CN114266349A - Load flow calculation method based on adaptive neural network - Google Patents
Load flow calculation method based on adaptive neural network Download PDFInfo
- Publication number
- CN114266349A CN114266349A CN202111486131.1A CN202111486131A CN114266349A CN 114266349 A CN114266349 A CN 114266349A CN 202111486131 A CN202111486131 A CN 202111486131A CN 114266349 A CN114266349 A CN 114266349A
- Authority
- CN
- China
- Prior art keywords
- neural network
- flow calculation
- calculation method
- adaptive neural
- adaptive
- 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
- 230000003044 adaptive effect Effects 0.000 title claims abstract description 48
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 45
- 238000004364 calculation method Methods 0.000 title claims abstract description 38
- 238000012549 training Methods 0.000 claims abstract description 18
- 238000009826 distribution Methods 0.000 claims abstract description 16
- 238000003062 neural network model Methods 0.000 claims abstract description 14
- 238000005259 measurement Methods 0.000 claims abstract description 9
- 230000001537 neural effect Effects 0.000 claims abstract description 4
- 210000002569 neuron Anatomy 0.000 claims abstract description 4
- 238000000034 method Methods 0.000 claims description 21
- 230000008569 process Effects 0.000 claims description 12
- 239000011159 matrix material Substances 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 4
- 238000009827 uniform distribution Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims 1
- 239000002994 raw material Substances 0.000 claims 1
- 239000000243 solution Substances 0.000 description 12
- 238000004088 simulation Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000000342 Monte Carlo simulation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000002347 injection Methods 0.000 description 1
- 239000007924 injection Substances 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003449 preventive effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a load flow calculation method based on a self-adaptive neural network, which comprises the steps of obtaining data of typical nodes of a power distribution network, and forming a training data set as subsequent input; defining the number of hidden layers and neurons, and establishing a neural self-adaptive neural network model framework; and training the adaptive neural network model by using the training data set, updating parameters in the neural network, and constructing the adaptive data trend model. The invention provides a load flow calculation method based on an adaptive neural network, which is characterized in that an adaptive load flow model is established based on an adaptive neural network model according to an acquired measurement data set, and load flow calculation under the scene of unknown topology and line parameters is realized.
Description
Technical Field
The invention relates to the technical field of power distribution of a power system, in particular to a load flow calculation method based on an adaptive neural network.
Background
The tidal current calculation is an important analysis calculation of the power system to study various problems in system planning and operation. For the power system in the planning, whether the proposed power system planning scheme can meet the requirements of various operation modes can be checked through load flow calculation; for an operating power system, various load changes and changes of a network structure can be predicted through load flow calculation, the safety of the system can not be endangered, whether the voltage of all buses in the system is within an allowable range, whether overload occurs to various elements in the system, and what preventive measures should be taken in advance when the overload possibly occurs, and the like.
However, accurate topology and line parameters are often difficult to obtain because, unlike transmission networks, distribution networks often have frequent topology changes. Some of these changes are artificial changes in the network topology in order to obtain an optimal power flow, but the means for monitoring topology changes may not be known because they cannot cover a large distribution network for economic reasons. Furthermore, many topology changes resulting from power outages or manual maintenance may also be unknown, as engineers in the field may not upload topology change information immediately after repairing the network. These factors make traditional power flow models impossible to solve.
Disclosure of Invention
This section is for the purpose of summarizing some aspects of embodiments of the invention and to briefly introduce some preferred embodiments. In this section, as well as in the abstract and the title of the invention of this application, simplifications or omissions may be made to avoid obscuring the purpose of the section, the abstract and the title, and such simplifications or omissions are not intended to limit the scope of the invention.
The present invention has been made in view of the above and/or other problems occurring in the existing adaptive neural network-based load flow calculation method.
Therefore, the problem to be solved by the present invention is how to provide a load flow calculation method based on an adaptive neural network.
In order to solve the technical problems, the invention provides the following technical scheme: a load flow calculation method based on an adaptive neural network comprises the steps of obtaining data of typical nodes of a power distribution network, and forming a training data set as subsequent input;
defining the number of hidden layers and neurons, and establishing a neural self-adaptive neural network model framework;
and training the adaptive neural network model by using the training data set, updating parameters in the neural network, and constructing the adaptive data trend model.
As a preferable solution of the adaptive neural network-based power flow calculation method of the present invention, wherein: acquiring data of a typical node of the power distribution network comprises acquiring voltage, active and reactive data.
As a preferable solution of the adaptive neural network-based power flow calculation method of the present invention, wherein: in the step of obtaining voltage, active and reactive data of the representative nodes, for a power distribution network with N representative nodes, all the representative nodes are represented by a set N ═ {1, 2, 3, …, N }, and voltage, active and reactive measurement data of all the representative nodes in T time sections are collected, and a data matrix is formed as follows:
wherein T is the total number of measurements(ii) a T is the time period, T is the [1, T ∈]；Vn,tThe voltage at the nth node t is shown; pn,tThe active power at the nth node t moment; qn,tThe reactive power at the nth node t.
As a preferable solution of the adaptive neural network-based power flow calculation method of the present invention, wherein: the forward propagation process of the power flow model with the depth of L +1 is as follows:
f(l)＝h(l)Θ(l),l＝0,...,L
h(l)＝max(0,W(l)h(l-1)+b(l)),l＝1,...,L
h(0)＝x＝[V1,ΔP2,...,ΔPn,ΔQ2,...,ΔQn]T
in the formula, x is the voltage amplitude of the node of the first-end transformer of 1 time section and a 2 n-1-dimensional vector formed by injecting active power and reactive power into the rest nodes; w (l) is the weight matrix of the l-th hidden layer, b (l) is the bias vector of the l-th hidden layer, Θ (l) is the weight matrix of the l-th output layer, α (l) is the weight of the l-th output layer. Unlike the conventional neural network, the final output result of the conventional neural network is h (l), and the output result f (x) of the adaptive neural network model is a weighted combination of h (0), …, h (l).
As a preferable solution of the adaptive neural network-based power flow calculation method of the present invention, wherein: the loss function of the data flow model is as follows:
y＝[ΔP1,ΔQ1,V2,…,Vn]T
in the formula, y is an n + 1-dimensional vector formed by injecting active power, reactive power and voltage amplitudes of other nodes into the first-end transformer nodes of 1 time section.
As a preferable solution of the adaptive neural network-based power flow calculation method of the present invention, wherein: in the process of training the model, the parameters to be learned are α (l), Θ (l), w (l), and b (l).
As a preferable solution of the adaptive neural network-based power flow calculation method of the present invention, wherein: at the first iteration, the weight α of each output layer obeys a uniform distribution, i.e.:
in subsequent iterations, the update process of the weights is as follows:
in the formula, beta epsilon (0, 1) is a discount rate parameter, and t is an iteration number.
As a preferable solution of the adaptive neural network-based power flow calculation method of the present invention, wherein: after each iteration, the weight α is normalized such that
As a preferable solution of the adaptive neural network-based power flow calculation method of the present invention, wherein: the updating process of the hidden layer weight parameters w (l) and b (l) and the output layer parameter Θ (l) is as follows:
in the formula, η is the learning rate.
The invention has the beneficial effects that: a load flow calculation method based on an adaptive neural network is provided, an adaptive load flow model is established based on the adaptive neural network model according to an acquired measurement data set, and load flow calculation under the scene of unknown topology and line parameters is realized.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive exercise. Wherein:
fig. 1 is a scene diagram of a load flow calculation method based on an adaptive neural network.
Fig. 2 is a topology of an IEEE 33 node distribution network.
FIG. 3 is a graph comparing the accuracy of a conventional neural network method and the method of the present invention.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways than those specifically described and will be readily apparent to those of ordinary skill in the art without departing from the spirit of the present invention, and therefore the present invention is not limited to the specific embodiments disclosed below.
Furthermore, reference herein to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one implementation of the invention. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments.
Example 1
Referring to fig. 1, a first embodiment of the present invention provides a power flow calculation method based on an adaptive neural network, where the power flow calculation method based on the adaptive neural network includes the following steps:
s1: and acquiring data of typical nodes of the power distribution network, and forming a training data set as subsequent input.
S2: defining the number of hidden layers and neurons, and establishing a neural self-adaptive neural network model framework.
S3: and training the adaptive neural network model by using the training data set, updating parameters in the neural network, and constructing the adaptive data trend model.
In step S1, acquiring data of a typical node of the distribution network mainly includes acquiring voltage, active and reactive data. In the step of obtaining voltage, active and reactive data of the representative nodes, for a power distribution network with N representative nodes, all the representative nodes are represented by a set N ═ {1, 2, 3, …, N }, and voltage, active and reactive measurement data of all the representative nodes in T time sections are collected, and a data matrix is formed as follows:
wherein T is the total number of times of measurement; t is the time period, T is the [1, T ∈]；Vn,tThe voltage at the nth node t is shown; pn,tThe active power at the nth node t moment; (ii) a Qn,tThe reactive power at the nth node t.
In step S2, an adaptive neural network framework with a depth of L +1 is first established, and then a loss function of the adaptive neural network is constructed.
The forward propagation process of the power flow model with the depth of L +1 is as follows:
f(l)＝h(l)Θ(l),l＝0,...,L
h(l)＝max(0,W(l)h(l-1)+b(l)),l＝1,...,L
h(0)＝x＝[V1,ΔP2,...,ΔPn,ΔQ2,...,ΔQn]T
in the formula, x is the voltage amplitude of the node of the first-end transformer of 1 time section and a 2 n-1-dimensional vector formed by injecting active power and reactive power into the rest nodes; w (l) is the weight matrix of the l-th hidden layer, b (l) is the bias vector of the l-th hidden layer, Θ (l) is the weight matrix of the l-th output layer, α (l) is the weight of the l-th output layer. Unlike the conventional neural network, the final output result of the conventional neural network is h (l), and the output result f (x) of the adaptive neural network model is a weighted combination of h (0), …, h (l).
Constructing a loss function of the adaptive neural network as follows:
y＝[ΔP1,ΔQ1,V2,…,Vn]T
in the formula, y is an n + 1-dimensional vector formed by injecting active power, reactive power and voltage amplitudes of other nodes into the first-end transformer nodes of 1 time section.
In step S3, the parameters to be learned in the process of training the model are α (l), Θ (l), w (l), and b (l). At the first iteration, the weight α of each output layer obeys a uniform distribution, i.e.:
in subsequent iterations, the update process of the weights is as follows:
in the formula, beta epsilon (0, 1) is a discount rate parameter, and t is an iteration number. After each iteration, the weight α is normalized such that
The updating process of the hidden layer weight parameters w (l) and b (l) and the output layer parameter Θ (l) is as follows:
in the formula, η is the learning rate.
And then updating model parameters in an online learning mode by utilizing the training data set to obtain a self-adaptive data load flow model, and realizing load flow calculation under the scene that the topology and the line parameters of the power distribution network are unknown.
Example 2
Referring to fig. 2 and 3, a second embodiment of the present invention is based on the previous embodiment.
Specifically, simulation and analysis are performed on an IEEE 33 node system, and the topology structure of the IEEE 33 node power distribution network is shown in fig. 2. In fig. 2, node 1 is a transformer node, and the voltage amplitude of the node is kept constant. The system was operated radially with a reference voltage of 12.66 kV.
In this embodiment, first, the injection active and reactive data of the node are randomly generated by using a monte carlo method. And then, carrying out forward-backward substitution load flow calculation to generate node voltage amplitude values under different scenes. Finally, the generated active, reactive and voltage amplitude data are input into the model for training the model, wherein 80% is the training data set and 20% is the testing data set. All simulations were done on a computer equipped with a 2.11GHz Intel Core i7 processor and 8G memory.
To verify the validity of the model, the difference between the estimated value and the true value of the model is described by Mean Relative Error (MRE), which is expressed as follows:
in the formula, yi is the ith element of the truth vector, i is the ith element of the model output vector, and n is the number of the elements.
Compared with the traditional method, the method has higher accuracy.
In this embodiment, the accuracy is compared by using the conventional neural network method and the method, and the comparison result is shown in fig. 3.
As can be seen from fig. 3, the accuracy of the method of the present invention is significantly higher than that of the conventional neural network method, which proves that the method of the present invention can establish an adaptive power flow model based on an adaptive neural network model according to the collected measurement data set, thereby realizing power flow calculation in a scenario where topology and line parameters are unknown.
It should be noted that the above-mentioned embodiments are only for illustrating the technical solutions of the present invention and not for limiting, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions may be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, which should be covered by the claims of the present invention.
Claims (9)
1. A load flow calculation method based on an adaptive neural network is characterized in that: comprises the steps of (a) preparing a mixture of a plurality of raw materials,
acquiring data of typical nodes of the power distribution network, and forming a training data set as subsequent input;
defining the number of hidden layers and neurons, and establishing a neural self-adaptive neural network model framework;
and training the adaptive neural network model by using the training data set, updating parameters in the neural network, and constructing the adaptive data trend model.
2. The adaptive neural network-based power flow calculation method according to claim 1, wherein: acquiring data of a typical node of the power distribution network comprises acquiring voltage, active and reactive data.
3. The adaptive neural network-based power flow calculation method according to claim 2, wherein: in the step of obtaining voltage, active and reactive data of the representative nodes, for a power distribution network with N representative nodes, all the representative nodes are represented by a set N ═ {1, 2, 3, …, N }, and voltage, active and reactive measurement data of all the representative nodes in T time sections are collected, and a data matrix is formed as follows:
wherein T is the total number of times of measurement; t is the time period, T is the [1, T ∈]；Vn,tThe voltage at the nth node t is shown; pn,tThe active power at the nth node t moment; qn,tThe reactive power at the nth node t.
4. The adaptive neural network-based power flow calculation method according to any one of claims 1 to 3, wherein: the forward propagation process of the power flow model with the depth of L +1 is as follows:
f(l)＝h(l)Θ(l),l＝0,...,L
h(l)＝max(0,W(l)h(l-1)+b(l)),l＝1,...,L
h(0)＝x＝[V1,ΔP2,...,ΔPn,ΔQ2,...,ΔQn]T
in the formula, x is the voltage amplitude of the node of the first-end transformer of 1 time section and a 2 n-1-dimensional vector formed by injecting active power and reactive power into the rest nodes; w (l) is the weight matrix of the l-th hidden layer, b (l) is the bias vector of the l-th hidden layer, Θ (l) is the weight matrix of the l-th output layer, α (l) is the weight of the l-th output layer. Unlike the conventional neural network, the final output result of the conventional neural network is h (l), and the output result f (x) of the adaptive neural network model is a weighted combination of h (0), …, h (l).
5. The adaptive neural network-based power flow calculation method according to claim 4, wherein: the loss function of the data flow model is as follows:
y＝[ΔP1,ΔQ1,V2,…,Vn]T
in the formula, y is an n + 1-dimensional vector formed by injecting active power, reactive power and voltage amplitudes of other nodes into the first-end transformer nodes of 1 time section.
6. The adaptive neural network-based power flow calculation method according to any one of claims 1 to 3 or 5, wherein: in the process of training the model, the parameters to be learned are α (l), Θ (l), w (l), and b (l).
7. The adaptive neural network-based power flow calculation method according to claim 6, wherein: at the first iteration, the weight α of each output layer obeys a uniform distribution, i.e.:
in subsequent iterations, the update process of the weights is as follows:
in the formula, beta epsilon (0, 1) is a discount rate parameter, and t is an iteration number.
8. The adaptive neural network-based power flow calculation method according to claim 7, wherein: after each iteration, the weight α is normalized such that
9. The adaptive neural network-based power flow calculation method according to claim 8, wherein: the updating process of the hidden layer weight parameters w (l) and b (l) and the output layer parameter Θ (l) is as follows:
in the formula, η is the learning rate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111486131.1A CN114266349A (en) | 2021-12-07 | 2021-12-07 | Load flow calculation method based on adaptive neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111486131.1A CN114266349A (en) | 2021-12-07 | 2021-12-07 | Load flow calculation method based on adaptive neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114266349A true CN114266349A (en) | 2022-04-01 |
Family
ID=80826448
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111486131.1A Pending CN114266349A (en) | 2021-12-07 | 2021-12-07 | Load flow calculation method based on adaptive neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114266349A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117650533A (en) * | 2024-01-26 | 2024-03-05 | 国网冀北电力有限公司 | Power system power flow analysis method and device based on graph network big data model |
-
2021
- 2021-12-07 CN CN202111486131.1A patent/CN114266349A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117650533A (en) * | 2024-01-26 | 2024-03-05 | 国网冀北电力有限公司 | Power system power flow analysis method and device based on graph network big data model |
| CN117650533B (en) * | 2024-01-26 | 2024-04-12 | 国网冀北电力有限公司 | Power system power flow analysis method and device based on graph network big data model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110046408B (en) | Interval uncertainty optimization method based on BP neural network differentiation and interval analysis | |
| CN111257934B (en) | Seismic oscillation peak acceleration prediction method based on second-order neuron deep neural network | |
| CN112310980B (en) | Safety and stability evaluation method and system for direct-current blocking frequency of alternating-current and direct-current series-parallel power grid | |
| CN105138717A (en) | Transformer state evaluation method by optimizing neural network with dynamic mutation particle swarm | |
| CN111415010A (en) | Bayesian neural network-based wind turbine generator parameter identification method | |
| CN112884236B (en) | Short-term load prediction method and system based on VDM decomposition and LSTM improvement | |
| CN109325613B (en) | Method for online predicting static voltage stability margin by using parallel sequential extreme learning machine | |
| CN106684885A (en) | Multi-scene analysis-based reactive power optimization method for power distribution network of wind power generation set | |
| CN110390461B (en) | Nonlinear fuzzy language power distribution network node vulnerability evaluation method based on complex network | |
| CN115940294B (en) | Multi-stage power grid real-time scheduling strategy adjustment method, system, equipment and storage medium | |
| CN113837432A (en) | Power system frequency prediction method driven by physics-data combination | |
| Stiasny et al. | Transient stability analysis with physics-informed neural networks | |
| CN114266349A (en) | Load flow calculation method based on adaptive neural network | |
| Su et al. | On the dynamical modeling with neural fuzzy networks | |
| CN114154753A (en) | Load prediction method and system | |
| CN113919221A (en) | Fan load prediction and analysis method and device based on BP neural network and storage medium | |
| CN117291069A (en) | LSTM sewage water quality prediction method based on improved DE and attention mechanism | |
| Mahala et al. | Active & reactive power rescheduling for congestion management using new PSO strategy | |
| CN110751328A (en) | High-proportion renewable energy power grid adaptive planning method based on joint weighted entropy | |
| CN111950765B (en) | Probabilistic transient stability prediction method based on stacked noise reduction self-encoder | |
| Tajalli et al. | Maximizing social welfare considering the uncertainty of wind power plants using a distributed consensus-based algorithm | |
| CN114142467A (en) | Power distribution network photovoltaic maximum access capacity measuring and calculating method based on non-precise modeling power flow model | |
| CN114330649A (en) | Voltage regulation method and system based on evolutionary learning and deep reinforcement learning | |
| CN113820954A (en) | Fault-tolerant control method of complex nonlinear system under generalized noise | |
| Xu | Risk assessment of green intelligent building based on artificial intelligence |
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 |