CN111242359B - Solar radiation online dynamic prediction method based on data drift - Google Patents
Solar radiation online dynamic prediction method based on data drift Download PDFInfo
- Publication number
- CN111242359B CN111242359B CN202010011980.0A CN202010011980A CN111242359B CN 111242359 B CN111242359 B CN 111242359B CN 202010011980 A CN202010011980 A CN 202010011980A CN 111242359 B CN111242359 B CN 111242359B
- Authority
- CN
- China
- Prior art keywords
- samples
- data
- data set
- prediction
- solar radiation
- 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
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
- 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
Abstract
The invention discloses a solar radiation online dynamic prediction method based on data drift, and belongs to the technical field of photovoltaic prediction. Dividing the L measured samples into two sections, calculating the two sectionsAverage value I (u)1) And I (u)2)；|I(u1)‑I(u2)|＞TdBy using MBPredicting if I (t) is data drift occurring continuously, storing X' (t) in DCIn, if DCNumber of data up to an upper limit, DCAlternative DAAnd training update MA(ii) a If data drift occurs for the first time during I (t), clearing DCAnd adding or replacing X' (t) to DBIn, and train update MB；|I(u1)‑I(u2)|≤TdBy using MAPredicting, adding or replacing X' (t) to DAIn, training update MA(ii) a And outputting the result to enter the next cycle. The invention realizes the on-line dynamic prediction and model update of solar radiation.
Description
Technical Field
The invention belongs to the technical field of photovoltaic prediction, and particularly relates to a solar radiation online dynamic prediction method based on data drift.
Background
Renewable energy sources, particularly wind and solar energy, are receiving increasing attention from people due to environmental pollution and climate change caused by the use of fossil energy. With the development of technology and the support of national policies, the photovoltaic industry develops rapidly. In the design of a photovoltaic system, solar radiation online prediction is a premise of power generation grid connection of a photovoltaic power station. Therefore, there is an urgent need to develop an online prediction technology of solar radiation.
For the articles disclosed by the online dynamic prediction of solar radiation, the prediction model is designed mainly in an offline manner and is debugged essentially at present, and the prediction model is used online, but a general dynamic prediction method capable of performing online training and updating on the prediction model according to the change of the site condition is not explicitly provided.
In addition, in the patent disclosure, no patent is disclosed for online dynamic prediction of solar radiation at present, and only CN 106815659A patent mentions an ultra-short-term solar radiation prediction method and device based on a hybrid model, but the patent only provides processing of solar radiation data and a corresponding prediction model, and cannot realize online update of the prediction model, so that the application range and precision of the method are limited to a certain extent.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a solar radiation on-line dynamic prediction method based on data drift, which can meet the demand of output prediction of a distributed or independent solar system, fully considers the difference of long-term and short-term regular changes of solar radiation data and the data attenuation characteristics, and realizes on-line dynamic prediction and model update of solar radiation by switching a prediction model on line by using a data drift technology.
In order to solve the technical problems, the invention adopts the technical scheme that:
a solar radiation online dynamic prediction method based on data drift comprises the following steps:
(1) measuring time period TiAverage value of internal solar radiation value I (T) and average value of corresponding meteorological parameter X (T), and TiFront continuously collected L-1 solar radiation values I (t)1)、I(t2)、I(t3)……I(tL-1) And corresponding meteorological parameter value X (t)1)、X(t2)、X(t3)……X(tL-1) (ii) a To form L data samples, which are: i (t)1) And X (t)1) Make up sample X' (t)1)，I(t2) And X (t)2) Make up sample X' (t)2)，I(t3) And X (t)3) Make up sample X' (t)3)，……，I(tL-1) And X (t)L-1) Make up sample X' (t)L-1) I (t) and X (t) form a sample X' (t), dividing L data samples into n1、n2Two sections;
(2) calculating n1、n2Average value of solar radiation values I (u) in both segments1) And I (u)2) (ii) a When I (u)1)-I(u2) | > threshold TdIn time, radiation data drift occurs, and an abnormal prediction model M is adoptedBPredicting otherwise, using conventional prediction model MACarrying out prediction;
wherein the abnormality prediction model MBThe prediction is as follows:
if data drift occurs for the first time during the period I (t), emptying the temporary data set DCJudging the abnormal data set DBWhether the number of samples reaches the upper limit or not, and if not, adding the sample X' (t) to the abnormal data set DBIn the middle, training and updating the abnormal prediction model MBComputing an anomaly data set D when it is yesBThe difference S between all the data samplesB(t), and X' (t) with an anomaly data set DBThe degree of difference S (t) between all the data samples in (1), when SBWhen the minimum value in (t) is less than the minimum value in S (t), X' (t) replaces the abnormal data set DBMiddle SB(t) the sample data with smaller difference degree with other samples in the two corresponding samples are according to the new abnormal data set DBTraining update anomaly prediction model MB；
Conventional predictive model MAThe prediction is as follows:
if I (t) is a data drift occurring continuously within L samples, then X' (t) is stored to the temporary data set DCUp to a temporary data set DCThe number of the middle samples reaches the set number, and the data set D is processedCReplacing the regular data set D with the middle sampleAAll samples in (1); if no data drift occurs during I (t), judging the conventional data set DAIf the number of middle samples reaches the upper limit, if not, X' (t) is directly added to the conventional data set DAIn updating the regular data set DA(ii) a Is time X' (t) replaced to the regular data set DAThe earliest data value is collected, and the regular data set D is updatedA(ii) a From the new conventional data set DATraining update routine predictive model MA；
(3) Outputting the prediction result, returning to the step (1), and entering the next time period Ti+1Solar radiation data prediction and model updating.
According to the solar radiation on-line dynamic prediction method based on data drift, a solar radiation sample with the continuous length L is divided into n1、n2Two segments, for non-overlapping division, i.e. n1+n2L; there may also be an overlapping division, i.e. n1+n2＞L。
The sun based on data driftRadiation on-line dynamic prediction method, the L-1 samples X' (t)1)、X(t2)、X′(t3)……X′(tL-1) The number of the collected L-1 samples was determined.
The solar radiation on-line dynamic prediction method based on the data drift comprises L-1 samples X' (t)1)、X(t2)、X′(t3)……X′(tL-1) For a continuous period of time TL-1Samples collected during the process.
According to the solar radiation online dynamic prediction method based on data drift, the difference degree between the samples is defined as the Euclidean distance between two multidimensional samples.
According to the solar radiation online dynamic prediction method based on data drift, the meteorological parameters are partial or all meteorological variables of temperature, humidity, wind speed, wind direction, aerosol or cloud amount corresponding to a photovoltaic power station or a radiation data station.
Has the advantages that: compared with the prior art, the invention has the advantages that:
according to the method, the data samples are divided into conventional data and abnormal data according to the characteristic change between the current solar radiation and the solar radiation data in a previous period of time, a conventional prediction model and an abnormal prediction model are constructed, and different prediction models are selected by combining a data drift technology, so that the on-line dynamic prediction and model update of the solar radiation are realized, and the prediction precision is improved.
Drawings
FIG. 1 is a flow chart illustrating the flow of executing prediction and outputting results of the online dynamic solar radiation prediction method based on data drift according to the present invention;
FIG. 2 is a conventional data set D in the solar radiation online dynamic prediction method based on data driftAAnd an abnormal data set DBAnd a conventional prediction model MAAnd an anomaly prediction model MBAnd updating the flow schematic diagram on line dynamically.
Detailed Description
The following is a preferred embodiment of the present invention, and does not limit the scope of the present invention.
Example 1
A solar radiation online dynamic prediction method based on data drift executes main process flows as shown in figures 1 and 2, and the specific processes are as follows:
measuring time period TiAverage value of solar radiation value I (T) and average value of corresponding meteorological parameter X (T), and TiFirst continuously collected L-1 solar radiation values I (t)1)、I(t2)、I(t3)……I(tL-1) And corresponding meteorological parameter value X (t)1)、X(t2)、X(t3)……X(tL-1) Forming L solar radiation samples by the same method, which respectively comprises the following steps: i (t)1) And X (t)1) Make up sample X' (t)1)，I(t2) And X (t)2) Make up sample X' (t)2)，I(t3) And X (t)3) Make up sample X' (t)3)，……，I(tL-1) And X (t)L-1) Make up sample X' (t)L-1) I (t) and X (t) form a sample X' (t), L solar radiation samples are obtained, and the L solar radiation samples are divided into n1、n2Two sections; where the data length L can be determined according to a prediction scale, such as when making ultra-short term predictions (one prediction point every 15 minutes, predicting 4 hours into the future radiation value), L is preferably set to 16; when short-term predictions are made (one prediction point per hour, predicting the amount of radiation in the future day), L is preferably set to 24; when making a medium-long term prediction (predicting the radiation dose one week, one month, or one year in the future at a prediction point of 3 hours, daily, or monthly), L is preferably set to a constant of 50 or more.
Take the prediction scale as ultra-short term prediction (one prediction point every 15 minutes, predicting the radiation value 4 hours in the future) as an example. During prediction, detecting the average value of the direct solar radiation value within 15 minutes at present, recording the average value as I (t), and recording the average value of the corresponding meteorological parameters as X (t); dividing the previous 15 samples and the current measured data into 16 samples without overlapping110 and n2Two segments 6 and the average I (u) of the samples of these two segments is calculated1) And I (u)2)；
When I (u)1)-I(u2) | threshold > valueTdUsing an anomaly prediction model MBMaking a prediction, otherwise, adopting a conventional prediction model MAAnd (5) predicting to obtain a prediction result of the solar radiation all day, and completing the prediction of the direct solar radiation 15 minutes ahead. Here the threshold value TdThe threshold value T can be set manually according to experience or according to a certain proportion, and is preferably selected in the casedIs set as I (u)1) And I (u)2) A medium or small half value.
When I (u)1)-I(u2)|≤TdAnd a conventional data set DAThe number of the middle samples is less than the set upper limit NAThe current measured value X' (t) is then added directly to the conventional data set DATraining and updating the conventional prediction model MA；
When I (u)1)-I(u2)|≤TdAnd a conventional data set DAThe number of the middle samples is more than or equal to a set upper limit NAWhen the current measured value X' (t) is replaced to the regular data set DAThe earliest data sample is collected and the conventional prediction model M is trained and updatedA；
When I (u)1)-I(u2)|＞TdAnd when continuous data drift occurs within L samples, the current measurement value X' (t) is stored in the temporary data set DCPerforming the following steps;
when the temporary data set DCThe number of the middle samples reaches the set number NCTemporal temporary data set DCSubstitution of the medium sample into the regular data set DAAll samples in and based on the new conventional data set DATraining update routine predictive model MA；
When I (u)1)-I(u2)|＞TdFirst time data drift occurs in L samples, and temporary data set D is emptiedC(ii) a If abnormal data set DBThe number of the middle samples does not reach the set upper limit NBThe current measured value X' (t) is then added directly to the abnormal data set DBForm a new anomaly data set DBAnd based on the new anomaly data set DBTraining update anomaly prediction model MB(ii) a If abnormal data set DBThe number of the middle samples reaches a set upper limit NBThen, the current measurement value X' (t) and the abnormal data set D are calculatedBThe degree of difference between the samples in (1), denoted as S (t); computing an anomalous data set DBThe difference between two samples is denoted as SB(t); if the minimum value in S (t) is greater than SB(t) minimum value, then the current measurement X' (t) replaces the outlier data set DBThe data sample with the minimum difference degree and the sum of the difference degrees of the two data with the minimum difference degree and other data are trained and updated to obtain the abnormal prediction model MB(ii) a Two data X' (t)m) And X' (t)n) The degree of difference is defined as
Wherein d is the number of elements in X'.
And after the next moment, namely the average value of the direct solar radiation in the future 15 minutes is predicted, updating the corresponding database and the prediction model on line, repeating the process, and predicting the next moment.
Conventional prediction model M in the present inventionAAnomaly prediction model MBThe existing model can be selected according to the requirement, and can also be constructed by self; in the present embodiment, the conventional prediction model MAPreference for linear model, anomaly prediction model MBA non-linear model is preferred.
Example 2
By adopting the method of example 1, the solar direct radiation data experiment in 2013 years in the national new energy laboratory database (NREL) in the united states verifies the prediction performance of the solar radiation online dynamic prediction method based on data drift, and the results are shown in table 1 compared with the existing method.
TABLE 1 Performance statistics of 15 minute Advance prediction results for different prediction methods
| Method of producing a composite material | Relative absolute mean error (%) | Relative root mean square error (%) |
| Continuous model | 20.94 | 31.10 |
| AR Linear model | 19.07 | 30.12 |
| ANN nonlinear model | 18.21 | 29.23 |
| Method of the invention | 16.47 | 26.67 |
Note: and (3) continuous model: marquez R, Coimbra C F m, intra-road DNI for acquiring a packetized image analysis [ J ]. Solar Energy, 2013, 91: 327-336.
The AR linear model: dimitris N.Politis.model-Based Prediction in Autotoregesion [ M ]// Model-Free Prediction and regression.2015.
ANN nonlinear model: weihaikun, theory and method of neural network structure design [ M ].2005.
By analyzing the table 1, it can be seen that the prediction method provided by the invention is relatively improved by 8.7% -14% in terms of relative root mean square error index compared with other prediction models.
The above-described online dynamic solar radiation prediction method based on data drift takes predicting 15 minutes of direct solar irradiance in advance as an example, and can also predict any solar radiation data in different time scales of 1 minute, 1 hour, one day, even one week and the like.
The above description is one of the present invention, and is not limited thereto, and other embodiments of the present invention are within the scope of the present invention.
Claims (6)
1. A solar radiation online dynamic prediction method based on data drift is characterized by comprising the following steps:
(1) measuring time periodT iMean value of internal solar radiation valuesI(t) And average value of corresponding meteorological parameterX(t) And is andT ifront continuously collected L-1 solar radiation valuesI(t 1)、I(t 2)、I(t 3)……I(t L-1) And corresponding meteorological parameter valuesX(t 1)、X(t 2)、X(t 3)……X(t L-1) L data samples are formed, respectively:I(t 1) AndX(t 1) Composition sampleX'(t 1)，I(t 2) AndX(t 2) Composition sampleX'(t 2)，I(t 3) AndX(t 3) Composition sampleX'(t 3)，……，I(t L-1) AndX(t L-1) Composition sampleX'(t L-1)，I(t) AndX(t) Composition sampleX'(t) Divide the L data samples inton 1、n 2Two sections;
(2) computingn 1、n 2Average of solar radiation values in both segmentsI(u 1) AndI(u 2) (ii) a When the oxygen deficiency is reachedI(u 1)-I(u 2)|>Threshold value TdIn time, radiation data drift occurs, and an abnormal prediction model M is adoptedBPredicting otherwise, using conventional prediction model MACarrying out prediction;
wherein the abnormality prediction model MBThe prediction is as follows:
if it isI(t) The first time data drift occurs, the temporary data set D is emptiedC，X'(t) And stores the temporary data set DCJudging the abnormal data set DBWhether the number of samples reaches the upper limit or not, and if not, the samples are takenX'(t) Addition to an abnormal data set DBIn the middle, training and updating the abnormal prediction model MBComputing an anomaly data set D when it is yesBThe difference S between all the data samplesB(t) andX'(t) With an abnormal data set DBThe degree of difference S (t) between all the data samples in (1), when SB(t) minimum value < S (t) minimum value,X'(t) Replacement of an anomalous data set DBMiddle SB(t) the sample data with smaller difference degree with other samples in the two samples corresponding to the minimum value in the (t) is obtained according to the new abnormal data set DBTraining update anomaly prediction model MB；
Conventional predictive model MAThe prediction is as follows:
if it isI(t) For data drift to occur continuously within L samples, thenX'(t) Stored to a temporary data set DCUp to a temporary data set DCThe number of the middle samples reaches the set number, and the data set D is processedCReplacing the regular data set D with the middle sampleAAll samples in (1); if it isI(t) Judging the conventional data set D without data driftAWhether the number of middle samples reaches the upper limit or not, if not, whether the number of middle samples reaches the upper limitX'(t) Direct addition to the regular data set DAIn updating the regular data set DA(ii) a Is whenX'(t) Replacement into regular data set DAThe earliest sample is collected and the conventional data set D is updatedA(ii) a From the new conventional data set DATraining update routine predictive model MA；
(3) Outputting the prediction result, returning to the step (1), and entering the next time periodT i+1Solar radiation data prediction and model updating.
2. According to the rightThe method for on-line dynamic prediction of solar radiation based on data drift of claim 1, wherein the solar radiation samples with continuous length L are divided inton 1、n 2Two segments, divided without overlap, i.e.n 1+n 2= L; or may be split with overlap, i.e.n 1+n 2＞L。
3. The online dynamic prediction method for solar radiation based on data drift according to claim 1, wherein the L-1 samplesX'(t 1)、X(t 2)、X'(t 3)……X'(t L-1) The number of the collected L-1 samples was determined.
4. The online dynamic prediction method for solar radiation based on data drift according to claim 1, wherein the L-1 samplesX'(t 1)、X(t 2)、X'(t 3)……X'(t L-1) For a continuous period of timeT L-1Samples collected during the process.
5. The method of claim 1, wherein the inter-sample difference is defined as Euclidean distance between two multidimensional samples.
6. The method of claim 1, wherein the meteorological parameters are partial or all meteorological variables of temperature, humidity, wind speed, wind direction, aerosol or cloud quantities corresponding to a photovoltaic power station or a radiation data station.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010011980.0A CN111242359B (en) | 2020-01-06 | 2020-01-06 | Solar radiation online dynamic prediction method based on data drift |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010011980.0A CN111242359B (en) | 2020-01-06 | 2020-01-06 | Solar radiation online dynamic prediction method based on data drift |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111242359A CN111242359A (en) | 2020-06-05 |
| CN111242359B true CN111242359B (en) | 2020-09-29 |
Family
ID=70872298
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010011980.0A Active CN111242359B (en) | 2020-01-06 | 2020-01-06 | Solar radiation online dynamic prediction method based on data drift |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111242359B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2023023114A (en) * | 2021-08-04 | 2023-02-16 | トヨタ自動車株式会社 | Power management system, power management server, and power management method |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8972221B2 (en) * | 2007-02-12 | 2015-03-03 | Locus Energy, Llc | Estimating solar irradiance components from plane of array irradiance and global horizontal irradiance |
| CN103955749B (en) * | 2014-03-19 | 2018-03-09 | 内蒙古电力勘测设计院有限责任公司 | Direct solar radiation value Forecasting Methodology and system |
| CN105447509A (en) * | 2015-11-10 | 2016-03-30 | 天津大学 | Short-term power prediction method for photovoltaic power generation system |
| CN106503833A (en) * | 2016-09-30 | 2017-03-15 | 天津大学 | Photovoltaic generation short term power rolling forecast method based on algorithm of support vector machine |
| CN108268961A (en) * | 2016-12-30 | 2018-07-10 | 中国电力科学研究院 | A kind of solar radiation ultra-short term prediction method based on weather pattern |
| CN106815659B (en) * | 2017-01-20 | 2020-09-18 | 国网浙江省电力公司电力科学研究院 | Ultra-short-term solar radiation prediction method and device based on hybrid model |
-
2020
- 2020-01-06 CN CN202010011980.0A patent/CN111242359B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN111242359A (en) | 2020-06-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhao et al. | Research and application based on the swarm intelligence algorithm and artificial intelligence for wind farm decision system | |
| Khan et al. | Genetic algorithm based optimized feature engineering and hybrid machine learning for effective energy consumption prediction | |
| Ye et al. | An ensemble method for short-term wind power prediction considering error correction strategy | |
| Pourhabib et al. | Short-term wind speed forecast using measurements from multiple turbines in a wind farm | |
| CN111488896B (en) | Distribution line time-varying fault probability calculation method based on multi-source data mining | |
| CN111832222A (en) | Pollutant concentration prediction model training method, prediction method and device | |
| Ding et al. | A novel seasonal adaptive grey model with the data-restacking technique for monthly renewable energy consumption forecasting | |
| CN110601250B (en) | Method for evaluating stability of static voltage of access point based on new energy power station | |
| CN103955749A (en) | Sun direct radiation value prediction method and system | |
| Alonso-Montesinos et al. | Economic effect of dust particles on photovoltaic plant production | |
| Elkazaz et al. | Performance assessment of an energy management system for a home microgrid with PV generation | |
| Jiang et al. | Ultra-short-term prediction of photovoltaic output based on an LSTM-ARMA combined model driven by EEMD | |
| Rahimi et al. | A comprehensive review on ensemble solar power forecasting algorithms | |
| CN110852477A (en) | Production water demand prediction method based on system dynamics | |
| CN108960522A (en) | A kind of photovoltaic power generation quantity prediction analysis method | |
| CN111242359B (en) | Solar radiation online dynamic prediction method based on data drift | |
| CN107256437B (en) | Photovoltaic power station ultra-short-term irradiance prediction method and system | |
| CN109388863B (en) | ARIMA model-based distributed photovoltaic output power prediction method | |
| Bou-Rabee et al. | BiLSTM Network-Based Approach for Solar Irradiance Forecasting in Continental Climate Zones | |
| Huang et al. | Techno-economic analysis of photovoltaic hydrogen production considering technological progress uncertainty | |
| CN111815051B (en) | GRNN photovoltaic power generation prediction method considering weather influence factors | |
| Agga et al. | Convolutional neural network (cnn) extended architectures for photovoltaic power production forecasting | |
| CN116565840A (en) | High-precision wind speed soft measurement method for wind power prediction of wind power plant | |
| CN111967652B (en) | Double-layer collaborative real-time correction photovoltaic prediction method | |
| Wei et al. | Short-term wind power prediction using an improved grey wolf optimization algorithm with back-propagation neural network |
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 | ||
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20200605 Assignee: Nanjing Ruiyan Intelligent Technology Co.,Ltd. Assignor: NANJING FORESTRY University Contract record no.: X2020320000387 Denomination of invention: An online dynamic prediction method of solar radiation based on data drift Granted publication date: 20200929 License type: Common License Record date: 20201218 Application publication date: 20200605 Assignee: NANJING FANGXINGWEIAI AGRICULTURAL TECHNOLOGY Co.,Ltd. Assignor: NANJING FORESTRY University Contract record no.: X2020320000377 Denomination of invention: An online dynamic prediction method of solar radiation based on data drift Granted publication date: 20200929 License type: Common License Record date: 20201218 |