CN114844048A - Power grid regulation and control demand-oriented adjustable load regulation capacity evaluation method - Google Patents

Power grid regulation and control demand-oriented adjustable load regulation capacity evaluation method Download PDF

Info

Publication number
CN114844048A
CN114844048A CN202210465933.2A CN202210465933A CN114844048A CN 114844048 A CN114844048 A CN 114844048A CN 202210465933 A CN202210465933 A CN 202210465933A CN 114844048 A CN114844048 A CN 114844048A
Authority
CN
China
Prior art keywords
user
target domain
adjustable load
domain
energy consumption
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
CN202210465933.2A
Other languages
Chinese (zh)
Inventor
刘刚
黄奇峰
杨冬梅
李波
阮文骏
杨世海
朱庆
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.)
State Grid Jiangsu Electric Power Co ltd Marketing Service Center
State Grid Corp of China SGCC
State Grid Jiangsu Electric Power Co Ltd
Nari Technology Co Ltd
State Grid Electric Power Research Institute
Original Assignee
State Grid Jiangsu Electric Power Co ltd Marketing Service Center
State Grid Corp of China SGCC
State Grid Jiangsu Electric Power Co Ltd
Nari Technology Co Ltd
State Grid Electric Power Research Institute
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 State Grid Jiangsu Electric Power Co ltd Marketing Service Center, State Grid Corp of China SGCC, State Grid Jiangsu Electric Power Co Ltd, Nari Technology Co Ltd, State Grid Electric Power Research Institute filed Critical State Grid Jiangsu Electric Power Co ltd Marketing Service Center
Priority to CN202210465933.2A priority Critical patent/CN114844048A/en
Publication of CN114844048A publication Critical patent/CN114844048A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/12—Circuit arrangements for ac mains or ac distribution networks for adjusting voltage in ac networks by changing a characteristic of the network load
    • H02J3/14—Circuit arrangements for ac mains or ac distribution networks for adjusting voltage in ac networks by changing a characteristic of the network load by switching loads on to, or off from, network, e.g. progressively balanced loading
    • H02J3/144—Demand-response operation of the power transmission or distribution network
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06312—Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/007—Arrangements for selectively connecting the load or loads to one or several among a plurality of power lines or power sources
    • H02J3/0075—Arrangements for selectively connecting the load or loads to one or several among a plurality of power lines or power sources for providing alternative feeding paths between load and source according to economic or energy efficiency considerations, e.g. economic dispatch
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
    • H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J2310/00—The network for supplying or distributing electric power characterised by its spatial reach or by the load
    • H02J2310/50—The network for supplying or distributing electric power characterised by its spatial reach or by the load for selectively controlling the operation of the loads
    • H02J2310/56—The network for supplying or distributing electric power characterised by its spatial reach or by the load for selectively controlling the operation of the loads characterised by the condition upon which the selective controlling is based
    • H02J2310/58—The condition being electrical
    • H02J2310/60—Limiting power consumption in the network or in one section of the network, e.g. load shedding or peak shaving
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S50/00—Market activities related to the operation of systems integrating technologies related to power network operation or related to communication or information technologies
    • Y04S50/14—Marketing, i.e. market research and analysis, surveying, promotions, advertising, buyer profiling, customer management or rewards

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Strategic Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Economics (AREA)
  • Evolutionary Computation (AREA)
  • Development Economics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Educational Administration (AREA)
  • Biomedical Technology (AREA)
  • Computing Systems (AREA)
  • Power Engineering (AREA)
  • General Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • Quality & Reliability (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • General Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Game Theory and Decision Science (AREA)
  • Health & Medical Sciences (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

The invention discloses a power grid regulation demand-oriented adjustable load regulation capacity evaluation method, and belongs to the technical field of power grid demand side management. The method comprises the steps of constructing an adjustable load generalized model according to resource characteristics of a demand side, fitting quadratic regression dynamic relations between energy consumption behaviors and various uncertain factors, extracting similar energy consumption behaviors among different users by using a clustering method, obtaining quadratic regression parameters of the energy consumption behaviors and the uncertain factors of each user in a source domain and a target domain according to a clustering result, calculating cosine similarity of parameter feature characteristics among the users in a class, obtaining probability distribution of adjustable load adjustment capability of the target domain according to the cosine similarity, aligning the energy consumption behavior characteristics of the users in the source domain and the target domain according to local maximum average difference, and evaluating the adjustable load adjustment capability of the target domain by using an online transfer learning network model based on feature alignment. The method and the device can realize the evaluation of the adjustable load regulation capacity of the demand side, and provide theoretical reference and technical support for the management of the demand side.

Description

Power grid regulation and control demand-oriented adjustable load regulation capacity evaluation method
Technical Field
The invention belongs to the technical field of power grid demand side management, and particularly relates to a power grid regulation demand-oriented adjustable load regulation capacity evaluation method.
Background
With the increase of the grid-connected proportion of distributed resources, a traditional unit is retired in a large amount, a large amount of flexible loads on a demand side are connected in an unordered mode, a peak load situation is increasingly prominent, the output of a distributed power supply is difficult to track a load curve, the phenomena of wind abandoning and light abandoning are serious, the power grid regulating capacity is seriously insufficient, safe and stable operation faces challenges, the power grid is easy to cause heavy-load overload operation, even the power grid is collapsed, and the frequent occurrence of major power failure accidents caused by the heavy-load overload operation is easy. The demand side implementation of demand response is an important direction for improving the system adjusting capacity and promoting new energy consumption to ensure the safety and stability of the system.
In the face of the operation characteristics of a novel power system, the adjustable load resources on the demand side are brought into the operation regulation and control of the normalized power grid, the conversion of the power grid from 'source load following motion' to 'source load interaction' is promoted, and the evaluation of the adjustable load regulation capacity is very urgent and necessary. In order to solve the problem of source-load imbalance caused by the double-height characteristic of the power grid, the adjustable load resource on the demand side needs to be excavated urgently, so that the method for evaluating the adjustable load adjusting capacity facing the power grid adjusting and controlling requirement has important research significance.
Disclosure of Invention
The invention aims to provide a power grid regulation demand-oriented adjustable load regulation capacity evaluation method, which is used for bringing demand side adjustable load resources into normalized power grid operation regulation and control and promoting the power grid to be changed from 'source load following movement' to 'source load interaction'.
In order to achieve the purpose, the invention adopts the following technical scheme:
a power grid regulation demand oriented adjustable load regulation capacity assessment method comprises the following steps:
acquiring user energy data to be evaluated at a demand side, wherein the user energy data comprises user energy behaviors, uncertain factors and electricity prices;
and inputting the energy data of the user to be evaluated into the trained online transfer learning network model, and outputting the adjustable load adjustment ratio of the user by the model.
Further, the online transfer learning network model is obtained by training through the following method:
acquiring historical energy data of a source domain and historical energy data of a target domain;
respectively constructing a source domain adjustable load generalized model and a target domain adjustable load generalized model according to user energy consumption behaviors in source domain data and target domain data and uncertain factors influencing the user energy consumption behaviors, and fitting based on the source domain adjustable load generalized model and the target domain adjustable load generalized model to obtain a quadratic regression dynamic relation between the source domain user energy consumption behaviors and the target domain user energy consumption behaviors and the uncertain factors;
clustering historical energy consumption data of a source domain and a target domain respectively to extract similar energy consumption behaviors among different users, and obtaining secondary regression parameters of the energy consumption behaviors and the uncertain factors of the similar source domain and the target domain users respectively according to clustering results of the source domain and the target domain and secondary regression dynamic relations of the energy consumption behaviors and the uncertain factors of the source domain and the target domain users;
calculating the cosine similarity of the quadratic regression parameter characteristics of the energy consumption behaviors of the users in the target domain and the source domain and the uncertain factors, and obtaining the probability distribution of the adjustable load regulation capacity of the target domain according to the cosine similarity;
based on the probability distribution of the adjustable load regulation capacity of the target domain, aligning the energy use behavior characteristics of the source domain and the target domain users according to the local maximum average difference, and minimizing the loss function through continuously training the model until the model converges.
Further, the adjustable load generalized model is:
P t ＝P b,t (R m )+ΔP t (ΔR)＝P b,t (R m )(1+f t (ΔR))
in the formula, P t The load can be adjusted for time t; p b,t (R m ) The load baseline load can be adjusted for time t, which is associated with an uncertainty factor R that affects the user's energy behavior m (ii) related; delta P t (Δ R) an adjustable load change value for time t, which is related to an uncertainty factor Δ R affecting the user energy behaviour; r m And Δ R is the average value and the variation value of the uncertain factors of a certain period of time respectively; f. of t (Δ R) is P b,t (R m ) And Δ P t (Δ R).
Further, the quadratic regression dynamic relationship between the user energy behavior and various uncertain factors is as follows:
f t (ΔR)＝f 1,t +f 2,t ΔR+f 3,t ΔR 2
in the formula (f) 1,t 、f 2,t 、f 3,t Are parameters of quadratic regression.
Further, the clustering historical energy consumption data of the source domain and the target domain to extract similar energy consumption behaviors among different users includes:
randomly selecting K user energy historical data as an initial clustering center;
calculating the distance of the remaining samples to the cluster center and assigning the samples to the nearest cluster:
Figure BDA0003624110630000031
wherein K is the number of clusters, v ik Indicates whether the ith user sample belongs to class k, v ik Is 1 denotes belonging to class k, v ik A value of 0 indicates not belonging to class k, d (c) m ,x i ) Is a sample x i To the center of the cluster c m The distance of (d);
updating a clustering center:
Figure BDA0003624110630000041
in the formula, N is the number of users;
the convergence condition is judged so that the following expression is minimized:
Figure BDA0003624110630000042
further, the similarity of the quadratic regression parameter characteristic cosine of the energy use behaviors of the users in the target domain and the source domain in the class and the uncertain factors is calculated according to the following steps:
the set of quadratic regression parameters of the kth class ith user energy consumption behaviors and uncertain factors of the source domain is as follows:
Figure BDA0003624110630000043
in the formula (I), the compound is shown in the specification,
Figure BDA0003624110630000044
the adjustable load baseline load for the ith user of the source domain,
Figure BDA0003624110630000045
Figure BDA0003624110630000046
respectively are quadratic regression parameters of the ith user of the source domain;
the quadratic regression parameter set of the kth class ith user energy consumption behavior and the uncertain factors of the target domain is as follows:
Figure BDA0003624110630000047
in the formula (I), the compound is shown in the specification,
Figure BDA0003624110630000048
the adjustable load baseline load for the ith user of the target domain,
Figure BDA0003624110630000049
Figure BDA00036241106300000410
the quadratic regression parameter of the ith user of the target domain;
calculating the cosine similarity between the ith user quadratic regression parameter of the kth class of the target domain and the jth user quadratic regression parameter of the kth class of the source domain according to the following formula:
Figure BDA0003624110630000051
in the formula (I), the compound is shown in the specification,
Figure BDA0003624110630000052
the quadratic regression parameters of the ith user in the source domain and the jth user in the target domainThe second regression parameter of (2).
Further, the obtaining of the probability distribution of the adjustable load adjustment capability of the target domain according to the cosine similarity includes:
according to the cosine similarity, calculating an adjustable load adjustment capacity pseudo label of a target domain user:
Figure BDA0003624110630000053
where N is the number of users, cos θ k,i,j In order to be the cosine similarity, the similarity between the cosine and the cosine is calculated,
Figure BDA0003624110630000054
the actual rate of adjustment for the adjustable load of the source domain user,
Figure BDA0003624110630000055
a load regulation capability adjustable pseudo label for a target domain user;
converting the pseudo label of the adjustable load regulation capacity of the target domain user into probability distribution:
Figure BDA0003624110630000056
in the formula (I), the compound is shown in the specification,
Figure BDA0003624110630000057
the load regulation capability probability distribution may be adjusted for the target domain.
Further, the aligning the energy use behavior characteristics of the source domain and the target domain users according to the local maximum average difference based on the probability distribution of the target domain adjustable load adjustment capability includes:
the unbiased estimate of the local maximum mean difference from the regenerated hilbert spatial kernel function is:
Figure BDA0003624110630000061
Figure BDA0003624110630000062
in the formula, p and q are response data set distribution, K is clustering number, and D s To contain the source field of the actual adjusted label exemplars, D t To contain the target domain of the actual adjusted label sample,
Figure BDA0003624110630000063
respectively source domain samples i and target domain samples j,
Figure BDA0003624110630000064
respectively the weight of the ith user sample in the source domain belonging to k classes and the weight of the target domain and the jth user sample belonging to k classes, wherein l is the l layer of the full connection layer,
Figure BDA0003624110630000065
the load adjustment capability probability distribution can be adjusted for the target domain user,
Figure BDA0003624110630000066
respectively regenerating Hilbert space mapping functions for a source domain user and a target domain user;
performing feature alignment on the full connection layers of the source domain network model and the target domain network model to obtain unbiased estimation as follows:
Figure BDA0003624110630000067
wherein p and q are response data set distribution, K is cluster number, N is user number,
Figure BDA0003624110630000068
respectively the weight that the jth user sample of the source domain belongs to the k class and the weight that the ith user sample of the target domain belongs to the k class,
Figure BDA0003624110630000069
is the source domain full connection layer lThe characteristics of the layers are such that,
Figure BDA00036241106300000610
Figure BDA00036241106300000611
for the l-th layer feature of the target domain full-link layer,
Figure BDA00036241106300000612
and
Figure BDA00036241106300000613
is a kernel function.
Further, the online transfer learning network model loss function is as follows:
Figure BDA0003624110630000071
where N is the number of users, H is the cross entropy loss, x i 、R i 、c i Respectively the energy using behavior, uncertain factors and electricity price of the ith user, f is a probability forecasting label of the adjustable load adjustment quantity of the target domain,
Figure BDA0003624110630000072
an adjustable load actual adjustment quantity label of the ith user in a source domain, wherein lambda is an adaptive loss coefficient, L is the number of layers of a full connection layer, and d l ' (p, q) is the adaptive loss.
Compared with the prior art, the invention has the beneficial effects that:
one of the beneficial effects of the invention is that based on the current power market environment and diversified benefit requirements, various uncertain factors influencing user energy behaviors are calculated, a pseudo label of the adjustable load adjustment capability of the target domain user is obtained according to the cosine similarity of the parameter features between users in the class, and the associated features between the typical user energy behaviors and historical adjustment data are effectively extracted.
One of the benefits of the invention is that different sample weights are considered, the empirical probability distribution of the related subdomains of the source domain and the target domain is measured by using the local maximum average difference, a regenerated Hilbert space kernel function is embedded, and the local maximum average difference is used as an alignment index of online migration learning characteristics on the basis of user classification, so that the problem of negative migration influence in online migration learning is reduced.
The method has the advantages that the energy utilization behavior characteristics of the source domain and the target domain users are aligned according to the local maximum average difference, the estimation of the adjustable load regulation capacity of the target domain is realized by utilizing the online transfer learning based on the characteristic alignment, the new energy consumption and the dynamic source load balance are facilitated, the safety and the stability of the system are guaranteed, the theoretical reference and the technical support are provided for the demand side management, the operation risk is effectively reduced, and the system energy efficiency is improved.
Drawings
FIG. 1 is a flowchart of an evaluation method for adjustable load regulation capacity for power grid regulation and control requirements according to an embodiment of the present invention;
fig. 2 is a flowchart of an online transfer learning network model training according to an embodiment of the present invention.
Detailed Description
The invention is further described with reference to specific examples. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.
As mentioned above, the adjustable load resource on the demand side is brought into the operation regulation and control of the normalized power grid, the power grid is pushed to be changed from 'source load following movement' to 'source load interaction', and the evaluation of the adjustable load regulation capacity is very urgent and necessary.
To this end, an embodiment of the present invention provides a method for evaluating an adjustable load regulation capability facing a power grid regulation demand, as shown in fig. 1, where the method includes:
step 11, acquiring user energy data to be evaluated at a demand side, wherein the user energy data comprises user energy behaviors, uncertain factors influencing the user energy behaviors and electricity prices;
and step 12, inputting the energy data of the user to be evaluated into the trained online transfer learning network model, and outputting the adjustable load adjustment ratio of the user, namely the percentage of the adjustment amount in the total load, by the model.
The online transfer learning network model comprises a source domain network model and a target domain network model, wherein the source domain network model and the target domain network model are identical in structure and are formed by sequentially connecting an input layer, an M-layer convolution layer, an L-layer full-connection layer and an output layer.
As shown in fig. 2, the online transfer learning network model is trained by the following method:
step 21, acquiring historical energy data of a source domain and historical energy data of a target domain;
the method comprises the steps of obtaining load resource data characteristics of a demand side, carrying out longitudinal layering based on a power grid structure, combining a heuristic dynamic partitioning method to carry out transverse partitioning and resource aggregation, and partitioning the demand side into a plurality of layered partitions. And selecting one of the hierarchical partitions as a target domain to acquire historical energy consumption data of the target domain.
In this embodiment, the target domain sample contains a small amount of historical energy data of intelligent buildings, industrial and commercial users and residents, and the used source domain sample contains a large amount of historical energy data of buildings. Wherein the target domain and the source domain energy consumption data respectively comprise user energy consumption behaviors, a plurality of uncertain factors influencing the user energy consumption behaviors and electricity prices.
Step 22, respectively constructing a source domain adjustable load generalized model and a target domain adjustable load generalized model according to the user energy consumption behaviors in the source domain data and the target domain data and uncertain factors influencing the user energy consumption behaviors, and respectively fitting based on the source domain adjustable load generalized model and the target domain adjustable load generalized model to obtain a quadratic regression dynamic relation between the source domain user energy consumption behaviors and the target domain user energy consumption behaviors and the multiple uncertain factors;
the adjustable load generalized model is used for describing the relation between the energy using behavior of the user and the change of the uncertain factors. To reflect the energy usage curve of a generalized adjustable load over time, the energy usage curve can be represented by the following equation:
P t ＝P b,t (R m )+ΔP t (ΔR)＝P b,t (R m )(1+f t (ΔR))
in the formula (I), the compound is shown in the specification,P t the load can be adjusted for time t; p b,t (R m ) The load baseline load can be adjusted for time t, which is associated with an uncertainty factor R that influences the user's energy behavior m (ii) related; delta P t (Δ R) an adjustable load change value for time t, which is related to an uncertainty factor Δ R affecting the user energy behaviour; r m And Δ R are respectively the average value and the variation value (such as meteorological data) of the uncertain factors of a certain period of time; f. of t (Delta R) is P b,t (R m ) And Δ P t The function of (Δ R) is closely related to the uncertainty factor.
For a user sensitive to influence of uncertain factors, the relationship between the adjustable load power change and the uncertain factors can be described through a quadratic regression equation, so that the quadratic regression dynamic relationship between the fitting energy consumption behavior and the various uncertain factors is as follows:
f t (ΔR)＝f 1,t +f 2,t ΔR+f 3,t ΔR 2
in the formula (f) 1,t 、f 2,t 、f 3,t Are parameters of quadratic regression.
Step 23, clustering the historical energy consumption data of the source domain and the target domain respectively to extract similar energy consumption behaviors among different users, and obtaining secondary regression parameters of the energy consumption behaviors of the source domain and the target domain users and uncertain factors respectively according to the clustering results of the source domain and the target domain and the secondary regression dynamic relation between the energy consumption behaviors of the source domain and the target domain users and the uncertain factors;
and respectively extracting similar energy consumption behaviors among different users from historical energy consumption data of the source domain and the target domain by using a K clustering algorithm.
And considering the adjustable load characteristic, selecting K sample points, wherein each sample point represents the initial clustering center of each cluster, calculating the distance from the rest samples to the clustering center, assigning the distance to the nearest cluster, recalculating the average value of each cluster, continuously and repeatedly adjusting the clustering center in the whole process, and when the clustering center is not changed, converging the clusters formed by data clustering and outputting the energy-using behaviors of K types of users. The method specifically comprises the following steps:
(1) randomly selecting K user energy historical data as an initial clustering center;
(2) calculating the distance of the remaining samples to the cluster center and assigning the samples to the nearest cluster:
Figure BDA0003624110630000101
wherein K is the number of clusters, v ik Indicates whether the ith user sample belongs to class k, v ik A value of 1 indicates belonging to class k, v ik A value of 0 indicates that it does not belong to class k, d (c) m ,x i ) Is a sample x i To the center of the cluster c m The distance of (d);
(3) updating a clustering center:
Figure BDA0003624110630000111
in the formula, N is the number of users;
(4) the convergence condition is judged so that the following expression is minimized:
Figure BDA0003624110630000112
step 24, calculating the cosine similarity of the quadratic regression parameter characteristics of the energy consumption behaviors of the users in the target domain and the source domain in the class and the uncertain factors, and obtaining the probability distribution of the adjustable load regulation capacity of the target domain according to the cosine similarity;
the similarity of the quadratic regression parameter characteristic cosine of the energy consumption behaviors of the users in the target domain and the source domain in the class and the uncertain factors is calculated according to the following steps:
the set of quadratic regression parameters of the kth class ith user energy consumption behaviors and uncertain factors of the source domain is as follows:
Figure BDA0003624110630000113
in the formula (I), the compound is shown in the specification,
Figure BDA0003624110630000114
the adjustable load baseline load for the ith user of the source domain,
Figure BDA0003624110630000115
Figure BDA0003624110630000116
respectively are quadratic regression parameters of the ith user of the source domain;
the quadratic regression parameter set of the kth class ith user energy consumption behavior and the uncertain factors of the target domain is as follows:
Figure BDA0003624110630000117
in the formula (I), the compound is shown in the specification,
Figure BDA0003624110630000118
the adjustable load baseline load for the ith user of the target domain,
Figure BDA0003624110630000119
Figure BDA00036241106300001110
the quadratic regression parameter of the ith user of the target domain;
calculating the cosine similarity between the ith user quadratic regression parameter of the kth class of the target domain and the jth user quadratic regression parameter of the kth class of the source domain according to the following formula:
Figure BDA0003624110630000121
in the formula (I), the compound is shown in the specification,
Figure BDA0003624110630000122
the quadratic regression parameters of the ith user in the source domain and the quadratic regression parameters of the jth user in the target domain are respectively.
The method comprises the following steps of obtaining probability distribution of adjustable load regulation capacity of a target domain according to cosine similarity, and specifically comprises the following steps:
according to the cosine similarity, calculating a pseudo label of the adjustable load regulation capacity of the target domain user:
Figure BDA0003624110630000123
where N is the number of users, cos θ k,i,j In order to be the cosine similarity, the similarity between the cosine and the cosine is calculated,
Figure BDA0003624110630000124
the actual rate of adjustment for the adjustable load of the source domain user,
Figure BDA0003624110630000125
a load regulation capability adjustable pseudo label for a target domain user;
converting the pseudo label of the adjustable load regulation capacity of the target domain user into probability distribution:
Figure BDA0003624110630000126
in the formula (I), the compound is shown in the specification,
Figure BDA0003624110630000127
the load regulation capability probability distribution may be adjusted for the target domain.
And 25, aligning the energy consumption behavior characteristics of the source domain and the target domain users according to the local maximum average difference based on the probability distribution of the adjustable load regulation capacity of the target domain, and minimizing the loss function by continuously training the model until the model converges.
The method specifically comprises the following steps:
the unbiased estimate of the local maximum mean difference from the regenerated hilbert spatial kernel function is:
Figure BDA0003624110630000131
Figure BDA0003624110630000132
in the formula, p and q are response data set distribution, K is clustering number, and D s To contain the source field of the actual adjusted label exemplars, D t To include the target domain of the actual adjusted label sample,
Figure BDA0003624110630000133
respectively source domain samples i and target domain samples j,
Figure BDA0003624110630000134
respectively the weight of the ith user sample in the source domain belonging to the k class and the weight of the target domain and the jth user sample belonging to the k class, wherein l is the l layer of the full connection layer,
Figure BDA0003624110630000135
the load adjustment capability probability distribution can be adjusted for the target domain user,
Figure BDA0003624110630000136
respectively regenerating Hilbert space mapping functions for a source domain user and a target domain user;
performing feature alignment on the full connection layers of the source domain network model and the target domain network model to obtain unbiased estimation as follows:
Figure BDA0003624110630000137
wherein p and q are response data set distribution, K is cluster number, N is user number,
Figure BDA0003624110630000138
respectively the weight that the jth user sample of the source domain belongs to the k class and the weight that the ith user sample of the target domain belongs to the k class,
Figure BDA0003624110630000139
is the first layer of the source domain full connection layerThe step of performing the sign operation,
Figure BDA00036241106300001310
Figure BDA0003624110630000141
for the l-th layer feature of the target domain full-link layer,
Figure BDA0003624110630000142
and
Figure BDA0003624110630000143
is a kernel function.
The online transfer learning network model loss function is as follows:
Figure BDA0003624110630000144
where N is the number of users, H is the cross entropy loss, x i 、R i 、c i Respectively the energy using behavior, uncertain factors and electricity price of the ith user, f is a probability forecasting label of the adjustable load adjustment quantity of the target domain,
Figure BDA0003624110630000145
the source domain ith user is an adjustable load actual adjustment quantity label, lambda is an adaptive loss coefficient, and L is the number of layers of a full connection layer, d' l (p, q) is the adaptive loss.
And repeating the steps, and minimizing the loss function by continuously training the model until the model converges to obtain the online transfer learning network model with aligned features. The model can be used for evaluating the adjustable load regulation capacity of the target domain.
The present invention has been disclosed in terms of the preferred embodiment, but is not intended to be limited to the embodiment, and all technical solutions obtained by substituting or converting equivalents thereof fall within the scope of the present invention.

Claims (9)

1. A power grid regulation demand oriented adjustable load regulation capacity assessment method is characterized by comprising the following steps:
acquiring user energy data to be evaluated at a demand side, wherein the user energy data comprises user energy behaviors, uncertain factors and electricity prices;
and inputting the energy data of the user to be evaluated into the trained online transfer learning network model, and outputting the adjustable load adjustment ratio of the user by the model.
2. The method for evaluating the adjustable load regulation capacity of the power grid regulation demand oriented system according to claim 1, wherein the online transfer learning network model is obtained by training through the following method:
acquiring historical energy data of a source domain and historical energy data of a target domain;
respectively constructing a source domain adjustable load generalized model and a target domain adjustable load generalized model according to user energy consumption behaviors in source domain data and target domain data and uncertain factors influencing the user energy consumption behaviors, and fitting based on the source domain adjustable load generalized model and the target domain adjustable load generalized model to obtain a quadratic regression dynamic relation between the source domain user energy consumption behaviors and the target domain user energy consumption behaviors and the uncertain factors;
clustering historical energy consumption data of a source domain and a target domain respectively to extract similar energy consumption behaviors among different users, and obtaining secondary regression parameters of the energy consumption behaviors and the uncertain factors of the similar source domain and the target domain users respectively according to clustering results of the source domain and the target domain and secondary regression dynamic relations of the energy consumption behaviors and the uncertain factors of the source domain and the target domain users;
calculating the cosine similarity of the quadratic regression parameter characteristics of the energy consumption behaviors of the users in the target domain and the source domain and the uncertain factors, and obtaining the probability distribution of the adjustable load regulation capacity of the target domain according to the cosine similarity;
based on the probability distribution of the adjustable load regulation capacity of the target domain, aligning the energy use behavior characteristics of the source domain and the target domain users according to the local maximum average difference, and minimizing the loss function through continuously training the model until the model converges.
3. The method for evaluating the adjustable load regulation capacity of the power grid regulation demand-oriented system according to claim 2, wherein the adjustable load generalized model is as follows:
P t ＝P b,t (R m )+ΔP t (ΔR)＝P b,t (R m )(1+f t (ΔR))
in the formula, P t The load can be adjusted for time t; p is b,t (R m ) The load baseline load can be adjusted for time t, which is associated with an uncertainty factor R that affects the user's energy behavior m (ii) related; delta P t (Δ R) an adjustable load change value for time t, which is related to an uncertainty factor Δ R affecting the user energy behaviour; r m And Δ R are respectively the average value and the variation value of the uncertain factors of a certain period of time; f. of t (Δ R) is P b,t (R m ) And Δ P t (Δ R).
4. The method for evaluating the adjustable load regulation capacity facing the power grid regulation demand according to claim 3, wherein the quadratic regression dynamic relationship between the user energy consumption behavior and various uncertain factors is as follows:
f t (ΔR)＝f 1,t +f 2,t ΔR+f 3,t ΔR 2
in the formula (f) 1,t 、f 2,t 、f 3,t Are parameters of quadratic regression.
5. The method for evaluating the adjustable load regulation capacity facing the power grid regulation demand as claimed in claim 2, wherein the clustering of the historical energy consumption data of the source domain and the target domain to extract similar energy consumption behaviors among different users comprises:
randomly selecting K user energy historical data as an initial clustering center;
calculating the distance of the remaining samples to the cluster center and assigning the samples to the nearest cluster:
Figure FDA0003624110620000031
wherein K is the number of clusters, v ik Indicates whether the ith user sample belongs to class k, v ik A value of 1 indicates belonging to class k, v ik A value of 0 indicates that it does not belong to class k, d (c) m ,x i ) Is a sample x i To the center of the cluster c m The distance of (a);
updating a clustering center:
Figure FDA0003624110620000032
in the formula, N is the number of users;
the convergence condition is judged so that the following expression is minimized:
Figure FDA0003624110620000033
6. the method for evaluating the adjustable load regulation capacity facing the power grid regulation and control demand is characterized in that the quadratic regression parameter characteristic cosine similarity between the energy consumption behavior of the users in the target domain and the source domain and the uncertain factors is calculated according to the following steps:
the set of quadratic regression parameters of the kth class ith user energy consumption behaviors and uncertain factors of the source domain is as follows:
Figure FDA0003624110620000034
in the formula (I), the compound is shown in the specification,
Figure FDA0003624110620000035
the adjustable load baseline load for the ith user of the source domain,
Figure FDA0003624110620000036
Figure FDA0003624110620000037
respectively are quadratic regression parameters of the ith user of the source domain;
the quadratic regression parameter set of the kth class ith user energy consumption behavior and the uncertain factors of the target domain is as follows:
Figure FDA0003624110620000041
in the formula (I), the compound is shown in the specification,
Figure FDA0003624110620000042
the adjustable load baseline load for the ith user of the target domain,
Figure FDA0003624110620000043
Figure FDA0003624110620000044
the quadratic regression parameter of the ith user of the target domain;
calculating the cosine similarity between the ith user quadratic regression parameter of the kth class of the target domain and the jth user quadratic regression parameter of the kth class of the source domain according to the following formula:
Figure FDA0003624110620000045
in the formula (I), the compound is shown in the specification,
Figure FDA0003624110620000046
the quadratic regression parameters of the ith user in the source domain and the quadratic regression parameters of the jth user in the target domain are respectively.
7. The method for evaluating the adjustable load regulation capacity of the power grid regulation demand oriented, according to claim 6, wherein the obtaining of the probability distribution of the adjustable load regulation capacity of the target domain according to the cosine similarity comprises:
according to the cosine similarity, calculating an adjustable load adjustment capacity pseudo label of a target domain user:
Figure FDA0003624110620000047
where N is the number of users, cos θ k,i,j In order to be the cosine similarity, the similarity between the cosine and the cosine is calculated,
Figure FDA0003624110620000048
the actual rate of adjustment for the adjustable load of the source domain user,
Figure FDA0003624110620000049
a load regulation capability adjustable pseudo label for a target domain user;
converting the pseudo label of the adjustable load regulation capacity of the target domain user into probability distribution:
Figure FDA0003624110620000051
in the formula (I), the compound is shown in the specification,
Figure FDA0003624110620000052
the load regulation capability probability distribution may be adjusted for the target domain.
8. The method for evaluating the adjustable load regulation capacity oriented to the power grid regulation demand according to claim 7, wherein the aligning the energy use behavior characteristics of the source domain and the target domain users according to the local maximum average difference based on the target domain adjustable load regulation capacity probability distribution comprises:
the unbiased estimate of the local maximum mean difference from the regenerated hilbert spatial kernel function is:
Figure FDA0003624110620000053
Figure FDA0003624110620000054
in the formula, p and q are response data set distribution, K is clustering number, and D s To contain the source field of the actual adjusted label exemplars, D t To include the target domain of the actual adjusted label sample,
Figure FDA0003624110620000055
respectively source domain samples i and target domain samples j,
Figure FDA0003624110620000056
respectively the weight of the ith user sample in the source domain belonging to k classes and the weight of the target domain and the jth user sample belonging to k classes, wherein l is the l layer of the full connection layer,
Figure FDA0003624110620000057
the load adjustment capability probability distribution can be adjusted for the target domain user,
Figure FDA0003624110620000058
respectively regenerating Hilbert space mapping functions for a source domain user and a target domain user;
and carrying out feature alignment on the full connection layers of the network models of the source domain and the target domain to obtain an unbiased estimation as follows:
Figure FDA0003624110620000061
wherein p and q are response data set distribution, K is cluster number, N is user number,
Figure FDA0003624110620000062
respectively belonging to k classes for the jth user sample in the source domain and the ith user sample in the target domainThe weight for which an individual user sample belongs to class k,
Figure FDA0003624110620000063
for the l-th layer feature of the source domain full-link layer,
Figure FDA0003624110620000064
Figure FDA0003624110620000065
for the l-th layer feature of the target domain full-link layer,
Figure FDA0003624110620000066
and
Figure FDA0003624110620000067
is a kernel function.
9. The method for evaluating the adjustable load regulation capacity of the power grid regulation demand oriented system according to claim 8, wherein the online transfer learning network model loss function is as follows:
Figure FDA0003624110620000068
where N is the number of users, H is the cross entropy loss, x i 、R i 、c i Respectively the energy using behavior, uncertain factors and electricity price of the ith user, f is a probability forecasting label of the adjustable load adjustment quantity of the target domain,
Figure FDA0003624110620000069
an adjustable load actual adjustment quantity label of the ith user in a source domain, wherein lambda is an adaptive loss coefficient, L is the number of layers of a full connection layer, and d l ' (p, q) is the adaptive loss.
CN202210465933.2A 2022-04-29 2022-04-29 Power grid regulation and control demand-oriented adjustable load regulation capacity evaluation method Pending CN114844048A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210465933.2A CN114844048A (en) 2022-04-29 2022-04-29 Power grid regulation and control demand-oriented adjustable load regulation capacity evaluation method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210465933.2A CN114844048A (en) 2022-04-29 2022-04-29 Power grid regulation and control demand-oriented adjustable load regulation capacity evaluation method

Publications (1)

Publication Number Publication Date
CN114844048A true CN114844048A (en) 2022-08-02

Family

ID=82567710

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210465933.2A Pending CN114844048A (en) 2022-04-29 2022-04-29 Power grid regulation and control demand-oriented adjustable load regulation capacity evaluation method

Country Status (1)

Country Link
CN (1) CN114844048A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117132022A (en) * 2023-10-20 2023-11-28 江苏瑞问科技有限公司 Digital power grid intelligent management system and method based on dynamic load

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117132022A (en) * 2023-10-20 2023-11-28 江苏瑞问科技有限公司 Digital power grid intelligent management system and method based on dynamic load
CN117132022B (en) * 2023-10-20 2023-12-29 江苏瑞问科技有限公司 Digital power grid intelligent management system and method based on dynamic load

Similar Documents

Publication Publication Date Title
CN107301472B (en) Distributed photovoltaic planning method based on scene analysis method and voltage regulation strategy
US11581740B2 (en) Method, system and storage medium for load dispatch optimization for residential microgrid
CN107947165A (en) A kind of power distribution network flexibility evaluation method towards regulatory demand
CN109359389A (en) City electric car charging decision method based on typical load dynamic game
CN108964050A (en) Micro-capacitance sensor dual-layer optimization dispatching method based on Demand Side Response
CN102999791A (en) Power load forecasting method based on customer segmentation in power industry
CN110429649A (en) Consider the high permeability renewable energy assemblage classification method of flexibility
CN106410781A (en) Power consumer demand response potential determination method
CN107122924A (en) A kind of intelligent distribution system bearing capacity evaluation method
CN112418496B (en) Power distribution station energy storage configuration method based on deep learning
CN104281986A (en) Micro-grid power prediction method
CN117272850B (en) Elastic space analysis method for safe operation scheduling of power distribution network
CN114498629A (en) New energy consumption-oriented user load alignment demand response method and device
CN110852495A (en) Site selection method for distributed energy storage power station
CN115411777A (en) Power distribution network flexibility evaluation and resource allocation method and system
CN109118120A (en) Consider the Multiobjective Decision Making Method of Reservoir Operation Scheme Substantial evaluation
CN114844048A (en) Power grid regulation and control demand-oriented adjustable load regulation capacity evaluation method
CN117353276A (en) Distributed energy cooperative control method and system
CN113991640B (en) Thermal power-based multi-energy complementary energy base energy configuration planning method
CN104036431A (en) Interactive multilevel decision method of comprehensive evaluation in power quality based on cloud model
CN109375506B (en) Cloud service resource accurate control method based on RBF neural network
CN118017573A (en) Source network charge storage resource planning method considering flexible supply and demand balance
Jin et al. Short-term net feeder load forecasting of microgrid considering weather conditions
CN117353399A (en) Uncertainty-considered AC/DC hybrid micro-grid flexibility assessment method
CN117013531A (en) Capacity domain assessment method of power distribution network

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