CN116610911A - Power consumption data restoration method and system based on Bayesian Gaussian tensor decomposition model - Google Patents

Power consumption data restoration method and system based on Bayesian Gaussian tensor decomposition model Download PDF

Info

Publication number
CN116610911A
CN116610911A CN202310884271.7A CN202310884271A CN116610911A CN 116610911 A CN116610911 A CN 116610911A CN 202310884271 A CN202310884271 A CN 202310884271A CN 116610911 A CN116610911 A CN 116610911A
Authority
CN
China
Prior art keywords
data
day
weather
repair
days
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.)
Granted
Application number
CN202310884271.7A
Other languages
Chinese (zh)
Other versions
CN116610911B (en
Inventor
王宗耀
屈浏强
许志浩
丁贵立
康兵
张兴旺
单惠敏
范师尧
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.)
Jiangxi Booway New Technology Co ltd
Nanchang Institute of Technology
Original Assignee
Jiangxi Booway New Technology Co ltd
Nanchang Institute of Technology
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 Jiangxi Booway New Technology Co ltd, Nanchang Institute of Technology filed Critical Jiangxi Booway New Technology Co ltd
Priority to CN202310884271.7A priority Critical patent/CN116610911B/en
Publication of CN116610911A publication Critical patent/CN116610911A/en
Application granted granted Critical
Publication of CN116610911B publication Critical patent/CN116610911B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/10—Pre-processing; Data cleansing
    • G06F18/15—Statistical pre-processing, e.g. techniques for normalisation or restoring missing data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • 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)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Business, Economics & Management (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Economics (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Software Systems (AREA)
  • Water Supply & Treatment (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Probability & Statistics with Applications (AREA)
  • Public Health (AREA)
  • Computational Linguistics (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention belongs to the field of data restoration, and discloses a power consumption data restoration method and a power consumption data restoration system based on a Bayesian tensor decomposition model, which are used for collecting weather, holidays, week types and power consumption data, constructing weather factors, holiday factors and week factors, and constructing a similarity objective function according to the weather factors, the holiday factors and the week factors; optimizing the similarity objective function by using a Trojan optimization algorithm, and searching M similarity days with the highest similarity with the repair day in the history days; and constructing a third-order tensor by using the electricity consumption data of M similar days with the highest similarity, and inputting the third-order tensor into a Bayesian Gaussian tensor decomposition model to carry out data restoration. According to the invention, the improved Trojan optimization algorithm is adopted to select similar days, and incomplete data is restored by inputting the similar days into the Bayesian Gaussian tensor decomposition model, so that the restoration accuracy is improved, the data quality is improved, and the accuracy of behaviors such as prediction and the like is improved.

Description

Power consumption data restoration method and system based on Bayesian Gaussian tensor decomposition model
Technical Field
The invention relates to the technical field of data restoration, in particular to a power consumption data restoration method and system based on a Bayesian Gaussian tensor decomposition model.
Background
With the improvement of the informatization degree of the power system and the acceleration construction of the smart grid, the user power data volume is in an exponential growth state. In order to better serve users and maintain sustainable development of national economy, the method has more and more important significance for analysis and mining of electricity consumption behaviors of massive users in large power consumption data, and meanwhile, the large load data generated by the massive power users also provides more serious challenges for the data restoration technology.
Currently, power consumption information acquisition systems may provide data support for a variety of services. However, due to the reasons of failure of the collecting device, unstable communication channel, external interference and the like, the electricity consumption information collecting system often collects the electric quantity data with data missing, which can cause that various application analysis based on the electric quantity data loses accuracy and even cannot be performed. Therefore, it is necessary to repair the power data in which the data is missing.
Disclosure of Invention
The invention aims to provide a power consumption data restoration method and system based on a Bayesian Gaussian tensor decomposition model, which are used for collecting weather, holidays, week types and power consumption data, constructing weather factors by using gray correlation, constructing holiday factors according to whether the weather factors are holidays, constructing week factors according to week types, and then constructing a similarity objective function. And optimizing the constructed objective function by using an improved wild horse optimization algorithm, so that a similar day with the highest similarity is found in a large amount of historical data, and the data on the similar day is input into a Bayesian tensor decomposition model to repair incomplete data. According to the invention, the improved wild horse optimization algorithm is adopted to select similar days, incomplete data is input into the Bayesian Gaussian tensor decomposition model to repair, the accuracy of repair data is improved, the data quality is improved, and the prediction accuracy is improved.
In order to achieve the above purpose, the technical scheme adopted by the invention is as follows: the power consumption data restoration method based on the Bayesian Gaussian tensor decomposition model comprises the following steps:
step 1: collecting weather data of repair days and history days, whether the weather data is holiday and week type data, and collecting electricity consumption data of incomplete data electricity meters and two adjacent electricity meters on repair days and history days;
step 2: normalizing the collected weather data to obtain normalized weather data;
step 3: constructing the normalized weather data into an weather factor by using gray correlation, constructing a holiday factor according to whether the weather factor is a holiday, constructing a week factor according to the week type, and constructing a similarity objective function according to the weather factor, the holiday factor and the week factor; optimizing the similarity objective function by using a Trojan optimization algorithm, and searching M similarity days with the highest similarity with the repair day in the history days;
step 4: and constructing a third-order tensor by using the electricity consumption data of M similar days with the highest similarity, and inputting the third-order tensor into a Bayesian Gaussian tensor decomposition model to carry out data restoration.
Further preferably, the normalized weather data of the repair day is used as a parent sequence, and the normalized weather data of the history day is used as a child sequence to perform gray correlation calculation:
；
in the method, in the process of the invention,for resolution factor +.>Normalizing weather data for the kth class of repair day; />Normalizing weather data for a kth class of an nth history day; />A gray correlation degree between the kth class normalized weather data of the nth history day and the repair day;
adding gray correlation degrees between various normalized weather data of the nth history day and the repair day to obtain a weather factor of the nth history dayThe formula is as follows: />K is the total number of weather data categories.
Further preferably, the holiday factor is constructed as follows:
；
；
in the method, in the process of the invention,holiday factor for the nth history day, +.>The correlation coefficient between the nth history day and the repair day is given, and q is the number of days between the repair day and the nth history day; mod is a remainder function; />Attenuation coefficient of whether the history day and the repair day are in the same holiday or non-holiday, ++>；/>Daily decay factor of every day is increased for each of the history day and repair day intervals, < >>；/>Is the number of days of the whole year.
Further preferably, the week factor is constructed as follows:
；
in the method, in the process of the invention,as a week factor for the nth history day,，as a coefficient of the weekly decay,is a proportionality coefficient which is used for adjusting the proportionality coefficient according to different daily electric quantity of each region, t 1 To repair the number of days, t 2 Is the number of weeks of similar days.
Further preferably, the similarity objective function is:
；
in the method, in the process of the invention,for the similarity of the nth history day and the repair day, < > for the nth history day and the repair day>Is the weight coefficient of weather factor, +.>Weight coefficient of holiday factor, +.>Is a weighting factor for the week factor.
Further preferably, the algorithm flow of the wild horse optimization algorithm is as follows:
step 3.5.1: creating an initial population:
；
；
wherein:the total number of candidate solutions in the search space, i.e., the total number of horses, d is the total number of decision variables,is the firstThe number of candidate solutions is chosen to be,is the firstThe first candidate solutionThe decision variables i=1, 2, …, N; j=1, 2, …, d;is the firstThe first candidate solutionThe minimum boundary of the individual decision variables,is the firstThe first candidate solutionMaximum boundaries of individual decision variables;is oneUniformly distributed random numbers;
dividing an initial population into several groups, wherein the grouping number is G= [ N x PS ], wherein PS is the percentage of horses in the population, taking PS as a control parameter of a wild horse algorithm, obtaining G leaders according to the number of the population, and uniformly distributing the remaining N-G members, randomly selecting the leaders of each group at the initial stage, and selecting the leaders according to the adaptability among the members at the subsequent stage;
step 3.5.2: grazing behavior:
the group members search for the center by taking the leader as the center, and the expression is:
；
wherein:for the updated location of the member,is the leader position, the firstCandidate solutionsIn the initial position of the device, the device is in the initial position,to take the value of [ -2,2]Random numbers between the member and the leader, for controlling the angle of the member and the leader,is an adaptive mechanism;
；
wherein:is a vector consisting of 0 and 1,andis [0,1]Random vectors uniformly distributed in space;to take the value of [0,1]]A random number between the two random numbers,for the purpose of the index,representing binary negation; satisfies the conditionRandom vector of (a)Returning an index;is an adaptive factor;
；
wherein: t is the number of iterations and,the maximum iteration number;
step 3.5.3: wild Ma Jiaopei: the wild horses leaving the u-th group and the wild horses leaving the v-th group are respectively added into a temporary group, and cross after entering puberty; updating the position of the wild Ma Zidai after crossing;
；
；
；
；
wherein:representing the position of the next generation of wild horses, +.>Representing the current location of the small trojan g within the u-th group,represents the little wild horse +.>Is>For crossing function +.>The position of the parent is indicated,represents the position of parent->Representing the position of globally optimal individuals, ">Representing the weight ratio of parent and transferred to offspring genes,/->Representing the weight ratio of parent to offspring genes;
step 3.5.4: team leaders: the location update formula of the leader is:
，
wherein:updated locations for the group leader,representing an optimal individual location;
step 3.5.5: the leader selects and pulls out:
，
wherein:the leader that appears is the optimal solution as an adaptive function.
Further preferably, the third-order tensor is formed by three matrices, and the first matrix is:
；
wherein N is u For the number of sampling points per day,1 st sampling point electricity consumption data of the day of ammeter repair representing data to be repaired, +.>The 1 st sampling point electricity consumption data of the data ammeter to be repaired of the M-th similar day,n-th day of ammeter repair day representing data to be repaired u Power consumption data of each sampling point, < >>Nth of data ammeter to be repaired representing mth similar day u The power consumption data of each sampling point is +.>To->Point missing data exists between the two;
the second matrix is:
；
in the method, in the process of the invention,1 st sampling point electricity consumption data representing the day of repair date of adjacent meter a of the data meter to be repaired,/>The 1 st sampling point electricity consumption data of the adjacent meter a of the data meter to be repaired of the M-th similar day,n-th day of repair day of adjacent meter a representing data meter to be repaired u Power consumption data of each sampling point, < >>N-th of adjacent meter a representing M-th similar day to be repaired data meter u The power consumption data of the sampling points;
the third matrix is:
；
in the method, in the process of the invention,1 st sampling point electricity consumption data representing the day of repair of adjacent meter c of the data meter to be repaired,/>The 1 st sampling point electricity consumption data of the adjacent ammeter c of the data ammeter to be repaired representing the M-th similar day,n-th day of repair day of adjacent ammeter c representing data ammeter to be repaired u Power consumption data of each sampling point, < >>N-th of adjacent meter c of the data meter to be repaired shown as M-th similar day u And the power consumption data of each sampling point.
Further preferably, the weather data includes a highest air temperature, a lowest air temperature, a rainfall, and a humidity.
The invention also discloses a system for realizing the electricity consumption data restoration method based on the Bayesian Gaussian tensor decomposition model, which comprises a data acquisition module, a normalization data processing module, a similar day calculation module and a data restoration module, wherein the data acquisition module comprises a weather and date data acquisition module and an electric quantity acquisition device, the weather and date data acquisition module is used for collecting weather data of restoration days and history days, whether the weather and date data are holiday data and week type data, and the electric quantity acquisition device is used for collecting electricity consumption data of incomplete data electricity meters and adjacent two electricity meters in restoration days and history days; the normalization data processing module performs normalization processing on the collected weather data to obtain normalized weather data; the similarity day calculation module is internally provided with a similarity objective function and a wild horse optimization algorithm; the data restoration module is internally provided with a Bayesian Gaussian tensor decomposition model.
The invention has the beneficial effects that: compared with the prior art, the method has the advantages that the accuracy of the selected similar days is improved by utilizing the historical data and the improved wild horse optimization algorithm, the data restoration is considered in terms of space and time, the information of surrounding data is utilized in space, and the accuracy of the data restoration is improved.
Drawings
FIG. 1 is a flow chart of a method for restoration of electrical data based on a Bayesian Gaussian tensor decomposition model.
FIG. 2 is a graph of improved algorithm for the optimization of the wild horse and the convergence of the optimal fitness before improvement.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.
Referring to fig. 1, the method for repairing the electricity consumption data based on the bayesian tensor decomposition model comprises the following steps:
step 1: weather data (including highest air temperature, lowest air temperature, rainfall and humidity) of repair days and history days, whether holiday and week type data are collected, and electricity consumption data of incomplete data electricity meters and two adjacent electricity meters repair days and history days are collected.
Step 2: and carrying out normalization processing on the collected weather data to obtain normalized weather data. The value range of the weather data is normalized to be between 0 and 1 by normalization, and the normalization mode is selected from the maximum and minimum modes:
；
wherein: z represents normalized weather data, Z represents weather data,representing maximum weather data, ++>Representing a weather data minimum;
step 3: constructing the normalized weather data into an weather factor by using gray correlation, constructing a holiday factor according to whether the weather factor is a holiday, constructing a week factor according to the week type, and constructing a similarity objective function according to the weather factor, the holiday factor and the week factor; and optimizing the similarity objective function by using a Trojan optimization algorithm, and searching M similarity days with the highest similarity with the repair day in the history days.
Step 3.1: taking normalized weather data of a repair day as a parent sequence, and taking normalized weather data of a history day as a subsequence to perform gray correlation calculation, wherein the formula is as follows:
；
in the method, in the process of the invention,，/>is a resolution factor. />The smaller the resolution, the greater the resolution, generally +.>The value interval of (1, 0),the specific values may be as appropriate. When->At the same time, the resolution is best, usually +.>。/>Normalized weather data (k=1, 2,3,4, corresponding to the highest air temperature, the lowest air temperature, the rainfall, the humidity, respectively) for the kth class of repair days; />Normalizing weather data for a kth class of an nth history day; />The gray correlation between the kth class normalized weather data for the nth history day and repair day.
Adding gray correlation degrees between various normalized weather data of the nth history day and the repair day to obtain a weather factor of the nth history dayThe formula is as follows: />K is the total number of weather data categories, and in this embodiment K is 4;
step 3.2: since electricity consumption of holiday residents is different from that of ordinary holiday residents, a holiday factor is constructed, and the formula is as follows:
；
；
in the method, in the process of the invention,is the firstHoliday factors for n history days, < ->The correlation coefficient between the nth history day and the repair day is given, and q is the number of days between the repair day and the nth history day; mod is a remainder function; />Attenuation coefficient of whether the history day and the repair day are in the same holiday or non-holiday, ++>；/>Daily decay factor of every day is increased for each of the history day and repair day intervals, < >>；/>The number of days of the whole year is 365.
Step 3.3: since domestic electricity has a periodic period, a week factor is constructed, and the formula is as follows:
；
in the method, in the process of the invention,as a week factor for the nth history day,，as a coefficient of the weekly decay,is a proportionality coefficient which is used for adjusting the proportionality coefficient according to different daily electric quantity of each region, t 1 To repair the number of days, t 2 Is the number of weeks of similar days.
Step 3.4: constructing a similarity objective function according to the weather factor, the holiday factor and the week factor:
；
in the method, in the process of the invention,for the similarity of the nth history day and the repair day, < > for the nth history day and the repair day>Is the weight coefficient of weather factor, +.>Weight coefficient of holiday factor, +.>Is a weighting factor for the week factor.
Step 3.5: optimizing the similarity objective function by using an improved pothorse optimization algorithm;
step 3.5.1: creating an initial population:
；
；
wherein:the total number of candidate solutions in the search space, i.e., the total number of horses, d is the total number of decision variables,is the firstThe number of candidate solutions is chosen to be,is the firstThe first candidate solutionThe decision variables i=1, 2, …, N; j=1, 2, …, d;is the firstThe first candidate solutionThe minimum boundary of the individual decision variables,is the firstThe first candidate solutionMaximum boundaries of individual decision variables;is oneUniformly distributed random numbers;
dividing an initial population into several groups, wherein the grouping number is G= [ N x PS ], wherein PS is the percentage of horses in the population, taking PS as a control parameter of a wild horse algorithm, obtaining G leaders according to the number of the population, and uniformly distributing the remaining N-G members, randomly selecting the leaders of each group at the initial stage, and selecting the leaders according to the adaptability among the members at the subsequent stage;
step 3.5.2: grazing behavior:
the group members search for the center by taking the leader as the center, and the expression is:
；
wherein:for the updated location of the member,is the leader position, the firstCandidate solutionsIn the initial position of the device, the device is in the initial position,to take the value of [ -2,2]Random numbers between the member and the leader, for controlling the angle of the member and the leader,is an adaptive mechanism;
，
wherein:is a vector consisting of 0 and 1,andis [0,1]Random vectors uniformly distributed in space;to take the value of [0,1]]A random number between the two random numbers,for the purpose of the index,representing binary negation; satisfies the conditionRandom vector of (a)Returning an index;is an adaptive factor;
；
wherein: t is the number of iterations and,the maximum iteration number;
step 3.5.3: wild Ma Jiaopei: the wild horses leaving the u-th group and the wild horses leaving the v-th group are respectively added into a temporary group, and cross after entering puberty; updating the position of the wild Ma Zidai after crossing;
；
；
；
；
wherein:representing the position of the next generation of wild horses, +.>Representing the field within the u-th groupThe current position of the horse g is determined,represents the little wild horse +.>Is>For crossing function +.>Representing the position of the parent,/->Represents the position of parent->Representing the position of globally optimal individuals, ">Representing the weight ratio of parent and transferred to offspring genes,/->The weight ratio of parent to offspring genes is expressed.
Step 3.5.4: team leaders: the location update formula of the leader is:
，
wherein:updated locations for the group leader,representing an optimal individual location;
step 3.5.5: the leader selects and pulls out:
，
wherein:the leader that appears is the optimal solution as an adaptive function.
The convergence curves of the optimal fitness of the improved and improved wild horse optimization algorithms are shown in fig. 2, which shows that the improved wild horse optimization algorithm can accelerate the optimizing efficiency and has higher convergence accuracy.
Step 4: the third-order tensor is formed by using the electricity consumption data of M (M=30) with the highest similarity days, and the third-order tensor is input into a Bayesian tensor decomposition model for data restoration.
The third-order tensor is composed of three matrices, wherein b is the ammeter of the data to be repaired, and the first matrix is composed of 1 repair day (missing data day) plus power consumption data of 96 sampling points of 30 similar days (15 minutes of sampling period)The electricity consumption data of 96 sampling points a day of 1 repair day (missing data day) plus 30 similar days of an adjacent electric meter of the electric meters to be repaired data form a second matrix +.>. The electricity consumption data of 96 sampling points a day of 1 repair day (missing data day) plus 30 similar days of another ammeter adjacent to the ammeter to be repaired data forms a third matrix->。
；
In the method, in the process of the invention,1 st sampling point electricity consumption data of the day of ammeter repair representing data to be repaired, +.>1 st sampling point electricity consumption data of the data ammeter to be repaired representing 30 th similar day, +.>96 th sampling point electricity consumption data of the day of ammeter restoration representing data to be restored, +.>And 96 th sampling point electricity consumption data of the data ammeter to be repaired representing the 30 th similar day. At->To->There is a point missing data in between.
In the method, in the process of the invention,1 st sampling point electricity consumption data representing the day of repair date of adjacent meter a of the data meter to be repaired,/>The 1 st sampling point electricity consumption data of the adjacent meter a of the data meter to be repaired representing the 30 th similar day,96 th sampling point electricity consumption data representing the day of repair date of adjacent meter a of the data meter to be repaired, < +.>96 th sampling point electricity consumption data of the adjacent meter a of the data meter to be repaired representing the 30 th similar day.
In the method, in the process of the invention,1 st sampling point electricity consumption data representing the day of repair of adjacent meter c of the data meter to be repaired,/>The 1 st sampling point electricity consumption data of the adjacent meter c of the data meter to be repaired representing the 30 th similar day,96 th sampling point electricity consumption data representing the day of repair date of adjacent meter c of the data meter to be repaired,/>96 th sampling point electricity consumption data of the adjacent meter c of the data meter to be repaired, which is shown as 30 th similar day.
The embodiment provides a system for realizing an electricity consumption data restoration method based on a Bayesian Gaussian tensor decomposition model, which comprises a data acquisition module, a normalization data processing module, a similar day calculation module and a data restoration module, wherein the data acquisition module comprises a weather and date data acquisition module and an electric quantity acquisition device, the weather and date data acquisition module is used for collecting weather data of restoration days and history days and whether the weather data are holiday data and week type data, and the electric quantity acquisition device is used for collecting electricity consumption data of incomplete data ammeter and two adjacent ammeter restoration days and history days; the normalization data processing module performs normalization processing on the collected weather data to obtain normalized weather data; the similarity day calculation module is internally provided with a similarity objective function and a wild horse optimization algorithm; the data restoration module is internally provided with a Bayesian Gaussian tensor decomposition model.
Finally, it should be noted that: the foregoing is merely a preferred example of the present invention, and the present invention is not limited thereto, but it is to be understood that modifications and equivalents of some of the technical features described in the foregoing embodiments may be made by those skilled in the art, although the present invention has been described in detail with reference to the foregoing embodiments. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (9)

1. The power consumption data restoration method based on the Bayesian Gaussian tensor decomposition model is characterized by comprising the following steps of:
step 1: collecting weather data of repair days and history days, whether the weather data is holiday and week type data, and collecting electricity consumption data of incomplete data electricity meters and two adjacent electricity meters on repair days and history days;
step 2: normalizing the collected weather data to obtain normalized weather data;
step 3: constructing the normalized weather data into an weather factor by using gray correlation, constructing a holiday factor according to whether the weather factor is a holiday, constructing a week factor according to the week type, and constructing a similarity objective function according to the weather factor, the holiday factor and the week factor; optimizing the similarity objective function by using a Trojan optimization algorithm, and searching M similarity days with the highest similarity with the repair day in the history days;
step 4: and constructing a third-order tensor by using the electricity consumption data of M similar days with the highest similarity, and inputting the third-order tensor into a Bayesian Gaussian tensor decomposition model to carry out data restoration.
2. The method for repairing electrical data based on a bayesian tensor decomposition model according to claim 1, wherein normalized weather data on repair days is used as a parent sequence, and normalized weather data on history days is used as a subsequence to perform gray correlation calculation:
；
in the method, in the process of the invention,for resolution factor +.>Normalizing weather data for the kth class of repair day; />Normalizing weather data for a kth class of an nth history day; />A gray correlation degree between the kth class normalized weather data of the nth history day and the repair day;
adding gray correlation degrees between various normalized weather data of the nth history day and the repair day to obtain a weather factor of the nth history dayThe formula is as follows: />K is the total number of weather data categories.
3. The method for restoring electricity consumption data based on a Bayesian Gaussian tensor decomposition model as set forth in claim 2, wherein the holiday factor is constructed as follows:
；
；
in the method, in the process of the invention,holiday factor for the nth history day, +.>The correlation coefficient between the nth history day and the repair day is given, and q is the number of days between the repair day and the nth history day; mod is a remainder function; />The attenuation coefficient of whether the history day and the repair day are in the same holiday or non-holiday is determined; />Daily decay coefficients are increased for each day of the history day and the repair day; />Is the number of days of the whole year.
4. The method for repairing electrical data based on a bayesian tensor decomposition model according to claim 3, wherein the week factor is constructed as follows:
；
in the method, in the process of the invention,for week factor of the nth history day, +.>Is a week attenuation coefficient>Is a proportionality coefficient which is used for adjusting the proportionality coefficient according to different daily electric quantity of each region, t 1 To repair the number of days, t 2 Is the number of weeks of similar days.
5. The method for restoring electrical data based on a bayesian tensor decomposition model according to claim 4, wherein the similarity objective function is:
；
in the method, in the process of the invention,for the similarity of the nth history day and the repair day, < > for the nth history day and the repair day>Is the weight coefficient of weather factor, +.>Weight coefficient of holiday factor, +.>Is a weighting factor for the week factor.
6. The method for repairing electricity consumption data based on a Bayesian Gaussian tensor decomposition model according to claim 1, wherein the algorithm flow of the wild horse optimization is as follows:
step 3.5.1: creating an initial population:
；
；
wherein:for the total number of candidate solutions in the search space, i.e. for the total number of horses, d for the total number of decision variables, +.>Is->Candidate solutions (L)>Is->First->The decision variables i=1, 2, …, N; j=1, 2, …, d; />Is->First->Minimum boundary of individual decision variables, +.>Is->First->Maximum boundaries of individual decision variables;is +.>Uniformly distributed random numbers;
dividing an initial population into several groups, wherein the grouping number is G= [ N x PS ], wherein PS is the percentage of horses in the population, taking PS as a control parameter of a wild horse algorithm, obtaining G leaders according to the number of the population, and uniformly distributing the remaining N-G members, randomly selecting the leaders of each group at the initial stage, and selecting the leaders according to the adaptability among the members at the subsequent stage;
step 3.5.2: grazing behavior:
the group members search for the center by taking the leader as the center, and the expression is:
；
wherein:updated location for member, +.>Is the leader position +.>Candidate solution->In the initial position->To take the value of [ -2,2]Random numbers between members and leader for controlling the angle of the members,/and>is an adaptive mechanism;
；
wherein:is a vector consisting of 0 and 1, < >>And->Is [0,1]Random vectors uniformly distributed in space; />To take the value of [0,1]]Random number between->For index (I)>Representing binary negation; satisfy condition->Random vector of->Returning an index; />Is an adaptive factor;
；
wherein: t is the number of iterations and,the maximum iteration number;
step 3.5.3: wild Ma Jiaopei: the wild horses leaving the u-th group and the wild horses leaving the v-th group are respectively added into a temporary group, and cross after entering puberty; updating the position of the wild Ma Zidai after crossing;
；
；
；
；
wherein:representing the position of the next generation of wild horses, +.>Representing the current position of the small wild g in the u-th group,/o->Represents the little wild horse +.>Is>For crossing function +.>Representing the position of the parent,/->Represents the position of parent->Representing the position of globally optimal individuals, ">Representing the weight ratio of parent and transferred to offspring genes,/->Representing the weight ratio of parent to offspring genes;
step 3.5.4: team leaders: the location update formula of the leader is:
，
wherein:updated location for the u-th group leader,/->Representing an optimal individual location;
step 3.5.5: the leader selects and pulls out:
，
wherein:the leader that appears is the optimal solution as an adaptive function.
7. The method for restoring electrical data based on bayesian tensor decomposition model according to claim 1, wherein the third-order tensor is composed of three matrices, and the first matrix is:
；
wherein N is u For the number of sampling points per day,1 st sampling point electricity consumption data of the day of ammeter repair representing data to be repaired, +.>1 st sampling point electricity consumption data of the data ammeter to be repaired representing the Mth similar day, +.>N-th day of ammeter repair day representing data to be repaired u Power consumption data of each sampling point, < >>Nth of data ammeter to be repaired representing mth similar day u The power consumption data of each sampling point is +.>To->Point missing data exists between the two;
the second matrix is:
；
in the method, in the process of the invention,the 1 st sampling point electricity consumption data representing the current day of repair of the adjacent meter a of the data meter to be repaired,1 st sampling point electricity consumption data of adjacent meter a of the data meter to be repaired representing the M-th similar day, ">N-th day of repair day of adjacent meter a representing data meter to be repaired u Power consumption data of each sampling point, < >>N-th of adjacent meter a representing M-th similar day to be repaired data meter u The power consumption data of the sampling points;
the third matrix is:
；
in the method, in the process of the invention,the 1 st sampling point electricity consumption data representing the current day of repair of the adjacent meter c of the data meter to be repaired,1 st sampling point electricity consumption data of adjacent meter c of M-th data meter to be repaired on similar day, ">N-th day of repair day of adjacent ammeter c representing data ammeter to be repaired u Power consumption data of each sampling point, < >>N-th of adjacent meter c of the data meter to be repaired shown as M-th similar day u And the power consumption data of each sampling point.
8. The method for restoration of electricity consumption data based on a bayesian tensor decomposition model according to claim 1, wherein the weather data includes a highest air temperature, a lowest air temperature, a rainfall and a humidity.
9. A system for implementing the power consumption data restoration method based on the bayesian tensor decomposition model according to any one of claims 1-8, which is characterized by comprising a data acquisition module, a normalization data processing module, a similar day calculation module and a data restoration module, wherein the data acquisition module comprises a weather and date data acquisition module and a power acquisition device, the weather and date data acquisition module is used for collecting weather data of restoration days and history days, whether the weather and date data are holiday and week type data, and the power acquisition device is used for collecting power consumption data of incomplete data ammeter and restoration days and history days of two adjacent ammeter; the normalization data processing module performs normalization processing on the collected weather data to obtain normalized weather data; the similarity day calculation module is internally provided with a similarity objective function and a wild horse optimization algorithm; the data restoration module is internally provided with a Bayesian Gaussian tensor decomposition model.
CN202310884271.7A 2023-07-19 2023-07-19 Power consumption data restoration method and system based on Bayesian Gaussian tensor decomposition model Active CN116610911B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310884271.7A CN116610911B (en) 2023-07-19 2023-07-19 Power consumption data restoration method and system based on Bayesian Gaussian tensor decomposition model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310884271.7A CN116610911B (en) 2023-07-19 2023-07-19 Power consumption data restoration method and system based on Bayesian Gaussian tensor decomposition model

Publications (2)

Publication Number Publication Date
CN116610911A true CN116610911A (en) 2023-08-18
CN116610911B CN116610911B (en) 2023-09-19

Family

ID=87685691

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310884271.7A Active CN116610911B (en) 2023-07-19 2023-07-19 Power consumption data restoration method and system based on Bayesian Gaussian tensor decomposition model

Country Status (1)

Country Link
CN (1) CN116610911B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104616075A (en) * 2015-01-30 2015-05-13 广西大学 Short-term load predicating method suitable for typhoon weather
CN105447509A (en) * 2015-11-10 2016-03-30 天津大学 Short-term power prediction method for photovoltaic power generation system
CN108830418A (en) * 2018-06-14 2018-11-16 东北电力大学 A kind of Short-Term Load Forecasting Method
CN109636003A (en) * 2018-11-14 2019-04-16 北京电力经济技术研究院有限公司 A kind of power-system short-term load forecasting method of high precision of prediction
CN112001554A (en) * 2020-08-26 2020-11-27 山东德佑电气股份有限公司 Short-term load prediction method based on parameter self-adaptive similar daily method
WO2022257243A1 (en) * 2021-06-07 2022-12-15 西安邮电大学 Water quality early-warning method and system
CN116432510A (en) * 2022-11-02 2023-07-14 贵州电网有限责任公司 Flexible load prediction method based on ensemble learning-longhorn beetle whisker optimization algorithm

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104616075A (en) * 2015-01-30 2015-05-13 广西大学 Short-term load predicating method suitable for typhoon weather
CN105447509A (en) * 2015-11-10 2016-03-30 天津大学 Short-term power prediction method for photovoltaic power generation system
CN108830418A (en) * 2018-06-14 2018-11-16 东北电力大学 A kind of Short-Term Load Forecasting Method
CN109636003A (en) * 2018-11-14 2019-04-16 北京电力经济技术研究院有限公司 A kind of power-system short-term load forecasting method of high precision of prediction
CN112001554A (en) * 2020-08-26 2020-11-27 山东德佑电气股份有限公司 Short-term load prediction method based on parameter self-adaptive similar daily method
WO2022257243A1 (en) * 2021-06-07 2022-12-15 西安邮电大学 Water quality early-warning method and system
CN116432510A (en) * 2022-11-02 2023-07-14 贵州电网有限责任公司 Flexible load prediction method based on ensemble learning-longhorn beetle whisker optimization algorithm

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
朱祥和;: "考虑气象因素的短期负荷预测模型研究", 数学的实践与认识, no. 03 *
李刚;刘佳林;王腾飞;丁坤;: "基于相似日理论和IPSO-Elman模型的短期光伏发电功率预测", 测控技术, no. 02 *

Also Published As

Publication number Publication date
CN116610911B (en) 2023-09-19

Similar Documents

Publication Publication Date Title
Gharehchopogh et al. A comprehensive survey on symbiotic organisms search algorithms
Liang et al. Evolutionary multi-task optimization for parameters extraction of photovoltaic models
Kong et al. Improved deep belief network for short-term load forecasting considering demand-side management
Høverstad et al. Short-term load forecasting with seasonal decomposition using evolution for parameter tuning
Cheng et al. Optimizing hydropower reservoir operation using hybrid genetic algorithm and chaos
Wu et al. A diploid genetic approach to short-term scheduling of hydro-thermal system
Li et al. An integrated missing-data tolerant model for probabilistic PV power generation forecasting
Muzumdar et al. Designing a robust and accurate model for consumer-centric short-term load forecasting in microgrid environment
CN112101521A (en) Building energy consumption prediction method based on long-term and short-term memory network hybrid model
CN114511132A (en) Photovoltaic output short-term prediction method and prediction system
CN110807508B (en) Bus peak load prediction method considering complex weather influence
Nama et al. A new parameter setting-based modified differential evolution for function optimization
CN117454939A (en) Wind power prediction method for optimizing BP neural network based on NSABO-Gold algorithm
Ibrahim et al. LSTM neural network model for ultra-short-term distribution zone substation peak demand prediction
CN116610911B (en) Power consumption data restoration method and system based on Bayesian Gaussian tensor decomposition model
CN112101612A (en) Prediction method of town gas usage, storage medium and terminal equipment
Shayeghi et al. STLF based on optimized neural network using PSO
Adika et al. Short term energy consumption prediction using bio-inspired fuzzy systems
CN115840794A (en) Photovoltaic system planning method based on GIS (geographic information System) and RL (Link State) models
Zuo Integrated forecasting models based on LSTM and TCN for short-term electricity load forecasting
Dash et al. Genetic optimization of a self organizing fuzzy-neural network for load forecasting
CN115471009A (en) Predictive optimized power system planning method
Hwang Daily electric load forecasting based on RBF neural network models
Hu et al. An approximate Annealing Search algorithm to global optimization and its connection to stochastic approximation
CN111931994A (en) Short-term load and photovoltaic power prediction method, system, equipment and medium thereof

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
GR01 Patent grant
GR01 Patent grant