Disclosure of Invention
The invention aims to provide a power distribution network data anomaly tracing method, which is used for carrying out anomaly terminal positioning through a line loss withdrawal method, carrying out additional identification on an anomaly terminal by combining a coefficient determined based on a line loss regression model and a pearson correlation coefficient index until the negative loss of a power distribution network is eliminated, solving the application defect that the user-variable relation of the terminal and the line-variable relation of the head end cannot be accurately judged in power distribution network anomaly monitoring, further the abnormal power grid data cannot be rapidly identified and reliably traced and analyzed, realizing efficient and accurate tracing of the power distribution network data, facilitating the quality management of the power distribution network data, effectively guaranteeing the construction of a novel intelligent power system and promoting the further development of the power industry and the sustainable prosperity of society.
In order to achieve the above objective, it is necessary to provide a method, a system, a computer device and a storage medium for tracing abnormality of power distribution network data according to the above technical problems.
In a first aspect, an embodiment of the present invention provides a method for tracing anomalies in power distribution network data, where the method includes the following steps:
acquiring corresponding power distribution network equipment according to the power distribution type of the target abnormal area; the power distribution network equipment comprises head-end equipment and a corresponding terminal equipment set;
Acquiring historical power data of the power distribution network equipment, obtaining corresponding power distribution network line loss data according to the historical power data, and carrying out regression analysis on the historical power data and the power distribution network line loss data to obtain a power distribution network line loss relation model and a model decision coefficient; the historical power data comprise head-end equipment power data and corresponding terminal equipment power data;
taking terminal equipment with regression coefficients smaller than a preset abnormal coefficient threshold value in the distribution network line loss relation model as an abnormal terminal to be confirmed, and obtaining an abnormal terminal set to be confirmed;
removing power data of terminal equipment corresponding to each abnormal terminal to be confirmed from the historical power data independently in sequence, carrying out regression analysis again to obtain a corresponding first updated model decision coefficient, and screening from the set of abnormal terminals to be confirmed to obtain a first withdrawn abnormal terminal according to the first updated model decision coefficient and the model decision coefficient;
updating the terminal equipment set, the line loss data and the historical power data of the power distribution network according to the first withdrawn abnormal terminal, and carrying out next-round abnormal terminal screening until a preset abnormal tracing stop condition is reached, so as to obtain an abnormal terminal set; and the preset abnormal tracing stop condition is that the line loss of the power distribution network is eliminated or an abnormal terminal to be confirmed cannot be obtained.
Further, the step of obtaining the corresponding power distribution network device according to the power distribution type of the target abnormal area includes:
when the distribution type of the target abnormal area is a low-voltage distribution network, using a distribution transformer as head-end equipment and using each low-voltage user as corresponding terminal equipment;
when the distribution type of the target abnormal area is a medium-voltage distribution network, a large feeder line is used as head-end equipment, and each distribution transformer is used as corresponding terminal equipment.
Further, the step of obtaining corresponding power distribution network line loss data according to the historical power data, and performing regression analysis on the historical power data and the power distribution network line loss data to obtain a power distribution network line loss relation model includes:
accumulating the power data of all the terminal equipment to obtain total power data of the terminal;
the head-end equipment power data and the terminal total power data are subjected to difference to obtain the power distribution network line loss data;
and taking the power distribution network line loss data and the power data of each terminal device as dependent variables and independent variables respectively to carry out multiple linear regression analysis, so as to obtain the power distribution network line loss relation model.
Further, the step of screening the first withdrawn abnormal terminal from the to-be-confirmed abnormal terminal set according to the first updated model decision coefficient and the model decision coefficient includes:
Judging whether each first updating model decision coefficient is smaller than the model decision coefficient, if so, making a difference between the model decision coefficient and the corresponding first updating model decision coefficient to obtain a corresponding first index descending amplitude;
and taking the abnormal terminal to be confirmed corresponding to the maximum value in the first index falling amplitude which is larger than the preset falling threshold as the first withdrawing abnormal terminal.
Further, the step of updating the terminal device set, the power distribution network line loss data and the historical power data according to the first abnormal terminal withdrawal, and performing the next abnormal terminal screening includes:
removing the first withdrawal anomaly terminal from the terminal device set;
adding the terminal equipment power data of the first withdrawal abnormal terminal to the power distribution network line loss data to obtain first line loss update data;
removing terminal equipment power data corresponding to the first withdrawing abnormal terminal from the historical power data to obtain first power update data;
and carrying out regression analysis according to the first line loss updating data and the first power updating data, and carrying out next round of abnormal terminal screening according to the obtained first abnormal exclusion line loss relation model and the first abnormal exclusion model decision coefficient.
Further, when the line loss of the power distribution network is not eliminated and the abnormal terminal to be confirmed cannot be obtained, the step of obtaining the abnormal terminal set after the preset abnormal tracing stop condition is reached includes:
taking the terminal equipment which currently meets the preset abnormality detection condition in the terminal equipment set as an abnormality terminal to be detected to obtain an abnormality terminal set to be detected; the preset abnormality detection condition is that the power data of the terminal equipment is larger than a preset power threshold value and the regression coefficient corresponding to the first abnormality elimination line loss relation model is a negative value;
and carrying out abnormality screening on each abnormal terminal to be detected in the abnormal terminal set to be detected by combining line loss and power correlation to obtain a second withdrawing abnormal terminal, and updating the terminal equipment set according to the second withdrawing abnormal terminal until the line loss of the power distribution network is eliminated to obtain the abnormal terminal set.
Further, the step of performing abnormality screening of combining line loss and power correlation on each abnormal terminal to be detected in the abnormal terminal set to be detected to obtain a second withdrawn abnormal terminal, and updating the terminal equipment set according to the second withdrawn abnormal terminal includes:
carrying out pearson correlation calculation on the sum of the terminal equipment power data of all the power distribution terminals in the terminal equipment set and the head-end equipment power data to obtain a first correlation coefficient;
Removing each abnormal terminal to be detected from the first abnormal exclusion line loss relation model, and performing regression analysis and Pearson correlation calculation to obtain a corresponding second updating model decision coefficient and a second correlation coefficient;
judging whether each second updating model decision coefficient is smaller than the first updating model decision coefficient, if so, making a difference between the first updating model decision coefficient and the corresponding second updating model decision coefficient to obtain a corresponding second index descent amplitude;
if the second index decreasing amplitude is larger than a preset decreasing threshold value and the second correlation coefficient is larger than the first correlation coefficient, judging the abnormal terminal to be detected as a second withdrawing abnormal terminal;
removing the second abnormal withdrawal terminal from the terminal equipment set, adding the terminal equipment power data of the second abnormal withdrawal terminal to the first line loss updating data to obtain second line loss updating data, and removing the terminal equipment power data corresponding to the second abnormal withdrawal terminal from the first power updating data to obtain second power updating data;
and carrying out regression analysis according to the second line loss updating data and the second power updating data, and carrying out the screening of the next round of second withdrawal abnormal terminals according to the obtained second abnormality rejection line loss relation model and the second abnormality rejection model decision coefficient, updating the terminal equipment set until the line loss of the power distribution network is eliminated, so as to obtain the abnormal terminal set.
In a second aspect, an embodiment of the present invention provides a power distribution network data anomaly tracing system, where the system includes:
the equipment acquisition module is used for acquiring corresponding power distribution network equipment according to the power distribution type of the target abnormal area; the power distribution network equipment comprises head-end equipment and a corresponding terminal equipment set;
the line loss analysis module is used for acquiring historical power data of the power distribution network equipment, obtaining corresponding power distribution network line loss data according to the historical power data, and carrying out regression analysis on the historical power data and the power distribution network line loss data to obtain a power distribution network line loss relation model and a model decision coefficient; the historical power data comprise head-end equipment power data and corresponding terminal equipment power data;
the abnormal initial selection module is used for taking terminal equipment with regression coefficients smaller than a preset abnormal coefficient threshold value in the distribution network line loss relation model as an abnormal terminal to be confirmed to obtain an abnormal terminal set to be confirmed;
the abnormality confirmation module is used for sequentially and independently removing the power data of the terminal equipment corresponding to each abnormal terminal to be confirmed from the historical power data, carrying out regression analysis again to obtain a corresponding first updated model decision coefficient, and screening the abnormal terminals to be confirmed in a centralized manner to obtain a first withdrawn abnormal terminal according to the first updated model decision coefficient and the model decision coefficient;
And the traversing backtracking module is used for updating the terminal equipment set, the power distribution network line loss data and the historical power data according to the first withdrawn abnormal terminal, and carrying out the next round of abnormal terminal screening until the preset abnormal backtracking stopping condition is reached, so as to obtain an abnormal terminal set.
In a third aspect, embodiments of the present invention further provide a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the above method when executing the computer program.
In a fourth aspect, embodiments of the present invention also provide a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the above method.
The method comprises the steps of obtaining power distribution network equipment comprising head-end equipment and corresponding terminal equipment sets according to power distribution types of a target abnormal area, obtaining corresponding power distribution network line loss data according to historical power data of the obtained power distribution network equipment, carrying out regression analysis on the historical power data and the power distribution network line loss data to obtain a power distribution network line loss relation model and model decision coefficients, taking terminal equipment with regression coefficients smaller than a preset abnormal coefficient threshold value in the power distribution network line loss relation model as abnormal terminals to be confirmed to obtain an abnormal terminal set to be confirmed, sequentially removing power data of the terminal equipment corresponding to each abnormal terminal to be confirmed from the historical power data, carrying out regression analysis again to obtain corresponding first update model decision coefficients, obtaining first abnormal terminals from the power distribution network line loss to be confirmed according to the first update model decision coefficients and the model decision coefficients, updating the terminal equipment sets, the power distribution network line loss data and the historical power data, and carrying out next round of terminal screening until the abnormal terminals to be confirmed are withdrawn from the power distribution network line loss to be confirmed or the abnormal terminals to be removed until the abnormal terminals to be confirmed cannot be removed. Compared with the prior art, the power distribution network data anomaly tracing method performs anomaly terminal positioning through a line loss withdrawal method, performs anomaly terminal additional identification by combining the coefficient determination based on the line loss regression model and the pearson correlation coefficient index, realizes efficient and accurate tracing of the power distribution network data anomaly, facilitates distribution network data quality management, effectively ensures the stability of power supply and user satisfaction, and is beneficial to realizing the construction of a novel intelligent power system and promoting the further development of the power industry and the sustainable prosperity of society.
Detailed Description
For the purpose of making the objects, technical solutions and advantageous effects of the present application more apparent, the present invention will be further described in detail with reference to the accompanying drawings and examples, and it should be understood that the examples described below are only illustrative of the present invention and are not intended to limit the scope of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The method for tracing the abnormal power distribution network data provided by the invention can be understood as a method for tracing the abnormal power distribution network data caused by the problems of user variable relation hanging errors, line variable relation hanging errors, meter metering errors, user electricity consumption errors and the like by carrying out abnormal terminal positioning through a line loss withdrawal method and carrying out additional identification on the abnormal terminal by combining with the coefficient and the Pearson correlation coefficient index determined based on the line loss regression model until the negative loss of the power distribution network is eliminated, finding and positioning all the power distribution network terminals causing the abnormal power distribution network line loss; the method can be applied to a terminal or a server as shown in fig. 1. The terminal may be, but not limited to, various personal computers, notebook computers, smartphones, tablet computers and portable wearable devices, and the server may be implemented by a separate server or a server cluster formed by a plurality of servers. The server can perform efficient and accurate power distribution network data anomaly tracing by adopting the power distribution network data anomaly tracing method architecture shown in the figure 2 according to actual application requirements, and the obtained anomaly terminal set is used for server follow-up study or transmitted to a terminal for a terminal user to check and analyze; the following embodiment will explain the power distribution network data anomaly tracing method in detail.
In one embodiment, as shown in fig. 3, a method for tracing abnormality of power distribution network data is provided, which includes the following steps:
s11, acquiring corresponding power distribution network equipment according to the power distribution type of the target abnormal area; the power distribution network equipment comprises head-end equipment and a corresponding terminal equipment set; the target abnormal area can be understood as an abnormal power distribution area which is obtained based on historical solar line loss screening and needs to identify an abnormal terminal, the corresponding power distribution type can be divided into low-voltage power distribution and medium-voltage power distribution according to the difference of actual application scenes, and the corresponding power distribution equipment which needs to be researched can be also distinguished; specifically, the step of obtaining the corresponding power distribution network device according to the power distribution type of the target abnormal area includes:
when the distribution type of the target abnormal area is a low-voltage distribution network, using a distribution transformer as head-end equipment and using each low-voltage user as corresponding terminal equipment;
when the distribution type of the target abnormal area is a medium-voltage distribution network, taking a large feeder line as head-end equipment and taking each distribution transformer as corresponding terminal equipment; the voltage of the large feeder line can be determined according to an actual application scenario, for example, a 10kv distribution large feeder line can be used as head-end equipment, and the method is not particularly limited herein.
S12, acquiring historical power data of the power distribution network equipment, obtaining corresponding power distribution network line loss data according to the historical power data, and carrying out regression analysis on the historical power data and the power distribution network line loss data to obtain a power distribution network line loss relation model and a model decision coefficient; the historical power data can be understood as historical daily power data of each power distribution network device, which is acquired according to a certain acquisition frequency (such as 15 minutes), and comprises head-end device power data and corresponding power data of a plurality of terminal devices;
specifically, the step of obtaining corresponding power distribution network line loss data according to the historical power data, and performing regression analysis on the historical power data and the power distribution network line loss data to obtain a power distribution network line loss relation model includes:
accumulating the power data of all the terminal equipment to obtain total power data of the terminal; the actual acquisition mode of the total power data of the terminals is also different due to the distribution type: for a low-voltage distribution network, considering that the line loss is mainly composed of heat loss generated by current flowing through a distribution wire, the contribution coefficient of each terminal to the line loss is smaller, so that the total power data of the terminals is direct accumulation of the measured power of a low-voltage user meter; for a medium-voltage distribution network, the transformer loss occupies most of the power loss from the feeder head end to the secondary side of the transformer, the heat loss of a line is relatively small, the power of the secondary side of the transformer is required to be converted to the primary side, namely, the iron loss (no-load loss) and copper loss (load loss) of the transformer are added on the basis of the measured power of the secondary side of the original transformer to be recorded as converted power of a terminal of the transformer, and the power of each transformer terminal is accumulated to obtain corresponding total power data of the terminal;
The head-end equipment power data and the terminal total power data are subjected to difference to obtain the power distribution network line loss data; the specific calculation formulas of the power distribution network line loss data also have all different distribution types:
1) The line loss calculation of the low-voltage distribution network is as shown in fig. 4, and is as follows:
wherein,P loss,t the line loss of the power distribution network at the time t of the low-voltage power distribution network is calculated;P s,t for the power number of head-end equipment at t moment of power distribution networkAccording to (distribution transformer power);p i,t measuring data of the power of the ith terminal equipment (power data of a low-voltage user) in the power distribution network at the moment t; n represents the number of terminal devices connected in the power distribution network area;
2) The line loss calculation of the medium-voltage distribution network is as shown in fig. 5-6, and is as follows:
in the method, in the process of the invention,
wherein,P loss,t the line loss of the power distribution network at the moment t of the medium-voltage power distribution network is calculated;P s,t the power data (large feeder line distribution power) of head-end equipment at the moment t of the distribution network is obtained;the power data of the j-th terminal equipment in the power distribution network at the moment t;p i,t measuring data of the power of the ith terminal equipment (measuring data of the power of a distribution transformer) in the power distribution network at the moment t;P i0,t andP ic,t the no-load loss and the load loss of a j-th distribution transformer in the power distribution network at the moment t are respectively represented;
the no-load loss is understood to be a fixed value generated by the core; load loss is understood to be the loss produced by the passage of load current through the transformer windings, the magnitude of which varies with load current, in proportion to the square of load current; firstly, the power of the secondary side of the distribution transformer is folded to the primary side as shown in fig. 5, namely, the model number, the no-load loss, the impedance voltage percentage and the no-load current percentage of the transformer are obtained from a transformer asset account to calculate the impedance of the transformer, after the impedance power loss of the transformer winding is calculated by the power of the secondary side and the impedance value, the power of the primary side is converted into the sum of the power of the secondary side, the reactive power loss and the no-load loss, and the line loss of the medium-voltage distribution network is calculated based on the mode shown in fig. 6 after the power loss of the transformer is considered; the impedance calculation formula of the transformer is as follows:
In the method, in the process of the invention,for transformer resistance>For transformer reactance>For transformer short-circuit loss->For short-circuit voltage>Is the rated capacity of the system.
Taking the power distribution network line loss data and the power data of each terminal device as dependent variables and independent variables respectively to carry out multiple linear regression analysis to obtain the power distribution network line loss relation model; the power distribution network line loss relation model can be understood as a multiple linear regression model which is used for obtaining a group of optimal weights based on least square method to enable error square sum minimum estimation to represent the relation between terminal power data and power distribution network line loss data, and is expressed as follows:
wherein,a i,t the linear regression coefficient of the ith terminal equipment in the power distribution network at the moment t can be also understood as the influence degree of each terminal on the line loss, namely the contribution coefficient; under normal conditions, the line loss of the low-voltage distribution network is mainlyThe power-saving type power supply circuit consists of heat loss generated by current flowing through a wire and is positively correlated with the power of each node terminal; since the wire resistance is relatively low (typically below 0.1 ohm), the contribution coefficient of each terminal to the wire loss is small; the line loss of the medium-voltage distribution network is mainly caused by the change loss caused by a transformer in operation, and after the copper loss and the iron loss of the transformer are superposed at the transformer terminals, the contribution coefficient of each terminal to the line loss is smaller; if the problems of user change relation hanging errors, line change relation hanging errors, meter measurement errors, user electricity consumption errors and the like exist, so that the data of the power distribution network terminal are abnormal, the abnormal terminal electric quantity is directly overlapped on the whole line loss in a negative loss mode, the theoretical coefficient is-1, the coefficient can fluctuate around-1 according to the data acquisition error in actual regression, but the absolute value is far higher than the contribution coefficient of a normal terminal to the line loss; therefore, the terminal with possible abnormality can be marked by whether the regression coefficient of the terminal power accords with the preset abnormality coefficient threshold value, namely, the regression coefficient of the normal terminal is close to 0, and the regression coefficient of the abnormal terminal is close to-1.
S13, taking terminal equipment with regression coefficients smaller than a preset abnormal coefficient threshold value in the distribution network line loss relation model as an abnormal terminal to be confirmed, and obtaining an abnormal terminal set to be confirmed; the regression coefficient is verified by a large amount of data, the absolute value of the regression coefficient of the normal terminal device fluctuates between 0 and 0.4, and in consideration of the influence of the actual measurement acquisition error on the model judgment accuracy, some elastic margins are reserved during calculation, and the preset abnormal coefficient threshold value is preferably set to-0.8, namely the terminal device smaller than the threshold value is likely to be an abnormal terminal; the number of the to-be-confirmed abnormal terminals in the corresponding obtained to-be-confirmed abnormal terminal set can be one or two or more, and the to-be-confirmed abnormal terminals can be determined by the value of the regression coefficient in the actually obtained power distribution network line loss relation model, which is not described in detail herein;
in a linear regression model consisting of a plurality of terminal powers and power distribution network line losses, a model decision coefficient (r 2score index) represents the interpretation capability of the model on observed data variances, namely the fitting degree of the model on actual data and the accuracy degree of relation between interpretation variables, wherein the value range of r2score is between 0 and 1, and the closer to 1, the stronger the interpretation capability of the model on data is represented, and the closer to 0, the weaker the interpretation capability of the model is represented; the calculation formula of the r2score index is:
The RMSE and the Var are the mean square error and the variance of the linear regression power loss of the terminal power distribution network; in general, the closer the r2score index is to 1, the better the model fits the data;
considering that the abnormal terminals are the main sources of the negative loss of the power distribution network, the correlation with the line loss is the highest, the situation that the regression effect is deteriorated is theoretically caused after the abnormal terminals are withdrawn, after the abnormal terminal set to be confirmed is obtained, the following line loss withdrawal method is adopted to carry out regression analysis after removing each abnormal terminal to be confirmed, the abnormal terminals are positioned under the condition of the descending amplitude of the corresponding model determination coefficient (r 2score index), and the abnormal terminal with the highest possibility is screened out.
S14, removing power data of terminal equipment corresponding to each abnormal terminal to be confirmed from the historical power data independently in sequence, carrying out regression analysis again to obtain a corresponding first updated model decision coefficient, and screening the abnormal terminals to be confirmed in a concentrated mode to obtain a first withdrawn abnormal terminal according to the first updated model decision coefficient and the model decision coefficient;
the first abnormal terminal is understood to be an abnormal terminal with the highest possibility of abnormality from all abnormal terminals to be confirmed through a line loss withdrawal method, the corresponding acquisition process is to remove the abnormal terminal to be confirmed from line loss regression analysis, and whether a decision coefficient corresponding to a new line loss relation model obtained through analysis is changed greatly relative to a model decision coefficient obtained through previous modeling or not is judged to further judge whether the abnormal terminal to be confirmed is the abnormal terminal or not; if the to-be-confirmed abnormal terminal is removed from the line loss regression analysis, the obtained first updated model determines that the coefficient index has no obvious change, and if the index is obviously reduced and exceeds a threshold value, the terminal is judged to be abnormal, and the abnormal terminal with the largest problem to be directly removed from the terminal equipment in a centralized manner is found from all the terminal equipment with the abnormality. Finding out an abnormal terminal with the maximum index decreasing amplitude;
Specifically, the step of intensively screening the abnormal terminals to be confirmed to obtain the first withdrawn abnormal terminal according to the first updated model decision coefficient and the model decision coefficient includes:
judging whether each first updating model decision coefficient is smaller than the model decision coefficient, if so, making a difference between the model decision coefficient and the corresponding first updating model decision coefficient to obtain a corresponding first index descending amplitude;
taking an abnormal terminal to be confirmed corresponding to the maximum value in the first index falling amplitude which is larger than a preset falling threshold value as the first withdrawing abnormal terminal; the preset drop threshold may be selected according to practical application requirements in principle, and in this embodiment, preferably, the preset drop threshold is set to 0.085 based on a large amount of data verification, that is, when the r2score drop amplitude obtained by the regression analysis of the to-be-confirmed abnormal terminal is greater than the threshold 0.085, the abnormal terminal may be abnormal, and one with the largest drop amplitude value is selected from all the abnormal terminals with the index drop amplitude greater than the threshold as the first withdrawal abnormal terminal.
After the first abnormal terminal is withdrawn after the line loss regression calculation is determined, withdrawing the first abnormal terminal from the power distribution network, and correspondingly adding the power value of the terminal to the line loss, and carrying out the next line loss regression calculation according to the following method until all the abnormal terminals are found out.
S15, updating the terminal equipment set, the power distribution network line loss data and the historical power data according to the first withdrawn abnormal terminal, and carrying out next-round abnormal terminal screening until a preset abnormal tracing stop condition is reached, so as to obtain an abnormal terminal set; the preset abnormal tracing stop condition is that the line loss of the power distribution network is eliminated or an abnormal terminal to be confirmed cannot be obtained, wherein the elimination of the line loss of the power distribution network can be understood as that the line loss is larger than zero at all moments, the failure to obtain the abnormal terminal to be confirmed can be understood as that the negative loss is not eliminated, and the terminal meeting the regression coefficient abnormal judgment threshold cannot be found;
specifically, the step of updating the terminal device set, the line loss data and the historical power data of the distribution network according to the first abnormal terminal withdrawn, and performing the next abnormal terminal screening includes:
removing the first withdrawal anomaly terminal from the terminal device set;
adding the terminal equipment power data of the first withdrawal abnormal terminal to the power distribution network line loss data to obtain first line loss update data;
removing terminal equipment power data corresponding to the first withdrawing abnormal terminal from the historical power data to obtain first power update data;
And carrying out regression analysis according to the first line loss updating data and the first power updating data, and carrying out next round of abnormal terminal screening according to the obtained first abnormal exclusion line loss relation model and the first abnormal exclusion model decision coefficient.
The step of performing the next round of abnormal terminal screening according to the obtained first abnormal exclusion line loss relation model and the first abnormal exclusion model determining coefficient may refer to S11-S14 to obtain the process of the first abnormal terminal withdrawal, which is not described herein again;
according to the method, the abnormal terminal positioning method is carried out through the line loss withdrawal method, so that the user transformation relation of the terminal and the line transformation relation of the head end can be accurately judged, abnormal data of the power grid can be efficiently and accurately identified, and reliable traceability analysis can be carried out; however, considering that in practical application, if the negative loss is not eliminated, it indicates that there is a possibility that an unidentified abnormal terminal may exist, in order to further ensure the comprehensiveness and reliability of anomaly tracing, in this embodiment, preferably, an anomaly detection method based on combination of a line loss regression model decision coefficient and a pearson correlation coefficient index is used to sequentially perform verification and screening on the terminals with high power remaining in the power distribution network and negative regression coefficients; specifically, when the line loss of the power distribution network is not eliminated and the abnormal terminal to be confirmed cannot be obtained, the step of obtaining the abnormal terminal set after the preset abnormal tracing stop condition is reached includes:
Taking the terminal equipment which currently meets the preset abnormality detection condition in the terminal equipment set as an abnormality terminal to be detected to obtain an abnormality terminal set to be detected; the preset abnormality detection condition is that the power data of the terminal equipment is larger than a preset power threshold value and the regression coefficient corresponding to the first abnormality elimination line loss relation model is a negative value; the preset power threshold value can be set according to practical application requirements in principle, and the embodiment preferably sets the preset power threshold value as an average value of the power of the remaining terminal devices in the current terminal device set, namely, terminal devices with the current terminal device set power larger than the preset power threshold value and the corresponding regression coefficient with a negative value in the previous round of regression analysis are listed as an abnormal terminal set to be inspected, and whether the abnormality is caused is confirmed by combining the change condition of the determined coefficient obtained by regression analysis according to the line loss withdrawal method and the correlation change of the sum sequence of the power of the head-end device and the power of the terminal device before and after withdrawal;
performing abnormality screening of combination of line loss and power correlation on each abnormal terminal to be detected in the abnormal terminal set to obtain a second withdrawing abnormal terminal, and updating the terminal equipment set according to the second withdrawing abnormal terminal until the line loss of the power distribution network is eliminated to obtain the abnormal terminal set; specifically, the step of performing abnormality screening of combining line loss and power correlation on each abnormal terminal to be detected in the abnormal terminal set to be detected to obtain a second withdrawn abnormal terminal, and updating the terminal equipment set according to the second withdrawn abnormal terminal includes:
Carrying out pearson correlation calculation on the sum of the terminal equipment power data of all the power distribution terminals in the terminal equipment set and the head-end equipment power data to obtain a first correlation coefficient; wherein, the formula of the pearson correlation calculation is as follows:
in the method, in the process of the invention,correlation between terminal power summation and power at head end of power distribution networkThe number is used for judging the correlation between the sum of the power measurement values of the intelligent ammeter of the low-voltage user terminal and the power measurement of the transformer in the transformer area or between the sum of the power measurement values of the transformer terminal and the feeder gateway power measurement sequence; />And->Average value of power of terminal equipment and power of head-end equipment respectively,/->And->Respectively collecting effective data of the power of the terminal equipment and the head-end equipment; n is the number of effective data of power collection;
removing each abnormal terminal to be detected from the first abnormal exclusion line loss relation model, and performing regression analysis and Pearson correlation calculation to obtain a corresponding second updating model decision coefficient and a second correlation coefficient;
judging whether each second updating model decision coefficient is smaller than the first updating model decision coefficient, if so, making a difference between the first updating model decision coefficient and the corresponding second updating model decision coefficient to obtain a corresponding second index descent amplitude;
If the second index decreasing amplitude is larger than a preset decreasing threshold value and the second correlation coefficient is larger than the first correlation coefficient, judging the abnormal terminal to be detected as a second withdrawing abnormal terminal;
removing the second abnormal withdrawal terminal from the terminal equipment set, adding the terminal equipment power data of the second abnormal withdrawal terminal to the first line loss updating data to obtain second line loss updating data, and removing the terminal equipment power data corresponding to the second abnormal withdrawal terminal from the first power updating data to obtain second power updating data;
performing regression analysis according to the second line loss updating data and the second power updating data, and performing next round of second abnormal terminal withdrawal screening according to the obtained second abnormal rejection line loss relation model and the second abnormal rejection model determining coefficient, updating the terminal equipment set until the line loss of the power distribution network is eliminated, and obtaining the abnormal terminal set;
the actual process of determining the second withdrawal anomaly terminal can be understood as follows: withdrawing the high-power terminals from the power distribution network in sequence, correspondingly adding terminal power to the line loss, and recording the change of r2score indexes each time; calculating the pearson correlation of the sum of the low-voltage user terminal power and the distribution transformer power in the low-voltage distribution network scene before and after the terminal is withdrawn, and the pearson correlation of the sum of the distribution transformer terminal power and the feeder power in the medium-voltage distribution network scene; if the index is obviously reduced and the power correlation is improved after a certain terminal in one round is withdrawn, determining that the terminal is abnormal, withdrawing the terminal from the power distribution network and the line loss, intensively picking out the terminal from the abnormal terminal to be detected, expanding the next round of calculation until the negative loss is eliminated or the abnormal terminal meeting the criterion cannot be found, integrating the withdrawn abnormal terminals in each round to obtain a finally determined abnormal terminal set, and further realizing tracing of the abnormal data of the power distribution network.
Through the steps, the power distribution network terminal which causes abnormal line loss of the power distribution network can be found and positioned, and abnormal tracing of power distribution network data caused by the problems of user variable relation hanging errors, line variable relation hanging errors, meter metering anomalies, user power consumption anomalies and the like is realized; it should be noted that, the scheme has the function of large-scale scanning calculation of the data of the power distribution network or the low-voltage power distribution station area, and can also calculate the area with negative line loss preferentially, namely, calculate the daily line loss screening negative loss area by adopting the historical data, find out the abnormal terminal with the greatest influence on the line loss index, and calculate the line loss, namely, the real-time line loss calculation method above.
According to the embodiment of the application, the power distribution network equipment comprising the head-end equipment and the corresponding terminal equipment sets is obtained according to the power distribution type of the target abnormal area, the corresponding power distribution network line loss data is obtained according to the obtained historical power data of the power distribution network equipment, after regression analysis is carried out on the historical power data and the power distribution network line loss data to obtain a power distribution network line loss relation model and a model decision coefficient, the terminal equipment with the regression coefficient smaller than the preset abnormal coefficient threshold value in the power distribution network line loss relation model is used as the abnormal terminal to be confirmed to obtain the abnormal terminal set to be confirmed, the power data of the terminal equipment corresponding to each abnormal terminal to be confirmed are removed from the historical power data independently, regression analysis is carried out again to obtain the corresponding first updated model decision coefficient, and the model decision coefficient are decided according to the first updated model, the first withdrawal abnormal terminal is obtained by centralized screening of the abnormal terminals to be confirmed, the terminal equipment set, the power distribution network line loss data and the historical power data are updated according to the first withdrawal abnormal terminal, the next round of abnormal terminal screening is carried out until the power distribution network line loss is eliminated or the abnormal terminals to be confirmed cannot be obtained, then the abnormal terminals of the high-power terminal equipment are additionally identified based on the line loss regression model decision coefficient and the pearson correlation coefficient index until the negative losses are completely eliminated, the scheme of the abnormal terminal set is obtained, the problem that the user-to-user relationship and the line-to-line relationship of the head end of the terminals cannot be accurately judged in the existing power distribution network abnormal monitoring is effectively solved, the application defects that the power grid abnormal data cannot be rapidly identified and the reliable tracing analysis is carried out can be realized, the efficient, accurate and comprehensive tracing of the power distribution network data abnormality can be realized, the quality management of the power distribution network data is facilitated, the intelligent power system has the advantages that the stability of power supply and the satisfaction of users are effectively guaranteed, meanwhile, the construction of a novel intelligent power system is facilitated, and the further development of the power industry and the sustainable prosperity of society are promoted.
In order to supplement the implementation effect of the method of the invention, the embodiment also combines the historical measurement data in the low-voltage distribution network and the medium-voltage distribution network examples to carry out simulation verification:
1) And (3) verifying low-voltage distribution network examples:
the practical power load historical data of the station area of seven stations of travel water is experimentally verified, the station area comprises 79 terminals, data of 96 time points (24 hours) are collected every day with 15 minutes as intervals, and the power value collected by a distribution transformer at the time t is recorded asThe power value collected by the low-voltage user terminal is +.>The method comprises the steps of carrying out a first treatment on the surface of the After data preprocessing, a multiple linear regression equation is established by carrying out regression analysis on terminal power and station line loss data:
wherein,is a low-voltage user terminal time sequence power matrix at t moment, < >>Is a line loss time sequence matrix of the power distribution network at the moment t>Is a feature matrix, i.e. a matrix of continuous unknown variables to be solved, < >>Is the residual vector at time t;
obtaining regression coefficients and r2score indexes of each terminal according to FIG. 7, wherein the index size is 0.536, and 11 terminals with regression coefficients lower than a set threshold are provided; calculating the r2score descending amplitude of each possible abnormal terminal through line loss regression, finding the abnormal terminal with the largest r2score descending, withdrawing the terminal, and changing the regression coefficient into 0.436 as shown in figure 8; removing the abnormal terminal from the station area, and adding the power to the corresponding line loss, wherein the regression coefficient is shown in a graph of FIG. 9;
Since the negative loss is not eliminated yet and the terminal meeting the regression coefficient abnormality judgment threshold cannot be found, sequentially checking the terminals which are left in the platform area and have high power and negative regression coefficients, wherein the regression coefficients are shown in a graph of FIG. 10;
until the negative loss is eliminated, the abnormal terminal detection is ended, the regression coefficient is shown in figure 11, the regression coefficient in the figure is a user exceeding-0.8, the influence value of r2score index change after the user is withdrawn is only 0.0024 according to the flow, and the magnitude of the influence value is smaller than the threshold value of the abnormal user on the r2score change amplitude set by the method, so that the normal terminal is judged;
and integrating the abnormal terminals withdrawn from each round, and realizing the tracing of the data abnormality of the low-voltage distribution transformer area, wherein the power change of the terminals before and after the abnormal terminals are withdrawn is shown as figure 12.
2) In medium voltage distribution network instance verification:
the practical power load historical data of the distribution transformer in the spring D238 line and the coverage area thereof are experimentally verified, the feeder line power supply area comprises 76 distribution transformer terminals, data of 96 time points (24 hours) are collected every day at 15 minutes intervals, and the power value collected by the spring D238 line at the time t is recorded asThe power value collected by the distribution transformer terminal hung below the power converter terminal is +.>;
First, the model of the transformer asset ledger is obtained and the impedance loss and the excitation loss are calculated according to the model of the transformer asset ledger, as shown in the partial model of table 1:
Table 1 transformer model, impedance loss and excitation loss examples
Secondly, calculating primary side power of the distribution transformer according to transformer loss, calculating feeder line loss through primary side converted power of a distribution transformer terminal, and then establishing a multiple linear regression model of feeder line loss-distribution transformer terminal power:
obtaining regression coefficients and r2score indexes of each distribution transformer terminal as shown in FIG. 13; finding an abnormal terminal with the largest r2score drop through line loss regression calculation, and withdrawing the terminal to obtain regression coefficients as shown in figure 14;
and integrating the abnormal terminals withdrawn from each round, and realizing the tracing of the data abnormality of the medium-voltage distribution transformer area, wherein the power change of the terminals before and after the abnormal terminals are withdrawn is shown as a figure 15.
On-site verification according to the model calculation results can be found: the accuracy of the line loss withdrawal anomaly terminal positioning method described in the power distribution network data anomaly tracing method is 98.24%. The main possible reasons for the conventional voltage-related method to reduce the accuracy to 83.42% are as follows: if the transformer areas belong to the same feeder line, the voltage fluctuation of each user terminal is similar; secondly, the voltage of the terminal is distorted due to the fact that a power supply line is long in rural areas of mountain areas, waveforms deviate from other terminals, and the fact that abnormal user terminals are difficult to identify by means of voltage criteria is limited greatly; the second is the conservation of power which is based on the principle of seeking conservation of energy between the distribution transformer and the low voltage subscribers of the bay, i.e. the sum of the subscriber power under the bay and the sum of the subscriber power about equal to the transformer power. The final accuracy is only 85.6%, which is because the model fits the power relation between the transformer and the user as much as possible, so that erroneous judgment is formed for part of low-power user terminals.
Table 2 comparison of the accuracy of the algorithm proposed by the present invention with the conventional algorithm
In summary, the invention provides the abnormal tracing method for the power distribution network data, which establishes the abnormal terminal positioning method model based on r2score line loss withdrawal aiming at mathematical characteristics of the power distribution network line loss and terminal power measurement data on the basis of the research based on a multiple linear regression model and the traditional power distribution network data abnormal tracing, has higher accuracy, effectively improves the real-time performance and flexibility of topology record file updating, simultaneously can adapt to the continuously-changed topology of a medium-low power distribution network, and can obtain good results in a novel power distribution network.
Although the steps in the flowcharts described above are shown in order as indicated by arrows, these steps are not necessarily executed in order as indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders.
In one embodiment, as shown in fig. 16, there is provided a power distribution network data anomaly tracing system, the system including:
the equipment acquisition module 1 is used for acquiring corresponding power distribution network equipment according to the power distribution type of the target abnormal area; the power distribution network equipment comprises head-end equipment and a corresponding terminal equipment set;
The line loss analysis module 2 is used for acquiring historical power data of the power distribution network equipment, obtaining corresponding power distribution network line loss data according to the historical power data, and carrying out regression analysis on the historical power data and the power distribution network line loss data to obtain a power distribution network line loss relation model and a model decision coefficient; the historical power data comprise head-end equipment power data and corresponding terminal equipment power data;
the abnormal initial selection module 3 is used for taking the terminal equipment with the regression coefficient smaller than a preset abnormal coefficient threshold value in the distribution network line loss relation model as an abnormal terminal to be confirmed to obtain an abnormal terminal set to be confirmed;
the anomaly confirmation module 4 is configured to sequentially remove power data of terminal devices corresponding to each anomaly terminal to be confirmed from the historical power data separately, re-perform regression analysis to obtain a corresponding first updated model decision coefficient, and screen out the anomaly terminal to be confirmed from the anomaly terminal set to obtain a first withdraw according to the first updated model decision coefficient and the model decision coefficient;
and the traversal backtracking module 5 is used for updating the terminal equipment set, the power distribution network line loss data and the historical power data according to the first withdrawal abnormal terminal, and carrying out the next round of abnormal terminal screening until the preset abnormal backtracking stopping condition is reached, so as to obtain an abnormal terminal set.
The specific limitation of the power distribution network data anomaly tracing system can be referred to the limitation of the power distribution network data anomaly tracing method, and the corresponding technical effects can be obtained equally, and are not repeated here. All or part of each module in the power distribution network data anomaly tracing system can be realized by software, hardware and a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
Fig. 17 shows an internal structural diagram of a computer device, which may be a terminal or a server in particular, in one embodiment. As shown in fig. 17, the computer device includes a processor, a memory, a network interface, a display, a camera, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by the processor to realize the power distribution network data anomaly tracing method. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, can also be keys, a track ball or a touch pad arranged on the shell of the computer equipment, and can also be an external keyboard, a touch pad or a mouse and the like.
It will be appreciated by those of ordinary skill in the art that the structure shown in fig. 17 is merely a block diagram of some of the structures associated with the present application and is not limiting of the computer device to which the present application may be applied, and that a particular computing device may include more or fewer components than those shown, or may combine certain components, or have the same arrangement of components.
In one embodiment, a computer device is provided comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the above method when the computer program is executed.
In one embodiment, a computer readable storage medium is provided having a computer program stored thereon, which when executed by a processor, implements the steps of the above method.
In summary, the embodiment of the invention provides a power distribution network data anomaly tracing method and system, the power distribution network data anomaly tracing method achieves power distribution type according to a target anomaly region, power distribution network equipment comprising head-end equipment and corresponding terminal equipment sets is obtained, corresponding power distribution network line loss data is obtained according to the obtained historical power data of the power distribution network equipment, regression analysis is carried out on the historical power data and the power distribution network line loss data to obtain a power distribution network line loss relation model and model decision coefficients, terminal equipment with the regression coefficients smaller than a preset anomaly coefficient threshold value in the power distribution network line loss relation model is taken as an anomaly terminal to be confirmed, an anomaly terminal set to be confirmed is obtained, power data of terminal equipment corresponding to each anomaly terminal to be confirmed are sequentially removed from the historical power data, regression analysis is carried out again to obtain corresponding first update model decision coefficients, a first withdrawal anomaly terminal is obtained from the anomaly terminal set to be confirmed according to the first update model decision coefficients and the model decision coefficients, next round of terminal screening is carried out on the terminal sets to be confirmed, the anomaly terminal data and the historical power data is updated until the power distribution network line loss relation model is completely withdrawn, the anomaly terminal is completely lost based on the basis of the absolute value of the anomaly terminal to be confirmed, the anomaly terminal is completely removed by the method, and the anomaly terminal is completely lost due to the fact that the anomaly terminal is completely is removed by the relative loss coefficient is identified based on the absolute method, and the anomaly terminal is completely lost due to the fact that the anomaly terminal is removed by the relative loss coefficient is lost, and the relative station is lost, and the anomaly terminal is based on the method is completely, the anomaly terminal loss is lost, and the quality loss is can is lost, the method is convenient for distribution network data quality management, effectively ensures the stability of power supply and user satisfaction, is beneficial to realizing the construction of a novel intelligent power system, and promotes the further development of the power industry and the sustainable prosperity of society.
In this specification, each embodiment is described in a progressive manner, and all the embodiments are directly the same or similar parts referring to each other, and each embodiment mainly describes differences from other embodiments. In particular, for system embodiments, since they are substantially similar to method embodiments, the description is relatively simple, as relevant to see a section of the description of method embodiments. It should be noted that, any combination of the technical features of the foregoing embodiments may be used, and for brevity, all of the possible combinations of the technical features of the foregoing embodiments are not described, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The foregoing examples represent only a few preferred embodiments of the present application, which are described in more detail and are not thereby to be construed as limiting the scope of the invention. It should be noted that modifications and substitutions can be made by those skilled in the art without departing from the technical principles of the present invention, and such modifications and substitutions should also be considered to be within the scope of the present application. Therefore, the protection scope of the patent application is subject to the protection scope of the claims.