CN117254472A - Power system probability power flow calculation method and system based on data fusion - Google Patents

Power system probability power flow calculation method and system based on data fusion Download PDF

Info

Publication number
CN117254472A
CN117254472A CN202311523012.8A CN202311523012A CN117254472A CN 117254472 A CN117254472 A CN 117254472A CN 202311523012 A CN202311523012 A CN 202311523012A CN 117254472 A CN117254472 A CN 117254472A
Authority
CN
China
Prior art keywords
component
time period
probability
time sequence
next adjacent
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
CN202311523012.8A
Other languages
Chinese (zh)
Other versions
CN117254472B (en
Inventor
陈爽
曾海燕
杨玺
高镇
彭子睿
王安龙
刘博特
张少谦
李航
林放
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.)
Wuhan Power Supply Co of State Grid Hubei Electric Power Co Ltd
Original Assignee
Wuhan Power Supply Co of State Grid Hubei Electric Power Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Wuhan Power Supply Co of State Grid Hubei Electric Power Co Ltd filed Critical Wuhan Power Supply Co of State Grid Hubei Electric Power Co Ltd
Priority to CN202311523012.8A priority Critical patent/CN117254472B/en
Publication of CN117254472A publication Critical patent/CN117254472A/en
Application granted granted Critical
Publication of CN117254472B publication Critical patent/CN117254472B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • H02J3/04—Circuit arrangements for ac mains or ac distribution networks for connecting networks of the same frequency but supplied from different sources
    • H02J3/06—Controlling transfer of power between connected networks; Controlling sharing of load between connected networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/18—Network design, e.g. design based on topological or interconnect aspects of utility systems, piping, heating ventilation air conditioning [HVAC] or cabling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/28—Design optimisation, verification or simulation using fluid dynamics, e.g. using Navier-Stokes equations or computational fluid dynamics [CFD]
    • 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
    • G06N3/047—Probabilistic or stochastic 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
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/04—Constraint-based CAD
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/08—Probabilistic or stochastic CAD
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/10—Numerical modelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2113/00—Details relating to the application field
    • G06F2113/04—Power grid distribution networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2113/00—Details relating to the application field
    • G06F2113/06—Wind turbines or wind farms
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2123/00—Data types
    • G06F2123/02—Data types in the time domain, e.g. time-series data
    • 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
    • H02J2203/10—Power transmission or distribution systems management focussing at grid-level, e.g. load flow analysis, node profile computation, meshed network optimisation, active network management or spinning reserve management
    • 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]

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Geometry (AREA)
  • Artificial Intelligence (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Computer Hardware Design (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Pure & Applied Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Mathematical Optimization (AREA)
  • Biophysics (AREA)
  • Mathematical Analysis (AREA)
  • Biomedical Technology (AREA)
  • Health & Medical Sciences (AREA)
  • Computational Mathematics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Fluid Mechanics (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Power Engineering (AREA)
  • Algebra (AREA)
  • Medical Informatics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Supply And Distribution Of Alternating Current (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The application relates to the field of power systems, in particular to a power system probability power flow calculation method and system based on data fusion, wherein the method comprises the following steps: acquiring a system state of a power system in a current time period, wherein the system state comprises a time sequence of photovoltaic output, wind output and load in the current time period; and inputting the system state into a trained probability power flow model, and outputting a probability power flow calculation result of the next adjacent time period, wherein the probability power flow calculation result comprises the mean value and the variance of node voltage and branch power in the next adjacent time period. By the technical scheme, the accuracy of the probability power flow calculation result in the power system can be improved.

Description

Power system probability power flow calculation method and system based on data fusion
Technical Field
The present disclosure relates generally to the field of power systems, and in particular, to a method and a system for calculating probability power flow of a power system based on data fusion.
Background
The probability power flow calculation can reflect the influence of random variation of various factors in the power system on the system operation, comprehensively considers the uncertain conditions of variable variation of a network topology structure, element parameters, node loads and the like of the power system, and is an important tool for power system planning and safe and reliable analysis.
With the rise of new energy sources such as photovoltaic power generation and wind power generation, a large amount of power output by the new energy sources is converged in a power system, but the output power of the photovoltaic power generation is greatly influenced by the intensity of solar radiation, the output power of the wind power generation is greatly influenced by wind speed, the uncertainty of variable change in the power system is enhanced, and how to obtain accurate probability trend calculation results and determine the distribution condition of power consumption parameters of all nodes and branches in the power system is a problem to be solved urgently.
At present, the patent application document with publication number of CN108336739A discloses a probability power flow on-line calculation method based on RBF neural network, and adopts a Monte Carlo method or an improved Monte Carlo method to sample random variables (wind speed, illumination radiance and load) of a power system for calculating the probability power flow so as to obtain a calculation sample; inputting the calculation sample into the RBF neural network probability power flow model after training, and judging the power flow solvability of the calculation sample; when the calculation sample is solvable, taking the mean value, variance and probability distribution of the RBF neural network probability power flow model output variables as probability power flow calculation results, wherein the output variables comprise voltage amplitude values and phase angles of all nodes of the power system, and active power and reactive power of each branch.
However, the method solves the problem that the probability power flow calculation result can be obtained, but the random variable of the power system is continuously changed at different moments, and the influence of the change of the random variable in different time periods on the electrical parameters in the power system is ignored, so that the probability power flow calculation result is inaccurate.
Disclosure of Invention
In order to solve the technical problems of the application, the application provides a power system probability power flow calculation method and system based on data fusion, and accuracy of a probability power flow calculation result in a power system is improved.
According to a first aspect of the present application, there is provided a power system probability power flow calculation method based on data fusion, including: acquiring a system state of a power system in a current time period, wherein the system state comprises a time sequence of photovoltaic output, wind output and load in the current time period; inputting the system state into a trained probability power flow model, and outputting a probability power flow calculation result of the next adjacent time period, wherein the probability power flow calculation result comprises the mean value and the variance of node voltage and branch power in the next adjacent time period; the training method of the probabilistic power flow model comprises the following steps: obtaining structural characteristics and component connection vectors of each calculation target according to the topological structure of the power system, wherein the component connection vectors comprise connection relations between the corresponding calculation targets and components in the power system, and the calculation targets comprise nodes and branches; collecting system state samples, probability tide tags and component states of a plurality of continuous historical time periods, wherein the component states comprise the running state of each component of the power system; for a historical time period, inputting a system state sample into a probabilistic power flow model, and extracting time sequence characteristics of the system state sample to obtain time sequence characteristics of the historical time period, and predicting time sequence prediction characteristics of the next adjacent historical time period based on the time sequence characteristics; performing dimension transformation on the time sequence prediction characteristics to output probability power flow prediction results of the next adjacent historical time period; building a training auxiliary model, wherein in the training auxiliary model, the component state of the historical time period is updated based on a time sequence difference characteristic to output a component prediction state of the next adjacent historical time period, and the real-time characteristic of each calculation target in the next adjacent historical time period is built based on the structural characteristic, the component connection vector and the component prediction state of the calculation target, wherein the time sequence difference characteristic is the difference between the time sequence characteristic and the time sequence prediction characteristic; calculating a loss function value based on the probabilistic power flow model and the training auxiliary model, the loss function value satisfying a relation:
Wherein,for the timing prediction feature, < >>For the timing characteristics of the next adjacent history period, +.>For the probabilistic load flow prediction result, +.>Probability flow tag for next adjacent history period +.>Predicting a state for said component, +.>Component status for next adjacent history period,/-for next adjacent history period>And->Probability trend prediction results +.>Middle->Calculation target and->Mean and variance corresponding to each calculation target, +.>And->Respectively the first +.>Calculation target and->Real-time characteristics of individual calculation targets, +.>For all calculation targets, +.>Is a loss function value; updating the probabilistic tide model and the training auxiliary model by using a gradient descent method to finish one-time training; and iteratively training the probability power flow model until the loss function value is smaller than a set value, and obtaining the trained probability power flow model.
In one embodiment, the probabilistic power flow model comprises a time sequence feature extraction layer, a first full connection layer and a second full connection layer; the time sequence feature extraction layer is used for extracting time sequence features of the system state in the current time period and obtaining the time sequence features in the current time period; the first full-connection layer is used for carrying out first dimension transformation on the time sequence characteristics so as to predict time sequence prediction characteristics in the next adjacent time period; and the second full-connection layer is used for carrying out second dimension transformation on the time sequence prediction characteristics and outputting probability power flow calculation results of the next adjacent time period.
In one embodiment, the timing feature is M rows and 1 columns in size; the first full-connection layer comprises a first input layer, a first hiding layer and a first output layer; the first input layer comprises M neurons and is used for receiving the time sequence characteristics output by the time sequence characteristic extraction layer; the number of neurons in the first hidden layer is not equal to M, and the neurons are used for carrying out dimension transformation on the time sequence characteristics; the first output layer comprises M neurons, and is used for carrying out dimension transformation on the output result of the first hidden layer again and outputting time sequence prediction features in the next adjacent time period, wherein the size of the time sequence prediction features is M rows and 1 column.
In one embodiment, the second fully-connected layer includes a second input layer, at least one second hidden layer, and a second output layer; the second input layer comprises M neurons and is used for receiving the time sequence prediction characteristics output by the first full-connection layer; the second hidden layers are used for carrying out multiple dimension transformation on the time sequence prediction characteristics, wherein the number of dimension transformation times is the same as the number of the second hidden layers; the second output layer comprises 2N neurons and is used for carrying out dimension transformation on the output result of the second hidden layer again and outputting the probability power flow calculation result of the next adjacent time period; and N is the total number of all nodes and branches in the power system, the size of a probability power flow calculation result is 2N rows and 1 column, one node corresponds to the mean value and variance of node voltage, and one branch corresponds to the mean value and variance of branch power.
In one embodiment, obtaining structural features and component connection vectors for each computing target according to a topology of the power system includes: for one node in the topological structure, acquiring an embedded vector of the node according to a graph embedding algorithm, wherein the embedded vector corresponds to the structural characteristics of the node; for one component in the topological structure, marking the connection state of the component as 1 in response to the connection of the component to the node, and marking the connection state of the component as 0 in response to the disconnection of the component to the node; traversing all components in the topological structure to obtain component connection vectors of the nodes, wherein the component connection vectors comprise connection states of all components; and for one branch in the topological structure, acquiring all nodes connected to the branch, taking the sum of structural features of all nodes as the structural features of the branch, and taking the union of component connection vectors of all nodes as the component connection vector of the branch.
In one embodiment, the training assistance model comprises a third fully connected layer comprising a third input layer, at least one third hidden layer, and at least one third output layer; the third input layer comprises B neurons for receiving the states of components in a historical time period, wherein B is the total number of all components in the power system; the third hiding layer performs multiple dimension reduction operations on the component state to obtain a component dimension reduction state, wherein the dimension of the component dimension reduction state is equal to the dimension of the time sequence difference characteristic, and the sum of the component dimension reduction state and the time sequence difference characteristic is used as an output result; the third output layer is used for carrying out multi-time dimension lifting operation on the output result of the third hidden layer to obtain the predicted state of the component in the next adjacent historical time period; the number of dimension-reducing operations is the same as the number of the third hidden layers, and the number of dimension-increasing operations is the same as the number of the third output layers.
In one embodiment, the constructing the real-time feature of each calculation target in the next adjacent historical time period based on the structural feature of the calculation target, the component connection vector and the component prediction state includes: for a calculation target, acquiring a component connection vector of the calculation target, and multiplying the component connection vector by the component prediction state to obtain the component real-time characteristics of the calculation target; and splicing the structural features of the calculation target with the real-time features of the components to obtain the real-time features of the calculation target in the next adjacent historical time period.
In one embodiment, the timing characteristics of the next adjacent historical periodThe acquisition method of (1) comprises the following steps: and inputting the system state samples of the next adjacent historical time period into the time sequence feature extraction layer, and outputting the time sequence features of the next adjacent historical time period.
The application also provides a power system probability power flow calculation system based on data fusion, which comprises a processor and a memory, wherein the memory stores computer program instructions, and the computer program instructions are executed by the processor to realize the power system probability power flow calculation method based on the data fusion according to the first aspect of the application.
The technical scheme of the application has the following beneficial technical effects:
according to the technical scheme provided by the application, firstly, the static characteristics of each node and each branch are calculated according to the topological structure of the power system, and the static characteristics comprise structural characteristics and component connection vectors; in the process of training the probabilistic power flow model, acquiring time sequence characteristics of a historical time period according to a system state sample of the historical time period, and predicting time sequence prediction characteristics of a next adjacent historical time period, and extracting the time sequence prediction characteristics to predict a probabilistic power flow prediction result of the next adjacent historical time period; further, in the training auxiliary model, the component state of the historical time period is updated according to the difference value of the time sequence characteristic and the time sequence prediction characteristic, so that the component prediction state of the next adjacent historical time period is obtained, and the real-time characteristic of each node and branch in the next adjacent historical time period can be accurately obtained according to the component prediction state; further, in the same time period, the nodes and branches with the same real-time characteristics have the same numerical value in the probability power flow prediction result, so that the loss function value is calculated according to the time sequence prediction characteristics, the probability power flow prediction result and the real-time characteristics of each node and branch so as to restrict the probability power flow model to output an accurate probability power flow calculation result; and outputting a probability power flow calculation result of the next continuous time period according to the trained probability power flow model, and improving the accuracy of the probability power flow calculation result.
Drawings
The above, as well as additional purposes, features, and advantages of exemplary embodiments of the present application will become readily apparent from the following detailed description when read in conjunction with the accompanying drawings. Several embodiments of the present application are illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings and in which like reference numerals refer to similar or corresponding parts and in which:
FIG. 1 is a flow chart of a method of power system probabilistic power flow calculation based on data fusion according to an embodiment of the present application;
FIG. 2 is a schematic diagram of a probabilistic power flow model according to an embodiment of the present application;
FIG. 3 is a schematic structural view of a first fully connected layer according to an embodiment of the present application;
FIG. 4 is a schematic structural view of a second fully connected layer according to an embodiment of the present application;
FIG. 5 is a flow chart of a method of training a probabilistic power flow model in accordance with an embodiment of the present application;
FIG. 6 is a schematic diagram of the structure of a training assistance model according to an embodiment of the application;
FIG. 7 is a schematic structural view of a third fully connected layer according to an embodiment of the present application;
fig. 8 is a block diagram of a power system probabilistic power flow calculation system based on data fusion according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all, of the embodiments of the present application. All other embodiments, which can be made by those skilled in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.
It should be understood that when the terms "first," "second," and the like are used in the claims, specification, and drawings of this application, they are used merely for distinguishing between different objects and not for describing a particular sequential order. The terms "comprises" and "comprising," when used in the specification and claims of this application, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
According to a first aspect of the application, the application provides a power system probability power flow calculation method based on data fusion. Fig. 1 is a flowchart of a method for calculating a probability flow of a power system based on data fusion according to an embodiment of the present application. As shown in fig. 1, the power system probability power flow calculation method includes steps S101 to S102, which will be described in detail below.
S101, acquiring a system state of a power system in a current time period, wherein the system state comprises a time sequence of photovoltaic output, wind output and load in the current time period.
In one embodiment, the length of the time period may be preset, and in this embodiment, the length of the set time period is 1 hour, that is, a probability flow calculation result is output every hour. And collecting the time sequence of the photovoltaic output, the wind output and the load in the current time period to obtain the system state of the power system in the current time period, wherein the system state is used for reflecting the power supply condition and the power consumption condition in the power system. The photovoltaic output is output power of the photovoltaic power generation system which is input into the power system, and the wind output is output power of the wind power generation system which is input into the power system.
For example, if the time interval between adjacent data in the time sequence is set to 1 minute, the system state of the power system in the current time period is a matrix of 3 rows and 60 columns, and the 3 rows respectively correspond to the time sequence of the photovoltaic output, the wind output and the load.
S102, inputting the system state into a trained probability power flow model, and outputting a probability power flow calculation result of the next adjacent time period, wherein the probability power flow calculation result comprises the mean value and the variance of node voltage and branch power in the next adjacent time period.
In one embodiment, the system state is input into a trained probabilistic power flow model, and a probability power flow calculation result of the next adjacent time period is output, wherein the probability power flow calculation result comprises a variance and a mean value of each node voltage in the power system and a variance and a mean value of each branch power.
It can be appreciated that, for a node in the power system, the probability distribution of the node voltage in the next adjacent time period can be determined by obtaining the mean and variance of the node voltage in the next adjacent time period, where the probability distribution of the node voltage satisfies the gaussian distribution; similarly, the probability distribution of each branch power in the next adjacent time period can be determined.
In one embodiment, please refer to fig. 2, which is a schematic structural diagram of a probabilistic power flow model according to an embodiment of the present application. The probabilistic power flow model comprises a time sequence feature extraction layer 21, a first full-connection layer 22 and a second full-connection layer 23; the time sequence feature extraction layer 21 performs time sequence feature extraction on the system state of the current time period to acquire the time sequence feature in the current time period; the first full connection layer 22 performs a first dimension transformation on the time sequence characteristics to predict time sequence prediction characteristics in a next adjacent time period; the second full-connection layer 23 performs a second dimension transformation on the time sequence prediction feature, and outputs a probability power flow calculation result of the next adjacent time period, where the probability power flow calculation result includes a mean value and a variance of node voltage and branch power in the next adjacent time period.
The timing feature extraction layer 21 may adopt any one of the existing timing feature extraction networks such as LSTM, BERT, and Transformer, which is not limited in this application, and the size of the timing feature output by the timing feature extraction layer 21 is denoted as M rows and 1 columns, where M is related to the specific structure of the timing feature extraction layer 21.
Fig. 3 is a schematic structural diagram of a first full connection layer according to an embodiment of the present application. The first fully-connected layer 22 includes a first input layer 221, a first hidden layer 222, and a first output layer 223; the first input layer 221 includes M neurons for receiving the timing characteristics output from the timing characteristics extraction layer 21; the number of neurons in the first hidden layer 222 is not equal to M, and the number of neurons in the first hidden layer 222 is the same as the number of neurons in the first hidden layer 222; the number of neurons in the first output layer 223 is also M, which is used to dimension-transform the output result of the first hidden layer 222 again, and output the time sequence prediction feature in the next adjacent time period, where the size of the time sequence prediction feature is also M rows and 1 column. It should be noted that the number of neurons in the first hidden layer 222 is greater than or equal to M, which is not limited in this application.
Fig. 4 is a schematic structural diagram of a second full-connection layer according to an embodiment of the present application. The second fully-connected layer 23 includes a second input layer 231, at least one second hidden layer 232, and a second output layer 233; the second input layer 231 includes M neurons for receiving the timing prediction features output by the first fully-connected layer 22; the number of second hidden layers 232 and the number of neurons in each second hidden layer 232 are not limited, and are used for performing multiple dimension transformation on the time sequence feature, wherein the number of dimension transformation is the same as the number of second hidden layers 232; the number of neurons in the second output layer 233 is 2N, which is used to perform dimension transformation on the output result of the second hidden layer 232 again, and output the probability flow calculation result of the next adjacent time period, where N is the total number of all nodes and branches in the power system, the size of the probability flow calculation result is 2N rows and 1 column, one node corresponds to the mean value and variance of the node voltage, and one branch corresponds to the mean value and variance of the branch power.
Thus, based on the trained probabilistic power flow model, the probability power flow calculation result of the next adjacent time period can be predicted based on the system state of the power system in the current time period.
In one embodiment, fig. 5 is a flow chart of a method of training a probabilistic power flow model according to an embodiment of the present application. As shown in fig. 5, the training method includes steps S201 to S207, which will be described in detail below.
S201, obtaining structural features and component connection vectors of each calculation target according to the topological structure of the power system, wherein the component connection vectors comprise connection relations between the corresponding calculation targets and components in the power system, and the calculation targets comprise nodes and branches.
In one embodiment, the topology structure of the power system comprises all components such as a generator, a circuit breaker, a motor and the like and the connection relation among any components, wherein any components are connected through wires; in the electric power system, a section of electric wire without branches is called a branch, the intersection points of 3 or more branches are nodes, and all the nodes and branches of the electric power system can be obtained through the topological structure of the electric power system. It can be understood that, for a node in the topology structure, the connection relationship between the node and the components and other nodes in the power system can be obtained, i.e. the topology structure can be regarded as a graph structure, and the graph structure includes connection relationships between any nodes and between the nodes and the components.
Specifically, obtaining structural features and component connection vectors of each calculation target according to the topology of the power system includes: for one node in the topological structure, acquiring an embedded vector of the node according to a graph embedding algorithm, wherein the embedded vector corresponds to the structural characteristics of the node; for one component in the topological structure, marking the connection state of the component as 1 in response to the connection of the component to the node, and marking the connection state of the component as 0 in response to the disconnection of the component to the node; traversing all components in the topological structure to obtain component connection vectors of the nodes, wherein the component connection vectors comprise connection states of all components; and for one branch in the topological structure, acquiring all nodes connected to the branch, taking the sum of structural features of all nodes as the structural features of the branch, and taking the union of component connection vectors of all nodes as the component connection vector of the branch.
The graph embedding algorithm can adopt any existing graph embedding algorithm such as Deep Walk, node2vec and the like, and is used for extracting the embedding vector of each Node in the topological structure; the number of the component connection vectors is the same as the number of all components in the power system.
Illustratively, when the number of components in the power system is 5, the connection relationship between the node 1 and the components 2 and 4 exists, and the component connection vector of the node 1 isThe method comprises the steps of carrying out a first treatment on the surface of the If the connection relation exists between the node 2 and the components 3 and 4, the component connection vector of the node 2 is +.>The method comprises the steps of carrying out a first treatment on the surface of the Branch 1 is connected to node 1 and node 2, and the component connection vector of branch 1 is +.>。
The size of the structural feature is determined by the graph embedding algorithm. In this embodiment, the size of the structural feature is also M rows and 1 column, that is, the size of the structural feature is the same as the timing feature and the timing prediction feature. The structural characteristics and the component connection vectors of the calculation targets are only related to the topological structure of the power system, and once the topological structure is determined, the structural characteristics and the component connection vectors cannot be changed.
In this way, the structural characteristics and the component connection vectors of the calculation target are determined according to the topological structure of the power system, and the calculation target comprises nodes and branches; the structural features and the component connection vectors are only related to the topological structure, can be regarded as static features of a calculation target, and can participate in training of the probabilistic power flow model.
S202, collecting system state samples, probability tide tags and component states of a plurality of continuous historical time periods, wherein the component states comprise the running state of each component of the power system.
In one embodiment, the system state sample is a time series of photovoltaic output, wind output, and load over a historical period of time; the probability tide label is the mean value and variance of all node voltages and branch power in a historical time period; the component states include an operating state of each component of the power system, the operating state including normal and abnormal, the normal being represented by a value of "1" and the abnormal being represented by a value of "0".
In this way, the system state samples, the probability flow labels and the component states of a plurality of continuous historical time periods are used as data for training the probability flow model.
S203, for a historical time period, inputting a system state sample into a probabilistic power flow model, and extracting time sequence characteristics of the system state sample to obtain time sequence characteristics of the historical time period, and predicting time sequence prediction characteristics of the next adjacent historical time period based on the time sequence characteristics; and carrying out dimension transformation on the time sequence prediction characteristics to output a probability power flow prediction result of the next adjacent historical time period.
In one embodiment, the probabilistic power flow model is trained using the data acquired in step S202, and the network structure of the probabilistic power flow model is already described in step S102, which is not described herein. After a system state sample in one historical time period is input into a probabilistic power flow model, the time sequence characteristics of the historical time period, the time sequence prediction characteristics of the next adjacent historical time period and the probabilistic power flow prediction results of the next adjacent historical time period can be obtained.
S204, building a training auxiliary model, wherein in the training auxiliary model, the component state of the historical time period is updated based on a time sequence difference characteristic to output a component prediction state of the next adjacent historical time period, and the real-time characteristic of each calculation target in the next adjacent historical time period is built based on the structural characteristic, the component connection vector and the component prediction state of the calculation target, wherein the time sequence difference characteristic is the difference between the time sequence characteristic and the time sequence prediction characteristic.
In one embodiment, to assist in the training of the probabilistic power flow model, a training assist model is built before training the probabilistic power flow model. Because the state of the component can influence the state of the calculation target in the power system, the state of the component can be regarded as the dynamic characteristic of the calculation target; in order to improve the accuracy of probability calculation results, the states of components in the next adjacent historical time period need to be predicted, and then the states of targets in a power system are accurately calculated, and therefore a training auxiliary model is built.
FIG. 6 is a schematic diagram of a training assistance model according to an embodiment of the present application; the input of the training auxiliary model is a time sequence difference characteristic and the state of a component in a historical time period, and the output is a real-time characteristic of each calculation target in the next adjacent historical time period, wherein the time sequence difference characteristic is the difference between the time sequence characteristic and the time sequence prediction characteristic.
Specifically, the training auxiliary model includes a third full-connection layer 31, please refer to fig. 7, which is a schematic structural diagram of the third full-connection layer according to an embodiment of the present application, wherein an input of the third full-connection layer 31 is a component state of a time sequence difference feature and a history period, and an output is a component prediction state of a next adjacent history period. The third fully-connected layer 31 includes a third input layer 311, at least one third hidden layer 312, and at least one third output layer 313; the third input layer 311 includes B neurons, which are configured to receive a component state in a historical period, where B is the total number of all components in the power system; the third hiding layer 312 performs multiple dimension reduction operations on the component state to obtain a component dimension reduction state, wherein the dimension of the component dimension reduction state is equal to the dimension of the time sequence difference characteristic, and the sum of the component dimension reduction state and the time sequence difference characteristic is used as an output result; the third output layer 313 is configured to perform a plurality of dimension-lifting operations on the output result of the third hidden layer 312, so as to obtain the predicted state of the component in the next adjacent historical time period. The number of the multiple dimension-reducing operations is the same as the number of layers of the third hidden layer 312, and the number of the multiple dimension-increasing operations is the same as the number of layers of the third output layer 313.
In one embodiment, the predicted states of components include operational states of all components in a next adjacent historical time period. After the third full connection layer 31 outputs the predicted state of the component in the next adjacent history period, the real-time feature of each calculation target in the next adjacent history period can be calculated, which is described in detail below. The construction of the real-time feature of each calculation target in the next adjacent historical time period based on the structural feature of the calculation target, the component connection vector and the component prediction state comprises the following steps: for a calculation target, acquiring a component connection vector of the calculation target, and multiplying the component connection vector by the component prediction state to obtain the component real-time characteristics of the calculation target; and splicing the structural features of the calculation target with the real-time features of the components to obtain the real-time features of the calculation target in the next adjacent historical time period.
It can be understood that, in the power system, the impact of the system state changes (such as load changes, photovoltaic output changes and wind output changes) on the components in the adjacent time periods affects the operation state of the components, so in the third full connection layer 31, the component prediction state in the next adjacent historical time period is accurately predicted according to the time sequence difference characteristic in the next adjacent historical time period and the component state in the current historical time period, so as to obtain the real-time characteristic of the calculation target (node and branch) in the next adjacent historical time period.
Thus, based on the training auxiliary model, the real-time characteristics of each calculation target in the next adjacent historical time period are obtained, and the structural characteristics, the component connection vectors and the component states of the calculation targets in the next adjacent historical time period are integrated by the real-time characteristics of one calculation target, so that the state information of the calculation targets in the power system is accurately reflected.
S205, calculating a loss function value based on the probability power flow model and the training auxiliary model.
In one embodiment, the loss function value satisfies the relationship:
wherein,for the timing prediction feature, < >>For the timing characteristics of the next adjacent history period, +.>For the probabilistic load flow prediction result, +.>Probability flow tag for next adjacent history period +.>Predicting a state for said component, +.>Component status for next adjacent history period,/-for next adjacent history period>And->Probability trend prediction results +.>Middle->Calculation target and->Mean and variance corresponding to each calculation target, +.>And->Respectively the first +.>Calculation target and->Real-time characteristics of individual calculation targets, +.>For all calculation targets, +.>Is the loss function value.
Wherein the timing characteristics of the next adjacent historical periodThe acquisition method of (1) comprises the following steps: and inputting the system state samples of the next adjacent historical time period into the time sequence feature extraction layer, and outputting the time sequence features of the next adjacent historical time period.
Wherein,the time sequence prediction features output by the first full-connection layer in the constraint probability power flow model are the same as the time sequence features of the next adjacent historical time period, so that the accuracy of the time sequence prediction features is improved; />The probability flow prediction result output by the second full-connection layer in the constraint probability flow model is the same as the probability flow label of the next adjacent historical time period, so that the accuracy of the probability flow prediction result is improved; />The predicted state of the components output by the third full-connection layer in the constraint training auxiliary model is the same as the state of the components in the next adjacent historical time period, so that the accuracy of the predicted state of the components is improved; further, for the->Calculation target and->Individual calculation targets, real-time features->Andthe greater the similarity between the probability trend prediction results, the +.>Calculation target and->The more similar the mean variance of the individual calculation targets, i.e +.>For constraint of +. >Individual calculation targets and the thThe real-time characteristics and mean variances among the calculation targets have positive correlation, so that the accuracy of the probability power flow prediction result is further improvedWherein->，/>Is->Mean value corresponding to each calculation target, +.>Is->The variance corresponding to the target is calculated.
Thus, the construction of the loss function value is completed, and the accuracy of the time sequence prediction characteristic, the probability power flow prediction result and the component prediction state can be restrained by the loss function value.
S206, updating the probability tide model and the training auxiliary model by using a gradient descent method, and completing one-time training.
S207, the training is executed iteratively until the loss function value is smaller than a set value, and a trained probability power flow model is obtained.
In one embodiment, a system state sample of a historical time period is continuously input into a probabilistic power flow model, the states of components of the historical time period are input into a training auxiliary model, the probabilistic power flow model and the training auxiliary model are continuously updated, and the training auxiliary model is abandoned until the loss function value is smaller than a set value, so that a trained probabilistic power flow model is obtained.
Thus, a trained probability power flow model is obtained, the system state of one time period is input into the trained probability power flow model, and the probability power flow calculation result of the next adjacent time period can be obtained.
Technical principles and implementation details of the power system probability power flow calculation method based on data fusion are introduced through the specific embodiments. According to the technical scheme provided by the application, firstly, the static characteristics of each node and each branch are calculated according to the topological structure of the power system, and the static characteristics comprise structural characteristics and component connection vectors; in the process of training the probabilistic power flow model, acquiring time sequence characteristics of a historical time period according to a system state sample of the historical time period, and predicting time sequence prediction characteristics of a next adjacent historical time period, and extracting the time sequence prediction characteristics to predict a probabilistic power flow prediction result of the next adjacent historical time period; further, in the training auxiliary model, the component state of the historical time period is updated according to the difference value of the time sequence characteristic and the time sequence prediction characteristic, so that the component prediction state of the next adjacent historical time period is obtained, and the real-time characteristic of each node and branch in the next adjacent historical time period can be accurately obtained according to the component prediction state; further, in the same time period, the nodes and branches with the same real-time characteristics have the same numerical value in the probability power flow prediction result, so that the loss function value is calculated according to the time sequence prediction characteristics, the probability power flow prediction result and the real-time characteristics of each node and branch so as to restrict the probability power flow model to output an accurate probability power flow calculation result; and outputting a probability power flow calculation result of the next continuous time period according to the trained probability power flow model, and improving the accuracy of the probability power flow calculation result.
According to a second aspect of the application, the application further provides a power system probability power flow calculation system based on data fusion. Fig. 8 is a block diagram of a power system probabilistic power flow calculation system based on data fusion according to an embodiment of the present application. As shown in fig. 8, the apparatus 50 comprises a processor and a memory storing computer program instructions which, when executed by the processor, implement a method of power system probability flow calculation based on data fusion according to the first aspect of the present application. The device also includes other components, such as a communication bus and a communication interface, which are well known to those skilled in the art, and the arrangement and function of which are known in the art and therefore not described in detail herein.
In the context of this application, the foregoing memory may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. For example, the computer readable storage medium may be any suitable magnetic or magneto-optical storage medium, such as, for example, resistance change Memory RRAM (Resistive Random Access Memory), dynamic Random Access Memory DRAM (Dynamic Random Access Memory), static Random Access Memory SRAM (Static Random-Access Memory), enhanced dynamic Random Access Memory EDRAM (Enhanced Dynamic Random Access Memory), high-Bandwidth Memory HBM (High-Bandwidth Memory), hybrid storage cube HMC (Hybrid Memory Cube), etc., or any other medium that may be used to store the desired information and that may be accessed by an application, a module, or both. Any such computer storage media may be part of, or accessible by, or connectable to, the device. Any of the applications or modules described herein may be implemented using computer-readable/executable instructions that may be stored or otherwise maintained by such computer-readable media.
The technical features of the above-described embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above-described embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples only represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the claims. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application is to be determined by the claims appended hereto.

Claims (9)

1. The power system probability power flow calculation method based on data fusion is characterized by comprising the following steps of:
acquiring a system state of a power system in a current time period, wherein the system state comprises a time sequence of photovoltaic output, wind output and load in the current time period;
inputting the system state into a trained probability power flow model, and outputting a probability power flow calculation result of the next adjacent time period, wherein the probability power flow calculation result comprises the mean value and the variance of node voltage and branch power in the next adjacent time period;
The training method of the probabilistic power flow model comprises the following steps:
obtaining structural characteristics and component connection vectors of each calculation target according to the topological structure of the power system, wherein the component connection vectors comprise connection relations between the corresponding calculation targets and components in the power system, and the calculation targets comprise nodes and branches;
collecting system state samples, probability tide tags and component states of a plurality of continuous historical time periods, wherein the component states comprise the running state of each component of the power system;
for a historical time period, inputting a system state sample into a probabilistic power flow model, and extracting time sequence characteristics of the system state sample to obtain time sequence characteristics of the historical time period, and predicting time sequence prediction characteristics of the next adjacent historical time period based on the time sequence characteristics; performing dimension transformation on the time sequence prediction characteristics to output probability power flow prediction results of the next adjacent historical time period;
building a training auxiliary model, wherein in the training auxiliary model, the component state of the historical time period is updated based on a time sequence difference characteristic to output a component prediction state of the next adjacent historical time period, and the real-time characteristic of each calculation target in the next adjacent historical time period is built based on the structural characteristic, the component connection vector and the component prediction state of the calculation target, wherein the time sequence difference characteristic is the difference between the time sequence characteristic and the time sequence prediction characteristic;
Calculating a loss function value based on the probabilistic power flow model and the training auxiliary model, the loss function value satisfying a relation:
wherein,for the timing prediction feature, < >>For the timing characteristics of the next adjacent history period, +.>For the probabilistic load flow prediction result, +.>Probability flow tag for next adjacent history period +.>Predicting a state for said component, +.>Component status for next adjacent history period,/-for next adjacent history period>And->Probability trend prediction results +.>Middle->Calculation target and->Mean and variance corresponding to each calculation target, +.>And->Respectively the first +.>Calculation target and->Real-time characteristics of individual calculation targets, +.>For all calculation targets, +.>Is a loss function value; updating the probabilistic tide model and the training auxiliary model by using a gradient descent method to finish one-time training;
and iteratively training the probability power flow model until the loss function value is smaller than a set value, and obtaining the trained probability power flow model.
2. The power system probability power flow calculation method based on data fusion according to claim 1, wherein the probability power flow model comprises a time sequence feature extraction layer, a first full connection layer and a second full connection layer;
The time sequence feature extraction layer is used for extracting time sequence features of the system state in the current time period and obtaining the time sequence features in the current time period;
the first full-connection layer is used for carrying out first dimension transformation on the time sequence characteristics so as to predict time sequence prediction characteristics in the next adjacent time period;
and the second full-connection layer is used for carrying out second dimension transformation on the time sequence prediction characteristics and outputting probability power flow calculation results of the next adjacent time period.
3. The power system probability power flow calculation method based on data fusion according to claim 2, wherein the size of the time sequence feature is M rows and 1 column; the first full-connection layer comprises a first input layer, a first hiding layer and a first output layer;
the first input layer comprises M neurons and is used for receiving the time sequence characteristics output by the time sequence characteristic extraction layer;
the number of neurons in the first hidden layer is not equal to M, and the neurons are used for carrying out dimension transformation on the time sequence characteristics;
the first output layer comprises M neurons, and is used for carrying out dimension transformation on the output result of the first hidden layer again and outputting time sequence prediction features in the next adjacent time period, wherein the size of the time sequence prediction features is M rows and 1 column.
4. A method of computing a probability flow of a power system based on data fusion according to claim 3, wherein the second fully connected layer comprises a second input layer, at least one second hidden layer and a second output layer;
the second input layer comprises M neurons and is used for receiving the time sequence prediction characteristics output by the first full-connection layer;
the second hidden layers are used for carrying out multiple dimension transformation on the time sequence prediction characteristics, wherein the number of dimension transformation times is the same as the number of the second hidden layers;
the second output layer comprises 2N neurons and is used for carrying out dimension transformation on the output result of the second hidden layer again and outputting the probability power flow calculation result of the next adjacent time period;
and N is the total number of all nodes and branches in the power system, the size of a probability power flow calculation result is 2N rows and 1 column, one node corresponds to the mean value and variance of node voltage, and one branch corresponds to the mean value and variance of branch power.
5. The method for calculating the probability power flow of the power system based on data fusion according to claim 1, wherein obtaining the structural feature and the component connection vector of each calculation target according to the topology of the power system comprises:
For one node in the topological structure, acquiring an embedded vector of the node according to a graph embedding algorithm, wherein the embedded vector corresponds to the structural characteristics of the node;
for one component in the topological structure, marking the connection state of the component as 1 in response to the connection of the component to the node, and marking the connection state of the component as 0 in response to the disconnection of the component to the node;
traversing all components in the topological structure to obtain component connection vectors of the nodes, wherein the component connection vectors comprise connection states of all components;
and for one branch in the topological structure, acquiring all nodes connected to the branch, taking the sum of structural features of all nodes as the structural features of the branch, and taking the union of component connection vectors of all nodes as the component connection vector of the branch.
6. The power system probability flow calculation method based on data fusion according to claim 1, wherein the training auxiliary model comprises a third full connection layer, the third full connection layer comprising a third input layer, at least one third hidden layer and at least one third output layer;
The third input layer comprises B neurons for receiving the states of components in a historical time period, wherein B is the total number of all components in the power system;
the third hiding layer performs multiple dimension reduction operations on the component state to obtain a component dimension reduction state, wherein the dimension of the component dimension reduction state is equal to the dimension of the time sequence difference characteristic, and the sum of the component dimension reduction state and the time sequence difference characteristic is used as an output result;
the third output layer is used for carrying out multi-time dimension lifting operation on the output result of the third hidden layer to obtain the predicted state of the component in the next adjacent historical time period;
the number of dimension-reducing operations is the same as the number of the third hidden layers, and the number of dimension-increasing operations is the same as the number of the third output layers.
7. The method for calculating the probability power flow of the power system based on the data fusion according to claim 6, wherein the construction of the real-time feature of each calculation target in the next adjacent historical time period based on the structural feature of the calculation target, the component connection vector and the component prediction state comprises:
for a calculation target, acquiring a component connection vector of the calculation target, and multiplying the component connection vector by the component prediction state to obtain the component real-time characteristics of the calculation target;
And splicing the structural features of the calculation target with the real-time features of the components to obtain the real-time features of the calculation target in the next adjacent historical time period.
8. The method for computing probability power flow of a power system based on data fusion of claim 1, wherein the timing characteristics of the next adjacent historical time periodThe acquisition method of (1) comprises the following steps:
and inputting the system state samples of the next adjacent historical time period into the time sequence feature extraction layer, and outputting the time sequence features of the next adjacent historical time period.
9. A data fusion based power system probability flow calculation system, comprising a processor and a memory, the memory storing computer program instructions which, when executed by the processor, implement the data fusion based power system probability flow calculation method according to any one of claims 1 to 8.
CN202311523012.8A 2023-11-16 2023-11-16 Power system probability power flow calculation method and system based on data fusion Active CN117254472B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311523012.8A CN117254472B (en) 2023-11-16 2023-11-16 Power system probability power flow calculation method and system based on data fusion

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311523012.8A CN117254472B (en) 2023-11-16 2023-11-16 Power system probability power flow calculation method and system based on data fusion

Publications (2)

Publication Number Publication Date
CN117254472A true CN117254472A (en) 2023-12-19
CN117254472B CN117254472B (en) 2024-01-30

Family

ID=89137221

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311523012.8A Active CN117254472B (en) 2023-11-16 2023-11-16 Power system probability power flow calculation method and system based on data fusion

Country Status (1)

Country Link
CN (1) CN117254472B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103236692A (en) * 2013-04-25 2013-08-07 网新创新研究开发有限公司 Method for evaluating operation status of power system by utilizing probability tide
CN109117951A (en) * 2018-01-15 2019-01-01 重庆大学 Probabilistic Load Flow on-line calculation method based on BP neural network
CN109412161A (en) * 2018-12-18 2019-03-01 国网重庆市电力公司电力科学研究院 A kind of Probabilistic Load calculation method and system
US20200184308A1 (en) * 2018-12-06 2020-06-11 University Of Tennessee Research Foundation Methods, systems, and computer readable mediums for determining a system state of a power system using a convolutional neural network
CN111463793A (en) * 2020-04-23 2020-07-28 国网上海市电力公司 Analytic probabilistic power flow calculation method and system
CN112994017A (en) * 2021-03-19 2021-06-18 国网江苏省电力有限公司南通供电分公司 Distributed photovoltaic power supply site selection optimization method based on power distribution network probability load flow calculation
CN113708376A (en) * 2021-08-23 2021-11-26 中国电力科学研究院有限公司 Power system probability optimal power flow calculation method, system, equipment and storage medium

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103236692A (en) * 2013-04-25 2013-08-07 网新创新研究开发有限公司 Method for evaluating operation status of power system by utilizing probability tide
CN109117951A (en) * 2018-01-15 2019-01-01 重庆大学 Probabilistic Load Flow on-line calculation method based on BP neural network
US20200184308A1 (en) * 2018-12-06 2020-06-11 University Of Tennessee Research Foundation Methods, systems, and computer readable mediums for determining a system state of a power system using a convolutional neural network
CN109412161A (en) * 2018-12-18 2019-03-01 国网重庆市电力公司电力科学研究院 A kind of Probabilistic Load calculation method and system
CN111463793A (en) * 2020-04-23 2020-07-28 国网上海市电力公司 Analytic probabilistic power flow calculation method and system
CN112994017A (en) * 2021-03-19 2021-06-18 国网江苏省电力有限公司南通供电分公司 Distributed photovoltaic power supply site selection optimization method based on power distribution network probability load flow calculation
CN113708376A (en) * 2021-08-23 2021-11-26 中国电力科学研究院有限公司 Power system probability optimal power flow calculation method, system, equipment and storage medium

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
KEJUN CHEN: "Physics-Guided Residual Learning for Probabilistic Power Flow Analysis", 《IEEE ACCESS》 *
PEI BIE: "Probabilistic dynamic load flow algorithm considering static security risk of the power system", 《2015 5TH INTERNATIONAL CONFERENCE ON ELECTRIC UTILITY DEREGULATION AND RESTRUCTURING AND POWER TECHNOLOGIES (DRPT)》 *
苑婷;王俊;朱浩?;王刚;张林;孙佳琪;余维涛;王欣然;段佳鑫;孟妍;: "基于随机变量状态时序模拟的光伏并网系统概率潮流研究", 沈阳农业大学学报, no. 06 *
陈新建;朱轶伦;洪道鉴;于杰;: "基于深度学习的概率性电网潮流快速计算方法", 沈阳师范大学学报(自然科学版), no. 06 *

Also Published As

Publication number Publication date
CN117254472B (en) 2024-01-30

Similar Documents

Publication Publication Date Title
Hua et al. Optimal energy management strategies for energy Internet via deep reinforcement learning approach
Di Piazza et al. An artificial neural network-based forecasting model of energy-related time series for electrical grid management
Müller et al. Genetic algorithm‐based phasor measurement unit placement method considering observability and security criteria
Yu et al. Solar photovoltaic model parameter estimation based on orthogonally-adapted gradient-based optimization
CN109412161B (en) Power system probability load flow calculation method and system
CN110232476A (en) A kind of micro-grid load prediction technique based on deep learning
Serttas et al. Short term solar power generation forecasting: A novel approach
Dong et al. Effective method to determine time‐delay stability margin and its application to power systems
Wu et al. Unrolled spatiotemporal graph convolutional network for distribution system state estimation and forecasting
Shayesteh et al. Scenario reduction, network aggregation, and DC linearisation: which simplifications matter most in operations and planning optimisation?
Pham et al. Neural network-based power flow model
CN105449667A (en) Prediction method for reliability of power generation system and power transmission system
Kang et al. CM-GAN: A cross-modal generative adversarial network for imputing completely missing data in digital industry
Mohammadi et al. A Deep Learning‐to‐learning Based Control system for renewable microgrids
CN117254472B (en) Power system probability power flow calculation method and system based on data fusion
Fang et al. Interval optimal reactive power reserve dispatch considering generator rescheduling
Tan et al. Explainable Artificial Intelligence Applied to Deep Reinforcement Learning Controllers for Photovoltaic Maximum Power Point Tracking
Mahto et al. Mpgcn-opf: A message passing graph convolution approach for optimal power flow for distribution network
Bogovič et al. Probabilistic three-phase power flow in a distribution system applying the pseudo-inverse and cumulant method
CN116522560A (en) Active power distribution network topology identification method and system based on probabilistic neural network
Carmichael et al. Application of deep neural networks to distribution system state estimation and forecasting
Lee et al. Uncertainty‐aware forecast interval for hourly PV power output
Ramesh et al. Electrical power system state estimation meter placement—a comparative survey report
Zhou et al. Transient stability assessment using gated recurrent unit
Alex Stanley Raja et al. Performance analysis and validation of intelligent tool based on Brownian random walk‐based sand cat swarm optimization algorithm for parameter identification of various solar photovoltaic mathematical models

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