CN115660132B - Photovoltaic power generation power prediction method and system - Google Patents
Photovoltaic power generation power prediction method and system Download PDFInfo
- Publication number
- CN115660132B CN115660132B CN202210938497.6A CN202210938497A CN115660132B CN 115660132 B CN115660132 B CN 115660132B CN 202210938497 A CN202210938497 A CN 202210938497A CN 115660132 B CN115660132 B CN 115660132B
- Authority
- CN
- China
- Prior art keywords
- power generation
- power
- predicted
- photovoltaic
- parameters
- 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
- 238000010248 power generation Methods 0.000 title claims abstract description 151
- 238000000034 method Methods 0.000 title claims abstract description 71
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 12
- 238000006243 chemical reaction Methods 0.000 claims description 7
- 238000004138 cluster model Methods 0.000 claims description 6
- 238000005516 engineering process Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000012407 engineering method Methods 0.000 claims description 3
- 238000003672 processing method Methods 0.000 claims description 2
- 238000013499 data model Methods 0.000 abstract description 4
- 230000009286 beneficial effect Effects 0.000 abstract 1
- 238000002310 reflectometry Methods 0.000 description 4
- 230000032683 aging Effects 0.000 description 3
- 230000005855 radiation Effects 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 239000000428 dust Substances 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 101000604223 Homo sapiens Nocturnin Proteins 0.000 description 1
- 102100038815 Nocturnin Human genes 0.000 description 1
- 229910021419 crystalline silicon Inorganic materials 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000001932 seasonal effect Effects 0.000 description 1
- 239000002210 silicon-based material Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention relates to a photovoltaic power generation power prediction method and a system, which belong to the technical field of power systems and comprise the following steps: generating a first predicted generated power of the photovoltaic module corresponding to the predicted time period according to the meteorological parameters, the equipment parameters of the photovoltaic module and the basic information of the power station; matching photovoltaic power generation parameters of the photovoltaic module based on a preset clustering model to obtain second predicted power corresponding to the predicted time period; and comparing the first predicted power with the second predicted power, and determining the final predicted power of photovoltaic power generation according to the comparison result. The beneficial effects are that: and adding a parallel preset clustering model as a data model, performing Euclidean distance similarity matching by using a K-means algorithm to obtain second predicted power generated in an average similarity period, and determining final photovoltaic power generation predicted power according to a comparison result of the two predictions, thereby improving the prediction precision and the robustness of the prediction method.
Description
Technical Field
The invention relates to the technical field of power systems, in particular to a photovoltaic power generation power prediction method and system.
Background
The photovoltaic power generation has strong periodic variability, the output power of the photovoltaic power generation is influenced by various meteorological factors, and parameters such as solar radiation intensity, atmospheric temperature, relative humidity, wind speed, wind direction, air pressure and the like have different degrees of influence on the photovoltaic power generation. The prediction algorithm for photovoltaic power generation is mainly used for predicting direct current power generation power of a photovoltaic power station and provides a reference for power generation performance for all photovoltaic power station assets.
Currently, the common practice in the industry falls into two main categories: prediction methods based on deep learning and prediction methods based on physical models. The prediction method based on deep learning uses historical photovoltaic power generation power data, and obtains prediction of future power generation through learning a mode rule in the historical data; the prediction method based on the physical model uses meteorological data and obtains the prediction of future generated power based on the photoelectric conversion principle of the photovoltaic module. The method completely depends on the accuracy and precision of prediction, and because of the characteristic of large randomness of photovoltaic power generation, when the difference between the predicted data and the real data is large, the overall robustness of the prediction is poor, and the fault tolerance is low.
Disclosure of Invention
In order to solve the technical problems, the invention provides a photovoltaic power generation power prediction method and a photovoltaic power generation power prediction system.
The technical problems solved by the invention can be realized by adopting the following technical scheme:
a photovoltaic power generation power prediction method, comprising:
generating a first predicted generated power corresponding to the predicted time period according to the meteorological parameters, the equipment parameters of the photovoltaic module and the basic information of the power station;
clustering photovoltaic power generation parameters in a photovoltaic power generation database through a K-means algorithm based on a preset clustering model to obtain second predicted power generation power corresponding to the predicted time period;
and comparing the first predicted power generation with the second predicted power generation, and determining the predicted power of photovoltaic power generation according to a comparison result.
Preferably, the method for generating the preset cluster model includes:
step A1, acquiring a data set comprising a plurality of samples to be trained, and randomly initializing a preset number of samples to be trained as clustering centers, wherein each clustering center corresponds to one class;
step A2, calculating the Euclidean distance from the rest samples to be trained to each clustering center, and matching each sample to be trained to the category corresponding to the smallest Euclidean distance;
a3, calculating the mass center of each category, and taking the calculated mass center as the updated clustering center;
and A4, repeating the steps A2-A3 until the samples to be trained in each category are not changed any more.
Preferably, before the photovoltaic power generation parameter is clustered, the method further comprises a characteristic engineering processing process, and the characteristic engineering processing process comprises the following steps:
combining the photovoltaic power generation parameters in the photovoltaic power generation database and the meteorological parameters in the meteorological database based on a characteristic engineering method to obtain statistical data;
and extracting feature data from the statistical data, and taking the extracted feature data as the data set comprising a plurality of samples to be trained, wherein the feature data comprises time feature information and meteorological feature information.
Preferably, in the step A2, the method for calculating the euclidean distance includes:
wherein s_i represents the ith sample to be trained; c_j represents the j-th cluster center; s_i m Representing the mth attribute of the ith sample to be trained; c_j m An mth attribute representing a jth of the cluster centers; d (s_i, c_j) represents the Euclidean distance from the ith sample to be trained to the jth cluster center.
Preferably, in the step A3, the updated formula of the centroid is as follows:
wherein r_i represents the i-th category; s represents the sample to be trained corresponding to the ith category; c_i new And representing the ith updated cluster center.
Preferably, the clustering processing method of the photovoltaic power generation parameters comprises the following steps:
acquiring meteorological parameters corresponding to the prediction time period, and carrying out the characteristic engineering processing process on the meteorological parameters to obtain characteristic data serving as meteorological prediction data;
calculating Euclidean distance from the weather forecast data to each clustering center;
the category corresponding to the smallest Euclidean distance is used as the attribution category obtained by matching the weather forecast data, and each attribution category comprises a plurality of characteristic data;
inputting the time characteristic information of a plurality of the characteristic data into the photovoltaic power generation database, and obtaining the photovoltaic instantaneous power corresponding to the time characteristic information;
and calculating the average value of the photovoltaic instantaneous power generation power to obtain the second predicted power generation power.
Preferably, the calculation method of the second predicted generated power includes:
wherein R represents the home class; p_pv_s_i represents the photovoltaic instantaneous power generation corresponding to the time characteristic information of the ith characteristic data; pc represents the second predicted generated power.
Preferably, the method for determining the predicted power of photovoltaic power generation includes:
calculating an absolute difference between the first predicted generated power and the second predicted generated power;
judging whether the absolute difference value is smaller than a preset threshold value or not:
if yes, outputting the first predicted power to serve as the photovoltaic power generation predicted power;
and if not, outputting the first predicted power generation power, the second predicted power generation power and an average value as the predicted power of the photovoltaic power generation.
Preferably, the method for generating the first predicted generated power includes:
P dc ＝η pv ×I t ×S×K 1 ÷1000；
wherein P is dc Generating power for said first prediction; η (eta) pv The photoelectric conversion efficiency of the photovoltaic module is; i t The total irradiance of the inclined plane of the photovoltaic module is; s is the area of the photovoltaic module; k (K) 1 And the direct current line loss coefficient of the photovoltaic module.
The invention also provides a photovoltaic power generation power prediction system, which is used for implementing the photovoltaic power generation power prediction method, and comprises the following steps:
the power generation power prediction module is used for generating first predicted power generation corresponding to the predicted time period according to the meteorological parameters, the equipment parameters of the photovoltaic module and the basic information of the power station;
the preset clustering model is used for carrying out clustering processing on photovoltaic power generation parameters in the photovoltaic power generation database through a K-means algorithm to obtain second predicted power generation power corresponding to the predicted time period;
and the comparison output module is respectively connected with the power generation power prediction module and the preset clustering model and is used for comparing the first predicted power generation power with the second predicted power generation power and determining the photovoltaic power generation predicted power according to a comparison result.
The technical scheme of the invention has the advantages that:
according to the invention, on the basis of obtaining the first predicted power by the traditional photovoltaic power generation power prediction method based on the physical model, a parallel preset clustering model is added as a data model, the K-means algorithm is used for carrying out Euclidean distance similarity matching, the second predicted power generation power in the average similarity period is obtained, the final photovoltaic power generation predicted power is determined by the comparison result of the two predictions, the problem that the prediction accuracy is reduced due to overlarge difference between the first predicted power generation power obtained by the prediction based on the physical model and the real situation is avoided, and the robustness of the prediction method is improved.
Drawings
FIG. 1 is a flow chart of a photovoltaic power generation power prediction method according to a preferred embodiment of the present invention;
FIG. 2 is a flow chart of a method for generating a preset cluster model according to a preferred embodiment of the present invention;
FIG. 3 is a flow chart of a feature engineering process according to a preferred embodiment of the invention;
FIG. 4 is a flow chart of a clustering method for photovoltaic power generation parameters according to a preferred embodiment of the present invention;
FIG. 5 is a flow chart of a method for determining the predicted power of photovoltaic power generation according to a preferred embodiment of the present invention;
fig. 6 is a block diagram of a photovoltaic power generation power prediction system according to a preferred embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
It should be noted that, without conflict, the embodiments of the present invention and features of the embodiments may be combined with each other.
The invention is further described below with reference to the drawings and specific examples, which are not intended to be limiting.
In a preferred embodiment of the present invention, based on the above-mentioned problems existing in the prior art, a photovoltaic power generation power prediction method is provided, which belongs to the technical field of power systems, as shown in fig. 1, and includes:
s1, generating first predicted generated power corresponding to a predicted time period according to meteorological parameters, equipment parameters of a photovoltaic module and basic information of a power station;
specifically, the weather parameters may be obtained from the weather database 1, and the weather parameters include, but are not limited to, air temperature corresponding to a certain time stamp, direct irradiance of the horizontal plane, scattered irradiance of the horizontal plane, total irradiance of the horizontal plane, and the like.
The device parameters of the photovoltaic module can be obtained from the device database 2, and the device parameters include, but are not limited to, the photoelectric conversion efficiency of the photovoltaic module under the standard test condition of the photovoltaic module, the temperature coefficient of the photovoltaic module, the rated operating temperature of the photovoltaic cell, the annual attenuation rate of the photovoltaic cell and the like.
The plant basic information can be obtained from the plant database 3, and the plant basic information includes, but is not limited to, the area of the photovoltaic module, the direct current line loss coefficient, the mismatch loss coefficient, the dust shielding loss coefficient, the years of using the photovoltaic module, the ratio of the irradiance of the direct radiation of the inclined plane to the horizontal plane, the inclination angle of the panel of the photovoltaic module, the ground reflectivity and the like.
Providing a physical model, wherein the physical model is connected with the meteorological database 1, the equipment database 2 and the power station database 3, and the first predicted power generation power is obtained based on the traditional photovoltaic power generation power prediction method according to the parameter prediction, and the first predicted power generation method comprises the following steps:
P dc ＝η pv ×I t ×S×K 1 ÷1000；
wherein P is dc Generating power for the first prediction; η (eta) pv The photoelectric conversion efficiency of the photovoltaic module is improved; i t The total irradiance of the inclined plane of the photovoltaic component is expressed in kilowatts per square meter (kW/m) 2 ) The method comprises the steps of carrying out a first treatment on the surface of the S is the area of the photovoltaic module in square meters (m 2 )；K 1 The direct current line loss coefficient of the photovoltaic module is dimensionless, and the default value can be 0.95.
Further, the method for generating the photoelectric conversion efficiency of the photovoltaic module comprises the following steps:
η pv ＝η s ×(1-α(T c -25))×K 2 ×K 3 ×K 4 ；
wherein eta pv The photoelectric conversion efficiency of the photovoltaic module is improved; η (eta) s Under standard test conditions (irradiance of incident light of 1 kW/m) 2 A temperature of 25 ℃ and an atmospheric mass of 1.5); alpha is photovoltaic componentIs related to the solar cell material and has the unit of DEG C -1 For crystalline silicon material, the value range is 0.003 DEG C -1 ～0.005℃ -1 ；T c The unit is the plate temperature of the photovoltaic module in a predicted time period; k (K) 2 The ageing loss coefficient of the photovoltaic module is dimensionless, and is decreased gradually according to a certain proportion each year; k (K) 3 The mismatch loss coefficient of the photovoltaic module is dimensionless, and the value range is 0.95-0.98; k (K) 4 The dust shielding loss coefficient of the photovoltaic module is dimensionless, and the value range is 0.9-0.95.
Further, the method for generating the plate temperature comprises the following steps:
wherein T is c The temperature is the plate temperature; t (T) a The unit is the temperature of the local air which is acquired according to the meteorological data and corresponds to the predicted time period; i t Is the total irradiance of the inclined plane, and is expressed in kilowatts per square meter (kW/m) 2 )；T NOCT The unit is the rated working temperature of the photovoltaic module.
Further, the method for generating the aging loss coefficient includes:
K 2 ＝(1-Y) i ；
wherein K is 2 Is the ageing loss coefficient; y is the annual attenuation rate of the photovoltaic module, and the default value can be 0.8%; iota is the number of years that the photovoltaic module is put into service.
Further, the method for generating the total irradiance of the inclined plane comprises the following steps:
wherein I is t Is the total irradiance of the inclined plane, and is expressed in kilowatts per square meter (kW/m) 2 )；R b The irradiance ratio of the direct radiation of the inclined plane to the horizontal plane is dimensionless, and the irradiance ratio is measured by declination angle, geographical latitude, inclination angle of the photovoltaic array and direction of the photovoltaic arrayThe azimuth angle and the time angle are comprehensively calculated, the calculation process can be obtained by adopting the prior art, and the details are not repeated here; i b Is the direct irradiance of the horizontal plane; i d For scattered irradiance in the horizontal plane, the units are kilowatts per square meter (kW/m) 2 ) The method comprises the steps of carrying out a first treatment on the surface of the I is the total irradiance in kilowatts per square meter (kW/m) 2 ) The method comprises the steps of carrying out a first treatment on the surface of the Beta is the panel inclination angle of the photovoltaic module; ρ is the ground reflectivity, which can be changed according to the ground material, and when the ground reflectivity has a default value, the value can be given according to the following table 1:
table 1 shows the ground reflectivities corresponding to different ground conditions
S2, clustering photovoltaic power generation parameters in a photovoltaic power generation database through a K-means algorithm based on a preset clustering model 6 to obtain second predicted power generation power corresponding to a predicted time period;
as a preferred embodiment, before the photovoltaic power generation parameter is clustered, the method further includes a feature engineering process, as shown in fig. 3, where the feature engineering process includes:
combining the photovoltaic power generation parameters in the photovoltaic power generation database 4 and the meteorological parameters in the meteorological database based on the characteristic engineering method to obtain statistical data;
and extracting feature data from the statistical data, wherein the extracted feature data is used as a data set comprising a plurality of samples to be trained, and the feature data comprises time feature information and meteorological feature information.
Specifically, the feature engineering is to extract features suitable for the input of a subsequent model from the original data. In this embodiment, before the clustering process, the photovoltaic power generation parameters in the photovoltaic power generation database 4 are subjected to feature engineering, so as to prepare data for generating the preset clustering model 6.
In an actual implementation process, the photovoltaic power generation parameter is historical power generation data of the photovoltaic module, and the historical power generation data comprises photovoltaic instantaneous power generation power corresponding to a certain moment of a certain month of a certain year. In a preferred embodiment, the storage format of the photovoltaic power generation parameters in the photovoltaic power generation database 4 may be { "timestamp": timestamp, "p_pv": p_pv }, having two attributes, wherein timestamp represents a timestamp, and may be specifically a certain time of a certain month of a certain year, and the predicted time period exists in a format of a timestamp; p_pv represents the photovoltaic instantaneous generation power corresponding to the time stamp.
Combining (merge) the photovoltaic power generation database and the weather database according to the timestamp to obtain the photovoltaic instantaneous power generation power and weather parameters corresponding to a certain moment of a certain month of a certain year, wherein the extracted characteristic data comprises time characteristic information and weather characteristic information, and the time characteristic information comprises but is not limited to: hour, month, season; preferably, the time feature information is extracted by the following steps: converting the above-mentioned timestamp attribute into standard readable datetime format; then extracting two attributes of hour and month from the datetime format; finally, a seasonal season is constructed based on the month Month attribute. Meteorological characteristic information includes, but is not limited to: temperature T a The method comprises the steps of carrying out a first treatment on the surface of the Horizontal plane direct irradiance I b The method comprises the steps of carrying out a first treatment on the surface of the Irradiance I of horizontal plane scattering d The method comprises the steps of carrying out a first treatment on the surface of the Total irradiance I in the horizontal plane.
Data set S after feature engineering fixed Is stored in the following format: { "Hour": hour, "montath": month, "season": season, "T a ”：T a ，”I b ”：I b ，”I d ”：I d "I": i, with 7 properties.
As a preferred embodiment, as shown in fig. 2, the method for generating the preset cluster model 6 includes:
step A1, acquiring a data set S comprising a plurality of samples to be trained fixed Randomly initializing a preset number of samples to be trainedAs cluster centers, each cluster center corresponds to a category;
step A2, calculating the Euclidean distance from the rest samples to be trained to each clustering center, and matching each sample to be trained to the category corresponding to the minimum Euclidean distance;
step A3, calculating the mass center of each category, and taking the calculated mass center as an updated clustering center;
and A4, repeating the steps A2-A3 until the samples to be trained in each category are not changed.
Specifically, in this embodiment, a preset cluster model 6 matched with a similar sequence based on euclidean distance is generated by using a data set subjected to feature engineering, and the specific generation method is as follows: (1) Randomly initializing k samples to be trained in a data set as a clustering center: a random number seed is set in {0, length (S fixed ) K random numbers, such as { a1, a2, a3 … ak }, are generated in the range of }, and are indexed in the data set S fixed The samples to be trained corresponding to the k random number indexes are found and used as initial clustering centers, and the k samples to be trained are marked as { s1, s2, s3 … sk }. Each category is denoted R, there are k categories: { R1, R2 … Rk }, category(s) ∈ { R1, R2 … Rk }, and k cluster centers are denoted { c1, c2, c3 … ck }. At this time, the k samples to be trained are corresponding cluster centers, that is, c1=s1, c2=s2, and ck=sk.
(2) And calculating Euclidean distances from the k initial clustering centers to other samples to be trained of the k initial clustering centers, and dividing the Euclidean distances into categories corresponding to the clustering centers with the smallest Euclidean distances.
(3) And calculating the mass centers of each class after the samples to be trained are reassigned, and updating the clustering centers according to the calculated mass centers.
(4) Repeating (2) and (3) until the internal elements of each category are not changed.
In a preferred embodiment, in step A2, the method for calculating the euclidean distance includes:
wherein s_i represents the ith sample to be trained; c_j represents a j-th cluster center; s_i m Representing the mth attribute of the ith sample to be trained; c_j m An mth attribute representing a jth cluster center; d (s_i, c_j) represents the Euclidean distance of the ith sample to be trained to the jth cluster center.
In a preferred embodiment, in step A3, the updated formula of the centroid is as follows:
wherein r_i represents the i-th category; s represents a sample to be trained corresponding to the ith category; c_i new Representing the ith updated cluster center.
In a preferred embodiment, the method for generating the second predicted generated power includes: acquiring meteorological parameters of a time period to be predicted, performing Euclidean distance similarity matching by using a K-means algorithm to obtain a belonging class of the meteorological parameters, and performing mean processing on instantaneous photovoltaic power generation power of the belonging class to obtain second predicted power generation power;
the second predicted generation power calculation method includes:
wherein R represents a home class; p_pv_s_i represents the photovoltaic instantaneous power generation power of the time characteristic information corresponding to the ith characteristic data; pc represents the second predicted generated power.
As a preferred embodiment, as shown in fig. 4, the clustering method of the photovoltaic power generation parameters includes:
acquiring weather parameters corresponding to a prediction time period, and performing characteristic engineering processing on the weather parameters to obtain characteristic data serving as weather prediction data;
calculating Euclidean distance from weather forecast data to each cluster center;
the category corresponding to the minimum Euclidean distance is used as a attribution category obtained by matching weather forecast data, and each attribution category comprises a plurality of characteristic data;
inputting the time characteristic information of a plurality of characteristic data into a photovoltaic power generation database, and obtaining the photovoltaic instantaneous power generation power corresponding to the time characteristic information;
and calculating the average value of the photovoltaic instantaneous power generation power to obtain a second predicted power generation power.
Specifically, in this embodiment, the period to be predicted is a future time of the photovoltaic power to be predicted in the photovoltaic power prediction method, and is denoted as tp; firstly, converting a time period tp to be predicted into a standard readable datetime format, and extracting time characteristic information; then, query matching is carried out on the meteorological data based on the time period tp to be predicted to obtain meteorological prediction data corresponding to the time period tp to be predicted, and the meteorological prediction data is recorded as sp: { "Hour": hour, "montath": month, "season": season, "T a ”：T a ，”I b ”：I b ，”I d ”：I d ，”I”：I}；
Inputting weather forecast data sp into a preset clustering model 6, and performing Euclidean distance similarity matching to obtain attribution class R of the weather parameter;
and acquiring time characteristic information of all objects in the attribution class, matching from a photovoltaic power generation database based on all time characteristic information to obtain photovoltaic instantaneous power generation power, and calculating the average value of all the matched photovoltaic instantaneous power generation power to be used as second predicted power generation power.
In the above scheme, the attribution class of the time period to be predicted is found through the K-means algorithm, then only the power of the samples in the attribution class is homogenized, only the sample points most similar in a small range (namely, a single attribution class) are considered, the essence of the method is equivalent to that the weights of other sample points with larger difference with the time period to be predicted are all forcedly set to 0, and the problem of excessive smoothness of the overall weighting method is eliminated.
And S3, comparing the first predicted power generation power with the second predicted power generation power, and determining the predicted power of the photovoltaic power generation according to a comparison result.
As a preferred embodiment, as shown in fig. 5, the method for determining the predicted power of photovoltaic power generation includes:
calculating an absolute difference between the first predicted power generation and the second predicted power generation;
judging whether the absolute difference value is smaller than a preset threshold value or not, wherein the preset threshold value is used for representing an acceptable deviation range of the first predicted generation power:
if yes, outputting first predicted power to be used as photovoltaic power generation predicted power;
if not, outputting the first predicted power and the second predicted power and an average value as the predicted power of the photovoltaic power generation.
The invention also provides a photovoltaic power generation power prediction system, which is used for implementing the photovoltaic power generation power prediction method, as shown in fig. 6, and comprises the following steps:
a generated power prediction module 5 for generating a first predicted generated power corresponding to the predicted time period based on the meteorological parameter, the equipment parameter of the photovoltaic module, and the station basic information;
the preset clustering model 6 is used for carrying out clustering processing on the photovoltaic power generation parameters in the photovoltaic power generation database through a K-means algorithm to obtain second predicted power generation power corresponding to the predicted time period;
the comparison output module 7 is respectively connected with the power generation power prediction module 5 and the preset clustering model 6 and is used for comparing the first predicted power generation power with the second predicted power generation power and determining the photovoltaic power generation predicted power according to the comparison result.
Specifically, aiming at the problems that the predicted power obtained by the traditional photovoltaic power generation power prediction method based on the physical model is easy to be different from the actual situation, so that the power generation performance is low and the efficiency is low, the embodiment of the invention adds a parallel preset clustering model 6 as a data model on the basis of the physical model, the data model uses a K-means algorithm to carry out Euclidean distance similarity matching to obtain second predicted power generated in an average similar period, then the second predicted power is compared with first predicted power output by the physical model, if the difference between the two predicted powers is large, the average value of the two predicted powers is taken as the photovoltaic power generation predicted power to be output, otherwise, the first predicted power is directly output as the photovoltaic power generation predicted power, and the final photovoltaic power generation predicted power is determined through the comparison result of the two predictions, so that the prediction accuracy is improved, and the prediction robustness is further improved.
The foregoing description is only illustrative of the preferred embodiments of the present invention and is not to be construed as limiting the scope of the invention, and it will be appreciated by those skilled in the art that equivalent substitutions and obvious variations may be made using the description and drawings, and are intended to be included within the scope of the present invention.
Claims (10)
1. A photovoltaic power generation power prediction method, comprising:
generating a first predicted generated power corresponding to the predicted time period according to the meteorological parameters, the equipment parameters of the photovoltaic module and the basic information of the power station;
clustering photovoltaic power generation parameters in a photovoltaic power generation database through a K-means algorithm based on a preset clustering model to obtain second predicted power generation power corresponding to the predicted time period;
comparing the first predicted power generation with the second predicted power generation, and determining the predicted power of photovoltaic power generation according to a comparison result;
the clustering processing method of the photovoltaic power generation parameters comprises the following steps:
acquiring meteorological parameters corresponding to the prediction time period, and performing a characteristic engineering processing process on the meteorological parameters to obtain characteristic data serving as meteorological prediction data;
inputting the weather forecast data into the preset clustering model, and performing Euclidean distance similarity matching to obtain attribution classes to which the weather parameters belong, wherein each attribution class comprises a plurality of characteristic data;
acquiring time characteristic information of a plurality of characteristic data in the attribution class, inputting the time characteristic information into the photovoltaic power generation database based on all the time characteristic information in the attribution class, and acquiring and obtaining photovoltaic instantaneous power generation corresponding to the time characteristic information;
and calculating the average value of the photovoltaic instantaneous power generation power to obtain the second predicted power generation power.
2. The photovoltaic power generation power prediction method according to claim 1, wherein the method for generating the preset cluster model includes:
step A1, acquiring a data set comprising a plurality of samples to be trained, and randomly initializing a preset number of samples to be trained as clustering centers, wherein each clustering center corresponds to one class;
step A2, calculating the Euclidean distance from the rest samples to be trained to each clustering center, and matching each sample to be trained to the category corresponding to the smallest Euclidean distance;
a3, calculating the mass center of each category, and taking the calculated mass center as the updated clustering center;
and A4, repeating the steps A2-A3 until the samples to be trained in each category are not changed any more.
3. The method for predicting photovoltaic power generation power according to claim 2, further comprising a feature engineering process before the clustering process is performed on the photovoltaic power generation parameters, wherein the feature engineering process comprises:
combining the photovoltaic power generation parameters in the photovoltaic power generation database and the meteorological parameters in the meteorological database based on a characteristic engineering method to obtain statistical data;
and extracting feature data from the statistical data, and taking the extracted feature data as the data set comprising a plurality of samples to be trained, wherein the feature data comprises time feature information and meteorological feature information.
4. The method according to claim 2, wherein in the step A2, the method for calculating the euclidean distance comprises:
wherein s_i represents the ith sample to be trained; c_j represents the j-th cluster center; s_i m Representing the mth attribute of the ith sample to be trained; c_j m An mth attribute representing a jth of the cluster centers; d (s_i, c_j) represents the Euclidean distance from the ith sample to be trained to the jth cluster center.
5. The method according to claim 2, wherein in the step A3, the update formula of the centroid is as follows:
wherein r_i represents the i-th category; s represents the sample to be trained corresponding to the ith category; c_i new And representing the ith updated cluster center.
6. The method for predicting photovoltaic power generation power according to claim 3, wherein the inputting the weather prediction data into the preset cluster model, performing similarity matching of euclidean distances, and obtaining the attribution class to which the weather parameter belongs, includes:
calculating Euclidean distance from the weather forecast data to each clustering center;
and taking the category corresponding to the smallest Euclidean distance as the attribution category obtained by matching the weather forecast data.
7. The photovoltaic power generation power prediction method according to claim 6, wherein the calculation method of the second predicted power generation power includes:
wherein R represents the home class; p_pv_s_i represents the photovoltaic instantaneous power generation corresponding to the time characteristic information of the ith characteristic data; pc represents the second predicted generated power.
8. The photovoltaic power generation power prediction method according to claim 1, characterized in that the method for determining photovoltaic power generation predicted power comprises:
calculating an absolute difference between the first predicted generated power and the second predicted generated power;
judging whether the absolute difference value is smaller than a preset threshold value or not:
if yes, outputting the first predicted power to serve as the photovoltaic power generation predicted power;
and if not, outputting the first predicted power generation power, the second predicted power generation power and an average value as the predicted power of the photovoltaic power generation.
9. The photovoltaic power generation power prediction method according to claim 1, wherein the first prediction power generation method includes:
P dc ＝η pv ×I t ×S×K 1 ÷1000；
wherein P is dc Generating power for said first prediction; η (eta) pv The photoelectric conversion efficiency of the photovoltaic module is; i t The total irradiance of the inclined plane of the photovoltaic module is; s is the area of the photovoltaic module; k (K) 1 And the direct current line loss coefficient of the photovoltaic module.
10. A photovoltaic power generation power prediction system for implementing the photovoltaic power generation power prediction method according to any one of claims 1 to 9, comprising:
the power generation power prediction module is used for generating first predicted power generation corresponding to the predicted time period according to the meteorological parameters, the equipment parameters of the photovoltaic module and the basic information of the power station;
the preset clustering model is used for carrying out clustering processing on photovoltaic power generation parameters in the photovoltaic power generation database through a K-means algorithm to obtain second predicted power generation power corresponding to the predicted time period;
and the comparison output module is respectively connected with the power generation power prediction module and the preset clustering model and is used for comparing the first predicted power generation power with the second predicted power generation power and determining the photovoltaic power generation predicted power according to a comparison result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210938497.6A CN115660132B (en) | 2022-08-05 | 2022-08-05 | Photovoltaic power generation power prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210938497.6A CN115660132B (en) | 2022-08-05 | 2022-08-05 | Photovoltaic power generation power prediction method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115660132A CN115660132A (en) | 2023-01-31 |
| CN115660132B true CN115660132B (en) | 2024-01-30 |
Family
ID=85024222
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210938497.6A Active CN115660132B (en) | 2022-08-05 | 2022-08-05 | Photovoltaic power generation power prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115660132B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117522156A (en) * | 2023-10-17 | 2024-02-06 | 江苏尚诚能源科技有限公司 | Distributed photovoltaic prediction evaluation method and system based on big data analysis |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109002915A (en) * | 2018-07-13 | 2018-12-14 | 福州大学 | Photovoltaic plant short term power prediction technique based on Kmeans-GRA-Elman model |
| CN109086928A (en) * | 2018-07-27 | 2018-12-25 | 福州大学 | Photovoltaic plant realtime power prediction technique based on SAGA-FCM-LSSVM model |
| CN111091139A (en) * | 2019-11-18 | 2020-05-01 | 特变电工西安电气科技有限公司 | Photovoltaic prediction method, device and equipment for similar day clustering and readable storage medium |
| CN111695601A (en) * | 2020-05-15 | 2020-09-22 | 特变电工西安电气科技有限公司 | Photovoltaic power prediction method, device, equipment and readable storage medium |
| CN112561139A (en) * | 2020-12-03 | 2021-03-26 | 安徽大学 | Short-term photovoltaic power generation power prediction method and system |
| CN114421530A (en) * | 2022-02-17 | 2022-04-29 | 科大智能科技股份有限公司 | Scheduling method and device of photovoltaic energy storage system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170286838A1 (en) * | 2016-03-29 | 2017-10-05 | International Business Machines Corporation | Predicting solar power generation using semi-supervised learning |
-
2022
- 2022-08-05 CN CN202210938497.6A patent/CN115660132B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109002915A (en) * | 2018-07-13 | 2018-12-14 | 福州大学 | Photovoltaic plant short term power prediction technique based on Kmeans-GRA-Elman model |
| CN109086928A (en) * | 2018-07-27 | 2018-12-25 | 福州大学 | Photovoltaic plant realtime power prediction technique based on SAGA-FCM-LSSVM model |
| CN111091139A (en) * | 2019-11-18 | 2020-05-01 | 特变电工西安电气科技有限公司 | Photovoltaic prediction method, device and equipment for similar day clustering and readable storage medium |
| CN111695601A (en) * | 2020-05-15 | 2020-09-22 | 特变电工西安电气科技有限公司 | Photovoltaic power prediction method, device, equipment and readable storage medium |
| CN112561139A (en) * | 2020-12-03 | 2021-03-26 | 安徽大学 | Short-term photovoltaic power generation power prediction method and system |
| CN114421530A (en) * | 2022-02-17 | 2022-04-29 | 科大智能科技股份有限公司 | Scheduling method and device of photovoltaic energy storage system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115660132A (en) | 2023-01-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20230136352A1 (en) | Method and system for predicting a day-ahead wind power of wind farms | |
| CN107766990B (en) | Method for predicting power generation power of photovoltaic power station | |
| CN112257941B (en) | Photovoltaic power station short-term power prediction method based on improved Bi-LSTM | |
| CN109002915B (en) | Photovoltaic power station short-term power prediction method based on Kmeans-GRA-Elman model | |
| CN107194495B (en) | Photovoltaic power longitudinal prediction method based on historical data mining | |
| CN105373857A (en) | Photovoltaic power station irradiance prediction method | |
| CN110766200A (en) | Method for predicting generating power of wind turbine generator based on K-means mean clustering | |
| CN111626473A (en) | Two-stage photovoltaic power prediction method considering error correction | |
| CN109902340A (en) | It is a kind of meter and complicated meteorology coupled characteristic multi-source-lotus combine scene generating method | |
| CN112070311A (en) | Day-ahead light power prediction method based on similar day clustering and meteorological factor weighting | |
| CN115660132B (en) | Photovoltaic power generation power prediction method and system | |
| CN116702937A (en) | Photovoltaic output day-ahead prediction method based on K-means mean value clustering and BP neural network optimization | |
| CN114676923A (en) | Method and device for predicting generated power, computer equipment and storage medium | |
| CN115759389A (en) | Day-ahead photovoltaic power prediction method based on weather type similar day combination strategy | |
| CN115759467A (en) | Time-division integrated learning photovoltaic prediction method for error correction | |
| Cattani | Combining data envelopment analysis and Random Forest for selecting optimal locations of solar PV plants | |
| CN110852492A (en) | Photovoltaic power ultra-short-term prediction method for finding similarity based on Mahalanobis distance | |
| CN114882373A (en) | Multi-feature fusion sandstorm prediction method based on deep neural network | |
| CN117578425A (en) | Regional power prediction correction method under extreme weather condition | |
| CN113095547A (en) | Short-term wind power prediction method based on GRA-LSTM-ICE model | |
| CN117744916A (en) | Method and device for predicting energy storage capacity, computer equipment and readable storage medium | |
| CN115456286A (en) | Short-term photovoltaic power prediction method | |
| CN112116127B (en) | Photovoltaic power prediction method based on association of meteorological process and power fluctuation | |
| CN112580899A (en) | Medium-and-long-term wind power generation prediction method and system fused with machine learning model | |
| TWI810750B (en) | Solar power forecasting method |
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 |