CN112101626A - Distributed photovoltaic power generation power prediction method and system - Google Patents
Distributed photovoltaic power generation power prediction method and system Download PDFInfo
- Publication number
- CN112101626A CN112101626A CN202010819076.2A CN202010819076A CN112101626A CN 112101626 A CN112101626 A CN 112101626A CN 202010819076 A CN202010819076 A CN 202010819076A CN 112101626 A CN112101626 A CN 112101626A
- Authority
- CN
- China
- Prior art keywords
- data
- live
- power generation
- prediction model
- photovoltaic power
- 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
- 238000010248 power generation Methods 0.000 title claims abstract description 121
- 238000000034 method Methods 0.000 title claims abstract description 32
- 239000011159 matrix material Substances 0.000 claims description 51
- 238000012549 training Methods 0.000 claims description 44
- 238000004364 calculation method Methods 0.000 claims description 32
- 238000012545 processing Methods 0.000 claims description 12
- 230000004913 activation Effects 0.000 claims description 3
- 230000008569 process Effects 0.000 description 7
- 230000006870 function Effects 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 238000003915 air pollution Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 239000011707 mineral Substances 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
- 230000003287 optical effect Effects 0.000 description 1
- 238000011160 research Methods 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/067—Enterprise or organisation modelling
-
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
-
- 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
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A30/00—Adapting or protecting infrastructure or their operation
-
- 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)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- General Business, Economics & Management (AREA)
- Biomedical Technology (AREA)
- Tourism & Hospitality (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Educational Administration (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method and a system for predicting distributed photovoltaic power generation power, which relate to the technical field of photovoltaic power generation power prediction and comprise the following steps: inputting the acquired current weather forecast data into a trained live weather data prediction model to obtain future weather live data corresponding to the current weather forecast data; inputting the future weather live data into a trained distributed photovoltaic power generation power prediction model to obtain photovoltaic power generation power corresponding to the future weather live data, so as to realize ultra-short-term prediction of the photovoltaic power generation power; the generated power prediction method realizes real-time prediction of the generated power of the distributed photovoltaic power station, and guarantees effective consumption of distributed photovoltaic power generation and safe and stable operation of a grid after grid connection.
Description
Technical Field
The invention relates to the technical field of photovoltaic power generation power prediction, in particular to a distributed photovoltaic power generation power prediction method and system.
Background
In order to reduce air pollution, protect human living environment, promote the reform of energy consumption mode, gradually change the existing energy structure, develop new renewable energy resources and effectively utilize the new renewable energy resources, has become a difficult task of the attack of various countries. From the sustainable development perspective, renewable energy sources can replace traditional mineral resources to become a main energy source channel of future international society, and most countries in the world have put renewable energy sources in a crucial position for the supply of future social energy sources.
Solar photovoltaic power generation is a renewable energy utilization mode with the fastest development speed and the largest scale, is greatly popularized and healthily developed in the international society, and promotes the wide expansion and application of distributed photovoltaic. However, uncontrollable factors such as sunlight, climate and current leads to the fact that photovoltaic power generation has the characteristics of discontinuity, unpredictability, instability and the like, the full consumption of new energy is seriously hindered, and the phenomenon of large-scale light abandon is caused; meanwhile, due to the instability of photovoltaic power generation, great challenges are brought to the stable operation of a power grid after photovoltaic grid connection.
In order to reduce the light abandoning rate of photovoltaic power generation, guarantee the stable operation of a power grid after distributed photovoltaic power generation is connected into a network, and develop researches such as the prediction of the power generation power of a distributed photovoltaic power supply and the like, the method is of great importance.
Disclosure of Invention
Based on the technical problems in the background art, the invention provides a method and a system for predicting the power of distributed photovoltaic power generation, which realize the real-time prediction of the power of a distributed photovoltaic power station and ensure the effective consumption of the distributed photovoltaic power generation and the safe and stable operation of a grid after grid connection.
The invention provides a method for predicting distributed photovoltaic power generation power, which comprises the following steps:
inputting the acquired current weather forecast data into a trained live weather data prediction model to obtain future weather live data corresponding to the current weather forecast data;
and inputting the future weather live data into the trained distributed photovoltaic power generation power prediction model to obtain the photovoltaic power generation power corresponding to the future weather live data, so as to realize the ultra-short-term prediction of the photovoltaic power generation power.
Further, in inputting the acquired current weather forecast data into the trained live weather data prediction model to obtain future weather live data corresponding to the current weather forecast data, training the constructed live weather data prediction model, including:
acquiring historical weather forecast data and historical weather live data, wherein the time of the historical weather forecast data corresponds to that of the historical weather live data, and the historical weather live data is acquired by a micro weather sensor;
training the model by taking historical meteorological forecast data as input of a live meteorological data prediction model and taking historical meteorological live data as output of the live meteorological data prediction model;
and acquiring current weather forecast data, and inputting the current weather forecast data into the trained live weather data prediction model to obtain future weather live data corresponding to the current weather forecast data.
Further, the training step of the constructed live meteorological data prediction model is as follows: the method for training the model by taking historical meteorological forecast data as input of a live meteorological data prediction model and historical meteorological live data as output of the live meteorological data prediction model comprises the following steps:
inputting a set of historical meteorological forecast data into a live meteorological data prediction model;
obtaining intermediate parameters f in a live meteorological data prediction modeltThe weight matrix and the offset of (2), the intermediate parameter itThe weight matrix and the offset of (1), the intermediate parameter otThe weight matrix and the offset of (2);
respectively calculating the output h of the t-th layer according to the weight matrix and the offsettIntermediate parameter f oft、it、ot；
Computing
Representing the computation of t-layer output CtAn intermediate parameter of (d);
according to an intermediate parameter ftIntermediate parameter itIntermediate parameters
Computing the intermediate output C of the t layert；
According to the intermediate parameter otIntermediate output CtCalculating t layer output ht；
The steps are circulated to obtain the total output of the live meteorological data prediction model;
and comparing the total output with the historical meteorological live data to correct the weight matrix and the offset so as to realize the training of the prediction model of the live meteorological data.
Further, in the step of inputting the future live meteorological data into the trained distributed photovoltaic power generation power prediction model to obtain the photovoltaic power generation power corresponding to the future live meteorological data and realize the ultra-short-term prediction of the photovoltaic power generation power, the mode training step includes:
acquiring historical meteorological actual data and photovoltaic actual power generation data;
and training the model by taking historical meteorological live data as input of the distributed photovoltaic power generation power prediction model and taking actual photovoltaic power generation data as output of the distributed photovoltaic power generation power prediction model.
Further, in training the model by taking historical meteorological live data as input of the distributed photovoltaic power generation power prediction model and taking actual photovoltaic power generation data as output of the distributed photovoltaic power generation power prediction model, the method comprises the following steps:
inputting a group of historical meteorological live data into a distributed photovoltaic power generation power prediction model;
obtaining a weight matrix w of the l layer in a distributed photovoltaic power generation power prediction modellAnd offset blAnd obtaining the output a of layer l-1l-1Wherein the input a of the first layer0A final output h for the live meteorological data prediction model;
according to weight matrix wlOffset blAn output amount al-1Computing the l-th layer output al；
The steps are circulated to obtain the total output of the distributed photovoltaic power generation power prediction model;
and comparing the total output with the actual photovoltaic power generation data to correct the weight matrix and the offset in the distributed photovoltaic power generation power prediction model so as to realize the training of the distributed photovoltaic power generation power prediction model.
Further, the live meteorological data prediction model training formula is as follows:
ft＝σ(wf·[ht-1,xt]+bf) (1)
it＝σ(wi·[ht-1,xt]+bi) (2)
ot＝σ(wo·[ht-1,xt]+bo) (3)
ht＝ot*tanh(Ct) (6)
wherein f istRepresents the computation of t-level output htFirst intermediate parameter of (d), wfRepresentation calculation ftWeight matrix of ht-1Represents the t-1 layer output, xtRepresenting the input of the t layer, bfRepresentation calculation ftOffset of itRepresents the computation of t-level output htSecond intermediate parameter of (d), wiRepresents the calculation itThe weight matrix of (a) is determined,birepresents the calculation itOffset of otRepresents the computation of t-level output htA third intermediate parameter of, woRepresents the calculation otWeight matrix of boRepresents the calculation otOffset of [ h ]t-1,xt]A matrix representing the output of layer t-1 with the input of layer t, σ () representing the activation function sigmoid function,
representing the computation of t-layer output CtIntermediate parameter of (d), wcRepresentation calculation
Weight matrix of bcRepresentation calculation
Of (d), tanh () represents a hyperbolic tangent function, htTo representtOutput of the layer, CtIntermediate output representing the calculation of t layers, Ct-1Represents the calculation of the intermediate output of the t-1 layer, htDenotes the output of the computation t layers, which denotes the product.
Further, a training formula corresponding to the distributed photovoltaic power generation power prediction model is as follows:
al＝σ(wlal-1+bl)
wherein, alDenotes the output of the l-th layer, wlWeight matrix representing the l-th layer, al-1Represents the output of layer l-1, blIndicating the offset of the l-th layer, input a of the first layer0H for final output of live meteorological data prediction modelt。
A distributed photovoltaic power generation power prediction system comprises a first processing module and a second processing module;
the first processing module is used for inputting the acquired current weather forecast data into the trained live weather data prediction model to obtain future weather live data corresponding to the current weather forecast data;
and the second processing module is used for inputting the future weather live data into the trained distributed photovoltaic power generation power prediction model to obtain photovoltaic power generation power corresponding to the future weather live data, so that the ultra-short-term prediction of the photovoltaic power generation power is realized.
The distributed photovoltaic power generation power prediction method and the system provided by the invention have the advantages that: the distributed photovoltaic power generation power prediction method and the system provided by the structure of the invention can realize centralized management and monitoring of large-scale and distributed photovoltaic stations; a user can provide an auxiliary decision according to the generated power information, the generated power prediction information and the meteorological data parameters of the display end for observing the distributed photovoltaic station in real time; on one hand, the new energy consumption and the yield level of the photovoltaic power station are improved; on the other hand, the accuracy of the photovoltaic power generation power prediction result is improved, reliable support can be provided for implementation of scheduling strategies and coordination control of a power grid scheduling department, and the influence of large-scale photovoltaic power station grid connection on stable operation of a power system is reduced.
Drawings
FIG. 1 is a schematic structural diagram of the present invention.
Detailed Description
The present invention is described in detail below with reference to specific embodiments, and in the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein, but rather should be construed as broadly as the present invention is capable of modification in various respects, all without departing from the spirit and scope of the present invention.
As shown in fig. 1, the distributed photovoltaic power generation power prediction method provided by the present invention includes the following steps S1 to S2:
s1: inputting the acquired current weather forecast data into a trained live weather data prediction model to obtain future weather live data corresponding to the current weather forecast data; the current weather forecast data can be directly obtained through weather forecast, and simultaneously, weather data at a certain time in the future can be obtained, the future weather live data is data corresponding to the current weather forecast data at a certain time in the future, and time correspondence exists between the data.
S2: and inputting the future weather live data into a trained distributed photovoltaic power generation power prediction model to obtain photovoltaic power generation power corresponding to the future weather live data, so as to realize ultra-short-term prediction of the photovoltaic power generation power.
In the present application, meteorological data parameters (current meteorological forecast data, future meteorological live data, historical meteorological forecast data, historical meteorological live data) include temperature, irradiance, wind speed, wind direction, weather conditions, rainfall, humidity, cloud cover, and the like.
Further, at step S1: inputting the acquired current weather forecast data into a trained live weather data prediction model to obtain future weather live data corresponding to the current weather forecast data, wherein the training process of the constructed live weather data prediction model comprises the following steps of S101 to S102:
s101: acquiring historical weather forecast data and historical weather live data, wherein the time of the historical weather forecast data corresponds to that of the historical weather live data, and the historical weather live data is acquired by a micro weather sensor;
historical weather forecast data can be directly obtained through weather forecast, the historical weather live data is data corresponding to the historical weather forecast data at a certain future moment, and time correspondence exists between the historical weather live data and the historical weather forecast data; the historical weather forecast data and the historical weather live data are both data before the corresponding moment of the future weather live data.
For example: collection of historical weather forecast data w'lAnd live data wlAccording to the time correlation, the correlation between the future weather live data at the time t and the historical weather forecast data and the historical weather live data at the first five times t-1, t-2, t-3, t-4 and t-5 is larger. Therefore, it is practical to collect data 5 moments before (but not limited to 5 moments) the future weather-live data, and output the data as the main parameter of the future weather-live data at the time t.
S102: training the model by taking historical meteorological forecast data as input of a live meteorological data prediction model and taking historical meteorological live data as output of the live meteorological data prediction model;
the model training at step S102 includes steps S112 to S182:
s112: inputting a set of historical meteorological forecast data into a live meteorological data prediction model;
in the training process of the live meteorological data prediction model, a plurality of groups of historical meteorological forecast data and corresponding historical meteorological live data are obtained, and the live meteorological data prediction model is trained one group by one group. The more data sets (historical weather forecast data and historical weather live data) the better theoretically, and the operator can select the number of the data sets as required in the actual training process.
S122: obtaining intermediate parameters f in a live meteorological data prediction modeltThe weight matrix and the offset of (2), the intermediate parameter itThe weight matrix and the offset of (1), the intermediate parameter otThe weight matrix and the offset of (2);
s132: respectively calculating the output h of the t-th layer according to the weight matrix and the offsettIntermediate parameter f oft、it、ot；
S142: computing
Representing the computation of t-layer output CtAn intermediate parameter of (d);
s152: according to an intermediate parameter ftIntermediate parameter itIntermediate parameters
Computing the intermediate output C of the t layert；
S162: according to the intermediate parameter otIntermediate output CtCalculating t layer output ht；
S172: looping steps S112 to S162 to obtain the total output of the live meteorological data prediction model;
according to steps S112 to S162, the input historical weather forecast data is sequentially processed through different layers in the live weather data prediction model, the output of each layer is used as the input of the next layer, and the total output of the historical weather forecast data after passing through the live weather data prediction model is finally realized, the total output is calculated through a formula in the live weather data prediction model, and may have a certain difference with the corresponding actual historical weather live data, so the process proceeds to step S182.
S182: and comparing the total output with the historical meteorological live data to correct the weight matrix and the offset so as to realize the training of the prediction model of the live meteorological data.
The training formula corresponding to the live meteorological data prediction model is as follows:
ft＝σ(wf·[ht-1,xt]+bf) (1)
it＝σ(wi·[ht-1,xt]+bi) (2)
ot＝σ(wo·[ht-1,xt]+bo) (3)
ht＝ot*tanh(Ct) (6)
wherein f istRepresents the computation of t-level output htFirst intermediate parameter of (d), wfRepresentation calculation ftWeight matrix of ht-1Represents the t-1 layer output, xtRepresenting the input of the t layer, bfRepresentation calculation ftOffset of itRepresents the computation of t-level output htSecond intermediate parameter of (d), wiRepresents the calculation itWeight matrix of biRepresents the calculation itOf (2)Dosage otRepresents the computation of t-level output htA third intermediate parameter of, woRepresents the calculation otWeight matrix of boRepresents the calculation otOffset of [ h ]t-1,xt]A matrix representing the output of layer t-1 with the input of layer t, σ () representing the activation function sigmoid function,
representing the computation of t-layer output CtIntermediate parameter of (d), wcRepresentation calculation
Weight matrix of bcRepresentation calculation
Of (d), tanh () represents a hyperbolic tangent function, htTo representtOutput of the layer, CtIntermediate output representing the calculation of t layers, Ct-1Represents the calculation of the intermediate output of the t-1 layer, htDenotes the output of the computation t layers, which denotes the product.
The method comprises the following steps: the weight matrix and the offset in the parameters are automatically adjusted and determined by model training and are not given artificially. The weight matrix and the offset correspond to all the weight matrices and the offsets in the parameters.
Specifically, in the process of comparing the total output with the historical weather live data, the weight matrix and the offset are dynamically adjusted, the next group of historical weather forecast data is calculated through the adjusted weight matrix and the adjusted offset, the calculation result is compared with the corresponding historical weather live data, the weight matrix and the offset are dynamically adjusted again until the difference between the total output and the historical weather live data is within an allowable range, and the training of a prediction model of the live weather data is completed.
According to the steps S101 to S102, when a real-time meteorological data prediction model is actually trained, firstly, a big data sample set comprising historical meteorological forecast data and historical meteorological real-time data is obtained, wherein the big data sample set comprises a sample set A and a sample set B, the sample set A is directly input and output training is carried out on the real-time meteorological data prediction model according to the principle that the historical meteorological forecast data is used as the input of the real-time meteorological data prediction model, and the historical meteorological real-time data is used as the output of the real-time meteorological data prediction model;
after the training is completed, N (e.g., 6) data are used as a group for the B sample set, and each group of data corresponds to N times respectively: t-N +1, t-N +2,. cndot.t. And taking the historical meteorological forecast data at the time of t-N +1, t-N +2, and t-1 as input, taking the historical meteorological actual data as output, transmitting the historical meteorological actual data into the actual meteorological data prediction model, simultaneously, taking the meteorological forecast data at the time of t as input, transmitting the input into the actual meteorological data prediction model, obtaining the meteorological actual data which are output by the actual meteorological data prediction model and correspond to the meteorological forecast data at the time of t, simultaneously, comparing the meteorological actual meteorological data with the meteorological actual meteorological data at the time of t of a sample set week worker to verify the training result of the actual meteorological data prediction model, and when the verification fails, continuing to enrich the sample set A to train the actual meteorological data prediction model until the actual meteorological data prediction model passes the verification.
Further, at step S2: inputting the future meteorological actual data into a trained distributed photovoltaic power generation power prediction model to obtain photovoltaic power generation power corresponding to the future meteorological actual data, and realizing ultra-short term prediction of the photovoltaic power generation power, wherein the training process of the constructed distributed photovoltaic power generation power prediction model comprises the following steps:
s201: acquiring historical meteorological actual data and photovoltaic actual power generation data;
the historical meteorological actual data and the photovoltaic actual power generation data are a group of data acquired at the same time, and multiple groups of data can be acquired at the same time to form a training sample set for training a distributed photovoltaic power generation power prediction model.
S202: and training the model by taking historical meteorological live data as input of the distributed photovoltaic power generation power prediction model and taking actual photovoltaic power generation data as output of the distributed photovoltaic power generation power prediction model.
The model training of step S202 includes the following steps S212 to S252:
s212: inputting a group of historical meteorological live data into a distributed photovoltaic power generation power prediction model;
the historical meteorological actual data are directly obtained through a microclimate sensor, and different quantities of historical meteorological actual data are obtained at different times to form a sample set used for training a distributed photovoltaic power generation power prediction model.
S222: obtaining a weight matrix w of the l layer in a distributed photovoltaic power generation power prediction modellAnd offset blAnd obtaining the output a of layer l-1l-1Wherein the input a of the first layer0A final output h for the live meteorological data prediction model;
s232: according to weight matrix wlOffset blAn output amount al-1Computing the l-th layer output al；
S242: the steps 212 to S232 are circulated, and the total output of the distributed photovoltaic power generation power prediction model is obtained;
according to steps 212 to S232, the input historical weather-scene data is sequentially processed through different layers in the distributed photovoltaic power generation power prediction model, the output of each layer is used as the input of the next layer, and finally the total output of the historical weather-scene data after passing through the distributed photovoltaic power generation power prediction model is realized, the total output is obtained through calculation by a formula in the distributed photovoltaic power generation power prediction model, and a certain difference may exist between the total output and the corresponding actual photovoltaic power generation data, so the step S252 is performed.
S252: and comparing the total output with the actual photovoltaic power generation data to correct the weight matrix and the offset in the distributed photovoltaic power generation power prediction model so as to realize the training of the distributed photovoltaic power generation power prediction model.
A training formula corresponding to the distributed photovoltaic power generation power prediction model is as follows:
al＝σ(wlal-1+bl)
wherein, alDenotes the output of the l-th layer, wlWeight matrix representing the l-th layer, al-1Represents the output of layer l-1, blIndicating the offset of the l-th layer, input a of the first layer0H for final output of live meteorological data prediction modelt. The weight matrix and the offset in the parameters are automatically adjusted and determined by model training and are not given artificially.
Specifically, in the process of comparing the total output with the photovoltaic actual power generation data, the weight matrix w is dynamically adjustedlAnd offset blAnd passes through the adjusted weight matrix wlAnd offset blCalculating the next group of historical meteorological actual data, comparing the calculation result with the corresponding photovoltaic actual power generation data, and dynamically adjusting the weight matrix w againlAnd offset blAnd completing the training of the distributed photovoltaic power generation power prediction model until the difference between the total output and the photovoltaic actual power generation data is within an allowable range.
The distributed photovoltaic power generation power prediction model is trained, so that future weather live data can be input into the distributed photovoltaic power generation power prediction model to obtain corresponding photovoltaic power generation power, and since the future weather live data is data corresponding to a certain time after the current time, the photovoltaic power generation power output according to the data is also predicted output power at the future time, and therefore ultra-short-term prediction of the photovoltaic power generation power can be achieved.
The live meteorological data prediction model is constructed based on an LSTM neural network (long-short time memory network), the distributed photovoltaic power generation power prediction model is constructed based on a BP neural network, and the BP neural network is a multilayer feedforward neural network trained according to an error back propagation algorithm.
As a specific embodiment, a method for predicting distributed photovoltaic power generation power includes:
s100: acquiring weather forecast data;
s200: and step S122 to step S172 are carried out by taking the weather forecast data as input, so as to obtain future weather live data.
At this time, the live meteorological data prediction models corresponding to steps S122 to S172 are already trained models.
S300: and (5) taking future weather live data as input, and entering steps S122 to S172 to obtain photovoltaic power generation power so as to realize prediction of the photovoltaic power generation power.
At this time, the distributed photovoltaic power generation power prediction model corresponding to steps S122 to S172 is a model that has been trained.
A distributed photovoltaic power generation power prediction system comprises a first processing module and a second processing module;
the first processing module is used for inputting the acquired current weather forecast data into the trained live weather data prediction model to obtain future weather live data corresponding to the current weather forecast data;
and the second processing module is used for inputting the future weather live data into the trained distributed photovoltaic power generation power prediction model to obtain photovoltaic power generation power corresponding to the future weather live data, so that the ultra-short-term prediction of the photovoltaic power generation power is realized.
Those of ordinary skill in the art will understand that: all or part of the steps for implementing the method embodiments may be implemented by hardware related to program instructions, and the program may be stored in a computer readable storage medium, and when executed, the program performs the steps including the method embodiments; and the aforementioned storage medium includes: various media that can store program codes, such as ROM, RAM, magnetic or optical disks.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention should be equivalent or changed within the scope of the present invention.
Claims (8)
1. A distributed photovoltaic power generation power prediction method is characterized by comprising the following steps:
inputting the acquired current weather forecast data into a trained live weather data prediction model to obtain future weather live data corresponding to the current weather forecast data;
and inputting the future weather live data into the trained distributed photovoltaic power generation power prediction model to obtain the photovoltaic power generation power corresponding to the future weather live data, so as to realize the ultra-short-term prediction of the photovoltaic power generation power.
2. The method for predicting distributed photovoltaic power generation power according to claim 1, wherein the training of the constructed live meteorological data prediction model in the case that the obtained current meteorological forecast data is input into the trained live meteorological data prediction model to obtain the future meteorological live data corresponding to the current meteorological forecast data comprises:
acquiring historical weather forecast data and historical weather live data, wherein the time of the historical weather forecast data corresponds to that of the historical weather live data, and the historical weather live data is acquired by a micro weather sensor;
training the model by taking historical meteorological forecast data as input of a live meteorological data prediction model and taking historical meteorological live data as output of the live meteorological data prediction model;
and acquiring current weather forecast data, and inputting the current weather forecast data into the trained live weather data prediction model to obtain future weather live data corresponding to the current weather forecast data.
3. The distributed photovoltaic power generation power prediction method of claim 1, wherein the training of the constructed live meteorological data prediction model comprises the following steps: the method for training the model by taking historical meteorological forecast data as input of a live meteorological data prediction model and historical meteorological live data as output of the live meteorological data prediction model comprises the following steps:
inputting a set of historical meteorological forecast data into a live meteorological data prediction model;
obtaining intermediate parameters f in a live meteorological data prediction modeltThe weight matrix and the offset of (2), the intermediate parameter itThe weight matrix and the offset of (1), the intermediate parameter otThe weight matrix and the offset of (2);
respectively calculating the output h of the t-th layer according to the weight matrix and the offsettIntermediate parameter f oft、it、ot；
Computing
Representing the computation of t-layer output CtAn intermediate parameter of (d);
according to an intermediate parameter ftIntermediate parameter itIntermediate parameters
Computing the intermediate output C of the t layert；
According to the intermediate parameter otIntermediate output CtCalculating t layer output ht；
The steps are circulated to obtain the total output of the live meteorological data prediction model;
and comparing the total output with the historical meteorological live data to correct the weight matrix and the offset so as to realize the training of the prediction model of the live meteorological data.
4. The method for predicting the power generation of the distributed photovoltaic power as claimed in claim 3, wherein the mode training step includes the steps of, in inputting the future live meteorological data into the trained distributed photovoltaic power generation power prediction model to obtain the photovoltaic power generation power corresponding to the future meteorological live meteorological data and realize the ultra-short-term prediction of the photovoltaic power generation power:
acquiring historical meteorological actual data and photovoltaic actual power generation data;
and training the model by taking historical meteorological live data as input of the distributed photovoltaic power generation power prediction model and taking actual photovoltaic power generation data as output of the distributed photovoltaic power generation power prediction model.
5. The distributed photovoltaic power generation power prediction method according to claim 4, wherein training the model by using the historical weather-live data as an input of the distributed photovoltaic power generation power prediction model and the photovoltaic actual power generation data as an output of the distributed photovoltaic power generation power prediction model comprises:
inputting a group of historical meteorological live data into a distributed photovoltaic power generation power prediction model;
obtaining a weight matrix w of the l layer in a distributed photovoltaic power generation power prediction modellAnd offset blAnd obtaining the output a of layer l-1l-1Wherein the input a of the first layer0A final output h for the live meteorological data prediction model;
according to weight matrix wlOffset blAn output amount al-1Computing the l-th layer output al；
The steps are circulated to obtain the total output of the distributed photovoltaic power generation power prediction model;
and comparing the total output with the actual photovoltaic power generation data to correct the weight matrix and the offset in the distributed photovoltaic power generation power prediction model so as to realize the training of the distributed photovoltaic power generation power prediction model.
6. The distributed photovoltaic power generation power prediction method of claim 3, wherein the training formula of the live meteorological data prediction model is as follows:
ft＝σ(wf·[ht-1,xt]+bf) (1)
it＝σ(wi·[ht-1,xt]+bi) (2)
ot＝σ(wo·[ht-1,xt]+bo) (3)
ht＝ot*tanh(Ct) (6)
wherein f istRepresents the computation of t-level output htFirst intermediate parameter of (d), wfRepresentation calculation ftWeight matrix of ht-1Represents the t-1 layer output, xtRepresenting the input of the t layer, bfRepresentation calculation ftOffset of itRepresents the computation of t-level output htSecond intermediate parameter of (d), wiRepresents the calculation itWeight matrix of biRepresents the calculation itOffset of otRepresents the computation of t-level output htA third intermediate parameter of, woRepresents the calculation otWeight matrix of boRepresents the calculation otOffset of [ h ]t-1,xt]A matrix representing the output of layer t-1 with the input of layer t, σ () representing the activation function sigmoid function,
representing the computation of t-layer output CtIntermediate parameter of (d), wcRepresentation calculation
Weight matrix of bcRepresentation calculation
Of (d), tanh () represents a hyperbolic tangent function, htTo representtOutput of the layer, CtRepresenting the intermediate output of the t layer, Ct-1Represents the calculation of the intermediate output of the t-1 layer, htDenotes the output of the computation t layers, which denotes the product.
7. The distributed photovoltaic power generation power prediction method according to claim 5, wherein the training formula of the distributed photovoltaic power generation power prediction model is as follows:
al＝σ(wlal-1+bl)
wherein, alDenotes the output of the l-th layer, wlWeight matrix representing the l-th layer, al-1Represents the output of layer l-1, blIndicating the offset of the l-th layer, input a of the first layer0H for final output of live meteorological data prediction modelt。
8. A distributed photovoltaic power generation power prediction system is characterized by comprising a first processing module and a second processing module;
the first processing module is used for inputting the acquired current weather forecast data into the trained live weather data prediction model to obtain future weather live data corresponding to the current weather forecast data;
and the second processing module is used for inputting the future weather live data into the trained distributed photovoltaic power generation power prediction model to obtain photovoltaic power generation power corresponding to the future weather live data, so that the ultra-short-term prediction of the photovoltaic power generation power is realized.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010819076.2A CN112101626A (en) | 2020-08-14 | 2020-08-14 | Distributed photovoltaic power generation power prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010819076.2A CN112101626A (en) | 2020-08-14 | 2020-08-14 | Distributed photovoltaic power generation power prediction method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112101626A true CN112101626A (en) | 2020-12-18 |
Family
ID=73753750
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010819076.2A Pending CN112101626A (en) | 2020-08-14 | 2020-08-14 | Distributed photovoltaic power generation power prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112101626A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113095562A (en) * | 2021-04-07 | 2021-07-09 | 安徽天能清洁能源科技有限公司 | Ultra-short term power generation prediction method and device based on Kalman filtering and LSTM |
| CN113689045A (en) * | 2021-08-27 | 2021-11-23 | 山东浪潮科学研究院有限公司 | Method, device and medium for predicting grid-connected electric quantity of photovoltaic region |
| CN115438839A (en) * | 2022-08-11 | 2022-12-06 | 海宁云多科技有限公司 | Photovoltaic power ultra-short term prediction system and method based on data of intelligent electric meter |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110059878A (en) * | 2019-04-15 | 2019-07-26 | 中国计量大学 | Based on CNN LSTM photovoltaic power generation power prediction model and its construction method |
| CN110348648A (en) * | 2019-08-02 | 2019-10-18 | 国网电子商务有限公司 | A kind of predicting power of photovoltaic plant method and device |
| CN110929963A (en) * | 2019-12-16 | 2020-03-27 | 深圳智润新能源电力勘测设计院有限公司 | Wind speed prediction method, wind speed prediction device, and storage medium |
| KR20200034015A (en) * | 2018-09-11 | 2020-03-31 | 광주과학기술원 | Stepwise solar power generation forecast apparatus using machine learning and the method thereof |
| CN111008728A (en) * | 2019-11-01 | 2020-04-14 | 深圳供电局有限公司 | Method for predicting short-term output of distributed photovoltaic power generation system |
| CN111461444A (en) * | 2020-04-07 | 2020-07-28 | 上海电气风电集团股份有限公司 | Prediction method, system, medium and electronic device for unit power of wind power plant |
-
2020
- 2020-08-14 CN CN202010819076.2A patent/CN112101626A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20200034015A (en) * | 2018-09-11 | 2020-03-31 | 광주과학기술원 | Stepwise solar power generation forecast apparatus using machine learning and the method thereof |
| CN110059878A (en) * | 2019-04-15 | 2019-07-26 | 中国计量大学 | Based on CNN LSTM photovoltaic power generation power prediction model and its construction method |
| CN110348648A (en) * | 2019-08-02 | 2019-10-18 | 国网电子商务有限公司 | A kind of predicting power of photovoltaic plant method and device |
| CN111008728A (en) * | 2019-11-01 | 2020-04-14 | 深圳供电局有限公司 | Method for predicting short-term output of distributed photovoltaic power generation system |
| CN110929963A (en) * | 2019-12-16 | 2020-03-27 | 深圳智润新能源电力勘测设计院有限公司 | Wind speed prediction method, wind speed prediction device, and storage medium |
| CN111461444A (en) * | 2020-04-07 | 2020-07-28 | 上海电气风电集团股份有限公司 | Prediction method, system, medium and electronic device for unit power of wind power plant |
Non-Patent Citations (1)
| Title |
|---|
| 张向荣 等: "《人工智能前沿技术丛书 模式识别》", 30 September 2019 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113095562A (en) * | 2021-04-07 | 2021-07-09 | 安徽天能清洁能源科技有限公司 | Ultra-short term power generation prediction method and device based on Kalman filtering and LSTM |
| CN113689045A (en) * | 2021-08-27 | 2021-11-23 | 山东浪潮科学研究院有限公司 | Method, device and medium for predicting grid-connected electric quantity of photovoltaic region |
| CN113689045B (en) * | 2021-08-27 | 2023-06-27 | 山东浪潮科学研究院有限公司 | Photovoltaic area grid-connected electric quantity prediction method, device and medium |
| CN115438839A (en) * | 2022-08-11 | 2022-12-06 | 海宁云多科技有限公司 | Photovoltaic power ultra-short term prediction system and method based on data of intelligent electric meter |
| CN115438839B (en) * | 2022-08-11 | 2023-08-18 | 海宁云多科技有限公司 | Photovoltaic power ultra-short-term prediction system and method based on intelligent ammeter data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ghaderi et al. | Deep forecast: Deep learning-based spatio-temporal forecasting | |
| Mahmoud et al. | An advanced approach for optimal wind power generation prediction intervals by using self-adaptive evolutionary extreme learning machine | |
| CN112614009B (en) | Power grid energy management method and system based on deep expectation Q-learning | |
| CN112101626A (en) | Distributed photovoltaic power generation power prediction method and system | |
| CN103489038A (en) | Photovoltaic ultra-short-term power prediction method based on LM-BP neural network | |
| Li et al. | Probabilistic charging power forecast of EVCS: Reinforcement learning assisted deep learning approach | |
| Mellit | ANN-based GA for generating the sizing curve of stand-alone photovoltaic systems | |
| Ashraf et al. | Artificial neural network based models for forecasting electricity generation of grid connected solar PV power plant | |
| CN104376389A (en) | Master-slave type micro-grid power load prediction system and master-slave type micro-grid power load prediction method based on load balancing | |
| CN112215428B (en) | Photovoltaic power generation power prediction method and system based on error correction and fuzzy logic | |
| Mukhoty et al. | Sequence to sequence deep learning models for solar irradiation forecasting | |
| CN102510059A (en) | Super short-term wind power forecasting method based on back propagation (BP) neural network | |
| CN116167531A (en) | Photovoltaic power generation prediction method based on digital twin | |
| Shah et al. | A novel prediction error-based power forecasting scheme for real pv system using pvusa model: A grey box-based neural network approach | |
| CN112819189A (en) | Wind power output prediction method based on historical predicted value | |
| Teferra et al. | Fuzzy-based prediction of solar PV and wind power generation for microgrid modeling using particle swarm optimization | |
| CN115423153A (en) | Photovoltaic energy storage system energy management method based on probability prediction | |
| CN111815051B (en) | GRNN photovoltaic power generation prediction method considering weather influence factors | |
| CN112116171A (en) | Novel photovoltaic power generation power prediction method based on neural network | |
| Sarathkumar et al. | Renewable energy resources forecasting model for virtual power plant in the deregulated electricity market using machine learning | |
| CN113283638A (en) | Load extreme curve prediction method and system based on fusion model | |
| CN111915084A (en) | Hybrid photovoltaic power generation power prediction method and system based on neural network | |
| CN117498400B (en) | Distributed photovoltaic and energy storage data processing method and system | |
| CN116227743B (en) | Photovoltaic power generation power abnormal rate determining method and system based on tuna swarm algorithm | |
| Sua et al. | Predicting Power Output of Solar Panels Using Machine Learning Algorithms |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201218 |
|
| RJ01 | Rejection of invention patent application after publication |