CN113936192A - Power distribution network synchronous measurement missing data repairing method, terminal and storage medium - Google Patents

Power distribution network synchronous measurement missing data repairing method, terminal and storage medium Download PDF

Info

Publication number
CN113936192A
CN113936192A CN202111231057.9A CN202111231057A CN113936192A CN 113936192 A CN113936192 A CN 113936192A CN 202111231057 A CN202111231057 A CN 202111231057A CN 113936192 A CN113936192 A CN 113936192A
Authority
CN
China
Prior art keywords
data
matrix
repaired
missing
time
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
CN202111231057.9A
Other languages
Chinese (zh)
Inventor
董增波
容春艳
王守相
柴林杰
胡诗尧
赵倩宇
郭佳
林荣
申永鹏
王中亮
郝军魁
李军阔
高立坡
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.)
Tianjin University
State Grid Corp of China SGCC
Economic and Technological Research Institute of State Grid Hebei Electric Power Co Ltd
Original Assignee
Tianjin University
State Grid Corp of China SGCC
Economic and Technological Research Institute of State Grid Hebei Electric Power Co Ltd
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 Tianjin University, State Grid Corp of China SGCC, Economic and Technological Research Institute of State Grid Hebei Electric Power Co Ltd filed Critical Tianjin University
Priority to CN202111231057.9A priority Critical patent/CN113936192A/en
Publication of CN113936192A publication Critical patent/CN113936192A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • 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
    • G06N3/045—Combinations of 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
    • G06N3/08—Learning methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Evolutionary Computation (AREA)
  • Biomedical Technology (AREA)
  • Pure & Applied Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Algebra (AREA)
  • Databases & Information Systems (AREA)
  • Remote Monitoring And Control Of Power-Distribution Networks (AREA)

Abstract

The invention provides a method, a terminal and a storage medium for repairing missing data in synchronous measurement of a power distribution network. The method comprises the following steps: acquiring to-be-repaired synchronous measurement data of the power distribution network, and constructing a to-be-repaired data matrix, a missing position matrix and a time interval matrix according to the to-be-repaired synchronous measurement data; the missing position matrix is used for representing the position of missing data in the data matrix to be repaired, and the time interval matrix is used for representing the time interval between the data in the data matrix to be repaired and the last data which is not missing; inputting a data matrix to be repaired, a missing position matrix and a time interval matrix into a pre-trained data repairing model to obtain a repaired data matrix; the data restoration model comprises a time sequence interpolation model based on a bidirectional GRU network and a cross-sequence interpolation model based on a full-connection network, and the output of the time sequence interpolation model is the input of the cross-sequence interpolation model; and obtaining the repaired synchronous measurement data according to the repaired data matrix. The invention can improve the data repair precision.

Description

Power distribution network synchronous measurement missing data repairing method, terminal and storage medium
Technical Field
The invention relates to the technical field of situation awareness of intelligent power distribution networks, in particular to a method, a terminal and a storage medium for repairing missing data in synchronous measurement of a power distribution network.
Background
The power distribution network directly faces to users and provides power energy for various users. With the massive access of distributed power sources and electric automobiles, the load types of the power distribution network are diversified, and gradually evolve to a huge-dimension dynamic large system, so that the development of an intelligent power distribution network is not slow. Aiming at the improvement of observable and controllable requirements of an intelligent power distribution network, students at home and abroad develop a micro-phasor measurement Unit (mu PMU) of the power distribution network according to the idea of a main Phasor Measurement Unit (PMU). The mu PMU has good measurement data synchronism and high measurement precision, can contain phase information, provides more comprehensive and accurate information for monitoring, protecting and controlling the system, and is an important source for situation perception of the power system.
However, mass data may encounter system failure or external interference in many links of acquisition, measurement, transmission, storage and the like, resulting in data loss. Incomplete data can influence modeling analysis and finally influence decision making, so that the acquisition of complete and reliable measurement data is very important.
At present, an interpolation method, for example, a lagrangian interpolation method, is mainly used to repair missing data, however, the method has a better repair effect on the missing data of the system under a static condition, and when the system is under a dynamic condition, the data repair accuracy is low.
Disclosure of Invention
The embodiment of the invention provides a method, a terminal and a storage medium for repairing missing data of synchronous measurement of a power distribution network, and aims to solve the problem of low data repairing precision when a system is in a dynamic condition in the prior art.
In a first aspect, an embodiment of the present invention provides a method for repairing missing data in power distribution network synchronous measurement, including:
acquiring to-be-repaired synchronous measurement data of the power distribution network, and constructing a to-be-repaired data matrix, a missing position matrix and a time interval matrix according to the to-be-repaired synchronous measurement data; the missing position matrix is used for representing the position of missing data in the data matrix to be repaired, and the time interval matrix is used for representing the time interval between the data in the data matrix to be repaired and the last data which is not missing;
inputting a data matrix to be repaired, a missing position matrix and a time interval matrix into a pre-trained data repairing model to obtain a repaired data matrix; the data restoration model comprises a time sequence interpolation model based on a bidirectional GRU network and a cross-sequence interpolation model based on a full-connection network, and the output quantity of the time sequence interpolation model is the input quantity of the cross-sequence interpolation model;
and obtaining the repaired synchronous measurement data according to the repaired data matrix.
In a possible implementation manner, constructing a data matrix to be repaired, a missing position matrix and a time interval matrix according to the synchronous measurement data to be repaired includes:
dividing the synchronous measurement data to be repaired into a plurality of groups of target synchronous measurement data, wherein the ith group of target synchronous measurement data comprises measurement data measured from time i to time i + T-1, T is the number of measurement data contained in each group of target synchronous measurement data, i is more than or equal to 1 and less than or equal to N, and N is the number of groups of the plurality of groups of target synchronous measurement data;
constructing a plurality of data matrixes to be repaired, a plurality of missing position matrixes and a plurality of time interval matrixes according to the plurality of groups of target synchronous measurement data; the target synchronous measurement data, the data matrix to be repaired, the missing position matrix and the time interval matrix are in one-to-one correspondence.
In one possible implementation, the number of the repaired data matrixes is N;
obtaining the repaired synchronous measurement data according to the repaired data matrix, including:
and splicing all the data in the first repaired data matrix and the last row of data of the rest N-1 repaired data matrices according to a time sequence to obtain the repaired synchronous measurement data.
In a possible implementation manner, after constructing a plurality of data matrices to be repaired, a plurality of missing position matrices, and a plurality of time interval matrices according to a plurality of sets of target synchronous measurement data, the method further includes:
respectively converting the plurality of data matrixes to be repaired, the plurality of missing position matrixes and the plurality of time interval matrixes into a tensor format to obtain a data tensor to be repaired, a missing position tensor and a time interval tensor;
correspondingly, inputting the data matrix to be repaired, the missing position matrix and the time interval matrix into a data repairing model trained in advance to obtain a repaired data matrix, which comprises the following steps:
and inputting the data tensor to be repaired, the missing position tensor and the time interval tensor into a data repairing model trained in advance to obtain a repaired data matrix.
In one possible implementation, in the time interval matrix,
Figure BDA0003315945810000031
wherein the content of the first and second substances,
Figure BDA0003315945810000032
is an element in the time interval matrix, representing the time interval between the data measured by channel d at time t and the last non-missing data;
Figure BDA0003315945810000033
sta timestamp at time t; st-1Is the timestamp at time t-1;
Figure BDA0003315945810000034
is an element in the missing position matrix and represents whether the data measured by the channel d at the time t-1 is missing or not;
Figure BDA0003315945810000035
indicating that channel d has data measured at time t-1 missing,
Figure BDA0003315945810000036
indicating that the data measured for channel d at time t-1 is not missing.
In one possible implementation, the formula of the time series interpolation model based on the bidirectional GRU network is as follows:
Figure BDA0003315945810000037
Figure BDA0003315945810000038
Figure BDA0003315945810000039
Figure BDA00033159458100000310
Figure BDA00033159458100000311
Figure BDA00033159458100000312
Figure BDA00033159458100000313
Figure BDA00033159458100000314
Figure BDA00033159458100000315
wherein σ and tanh are activation functions;
Figure BDA00033159458100000316
is the number of the channel d of the data matrix to be repaired at the time t-1According to the value of the one or more parameters,
Figure BDA00033159458100000317
the data value of a channel d of the data matrix to be repaired at the moment t +1 is obtained;
Figure BDA00033159458100000318
and
Figure BDA00033159458100000319
the values of the hidden layer, the update gate, the new memory cell and the reset gate, respectively, for channel d propagating forward at time t;
Figure BDA00033159458100000320
and
Figure BDA00033159458100000321
the values of the hidden layer, the update gate, the new memory cell and the reset gate, respectively, for channel d counter-propagating at time t;
Figure BDA0003315945810000041
the output value of the time series interpolation model based on the bidirectional GRU network at time t for channel d,
Figure BDA0003315945810000042
Figure BDA0003315945810000043
respectively representing the element values of a data matrix to be repaired, a missing position matrix and a time interval matrix which are repaired by a time sequence interpolation model based on a bidirectional GRU network; wo、
Figure BDA0003315945810000044
And
Figure BDA0003315945810000045
are all weight parameters; bo、
Figure BDA0003315945810000046
And
Figure BDA0003315945810000047
are all bias parameters.
In one possible implementation, the formula of the cross-sequence interpolation model based on the fully-connected network is as follows:
Figure BDA0003315945810000048
ht＝φ(Uxt+Vzt+β)
wherein σ and φ are activation functions;
Figure BDA0003315945810000049
the output value of the cross-sequence interpolation model at the moment t based on the full-connection network is obtained; h istIs an intermediate parameter at time t; x is the number oftThe data value of the data matrix to be repaired at the moment t is obtained;
Figure BDA00033159458100000410
Figure BDA00033159458100000411
and mtRespectively representing the element values of a data matrix to be repaired and a missing position matrix which are repaired by a time sequence interpolation model based on a bidirectional GRU network; w, U and V are both weight parameters; both α and β are bias parameters.
In a possible implementation manner, before inputting the data matrix to be repaired, the missing position matrix, and the time interval matrix into a data repair model trained in advance to obtain a repaired data matrix, the method further includes:
acquiring a training sample set;
performing iterative training on a pre-established data restoration model according to a training sample set to obtain a trained data restoration model;
in each iterative training, evaluating the training result by adopting a root mean square error, and finishing the training when the root mean square error meets a preset condition to obtain a trained data restoration model;
root mean square error
Figure BDA00033159458100000412
Is calculated by the formula
Figure BDA00033159458100000413
Figure BDA00033159458100000414
Matrix output for data recovery model
Figure BDA00033159458100000415
An element of (1);
Figure BDA00033159458100000416
elements of the actual synchronization measurement data matrix x that do not contain missing data;
Figure BDA00033159458100000417
is an element of the missing position matrix; d is the number of channels; t is the time sequence length.
In a second aspect, an embodiment of the present invention provides a terminal, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor, when executing the computer program, implements the steps of the power distribution network synchronization measurement missing data repairing method according to the first aspect or any possible implementation manner of the first aspect.
In a third aspect, an embodiment of the present invention provides a computer-readable storage medium, where a computer program is stored, and when the computer program is executed by a processor, the method for repairing synchronization measurement missing data of a power distribution network according to the first aspect or any one of the possible implementation manners of the first aspect is implemented.
The embodiment of the invention provides a method, a terminal and a storage medium for repairing lost data in synchronous measurement of a power distribution network, wherein the time delay of synchronous measurement data of the power distribution network is considered by constructing a time interval matrix corresponding to a data matrix to be repaired; the data restoration method comprises the steps that missing data are restored through a data restoration model, the data restoration model comprises a time sequence interpolation model based on a bidirectional GRU network and a cross sequence interpolation model based on a full-connection network, the output quantity of the time sequence interpolation model is the input quantity of the cross sequence interpolation model, missing data of a copy channel can be restored according to the time sequence relation of measured data of each channel through the time sequence interpolation model based on the bidirectional GRU network, the output quantity of the time sequence interpolation model is used as the input quantity of the cross sequence interpolation model based on the full-connection network, data can be further restored according to the space correlation between different channel data at the same moment, the space-time correlation information of the data can be fully mined, and the data restoration precision can be improved even when the system is in a dynamic condition.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the embodiments or the prior art descriptions will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive exercise.
Fig. 1 is a flowchart illustrating an implementation of a method for repairing missing data in synchronous measurement of a power distribution network according to an embodiment of the present invention;
fig. 2 is a schematic diagram of splicing a repaired data matrix according to an embodiment of the present invention;
FIG. 3 is a schematic structural diagram of a data recovery model provided by an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a cross-sequence interpolation model based on a fully-connected network according to an embodiment of the present invention;
FIG. 5 is a graph illustrating the repairing effect of the voltage amplitude of the time sequence of a certain channel when the data loss rate is 20% according to the embodiment of the present invention;
FIG. 6 is a graph showing the repairing effect of the voltage amplitude of the time sequence of a certain channel when the data loss rate is 30% according to the embodiment of the present invention;
fig. 7 is a graph illustrating the repairing effect of the voltage amplitudes of different channels at a certain time when the data loss rate is 20% according to the embodiment of the present invention;
fig. 8 is a graph illustrating the repairing effect of the voltage amplitudes of different channels at a certain time when the data loss rate is 30% according to the embodiment of the present invention;
fig. 9 is a schematic diagram of absolute error distribution of voltage amplitude restoration under the condition that the data loss rate is 20% according to the embodiment of the present invention;
fig. 10 is a schematic structural diagram of a power distribution network synchronous measurement missing data recovery apparatus according to an embodiment of the present invention;
fig. 11 is a schematic diagram of a terminal according to an embodiment of the present invention.
Detailed Description
In the following description, for purposes of explanation and not limitation, specific details are set forth, such as particular system structures, techniques, etc. in order to provide a thorough understanding of the embodiments of the invention. It will be apparent, however, to one skilled in the art that the present invention may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.
In order to make the objects, technical solutions and advantages of the present invention more apparent, the following description is made by way of specific embodiments with reference to the accompanying drawings.
Referring to fig. 1, it shows an implementation flowchart of the method for repairing missing data in synchronous measurement of a power distribution network according to the embodiment of the present invention, where an execution subject of the method for repairing missing data in synchronous measurement of a power distribution network may be a terminal.
Referring to fig. 1, the method for repairing the loss data in the synchronous measurement of the power distribution network is detailed as follows:
in S101, acquiring to-be-repaired synchronous measurement data of the power distribution network, and constructing a to-be-repaired data matrix, a missing position matrix and a time interval matrix according to the to-be-repaired synchronous measurement data; the missing position matrix is used for representing the position of missing data in the data matrix to be repaired, and the time interval matrix is used for representing the time interval between the missing data and the last non-missing data in the data matrix to be repaired.
The embodiment can synchronously measure measurement data, such as voltage data, current data and the like, of a plurality of nodes of the power distribution network through the mu PMU. Each node acts as a channel. The to-be-repaired synchronous measurement data of the power distribution network may include data obtained by synchronous measurement of a plurality of channels at a plurality of times.
In this embodiment, a data matrix to be repaired may be constructed according to the to-be-repaired synchronous measurement data of the power distribution network, a missing position matrix may be constructed according to a position where missing data in the data matrix to be repaired is located, and a time interval matrix may be constructed according to a time interval between data at each time in each channel in the data matrix to be repaired and last non-missing data of the channel. The last data of the channel is the data with the latest time interval.
The missing position matrix may be a two-dimensional mask matrix, and an element value of 0 at a corresponding position indicates that data is missing, and an element value of 1 at a corresponding position indicates that data is not missing.
In S102, inputting a data matrix to be repaired, a missing position matrix and a time interval matrix into a pre-trained data repair model to obtain a repaired data matrix; the data restoration model comprises a time sequence interpolation model based on a bidirectional GRU (gated recurrent unit) network and a cross-sequence interpolation model based on a full-connection network, and the output quantity of the time sequence interpolation model is the input quantity of the cross-sequence interpolation model.
In this embodiment, a data matrix to be repaired is subjected to data repair by constructing a data repair model. And inputting the data matrix to be repaired, the missing position matrix and the time interval matrix into a pre-trained data repairing model to obtain the data matrix after repairing the data matrix to be repaired.
The data restoration model can be a multidirectional cyclic neural network model, and the model comprises a time sequence interpolation model based on a bidirectional GRU network and a cross-sequence interpolation model based on a full-connection network, wherein the time sequence interpolation model based on the bidirectional GRU network is used for time sequence interpolation, and the cross-sequence interpolation model based on the full-connection network is used for cross-channel data interpolation. In other words, in the model, the bidirectional GRU is used as its interpolation block, missing data of the duplicate channel can be repaired according to the time sequence relation of the measured data of each channel, the full connection layer is used as the interpolation block, the interpolation block takes the output of the interpolation block as the input, the missing data is further repaired according to the spatial correlation between different channel data at the same time, and the reconstruction accuracy is improved. The data restoration model can fully mine the time-space correlation information of the data, and the data restoration precision is obviously improved.
The long and short term memory network and the GRU are both variants of the RNN (Recurrent Neural Networks) after optimization, and the RNN with the door mechanism can solve the problems of gradient disappearance and gradient explosion of the traditional RNN and has better performance in practical application. And compared with a long-term and short-term memory network, the GRU is simpler, has less parameters and high convergence speed, and can accelerate the experimental process. Therefore, the present embodiment selects bidirectional GRUs as the interpolation blocks.
In S103, the repaired synchronous measurement data is obtained according to the repaired data matrix.
In this embodiment, the reverse operation may be performed by the method of constructing the data matrix to be repaired according to the synchronous measurement data to be repaired in S101, so that the repaired synchronous measurement data may be obtained according to the repaired data matrix. The repaired synchronous measurement data is the data after the repair of the to-be-repaired synchronous measurement data of the power distribution network is completed.
According to the embodiment of the invention, the time delay of the synchronous measurement data of the power distribution network is considered by constructing the time interval matrix corresponding to the data matrix to be repaired; the missing data is repaired through a data repairing model, the data repairing model comprises a time sequence interpolation model based on a bidirectional GRU network and a cross sequence interpolation model based on a full connection network, the output quantity of the time sequence interpolation model is the input quantity of the cross sequence interpolation model, through the time sequence interpolation model based on the bidirectional GRU network, the missing data of the duplicate channels can be corrected according to the time sequence relation of the measured data of each channel, the output quantity of the time sequence interpolation model is used as the input quantity based on the cross sequence interpolation model of the full-connection network, the data can be further repaired according to the spatial correlation among different channel data at the same moment, the time-space correlation information of the data can be fully mined, the data repairing precision can be improved when the system is in a static condition, and the data repairing precision can also be improved when the system is in a dynamic condition.
In some embodiments, the "constructing a data matrix to be repaired, a missing position matrix and a time interval matrix according to the synchronous measurement data to be repaired" in S101 may include:
dividing the synchronous measurement data to be repaired into a plurality of groups of target synchronous measurement data, wherein the ith group of target synchronous measurement data comprises measurement data measured from time i to time i + T-1, T is the number of measurement data contained in each group of target synchronous measurement data, i is more than or equal to 1 and less than or equal to N, and N is the number of groups of the plurality of groups of target synchronous measurement data;
constructing a plurality of data matrixes to be repaired, a plurality of missing position matrixes and a plurality of time interval matrixes according to the plurality of groups of target synchronous measurement data; the target synchronous measurement data, the data matrix to be repaired, the missing position matrix and the time interval matrix are in one-to-one correspondence.
In this embodiment, the grouped synchronous measurement data to be repaired is referred to as target synchronous measurement data. And dividing the synchronous measurement data to be repaired into N groups of target synchronous measurement data. Each time, a piece of synchronous data measured by D channels may be obtained, and thus, each set of target synchronous measurement data may include T pieces of data, and each matrix of data to be repaired may be a matrix of T × D, i.e., T rows and D columns. Where D is the number of channels, and T is the number of measurement data included in each set of target synchronous measurement data, where the included measurement data is measurement data measured at a time, and may also represent a time sequence length, that is, a sequence length of a time sequence. The values of T and D can be set according to actual requirements. In one possible implementation, T may be 12.
For each data matrix to be repaired, a missing position matrix may be constructed according to the position of missing data in the matrix, for example, in the missing position matrix, the value of the missing data position is 0, and the value of the position of data that is not missing is 1, and of course, the two values may be opposite, and are not limited specifically here.
When the missing data is delayed irregularly, a time interval matrix needs to be constructed, and the time interval matrix is used for representing the time interval between the acquisition time of each piece of data of each channel and the acquisition time of the last piece of un-missing data of the channel. The time interval may be represented by a number of time intervals, for example, the number of time intervals between time 1 and time 3 is 2, the number of time intervals between time 2 and time 5 is 5, and so on.
The deletion location matrix and the time interval matrix may both be matrices of T x D.
In a possible implementation manner, before the dividing the synchronous measurement data to be repaired into multiple sets of target synchronous measurement data, the method for repairing the synchronous measurement missing data of the power distribution network further includes:
and performing data normalization on the synchronous measurement data to be repaired by adopting a maximum and minimum normalization method.
In this embodiment, the normalized data may be subdivided into a plurality of sets of target synchronous metrology data.
In some embodiments, the number of repaired data matrices is N;
the S103 may include:
and splicing all the data in the first repaired data matrix and the last row of data of the rest N-1 repaired data matrices according to a time sequence to obtain the repaired synchronous measurement data.
The scale of the repaired data matrix is the same as that of the data matrix to be repaired, the quantity of the repaired data matrix is the same, and the repaired data matrix and the data matrix are in one-to-one correspondence. The data repaired by the data repairing model are still N matrixes with the time series length of T. Considering the memory function of the recurrent neural network, the last line of data of each matrix is the most accurate to repair, so that when the data are grouped, the i-th group of target synchronous measurement data includes measurement data measured from time i to time i + T-1, and after the repair is finished, all data in the first repaired data matrix and the last line of data in the remaining N-1 repaired data matrices are spliced according to the time sequence to obtain repaired synchronous measurement data, which can be referred to fig. 2 specifically. The repaired data matrixes are also arranged according to a time sequence, namely the first repaired data matrix is a matrix obtained after the data matrix to be repaired corresponding to the 1 st group of target synchronous measurement data is repaired, and the rest N-1 repaired data matrixes are all the repaired data matrixes except the first repaired data matrix.
In some embodiments, after S101, the method for repairing the power distribution grid synchronization measurement missing data further includes:
respectively converting the plurality of data matrixes to be repaired, the plurality of missing position matrixes and the plurality of time interval matrixes into a tensor format to obtain a data tensor to be repaired, a missing position tensor and a time interval tensor;
accordingly, the S102 may include:
and inputting the data tensor to be repaired, the missing position tensor and the time interval tensor into a data repairing model trained in advance to obtain a repaired data matrix.
In this embodiment, the multiple data matrixes to be repaired, the multiple missing position matrixes and the multiple time interval matrixes may be respectively converted into a three-dimensional tensor format, so as to obtain a data tensor to be repaired, a missing position tensor and a time interval tensor. The repaired data matrix may also be a three-dimensional matrix in a tensor format.
In some embodiments, the time interval matrix may include, in the time interval matrix,
Figure BDA0003315945810000111
wherein the content of the first and second substances,
Figure BDA0003315945810000112
is an element in the time interval matrix, representing the time interval between the data measured by channel d at time t and the last non-missing data;
Figure BDA0003315945810000113
sta timestamp at time t; st-1Is the timestamp at time t-1;
Figure BDA0003315945810000114
is an element in the missing position matrix and represents whether the data measured by the channel d at the time t-1 is missing or not;
Figure BDA0003315945810000115
indicating that channel d has data measured at time t-1 missing,
Figure BDA0003315945810000116
indicating that the data measured for channel d at time t-1 is not missing.
Figure BDA0003315945810000117
Representing the time interval between the data measured by channel d at time t-1 and the last non-missing data.
For each time interval matrix, the corresponding time interval matrix can be obtained by recursion by adopting the formula. In each time interval matrix, T takes a value from 1 to T.
In some embodiments, the formula of the time series interpolation model based on the bidirectional GRU network is:
Figure BDA0003315945810000118
Figure BDA0003315945810000119
Figure BDA00033159458100001110
Figure BDA00033159458100001111
Figure BDA00033159458100001112
Figure BDA00033159458100001113
Figure BDA00033159458100001114
Figure BDA00033159458100001115
Figure BDA00033159458100001116
wherein σ and tanh are activation functions;
Figure BDA00033159458100001117
for the data value of channel d of the data matrix to be repaired at time t-1,
Figure BDA00033159458100001118
the data value of a channel d of the data matrix to be repaired at the moment t +1 is obtained;
Figure BDA00033159458100001119
and
Figure BDA00033159458100001120
the values of the hidden layer, the update gate, the new memory cell and the reset gate, respectively, for channel d propagating forward at time t;
Figure BDA00033159458100001121
and
Figure BDA00033159458100001122
respectively, the channel d is inverted at the time tValues to the propagated hidden layer, the update gate, the new memory cell, and the reset gate;
Figure BDA00033159458100001123
the output value of the time series interpolation model based on the bidirectional GRU network at time t for channel d,
Figure BDA00033159458100001124
Figure BDA00033159458100001125
respectively representing the element values of a data matrix to be repaired, a missing position matrix and a time interval matrix which are repaired by a time sequence interpolation model based on a bidirectional GRU network; wo、
Figure BDA0003315945810000121
And
Figure BDA0003315945810000122
are all weight parameters; bo、
Figure BDA0003315945810000123
And
Figure BDA0003315945810000124
are all bias parameters.
The new memory unit may be a memory unit in an existing GRU, or may be a modified memory unit, and is not particularly limited herein.
The values of the missing location matrix and the time interval matrix may be constant during data repair by the data repair model.
In some embodiments, the formula of the cross-sequence interpolation model based on the fully-connected network is:
Figure BDA0003315945810000125
ht＝φ(Uxt+Vzt+β)
wherein σ and φ are activation functions;
Figure BDA0003315945810000126
the output value of the cross-sequence interpolation model at the moment t based on the full-connection network is obtained; h istIs an intermediate parameter at time t; x is the number oftThe data value of the data matrix to be repaired at the moment t is obtained;
Figure BDA0003315945810000127
Figure BDA0003315945810000128
and mtRespectively representing the element values of a data matrix to be repaired and a missing position matrix which are repaired by a time sequence interpolation model based on a bidirectional GRU network; w, U and V are both weight parameters; both α and β are bias parameters.
Referring to FIGS. 3 and 4, FC is the full link layer, Xt-1、Xt、Xt+1All are inputs of a data recovery model;
Figure BDA0003315945810000129
all are the output of a time series interpolation model;
Figure BDA00033159458100001210
the data restoration method is the output of a cross-sequence interpolation model and the output of a data restoration model.
In fig. 3, a dashed box is a time series interpolation model based on a bidirectional GRU network. The model processes input sequences sequentially in sequence and in reverse order in a time dimension, and context information of the current time is contained in an output node of each time step. The bidirectional GRU in this embodiment is different from the conventional bidirectional RNN in that the time of its input to the hidden layer is delayed in the forward propagation and advanced in the backward propagation, for the purpose of avoiding the future
Figure BDA00033159458100001211
For estimating
Figure BDA00033159458100001212
See fig. 4, which is a schematic structural diagram of a cross-sequence interpolation model based on a fully-connected network. As shown, the input data not only propagates in the time dimension, but also passes between different channel data at the same time. In the embodiment, the full-connection interpolation part is arranged after the time-series interpolation part of the bidirectional GRU network, so that the data restoration precision can be improved.
In some embodiments, before inputting the data matrix to be repaired, the missing position matrix, and the time interval matrix into the pre-trained data repair model to obtain the repaired data matrix, the method further includes:
acquiring a training sample set;
performing iterative training on a pre-established data restoration model according to a training sample set to obtain a trained data restoration model;
in each iterative training, evaluating the training result by adopting Root Mean Square Error (RMSE), and finishing the training when the Root Mean square Error meets a preset condition to obtain a trained data restoration model;
root mean square error
Figure BDA0003315945810000131
Is calculated by the formula
Figure BDA0003315945810000132
Figure BDA0003315945810000133
Matrix output for data recovery model
Figure BDA0003315945810000134
An element of (1);
Figure BDA0003315945810000135
elements of the actual synchronization measurement data matrix x that do not contain missing data;
Figure BDA0003315945810000136
is an element of the missing position matrix; d is the number of channels; t is the time sequence length.
The training sample set may include a plurality of training samples, each of which includes complete non-missing metrology data and metrology data including missing data corresponding to the complete non-missing metrology data. Random missing proportion can be set for the complete non-missing measurement data to obtain the corresponding measurement data containing the missing data.
The preset condition includes that the root mean square error is smaller than a preset value, or the difference value of the root mean square errors obtained by two adjacent training is within a preset range, and the like.
In one possible implementation, the condition for the end of the iterative training may be that the number of iterations is greater than a preset number.
The preset value, the preset range and the preset times can be set according to actual requirements, and are not particularly limited herein.
In this example, the accuracy of the repair data was evaluated using RMSE as a performance evaluation index. And updating the optimization parameters one by one, and storing the trained data restoration model so as to be directly applied in the subsequent data restoration.
The purpose of this embodiment is to minimize the repair error of the missing data, so the calculation formula of the root mean square error is adopted to perform the calculation, and the non-missing data is not involved in the evaluation.
In one possible implementation, an objective function that minimizes the repair error may be constructed, and after solving the objective function, the training may be stopped. The objective function is:
Figure BDA0003315945810000141
Figure BDA0003315945810000142
in the above formula, phidThe function represents a bi-directional GRU network,the Ψ function represents a full connectivity layer network.
In a possible implementation manner, a time sequence interpolation model based on a bidirectional GRU network may be trained, after the model is trained, the model is stored, then, based on the trained model, a cross-sequence interpolation model based on a full-connection network is trained, and after the training is finished, a repaired data repair model is obtained.
In this embodiment, a multidirectional recurrent neural network model composed of a bidirectional GRU network construction interpolation block and a full-connection network construction cross-channel data interpolation block is established in consideration of the time-space correlation of mu PMU measurement data; a time interval matrix is used for storing time delay information between two measurements, and time delay characteristics are stored, so that measured data are not required to be strictly synchronous; the structure of the whole multidirectional cyclic neural network model utilizes the three-dimensional characteristics of data; and restoring all data of the first repaired data matrix and the last row of data of the rest N-1 repaired data matrices into data for practical application.
In a specific application scenario, voltage amplitude in synchronous phasor measurement data of a power distribution network is used for testing, initial amplitude data is in a per unit value form, 0 is randomly set under a certain probability through a missing position matrix, and the whole data set is set as to-be-repaired synchronous measurement data under a given data missing rate.
Fig. 5 and fig. 6 respectively show the repairing effect graphs of the voltage amplitudes of the same channel at different data loss rates, fig. 7 and fig. 8 respectively show the repairing effect graphs of the voltage amplitudes of different channels at the same time at different data loss rates, which show the repairing effect of the method provided by the embodiment on time and space, wherein ∘ represents an actual value, and x is a value after repairing. The violin diagram in fig. 9 is a distribution of absolute errors of voltage amplitude repair values under the condition that the data loss rate is 20%, and it can be seen from fig. 9 that the absolute errors are intensively distributed near the median, and under the condition that the data loss rate is 20%, the repair result still maintains very high precision, and the median of the absolute errors is only 0.0006/p.u., which verifies the validity of the method provided by the embodiment.
It should be understood that, the sequence numbers of the steps in the foregoing embodiments do not imply an execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.
The following are embodiments of the apparatus of the invention, reference being made to the corresponding method embodiments described above for details which are not described in detail therein.
Fig. 10 is a schematic structural diagram of a power distribution network synchronization measurement missing data recovery apparatus according to an embodiment of the present invention, and for convenience of description, only the relevant portions of the embodiment of the present invention are shown, which is detailed as follows:
as shown in fig. 10, the distribution network synchronization measurement missing data recovery apparatus 100 includes: a matrix building module 101, a repair module 102 and a restore module 103.
The matrix construction module 101 is configured to acquire to-be-repaired synchronous measurement data of the power distribution network, and construct a to-be-repaired data matrix, a missing position matrix and a time interval matrix according to the to-be-repaired synchronous measurement data; the missing position matrix is used for representing the position of missing data in the data matrix to be repaired, and the time interval matrix is used for representing the time interval between the data in the data matrix to be repaired and the last data which is not missing;
the repairing module 102 is configured to input the data matrix to be repaired, the missing position matrix and the time interval matrix into a data repairing model trained in advance, so as to obtain a repaired data matrix; the data restoration model comprises a time sequence interpolation model based on a bidirectional GRU network and a cross-sequence interpolation model based on a full-connection network, and the output quantity of the time sequence interpolation model is the input quantity of the cross-sequence interpolation model;
and the restoring module 103 is configured to obtain the repaired synchronous measurement data according to the repaired data matrix.
In a possible implementation manner, the matrix building module 101 is specifically configured to:
dividing the synchronous measurement data to be repaired into a plurality of groups of target synchronous measurement data, wherein the ith group of target synchronous measurement data comprises measurement data measured from time i to time i + T-1, T is the number of measurement data contained in each group of target synchronous measurement data, i is more than or equal to 1 and less than or equal to N, and N is the number of groups of the plurality of groups of target synchronous measurement data;
constructing a plurality of data matrixes to be repaired, a plurality of missing position matrixes and a plurality of time interval matrixes according to the plurality of groups of target synchronous measurement data; the target synchronous measurement data, the data matrix to be repaired, the missing position matrix and the time interval matrix are in one-to-one correspondence.
In one possible implementation, the number of the repaired data matrixes is N;
the restoring module 103 is specifically configured to:
and splicing all the data in the first repaired data matrix and the last row of data of the rest N-1 repaired data matrices according to a time sequence to obtain the repaired synchronous measurement data.
In a possible implementation manner, the distribution network synchronization measurement missing data restoration apparatus 100 further includes a format conversion module.
The format conversion module is used for respectively converting the multiple data matrixes to be repaired, the multiple missing position matrixes and the multiple time interval matrixes into a tensor format to obtain a data tensor to be repaired, a missing position tensor and a time interval tensor;
accordingly, the repair module 102 is specifically configured to:
and inputting the data tensor to be repaired, the missing position tensor and the time interval tensor into a data repairing model trained in advance to obtain a repaired data matrix.
In one possible implementation, in the time interval matrix,
Figure BDA0003315945810000161
wherein the content of the first and second substances,
Figure BDA0003315945810000162
is an element in the time interval matrix, representing the time interval between the data measured by channel d at time t and the last non-missing data;
Figure BDA0003315945810000163
sta timestamp at time t; st-1Is the timestamp at time t-1;
Figure BDA0003315945810000164
is an element in the missing position matrix and represents whether the data measured by the channel d at the time t-1 is missing or not;
Figure BDA0003315945810000165
indicating that channel d has data measured at time t-1 missing,
Figure BDA0003315945810000166
indicating that the data measured for channel d at time t-1 is not missing.
In one possible implementation, the formula of the time series interpolation model based on the bidirectional GRU network is as follows:
Figure BDA0003315945810000171
Figure BDA0003315945810000172
Figure BDA0003315945810000173
Figure BDA0003315945810000174
Figure BDA0003315945810000175
Figure BDA0003315945810000176
Figure BDA0003315945810000177
Figure BDA0003315945810000178
Figure BDA0003315945810000179
wherein σ and tanh are activation functions;
Figure BDA00033159458100001710
for the data value of channel d of the data matrix to be repaired at time t-1,
Figure BDA00033159458100001711
the data value of a channel d of the data matrix to be repaired at the moment t +1 is obtained;
Figure BDA00033159458100001712
and
Figure BDA00033159458100001713
the values of the hidden layer, the update gate, the new memory cell and the reset gate, respectively, for channel d propagating forward at time t;
Figure BDA00033159458100001714
and
Figure BDA00033159458100001715
the values of the hidden layer, the update gate, the new memory cell and the reset gate, respectively, for channel d counter-propagating at time t;
Figure BDA00033159458100001716
the output value of the time series interpolation model based on the bidirectional GRU network at time t for channel d,
Figure BDA00033159458100001717
Figure BDA00033159458100001718
respectively representing the element values of a data matrix to be repaired, a missing position matrix and a time interval matrix which are repaired by a time sequence interpolation model based on a bidirectional GRU network; wo、
Figure BDA00033159458100001719
And
Figure BDA00033159458100001720
are all weight parameters; bo、
Figure BDA00033159458100001721
And
Figure BDA00033159458100001722
are all bias parameters.
In one possible implementation, the formula of the cross-sequence interpolation model based on the fully-connected network is as follows:
Figure BDA00033159458100001723
ht＝φ(Uxt+Vzt+β)
wherein σ and φ are activation functions;
Figure BDA00033159458100001724
the output value of the cross-sequence interpolation model at the moment t based on the full-connection network is obtained; h istIs an intermediate parameter at time t; x is the number oftThe data value of the data matrix to be repaired at the moment t is obtained;
Figure BDA00033159458100001725
Figure BDA00033159458100001726
and mtRespectively, time of passing through the bidirectional GRU-based networkElement values of a data matrix to be repaired and a missing position matrix after the restoration of the sequence interpolation model; w, U and V are both weight parameters; both α and β are bias parameters.
In a possible implementation manner, the distribution network synchronization measurement missing data restoration apparatus 100 further includes a training module.
The training module is used for:
acquiring a training sample set;
performing iterative training on a pre-established data restoration model according to a training sample set to obtain a trained data restoration model;
in each iterative training, evaluating the training result by adopting a root mean square error, and finishing the training when the root mean square error meets a preset condition to obtain a trained data restoration model;
root mean square error
Figure BDA0003315945810000181
Is calculated by the formula
Figure BDA0003315945810000182
Figure BDA0003315945810000183
Matrix output for data recovery model
Figure BDA0003315945810000184
An element of (1);
Figure BDA0003315945810000185
elements of the actual synchronization measurement data matrix x that do not contain missing data;
Figure BDA0003315945810000186
is an element of the missing position matrix; d is the number of channels; t is the time sequence length.
Fig. 11 is a schematic diagram of a terminal according to an embodiment of the present invention. As shown in fig. 11, the terminal 11 of this embodiment includes: a processor 110, a memory 111 and a computer program 112 stored in said memory 111 and executable on said processor 110. The processor 110 executes the computer program 112 to implement the steps in the embodiments of the power distribution network synchronization measurement missing data repairing method, such as S101 to S103 shown in fig. 1. Alternatively, the processor 110, when executing the computer program 112, implements the functions of each module/unit in the above-mentioned device embodiments, for example, the functions of the modules/units 101 to 103 shown in fig. 10.
Illustratively, the computer program 112 may be partitioned into one or more modules/units that are stored in the memory 111 and executed by the processor 110 to implement the present invention. The one or more modules/units may be a series of computer program instruction segments capable of performing specific functions, which are used to describe the execution process of the computer program 112 in the terminal 11. For example, the computer program 112 may be divided into the modules/units 101 to 103 shown in fig. 10.
The terminal 11 may be a desktop computer, a notebook, a palm computer, a cloud server, or other computing devices. The terminal 11 may include, but is not limited to, a processor 110, a memory 111. Those skilled in the art will appreciate that fig. 11 is merely an example of a terminal 11 and does not constitute a limitation of terminal 11 and may include more or less components than those shown, or combine certain components, or different components, e.g., the terminal may also include input-output devices, network access devices, buses, etc.
The Processor 110 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, discrete hardware component, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The storage 111 may be an internal storage unit of the terminal 11, such as a hard disk or a memory of the terminal 11. The memory 111 may also be an external storage device of the terminal 11, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), and the like, which are provided on the terminal 11. Further, the memory 111 may also include both an internal storage unit and an external storage device of the terminal 11. The memory 111 is used for storing the computer program and other programs and data required by the terminal. The memory 111 may also be used to temporarily store data that has been output or is to be output.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units and modules is illustrated, and in practical applications, the above-mentioned function distribution may be performed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-mentioned functions. Each functional unit and module in the embodiments may be integrated in one processing unit, or each unit may exist alone physically, or two or more units are integrated in one unit, and the integrated unit may be implemented in a form of hardware, or in a form of software functional unit. In addition, specific names of the functional units and modules are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present application. The specific working processes of the units and modules in the system may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and reference may be made to the related descriptions of other embodiments for parts that are not described or illustrated in a certain embodiment.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
In the embodiments provided in the present invention, it should be understood that the disclosed apparatus/terminal and method may be implemented in other ways. For example, the above-described apparatus/terminal embodiments are merely illustrative, and for example, the division of the modules or units is only one logical division, and there may be other divisions when actually implemented, for example, a plurality of units or components may be combined or may be 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 invention 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 modules/units, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the method according to the embodiments of the present invention may also be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the computer program may implement the steps of the embodiments of the method for repairing missing data in synchronous measurement of a distribution network. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, and the like. It should be noted that the computer readable medium may contain other components which may be suitably increased or decreased as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer readable media which may not include electrical carrier signals and telecommunications signals in accordance with legislation and patent practice.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will 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; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present invention, and are intended to be included within the scope of the present invention.

Claims (10)

1. A method for repairing loss data in synchronous measurement of a power distribution network is characterized by comprising the following steps:
acquiring to-be-repaired synchronous measurement data of the power distribution network, and constructing a to-be-repaired data matrix, a missing position matrix and a time interval matrix according to the to-be-repaired synchronous measurement data; the missing position matrix is used for representing the position of missing data in the data matrix to be repaired, and the time interval matrix is used for representing the time interval between the data in the data matrix to be repaired and the last data which is not missing;
inputting the data matrix to be repaired, the missing position matrix and the time interval matrix into a pre-trained data repairing model to obtain a repaired data matrix; the data restoration model comprises a time sequence interpolation model based on a bidirectional GRU network and a cross sequence interpolation model based on a full connection network, and the output quantity of the time sequence interpolation model is the input quantity of the cross sequence interpolation model;
and obtaining the repaired synchronous measurement data according to the repaired data matrix.
2. The method for repairing the missing data in the synchronous measurement of the power distribution network according to claim 1, wherein the constructing a data matrix to be repaired, a missing position matrix and a time interval matrix according to the synchronous measurement data to be repaired includes:
dividing the synchronous measurement data to be repaired into a plurality of groups of target synchronous measurement data, wherein the ith group of target synchronous measurement data comprises measurement data measured from time i to time i + T-1, T is the number of measurement data contained in each group of target synchronous measurement data, i is more than or equal to 1 and is less than or equal to N, and N is the group number of the plurality of groups of target synchronous measurement data;
constructing a plurality of data matrixes to be repaired, a plurality of missing position matrixes and a plurality of time interval matrixes according to the plurality of groups of target synchronous measurement data; the target synchronous measurement data, the data matrix to be repaired, the missing position matrix and the time interval matrix are in one-to-one correspondence.
3. The method according to claim 2, wherein the number of the repaired data matrices is N;
the obtaining of the repaired synchronous measurement data according to the repaired data matrix includes:
and splicing all the data in the first repaired data matrix and the last row of data of the rest N-1 repaired data matrices according to a time sequence to obtain the repaired synchronous measurement data.
4. The method for repairing the missing data in the synchronous measurement of the power distribution network according to claim 2, wherein after the step of constructing a plurality of data matrices to be repaired, a plurality of missing position matrices and a plurality of time interval matrices according to the plurality of sets of target synchronous measurement data, the method further comprises:
respectively converting the plurality of data matrixes to be repaired, the plurality of missing position matrixes and the plurality of time interval matrixes into a tensor format to obtain a data tensor to be repaired, a missing position tensor and a time interval tensor;
correspondingly, the inputting the data matrix to be repaired, the missing position matrix and the time interval matrix into a pre-trained data repair model to obtain a repaired data matrix includes:
and inputting the data tensor to be repaired, the missing position tensor and the time interval tensor into a data repairing model trained in advance to obtain a repaired data matrix.
5. The method as claimed in claim 1, wherein in the time interval matrix,
Figure FDA0003315945800000021
wherein the content of the first and second substances,
Figure FDA0003315945800000022
is an element in the time interval matrix, representing the time interval between the data measured by channel d at time t and the last non-missing data;
Figure FDA0003315945800000023
sta timestamp at time t; st-1Is the timestamp at time t-1;
Figure FDA0003315945800000024
is absent ofAn element in the position matrix, which indicates whether the data measured by the channel d at the time t-1 is missing;
Figure FDA0003315945800000025
indicating that channel d has data measured at time t-1 missing,
Figure FDA0003315945800000026
indicating that the data measured for channel d at time t-1 is not missing.
6. The method for repairing synchronization measurement missing data in a power distribution network according to claim 1, wherein the formula of the time series interpolation model based on the bidirectional GRU network is as follows:
Figure FDA0003315945800000031
Figure FDA0003315945800000032
Figure FDA0003315945800000033
Figure FDA0003315945800000034
Figure FDA0003315945800000035
Figure FDA0003315945800000036
Figure FDA0003315945800000037
Figure FDA0003315945800000038
Figure FDA0003315945800000039
wherein σ and tanh are activation functions;
Figure FDA00033159458000000310
for the data value of channel d of the data matrix to be repaired at time t-1,
Figure FDA00033159458000000311
the data value of the channel d of the data matrix to be repaired at the time t +1 is obtained;
Figure FDA00033159458000000312
Figure FDA00033159458000000313
and
Figure FDA00033159458000000314
the values of the hidden layer, the update gate, the new memory cell and the reset gate, respectively, for channel d propagating forward at time t;
Figure FDA00033159458000000315
and
Figure FDA00033159458000000316
the values of the hidden layer, the update gate, the new memory cell and the reset gate, respectively, for channel d counter-propagating at time t;
Figure FDA00033159458000000317
the output value of the time series interpolation model based on the bidirectional GRU network at time t for channel d,
Figure FDA00033159458000000318
respectively is the element value of the data matrix to be repaired, the missing position matrix and the time interval matrix which are repaired by the time sequence interpolation model based on the bidirectional GRU network
Figure FDA00033159458000000319
And
Figure FDA00033159458000000320
are all weight parameters;
Figure FDA00033159458000000321
and
Figure FDA00033159458000000322
are all bias parameters.
7. The method for repairing the missing data in the synchronous measurement of the power distribution network according to claim 1, wherein the formula of the cross-sequence interpolation model based on the fully-connected network is as follows:
Figure FDA00033159458000000323
ht＝φ(Uxt+Vzt+β)
wherein σ and φ are activation functions;
Figure FDA00033159458000000324
the output value of the cross-sequence interpolation model at the moment t based on the full-connection network is obtained; h istIs an intermediate parameter at time t; x is the number oftObtaining a data value of the data matrix to be repaired at a time t;
Figure FDA00033159458000000325
Figure FDA00033159458000000326
and mtRespectively representing the element values of a data matrix to be repaired and a missing position matrix which are repaired by a time sequence interpolation model based on a bidirectional GRU network; w, U and V are both weight parameters; both α and β are bias parameters.
8. The method for repairing the missing data in the synchronous measurement of the power distribution network according to any one of claims 1 to 7, wherein before the step of inputting the data matrix to be repaired, the missing position matrix and the time interval matrix into a pre-trained data repairing model to obtain a repaired data matrix, the method further comprises:
acquiring a training sample set;
performing iterative training on a pre-established data restoration model according to the training sample set to obtain a trained data restoration model;
in each iterative training, evaluating the training result by adopting a root mean square error, and finishing the training when the root mean square error meets a preset condition to obtain the trained data restoration model;
root mean square error
Figure FDA0003315945800000041
Is calculated by the formula
Figure FDA0003315945800000042
Figure FDA0003315945800000043
Matrix output for data recovery model
Figure FDA0003315945800000044
An element of (1);
Figure FDA0003315945800000045
elements of the actual synchronization measurement data matrix x that do not contain missing data;
Figure FDA0003315945800000046
is an element of the missing position matrix; d is the number of channels; t is the time sequence length.
9. A terminal comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the power distribution network synchronization measurement missing data restoration method according to any one of the above claims 1 to 8.
10. A computer-readable storage medium, which stores a computer program, wherein the computer program, when executed by a processor, implements the steps of the power distribution network synchronization measurement missing data recovery method according to any one of claims 1 to 8.
CN202111231057.9A 2021-10-22 2021-10-22 Power distribution network synchronous measurement missing data repairing method, terminal and storage medium Pending CN113936192A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111231057.9A CN113936192A (en) 2021-10-22 2021-10-22 Power distribution network synchronous measurement missing data repairing method, terminal and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111231057.9A CN113936192A (en) 2021-10-22 2021-10-22 Power distribution network synchronous measurement missing data repairing method, terminal and storage medium

Publications (1)

Publication Number Publication Date
CN113936192A true CN113936192A (en) 2022-01-14

Family

ID=79283663

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111231057.9A Pending CN113936192A (en) 2021-10-22 2021-10-22 Power distribution network synchronous measurement missing data repairing method, terminal and storage medium

Country Status (1)

Country Link
CN (1) CN113936192A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114550945A (en) * 2022-02-21 2022-05-27 湖北省疾病预防控制中心（湖北省预防医学科学院） Method for repairing missing data in pulmonary function detection
CN116451006A (en) * 2023-06-12 2023-07-18 湖南大学 PMU data recovery method and system based on enhanced time sequence mode attention
CN116627953A (en) * 2023-05-24 2023-08-22 首都师范大学 Method for repairing loss of groundwater level monitoring data
CN116861347A (en) * 2023-05-22 2023-10-10 青岛海洋地质研究所 Magnetic force abnormal data calculation method based on deep learning model

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114550945A (en) * 2022-02-21 2022-05-27 湖北省疾病预防控制中心（湖北省预防医学科学院） Method for repairing missing data in pulmonary function detection
CN116861347A (en) * 2023-05-22 2023-10-10 青岛海洋地质研究所 Magnetic force abnormal data calculation method based on deep learning model
CN116627953A (en) * 2023-05-24 2023-08-22 首都师范大学 Method for repairing loss of groundwater level monitoring data
CN116627953B (en) * 2023-05-24 2023-10-27 首都师范大学 Method for repairing loss of groundwater level monitoring data
CN116451006A (en) * 2023-06-12 2023-07-18 湖南大学 PMU data recovery method and system based on enhanced time sequence mode attention
CN116451006B (en) * 2023-06-12 2023-08-25 湖南大学 PMU data recovery method and system based on enhanced time sequence mode attention

Similar Documents

Publication Publication Date Title
CN113936192A (en) Power distribution network synchronous measurement missing data repairing method, terminal and storage medium
CN111130839B (en) Flow demand matrix prediction method and system
CN110147367B (en) Temperature missing data filling method and system and electronic equipment
CN115185804A (en) Server performance prediction method, system, terminal and storage medium
CN114168574A (en) Industrial load oriented data missing processing method and device
CN112131638B (en) Wind-induced dynamic characteristic type determination method of large-span roof structure and terminal equipment
CN111931939A (en) Single-amplitude quantum computation simulation method
CN109117352B (en) Server performance prediction method and device
CN113537614A (en) Construction method, system, equipment and medium of power grid engineering cost prediction model
CN113010954A (en) Bridge structure damage identification method and device and terminal equipment
CN115796338A (en) Photovoltaic power generation power prediction model construction and photovoltaic power generation power prediction method
CN107562695B (en) Load data processing method and device for distribution transformer
CN114519429B (en) Method, device and medium for obtaining observability quantity of target system
CN115953907A (en) Traffic speed prediction method based on time-space gating graph convolution network and application thereof
CN111738426A (en) Quantum cryptography system stability control method based on GRU model
CN105160069A (en) Improved compact teaching optimization algorithm based mechanical parameter soft measurement method
CN116227239B (en) Acceleration test data analysis method, device and equipment based on gray prediction model
CN109164496A (en) A kind of earth's magnetic field signal estimation method based on chaos phase space theory
Kochan et al. Integral nonlinearity of third order single bit sigma-delta modulator
CN116842663B (en) Machine learning-based nonlinear time-varying stress value measuring method for automobile leaf springs
Berezin et al. Application of Recurrent Graph Convolutional Networks to the Neural State Estimation Problem
CN112232489A (en) Method and device for gating cycle network and method and device for link prediction
Brzhevska et al. Recurrent Estimation of the Information State Vector and the Correlation of Measuring Impact Matrix using a Multi-Agent Model
CN117935558A (en) Traffic flow prediction method and device integrating autocorrelation mechanism and storage medium
CN114500045A (en) Smart grid false data injection attack detection method and device based on structure sparse matrix separation

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