CN112785456A - High-loss line electricity stealing detection method based on vector autoregressive model - Google Patents
High-loss line electricity stealing detection method based on vector autoregressive model Download PDFInfo
- Publication number
- CN112785456A CN112785456A CN202110117959.3A CN202110117959A CN112785456A CN 112785456 A CN112785456 A CN 112785456A CN 202110117959 A CN202110117959 A CN 202110117959A CN 112785456 A CN112785456 A CN 112785456A
- Authority
- CN
- China
- Prior art keywords
- loss
- user
- line
- electric quantity
- electricity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000005611 electricity Effects 0.000 title claims abstract description 83
- 238000001514 detection method Methods 0.000 title claims abstract description 11
- 238000000034 method Methods 0.000 claims abstract description 29
- 230000007774 longterm Effects 0.000 claims abstract description 27
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 12
- 238000005316 response function Methods 0.000 claims abstract description 12
- 230000000694 effects Effects 0.000 claims description 17
- 238000004458 analytical method Methods 0.000 claims description 13
- 238000012353 t test Methods 0.000 claims description 13
- 238000012937 correction Methods 0.000 claims description 12
- 230000008859 change Effects 0.000 claims description 9
- 238000000692 Student's t-test Methods 0.000 claims description 4
- 238000007689 inspection Methods 0.000 abstract description 6
- 230000008846 dynamic interplay Effects 0.000 abstract description 5
- 238000011002 quantification Methods 0.000 abstract 1
- 238000012360 testing method Methods 0.000 description 13
- 230000002159 abnormal effect Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 230000006399 behavior Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 230000004044 response Effects 0.000 description 4
- 238000010998 test method Methods 0.000 description 4
- 230000005856 abnormality Effects 0.000 description 3
- 238000007796 conventional method Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000011835 investigation Methods 0.000 description 2
- 238000013112 stability test Methods 0.000 description 2
- 238000001744 unit root test Methods 0.000 description 2
- 230000018199 S phase Effects 0.000 description 1
- 238000001772 Wald test Methods 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000012098 association analyses Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010224 classification analysis Methods 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 230000005612 types of electricity Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Computational Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Economics (AREA)
- Mathematical Physics (AREA)
- Databases & Information Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Operations Research (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Algebra (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A high-loss line electricity stealing detection method based on a vector autoregressive model is characterized in that a long-term dynamic interaction relation exists between line loss electricity quantity of a line and electricity consumption of each subordinate user, and firstly, a long-term equilibrium relation between the line loss electricity quantity and the electricity consumption of the user is analyzed by using edge constraint co-integration inspection; then, constructing a vector autoregressive model of the line loss electric quantity of the line and the electric quantity of each subordinate user, and calculating an impulse response function; and finally, analyzing fluctuation contribution degree between the line loss electricity quantity and the electricity consumption of the user through variance decomposition and quantification, and identifying the user which has obvious influence on the line loss electricity quantity and has the largest fluctuation contribution degree as a suspected electricity stealing user. The method does not require that the user and the line have the same single integer order, and can give a quantitative result of the contribution of the user to the line loss.
Description
Technical Field
The invention belongs to the field of power grid line loss analysis, and relates to a method for establishing a vector autoregressive model for analyzing, detecting and positioning users with abnormal power consumption based on line loss abnormality, so as to perform positioning identification on suspected users with electricity stealing (users with abnormal power consumption).
Background
Under the economic condition of the market, part of illegal operators steal electric energy by various means, and the income loss of power supply enterprises is directly caused. Traditionally, power consumption anomaly detection mainly depends on manual investigation, and due to the lack of data and the lack of directivity of anomaly detection, a large amount of manpower and material resources are consumed, but the effect is poor. At present, the power grid enterprises in China basically realize complete collection of power utilization information, can timely and accurately master power utilization data and customer information of power users, and provide effective technical support for mining, analyzing and identifying electricity stealing users by utilizing the power utilization data.
Production technicians of power supply enterprises summarize indexes with exact physical significance such as zero sequence current, power reversal and electricity meter voltage loss of low-voltage users according to experience, and can accurately identify abnormal electricity utilization behaviors, but the method is only suitable for specific types of electricity stealing methods, and other forms of electricity stealing such as electricity around meters cannot be detected.
Power workers have conducted a great deal of research around data-driven power anomaly detection, both from unsupervised cluster analysis and supervised classification analysis. The researches generally design characteristic index items according to common characteristics that a trend of electricity consumption is reduced, a daily load curve is abnormal, the reporting capacity utilization rate is low and the like, which are shown by electricity stealing of a user, and information such as customer information, payment records and checked times is combined, and then electricity utilization abnormality is identified by pertinently selecting a classification or clustering algorithm. In these researches, the characteristic index item is often designed by using the abnormal change of the power consumption as a core element, and the characteristic index item is easily mistakenly reported for the following reasons: the power consumption behavior characteristics of users in different industries are remarkably different, the power consumption requirements of a considerable part of industries directly depend on the order demand, and the large-amplitude or trend fluctuation of the power consumption of the users is a normal state; an industrial user with a relatively stable power consumption demand may have low power consumption abnormality under external interference such as environmental protection inspection and safety inspection.
In line loss management of power supply enterprises, line loss management of distribution lines has been a key point and a difficulty of line loss management. Investigation shows that in the power distribution link, line loss of 10kV and below accounts for 65% -70% of the total line loss. Because the electricity stealing is an important reason that the line loss rate is high, in actual work, marketing personnel often select the distribution line with the highest line loss rate to check the electricity utilization, and if the line loss rate is more than 5%, the electricity stealing behavior is generally considered to exist. Although detailed power consumption data of all distribution areas under the distribution line are recorded in the integrated line loss management system, due to the lack of an effective data mining analysis method, marketers can only select high-loss lines and then check the power stealing users one by one according to experience, and therefore, the research of an applicable high-loss line power stealing user positioning identification algorithm is urgently needed to improve the power consumption checking work efficiency.
Disclosure of Invention
Although electricity stealing methods are various, most electricity stealing users often adopt a voltage division method or a shunt method to realize equal proportion electricity stealing in order to avoid inspection, the change of the branch line electricity loss quantity is basically in direct proportion to the electricity consumption of the electricity stealing users, and the load curve of the electricity stealing users can obviously show the synchronous change of the line electricity loss quantity. Therefore, the line loss capacity and the load capacity of the electricity stealing users have strong correlation. By utilizing the correlation characteristics, the long-term dynamic interaction action relationship between the line loss electric quantity and the load electric quantity of the electricity stealing users can be mined and analyzed, and the electricity stealing users causing the abnormal fluctuation of the line loss electric quantity are identified.
Therefore, an object of the present invention is to provide a method for detecting electricity stealing on a high-loss line based on a vector autoregressive model, so as to locate and identify suspected users of electricity stealing.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows: a high-loss line electricity stealing detection method based on a vector autoregressive model comprises the following steps:
step 1, determining an electricity stealing high-loss line, acquiring unit time loss electricity quantity data of the electricity stealing high-loss line and unit time electricity quantity data of each subordinate user of the high-loss line in the same time period, and respectively establishing a unit time loss electricity quantity time sequence S ═ S1，S2，…，StAnd the unit time electricity consumption time series Y of each subordinate useri＝{Yi1，Yi2，…YitWherein, i is subordinate user, i is 1, 2, …, n, n is the number of all subordinate users, and t is the length of time sequence;
the above-mentioned determination of the electricity stealing high-loss line is determined according to the line loss rate or line loss fluctuation rate of each line, and is a conventional technique in the art. When the line loss rate is used for determining, a power supply enterprise determines according to local actual conditions that the theoretical line loss of a 10kV distribution line is more than 2% -3%, and generally, a line with the line loss higher than 5% is considered as a high-loss line.
The above-mentioned time series S of the amount of electricity lost per unit time and the time series Y of the amount of electricity used per unit timeiThe same unit time is used. The unit time is a specific time interval, and when the unit time is a day (namely 24 hours), the acquired electricity consumption data of each user directly comprises daily electricity consumption data, so that the daily electricity consumption time sequence of each subordinate user can be directly established; when the unit time is other time intervals such as 15 minutes, 30 minutes, 60 minutes or 12 hours, the time interval electricity consumption data can be calculated according to the obtained electricity consumption data of each user, and then the hour electricity consumption time sequence of each subordinate user is established.
Step 2, respectively constructing the unit time electricity consumption time series Y of each user under the high-loss line1，Y2，Y3...YnUnconstrained error correction of time series S with unit time loss of electric quantityThe positive model is used for checking whether a long-term coordination relation exists between the loss electric quantity of the high-loss line and the electric quantity of each subordinate user through a limit coordination checking method, and if the long-term coordination relation exists, the step 3 is carried out;
the above-mentioned process of establishing the unconstrained error correction model is conventional in the art.
The above mentioned edge-fitting test method is a conventional technique in the art. The constraint condition of the common coordination check method (the Engle-Granger coordination check and the Johansen coordination check) is that the time series of two variables have the same single integer order, and the method cannot be applied when the line loss and the user power consumption are different orders. As dozens of users may exist under a line or a platform area, the difference of the single integral order of the power loss of each user and the line is a common phenomenon, the limit coordination test can simultaneously test the coordination relation among a plurality of variables, and compared with the common coordination test method, the limit coordination test method does not require the single integral of the same order of the variables and breaks through the limitation of the single integral order, so that the method utilizes the limit coordination test method to test the relation among the sequences, and can more stably identify the coordination relation between the power loss and the power consumption of each subordinate user.
Step 3, constructing a unit time electricity consumption time sequence Y of each user under the high-loss line1，Y2，Y3...YnA vector autoregressive model of the unit time loss electric quantity time sequence S (which is beneficial to analyzing dynamic interaction relations between the loss electric quantity of the high-loss line and the electric quantity of each subordinate user in the step 4, namely impulse response function analysis and dynamic variance decomposition), and determining the optimal lag period number of the model to stabilize the vector autoregressive model;
the optimal hysteresis period number of the model is determined according to AIC, SC and HQ information criteria in the vector autoregressive model, and is conventional in the art.
The above-mentioned process of verifying the stability of the vector autoregressive model is a routine technique in the art.
And 4, on the basis of the established vector autoregressive model, analyzing users with long-term and short-term significant influences on the loss electric quantity by adopting a pulse response function, then performing dynamic variance decomposition on the loss electric quantity and each subordinate user, calculating the contribution degree of each user to the change of the loss electric quantity, identifying the user with the significant influence on the loss electric quantity of the high-loss line and the largest fluctuation contribution degree as a suspected electricity stealing user, and performing home inspection.
The above-mentioned impulse response function analysis, dynamic variance decomposition and fluctuation contribution calculation processes in the vector autoregressive model are all conventional in the art.
The long-term significant influence of each user on the loss electric quantity is judged according to the magnitude relation between the t-test statistic of each user variable in the long-term co-integration equation and the 5% significance level: if the t test statistic is more than 5%, indicating that the user has no long-term significant influence on the loss of the electric quantity; if the t test statistic is less than 5%, the user is indicated to have long-term significant influence on the loss of the electric quantity. The sign of the estimates of the individual user variable coefficients also represents whether the long term significant effect is a promotion or a reduction.
The short-term significant influence of each user on the loss electric quantity is judged according to the relation between the t test statistic of each user lag variable in the unconstrained error correction model equation and the 5% significance level by judging whether each user lag variable in the unconstrained error correction model equation has a lag effect on the loss variable, namely the short-term significant influence: if the t test statistic of the user lag variable is more than 5%, indicating that the user has no short-term significant influence on the lost electric quantity; if the t test statistic of the user hysteresis variable is less than 5%, the user is indicated to have hysteresis effect on the loss of the electric quantity, namely short-term significant influence. The positive and negative coefficient estimates for each user lag variable also represent the positive and negative hysteresis effect.
The long-term co-integration equation and the unconstrained error correction model equation are derived from the margin co-integration test.
When the line loss of the distribution line is abnormally fluctuated, the influence of each user of the line loss line on the line loss electric quantity is quantitatively analyzed through the margin covariance check and the impulse response function analysis and the dynamic variance decomposition in the vector autoregressive model. Compared with the traditional association analysis and the granger attribution analysis, the dynamic influence path and fluctuation contribution degree of each user of the line loss line on the line loss electric quantity can be given, the influence degree of each user on the line loss electric quantity is quantized, and the constraint of stationarity is avoided. Compared with the method that marketing departments check on the door one by one, the method reduces checking cost and reduces unnecessary economic loss of power supply enterprises. The method can detect the electricity stealing users in the electricity information acquisition master station, can detect the electricity stealing in an edge calculation mode in an intelligent gateway or an acquisition terminal of a distribution room, and can avoid the problem that the electricity stealing detection accuracy is influenced by negative line loss caused by failure of uploading electricity data of individual users when the distribution room is realized in the edge calculation mode.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
FIG. 2 is a graph of a unit root test of a vector autoregressive model according to an embodiment of the present invention.
Fig. 3 is a graph of the impulse response of the line loss and the power consumption of each individual user according to the embodiment of the present invention.
In the figure, the abscissa represents the impact duration and the ordinate represents the positive and negative reflection.
Detailed Description
Referring to fig. 1 in combination, the invention relates to a high-loss line electricity stealing detection method based on a vector autoregressive model, which comprises the following specific steps:
step 1, determining an electricity stealing high-loss line, acquiring unit time loss electricity quantity data of the electricity stealing high-loss line and unit time electricity quantity data of each subordinate user of the high-loss line in the same time period, and respectively establishing a unit time loss electricity quantity time sequence S ═ S1，S2，…，StAnd the unit time electricity consumption time series Y of each subordinate useri＝{Yi1，Yi2，…YitWherein, i is subordinate user, i is 1, 2, …, n, n is the number of all subordinate users, and t is the length of time sequence;
the above-mentioned unit time may be 15 minutes, 30 minutes, 60 minutes, 12 hours, 24 hours or the like.
Step 2, respectively constructing the subordinate utilities of the high-loss lineElectricity consumption time series Y for user unit time1，Y2，Y3...YnA non-constrained error correction model of the unit time electric quantity loss time sequence S is used for checking whether a long-term coordination relation exists between the electric quantity loss of the high-loss line and the electric quantity consumption of each subordinate user through a margin coordination checking method, and if the long-term coordination relation exists, the step 3 is carried out;
the above mentioned unconstrained error correction model expression is as follows:
wherein, Delta StAnd Δ YtFirst order difference, i, between the branch line loss and the user power0，i1，i2，...，imRespectively representing line loss capacity and lag period numbers of m users, c, gamma and alphakAnd betakRepresenting the parameter to be estimated, mutAre residual terms.
Margin-covariance check by correcting the lag variable S in the model for unconstrained errorst，YtAnd the coefficient is realized by combining the Wald test, and the output result is F statistic. If the value of the F statistic is larger than the upper critical value, the fact that the co-integration relation exists among the variables can be judged without carrying out variable stability test; if the value of the F statistic is smaller than the lower critical value, directly judging that the original hypothesis is true, namely, no co-integration relation exists among the variables; if the value of the F statistic is between the upper critical value and the lower critical value, the conclusion cannot be directly drawn, and the unit root test of the variable needs to be carried out.
Step 3, constructing a unit time electricity consumption time sequence Y of each user under the high-loss line1，Y2，Y3...YnAnd a vector autoregressive model of the unit time loss electric quantity time series S is used for analyzing the dynamic interaction relation between variables (the loss electric quantity of the high-loss line and the electric quantity of each subordinate user). Determining the optimal lag period number of the model according to the AIC, SC and HQ information criteria; after the optimal number of lag periods of the model is determined, the stability of the vector autoregressive model is checked.
The construction of the above-mentioned vector autoregressive model is a conventional technique in the art. Specifically, n user electricity consumptions Y are analyzed by constructing a vector autoregressive modeliThe dynamic interaction relationship between (i ═ 1, 2, …, n) and the loss electric quantity S, and the vector autoregressive (p) model containing n +1 variables can be expressed as:
Zt＝c+A1Zt-1+A2Zt-2...+ApZt-p+εt
wherein T is 1, 2tIs a k-dimensional internal variable vector consisting of n user electric quantities and 1 line loss electric quantity, p is a lag order, T is the number of samples, A1，...，ApIs a matrix of n x n dimensional coefficients, epsilontIs a k-dimensional random perturbation vector.
The selection of the optimal number of lag periods for the model and the testing of the stability of the model are also conventional in the art.
1) Model optimal lag period number selection
The information criteria AIC and SC are used as a standard for measuring the superiority and inferiority of the regression model, and the smaller the value of the information criteria AIC and SC, the more accurate the constructed model is, and the better the conciseness is. AIC and SC are defined as:
wherein
Is the sum of the squares of the residuals of the model, k is the number of parameters in the model, and n is the sample size.
2) Model stability test
The condition that the vector autoregressive (p) model is stable is that all roots of the following characteristic equation fall within a unit circle, and then the vector autoregressive model is stable, and the overall fitting is good and the stability is high.
|Inλp-A1λp-1-A2λp-2-L-Ap|＝0
Wherein InIs an n × n dimensional identity matrix, A1,…,ApIs an n x n dimensional coefficient matrix, with λ being the root of the characteristic equation.
And 4, on the basis of the established vector autoregressive model, analyzing users with long-term and short-term significant influences on the loss electric quantity by adopting a pulse response function, then performing dynamic variance decomposition on the loss electric quantity and each subordinate user, calculating the contribution degree of each user to the change of the loss electric quantity, and identifying the user with the significant influence on the loss electric quantity of the high-loss line and the largest fluctuation contribution degree as a suspected electricity stealing user.
The impulse response function is derived by the following procedure:
any one vector autoregressive model can be represented as an infinite order VMA (∞) process:
Zt+s＝εt+s+C1εt+s-1+C2εt+s-2+L
Cs＝δZt+s/δεt t＝1,2,L T
Csthe ith row and jth column elements of (1) are equal to the disturbance items of the jth user variable in the period t, and the unit impact is added, and when the disturbance in other periods is constant, the influence on the ith variable value of the period t + s is obtained.
Handle CsThe ith row and jth column elements in the middle row are regarded as functions of the lag phase s:
δZi,t+s/δεjt,s＝1,2,3,L
the above equation is an impulse response function describing Z with time t, other variables and early variables unchangedi,t+sTo ZjtThe reaction process of (1). The magnitude of the impulse response function value represents the magnitude of long-term effect of the electric quantity of the user on the electric quantity loss, and the positive and negative of the function value represent the positive and negative of the influence of the electric quantity of the user on the electric quantity loss. If the impulse response function value of a certain user is in a negative value in a short period and in a larger positive value in a long period, the user has a hysteresis effect on the loss electric quantity in the short period, but has a positive significant effect on the loss electric quantity in the long period.
The dynamic variance decomposition and fluctuation contribution degree calculation process is as follows:
the variance decomposition is to further evaluate the importance of different structural impacts by analyzing the contribution degree of each structural impact to the change of an endogenous variable. ZiThe expression of the corresponding variance in the first s-phase is as follows:
wherein: sigmajj＝E(εtεt')
ZiCan be decomposed into k uncorrelated effects to determine the individual perturbation terms relative to ZiThe variance of (a) defines the following equation:
the relative variance contribution Ratio (RVC) is the impact-based variance pair Z according to the jth variableiThe influence of the jth variable on the ith variable is analyzed by the relative contribution of the variance of (a). If RVCj→i(s) a large value means that the j-th variable has a large influence on the i-th variable, i.e., a large contribution; conversely, if RVCj→iWhen(s) is small, the influence of the jth variable on the ith variable is considered to be small, that is, the contribution degree is small.
Example 1
Taking the branch line power loss of a high loss line and the power consumption data of the subordinate private variable users as examples, the daily line power loss of the line from 4 months 1 to 7 months 9 days and the daily power consumption of the subordinate 6 private variable users are collected, and a sequence S and a daily power consumption sequence Y of the subordinate 6 private variable users are respectively established1,Y2,Y3…Y6。
1) Margin integration test
And performing margin co-integration test on the line loss electricity quantity of the line and the electricity consumption sequence of the subordinate dedicated transformer user to obtain the margin test F statistic of the corresponding estimation equation. At this time, the statistics of the margin test F in the case of 3 cases without the time trend item, the time trend item with constraint, and the time trend item with constraint are respectively: f1＝7.0819，F2＝10.053，F3The original assumption that no coordination relationship exists between the variables is rejected at a significance level of 5%, and the long-term coordination relationship exists between the line loss capacity and the power consumption of the next 6 special variable users. And simplifying the processing estimation equation according to the AIC and SC information criteria and the sequence-related LM test statistic to eliminate the unnoticeable orders, and finally setting the model as ARDL (1,0,0,1,0,0, 4). Accordingly, a long-term co-integration relation equation and a corresponding error correction model are obtained as shown in formulas 1 and 2.
S＝0.0301Y1+9.6807Y2-0.8857Y3+23.1207Y4+1.8884Y5-6.4857Y6(1)
The error correction model of the power loss of the distribution line and the following 6 special variable users is as follows:
ΔS＝-2961.5383+10.0099T-0.4326ECM(-1)-0.6178S(-1)+0.0186Y1
+5.9817Y2-0.5472Y3(-1)+14.2863Y4+1.1669Y5-4.0075Y6(-1)
+0.1088ΔY3+3.6154ΔY6(-1)+6.0528ΔY6(-2)+3.6032ΔY6(-3)(2)
wherein ECM (-1) ═ S-0.0301Y1(-1)-9.6807Y2(-1)+0.8857Y3(-1)
-23.1207Y4(-1)-1.8884Y5(-1)+6.4857Y6(-1)
From the long-term covariance equation estimation results, the t-test statistic of each variable accounts for Y5，Y6The coefficient estimation values are all very obvious and have obvious influence on the line loss electricity quantity; y is1，Y2，Y3，Y4The coefficient estimated value of (2) is not significant, and has no significant influence on the line loss capacity. The direction of the estimated value of each variable coefficient indicates Y5Has obvious promoting influence on line loss capacity, Y6Has a significant attenuation effect on the line loss capacity.
From the estimation result of the error correction model, divide by Y1，Y2，Y4，Y5Besides the sequence, the influence of other sequences on the line loss capacity existsA significant hysteresis effect. Wherein Y is3There is a weak negative hysteresis effect; y is6The effect is large and has both positive hysteresis effect and negative hysteresis effect.
2) Construction and evaluation of vector autoregressive model
Five evaluation statistical criteria were obtained by setting different hysteresis periods, and the test values are shown in table 1, where the shaded portion indicates significance at a significance level of 5%.
TABLE 1 model hysteresis determination test
| Lag | LogL | LR | FPE | AIC | SC | HQ |
| 1 | -3496.579 | NA | 9.47e+25 | 79.675 | 81.046* | 80.228* |
| 2 | -3432.570 | 107.881 | 6.87e+25 | 79.338 | 82.078 | 80.443 |
| 3 | -3370.824 | 94.354* | 5.40e+25* | 79.052* | 83.163* | 80.709* |
| 4 | -3339.314 | 43.193 | 8.76e+25 | 79.445 | 84.926 | 81.654 |
As can be seen from the test results in Table 1, when the lag period p is 3, all statistics are significant at the significance level of 5%, i.e. the five evaluation statistical criteria of LR, FPE, AIC, SC, HQ are all that the optimal lag period p is 3. Therefore, it is reasonable to choose to build a vector autoregressive model with lag period number of 3. The model constructed in this example has 7 variables, the optimal hysteresis order is 3, so the roots of the 21 characteristic equations should lie within the unit circle. As can be seen from FIG. 2, the reciprocal of all roots are located within the unit circle, which shows that the vector autoregressive model with lag period number of 3 is well fitted and has high stability.
3) Impulse response analysis and variance decomposition
Analysis of Y with long and short term significant effects on line loss using impulse response functions3，Y5，Y6. The analysis result is shown in fig. 3, and the 3 curves with different colors respectively represent the pulse function change curves of 3 special variable users, that is, respectively represent the responses of the line power loss after the power consumption of the 3 special variable users is impacted in the positive unit.
And carrying out dynamic variance decomposition on the line loss electric quantity and the subordinate special variable users. The results are shown in table 2, and in the first few periods, the predicted variance of the line loss power is mainly explained by the change of the line loss power, and meets the characteristics of strong volatility and difficult prediction. In the long term, about 77% of the variance of the line loss power can be interpreted by the following 6 users with specific variables. Specifically, in the six-factor interpretable portion, Y is5The interpretation degree is the highest and is about 30 percent; second is Y3And Y6The two are roughly equivalent in interpretation, about 17% and 14%, respectively; is again Y2About 8%; finally Y1And Y4All are only about 4%. Summarizing, among six factors that vary the amount of power loss of the driving line, Y5Maximum influence, Y3And Y6Secondly, other factors have a very weak influence.
The analysis shows that the technical loss of the line plays a leading role at the initial stage of the electric energy transmission process, and the variance contribution of the line loss electric quantity mainly comes from Y along with the influence of subordinate special variable users3，Y5，Y6. Wherein Y is5The impact on the amount of power loss is significant. Combining the impulse response analysis described above with Y5The user is determined as the user with the largest suspicion of electricity stealing under the line. Through on-site inspection, the Y is confirmed5The user has illegal electricity stealing behavior.
TABLE 2 variance decomposition chart of line loss and power consumption
Claims (5)
1. A high-loss line electricity stealing detection method based on a vector autoregressive model is characterized by comprising the following steps:
step 1. determinationThe method comprises the steps of obtaining unit time loss electric quantity data of the electricity stealing high-loss line and unit time power consumption data of each user under the high-loss line in the same time period, and respectively establishing a unit time loss electric quantity time sequence S ═ Si,S2,...,StAnd the unit time electricity consumption time series Y of each subordinate useri＝{Yi1，Yi2，YitWherein, i is subordinate user, i is 1, 2, …, n, n is the number of all subordinate users, and t is the length of time sequence;
step 2, respectively constructing the unit time electricity consumption time series Y of each user under the high-loss line1，Y2，Y3...YnA non-constrained error correction model of the unit time electric quantity loss time sequence S is used for checking whether a long-term coordination relation exists between the electric quantity loss of the high-loss line and the electric quantity consumption of each subordinate user through a margin coordination checking method, and if the long-term coordination relation exists, the step 3 is carried out;
step 3, constructing a unit time electricity consumption time sequence Y of each user under the high-loss line1,Y2,Y3…YnDetermining the optimal lag period number of the model to stabilize the vector autoregressive model with the vector autoregressive model of the unit time loss electric quantity time sequence S;
and 4, on the basis of the established vector autoregressive model, analyzing users with long-term and short-term significant influences on the loss electric quantity by adopting a pulse response function, then performing dynamic variance decomposition on the loss electric quantity and each subordinate user, calculating the contribution degree of each user to the change of the loss electric quantity, and identifying the user with the significant influence on the loss electric quantity of the high-loss line and the largest fluctuation contribution degree as a suspected electricity stealing user.
2. The vector autoregressive model-based high-loss line electricity stealing detection method according to claim 1, wherein the determination of the electricity stealing high-loss line in step 1 is determined according to the line loss rate or the line loss fluctuation rate of the line.
3. The method as claimed in claim 1, wherein the model optimal hysteresis period in step 3 is determined according to AIC, SC and HQ information criteria.
4. The method as claimed in claim 1, wherein the analysis of the users with significant long-term influence on the power loss in step 4 is determined according to the magnitude relationship between the t-test statistic and the 5% significance level of each user variable in the long-term co-integration equation: if the t test statistic is more than 5%, indicating that the user has no long-term significant influence on the loss of the electric quantity; if the t test statistic is less than 5%, the user is indicated to have long-term significant influence on the loss of the electric quantity.
5. The method as claimed in claim 1, wherein the analysis of the users with short-term significant effect on the power loss in step 4 is determined according to the magnitude relationship between t-test statistic and 5% significance level of lag variable of each user in the unconstrained error correction model: if the t test statistic of the user lag variable is more than 5%, indicating that the user has no short-term significant influence on the lost electric quantity; if the t test statistic of the user hysteresis variable is less than 5%, the user is indicated to have hysteresis effect on the loss of the electric quantity, namely short-term significant influence.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110117959.3A CN112785456A (en) | 2021-01-28 | 2021-01-28 | High-loss line electricity stealing detection method based on vector autoregressive model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110117959.3A CN112785456A (en) | 2021-01-28 | 2021-01-28 | High-loss line electricity stealing detection method based on vector autoregressive model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112785456A true CN112785456A (en) | 2021-05-11 |
Family
ID=75759377
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110117959.3A Pending CN112785456A (en) | 2021-01-28 | 2021-01-28 | High-loss line electricity stealing detection method based on vector autoregressive model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112785456A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114218522A (en) * | 2021-12-02 | 2022-03-22 | 清华大学 | Station user contribution degree measuring and calculating method based on information transfer entropy and electricity stealing troubleshooting method |
| CN116340441A (en) * | 2023-05-29 | 2023-06-27 | 中山大学 | Rainfall-air temperature relationship analysis method and device based on vector autoregressive model |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110231503A (en) * | 2019-07-08 | 2019-09-13 | 南方电网科学研究院有限责任公司 | Stealing user recognition positioning method of the area Gao Suntai based on Granger CaFpngerusality test |
| CN110322371A (en) * | 2019-07-08 | 2019-10-11 | 长沙理工大学 | The area Gao Suntai multiplexing electric abnormality user based on multiple linear regression analysis detects localization method |
-
2021
- 2021-01-28 CN CN202110117959.3A patent/CN112785456A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110231503A (en) * | 2019-07-08 | 2019-09-13 | 南方电网科学研究院有限责任公司 | Stealing user recognition positioning method of the area Gao Suntai based on Granger CaFpngerusality test |
| CN110322371A (en) * | 2019-07-08 | 2019-10-11 | 长沙理工大学 | The area Gao Suntai multiplexing electric abnormality user based on multiple linear regression analysis detects localization method |
Non-Patent Citations (6)
| Title |
|---|
| 唐冬来;刘友波;熊智临;马铁丰;苏童;: "基于时空关联矩阵的配电台区反窃电预警方法", 电力系统自动化, no. 19 * |
| 嵇灵;牛东晓;劳咏昶;路妍;: "基于ARDL边限协整检验的中国电力消费与经济增长关系实证研究", 华东电力, no. 08, 24 August 2012 (2012-08-24), pages 1299 - 1302 * |
| 殷涛;薛阳;杨艺宁等: ""基于向量自回归模型的高损线路窃电监测"", 《中国电机工程学报》, vol. 42, no. 3, 5 February 2022 (2022-02-05), pages 1015 - 1023 * |
| 裴丽倩、刘伊生、原境彪: ""基于VAR模型的民用建筑运行能耗影响因素分析"", 《建筑节能》, vol. 48, no. 353, 25 July 2020 (2020-07-25), pages 156 - 162 * |
| 裴丽倩;刘伊生;原境彪;: "基于VAR模型的民用建筑运行能耗影响因素分析", 建筑节能, no. 07, 25 July 2020 (2020-07-25) * |
| 金晟: ""基于线损电量归因分析的低误报率窃电检测方法"", 《中国优秀硕士学位论文全文数据库 (工程科技Ⅱ辑)》, 15 January 2023 (2023-01-15), pages 042 - 2083 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114218522A (en) * | 2021-12-02 | 2022-03-22 | 清华大学 | Station user contribution degree measuring and calculating method based on information transfer entropy and electricity stealing troubleshooting method |
| CN114218522B (en) * | 2021-12-02 | 2024-04-09 | 清华大学 | Method for measuring and calculating contribution degree of users in area based on information transfer entropy and method for checking fraudulent use of electricity |
| CN116340441A (en) * | 2023-05-29 | 2023-06-27 | 中山大学 | Rainfall-air temperature relationship analysis method and device based on vector autoregressive model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110223196B (en) | Anti-electricity-stealing analysis method based on typical industry feature library and anti-electricity-stealing sample library | |
| CN110097297A (en) | A kind of various dimensions stealing situation Intellisense method, system, equipment and medium | |
| CN110763929A (en) | Intelligent monitoring and early warning system and method for convertor station equipment | |
| CN103678766B (en) | A kind of abnormal Electricity customers detection method based on PSO algorithm | |
| CN110222991B (en) | Metering device fault diagnosis method based on RF-GBDT | |
| CN110264107B (en) | Large data technology-based abnormal diagnosis method for line loss rate of transformer area | |
| CN113884975B (en) | Out-of-tolerance ammeter detection method based on improved dynamic line loss estimation | |
| CN111008193B (en) | Data cleaning and quality evaluation method and system | |
| CN103062076A (en) | Method for calibrating performance curve of single water pump in urban water plant pump station | |
| CN113032454A (en) | Interactive user power consumption abnormity monitoring and early warning management cloud platform based on cloud computing | |
| CN112785456A (en) | High-loss line electricity stealing detection method based on vector autoregressive model | |
| CN114004296A (en) | Method and system for reversely extracting monitoring points based on power load characteristics | |
| CN111275295B (en) | Distributed photovoltaic fault diagnosis method based on inverse distance weight interpolation | |
| CN114065605A (en) | Intelligent electric energy meter running state detection and evaluation system and method | |
| CN111949939A (en) | Intelligent electric meter running state evaluation method based on improved TOPSIS and cluster analysis | |
| CN114862139A (en) | Data-driven transformer area line loss rate abnormity diagnosis method | |
| CN111612371A (en) | Intelligent ammeter quality evaluation method based on analytic hierarchy process | |
| CN111310854A (en) | Low false alarm rate electricity stealing detection secondary screening method based on electricity load fingerprint | |
| CN110852906B (en) | Method and system for identifying electricity stealing suspicion based on high-dimensional random matrix | |
| CN114878934A (en) | Electric energy consumption data abnormity early warning method | |
| CN115293257A (en) | Detection method and system for abnormal electricity utilization user | |
| CN112417627A (en) | Power distribution network operation reliability analysis method based on four-dimensional index system | |
| CN117851908B (en) | Improved on-line low-voltage transformer area electric energy meter misalignment monitoring method and device | |
| CN111612019A (en) | Method for identifying and analyzing fault abnormality of intelligent electric meter based on big data model | |
| CN106709623B (en) | Power grid marketing inspection risk control method based on risk calculation model |
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 |