average center of being asked for taking each steady state data section and error radius, as foundation, are carried out stable state and bornThe correction of lotus;

S104, Running Status of Electrical Appliances identification: based on the method for integer programming, solve optimum Running Status of Electrical Appliances groupClose;

S105, output: the electric equipment sequence of operation.

2. a kind of Running Status of Electrical Appliances recognition methods based on two sliding windows according to claim 1, its feature existsIn, step S101 specifically comprises the following steps: initialize first sliding window with data initial starting point: set length of window NWith data variation threshold value Th%, pointwise is analyzed, when the absolute value of k point data and k-1 point data difference is less than k-1 point dataTh%, k is counted and is refused to take a passenger into sliding window; If be greater than, taking k point as starting point, re-start the initial of sliding windowChange, and be zero by front k-1 some assignment in total data, until window is written into N data; The average of calculation window dataThisTime data in window be a steady-state process; If last data being written in window after initializing are k data,When k+1 data are less than window averageTh%, by window to unit of front slide; Otherwise, introduce second window,Same, length of window N and data variation threshold value Th%, re-start initialization taking these data as starting point; At second windowInitialization procedure in, meter window in data amount check be j, establishing j data that are written in window is k data in total data,Be greater than threshold value Th% if there is data, i.e. the initialization of second window failure, now that first window is single to front slide jPosition, and with average in windowReplace k-j+1 to the k data in total data; If initialized successfully, i.e. second windowMouth is written into N data, is now judged as second stable state; Window exchanges afterwards, and second window continued as first windowContinuous slip, first window empties, as second new follow-up use of window; By above step, load steady state point is enteredRow is differentiated, and has accidental error or transition point all to reject data origination; The accidental error point that exceeds setting threshold weeding out andTransition point original position, fills up by the average of sliding window; Wherein, N=4, Th=15; Now, k >=2,0≤j≤4.

3. a kind of Running Status of Electrical Appliances recognition methods based on two sliding windows according to claim 2, its feature existsIn, the mode that obtains stable state load dividing turning point in step S102 is the total power consumption information after treatment that step S101 is obtainedPointwise input, later a bit before any jump degree be foundation, after judge, whether any is the turning point of steady state data section,Setting threshold Th%, in the time that rear data exceed the Th% of last some data, is trapezoidal segmentation turning point more afterwards.

4. a kind of Running Status of Electrical Appliances recognition methods based on two sliding windows according to claim 1, its feature existsIn, step S103 specifically comprises the following steps: obtained each steady state data section average is marked on number axis, and taking average asThe circle of uncertainty is made at center, the ratio of that less radius in the length by two circle of uncertainty intersection on number axis and two error radiusAs the standard of evaluating coincidence degree, in these averages that judgement is tried to achieve, whether can merge; When coincidence degree exceedes settingCTh% time, two averages are merged into one, using two means of mean as merge after average, represent same stable state;Otherwise, do not carry out the correction at average center.

5. a kind of Running Status of Electrical Appliances recognition methods based on two sliding windows according to claim 4, its feature existsIn: CTh%=60%.

6. a kind of Running Status of Electrical Appliances recognition methods based on two sliding windows according to claim 1, its feature existsIn: in step S104, solve optimum Running Status of Electrical Appliances and combine the model adopting and be:

Ladder total load after treatment is L^t；

Constraints:

${\Σ}_{j=1}^{K_m}{Z}_{m,j}^t\≤1,m=1,2,3...,M---\left(2\right)$

$Z_{m,j}^t\∈\{0,1\}---\left(3\right)$

$|{\mathrm{\Σ}}_{m=1}^M{\mathrm{\Σ}}_{j=1}^{K_m}{P}_{m,j}\·Z_{m,j}^t-L^t|\≤\mathrm{\ϵ}---\left(4\right)$

Target:

minε(5)

Wherein, M represents electrical equipment number, K_mRepresent the state number of m electrical equipment,Represent the K of m electrical equipment_mIndividualThe corresponding power consumption number of state; K_mWithFor known, obtain according to the actual electrical equipment quantity of family and power consumption of electrical appliances situation;

Finally solve according to model and constraints the optimum electrical appliance state combination that meets target, each electrical equipment is correspondingObtain required household electrical appliance running state information.