CN110956330A - Method and system for predicting line loss of power transmission line based on multidimensional influence quantity - Google Patents

Method and system for predicting line loss of power transmission line based on multidimensional influence quantity Download PDF

Info

Publication number
CN110956330A
CN110956330A CN201911213450.8A CN201911213450A CN110956330A CN 110956330 A CN110956330 A CN 110956330A CN 201911213450 A CN201911213450 A CN 201911213450A CN 110956330 A CN110956330 A CN 110956330A
Authority
CN
China
Prior art keywords
data
tree model
training
transmission line
gradient lifting
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
Application number
CN201911213450.8A
Other languages
Chinese (zh)
Inventor
余佶成
岳长喜
朱凯
李鹤
李登云
熊魁
周峰
李智成
田爽
余也凤
胡浩亮
李小飞
黄俊昌
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Electric Power Research Institute Co Ltd CEPRI
Original Assignee
China Electric Power Research Institute Co Ltd CEPRI
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China Electric Power Research Institute Co Ltd CEPRI filed Critical China Electric Power Research Institute Co Ltd CEPRI
Priority to CN201911213450.8A priority Critical patent/CN110956330A/en
Publication of CN110956330A publication Critical patent/CN110956330A/en
Pending legal-status Critical Current

Links

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
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • 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
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/067—Enterprise or organisation modelling
    • 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Electricity, gas or water supply

Abstract

The invention discloses a method and a system for predicting transmission line loss based on multidimensional influence quantity, wherein the method comprises the following steps: collecting the multidimensional influence quantity of the line loss of the transmission line; performing data preprocessing on line body data in a power transmission line and non-time sequence data in tower data of each level of the line to enable the non-time sequence data to form fixed parameters of the power transmission line, merging the fixed parameters into a time sequence data table of the power transmission line, and generating a multi-dimensional influence quantity data table; dividing data in the multi-dimensional influence quantity data table into training set data, verification set data and test set data; training the gradient lifting tree model through training set data, and outputting a training result; verifying the training result through the verification set data, and generating a trained gradient lifting tree model when the training result passes the verification; and predicting the test set data through the trained gradient lifting tree model, and outputting a line loss prediction result of the power transmission line.

Description

Method and system for predicting line loss of power transmission line based on multidimensional influence quantity
Technical Field
The invention relates to the technical field of electric energy metering in electrical engineering, in particular to a method and a system for predicting line loss of a power transmission line based on multidimensional influence quantity.
Background
The line loss reflects the planning, production and management level of the power grid and is an important standard for assessing power departments. However, errors in theoretical line loss calculation can cause that a report cannot accurately reflect the actual line loss condition, and great obstacles are brought to line loss management. With the advance of fine management work of line loss, an accurate synchronous line loss calculation method is urgently needed.
Therefore, a technique is needed to realize a technique for predicting the line loss of the transmission line based on the multidimensional influence quantity.
Disclosure of Invention
The technical scheme of the invention provides a method and a system for predicting line loss of a power transmission line based on multidimensional influence quantity, which aim to solve the problem of low line loss prediction precision of the existing power transmission line.
In order to solve the above problem, the present invention provides a method for predicting line loss of a power transmission line based on a multidimensional influence quantity, the method comprising:
collecting the multidimensional influence quantity of the line loss of the transmission line;
performing data preprocessing on line body data in the power transmission line and non-time sequence data in tower data of each level of the line to enable the non-time sequence data to form fixed parameters of the power transmission line, and merging the fixed parameters into a time sequence data table of the power transmission line to generate a multi-dimensional influence quantity data table;
dividing data in the multi-dimensional influence quantity data table into training set data, verification set data and test set data; training the gradient lifting tree model through the training set data, and outputting a training result;
verifying the training result through verification set data, and generating a trained gradient lifting tree model when the training result passes the verification;
and predicting the test set data through the trained gradient lifting tree model, and outputting a line loss prediction result of the power transmission line.
Preferably, the training the gradient lifting tree model through the training set data and outputting a training result further includes:
and analyzing the weight of the multi-dimensional influence quantity through a gradient lifting tree model.
Preferably, the analyzing the weight of the multi-dimensional influence quantity through the gradient lifting tree model further includes:
training the gradient lifting tree model through the training set data to obtain a trained gradient lifting tree model;
extracting all split node data n of the trained gradient lifting tree model;
calculating the number n of split nodes generated by each dimension influence quantity in the k dimension influence quantityk；
According to the total scoreThe data n of the split nodes and the number nk of the split nodes generated by each dimension of influence quantity are calculated, and the weight Q of the k dimension of influence quantity is calculatedk，Qk＝nk/n×100％。
Preferably, the verifying the training result by the verification set data, and when the training result passes the verification, generating a trained gradient lifting tree model, further includes:
initializing the gradient lifting tree model, estimating a model parameter which minimizes a loss function, and establishing an initial gradient lifting tree model based on the model parameter;
within the predicted maximum number of iterations, steps S1, S2, S3, and S4 are iteratively performed in order:
s1, calculating a loss function and a negative gradient of the initial gradient lifting tree model;
s2, taking the negative gradient as new training set data, and fitting a regression tree model according to the new training set data;
s3, calculating each leaf node of the regression tree model to obtain a best fitting value;
s4, fusing the best fitting value to a regression tree model;
and when the maximum iteration times are reached, outputting the trained gradient lifting tree model.
Preferably, the multi-dimensional influence quantity comprises: the method comprises the steps of selling electric quantity of a power transmission line, voltage, current, temperature, humidity, air pressure, wind speed and wind direction of the environment where the power transmission line is located, commissioning time of line body data of the power transmission line, length of an overhead line, type of a lead, section of the lead, number of split roots, span of each stage of tower of the line, nominal height, pole height, number of loops erected on the same pole, phase sequence, terrain geology, pole tower properties and corner direction.
According to another aspect of the present invention, there is provided a system for predicting a line loss of a power transmission line based on a multidimensional influence quantity, the system comprising:
the acquisition unit is used for acquiring the multidimensional influence quantity of the line loss of the transmission line;
the generating unit is used for carrying out data preprocessing on the line body data in the power transmission line and the non-time sequence data in the tower data of each level of the line, enabling the non-time sequence data to form fixed parameters of the power transmission line, merging the fixed parameters into a time sequence data table of the power transmission line, and generating a multi-dimensional influence quantity data table;
the training unit is used for dividing the data in the multi-dimensional influence quantity data table into training set data, verification set data and test set data; training the gradient lifting tree model through the training set data, and outputting a training result;
the verification unit is used for verifying the training result through verification set data, and when the training result passes the verification, a trained gradient lifting tree model is generated;
and the prediction unit is used for predicting the test set data through the trained gradient lifting tree model and outputting a line loss prediction result of the power transmission line.
Preferably, the training unit is configured to train the gradient lifting tree model through the training set data, output a training result, and further configured to:
and analyzing the weight of the multi-dimensional influence quantity through a gradient lifting tree model.
Preferably, the analyzing the weight of the multi-dimensional influence quantity through the gradient lifting tree model further includes:
training the gradient lifting tree model through the training set data to obtain a trained gradient lifting tree model;
extracting all split node data n of the trained gradient lifting tree model;
calculating the number n of split nodes generated by each dimension influence quantity in the k dimension influence quantityk；
Calculating the weight Q of the k-dimensional influence quantity according to the data n of all the split nodes and the number nk of the split nodes generated by each-dimensional influence quantityk，Qk＝nk/n×100％。
Preferably, the verification unit is configured to verify the training result through verification set data, and when the training result passes the verification, generate a trained gradient lifting tree model, and further configured to:
initializing the gradient lifting tree model, estimating a model parameter which minimizes a loss function, and establishing an initial gradient lifting tree model based on the model parameter;
within the predicted maximum number of iterations, steps S1, S2, S3, and S4 are iteratively performed in order:
s1, calculating a loss function and a negative gradient of the initial gradient lifting tree model;
s2, taking the negative gradient as new training set data, and fitting a regression tree model according to the new training set data;
s3, calculating each leaf node of the regression tree model to obtain a best fitting value;
s4, fusing the best fitting value to a regression tree model;
and when the maximum iteration times are reached, outputting the trained gradient lifting tree model.
Preferably, the multi-dimensional influence quantity comprises: the method comprises the steps of selling electric quantity of a power transmission line, voltage, current, temperature, humidity, air pressure, wind speed and wind direction of the environment where the power transmission line is located, commissioning time of line body data of the power transmission line, length of an overhead line, type of a lead, section of the lead, number of split roots, span of each stage of tower of the line, nominal height, pole height, number of loops erected on the same pole, phase sequence, terrain geology, pole tower properties and corner direction.
The technical scheme of the invention provides a method and a system for predicting transmission line route loss based on multidimensional influence quantity, wherein the method comprises the following steps: collecting the multidimensional influence quantity of the line loss of the transmission line; performing data preprocessing on line body data in a power transmission line and non-time sequence data in tower data of each level of the line to enable the non-time sequence data to form fixed parameters of the power transmission line, merging the fixed parameters into a time sequence data table of the power transmission line, and generating a multi-dimensional influence quantity data table; dividing data in the multi-dimensional influence quantity data table into training set data, verification set data and test set data; training the gradient lifting tree model through training set data, and outputting a training result; verifying the training result through the verification set data, and generating a trained gradient lifting tree model when the training result passes the verification; and predicting the test set data through the trained gradient lifting tree model, and outputting a line loss prediction result of the power transmission line. In order to improve the accuracy of the line loss prediction of the 500kV overhead transmission line, the technical scheme of the invention provides a 500kV overhead transmission line loss prediction method based on multidimensional influence quantity, solves the problem that the line loss prediction precision of the existing 500kV overhead transmission line is not high, and provides certain reference for the line loss treatment direction.
Drawings
A more complete understanding of exemplary embodiments of the present invention may be had by reference to the following drawings in which:
fig. 1 is a flowchart of a method for predicting line loss of a power transmission line based on multidimensional influence quantities according to a preferred embodiment of the present invention;
FIG. 2 is a flow chart of a gradient lifting tree model training process according to a preferred embodiment of the present invention;
FIG. 3 is a flow chart of a gradient lift tree model training process according to a preferred embodiment of the present invention; and
fig. 4 is a diagram illustrating a system for predicting transmission line loss based on multidimensional influence quantities according to a preferred embodiment of the present invention.
Detailed Description
The exemplary embodiments of the present invention will now be described with reference to the accompanying drawings, however, the present invention may be embodied in many different forms and is not limited to the embodiments described herein, which are provided for complete and complete disclosure of the present invention and to fully convey the scope of the present invention to those skilled in the art. The terminology used in the exemplary embodiments illustrated in the accompanying drawings is not intended to be limiting of the invention. In the drawings, the same units/elements are denoted by the same reference numerals.
Unless otherwise defined, terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Further, it will be understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal sense.
Fig. 1 is a flowchart of a method for predicting transmission line loss based on multidimensional influence quantity according to a preferred embodiment of the present invention. The application provides a 500kV overhead transmission line route loss prediction method based on multidimensional influence quantity, which comprises the following steps: collecting multidimensional influence quantity data of a 500kV overhead transmission line; preprocessing the characteristic data of the line tower; and dividing the preprocessed line data, training a gradient lifting tree model to predict line loss and calculating the weight of the multidimensional influence quantity according to a prediction result. The method comprises the steps of monitoring electric energy metering data and environmental data collection line body data of the 500kV overhead transmission line; and the line loss of the 500kV overhead transmission line is predicted according to the monitoring data, and the line loss treatment direction is guided through the calculation result of the multidimensional influence weight in the prediction model, so that the economic loss caused by abnormal line loss is effectively avoided, and the line loss management level of the 500kV overhead transmission line is obviously provided. As shown in fig. 1, a method for predicting line loss of a power transmission line based on multidimensional influence quantity includes:
preferably, in step 101: and collecting the multidimensional influence quantity of the line loss of the transmission line. Preferably, the amount of multi-dimensional influence comprises: the method comprises the steps of selling electric quantity of a power transmission line, voltage, current, temperature, humidity, air pressure, wind speed and wind direction of the environment where the power transmission line is located, commissioning time of line body data of the power transmission line, length of an overhead line, type of a lead, section of the lead, number of split roots, span of each stage of tower of the line, nominal height, pole height, number of loops erected on the same pole, phase sequence, terrain geology, pole tower properties and corner direction. The method and the device collect the related multidimensional influence quantity of the 500kV overhead transmission line route loss.
Preferably, at step 102: and performing data preprocessing on the line body data in the power transmission line and the non-time sequence data in the tower data of each level of the line to enable the non-time sequence data to form fixed parameters of the power transmission line, and merging the fixed parameters into a time sequence data table of the power transmission line to generate a multi-dimensional influence quantity data table. According to the method and the device, data preprocessing is carried out on two types of non-time sequence data of line body data and tower data of all levels of lines, so that fixed parameters of the lines are formed and merged into a time sequence data table of the lines, and a multi-dimensional influence quantity data table is formed. The non-time sequence data preprocessing method in the application is used for numerical data such as: the length, the section of a wire, the span, the call scale and the like of the overhead line are averaged, and the type data are as follows: the phase sequence, the terrain geology and the like adopt ONE-HOT coding, namely, each possible category of category data is used as a single line fixed parameter, and the numerical value of the line fixed parameter is the number of the towers belonging to the category in the line.
Preferably, in step 103: dividing data in the multi-dimensional influence quantity data table into training set data, verification set data and test set data; and training the gradient lifting tree model through the training set data, and outputting a training result. Preferably, training the gradient lifting tree model through training set data, and outputting a training result, further comprising: the weight of the multidimensional influence quantity is analyzed through a gradient lifting tree model. Preferably, the gradient lifting tree model analyzes the weight of the multidimensional influence quantity, and further comprises: training the gradient lifting tree model through training set data to obtain a trained gradient lifting tree model; extracting all split node data n of the trained gradient lifting tree model; calculating the number n of split nodes generated by each dimension influence quantity in the k dimension influence quantityk(ii) a The number n of the split nodes generated according to all the split node data n and the influence quantity of each dimensionkCalculating the weight Q of the k-dimensional influence quantityk，Qk＝nkN × 100%. The application divides the preprocessed data. Based on the data recording date, the multidimensional influence amount data table was divided into 6 uniform parts, of which 5 parts were training verification data of the model and 1 part was test data of the training model. 5 parts of training and verifying data are divided into 5 types of training sets and verifying set combinations by using a 5-fold cross inspection method, 4 parts of 5 parts which are not completely the same are taken from each training set, 1 part of data is taken from the verifying set, and 5 times of circular training is carried out. The method divides the preprocessed data to form training set data, verification set data and test set data, trains the gradient lifting tree model by using the training set data, screens the scores of the verification set data, and forms the modelAnd predicting the data of the test set by the set, averaging the results, outputting the results as line loss prediction results, obtaining a 500kV overhead transmission line loss prediction model, and finally analyzing the weight of the multidimensional influence quantity in the model. The input of the 500kV overhead transmission line route loss prediction model is a multidimensional influence quantity data table in training set data, and the output is a 500kV overhead transmission line route loss value.
Extracting training set data and training a gradient lifting tree model; where a finite set of data D ═ { Z, y } is given, where Z ═ Z1,…,Zi,…ZN]To input, y ═ y1,…yN]Is the output.
As shown in fig. 2, extracting the number n of all split nodes of the trained gradient lifting tree model;
calculating the number n of split nodes generated by each dimension of k-dimension influence quantityk；
Calculating an influence weight Q of a k-dimensional influencek，Qk＝nk/n×100％。
Preferably, at step 104: and verifying the training result through the verification set data, and generating a trained gradient lifting tree model when the training result passes the verification. Preferably, the training result is verified through the verification set data, and when the training result passes the verification, the trained gradient lifting tree model is generated, further comprising: initializing the gradient lifting tree model, estimating a model parameter which minimizes a loss function, and establishing an initial gradient lifting tree model based on the model parameter; iteratively performing steps S1, S2, S3 and S4 for a predicted maximum number of iterations: s1, calculating a loss function and a negative gradient of the initial gradient lifting tree model; s2, taking the negative gradient as new training set data, and fitting a regression tree model according to the new training set data; s3, calculating each leaf node of the regression tree model to obtain a best fitting value; s4, fusing the best fitting value to the regression tree model; and when the maximum iteration times are reached, outputting the trained gradient lifting tree model.
As shown in fig. 3, the present application first initializes the gradient lifting tree model, where D ═ { Z, y } for a given finite dataset, where Z ═ Z1,…,Zi,…ZN]To input, y ═ y1,…yN]Is the output. The model parameter γ that minimizes the loss function L (y, γ) is estimated as the number of data samples, and is used as the initial model f0(Zi) Namely:
Figure BDA0002298801970000081
assuming that C is the number of iterations, for the C-th iteration, C is 1,2, … C, steps S1-S4 are performed;
s1 the negative gradient r of the current model loss function and model is calculated as followsicNamely, the residual:
Figure BDA0002298801970000082
s2 dividing ricAs sample ZiNovel labels of formula wherein f (Z)i) For the model obtained for c-1 iterations, a new sample data set is obtained [ (Z)i,ric),i＝1,2,…N]Fitting the new training data to obtain a next regression tree model, wherein the new tree model is formed by leaf nodes Rjc(J-1, 2, … J). J is the number of leaf nodes of the regression tree model.
S3 for each leaf node RjcCalculating the best fit value gamma of the samplejc。
Figure BDA0002298801970000083
Wherein f isc-1(Zi) A model obtained for c-1 iterations;
s4 updates the model for the mth iteration:
Figure BDA0002298801970000091
I(Zi∈Rjc) For indicating the function, when the sample ZiBelong to leaf node RjcIf so, the function value is 1, otherwise, the function value is 0;
outputting a final gradient lifting tree model fC(Zi)。
Figure BDA0002298801970000092
fo(Zi) Is an initial model; i (Z)i∈Rjc) Is an indicator function;
preferably, at step 105: and predicting the test set data through the trained gradient lifting tree model, and outputting a line loss prediction result of the power transmission line.
Fig. 4 is a diagram illustrating a system for predicting transmission line loss based on multidimensional influence quantities according to a preferred embodiment of the present invention. As shown in fig. 4, the present application provides a system for predicting transmission line loss based on multidimensional influence quantity, the system comprising:
the acquisition unit 401 is configured to acquire a multidimensional influence amount of the transmission line loss. Preferably, the amount of multi-dimensional influence comprises: the method comprises the steps of selling electric quantity of a power transmission line, voltage, current, temperature, humidity, air pressure, wind speed and wind direction of the environment where the power transmission line is located, commissioning time of line body data of the power transmission line, length of an overhead line, type of a lead, section of the lead, number of split roots, span of each stage of tower of the line, nominal height, pole height, number of loops erected on the same pole, phase sequence, terrain geology, pole tower properties and corner direction. The method and the device collect the related multidimensional influence quantity of the 500kV overhead transmission line route loss.
The generating unit 402 is configured to perform data preprocessing on the line body data in the power transmission line and the non-time-series data in the tower data of each stage of the line, so that the non-time-series data form a fixed parameter of the power transmission line, and merge the fixed parameter into the time-series data table of the power transmission line to generate a multidimensional influence quantity data table. According to the method and the device, data preprocessing is carried out on two types of non-time sequence data of line body data and tower data of all levels of lines, so that fixed parameters of the lines are formed and merged into a time sequence data table of the lines, and a multi-dimensional influence quantity data table is formed. The non-time sequence data preprocessing method in the application is used for numerical data such as: the length, the section of a wire, the span, the call scale and the like of the overhead line are averaged, and the type data are as follows: the phase sequence, the terrain geology and the like adopt ONE-HOT coding, namely, each possible category of category data is used as a single line fixed parameter, and the numerical value of the line fixed parameter is the number of the towers belonging to the category in the line.
A training unit 403, configured to divide data in the multidimensional influence quantity data table into training set data, validation set data, and test set data; and training the gradient lifting tree model through the training set data, and outputting a training result. Preferably, the training unit is configured to train the gradient lifting tree model through the training set data, output a training result, and further configured to: the weight of the multidimensional influence quantity is analyzed through a gradient lifting tree model. Preferably, analyzing the weight of the multidimensional influence quantity through the gradient lifting tree model further comprises: training the gradient lifting tree model through training set data to obtain a trained gradient lifting tree model; extracting all split node data n of the trained gradient lifting tree model; calculating the number n of split nodes generated by each dimension influence quantity in the k dimension influence quantityk(ii) a The number n of the split nodes generated according to all the split node data n and the influence quantity of each dimensionkCalculating the weight Q of the k-dimensional influence quantityk，Qk＝nkN × 100%. The application divides the preprocessed data. Based on the data recording date, the multidimensional influence amount data table was divided into 6 uniform parts, of which 5 parts were training verification data of the model and 1 part was test data of the training model. 5 parts of training and verifying data are divided into 5 types of training sets and verifying set combinations by using a 5-fold cross inspection method, 4 parts of 5 parts which are not completely the same are taken from each training set, 1 part of data is taken from the verifying set, and 5 times of circular training is carried out. The method comprises the steps of dividing preprocessed data to form training set data, verification set data and test set data, training a gradient lifting tree model by utilizing the training set data, screening according to scores of the verification set data, predicting the test set data by the formed model set, averaging results, outputting the results as line loss prediction results, obtaining a 500kV overhead transmission line loss prediction model, and finally analyzing the weight of multi-dimensional influence in the model. 500kV overhead transmission line route of this application decreasesThe input of the prediction model is a multidimensional influence data table in training set data, and the output is a 500kV overhead transmission line loss value.
Extracting training set data and training a gradient lifting tree model; where a finite set of data D ═ { Z, y } is given, where Z ═ Z1,…,Zi,…ZN]To input, y ═ y1,…yN]Is the output.
As shown in fig. 2, extracting the number n of all split nodes of the trained gradient lifting tree model;
calculating the number n of split nodes generated by each dimension of k-dimension influence quantityk；
Calculating an influence weight Q of a k-dimensional influencek，Qk＝nk/n×100％。
And the verification unit 404 is configured to verify the training result through the verification set data, and when the training result passes the verification, generate a trained gradient lifting tree model. Preferably, the verification unit 404 is configured to verify the training result through the verification set data, and when the training result passes the verification, generate a trained gradient lifting tree model, and further configured to: initializing the gradient lifting tree model, estimating a model parameter which minimizes a loss function, and establishing an initial gradient lifting tree model based on the model parameter; within the predicted maximum number of iterations, steps S1, S2, S3, and S4 are iteratively performed in order: s1, calculating a loss function and a negative gradient of the initial gradient lifting tree model; s2, taking the negative gradient as new training set data, and fitting a regression tree model according to the new training set data; s3, calculating each leaf node of the regression tree model to obtain a best fitting value; s4, fusing the best fitting value to the regression tree model; and when the maximum iteration times are reached, outputting the trained gradient lifting tree model.
As shown in fig. 3, the present application first initializes a gradient lifting tree model, estimates a model parameter γ that minimizes a loss function L (y, γ), where N is the number of data samples, and uses it as an initial model f0(Zi) And, namely:
Figure BDA0002298801970000111
assuming that C is the number of iterations, for the C-th iteration, C is 1,2, … C, steps S1-S4 are performed;
s1 the negative gradient r of the current model loss function and model is calculated as followsicNamely, the residual:
Figure BDA0002298801970000112
s2 dividing ricAs sample ZiNovel labels of formula wherein f (Z)i) For the model obtained for c-1 iterations, a new sample data set is obtained [ (Z)i，ric),i＝1,2,…N]Fitting the new training data to obtain a next regression tree model, wherein the new tree model is formed by leaf nodes Rjc(J-1, 2, … J). J is the number of leaf nodes of the regression tree model.
S3 for each leaf node RjcCalculating the best fit value gamma of the samplejc。
Figure BDA0002298801970000121
Wherein f isc-1(Zi) A model obtained for c-1 iterations;
s4 updates the model for the mth iteration:
Figure BDA0002298801970000122
Figure BDA0002298801970000123
I(Zi∈Rjc) For indicating the function, when the sample ZiBelong to leaf node RjcIf so, the function value is 1, otherwise, the function value is 0;
outputting a final gradient lifting tree model fC(Zi)。
Figure BDA0002298801970000124
fo(Zi) Is an initial model; i (Z)i∈Rjc) Is an indicator function;
and the prediction unit 405 is configured to predict the test set data through the trained gradient lifting tree model, and output a line loss prediction result of the power transmission line.
A system 400 for predicting transmission line loss based on multidimensional influence measures according to the preferred embodiment of the present invention corresponds to the method 100 for predicting transmission line loss based on multidimensional influence measures according to the preferred embodiment of the present invention, and is not described herein again.
The invention has been described with reference to a few embodiments. However, other embodiments of the invention than the one disclosed above are equally possible within the scope of the invention, as would be apparent to a person skilled in the art from the appended patent claims.
Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise herein. All references to "a/an/the [ device, component, etc ]" are to be interpreted openly as referring to at least one instance of said device, component, etc., unless explicitly stated otherwise. The steps of any method disclosed herein do not have to be performed in the exact order disclosed, unless explicitly stated.

Claims (10)

1. A method for predicting line loss of a power transmission line based on multidimensional influence quantity, the method comprising:
collecting the multidimensional influence quantity of the line loss of the transmission line;
performing data preprocessing on line body data in the power transmission line and non-time sequence data in tower data of each level of the line to enable the non-time sequence data to form fixed parameters of the power transmission line, and merging the fixed parameters into a time sequence data table of the power transmission line to generate a multi-dimensional influence quantity data table;
dividing data in the multi-dimensional influence quantity data table into training set data, verification set data and test set data; training the gradient lifting tree model through the training set data, and outputting a training result;
verifying the training result through verification set data, and generating a trained gradient lifting tree model when the training result passes the verification;
and predicting the test set data through the trained gradient lifting tree model, and outputting a line loss prediction result of the power transmission line.
2. The method of claim 1, wherein training a gradient-boosted tree model with the training set data and outputting a training result, further comprises:
and analyzing the weight of the multi-dimensional influence quantity through a gradient lifting tree model.
3. The method of claim 2, the analyzing weights of the multi-dimensional influence quantities by a gradient lifting tree model, further comprising:
training the gradient lifting tree model through the training set data to obtain a trained gradient lifting tree model;
extracting all split node data n of the trained gradient lifting tree model;
calculating the number n of split nodes generated by each dimension influence quantity in the k dimension influence quantityk；
The number n of the split nodes generated according to the data n of all the split nodes and the influence quantity in each dimensionkCalculating the weight Q of the k-dimensional influence quantityk，Qk＝nk/n×100％。
4. The method of claim 1, wherein the training results are validated by validation set data, and when the training results are validated, a trained gradient-boosted tree model is generated, further comprising:
initializing the gradient lifting tree model, estimating a model parameter which minimizes a loss function, and establishing an initial gradient lifting tree model based on the model parameter;
within the predicted maximum number of iterations, steps S1, S2, S3, and S4 are iteratively performed in order:
s1, calculating a loss function and a negative gradient of the initial gradient lifting tree model;
s2, taking the negative gradient as new training set data, and fitting a regression tree model according to the new training set data;
s3, calculating each leaf node of the regression tree model to obtain a best fitting value;
s4, fusing the best fitting value to a regression tree model;
and when the maximum iteration times are reached, outputting the trained gradient lifting tree model.
5. The method of claim 1, the multi-dimensional impact quantity comprising: the method comprises the steps of selling electric quantity of a power transmission line, voltage, current, temperature, humidity, air pressure, wind speed and wind direction of the environment where the power transmission line is located, commissioning time of line body data of the power transmission line, length of an overhead line, type of a lead, section of the lead, number of split roots, span of each stage of tower of the line, nominal height, pole height, number of loops erected on the same pole, phase sequence, terrain geology, pole tower properties and corner direction.
6. A system for predicting transmission line loss based on multidimensional impact quantities, the system comprising:
the acquisition unit is used for acquiring the multidimensional influence quantity of the line loss of the transmission line;
the generating unit is used for carrying out data preprocessing on the line body data in the power transmission line and the non-time sequence data in the tower data of each level of the line, enabling the non-time sequence data to form fixed parameters of the power transmission line, merging the fixed parameters into a time sequence data table of the power transmission line, and generating a multi-dimensional influence quantity data table;
the training unit is used for dividing the data in the multi-dimensional influence quantity data table into training set data, verification set data and test set data; training the gradient lifting tree model through the training set data, and outputting a training result;
the verification unit is used for verifying the training result through verification set data, and when the training result passes the verification, a trained gradient lifting tree model is generated;
and the prediction unit is used for predicting the test set data through the trained gradient lifting tree model and outputting a line loss prediction result of the power transmission line.
7. The system of claim 6, the training unit is configured to train a gradient lifting tree model through the training set data, output a training result, and further configured to:
and analyzing the weight of the multi-dimensional influence quantity through a gradient lifting tree model.
8. The system of claim 7, the analyzing weights of the multi-dimensional influence quantities by a gradient lifting tree model, further comprising:
training the gradient lifting tree model through the training set data to obtain a trained gradient lifting tree model;
extracting all split node data n of the trained gradient lifting tree model;
calculating the number n of split nodes generated by each dimension influence quantity in the k dimension influence quantityk；
The number n of the split nodes generated according to the data n of all the split nodes and the influence quantity in each dimensionkCalculating the weight Q of the k-dimensional influence quantityk，Qk＝nk/n×100％。
9. The system of claim 6, wherein the validation unit is configured to validate the training result with validation set data, and when the training result is validated, generate a trained gradient-boosted tree model, and further configured to:
initializing the gradient lifting tree model, estimating a model parameter which minimizes a loss function, and establishing an initial gradient lifting tree model based on the model parameter;
within the predicted maximum number of iterations, steps S1, S2, S3, and S4 are iteratively performed in order:
s1, calculating a loss function and a negative gradient of the initial gradient lifting tree model;
s2, taking the negative gradient as new training set data, and fitting a regression tree model according to the new training set data;
s3, calculating each leaf node of the regression tree model to obtain a best fitting value;
s4, fusing the best fitting value to a regression tree model;
and when the maximum iteration times are reached, outputting the trained gradient lifting tree model.
10. The system of claim 6, the multi-dimensional impact quantity comprising: the method comprises the steps of selling electric quantity of a power transmission line, voltage, current, temperature, humidity, air pressure, wind speed and wind direction of the environment where the power transmission line is located, commissioning time of line body data of the power transmission line, length of an overhead line, type of a lead, section of the lead, number of split roots, span of each stage of tower of the line, nominal height, pole height, number of loops erected on the same pole, phase sequence, terrain geology, pole tower properties and corner direction.
CN201911213450.8A 2019-12-02 2019-12-02 Method and system for predicting line loss of power transmission line based on multidimensional influence quantity Pending CN110956330A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911213450.8A CN110956330A (en) 2019-12-02 2019-12-02 Method and system for predicting line loss of power transmission line based on multidimensional influence quantity

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911213450.8A CN110956330A (en) 2019-12-02 2019-12-02 Method and system for predicting line loss of power transmission line based on multidimensional influence quantity

Publications (1)

Publication Number Publication Date
CN110956330A true CN110956330A (en) 2020-04-03

Family

ID=69979307

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911213450.8A Pending CN110956330A (en) 2019-12-02 2019-12-02 Method and system for predicting line loss of power transmission line based on multidimensional influence quantity

Country Status (1)

Country Link
CN (1) CN110956330A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113533882A (en) * 2021-06-29 2021-10-22 国网新疆电力有限公司信息通信公司 Height measurement equipment for detecting line loss of power transmission line of power tower
WO2023221426A1 (en) * 2022-05-16 2023-11-23 中国南方电网有限责任公司超高压输电公司检修试验中心 Direct-current transmission project loss evaluation method and apparatus, device, and storage medium

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113533882A (en) * 2021-06-29 2021-10-22 国网新疆电力有限公司信息通信公司 Height measurement equipment for detecting line loss of power transmission line of power tower
CN113533882B (en) * 2021-06-29 2023-05-12 国网新疆电力有限公司信息通信公司 Height measurement equipment for detecting line loss of power transmission line of power tower
WO2023221426A1 (en) * 2022-05-16 2023-11-23 中国南方电网有限责任公司超高压输电公司检修试验中心 Direct-current transmission project loss evaluation method and apparatus, device, and storage medium

Similar Documents

Publication Publication Date Title
CN108375715B (en) Power distribution network line fault risk day prediction method and system
CN110555230B (en) Rotary machine residual life prediction method based on integrated GMDH framework
CN106059661A (en) Time sequence analysis based optical transmission network trend prediction method
JP2014105989A (en) Energy consumption prediction method of building power equipment
CN106229972B (en) A kind of wind power forecasting method integrated based on more meteorological sources and segmentation modeling
CN102789545A (en) Method for predicating remaining life of turbine engine based on degradation model matching
CN110059845B (en) Metering device clock error trend prediction method based on time sequence evolution gene model
CN107545110B (en) Dynamic stress accelerated life test profile compiling method
CN103631681A (en) Method for online restoring abnormal data of wind power plant
CN101477340A (en) Progressive health monitoring method for cable system in cable structure based on cable force monitoring
CN102521080A (en) Computer data recovery method for electricity-consumption information collecting system for power consumers
US20160282400A1 (en) Equipment failure prediction system, equipment failure prediction device and equipment failure prediction method
CN110956330A (en) Method and system for predicting line loss of power transmission line based on multidimensional influence quantity
CN111881124A (en) Data processing method and system based on state estimation of improved algorithm
CN114265001B (en) Smart electric meter metering error evaluation method
CN111612149A (en) Main network line state detection method, system and medium based on decision tree
CN105488572A (en) Health state evaluation method of power distribution equipment
CN105259398A (en) Method for analyzing random error characteristic of optical fiber current transformer based on total variance
CN103279030B (en) Dynamic soft measuring modeling method and device based on Bayesian frame
CN112345972B (en) Power distribution network line transformation relation abnormity diagnosis method, device and system based on power failure event
CN113269419A (en) Building engineering project material incoming acceptance management method based on intellectualization and cloud management platform
CN103400201A (en) Method for solving state estimation problem taking maximum normal rate of measurement point as target
CN116957534A (en) Method for predicting replacement number of intelligent electric meter
CN116756199A (en) Wind measurement data processing recommendation method and system
CN114583767B (en) Data-driven wind power plant frequency modulation response characteristic modeling method and system

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