CN115313510A - Adaptive reactive compensation photovoltaic inverter control method and system - Google Patents
Adaptive reactive compensation photovoltaic inverter control method and system Download PDFInfo
- Publication number
- CN115313510A CN115313510A CN202210872199.1A CN202210872199A CN115313510A CN 115313510 A CN115313510 A CN 115313510A CN 202210872199 A CN202210872199 A CN 202210872199A CN 115313510 A CN115313510 A CN 115313510A
- Authority
- CN
- China
- Prior art keywords
- photovoltaic
- matrix
- node
- training
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
- H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers
- H02J3/46—Controlling of the sharing of output between the generators, converters, or transformers
-
- 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
-
- 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
- H02J13/00—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network
- H02J13/00002—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network characterised by monitoring
-
- 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
- H02J13/00—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network
- H02J13/00006—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network characterised by information or instructions transport means between the monitoring, controlling or managing units and monitored, controlled or operated power network element or electrical equipment
- H02J13/00028—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network characterised by information or instructions transport means between the monitoring, controlling or managing units and monitored, controlled or operated power network element or electrical equipment involving the use of Internet protocols
-
- 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
- H02J3/001—Methods to deal with contingencies, e.g. abnormalities, faults or failures
-
- 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
- H02J3/18—Arrangements for adjusting, eliminating or compensating reactive power in networks
-
- 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
- H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers
- H02J3/46—Controlling of the sharing of output between the generators, converters, or transformers
- H02J3/48—Controlling the sharing of the in-phase component
-
- 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
- H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers
- H02J3/46—Controlling of the sharing of output between the generators, converters, or transformers
- H02J3/50—Controlling the sharing of the out-of-phase component
-
- 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
- H02J2300/00—Systems for supplying or distributing electric power characterised by decentralized, dispersed, or local generation
- H02J2300/20—The dispersed energy generation being of renewable origin
- H02J2300/22—The renewable source being solar energy
- H02J2300/24—The renewable source being solar energy of photovoltaic origin
- H02J2300/26—The renewable source being solar energy of photovoltaic origin involving maximum power point tracking control for photovoltaic sources
-
- 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
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/50—Photovoltaic [PV] energy
- Y02E10/56—Power conversion systems, e.g. maximum power point trackers
-
- 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
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E40/00—Technologies for an efficient electrical power generation, transmission or distribution
- Y02E40/30—Reactive power compensation
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention belongs to the technical field of electric power, and particularly relates to a photovoltaic inverter control method and system for adaptive reactive power compensation. The electrical part of the invention is connected with a photovoltaic DC-AC converter by a photovoltaic array through a cable, and the photovoltaic DC-AC converter converts direct current output by the photovoltaic array into alternating current and transmits the alternating current to an alternating current bus; the control part is communicated with the environmental data detection system through a self-learning optimization controller to acquire environmental irradiance and environmental temperature data acquired by the environmental detection system in real time; acquiring current and voltage data output by the photovoltaic array through a current and voltage transformer; acquiring voltage data of an alternating current bus through a voltage transformer; the self-learning optimization controller takes voltage and power information communicated with a data acquisition device of an adjacent network node as calculation input data of the self-learning optimization controller, and a calculation result is sent to the photovoltaic DC-AC converter to control operation. The problems of voltage out-of-limit and overlarge parameter deviation of the controller can be effectively avoided.
Description
Technical Field
The invention belongs to the technical field of electric power, and particularly relates to a photovoltaic inverter control method and system with adaptive reactive power compensation.
Background
The solar photovoltaic technology enters a rapid development stage, and photovoltaic grid-connected power generation becomes the most main way for people to utilize solar energy.
With the development of projects such as photovoltaic poverty alleviation, a large amount of photovoltaic equipment is accessed in more and more rural areas. The rural areas have extensive areas and sparse people and are dispersed in load, and the conditions of few distribution points in a distribution area, long distribution distance and small line diameter design are common in consideration of the economy of distribution network construction; in the daytime, local voltage is higher due to the fact that local load is small, photovoltaic output is large, local load is large in the evening, photovoltaic output is not large, local voltage is lower due to the fact that photovoltaic output is not large, daily voltage deviation of a power grid of a transformer area is more than 30% due to the fact that a large number of distributed photovoltaic devices are connected, the voltage is higher in the daytime and lower in the nighttime, and safe and stable operation of a low-voltage rural power grid is seriously influenced. Meanwhile, the grid voltage is too high and exceeds the operation limit of the distributed photovoltaic inverter, and partial photovoltaic inverters are disconnected due to overvoltage.
The electric energy generated by the photovoltaic module is incorporated into an alternating current power grid through a photovoltaic inverter (PV inverter or solar inverter), and the photovoltaic inverter can convert the variable direct current voltage generated by the photovoltaic PV solar panel into alternating current AC with commercial power frequency.
At present, a photovoltaic inverter generally operates in a Maximum Power Point Tracking (MPPT) mode, and the problems of reactive power, voltage out-of-limit, harmonic waves and the like in the operation of a power grid are not considered.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a photovoltaic inverter control method and system with adaptive reactive power compensation. The method aims to optimally control the operation parameters of the photovoltaic inverter based on a graph convolution neural network method and solve the problems of grid voltage out-of-limit and the like caused by the fact that the photovoltaic power output is controlled by only depending on a Maximum Power Point Tracking (MPPT) method at present.
The technical scheme adopted by the invention for realizing the purpose is as follows:
a self-adaptive reactive compensation photovoltaic inverter control system comprises an electric part and a control part;
wherein the electrical part: the photovoltaic array is connected with a photovoltaic DC-AC converter through a cable, and the photovoltaic DC-AC converter converts direct current output by the photovoltaic array into alternating current and transmits the alternating current to an alternating current bus through the cable;
wherein the control section: the self-learning optimization controller is communicated with the environmental data detection system through a communication protocol to acquire environmental irradiance and environmental temperature data acquired by the environmental detection system in real time; acquiring current and voltage data output by the photovoltaic array in real time through a current transformer and a voltage transformer; acquiring voltage data of an alternating current bus through a voltage transformer; meanwhile, the self-learning optimization controller is communicated with a data acquisition device of an adjacent network node to acquire voltage and power information of the adjacent node; and the information is used as calculation input data of the self-learning optimization controller, and a calculation result of the self-learning optimization controller is used as a control quantity and is sent to the photovoltaic DC-AC converter to control the photovoltaic DC-AC converter to operate.
Furthermore, the environmental data detection system is used for detecting the environmental temperature and the environmental irradiance.
Furthermore, the self-learning optimization controller performs optimization calculation by using data acquired by the environment data detection system and the power grid node power detection sensor to obtain a control quantity required by the photovoltaic DC-AC converter, and controls the DC-AC converter to transmit electric energy generated by the photovoltaic array to the power grid.
A control method of a photovoltaic inverter with adaptive reactive compensation comprises the following steps:
step 1, an undirected graph is constructed according to an electrical network to describe the topological relation between a photovoltaic inverter access point and other electrical branches and electrical nodes, an incidence matrix is formed, and a Laplace matrix is constructed;
step 2, constructing a graph convolution neural network model by using the Laplace matrix;
step 3, constructing a simulation system model and further constructing a training data set so as to initially train the constructed graph convolution neural network model;
step 4, carrying out repeated iterative training on the graph convolution neural network by using the extracted training data;
step 5, calculating PWM duty ratio, active power and reactive power through the trained graph convolution neural network;
step 6, controlling the inverter to operate by taking the calculation result as a photovoltaic inverter control parameter;
step 7, collecting inverter operation data in real time;
and 8, triggering the self-learning optimization controller to perform self-learning training by using the newly added data every 3-7 days.
Furthermore, step 1, an undirected graph is constructed according to an electrical network to describe the topological relation between the photovoltaic inverter access point and other electrical branches and electrical nodes, an incidence matrix is formed, and a Laplace matrix is constructed;
setting the incidence matrix A, wherein the Laplace matrix is Laplacian matrix: for the graph G = (V, E), its laplacian matrix is defined as L = D-a, where L is the laplacian matrix, D = diag (D) is the degree matrix of vertices, i.e. the diagonal matrix, D = rowSum (a), with the diagonal elements being the degrees of the respective vertices in turn, for each element in the a matrix a ij If nodes i and j are adjacent, then A ij =1, otherwise A ij ＝0。
Furthermore, in the step 2, in constructing the graph convolution neural network model by using the laplacian matrix, the laplacian matrix is responsible for introducing the connection relation between the nodes into the model in a matrix manner, and the connection relation is used as a characteristic parameter of the graph and participates in calculation in each iteration of training the graph convolution neural network;
the graph convolution neural network model comprises an input layer, a hidden layer, a pooling layer, a full-link layer and an output layer;
the input layer includes: irradiance, temperature, photovoltaic array output current, photovoltaic array output voltage, voltage of adjacent electrical connection nodes, input active power and input reactive power of adjacent electrical nodes;
the hidden layer: the number of hidden layers is determined by the maximum value of the shortest path from any node in the S1 to other nodes, and is usually 1-N layers; using a parameterized modified linear unit PReLU as an activation function:
in the above formula, a is a learnable parameter, r (x) is an activation function value, and x is an output matrix of an upper model; the learnable parameter a is updated in the training process;
the pooling layer is as follows: the method is used for reducing the size of the model, improving the calculation speed and improving the robustness of the extracted features;
the full connection layer: wherein the number of the nodes is l N ＝log 2 V, wherein V is the number of the output nodes of the upper layer, and a Sigmoid function is adopted as an activation function:
in the above formula, σ (z) is an output value of the activation function, and z is an output matrix of the upper layer model;
the output layer: PWM duty cycle, output active power, output reactive power.
Furthermore, in step 3, the simulation system model is constructed, and then the training data set is constructed, so that the constructed convolution neural network model is initially trained, and the initial training data set is generated through the simulation system model, so that the convolution neural network model constructed in step 2 can be initially trained;
the simulation system model is constructed by simulating voltage out-of-limit conditions and normal operation conditions of different network positions under different illumination conditions respectively to construct a training data set; and extracting data required by the convolutional neural network of the training graph from the data set according to the constructed training data set.
Further, the multiple iterative training is performed by multiple iterative training, and in each iteration, the multiple iterative training includes:
step 4-1, training hidden layer parameters of the graph convolutional neural network;
according to the adjacent matrix A, the node characteristics of each object node and the node characteristics of the adjacent nodes are aggregated to form a new aggregated characteristic matrix, a new Laplace matrix is calculated, a weighted average value fusion mode is adopted in an aggregation algorithm, weighted average values are obtained from the node characteristics and the node characteristics of the adjacent nodes, attention weights are distributed to different adjacent nodes when characteristic information is aggregated, and the method comprises the following steps:
wherein:
is the characteristic value of the current node of the current layer, sigma u∈N(v)∪{v} (. Cndot.) denotes all neighboring nodes and self node, weight
For the attention weight coefficient between the u-th node and the v-th node, the attention weight coefficient, W, of each node needs to be learned during training k In order to aggregate the weights, the weights are,
the node is the neighbor node characteristic of the upper layer;
determining embedded vectors of nodes by using N hidden layers contained in the graph convolution neural network, carrying out graph convolution operation on node characteristics, obtaining a graph signal matrix after activating a function, and determining a training gradient and an attention weight of parameters in the graph convolution neural network; updating parameters in a graph convolution neural network based on the training gradient and the attention weight;
step 4-2, training parameters of a full connection layer and an output layer by using a random gradient descent method of Momentum;
the method is obtained through multiple times of simulation: initial learning rate and momentum over-parameters.
A computer device comprising a storage medium, a processor and a computer program stored on the storage medium and executable on the processor, the processor implementing any of the steps of an adaptive reactive compensation photovoltaic inverter control method when executing the computer program.
A computer storage medium having a computer program stored thereon, the computer program when executed by a processor implementing the steps of any of the methods for adaptive reactive compensation photovoltaic inverter control described herein.
Compared with the prior art, the invention has the following beneficial effects and advantages:
according to the method, the voltage and power of the photovoltaic access point and the adjacent distribution network nodes are considered when the operation parameters of the photovoltaic inverter are calculated, the problems of voltage out-of-limit and the like caused by large-scale photovoltaic access to the distribution network can be effectively avoided, and meanwhile, the active power output as large as possible is ensured.
The self-learning optimization controller is designed by adopting a method based on the graph convolution neural network, and compared with the traditional scheme based on a mechanism model, the neural network method has the advantage of high speed in calculating an output result; meanwhile, compared with other deep neural networks such as RNN and LSTM, the method of the graph convolution neural network considers the incidence relation of the power grid nodes and the mutual influence among the incidence nodes.
The self-learning optimization controller designed by the invention comprises a self-learning function, and can perform supplementary training by using historical operating data so as to avoid the problems of overlarge parameter deviation of the controller and the like caused by the change of external factors.
Drawings
The above and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a block diagram of the control apparatus of the present invention;
FIG. 3 is a diagram of a convolutional neural network model of the present invention.
Detailed Description
In order that the above objects, features and advantages of the present invention can be more clearly understood, a more particular description of the invention will be rendered by reference to the appended drawings. It should be noted that the embodiments and features of the embodiments of the present application may be combined with each other without conflict.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced otherwise than as specifically described herein and, therefore, the scope of the present invention is not limited by the specific embodiments disclosed below.
The solution of some embodiments of the invention is described below with reference to fig. 1-3.
Example 1
The invention provides an embodiment, which is a photovoltaic inverter control system with adaptive reactive compensation, as shown in fig. 2, and fig. 2 is a block diagram of a control device of the invention.
The self-learning optimization system comprises a photovoltaic array, an environmental data detection system, a photovoltaic DC-AC converter, a self-learning optimization controller, a power grid node voltage and power grid node power detection sensor.
The electrical part includes: the photovoltaic array is connected with the photovoltaic DC-AC converter through a cable, and the photovoltaic DC-AC converter converts direct current output by the photovoltaic array into alternating current and transmits the alternating current to the alternating current bus through the cable.
The control section includes: the self-learning optimization controller is communicated with the environmental data detection system through a communication protocol to acquire environmental irradiance and environmental temperature data acquired by the environmental detection system in real time; acquiring current and voltage data output by the photovoltaic array in real time through a current transformer and a voltage transformer; acquiring voltage data of an alternating current bus through a voltage transformer; meanwhile, the self-learning optimization controller is communicated with the data acquisition device of the adjacent network node to acquire the voltage and power information of the adjacent node.
And the information is used as calculation input data of the self-learning optimization controller, and a calculation result of the self-learning optimization controller is used as a control quantity and is sent to the photovoltaic DC-AC converter to control the photovoltaic DC-AC converter to operate.
The environment data detection system is used for detecting the environment temperature and the environment irradiance.
The self-learning optimization controller performs optimization calculation by using data acquired by the environment data detection system and the power grid node power detection sensor to obtain control quantity required by the photovoltaic DC-AC converter, and controls the DC-AC converter to transmit electric energy generated by the photovoltaic array to the power grid.
Example 2
The invention also provides an embodiment, and provides a photovoltaic inverter control method with adaptive reactive power compensation. As shown in FIG. 1, FIG. 1 is a flow chart of the method of the present invention.
The control method is improved in that the method is designed and implemented based on a graph convolution neural network model, and specifically comprises the following steps:
step 1, according to an electrical network, constructing an undirected graph to describe the topological relation between a photovoltaic inverter access point and other electrical branches and electrical nodes, forming an incidence matrix, and constructing a Laplace matrix.
Setting the incidence matrix A, wherein the Laplace matrix is Laplacian matrix: for the graph G = (V, E), the definition of its laplacian matrix is L = D-a, where L is the laplacian matrix, D = diag (D) is the degree matrix of vertices, i.e. diagonal matrix, D = rowSum (a), the elements on the diagonal are the degrees of the respective vertices in turn, for each element in the a matrix a ij If nodes i and j are adjacent, then A ij =1, otherwise A ij ＝0；
And 2, constructing a graph convolution neural network model by using the Laplace matrix.
The Laplace matrix is responsible for introducing the connection relation between the nodes in the graph into the model in a matrix mode, and the connection relation is used as a characteristic parameter of the graph and participates in calculation in each iteration of training graph convolutional neural network.
The graph convolution neural network model is shown in fig. 3, and fig. 3 is a graph of the graph convolution neural network model of the invention. Including 1 input layer, 2 hidden layers, 1 pooling layer, 2 full-link layers and 1 output layer.
The input layer includes: irradiance, temperature, photovoltaic array output current, photovoltaic array output voltage, voltage of adjacent electrical connection nodes, input active power and input reactive power of adjacent electrical nodes;
the hidden layer: the number of hidden layers is determined by the maximum value of the shortest path from any node to other nodes in the S1, and is usually 1-N layers. The N is the maximum value of the shortest path from any node to other nodes in the S1, where N is related to the scale of the graph.
Using a parameterized modified linear unit PReLU as an activation function:
in the above formula, a is a learnable parameter, r (x) is the activation function value, and x is the output matrix of the upper model.
a is used as a learnable parameter and is updated in the training process.
The pooling layer: the method is used for reducing the size of the model, improving the calculation speed and improving the robustness of the extracted features;
the full connection layer: wherein the number of the nodes is l N ＝log 2 V, wherein V is the number of the output nodes of the upper layer, and a Sigmoid function is adopted as an activation function:
in the above equation, σ (z) is an activation function output value, and z is an output matrix of the upper layer model.
The output layer: PWM duty cycle, output active power, output reactive power.
The PWM duty cycle is a term commonly used by practitioners in the technical field, and is a value between 0% and 100%, and is one of important parameters for controlling the photovoltaic inverter.
And 3, constructing a simulation system model and further constructing a training data set so as to initially train the constructed graph convolution neural network model.
An initial training data set is generated through the simulation system model, so that the graph convolution neural network model constructed in the step 2 can be initially trained.
The simulation system model is constructed by simulating the voltage out-of-limit condition and the normal operation condition of different network positions under different illumination conditions respectively to construct a training data set.
And extracting data required by the convolutional neural network of the training graph from the data set according to the constructed training data set.
And 4, carrying out repeated iterative training on the graph convolution neural network by using the extracted training data.
As shown in FIG. 3, FIG. 3 is a diagram of a convolutional neural network model of the present invention.
The multiple iterative training is specifically performed through multiple iterative training, and in each iteration:
step 4-1, training hidden layer parameters of the graph convolutional neural network;
step 4-2, training parameters of a full connection layer and an output layer by using a random gradient descent method of Momentum;
step 5, calculating PWM duty ratio, active power and reactive power through the trained graph convolution neural network;
step 6, controlling the inverter to operate by taking the calculation result of the step 5 as a photovoltaic inverter control parameter;
and 7, acquiring the operation data of the inverter in real time.
The collecting inverter operation data comprises: the method comprises the steps of controlling parameters of an inverter, voltage and input power of nodes of a power grid, output current of a photovoltaic array, output voltage of the photovoltaic array, ambient irradiance and ambient temperature, storing running data into a training data set, and using the running data for self-learning training of a graph convolution neural network.
And 8, triggering the self-learning optimization controller to perform self-learning training by utilizing the newly added data at certain intervals, such as 3 days or 1 week.
Example 3
The invention further provides an embodiment, and provides a photovoltaic inverter control method and system with adaptive reactive power compensation. The hidden layer is 2 layers selected from the hidden layers through experimental verification, and the hidden layer is the optimal priority layer, so that the effect is ideal.
Example 4
The invention also provides an embodiment, which is a photovoltaic inverter control method of adaptive reactive power compensation,
the method of the present invention is designed and implemented based on a convolutional neural network model, as in embodiment 2, wherein in step 4, the extracted training data is used to perform a plurality of iterative trainings on the convolutional neural network, and in each iteration, the method includes the following steps:
step 4-1, training hidden layer parameters of the graph convolutional neural network:
for each object node, according to the adjacent matrix A, aggregating the node characteristics of the object node and the node characteristics of the neighbor nodes of the object node to form a new aggregated characteristic matrix, and calculating a new Laplace matrix, wherein the aggregation algorithm can adopt a weighted mean fusion mode, namely, the node characteristics and the node characteristics of the neighbor nodes of the object node obtain a weighted average value, and the attention weight is distributed to different neighbor nodes when characteristic information is aggregated, and the method comprises the following steps:
wherein:
is the characteristic value of the current node of the current layer, sigma u∈N(v)∪{v} (. To) all neighbor nodes and self-node, weight
For the attention weight coefficient between the u-th node and the v-th node, the attention weight coefficient, W, of each node needs to be learned during training k In order to aggregate the weights, the weights are,
the node is characterized by the neighbor node of the previous layer.
Determining embedded vectors of nodes by using N hidden layers contained in the graph convolution neural network, carrying out graph convolution operation on node characteristics, obtaining a graph signal matrix after activating a function, and determining a training gradient and an attention weight of parameters in the graph convolution neural network; updating parameters in the graph convolutional neural network based on the training gradients and attention weights;
and 4-2, training parameters of the full connection layer and the output layer by using a random gradient descent method of Momentum.
The method is obtained through multiple times of simulation: the initial learning rate is 0.12, and the momentum over-parameter is 0.78.
The other steps are the same as in example 2.
Example 5
Based on the same inventive concept, the embodiment of the present invention also provides a computer device, which includes a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor. The processor, when executing the computer program, implements the steps of the adaptive reactive power compensation photovoltaic inverter control method according to any one of embodiments 2, 3, or 4.
Example 6
Based on the same inventive concept, an embodiment of the present invention further provides a computer storage medium, where a computer program is stored on the computer storage medium, and when the computer program is executed by a processor, the steps of the adaptive reactive power compensation photovoltaic inverter control method according to any one of embodiments 2, 3, and 4 are implemented.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting the same, and although the present invention is described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications and equivalents may be made to the embodiments of the invention without departing from the spirit and scope of the invention, which is to be covered by the claims.
Claims (10)
1. The utility model provides a photovoltaic inverter control system of self-adaptation reactive compensation which characterized by: comprises an electric part and a control part;
wherein the electrical part: the photovoltaic array is connected with a photovoltaic DC-AC converter through a cable, and the photovoltaic DC-AC converter converts direct current output by the photovoltaic array into alternating current and transmits the alternating current to an alternating current bus through the cable;
wherein the control section: the self-learning optimization controller is communicated with the environmental data detection system through a communication protocol to acquire environmental irradiance and environmental temperature data acquired by the environmental detection system in real time; acquiring current and voltage data output by the photovoltaic array in real time through a current transformer and a voltage transformer; acquiring voltage data of an alternating current bus through a voltage transformer; meanwhile, the self-learning optimization controller is communicated with a data acquisition device of an adjacent network node to acquire voltage and power information of the adjacent node; and the information is used as calculation input data of the self-learning optimization controller, and a calculation result of the self-learning optimization controller is used as a control quantity and is sent to the photovoltaic DC-AC converter to control the photovoltaic DC-AC converter to operate.
2. The adaptive reactive compensation photovoltaic inverter control system of claim 1, wherein: the environment data detection system is used for detecting the environment temperature and the environment irradiance.
3. The adaptive reactive compensation photovoltaic inverter control system of claim 1, wherein: the self-learning optimization controller performs optimization calculation by using data acquired by the environment data detection system and the power grid node power detection sensor to obtain control quantity required by the photovoltaic DC-AC converter, and controls the DC-AC converter to transmit electric energy generated by the photovoltaic array to the power grid.
4. A self-adaptive reactive compensation photovoltaic inverter control method is characterized by comprising the following steps: the method comprises the following steps:
step 1, an undirected graph is constructed according to an electrical network to describe the topological relation between a photovoltaic inverter access point and other electrical branches and electrical nodes, an incidence matrix is formed, and a Laplace matrix is constructed;
step 2, constructing a graph convolution neural network model by using the Laplace matrix;
step 3, constructing a simulation system model and further constructing a training data set so as to initially train the constructed graph convolution neural network model;
step 4, carrying out repeated iterative training on the graph convolution neural network by using the extracted training data;
step 5, calculating PWM duty ratio, active power and reactive power through the trained graph convolution neural network;
step 6, controlling the inverter to operate by taking the calculation result as a photovoltaic inverter control parameter;
step 7, collecting the operation data of the inverter in real time;
and 8, triggering the self-learning optimization controller to perform self-learning training by using the newly added data every 3-7 days.
5. The adaptive reactive compensation photovoltaic inverter control method according to claim 1, wherein: step 1, constructing an undirected graph according to an electrical network to describe the topological relation between a photovoltaic inverter access point and other electrical branches and electrical nodes, forming an incidence matrix, and constructing a Laplace matrix;
setting the incidence matrix A, wherein the Laplacian matrix is Laplacianmatrix: for the graph G = (V, E), its laplacian matrix is defined as L = D-a, where L is the laplacian matrix, D = diag (D) is the degree matrix of vertices, i.e. the diagonal matrix, D = rowSum (a), with the diagonal elements being the degrees of the respective vertices in turn, for each element in the a matrix a ij If nodes i and j are adjacent, then A ij =1, otherwise A ij ＝0。
6. The adaptive reactive compensation photovoltaic inverter control method according to claim 1, wherein: 2, in the construction of the graph convolution neural network model by using the Laplace matrix, the Laplace matrix is responsible for introducing the connection relation between the nodes into the model in a matrix mode and is used as a characteristic parameter of the graph, and the Laplace matrix participates in calculation in each iteration of the training graph convolution neural network;
the graph convolution neural network model comprises an input layer, a hidden layer, a pooling layer, a full-link layer and an output layer;
the input layer includes: irradiance, temperature, photovoltaic array output current, photovoltaic array output voltage, voltage of adjacent electrical connection nodes, input active power and input reactive power of adjacent electrical nodes;
the hidden layer is as follows: the number of hidden layers is determined by the maximum value of the shortest path from any node in the S1 to other nodes, and is usually 1-N layers; using a parameterized modified linear unit PReLU as an activation function:
in the above formula, a is a learnable parameter, r (x) is an activation function value, and x is an output matrix of an upper model; the learnable parameter a is updated in the training process;
the pooling layer is as follows: the method is used for reducing the size of the model, improving the calculation speed and improving the robustness of the extracted features;
the full connection layer: wherein the number of the nodes is l N ＝log 2 V, wherein V is the number of the output nodes of the upper layer, and a Sigmoid function is adopted as an activation function:
in the above formula, σ (z) is an output value of the activation function, and z is an output matrix of the upper layer model;
the output layer: PWM duty cycle, output active power, output reactive power.
7. The adaptive reactive compensation photovoltaic inverter control method according to claim 1, wherein: 3, constructing a simulation system model and further constructing a training data set so as to initially train the constructed convolutional neural network model, wherein the simulation system model is used for generating the initial training data set so as to initially train the convolutional neural network model constructed in the step 2;
the simulation system model is constructed by simulating voltage out-of-limit conditions and normal operation conditions of different network positions under different illumination conditions respectively to construct a training data set; and extracting data required by the convolutional neural network of the training graph from the data set according to the constructed training data set.
8. The adaptive reactive compensation photovoltaic inverter control method according to claim 1, wherein: the multiple iterative training is performed through multiple iterative training, and each iteration comprises the following steps:
step 4-1, training hidden layer parameters of the graph convolutional neural network;
for each object node, according to the adjacent matrix A, aggregating the node characteristics of the object node and the node characteristics of the neighbor nodes of the object node to form a new aggregated characteristic matrix, and calculating a new Laplace matrix, wherein the aggregation algorithm adopts a weighted mean fusion mode to calculate a weighted mean value of the node characteristics and the node characteristics of the neighbor nodes of the object node, and distributes attention weights to different neighbor nodes when aggregating characteristic information, and the method comprises the following steps:
wherein:
is a characteristic value of a current node of a current layer, sigma u∈N(v)∪{v} (. Cndot.) denotes all neighboring nodes and self node, weight
For the attention weight coefficient between the u-th node and the v-th node, the attention weight coefficient, W, of each node needs to be learned during training k In order to aggregate the weights, the weights are,
the node is the neighbor node characteristic of the upper layer;
determining embedded vectors of nodes by utilizing N hidden layers contained in the graph convolution neural network, performing graph convolution operation on node characteristics, obtaining a graph signal matrix after activating a function, and determining training gradients and attention weights of parameters in the graph convolution neural network; updating parameters in a graph convolution neural network based on the training gradient and the attention weight;
4-2, training parameters of the full connection layer and the output layer by using a random gradient descent method of Momentum;
the method is obtained through multiple times of simulation: initial learning rate and momentum over-parameters.
9. A computer arrangement comprising a storage medium, a processor and a computer program stored on the storage medium and executable on the processor, wherein the processor when executing the computer program performs the steps of the method for adaptive reactive power compensated photovoltaic inverter control according to any of claims 4-8.
10. A computer storage medium, comprising: the computer storage medium has a computer program stored thereon, and the computer program, when executed by a processor, implements the steps of the method for controlling an adaptive reactive power compensation pv inverter according to any of claims 4 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210872199.1A CN115313510A (en) | 2022-07-23 | 2022-07-23 | Adaptive reactive compensation photovoltaic inverter control method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210872199.1A CN115313510A (en) | 2022-07-23 | 2022-07-23 | Adaptive reactive compensation photovoltaic inverter control method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115313510A true CN115313510A (en) | 2022-11-08 |
Family
ID=83858490
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210872199.1A Pending CN115313510A (en) | 2022-07-23 | 2022-07-23 | Adaptive reactive compensation photovoltaic inverter control method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115313510A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116760055A (en) * | 2023-06-07 | 2023-09-15 | 东南大学 | Dynamic reactive compensation method based on neural network |
-
2022
- 2022-07-23 CN CN202210872199.1A patent/CN115313510A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116760055A (en) * | 2023-06-07 | 2023-09-15 | 东南大学 | Dynamic reactive compensation method based on neural network |
| CN116760055B (en) * | 2023-06-07 | 2024-03-12 | 东南大学 | Dynamic reactive compensation method based on neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114362196B (en) | Multi-time-scale active power distribution network voltage control method | |
| CN105870949B (en) | A kind of micro-capacitance sensor energy-storage units optimal control method based on distributed gradient algorithm | |
| JP2024008787A (en) | Control method of hybrid energy storage system, device and electronic apparatus | |
| Wang et al. | Design and analysis of genetic algorithm and BP neural network based PID control for boost converter applied in renewable power generations | |
| CN105896575B (en) | Hundred megawatt energy storage power control method and system based on self-adaptive dynamic programming | |
| CN111146821A (en) | DSTATCOM optimal configuration method considering photovoltaic uncertainty | |
| CN113378100B (en) | Power distribution network source network load storage collaborative optimization scheduling model and method considering carbon emission | |
| CN104734175A (en) | Method for intelligently modifying wind speed and power curve of wind turbine generator | |
| CN116667325B (en) | Micro-grid-connected operation optimization scheduling method based on improved cuckoo algorithm | |
| Yang et al. | Deep learning-based distributed optimal control for wide area energy Internet | |
| CN104915788B (en) | A method of considering the Electrical Power System Dynamic economic load dispatching of windy field correlation | |
| CN112330021A (en) | Network coordination control method of distributed optical storage system | |
| CN115313510A (en) | Adaptive reactive compensation photovoltaic inverter control method and system | |
| CN105207255B (en) | A kind of power system peak regulation computational methods suitable for wind power output | |
| Jegajothi et al. | Combination BFPSO Tuned Intelligent Controller for Maximum Power Point Tracking in Solar Photovoltaic Farm Interconnected to Grid Supply | |
| CN116031889A (en) | Multi-agent distributed voltage control method based on robust deep reinforcement learning | |
| Hongfei et al. | Optimal control virtual inertia of optical storage microgrid based on improved sailfish algorithm | |
| CN115133540A (en) | Power distribution network model-free real-time voltage control method | |
| CN114298383A (en) | Virtual power plant energy storage capacity configuration method and device | |
| CN114204613A (en) | Reactive compensation method and system for offshore wind farm access power system | |
| Sulaiman et al. | Evolutionary programming versus artificial immune system in evolving neural network for grid-connected photovoltaic system output prediction | |
| CN116231747A (en) | Distributed photovoltaic inverter control method, device and storage medium | |
| CN113708390B (en) | Intelligent equipment real-time control method and system for three-phase imbalance treatment of power distribution station area | |
| Shen et al. | Power control of wind energy conversion system under multiple operating regimes with deep residual recurrent neural network: theory and experiment | |
| CN112134304B (en) | Micro-grid full-automatic navigation method, system and device based on deep learning |
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 |