CN113988481B - Wind power prediction method based on dynamic matrix prediction control - Google Patents
Wind power prediction method based on dynamic matrix prediction control Download PDFInfo
- Publication number
- CN113988481B CN113988481B CN202111585752.5A CN202111585752A CN113988481B CN 113988481 B CN113988481 B CN 113988481B CN 202111585752 A CN202111585752 A CN 202111585752A CN 113988481 B CN113988481 B CN 113988481B
- Authority
- CN
- China
- Prior art keywords
- wind power
- prediction
- time
- neural network
- dynamic matrix
- 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.)
- Active
Links
- 239000011159 matrix material Substances 0.000 title claims abstract description 87
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000013528 artificial neural network Methods 0.000 claims abstract description 59
- 210000002569 neuron Anatomy 0.000 claims description 44
- 238000012937 correction Methods 0.000 claims description 37
- 230000004044 response Effects 0.000 claims description 22
- 230000009471 action Effects 0.000 claims description 14
- 239000000126 substance Substances 0.000 claims description 8
- 230000004913 activation Effects 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 5
- 150000001875 compounds Chemical class 0.000 claims description 3
- 210000002364 input neuron Anatomy 0.000 claims description 3
- 230000003287 optical effect Effects 0.000 claims description 3
- 210000004205 output neuron Anatomy 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 238000010248 power generation Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000035772 mutation Effects 0.000 description 2
- 230000008121 plant development Effects 0.000 description 2
- 238000005096 rolling process Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- 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
- H02J3/004—Generation forecast, e.g. methods or systems for forecasting future energy generation
-
- 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
- H02J2300/28—The renewable source being wind energy
-
- 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
- Y02E40/70—Smart grids as climate change mitigation technology in the energy generation sector
-
- 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
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P80/00—Climate change mitigation technologies for sector-wide applications
- Y02P80/10—Efficient use of energy, e.g. using compressed air or pressurized fluid as energy carrier
-
- 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
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P80/00—Climate change mitigation technologies for sector-wide applications
- Y02P80/20—Climate change mitigation technologies for sector-wide applications using renewable energy
-
- 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
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- General Business, Economics & Management (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Tourism & Hospitality (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Primary Health Care (AREA)
- Development Economics (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Power Engineering (AREA)
- Wind Motors (AREA)
Abstract
The invention discloses a wind power prediction method based on dynamic matrix prediction control, which adopts more than two initial wind power prediction submodels, then adopts a first dynamic matrix prediction control system to correct the predicted values of the more than two initial wind power prediction submodels, secondly dynamically determines the weight of the predicted value of the initial wind power prediction submodel through an echo state neural network, and finally adopts a second dynamic matrix prediction control system to correct the predicted value of the echo state neural network, and then utilizes the dynamic matrix prediction control.
Description
Technical Field
The invention relates to a wind power prediction method based on dynamic matrix prediction control, and belongs to the technical field of wind power prediction.
Background
The wind power prediction is that a prediction model of wind power plant output power is established by using data of historical power, historical wind speed, landform, numerical weather forecast, wind turbine generator operating state and the like of a wind power plant, the data of the wind speed, the power, the numerical weather forecast and the like are used as input of the model, and the future active power of the wind power plant is predicted by combining the equipment state and the operating condition of the wind power plant generator.
The wind power prediction system has important significance for the operation of a power system connected with a large amount of wind power. The power system is a complex dynamic system, and it is the responsibility of the power grid to maintain the power balance among power generation, power transmission and power utilization. Without a wind power system, the power grid dispatching mechanism can make a power generation plan according to the daily load curve, so that the demand of the next day power is met. The output power of the wind power plant has volatility and intermittency, the difficulty of making a power generation plan is greatly increased due to large-scale access of wind power, and the wind power brings huge challenges to the dispatching and operation of a power system. The method is one of effective means for relieving peak load and frequency modulation pressure of the power system and improving wind power receiving capacity, meanwhile, wind power plant development enterprises can also reasonably arrange the maintenance of wind turbine equipment by selecting weather with smaller wind power by using wind power forecast, and the generated energy loss caused by the fact that the wind turbine cannot generate power is reduced as far as possible.
The existing wind power prediction methods can be mainly divided into methods such as a physical model method, a statistical method, a learning method and the like, but the methods all consider the wind power as a stable time sequence, and are difficult to predict the power mutation phenomenon caused by mutation events such as fan blade icing caused by strong wind shutdown and temperature reduction, so the wind power prediction accuracy is greatly reduced under extreme disaster conditions. In order to solve the problems, a relatively wide idea is to adopt a combined prediction model, so that the problem that a single model has a large prediction point error is overcome, but the combined prediction model is particularly important for determining the weight coefficient of each sub-model, the existing methods include an equal weight average method, a minimum variance method, an unconstrained least square method and the like, the weight coefficient of the combined prediction mode is mostly kept unchanged, and the prediction accuracy is limited when the combined prediction mode meets an uncertain weather state. Other methods for determining the weight coefficient include a probabilistic weight method, an information entropy method, a wavelet decomposition, a genetic algorithm, a neural network, and the like, and although these methods can achieve dynamic adjustment of the weight of each submodel and also achieve wind power prediction accuracy, the wind power prediction accuracy depends on the construction of the submodel, and is also limited in the case of an uncertain weather state.
Disclosure of Invention
The purpose of the invention is as follows: in order to overcome the defects in the prior art, the invention provides the wind power prediction method based on dynamic matrix prediction control, which has high accuracy and high calculation speed.
The technical scheme is as follows: in order to achieve the purpose, the invention adopts the technical scheme that:
a wind power prediction method based on dynamic matrix predictive control comprises the following steps:
And 2, establishing more than two initial wind power prediction submodels according to the historical prediction data set of the initial wind power. Obtaining a prediction data set according to the history of the initial wind powerAt the first momentWind power prediction value of initial wind power prediction submodel,,Representing the number of sub-models of the initial wind power predictor.
Step 3, the product obtained in the step 2 is processedAt the first momentWind power prediction value of initial wind power prediction submodelAnd detectedActual wind power at time of dayComparing to obtain the prediction error of the sub-model of the initial wind power predictor,。
Step 4, predicting errors of the initial wind power prediction submodels obtained in the step 4 through a first dynamic matrix prediction control systemFeeding back the wind power predicted value of the initial wind power prediction submodel, and carrying out feedback correction on the wind power predicted value of the initial wind power prediction submodel to obtain a wind power prediction correction value of the initial wind power prediction submodelAnd further obtaining a wind power prediction correction matrix of the initial wind power prediction submodel,Indicating the prediction time.
And 5, constructing an echo state neural network, wherein the echo state neural network comprises an input layer, a reserve pool and an output layer which are sequentially connected.
During training, inputting the wind power prediction correction matrix of the initial wind power prediction submodel obtained in the step 4 into an echo state neural network, and training the echo state neural network, wherein a least square method is adopted to estimate a connection weight matrix from the neuron in the reserve pool to the neuron in the output layerAnd obtaining the trained echo state neural network.
And 6, processing the historical prediction data set of the echo state neural network through the steps 2 to 4 to obtain an echo state neural network correction set. Inputting the obtained echo state neural network correction set into the trainedObtaining the predicted value of the neural network in the echo state。
Step 7, the echo state neural network predicted value obtained in the step 6 is usedAnd detectedActual wind power at timeComparing to obtain wind power prediction error,。
Step 8, the wind power prediction error obtained in the step 7 is subjected to prediction by a second dynamic matrix prediction control systemIs fed back toNeural network prediction of echo state at time, pairThe wind power predicted value of the echo state neural network predicted value at the moment is subjected to feedback correction to obtain the wind power predicted value after the second correctionTemporal echo state neural network prediction correction。
Preferably: the step input of the first dynamic matrix predictive control system is the prediction error of the initial wind power predictor sub-model, and the sampling data of the corresponding wind power unit step response is obtained,,Modeling a time domain for a step response prediction model of a first dynamic matrix predictive control system, the step response prediction model of the first dynamic matrix predictive control system:
wherein the content of the first and second substances,is at the same timeTime of day prediction futureAt a moment in time haveA continuous control increment,,… , Wind power prediction under influenceThe output of the value is carried out,to control the time domain length.Is at the same timeTime of day prediction futureOutputting the wind power predicted value under the action of no control increment at each moment,is a dynamic matrix whose elements are step response coefficients describing the dynamic behavior of the system,for the future from now onThe control variable at each instant.
In thatWind power predicted at time without control incrementUnder the action ofThe output predicted value at each time is。
In thatPredicted wind power at the momentWith control incrementsUnder the action ofThe output predicted value at each time is。
Preferably: the feedback correction of the first dynamic matrix predictive control system is as follows: according toConstantly exerting control action on wind power valueThen, atThe actual output is collected at all timesAndoutput prediction value of step response prediction model based on first dynamic matrix prediction control system at momentComparing to obtain real-time prediction error:
By predicting the error in real timeThe weighting modifies the prediction of the future output, i.e.:
in the formula (I), the compound is shown in the specification,is composed ofPredicted future of time error correctedThe predicted wind power at each moment in time,is an error correction vector, i.e. a weight coefficient applied when error correcting the predicted values at different time instants, wherein,。
preferably: the optimized performance indexes of the first dynamic matrix predictive control system are as follows:
wherein the content of the first and second substances,the performance index of the optimization is shown,a weighting coefficient representing a tracking error of the optical disc,show in the futureThe desired value of the output wind power is rolled in time,a weighting factor representing a control increment.
A reserve pool: is composed ofSparse network formed by connecting individual neurons, and weight matrix for connection between neurons in reserve poolIt is shown that,representing neurons within the reservoirAnd neuronsConnection weight, connection weight matrixThe connection weight between the neuron in the input layer and the neuron in the dynamic reserve pool is expressed by a matrixAnd (4) showing. The pool accepts inputs in two directions, one from the input layer and the other from the state of the previous output layer.
An output layer: comprises 1 output neuron, a matrix for connecting weights from neuron in reserve pool to neuron in output layerRepresenting a matrix of connection weights from neurons within the output layer to neurons within the reservoirAnd (4) showing.
wherein the content of the first and second substances,is shown inThe input layer input is made at the time of day,is shown inThe output of the output layer at the time instant,is shown inThe neural network prediction value of the echo state at the moment,is shown inAt the moment, the first part inside the reserve tankThe state of the individual neurons is known,is shown inAt that moment, the state in the reserve tank. Initializing the state of a reserve pool,。
The internal state of the reserve pool neurons is updated as follows:
the internal state of the output layer neurons is updated as follows:
which represents a non-linear activation function,which represents a linear activation function, is shown,indicating the amount of offset.
Preferably: the sparse linking rate is 0.01-0.05.
Preferably:the step input of the second dynamic matrix prediction control system is wind power prediction error, and corresponding sampling data of wind power unit step response are obtained,,Modeling a time domain for a step response prediction model of a second dynamic matrix predictive control system.
Compared with the prior art, the invention has the following beneficial effects:
the method adopts more than two initial wind power prediction submodels, then adopts a first dynamic matrix prediction control system to correct the predicted values of the more than two initial wind power prediction submodels, and makes up the inaccuracy of the traditional wind power prediction method in wind power prediction calculation under extreme conditions through feedback correction and limited time domain rolling optimization. Secondly, dynamically determining the weight of the predicted value of the initial wind power prediction submodel through an echo state neural network, preventing uncertain data from influencing the predicted value, and well integrating the characteristics of each initial wind power prediction submodel. Finally, the second dynamic matrix prediction control system is adopted to correct the predicted value of the echo state neural network, the dynamic matrix prediction control is utilized again, based on feedback correction and finite time domain rolling optimization, the prediction failure of the echo state neural network is avoided, and by the means, the influence on the inaccuracy of wind power prediction under extreme conditions can be well relieved, the optimal prediction of future wind power of the wind power station is realized, and meanwhile, the wind power prediction precision is high.
Drawings
FIG. 1 is a schematic flow chart of an embodiment of the present invention.
Detailed Description
The present invention is further illustrated by the following description in conjunction with the accompanying drawings and the specific embodiments, it is to be understood that these examples are given solely for the purpose of illustration and are not intended as a definition of the limits of the invention, since various equivalent modifications will occur to those skilled in the art upon reading the present invention and fall within the limits of the appended claims.
A wind power prediction method based on dynamic matrix predictive control is disclosed, as shown in FIG. 1, and comprises the following steps:
And 2, establishing more than two initial wind power prediction submodels according to the historical prediction data set of the initial wind power. Obtaining a prediction data set according to the history of the initial wind powerAt the first momentWind power prediction value of initial wind power prediction submodel,,Representing the number of sub-models of the initial wind power predictor. The initial wind power prediction submodel includes a numerical weather wind power prediction model for simulating atmospheric dynamics based on a physical principle and boundary conditions, a wind power prediction model based on autoregressive moving average, a wind power prediction model based on kalman filtering, a wind power prediction model based on a markov chain, a wind power prediction model based on a gray theory, and a wind power prediction model based on artificial intelligence (neural network, support vector machine, limit learning machine, etc.), that is, the initial wind power prediction submodel in this embodiment may be an existing wind power prediction model.
Step 3, the product obtained in the step 2 is processedAt the first momentWind power prediction value of initial wind power prediction submodelAnd detectedActual wind power at time of dayComparing to obtain the prediction error of the sub model of the initial wind power predictor,。
Step 4, predicting errors of the initial wind power prediction submodels obtained in the step 4 through a first dynamic matrix prediction control systemFeeding back the wind power predicted value of the initial wind power prediction submodel, and carrying out feedback correction on the wind power predicted value of the initial wind power prediction submodel to obtain a wind power prediction correction value of the initial wind power prediction submodelAnd further obtaining a wind power prediction correction matrix of the initial wind power prediction submodel,Indicating the prediction time.
The step input of the first dynamic matrix predictive control system is the prediction error of the initial wind power prediction submodel, and the sampling data of the corresponding wind power unit step response is obtained,,Modeling a time domain for a step response prediction model of a first dynamic matrix predictive control system, the step response prediction model of the first dynamic matrix predictive control system:
wherein the content of the first and second substances,is at the same timeTime of day prediction futureAt a moment in time haveA continuous control increment,,… , Outputting the wind power predicted value under the action,in order to control the length of the time domain,indicating a roll-optimized temporal length.Is at the same timeTime of day prediction futureOutputting the wind power predicted value under the action of no control increment at each moment,is a dynamic matrix whose elements are step response coefficients describing the dynamic behavior of the system,for the future from now onThe control variable at each moment.
In thatWind power predicted at time without control incrementUnder the action ofThe output predicted value at each time is。
In thatThe wind power predicted at the moment has control incrementUnder the action ofThe output predicted value at each time is。
The feedback correction of the first dynamic matrix predictive control system is as follows: according toConstantly exerting control action on wind power valueThen, atThe actual output is collected at all timesAndoutput prediction value of step response prediction model based on first dynamic matrix prediction control system at momentComparing to obtain real-time prediction error:
By predicting the error in real timeThe weighting modifies the prediction of the future output, i.e.:
in the formula (I), the compound is shown in the specification,is composed ofPredicted future of time corrected by errorThe predicted wind power at each moment in time,is an error correction vector, i.e. a weight coefficient applied when error correcting the predicted values at different time instants, wherein,。
the optimized performance indexes of the first dynamic matrix predictive control system are as follows:
wherein the content of the first and second substances,the performance index of the optimization is shown,a weighting coefficient representing a tracking error of the optical disc,show in the futureThe desired value of the output wind power is rolled in time,a weighting factor representing a control increment.
Step 5, constructing an echo state neural network, wherein the echo state neural network comprises an input layer, a reserve pool and an output layer, and the method comprises the following steps:
A reserve pool: is composed ofA sparse network formed by connecting a plurality of neurons,for connection weight matrix between neurons in reserve poolIt is shown that,representing neurons within the reservoirAnd neuronsConnection weight, connection weight matrixIs a sparse matrix, the sparsity of the connection between neurons in the reserve pool is expressed by a sparse connection rate, which is generally selected to be 0.01-0.05, and in order to make the network have the property of echo state to ensure the stability of the network, a connection weight matrixMust be smaller than 1. Matrix for connection weights between neurons in input layer and neurons in dynamic reservoirAnd (4) showing. The pool accepts inputs in two directions, one from the input layer and the other from the state of the previous output layer.
An output layer: comprises 1 output neuron, a matrix for connecting weights from neuron in reserve pool to neuron in output layerRepresenting a matrix of connection weights from neurons within the output layer to neurons within the reservoirAnd (4) showing.
is shown inThe input layer input is made at the time of day,is shown inThe output of the output layer at the time instant,is shown inThe neural network prediction value of the echo state at the moment,is shown inAt the moment, the first part inside the reserve tankThe state of the individual neurons is known,is shown inAt that moment, the state in the reserve tank. Initializing the status of a reserve pool,。
The internal state of the reserve pool neurons is updated as follows:
the internal state of the output layer neurons is updated as follows:
which is representative of a non-linear activation function,which represents a linear activation function, is shown,indicating the amount of offset.
During training, inputting the wind power prediction correction matrix of the initial wind power prediction submodel obtained in the step 4 into the neural network in the echo state, and inputting the correction matrix into the neural network in the echo stateTraining the channels, wherein a least square method is adopted to estimate a connection weight matrix from the neuron in the reserve pool to the neuron in the output layerAnd obtaining the trained echo state neural network.
And 6, processing the historical prediction data set of the neural network in the echo state through the steps 2 to 4 to obtain a correction set of the neural network in the echo state. Inputting the obtained echo state neural network correction set into the trained echo state neural network to obtain the predicted value of the echo state neural network。
Step 7, the echo state neural network predicted value obtained in the step 6 is usedAnd detectedActual wind power at time of dayComparing to obtain wind power prediction error,。
Step 8, the wind power prediction error obtained in the step 7 is subjected to prediction by a second dynamic matrix prediction control systemIs fed back toNeural network prediction of echo state at time, pairThe wind power predicted value of the echo state neural network predicted value at the moment is subjected to feedback correction to obtain the wind power predicted value after the second correctionTemporal echo state neural network prediction correction。
The step input of the second dynamic matrix prediction control system is wind power prediction error, and corresponding sampling data of wind power unit step response are obtained,,And modeling a time domain for the step response prediction model of the second dynamic matrix prediction control system, wherein other step response prediction models, feedback correction and optimization performance indexes are similar to those of the first dynamic matrix prediction control system and are not repeated herein.
The embodiment carries out dynamic matrix control prediction for the submodels once through twice dynamic matrix control prediction, carries out dynamic matrix control prediction for the echo state neural network once, and simultaneously adopts the echo state neural network to carry out dynamic weight determination for each submodel, therefore, the influence of extreme conditions on wind power can be effectively prevented, the inherent volatility and intermittent defects of wind power are reduced simultaneously, the difficulty of making a power generation plan caused by large-scale access of the wind power is reduced, the challenge brought by the wind power to the dispatching operation of a power system is reduced, the peak regulation and the frequency modulation pressure of the power system are relieved, and the wind power receiving capacity is improved. In addition, wind power plant development enterprises can also reasonably arrange the maintenance of the wind turbine equipment by selecting weather with smaller wind power by using wind power forecast, and the power generation loss caused by the fact that the wind turbine cannot generate power due to maintenance is reduced as much as possible.
The above description is only of the preferred embodiments of the present invention, and it should be noted that: it will be apparent to those skilled in the art that various modifications and adaptations can be made without departing from the principles of the invention and these are intended to be within the scope of the invention.
Claims (8)
1. A wind power prediction method based on dynamic matrix predictive control is characterized by comprising the following steps:
step 1, collecting historical wind speed, topographic features, numerical weather forecast and historical wind power of a wind power plant as a historical data set, and dividing the historical data set into an initial wind power historical prediction data set and an echo state neural network historical prediction data set;
step 2, establishing more than two initial wind power prediction submodels according to the historical prediction data set of the initial wind power; obtaining a prediction data set according to the history of the initial wind powerAt the first momentWind power prediction value of initial wind power prediction submodel,,Representing the number of the initial wind power prediction submodels;
step 3, the product obtained in the step 2At the first momentWind power prediction value of initial wind power prediction submodelAnd detectedActual wind power at time of dayComparing to obtain the prediction error of the sub-model of the initial wind power predictor,;
Step 4, predicting errors of the initial wind power prediction submodels obtained in the step 4 through a first dynamic matrix prediction control systemFeeding back the wind power predicted value of the initial wind power prediction submodel, and carrying out feedback correction on the wind power predicted value of the initial wind power prediction submodel to obtain a wind power prediction correction value of the initial wind power prediction submodelAnd further obtaining a wind power prediction correction matrix of the initial wind power prediction submodel,Representing a predicted time;
step 5, constructing an echo state neural network, wherein the echo state neural network comprises an input layer, a reserve pool and an output layer which are connected in sequence;
during training, inputting the wind power prediction correction matrix of the initial wind power prediction submodel obtained in the step 4 into an echo state neural network, and training the echo state neural network, wherein a least square method is adopted to estimate a connection weight matrix from the neuron in the reserve pool to the neuron in the output layerObtaining a trained echo state neural network;
step 6, processing the historical prediction data set of the echo state neural network through the steps 2 to 4 to obtain an echo state neural network correction set; inputting the obtained echo state neural network correction set into the trained echo state neural network to obtain the predicted value of the echo state neural network;
Step 7, the echo state neural network predicted value obtained in the step 6 is usedAnd detectedActual wind power at time of dayComparing to obtain wind power prediction error,;
Step 8, the wind power prediction error obtained in the step 7 is subjected to prediction by a second dynamic matrix prediction control systemIs fed back toNeural network prediction of echo state at time, pairThe wind power predicted value of the echo state neural network predicted value at the moment is subjected to feedback correction to obtain the wind power predicted value after the second correctionTemporal echo state neural network prediction correction。
2. The wind power prediction method based on dynamic matrix predictive control according to claim 1, characterized in that: the step input of the first dynamic matrix predictive control system is the prediction error of the initial wind power predictor sub-model, and the sampling data of the corresponding wind power unit step response is obtained,,Modeling the time domain for a step response prediction model of a first dynamic matrix predictive control system, firstStep response prediction model of dynamic matrix predictive control system:
wherein the content of the first and second substances,is at leastTime of day prediction futureAt a moment in time haveA continuous control increment,,… , Outputting the wind power predicted value under the action,to control the time domain length;is at the same timeTime of day prediction futureOutputting the wind power predicted value under the action of no control increment at each moment,is a dynamic matrix whose elements are step response coefficients describing the dynamic behavior of the system,for the future from now onA control variable at each time;
in thatWind power predicted at time without control incrementUnder the action ofThe output predicted value at each time is;
3. The wind power prediction method based on dynamic matrix predictive control according to claim 2, characterized in that: the feedback correction of the first dynamic matrix predictive control system is as follows: according toConstantly exerting control action on wind power valueThen, atThe actual output is collected at all timesAndoutput prediction value of step response prediction model based on first dynamic matrix prediction control system at momentComparing to obtain real-time prediction error:
in the formula (I), the compound is shown in the specification,is composed ofPredicted future of time error correctedThe predicted wind power at each moment in time,is an error correction vector, i.e. a weight coefficient applied when error correcting the predicted values at different time instants, wherein,。
4. the wind power prediction method based on dynamic matrix predictive control according to claim 3, characterized in that: the optimized performance indexes of the first dynamic matrix predictive control system are as follows:
wherein the content of the first and second substances,the performance index of the optimization is shown,a weighting coefficient representing a tracking error of the optical disc,show in the futureThe desired value of the output wind power is rolled in time,a weighting factor representing a control increment.
5. The wind power prediction method based on dynamic matrix predictive control according to claim 4, characterized in that: the input layer: comprises thatAn input neuron;
a reserve pool: is composed ofSparse network formed by connecting individual neurons, and connection weight matrix between neurons in reserve pool, Representing neurons within the reservoirAnd neuronsConnection weight, connection weight matrixIs a sparse matrix, the sparsity of connection between neurons in the reservoir is expressed by sparse connection rate, and the input layerMatrix for connection weights between internal neurons and internal neurons in dynamic reservoirRepresents; the reserve pool accepts inputs in two directions, one from the input layer and the other from the output of the state of the previous output layer;
an output layer: comprises 1 output neuron, a matrix for connecting weights from neuron in reserve pool to neuron in output layerRepresenting a matrix of connection weights from neurons within the output layer to neurons within the reservoirRepresents;
wherein the content of the first and second substances,is shown inThe input layer input is made at the time of day,is shown inThe output of the output layer at the time instant,is shown inThe neural network prediction value of the echo state at the moment,is shown inAt the moment, the first part inside the reserve tankThe state of the individual neurons is known,is shown inThe state in the reserve tank at the moment; initializing the status of a reserve pool,;
The internal state of the reserve pool neurons is updated as follows:
the internal state of the output layer neurons is updated as follows:
6. The wind power prediction method based on dynamic matrix predictive control according to claim 5, characterized in that: the sparse linking rate is 0.01-0.05.
8. The wind power prediction method based on dynamic matrix predictive control according to claim 7, characterized in that: the step input of the second dynamic matrix predictive control system is a wind power prediction error, and sampling data of corresponding wind power unit step response are obtained,,Modeling a time domain for a step response prediction model of a second dynamic matrix predictive control system.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111585752.5A CN113988481B (en) | 2021-12-23 | 2021-12-23 | Wind power prediction method based on dynamic matrix prediction control |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111585752.5A CN113988481B (en) | 2021-12-23 | 2021-12-23 | Wind power prediction method based on dynamic matrix prediction control |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113988481A CN113988481A (en) | 2022-01-28 |
CN113988481B true CN113988481B (en) | 2022-05-03 |
Family
ID=79734108
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111585752.5A Active CN113988481B (en) | 2021-12-23 | 2021-12-23 | Wind power prediction method based on dynamic matrix prediction control |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113988481B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114583767B (en) * | 2022-03-10 | 2023-03-17 | 中国电力科学研究院有限公司 | Data-driven wind power plant frequency modulation response characteristic modeling method and system |
CN116316617B (en) * | 2023-05-26 | 2023-09-15 | 南方电网数字电网研究院有限公司 | Multi-station intelligent fusion new energy generation power region prediction method and system |
CN117748501B (en) * | 2024-02-19 | 2024-05-07 | 西安热工研究院有限公司 | Wind power prediction method and system for energy storage auxiliary black start |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102831475A (en) * | 2012-08-01 | 2012-12-19 | 华北电力大学 | Sine normalization method for power forecast model of wind power plant |
CN107947166A (en) * | 2017-12-01 | 2018-04-20 | 湖南水利水电职业技术学院 | Dispatching method and device when a kind of multipotency microgrid based on dynamic matrix control becomes |
CN113193556A (en) * | 2021-05-12 | 2021-07-30 | 燕山大学 | Short-term wind power prediction method based on probability prediction model |
CN113610665A (en) * | 2021-08-12 | 2021-11-05 | 东北电力大学 | Wind power generation power prediction method based on multi-delay output echo state network |
-
2021
- 2021-12-23 CN CN202111585752.5A patent/CN113988481B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102831475A (en) * | 2012-08-01 | 2012-12-19 | 华北电力大学 | Sine normalization method for power forecast model of wind power plant |
CN107947166A (en) * | 2017-12-01 | 2018-04-20 | 湖南水利水电职业技术学院 | Dispatching method and device when a kind of multipotency microgrid based on dynamic matrix control becomes |
CN113193556A (en) * | 2021-05-12 | 2021-07-30 | 燕山大学 | Short-term wind power prediction method based on probability prediction model |
CN113610665A (en) * | 2021-08-12 | 2021-11-05 | 东北电力大学 | Wind power generation power prediction method based on multi-delay output echo state network |
Also Published As
Publication number | Publication date |
---|---|
CN113988481A (en) | 2022-01-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113988481B (en) | Wind power prediction method based on dynamic matrix prediction control | |
Kariniotakis et al. | Wind power forecasting using advanced neural networks models | |
CN108564206B (en) | Wind power prediction method based on distributed optimization and spatial correlation | |
CN111461463B (en) | Short-term load prediction method, system and equipment based on TCN-BP | |
CN102478584B (en) | Wind power station wind speed prediction method based on wavelet analysis and system thereof | |
CN112598180A (en) | Distributed regional wind power prediction method | |
CN110555548B (en) | ICS-ELM ultra-short-term wind power prediction method based on data mining original error correction | |
CN104036328A (en) | Self-adaptive wind power prediction system and prediction method | |
US11808915B2 (en) | Wind power prediction method and system based on deep deterministic policy gradient algorithm | |
CN114757427A (en) | Autoregressive corrected LSTM intelligent wind power plant ultra-short term power prediction method | |
CN116885691B (en) | Wind power climbing event indirect prediction method | |
CN115860177A (en) | Photovoltaic power generation power prediction method based on combined machine learning model and application thereof | |
CN112001537A (en) | Short-term wind power prediction method based on gray model and support vector machine | |
CN114429248A (en) | Transformer apparent power prediction method | |
Chi et al. | Comparison of two multi-step ahead forecasting mechanisms for wind speed based on machine learning models | |
CN113344283A (en) | Energy internet new energy consumption capacity assessment method based on edge intelligence | |
CN117293809A (en) | Multi-time space scale new energy generation power prediction method based on large model | |
CN115544895B (en) | Photovoltaic power station annual output guarantee rate model optimization method | |
Mendes et al. | Very short-term wind power forecasting: state-of-the-art | |
CN116502747A (en) | Power system standby quantification method based on probability prediction and considering uncertainty of multiple time scales of new energy | |
CN111950811B (en) | Regional photovoltaic power prediction method and system based on double-layer artificial neural network | |
CN113112085A (en) | New energy station power generation load prediction method based on BP neural network | |
CN112183814A (en) | Short-term wind speed prediction method | |
CN112036672A (en) | New energy power generation ultra-short term power prediction method and system based on iterative correction | |
Ramakrishnan et al. | Deep Feature Selection for Wind Forecasting‐I |
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 |