Load identification based on optimal matching and decomposition to household systems

Abstract

The present invention discloses a system for decomposing loads to households based on optimal matching, including a data acquisition module, the data acquisition module including an acquisition terminal and an electric meter, the acquisition terminal being used to acquire current data and voltage data at the electric meter box; the acquisition terminal including a load identification module and a decomposition to households module, the load identification module being used to identify the loads of all users mounted under the electric meter box to obtain load identification results; the decomposition to households module attributing the loads in the load identification results to the users under the electric meter box based on the load identification results and electricity consumption data. The system for decomposing loads to households based on optimal matching of the present invention improves the accuracy of decomposition to households compared to the previous method of matching only by power. By adopting a method of optimizing the matching of electric quantity, the method with the highest degree of matching of electric quantity among all decomposition methods can be obtained, thereby improving the accuracy of decomposing electric quantity.

Description

Technical Field

[0001] The invention relates to a load identification and decomposition system based on optimal matching, belonging to the technical field of load identification. Background Art

[0002] Under the background of vigorously developing the global energy Internet, my country is promoting the construction of smart grids in an orderly manner. Realizing two-way interactive smart electricity consumption has become the main development goal of the power system. The key to developing smart electricity consumption is to obtain residential electricity load information and conduct intelligent analysis. Residential electricity load monitoring is mainly divided into two categories: invasive and non-invasive. Invasive load monitoring obtains load information by installing monitoring devices on residential electrical equipment. Although the load monitoring is highly accurate, it is costly and affects the daily life of residents, and it is difficult to maintain in the later stage. Non-invasive load monitoring does not require entering the residents' homes. It only installs collection devices on the outside to obtain the user's total electricity load information, and identifies the load status of each electrical equipment in the home through intelligent analysis, which is relatively simple and convenient. However, the existing conventional non-invasive load monitoring method is realized by modifying or adding monitoring devices, and still faces problems such as high investment, high cost, wide scope, and difficulty in promotion. The accuracy and timeliness of intelligent analysis cannot meet actual needs.

[0003] In the prior art, the data granularity that can be provided by the down-hung electricity meter is too large and the data is relatively small. In simple scenarios with fewer electrical appliances, load identification can be better broken down to households. However, in complex scenarios with more electrical appliances, only using the minute-level data of the electricity meter for decomposition will result in a low decomposition rate or misdecomposition.

[0004] Therefore, a new load identification method based on optimal matching and decomposition to household systems is needed to solve the above problems. Summary of the invention

[0005] Purpose of the invention: In view of the problems existing in the prior art, the present invention provides a load identification and decomposition system based on optimal matching.

[0006] A load identification and acquisition system based on optimal matching includes a data acquisition module, wherein the data acquisition module includes an acquisition terminal and an electric meter.

[0007] The acquisition terminal is arranged at the meter box on the user's power supply side, and is used to collect current data and voltage data at the meter box;

[0008] The electric meter is a hanging electric meter of the electric meter box, which is used to collect the user's electricity consumption data;

[0009] The acquisition terminal includes a load identification module and a household decomposition module.

[0010] The load identification module is used to perform load identification on the loads of all users mounted under the meter box according to the current data and voltage data at the meter box to obtain a load identification result;

[0011] The household decomposition module attributes the load in the load identification result to the user under the meter box based on the load identification result and the electricity consumption data.

[0012] Furthermore, the load identification result includes the type of electrical appliance and the decomposed power consumption.

[0013] Furthermore, the electricity consumption data includes voltage, current, active power, reactive power and positive active electrical energy.

[0014] Furthermore, the household decomposition module attributes the load in the load identification result to the user under the meter box based on the load identification result and the power consumption data, including the following steps:

[0015] Step 1: Data preprocessing: Store the load identification data set D with a startup time T0 before the current time n and a power consumption data sequence, wherein the power consumption data sequence includes an active power sequence P and a forward active electric energy sequence E, and adjacent data in the active power sequence P are differentiated to obtain an active power difference sequence P Δ , the adjacent data in the positive active electric energy sequence E are differentiated to obtain the positive active electric energy difference sequence E Δ ;

[0016] Step 2: Match based on power similarity:

[0017] Traverse the load identification data set D n If a load has a difference with the active power sequence P Δ One of the items satisfies the following conditions at the same time: the time difference between the two is less than T s , the power difference is less than P s And the percentage difference is less than Th s , then the two are considered to be matched successfully: the load belongs to the user, the first decomposition result R1 is obtained, and the load is removed from the load identification data set D n The second load identification data set D′ is obtained by eliminating n , removing the user's electricity usage data from the electricity usage data sequence to obtain a second electricity usage data sequence;

[0018] Step 3: Power-based optimization matching:

[0019] The third data set and the fourth data set are optimized and matched by using the optimization matching model to obtain a power selection vector, wherein the third data set is the second load identification data set D′n The fourth data set is the second active power difference sequence P′ of the second power consumption data sequence Δ , use the power selection vector to determine the user to which the load belongs, obtain the second decomposition result R2, and remove the load from the second load identification data set D′ n The third load identification data set D″ is obtained by eliminating n , removing the user's electricity usage data from the second electricity usage data sequence to obtain a third electricity usage data sequence;

[0020] Step 4: Optimized matching compensation based on power consumption:

[0021] Traverse the third load identification data set D″ n If a load meets the following conditions: the continuous operation time of the load is less than the minimum collection time and the load operation period does not cross the collection time point, the fifth data set and the sixth data set are optimized and matched by using the optimization matching model to obtain the power selection vector, where the fifth data set is the third load identification data set D″ n The sixth data set is the third positive active electric energy difference sequence E″ of the third electric power data sequence Δ , use the power selection vector to determine the user to which the load belongs, and obtain the third decomposition result R3;

[0022] Step 5: Optimization matching based on T time power:

[0023] Traverse the load identification data set D n All loads in , the seventh data set and the eighth data set are optimized and matched by using the optimization matching model to obtain the T time power selection vector, where the seventh data set is the load identification data set D n The eighth data set is the remaining active power energy sequence {e″′1-∑e r1 ,e″′2-∑e r2 ,...,e″′ n -∑e rn}, where e″′ n is the nth member in the positive active energy difference sequence of the power consumption data sequence, ∑e rn The sum of the load identification data power decomposed to user n in the first decomposition result R1, the second decomposition result R2 and the third decomposition result R3 is used to determine the user to which the load belongs using the T time power selection vector, and the fourth decomposition result R4 is obtained;

[0024] Step 6: Determine the load decomposition result according to the priority: the priority of the first decomposition result R1>the priority of the second decomposition result R2=the priority of the third decomposition result R3>the priority of the fourth decomposition result R4.

[0025] The present invention improves the accuracy of household decomposition: compared with the previous method of matching only by power, the accuracy of household decomposition is improved. Improve the accuracy of household electricity decomposition: by optimizing the matching of electricity, the method with the highest matching degree of electricity among all decomposition methods can be obtained, thereby improving the accuracy of decomposition of electricity.

[0026] Furthermore, in step 1, the T time is 15 minutes, and the electricity consumption data is minute-level data.

[0027] Furthermore, in step 2, Ts is 120 seconds, Ps is 30w, and Th s is 3%.

[0028] Furthermore, the optimized matching model in step 3 and step 4 optimizes the matching of the first data set and the second data set by the following formula:

[0029]

[0030] Where X is the first data set, X=[x1 x2 … x n ] T , Y is the second data set, Y=[y1 y2 … y m ] T ,y i is the i-th data in the second data set, R is the selection vector, R=[r1 r2 … r n ] T , n is the data volume of the first data set X and the selection vector R, m is the data volume of the second data set Y, and the selection vector with the smallest value in the above formula is the optimal matching vector.

[0031] Furthermore, the first data set in step 3 is a load identification data set D n The second data set is the active power difference sequence P Δ A dataset consisting of the largest m1 members.

[0032] Furthermore, the minimum acquisition time in step 4 is 1 minute.

[0033] Beneficial effects: The load identification and decomposition system based on optimal matching of the present invention adopts multiple load decomposition methods to improve the accuracy of decomposition to households and the accuracy of decomposition to households. Compared with the previous method of matching only by power, the accuracy of decomposition to households is improved. By adopting the method of optimizing the matching of electricity, the method with the highest degree of electricity matching among all decomposition methods can be obtained, thereby improving the accuracy of decomposition of electricity. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1It is a flow chart of the load identification and decomposition to household system based on optimal matching of the present invention;

[0035] Figure 2 This is a flowchart of how the household module attributes the load to the user under the meter box based on the load identification results and electricity consumption data. DETAILED DESCRIPTION

[0036] The preferred embodiments of the present invention will be described below in conjunction with the accompanying drawings to more clearly and completely illustrate the technical solutions of the present invention.

[0037] See also Figure 1 and Figure 2 As shown, the load identification and household decomposition system based on optimal matching of the present invention is intended to improve the low decomposition rate of the original household decomposition solution in complex environments.

[0038] The specific technical solutions are as follows:

[0039] Overall functional architecture: The system includes data acquisition module, load identification module, and household decomposition module.

[0040] Data acquisition module: includes real-time voltage and current signal acquisition on the meter box side and minute-level data acquisition function of the electric meter hanging under the meter box.

[0041] The real-time voltage and current signal collection on the meter box side is required to be continuously collected at a frequency of at least 6400 Hz to ensure the accuracy of the collected signals. The minute-level data collection of the electric meter under the meter box requires the real-time voltage, current, active power, reactive power, and positive active electric energy of the electric meter under the meter box to be collected every minute.

[0042] Load identification module: Based on the real-time voltage and current signals collected from the meter box, load identification of all users' loads mounted under the meter box is realized, including the identification of electrical appliance types and the decomposition of power consumption.

[0043] Decomposition to household module: Based on the load identification results output by the load identification module and the minute-level data of the electric meter under the meter box collected, the specific individual load in the load identification results is attributed to a specific user under the meter box.

[0044] Core technologies:

[0045] The innovation mainly focuses on the decomposition into household modules. The detailed steps of the decomposition into household modules are now described.

[0046] Step 1: Data preprocessing:

[0047] Load identification data:

[0048] Considering the deviation between load identification data and minute-level data, the load identification data set D with a start time within 15 minutes of the current time is retained. n , the rest of the data is cleaned up regularly.

[0049] Minute-level data:

[0050] Store the minute-level data of the meter box hanging electric meter obtained from the data acquisition module, and save the latest 15 minutes of data, among which the active power sequence is recorded as P, the positive active electric energy sequence is recorded as E, and the difference between the two sequences is calculated, and the active power difference sequence is recorded as P Δ , the positive active electric energy difference sequence is recorded as E Δ .

[0051] Step 2: Matching based on power similarity:

[0052] Traverse the load identification data set D n All members in if they are consistent with the active power difference sequence P Δ One of the following conditions is met:

[0053] 1. The time difference is less than T s (T s According to the actual situation, 120 seconds can be used);

[0054] 2. The power difference is less than P s (P s The value is 30w), and the percentage difference is less than Th s (Th s Take the value 3%);

[0055] The match is considered successful. The load identification data member is considered to belong to the user, and both are removed from the sequence.

[0056] Among them, the load identification data set D n , denoted as D n ={D1,D2,...,D k ,...D m} is a collection of load identification data, and a single load identification data member D k ={p dk ,t dk ,e dk ,...}, where p dk Represents the power of the load identification data, t dk represents its starting time, e dk Represents its power consumption.

[0057] Active power difference sequence P Δ Described as P Δ={p1,p2,...,p q ,...p m};

[0058] Positive active electric energy difference sequence E Δ Described as E Δ ={e1,e2,...,e q ,...e m};

[0059] The data sampling time series T is described as T = {t1, t2, ..., t q ,...t m};

[0060] like k makes |t dj -t k | <T s And|p dj -p k | <P s and ---Condition 1

[0061] It is considered that the load identification data D j Belongs to user k. And the load identification data D j The power consumption data of user k is obtained from the load identification dataset D n and electricity consumption data series. The decomposition result R1={{k,D j},...}, where k and j make the corresponding load identification data set D n And the data in the electricity consumption data sequence meets condition 1.

[0062] After eliminating all load identification data and power consumption data that meet condition 1, a second load identification data set D′ will be generated. n and the second power consumption data sequence. Recorded as D′ n ={d∈D n |The electricity consumption data of any user does not meet condition 1}.

[0063] The active power difference sequence after elimination is recorded as P′ Δ ; The data sampling time series after elimination is recorded as T'; the positive active electric energy difference series after elimination is recorded as E' Δ .

[0064] Step 3: Power-based optimization matching:

[0065] First, let's introduce the calculation model used for optimization matching: the optimization matching model:

[0066] Assume that data set 1 is Ω1, which contains n data and has feature x. Its feature sequence can be expressed as X = [x1 x2 …x n ] T Data set 2 is Ω2, which contains m data. It has feature y, and its feature sequence can be expressed as Y = [y1 y2 … y m ] T R is a matrix of m*n, which can be expressed as R=[r1 r2 … r n ] means, where r1...r n is a column vector, r n It can be expressed as r n =[r n1 r n2 ... r nm ]. The entire matrix is ​​expressed as:

[0067]

[0068] The value can be calculated according to the following formula:

[0069]

[0070] In the formula, r i represents the device selected by the i-th device, and X is a column vector.

[0071] Constraints: And r ks is a positive integer, k<=n;

[0072] In this model, we can always find a selection vector R such that The value represented by is the smallest, then this vector is called its optimal matching vector.

[0073] When applied to this problem, the Y sequence is transformed from the active power difference sequence P Δ Instead, the X sequence is composed of the identification data set D n Power replacement.

[0074] Considering the limited amount of calculation, we can use the power difference sequence P Δ Select the largest m members from the second load identification data set D′ n Select the largest n members from the . By comparing and judging the values, we can get the R vector with the smallest value, which represents which user it belongs to. After the decomposition to the household is completed, it is removed from the corresponding sequence accordingly to obtain the third load identification data set D″ n and a third electricity usage data sequence.

[0075] Step 4: Optimized matching compensation based on power:

[0076] Since the active power sequence is recorded as P and the collection granularity is once per minute, it may not be possible to collect the active power fluctuation caused by the load running for less than 1 minute. However, the positive active electric energy sequence is recorded as E as the accumulated amount, which can reflect the power fluctuation of the short-term load. So traverse the third load identification data set D″ n All members of , if the following conditions are met:

[0077] 1. The load continuous operation time is less than the collection granularity (1 minute)

[0078] 2. The load operation period does not span across the collection time point

[0079] Using the optimal matching method, the Y sequence is transformed from the positive active electric energy difference sequence E Δ Instead, the X sequence is replaced by the load power in the identification data set Dn. Optimization matching is performed.

[0080] Step 5: Optimization matching based on 15 minutes of power consumption:

[0081] For each meter box connected user corresponding to the active electric energy difference sequence E Δ Summing, we get ΣE Δ , which is the 15-minute total active electric energy difference sequence. At the same time, subtract the load that has been decomposed in step 2, and finally get the remaining active electric energy sequence {e″′1-∑e r1 ,e″′2-∑e r2 ,...,e″′ n -∑e rn}, where e″′ n is the nth member in the positive active energy difference sequence, ∑e rn is the sum of the electricity e of the load identification data decomposed to user n in the decomposition result R1+R2+R3. It replaces the Y sequence in the optimal matching method, and the X sequence is composed of the load identification data set D″ n "The load power is replaced to optimize the matching.

[0082] Step 6 Final decision based on historical data and optimized matching results

[0083] After following steps 2 to 5, each load may obtain multiple decomposition results (e.g., multiple starts and stops of an electrical appliance). The multiple decomposition results are finally decided according to the priority to determine the final user. When the present invention is implemented, the priority is:

[0084] 1. Step 2 results;

[0085] 2. The results of steps 3 and 4;

[0086] 3. The result of step 5.

[0087] Experimental protocol:

[0088] Step 1:

[0089] The previously collected load voltage and current data are randomly superimposed to simulate the voltage and current sampling data collected at the meter box. The sampling is performed at intervals of 1 minute to simulate the minute-level data of the electric meter under the meter box collected by the collector. 2 appliances are superimposed, 4 appliances are superimposed, and 6 appliances are superimposed, with 20 sets of simulation data for each structure. The load samples used include air conditioners, kettles, electric heating, battery cars, etc.

[0090] Step 2:

[0091] A comparative experiment was conducted using a household-based decomposition method based on optimized matching and a method that only used power for matching.

[0092] Note: The load identification module and the acquisition module use the same module.

[0093] Step 3:

[0094] Scoring Methodology:

[0095] The correct decomposition of household electricity is selected as the scoring standard. Since the accuracy of load identification will also affect the final decomposition of electricity, some samples with errors in the load identification module are eliminated, and the final result is shown in Table 1.

[0096] Table 1. Test results

[0097]

[0098] It can be seen that compared with only power matching, the present invention can greatly improve the accuracy of load identification and decomposition into household power consumption, especially in complex electrical environment.

[0099] The present invention can improve the accuracy of household decomposition: compared with the previous method of matching only by power, the accuracy of household decomposition is improved. It can also improve the accuracy of household electricity decomposition: by optimizing the matching of electricity, the method with the highest electricity matching degree among all decomposition methods can be obtained, thereby improving the accuracy of decomposition of electricity.

Claims

1. Load identification and acquisition system based on optimal matching, characterized in that: It includes a data acquisition module, which includes a data acquisition terminal and an electric meter. The acquisition terminal is arranged at the meter box on the user's power supply side, and is used to collect current data and voltage data at the meter box; The electric meter is a hanging electric meter of the electric meter box, which is used to collect the user's electricity consumption data; The acquisition terminal includes a load identification module and a household decomposition module. The load identification module is used to perform load identification on the loads of all users mounted under the meter box according to the current data and voltage data at the meter box to obtain a load identification result; The household decomposition module attributes the load in the load identification result to the user under the meter box based on the load identification result and the electricity consumption data.

2. The load identification and decomposition system based on optimal matching according to claim 1, characterized in that: The load identification result includes the type of electrical appliance and the decomposed power.

3. The load identification and decomposition system based on optimal matching according to claim 1, characterized in that: The electricity consumption data includes voltage, current, active power, reactive power and positive active electric energy.

4. The load identification and decomposition system based on optimal matching according to claim 1, characterized in that: The decomposition module to households attributes the load in the load identification result to the user under the meter box based on the load identification result and the power consumption data, including the following steps: Step 1: Data preprocessing: Store the load identification data set D with a startup time T0 before the current time n and a power consumption data sequence, wherein the power consumption data sequence includes an active power sequence P and a forward active electric energy sequence E, and adjacent data in the active power sequence P are differentiated to obtain an active power difference sequence P Δ , the adjacent data in the positive active electric energy sequence E are differentiated to obtain the positive active electric energy difference sequence E Δ ; Step 2: Match based on power similarity: Traverse the load identification data set D n If a load has a difference with the active power sequence P Δ One of the items satisfies the following conditions at the same time: the time difference between the two is less than T s , the power difference is less than P s And the percentage difference is less than Th s , then the two are considered to be matched successfully: the load belongs to the user, the first decomposition result R1 is obtained, and the load is removed from the load identification data set D n The second load identification data set D' is obtained by eliminating n , removing the user's electricity usage data from the electricity usage data sequence to obtain a second electricity usage data sequence; Step 3: Power-based optimization matching: The third data set and the fourth data set are optimized and matched by using the optimization matching model to obtain a power selection vector, wherein the third data set is the second load identification data set D' n The fourth data set is the second active power difference sequence P' of the second power consumption data sequence. Δ , use the power selection vector to determine the user to which the load belongs, obtain the second decomposition result R2, and remove the load from the second load identification data set D' n The third load identification data set D" is obtained by eliminating n , removing the user's electricity usage data from the second electricity usage data sequence to obtain a third electricity usage data sequence; Step 4: Optimized matching compensation based on power consumption: Traverse the third load identification data set D" n If a load meets the following conditions: the continuous operation time of the load is less than the minimum collection time and the load operation period does not cross the collection time point, the fifth data set and the sixth data set are optimized and matched using the optimization matching model to obtain the power selection vector, where the fifth data set is the third load identification data set D" n The sixth data set is the third positive active electric energy difference sequence E" of the third electric power data sequence Δ , use the power selection vector to determine the user to which the load belongs, and obtain the third decomposition result R3; Step 5: Optimization matching based on T time power: Traverse the load identification data set D n All loads in , the seventh data set and the eighth data set are optimized and matched using the optimization matching model to obtain the T time power selection vector, where the seventh data set is the load identification data set D n The eighth data set is the remaining active power energy sequence {e”'1-∑e r1 ,e”'2-∑e r2 ,...,e"' n -∑e rn }, where e'' n is the nth member in the positive active energy difference sequence of the power consumption data sequence, ∑e rn The sum of the load identification data power decomposed to user n in the first decomposition result R1, the second decomposition result R2 and the third decomposition result R3 is used to determine the user to which the load belongs using the T time power selection vector, and the fourth decomposition result R4 is obtained; Step 6: Determine the load decomposition result according to the priority: the priority of the first decomposition result R1>the priority of the second decomposition result R2=the priority of the third decomposition result R3>the priority of the fourth decomposition result R4.

5. The load identification and decomposition system based on optimal matching according to claim 4 is characterized in that: In step 1, the T time is 15 minutes, and the electricity consumption data is minute-level data.

6. The load identification and decomposition system based on optimal matching according to claim 4, characterized in that: In step 2, Ts is 120 seconds, Ps is 30w, Th s is 3%.

7. The load identification and decomposition system based on optimal matching according to claim 4, characterized in that: In step 3 and step 4, the optimized matching model optimizes the matching of the first data set and the second data set by the following formula: Where X is the third data set or the fifth data set, X=[x1 x2 … x n ] T , Y is the fourth data set or the sixth data set, Y=[y1 y2 … y m ] T ,y i is the i-th data in the data set Y, R is the selection vector, R=[r1 r2 … r n ] T , n is the data volume of data set X and selection vector R, m is the data volume of data set Y, and the selection vector with the smallest value in the above formula is the optimal matching vector.

8. The load identification and decomposition system based on optimal matching according to claim 4, characterized in that: The third data set in step 3 is the second load identification data set D' n The fourth data set is a second active power difference sequence P' of the second power consumption data sequence. Δ A dataset consisting of the largest m1 members.

9. The load identification and decomposition system based on optimal matching according to claim 4, characterized in that: The minimum acquisition time in step 4 is 1 minute.

Images