CN112801365A - Power distribution network reliability prediction method, device, equipment and medium - Google Patents
Power distribution network reliability prediction method, device, equipment and medium Download PDFInfo
- Publication number
- CN112801365A CN112801365A CN202110112126.8A CN202110112126A CN112801365A CN 112801365 A CN112801365 A CN 112801365A CN 202110112126 A CN202110112126 A CN 202110112126A CN 112801365 A CN112801365 A CN 112801365A
- Authority
- CN
- China
- Prior art keywords
- households
- power failure
- fault
- predicted
- month
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000009826 distribution Methods 0.000 title claims abstract description 57
- 230000015654 memory Effects 0.000 claims abstract description 35
- 238000000605 extraction Methods 0.000 claims abstract description 30
- 238000003062 neural network model Methods 0.000 claims abstract description 20
- 238000013528 artificial neural network Methods 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 11
- 238000007781 pre-processing Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 8
- 238000003860 storage Methods 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 230000007774 longterm Effects 0.000 description 5
- 230000004913 activation Effects 0.000 description 4
- 238000005065 mining Methods 0.000 description 4
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000012706 support-vector machine Methods 0.000 description 3
- ZOJBYZNEUISWFT-UHFFFAOYSA-N allyl isothiocyanate Chemical compound C=CCN=C=S ZOJBYZNEUISWFT-UHFFFAOYSA-N 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000006403 short-term memory Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- 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
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- General Health & Medical Sciences (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Game Theory and Decision Science (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Educational Administration (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The application discloses a method, a device, equipment and a medium for predicting the reliability of a power distribution network, wherein the method comprises the following steps: acquiring historical data of the number of households when a first fault of a plurality of district power supply offices before a month to be predicted is powered off; performing characteristic extraction on the historical data of the number of the users when the first fault has power failure to obtain a first time sequence characteristic variable; inputting the first time sequence characteristic variable into the long-short term memory neural network model to predict the number of households in the case of power failure, and obtaining a predicted value of the number of households in the case of power failure in each district power supply office or the whole situation in the month to be predicted; and calculating the reliability index prediction result of each district power supply office or the whole in the month to be predicted according to the number of households in the prearranged period of the month to be predicted or the whole in the month to be predicted and the number of households in the failure power failure. The method solves the technical problem that the reliability prediction result is low in precision due to the fact that the existing power distribution network reliability prediction method based on the regression prediction model lacks consideration of time sequence correlation.
Description
Technical Field
The application relates to the technical field of power distribution networks, in particular to a method, a device, equipment and a medium for predicting reliability of a power distribution network.
Background
The power supply reliability is a main index for measuring the quality and the function of the power system, and with the development of science and technology and the continuous improvement of the social and economic levels, the continuous and reliable power supply cannot be realized in production and life of people, the attention degree of a user to the power supply reliability is continuously improved, and the customer satisfaction degree of a power supply enterprise is closely related to the power supply reliability index.
The reliability of the distribution system, which is an important link between the transmission and transformation system and the power consumers, directly affects the reliability and satisfaction degree of the consumers on the power supply capacity and the operation level of the power grid company. The reliability of the power distribution network is comprehensively embodied in various aspects such as power grid planning, grid structure level, equipment health state, operation management level and the like. With the continuous expansion of the scale of the urban power distribution network, how to quickly and effectively carry out reliability assessment and prediction on the power distribution network, and a constructive suggestion is provided according to a prediction result to guide reliability work to be a research hotspot, so that the method has important theoretical significance and practical value.
In the prior art, when the reliability of the power distribution network is evaluated and predicted, the reliability of the power distribution network is predicted usually based on regression prediction models such as an artificial neural network, a support vector machine or an extreme learning machine, and a mapping relation is established between reliability influencing factors and reliability indexes, but the method lacks consideration of time series correlation, so that the accuracy of a reliability prediction result is low.
Disclosure of Invention
The application provides a method, a device, equipment and a medium for predicting the reliability of a power distribution network, which are used for solving the technical problem that the reliability prediction result is low in precision due to the fact that the existing method for predicting the reliability of the power distribution network based on a regression prediction model lacks consideration of time sequence correlation.
In view of this, the first aspect of the present application provides a method for predicting reliability of a power distribution network, including:
acquiring historical data of the number of households when a first fault of a plurality of district power supply offices before a month to be predicted is powered off;
performing characteristic extraction on the historical data of the number of the users when the first fault has power failure to obtain a first time sequence characteristic variable;
inputting the first time sequence characteristic variable into a long-short term memory neural network model to predict the number of households in the power failure, so as to obtain a predicted value of the number of households in the power failure of each district power supply office or the whole situation in the month to be predicted;
and calculating the reliability index prediction result of each district power supply office or the whole in the month to be predicted according to the number of the households in the prearranged period of the month to be predicted of each district power supply office or the whole in the month to be predicted and the predicted value of the number of the households in the fault power failure period.
Optionally, the performing feature extraction on the historical data of the number of users during the first fault power failure to obtain a first time-sequence feature variable includes:
extracting date features, area names, the number of days of fault power failure in a preset time period, the minimum number of households in fault power failure, the average number of households in fault power failure, the median number of households in fault power failure, the maximum number of households in fault power failure, the standard deviation of the number of households in fault power failure, the bias of the number of households in fault power failure, the average value of daily variation of the number of households in time, the maximum value of daily variation of the number of households in time, the minimum value of daily variation of the number of households in time, the standard deviation of daily variation of the number of households in time and/or the number of households in day fault power failure to obtain a first time sequence feature variable.
Optionally, the performing feature extraction on the historical data of the number of users during the first fault power failure to obtain a first time-sequence feature variable, and then further including:
and preprocessing the first time sequence characteristic variable.
Optionally, the configuration process of the long-term and short-term memory neural network model is as follows:
acquiring historical data of the number of households when a second fault of a plurality of regional power supply offices fails;
performing characteristic extraction on the historical data of the number of the users when the second fault has power failure to obtain a second time sequence characteristic variable;
and training the long-short term memory neural network through the second time sequence characteristic variable to obtain the long-short term memory neural network model.
This application second aspect provides a distribution network reliability prediction device, includes:
the acquisition unit is used for acquiring the historical data of the number of households when the power failure occurs in the first fault of the power supply bureaus of the plurality of areas before the month to be predicted;
the characteristic extraction unit is used for extracting the characteristics of the historical data of the number of the users when the first fault is in power failure to obtain a first time sequence characteristic variable;
the prediction unit is used for inputting the first time sequence characteristic variable into the long-short term memory neural network model to predict the number of households in the fault power failure, so as to obtain the predicted value of the number of households in the fault power failure of each district power supply office or the whole situation in the month to be predicted;
and the calculation unit is used for calculating the reliability index prediction result of each district power supply office or the whole in the month to be predicted according to the number of the households during the prearrangement of the month to be predicted by each district power supply office or the whole in the month to be predicted and the predicted value of the number of the households during the fault power failure.
Optionally, the feature extraction unit is specifically configured to:
extracting date features, area names, the number of days of fault power failure in a preset time period, the minimum number of households in fault power failure, the average number of households in fault power failure, the median number of households in fault power failure, the maximum number of households in fault power failure, the standard deviation of the number of households in fault power failure, the bias of the number of households in fault power failure, the average value of daily variation of the number of households in time, the maximum value of daily variation of the number of households in time, the minimum value of daily variation of the number of households in time, the standard deviation of daily variation of the number of households in time and/or the number of households in day fault power failure to obtain a first time sequence feature variable.
Optionally, the method further includes:
and the preprocessing unit is used for preprocessing the first time sequence characteristic variable.
Optionally, the method further includes:
a configuration unit, specifically configured to:
acquiring historical data of the number of households when a second fault of a plurality of regional power supply offices fails;
performing characteristic extraction on the historical data of the number of the users when the second fault has power failure to obtain a second time sequence characteristic variable;
and training the long-short term memory neural network through the second time sequence characteristic variable to obtain the long-short term memory neural network model.
A third aspect of the present application provides a power distribution network reliability prediction device, which includes a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is configured to execute the method for predicting reliability of a power distribution network according to any one of the first aspect according to instructions in the program code.
A fourth aspect of the present application provides a computer-readable storage medium for storing program code for executing the method for predicting reliability of a power distribution network according to any one of the first aspect.
According to the technical scheme, the method has the following advantages:
the application provides a power distribution network reliability prediction method, which comprises the following steps: acquiring historical data of the number of households when a first fault of a plurality of district power supply offices before a month to be predicted is powered off; performing characteristic extraction on the historical data of the number of the users when the first fault has power failure to obtain a first time sequence characteristic variable; inputting the first time sequence characteristic variable into the long-short term memory neural network model to predict the number of households in the case of power failure, and obtaining a predicted value of the number of households in the case of power failure in each district power supply office or the whole situation in the month to be predicted; and calculating the reliability index prediction result of each district power supply office or the whole in the month to be predicted according to the number of households in the prearranged period of the month to be predicted or the whole in the month to be predicted and the number of households in the failure power failure.
In the application, after historical data of the number of households when the power failure of the first fault of each district power supply office is in front of the month to be predicted is acquired, extracting the characteristic variable related to the time sequence of the historical data of the number of users when the first fault has power failure to obtain the first time sequence characteristic variable, fully mining the internal association between the time sequence data, and finally, calculating a reliability index prediction result of each district power supply office or the whole in the month to be predicted according to the number of the households in the prearranged month of each district power supply office or the whole in the month to be predicted and the number of the households in the power failure predicted value, so that the technical problem that the reliability prediction result is low in precision due to lack of consideration of time sequence correlation in the conventional power distribution network reliability prediction method based on a regression prediction model is solved.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without inventive exercise.
Fig. 1 is a schematic flowchart of a method for predicting reliability of a power distribution network according to an embodiment of the present disclosure;
fig. 2 is a schematic diagram of a power outage nature division according to an embodiment of the present application;
fig. 3 is a schematic structural diagram of a basic unit of an LSTM network according to an embodiment of the present application;
FIG. 4 is a schematic diagram showing comparison of predicted results of 3 prediction methods provided in the present embodiment in 8 months of 2020;
FIG. 5 is a schematic diagram illustrating comparison of predicted results of global and regional offices in different months according to an embodiment of the present disclosure;
fig. 6 is a schematic structural diagram of a power distribution network reliability prediction apparatus according to an embodiment of the present application.
Detailed Description
The application provides a method, a device, equipment and a medium for predicting the reliability of a power distribution network, which are used for solving the technical problem that the reliability prediction result is low in precision due to the fact that the existing method for predicting the reliability of the power distribution network based on a regression prediction model lacks consideration of time sequence correlation.
In order to make the technical solutions of the present application better understood, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
The existing power distribution network reliability prediction method can be classified from two aspects of power distribution network reliability indexes and prediction methods.
1. The reliability index of the power distribution network is as follows:
the reliability index of the power distribution network measures the capacity of the power distribution network for continuously providing electric energy for users, and is an important basis for a power distribution network manager to evaluate the self management level, draw up a planning scheme and determine an investment strategy. The statistical indexes of the reliability of the power distribution network mainly comprise: a Reliability on Service (RS), an Average user outage time (AIHC), an Average user outage time (AITC), and the like. As shown in fig. 2, the power failure properties include pre-scheduled power failure and fault power failure, the reliability indexes are further subdivided according to the influence of different power failure properties, such as average pre-scheduled power failure time of a user, average pre-scheduled power failure time of the user, average pre-scheduled power failure times of the user, and the like, and a single index or weighted integration of multiple indexes can be selected according to different research requirements.
2. The method for predicting the reliability of the power distribution network comprises the following steps:
the principle of power distribution network reliability prediction can be described as: historical operating state data and fault data are modeled to infer reliability that will be exhibited later. At present, most mainstream reliability prediction technical means are based on a machine learning algorithm, for example, a regression prediction model based on an Artificial Neural Network (ANN), a Support Vector Machine (SVM), and an Extreme Learning Machine (ELM), and a basic idea is to establish a mapping relationship between a reliability influence factor and a reliability index. The method lacks consideration on the time sequence correlation, and has the problem of ineffective convergence when a plurality of samples exist, so that the accuracy of a prediction result is not high; moreover, the model needs to depend on timeliness and accuracy of various reliability influence factors, and the application difficulty is high.
Based on the reasons, the power distribution network reliability prediction method based on time series prediction is provided, modeling is performed only based on the change trend of the reliability index along with the time characteristic, timeliness and accuracy of other characteristic variables are not relied on, the problem that effective convergence cannot be achieved when a large number of samples exist is avoided, and therefore accuracy of prediction results is improved.
For easy understanding, referring to fig. 1, an embodiment of a method for predicting reliability of a power distribution network provided in the present application includes:
In consideration of the fact that the existing distribution network reliability prediction indexes are not strong in selection pertinence, one or more of reliability indexes such as system power supply reliability, user average power failure time and user average power failure times are mostly directly used as prediction targets, and are not divided according to power failure properties, so that the pertinence is insufficient. The index of the number of the users in the time can be classified into the number of the users in the scheduled power failure and the number of the users in the failure power failure according to the power failure property. Due to the fact that prearranged measures such as scheduled maintenance, uninterrupted operation, load transfer scheme and the like can be 'calculated first and then stopped', prearranged reliability indexes can be accurately estimated. Therefore, the difficulty of reliability prediction mainly lies in fault power Failure, so that the embodiment of the application directly adopts a total Number of customer homes of Interruption to Failure (NCHI-F) index as a prediction object, and the NCHI-F index has the characteristics of intuition and easy quantification, and can be respectively modeled, predicted, evaluated and controlled by global decomposition to a regional power supply office and a power supply station. Where NCHI-F (h.household) is defined as follows:
NCHI-F is equal to sigma, and the time of each fault power failure is multiplied by the number of users influenced by each fault power failure.
In the embodiment of the application, the historical data of the number of households when the power supply bureaus of a plurality of areas in a certain area have the first failure before the month to be predicted and have power failure can be obtained, and the historical data of the number of households when the power supply bureaus have the failure 60 days before the month to be predicted and have power failure can be obtained as the historical data of the number of households when the power supply bureaus have the first failure and power failure. For example, the number of households during the failure and power failure of each district power supply office in the area in month 8 needs to be predicted, the number of households during the failure and power failure per unit in month 6 to month 7 can be obtained, and by taking the day as a unit, the power supply offices in each district can be further aggregated respectively, and the outlier sample such as the major event day is removed.
And 102, performing feature extraction on the historical data of the number of the users when the first fault has power failure to obtain a first time sequence feature variable.
In the embodiment of the application, the date feature, the district name, the number of days of the fault power failure in the preset time period, the minimum number of households in the fault power failure, the average number of households in the fault power failure, the median number of households in the fault power failure, the maximum number of households in the fault power failure, the standard deviation of the number of households in the fault power failure, the average value of daily variation of the number of households, the maximum value of daily variation of the number of households, the minimum value of daily variation of the number of households, the standard deviation of daily variation of the number of households and/or the number of households in the day of the fault power failure are extracted, and the first time sequence feature variable.
Specifically, the parameter end _ date may be set as the last date of the historical data of the number of households when the first failure has failed, the parameter n _ pre _ days may be set as 60, that is, the number of households when the failure has failed and failed 60 days before the predicted date is used for feature extraction, the time sequence information of the number of households when the failure has failed and failed recently is mined, and the extracted first time sequence feature variable may specifically refer to table 1.
TABLE 1 first timing characteristic variables
After the first time sequence feature variable is extracted, the first time sequence feature variable may be preprocessed, and specifically, the first time sequence feature variable may be subjected to normalization processing or one-hot coding according to a feature type of the first time sequence feature variable and then used as an input of the long-short term memory neural network model. By extracting the time sequence characteristic variables, the time sequence correlation of the daily characteristic of the reliability index can be fully mined.
After the first time sequence characteristic variable is extracted and obtained, the first time sequence characteristic variable is input into the long-short term memory neural network model to predict the number of households in the fault power failure, a predicted value of the number of households in the fault power failure of each district power supply station or the whole situation in the month to be predicted is obtained, and the predicted value of the number of households in the fault power failure of each day of the month to be predicted in the power supply station or the whole situation in the region.
Further, the configuration process of the long-short term memory neural network model in the embodiment of the present application is as follows:
and S1, acquiring the historical data of the number of the households when the second fault of the power supply bureaus of the plurality of areas is in power failure.
The acquired historical data of the number of the households during the second fault power failure of the power supply bureaus of the plurality of areas is longer than the time span of the historical data of the number of the households during the first fault power failure, and the historical data of the number of the households during the second fault power failure can be the number of the households during the fault power failure of nearly two years.
And S2, performing feature extraction on the historical data of the number of users during the second fault power failure to obtain a second time sequence feature variable.
And performing feature extraction on the historical data of the number of users in the second fault power failure to obtain a second time sequence feature variable, wherein the specific process of performing feature extraction on the historical data of the number of users in the first fault power failure to obtain the first time sequence feature variable is similar to the specific process of performing feature extraction on the historical data of the number of users in the first fault power failure to obtain the first time sequence feature variable. And selecting end _ date by sliding, and establishing a training set sample by utilizing the number of households in the power failure of the fault in the last two years. After the second time series characteristic variable is extracted, the second time series characteristic variable may be preprocessed, and specifically, normalization processing or one-hot encoding may be performed according to a characteristic type of the second time series characteristic variable.
And S3, training the long-short term memory neural network through the second time sequence characteristic variable to obtain a long-short term memory neural network model.
Long Short-term memory neural network (LSTM) is an improved Recurrent Neural Network (RNN) that adds an additional forgetting gate. The improved LSTM network solves the problems of gradient disappearance and gradient explosion of the model in training and can effectively learn the long-term dependence and correlation among data. Therefore, the LSTM network can be applied in the fields of time series and the like.
The basic unit of the LSTM network is shown in fig. 3, and the LSTM network records the long-term dependency (long-term information) between time sequence data through a memory unit composed of an internal forgetting gate, an input gate and an output gate, and temporally records the long-term dependency (long-term information) between the time sequence dataFeatures are captured in the neighboring information (short-term information). The LSTM unit accepts the current data input x through 3 gates at each instanttState h from the last hidden layert-1And a memory cell ct-1. The forward propagation process is as follows:
ft＝σ(Wf[ht-1,xt]+bf)；
in the formula, xtIs a first timing characteristic variable of the input, ftA forgetting gate value; i.e. itIs an input gate; stIs a temporary state quantity; c. CtIs a state quantity; otIntermediate output; h istIs the output gate value; y istIs an output; wf、Wi、Ws、WoRespectively corresponding weight matrixes; bf、bi、bs、boRespectively corresponding bias terms; an array product method, i.e., multiplication of elements of a vector; σ is the activation function.
The expression of the sigmoid activation function is as follows:
the expression of the tanh activation function is:
the expression of the softmax activation function is:
and training the long-short term memory neural network through the preprocessed second time sequence characteristic variable to obtain a long-short term memory neural network model. The prediction performance of the model depends on selection of the hyper-parameters to a great extent, and the hyper-parameters in the embodiment of the application mainly comprise: the network result hyper-parameter and the optimization algorithm hyper-parameter are tested, and the optimal hyper-parameter is selected as shown in table 2.
TABLE 2 Superparameter selection case
And 104, calculating a reliability index prediction result of each district power supply office or the whole in the month to be predicted according to the number of the households in the prearranged period of the month to be predicted of each district power supply office or the whole in the month to be predicted and the predicted value of the number of the households in the failure power failure period.
Calculating the sum of the predicted values of the number of households when the power supply bureaus of each area are pre-arranged in the month to be predicted and the number of households when the power supply bureaus of each area are in failure power failure, and obtaining the reliability index prediction result of the power supply bureaus of each area in the month to be predicted; and calculating the sum of the predicted values of the number of the households when the month to be predicted is prearranged and the number of the households when the power fails, and obtaining the reliability index prediction result of the month to be predicted. The method and the device can realize the reliability prediction of large time scale crossing and also can realize the monthly reliability index prediction of the global and regional power supply bureau.
In the embodiment of the application, after the historical data of the number of households when the power failure of the first fault before the month to be predicted of each power supply office is obtained, extracting the characteristic variable related to the time sequence of the historical data of the number of users when the first fault has power failure to obtain the first time sequence characteristic variable, fully mining the internal association between the time sequence data, and finally, calculating a reliability index prediction result of each district power supply office or the whole in the month to be predicted according to the number of the households in the prearranged month of each district power supply office or the whole in the month to be predicted and the number of the households in the power failure predicted value, so that the technical problem that the reliability prediction result is low in precision due to lack of consideration of time sequence correlation in the conventional power distribution network reliability prediction method based on a regression prediction model is solved.
The above is an embodiment of the power distribution network reliability prediction method provided by the present application, and the following is an application example of the power distribution network reliability prediction method provided by the present application.
According to the embodiment of the application, the number of households in daily fault power failure of 11 district power supply offices in Guangzhou in the period from 1/2018 to 7/31/2020 is obtained, the districts are aggregated respectively, and the daily outlier sample of a major event is removed. Firstly, setting the parameter end _ date to be 7 months and 31 days, setting the parameter n _ pre _ days to be 60, namely, using the number of households during power failure of the fault 60 days before the prediction day to carry out feature extraction, and mining the time sequence information of the number of households during recent fault. Selecting end _ date in a sliding mode, establishing a training set sample by utilizing data of the last two years, carrying out normalization and one-hot code conversion according to the feature type of the extracted time sequence feature variable, then using the normalized and one-hot code converted data as input, and training a multi-step prediction LSTM model in the next step.
The embodiment of the application adopts three schemes for comparison and test: (1) model 1: the BPNN model is trained by a single characteristic variable without characteristic extraction; (2) model 2: the LSTM model is trained by a single characteristic variable without characteristic extraction; (3) model 3: and (3) carrying out feature extraction on the LSTM model with multiple feature variables (the scheme provided by the application). The actual number of households and the predicted number of households and the error condition in 8 months in 2020 are shown in fig. 4 and table 3.
The method comprises the following steps of (1) taking an absolute percentage error as an evaluation index of a prediction result of the number of households in the event of monthly fault power failure:
TABLE 3 comparison of prediction results of different prediction methods in 8 months of 2020
The average error for the prediction in 11 regions compared to the three schemes by figure 4 and table 3 is: 67%, 37%, 25%, from which it can be seen:
(1) on the task of predicting the reliability index, the LSTM algorithm performs better than the traditional BPNN algorithm.
(2) The characteristic extraction method adopted by the embodiment of the application converts single-variable time sequence prediction into multi-variable multi-step prediction, can more fully excavate effective information of time sequence characteristics, and improves prediction accuracy.
The prediction results are shown in table 4 and fig. 5, in which 4 months from 5 months to 8 months in 2020 are used as test sets for prediction comparison.
TABLE 4 comparison of global and regional local predicted results in different months
The reliability of the method for predicting the number of users in the fault power failure provided by the embodiment of the present application is further described by using table 4 and fig. 5, so that the reliability of the reliability index prediction result obtained by calculating the predicted number of users in the fault power failure can be ensured.
The foregoing is an application example of the power distribution network reliability prediction method provided by the present application, and the following is an embodiment of a power distribution network reliability prediction device provided by the present application.
Referring to fig. 6, an apparatus for predicting reliability of a power distribution network provided in an embodiment of the present application includes:
the acquisition unit is used for acquiring the historical data of the number of households when the power failure occurs in the first fault of the power supply bureaus of the plurality of areas before the month to be predicted;
the characteristic extraction unit is used for extracting the characteristics of the historical data of the number of the users when the first fault is in power failure to obtain a first time sequence characteristic variable;
the prediction unit is used for inputting the first time sequence characteristic variable into the long-short term memory neural network model to predict the number of households in the fault power failure, and obtaining the predicted value of the number of households in the fault power failure of each district power supply office or the whole situation in the month to be predicted;
and the computing unit is used for computing the reliability index prediction result of each district power supply office or the whole in the month to be predicted according to the number of the households during the prearranged time of the month to be predicted of each district power supply office or the whole in the month to be predicted and the predicted value of the number of the households during the fault power failure.
As a further refinement, the feature extraction unit is specifically configured to:
extracting date features, area names, days of failure power failure in a preset time period, the minimum value of the number of households in failure power failure, the average value of the number of households in failure power failure, the median of the number of households in failure power failure, the maximum value of the number of households in failure power failure, the standard deviation of the number of households in failure power failure, the skewness of the number of households in failure power failure, the average value of daily variation of the number of households in failure, the maximum value of daily variation of the number of households in failure, the minimum value of daily variation of the number of households in failure, the standard deviation of daily variation of the number of households in failure power failure and/or the number of households in failure power failure in day to obtain a.
As a further improvement, the method further comprises the following steps:
and the preprocessing unit is used for preprocessing the first time sequence characteristic variable.
As a further improvement, the method further comprises the following steps:
a configuration unit, specifically configured to:
acquiring historical data of the number of households when a second fault of a plurality of regional power supply offices fails;
performing characteristic extraction on the historical data of the number of the users during the power failure of the second fault to obtain a second time sequence characteristic variable;
and training the long-short term memory neural network through the second time sequence characteristic variable to obtain a long-short term memory neural network model.
In the embodiment of the application, after the historical data of the number of households when the power failure of the first fault before the month to be predicted of each power supply office is obtained, extracting the characteristic variable related to the time sequence of the historical data of the number of users when the first fault has power failure to obtain the first time sequence characteristic variable, fully mining the internal association between the time sequence data, and finally, calculating a reliability index prediction result of each district power supply office or the whole in the month to be predicted according to the number of the households in the prearranged month of each district power supply office or the whole in the month to be predicted and the number of the households in the power failure predicted value, so that the technical problem that the reliability prediction result is low in precision due to lack of consideration of time sequence correlation in the conventional power distribution network reliability prediction method based on a regression prediction model is solved.
The embodiment of the application also provides a power distribution network reliability prediction device, which comprises a processor and a memory;
the memory is used for storing the program codes and transmitting the program codes to the processor;
the processor is configured to execute the power distribution network reliability prediction method in the foregoing method embodiments according to instructions in the program code.
The embodiment of the application also provides a computer-readable storage medium, which is used for storing program codes, and the program codes are used for executing the reliability prediction method of the power distribution network in the foregoing method embodiment.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be substantially implemented or contributed to by the prior art, or all or part of the technical solution may be embodied in a software product, which is stored in a storage medium and includes instructions for executing all or part of the steps of the method described in the embodiments of the present application through a computer device (which may be a personal computer, a server, or a network device). And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.
Claims (10)
1. A method for predicting the reliability of a power distribution network is characterized by comprising the following steps:
acquiring historical data of the number of households when a first fault of a plurality of district power supply offices before a month to be predicted is powered off;
performing characteristic extraction on the historical data of the number of the users when the first fault has power failure to obtain a first time sequence characteristic variable;
inputting the first time sequence characteristic variable into a long-short term memory neural network model to predict the number of households in the power failure, so as to obtain a predicted value of the number of households in the power failure of each district power supply office or the whole situation in the month to be predicted;
and calculating the reliability index prediction result of each district power supply office or the whole in the month to be predicted according to the number of the households in the prearranged period of the month to be predicted of each district power supply office or the whole in the month to be predicted and the predicted value of the number of the households in the fault power failure period.
2. The method for predicting the reliability of the power distribution network according to claim 1, wherein the step of performing feature extraction on the historical data of the number of users during the first fault power outage to obtain a first time sequence feature variable comprises the following steps:
extracting date features, area names, the number of days of fault power failure in a preset time period, the minimum number of households in fault power failure, the average number of households in fault power failure, the median number of households in fault power failure, the maximum number of households in fault power failure, the standard deviation of the number of households in fault power failure, the bias of the number of households in fault power failure, the average value of daily variation of the number of households in time, the maximum value of daily variation of the number of households in time, the minimum value of daily variation of the number of households in time, the standard deviation of daily variation of the number of households in time and/or the number of households in day fault power failure to obtain a first time sequence feature variable.
3. The method according to claim 1, wherein the characteristic extraction is performed on the historical data of the number of users during the first fault power outage to obtain a first time sequence characteristic variable, and then the method further comprises:
and preprocessing the first time sequence characteristic variable.
4. The method for predicting the reliability of the power distribution network according to claim 1, wherein the configuration process of the long-short term memory neural network model is as follows:
acquiring historical data of the number of households when a second fault of a plurality of regional power supply offices fails;
performing characteristic extraction on the historical data of the number of the users when the second fault has power failure to obtain a second time sequence characteristic variable;
and training the long-short term memory neural network through the second time sequence characteristic variable to obtain the long-short term memory neural network model.
5. A power distribution network reliability prediction device, comprising:
the acquisition unit is used for acquiring the historical data of the number of households when the power failure occurs in the first fault of the power supply bureaus of the plurality of areas before the month to be predicted;
the characteristic extraction unit is used for extracting the characteristics of the historical data of the number of the users when the first fault is in power failure to obtain a first time sequence characteristic variable;
the prediction unit is used for inputting the first time sequence characteristic variable into the long-short term memory neural network model to predict the number of households in the fault power failure, so as to obtain the predicted value of the number of households in the fault power failure of each district power supply office or the whole situation in the month to be predicted;
and the calculation unit is used for calculating the reliability index prediction result of each district power supply office or the whole in the month to be predicted according to the number of the households during the prearrangement of the month to be predicted by each district power supply office or the whole in the month to be predicted and the predicted value of the number of the households during the fault power failure.
6. The distribution network reliability prediction device according to claim 5, wherein the feature extraction unit is specifically configured to:
extracting date features, area names, the number of days of fault power failure in a preset time period, the minimum number of households in fault power failure, the average number of households in fault power failure, the median number of households in fault power failure, the maximum number of households in fault power failure, the standard deviation of the number of households in fault power failure, the bias of the number of households in fault power failure, the average value of daily variation of the number of households in time, the maximum value of daily variation of the number of households in time, the minimum value of daily variation of the number of households in time, the standard deviation of daily variation of the number of households in time and/or the number of households in day fault power failure to obtain a first time sequence feature variable.
7. The distribution network reliability prediction device of claim 5, further comprising:
and the preprocessing unit is used for preprocessing the first time sequence characteristic variable.
8. The distribution network reliability prediction device of claim 5, further comprising:
a configuration unit, specifically configured to:
acquiring historical data of the number of households when a second fault of a plurality of regional power supply offices fails;
performing characteristic extraction on the historical data of the number of the users when the second fault has power failure to obtain a second time sequence characteristic variable;
and training the long-short term memory neural network through the second time sequence characteristic variable to obtain the long-short term memory neural network model.
9. A power distribution network reliability prediction device, characterized in that the device comprises a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is configured to execute the method for power distribution network reliability prediction according to any one of claims 1 to 4 according to instructions in the program code.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium is configured to store program code for performing the method for reliability prediction of power distribution networks according to any of claims 1-4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110112126.8A CN112801365A (en) | 2021-01-27 | 2021-01-27 | Power distribution network reliability prediction method, device, equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110112126.8A CN112801365A (en) | 2021-01-27 | 2021-01-27 | Power distribution network reliability prediction method, device, equipment and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112801365A true CN112801365A (en) | 2021-05-14 |
Family
ID=75812214
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110112126.8A Pending CN112801365A (en) | 2021-01-27 | 2021-01-27 | Power distribution network reliability prediction method, device, equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112801365A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102722759A (en) * | 2012-05-17 | 2012-10-10 | 河海大学 | Method for predicting power supply reliability of power grid based on BP neural network |
| CN109659933A (en) * | 2018-12-20 | 2019-04-19 | 浙江工业大学 | A kind of prediction technique of power quality containing distributed power distribution network based on deep learning model |
| CN110929853A (en) * | 2019-12-11 | 2020-03-27 | 国网河南省电力公司洛阳供电公司 | Power distribution network line fault prediction method based on deep learning |
| CN111553535A (en) * | 2020-05-06 | 2020-08-18 | 东华大学 | Navigation reference method based on AE-LSTM-BO traffic flow prediction |
-
2021
- 2021-01-27 CN CN202110112126.8A patent/CN112801365A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102722759A (en) * | 2012-05-17 | 2012-10-10 | 河海大学 | Method for predicting power supply reliability of power grid based on BP neural network |
| CN109659933A (en) * | 2018-12-20 | 2019-04-19 | 浙江工业大学 | A kind of prediction technique of power quality containing distributed power distribution network based on deep learning model |
| CN110929853A (en) * | 2019-12-11 | 2020-03-27 | 国网河南省电力公司洛阳供电公司 | Power distribution network line fault prediction method based on deep learning |
| CN111553535A (en) * | 2020-05-06 | 2020-08-18 | 东华大学 | Navigation reference method based on AE-LSTM-BO traffic flow prediction |
Non-Patent Citations (1)
| Title |
|---|
| 张桦: "配电系统可靠性预测模型及其算法研究", 《工程科技Ⅱ辑》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Pao | Forecasting energy consumption in Taiwan using hybrid nonlinear models | |
| Zeng et al. | Hybrid probabilistic-possibilistic approach for capacity credit evaluation of demand response considering both exogenous and endogenous uncertainties | |
| CN109308571B (en) | Distribution line variable relation detection method | |
| CN112633316B (en) | Load prediction method and device based on boundary estimation theory | |
| Nazir et al. | Forecasting energy consumption demand of customers in smart grid using Temporal Fusion Transformer (TFT) | |
| CN113554466A (en) | Short-term power consumption prediction model construction method, prediction method and device | |
| Mokarram et al. | Net-load forecasting of renewable energy systems using multi-input LSTM fuzzy and discrete wavelet transform | |
| CN104834975A (en) | Power network load factor prediction method based on intelligent algorithm optimization combination | |
| CN111415068A (en) | Power distribution decision modeling method based on relevance of transformation measures and loss load index | |
| Zhao et al. | Short-term microgrid load probability density forecasting method based on k-means-deep learning quantile regression | |
| Kong et al. | Power load forecasting method based on demand response deviation correction | |
| CN116757465A (en) | Line risk assessment method and device based on double training weight distribution model | |
| Lao et al. | The optimized multivariate grey prediction model based on dynamic background value and its application | |
| Gao et al. | A two-layer SSA-XGBoost-MLR continuous multi-day peak load forecasting method based on hybrid aggregated two-phase decomposition | |
| Mishra et al. | Performance evaluation of prophet and STL-ETS methods for load forecasting | |
| CN112836871A (en) | Power distribution network reliability prediction method, device, equipment and medium | |
| Chen et al. | Residential short term load forecasting based on federated learning | |
| Grandón et al. | Electricity demand forecasting with hybrid classical statistical and machine learning algorithms: Case study of Ukraine | |
| CN112508254A (en) | Method for determining investment prediction data of transformer substation engineering project | |
| CN112465266A (en) | Bus load prediction accuracy analysis method and device and computer equipment | |
| Li et al. | Research on population development trend in Huizhou of China forecast based on optimal weighted combination method and fractional grey model | |
| Połap et al. | Energy consumption prediction model for smart homes via decentralized federated learning with LSTM | |
| Cao et al. | A hybrid electricity load prediction system based on weighted fuzzy time series and multi-objective differential evolution | |
| CN112801365A (en) | Power distribution network reliability prediction method, device, equipment and medium | |
| CN115759418A (en) | Adjustable capacity prediction method and system for electric power retail customer |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210514 |