A method and system for predicting photovoltaic power station output based on model output prediction sequence

Abstract

The present invention discloses a method for predicting the output of a photovoltaic power station based on a model output prediction sequence, comprising the following steps: preprocessing is performed based on output data information of the photovoltaic power station to be predicted and meteorological forecast data information to obtain a total set of candidate training samples and a prediction model input feature vector; model training is performed on a first data-driven model based on the total set of candidate training samples to obtain a first model predicted output sequence; mutual information processing is performed between output prediction sequences based on the first model predicted output sequence, and a better output sequence and a prediction input feature vector are selected to perform model training on a second data-driven model to obtain a second model output prediction sequence; a weight vector is constructed through a better output sequence, and a final output prediction sequence is determined based on the weight vector and the second model output prediction sequence; the output of the photovoltaic power station is accurately predicted, thereby improving the stability of the output of the photovoltaic power station.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power station output prediction, and in particular to a method and system for predicting photovoltaic power station output based on a model output prediction sequence. Background Art

[0002] At present, the output prediction system of photovoltaic power stations cannot promote the dispatching department to make reasonable decisions, such as formulating power generation plans, standby configurations and optimal unit combinations, which is not conducive to the operation and maintenance of power stations. Moreover, it is not possible to reasonably deploy related work such as inspection and maintenance of power stations based on the output prediction results, thereby reducing the power loss caused by operation and maintenance. Therefore, in the grid-connected operation stage of photovoltaic power stations, in order to ensure the reasonable consumption of output power, reduce the loss of abandoned light, and meet the needs of more refined power dispatching, an accurate photovoltaic power station output prediction system should be established. Summary of the invention

[0003] In order to overcome the deficiencies of the prior art, the present invention provides a method and system for predicting the output of a photovoltaic power station based on a model output prediction sequence, which achieves accurate prediction of the output of the photovoltaic power station, thereby improving the stability of the output of the photovoltaic power station.

[0004] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical solutions:

[0005] In a first aspect, the present application provides a method for predicting the output of a photovoltaic power station based on a model output prediction sequence, comprising the following steps:

[0006] S101, preprocessing is performed based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information to obtain a total set of candidate training samples and a prediction model input feature vector;

[0007] S102, performing model training on a first data-driven model based on the total candidate set of training samples to obtain a first model predicted output sequence;

[0008] S103, performing mutual information processing between output prediction sequences based on the first model output prediction sequence, and selecting a better output sequence and a prediction input feature vector to perform model training on a second data-driven model to obtain a second model output prediction sequence;

[0009] S104, constructing a weight vector through a relatively optimal output sequence, and determining a final output prediction sequence based on the weight vector and the second model output prediction sequence.

[0010] Further, preprocessing is performed based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information to obtain the total set of training sample candidates and the prediction model input feature vector, including the following steps:

[0011] Constructing a prediction original data set based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information;

[0012] Perform linear interpolation on the original prediction data set;

[0013] Perform feature extraction on the predicted original data set processed by linear interpolation;

[0014] The feature extraction results are normalized, and the normalized expression is as follows:

[0015]

[0016] Among them, x nor are the normalized input and output eigenvalues, x is the original eigenvalue, and x max is the maximum eigenvalue, x min is the minimum eigenvalue;

[0017] Based on the normalized processing results, the output of the photovoltaic power station to be predicted and the astronomical total solar irradiance during the measured period are calculated.

[0018] Furthermore, the first data-driven model is trained based on the total candidate set of training samples to obtain a first model predicted output sequence, which includes the following steps:

[0019] Based on the first training sample candidate set, n training sample sets with the same weather type within the same output measurement period are selected to perform model training on the first data-driven model;

[0020] Based on the first data-driven model training results, a first model output prediction sequence is obtained.

[0021] Further, mutual information processing is performed between output prediction sequences based on the output sequence predicted by the first model, and a better output sequence and a prediction input feature vector are selected to perform model training on the second data-driven model, so as to obtain the output prediction sequence of the second model, including the following steps:

[0022] Based on the output sequence predicted by the first model, mutual information processing is performed between n output prediction sequences, and mutual information between the n output prediction sequences and the actual output sequence c is calculated;

[0023] Selecting a better output sequence and a predicted input feature vector to perform model training on the second data-driven model;

[0024] The model training result is performed based on the second data-driven model to obtain a second model output prediction sequence.

[0025] Furthermore, selecting a better output sequence specifically includes the following steps:

[0026] Let F = {pq [1],p q [2],…,p q [n]}, select the set {I(p q [i],c)|i=1,2,…n}, the p corresponding to the largest element q [i], let S = {p q [i]}, F = F-{p q [i]};

[0027] Select p from F that satisfies the maximum relevance and minimum redundancy criterion (maxΦ(D,R)) q [j], let S = S + {p q [j]}, F = F-{p q [j]}, where Φ(D,R)=DR, {p F [k]|k=1,…,K}=F, K is the number of elements in the set F;

[0028] When the number of elements in the set S reaches m, we get the set {p q [1],p q [2],…,p q [n]}, the set S consisting of the first m best prediction sequences that meet the maximum correlation and minimum redundancy criteria r ={p r [1],p r [2],…,p r [m]}, and record the corresponding fixed training sample collection period set T r ={t1,t2,…,t m} and the mutual information set I(P r ,c)={I(p r [1],c),I(p r [2],c),…,I(p r [m],c)}.

[0029] Further, selecting a better output sequence and a predicted input feature vector to perform model training on the second data-driven model includes the following steps:

[0030] Based on the second training sample candidate set, a plurality of training sample sets with the same weather type in the same output measurement period and a training sample set with a better output sequence are selected to perform model training on the second data-driven model;

[0031] The predicted input feature vector is used as input data of the base predictor of the second data-driven model to obtain the second model output prediction sequence.

[0032] Furthermore, constructing a weight vector through a better output sequence, and determining a final output prediction sequence based on the weight vector and the second model output prediction sequence includes the following steps:

[0033] The weight vector is constructed based on the optimal output sequence. The weight factor calculation expression of the weight vector is as follows:

[0034]

[0035] Among them, ω i is the weight factor, I(p r [i],c) is the mutual information set I(P r ,c) in the i-th mutual information, I(p r [j], c) is the jth mutual information in the mutual information set of the better output sequence, and the weight vector w = [ω1, ω2, …, ω m ] T .

[0036] The final output prediction sequence is calculated based on the weight vector and the second model output prediction sequence to obtain the final output prediction sequence. The expression for the final output prediction sequence calculation is as follows:

[0037] p ω =P×w

[0038] Among them, p ω is the output prediction sequence of the photovoltaic power station, P is the output prediction matrix {p p [1],p p [2],…,p p [m]}, the output prediction matrix is ​​formed by the output prediction sequence of the second model, and w is the weight vector.

[0039] A second aspect of the present application provides a system for predicting photovoltaic power station output based on a model output prediction sequence, comprising:

[0040] The local communication subsystem is used to communicate with the centralized ground power station control system through a wired or wireless local area network to obtain the output data information of the photovoltaic power station to be predicted and transmit it to the embedded prediction host subsystem;

[0041] The multi-source wide-area communication subsystem is used to communicate with the meteorological cloud service station, obtain meteorological forecast data information, and transmit it to the embedded prediction host subsystem;

[0042] Embedded prediction host subsystem, including:

[0043] The first processing unit is used to perform preprocessing based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information to obtain a total set of candidate training samples and a prediction model input feature vector;

[0044] A second processing unit is used to perform model training on the first data-driven model based on the total candidate set of training samples to obtain a first model predicted output sequence;

[0045] A third processing unit is used to perform mutual information processing between output prediction sequences based on the output sequence predicted by the first model, and select a better output sequence and a prediction input feature vector to perform model training on the second data-driven model to obtain a second model output prediction sequence;

[0046] The fourth processing unit is used to construct a weight vector through a relatively optimal output sequence, and determine a final output prediction sequence based on the weight vector and the second model output prediction sequence.

[0047] The beneficial effect of the present application is to achieve accurate prediction of the output of a photovoltaic power station, thereby improving the stability of the output of the photovoltaic power station. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0049] Figure 1 It is a schematic diagram of the steps of a method for predicting the output of a photovoltaic power station based on a model output prediction sequence of the present invention;

[0050] Figure 2 The present invention is a schematic diagram of an output prediction process of a method for predicting the output of a photovoltaic power station based on a model output prediction sequence. DETAILED DESCRIPTION

[0051] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0052] The following describes the embodiments of the present invention through specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0053] Embodiment 1:

[0054] A method for predicting the output of a photovoltaic power station based on a model output prediction sequence comprises the following steps:

[0055] S101, preprocessing is performed based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information to obtain a total set of candidate training samples and a prediction model input feature vector;

[0056] The output data information of the photovoltaic power station to be predicted and the weather forecast data information are obtained, and the output data information of the photovoltaic power station to be predicted and the weather forecast data information are preprocessed to obtain the total set of training sample candidates and the prediction model input feature vector.

[0057] Preprocessing based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information to obtain the total set of training sample candidates and the input feature vector of the prediction model includes the following steps:

[0058] Constructing a prediction original data set based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information;

[0059] Perform linear interpolation on the original prediction data set;

[0060] Perform feature extraction on the predicted original data set processed by linear interpolation;

[0061] The feature extraction results are normalized, and the normalized expression is as follows:

[0062]

[0063] Among them, x nor are the normalized input and output eigenvalues, x is the original eigenvalue, and x max is the maximum eigenvalue, x min is the minimum eigenvalue;

[0064] Based on the normalized processing results, the output of the photovoltaic power station to be predicted and the astronomical total solar irradiance during the measured period are calculated.

[0065] It should be noted that the total set of candidate training samples and the prediction model input feature vector are obtained by preprocessing the output data information of the photovoltaic power station to be predicted and the weather forecast data information, and the total set of candidate training samples includes 2 or more candidate training samples. For example, the total set of candidate training samples includes a first candidate training sample set and a second candidate training sample set, the first candidate training sample set includes n training sample sets, and the second candidate training sample set includes w training sample sets.

[0066] S102, performing model training on a first data-driven model based on the total candidate set of training samples to obtain a first model predicted output sequence;

[0067] By preprocessing the output data information of the photovoltaic power station to be predicted and the weather forecast data information, a total set of training sample candidates and a prediction model input feature vector are obtained, and a first data-driven model is trained based on the total set of training sample candidates to obtain a first model predicted output sequence.

[0068] Model training of the first data-driven model based on the total candidate set of training samples to obtain the first model predicted output sequence includes the following steps:

[0069] Based on the first training sample candidate set, n training sample sets with the same weather type within the same output measurement period are selected to perform model training on the first data-driven model;

[0070] Based on the first data-driven model training results, a first model output prediction sequence is obtained.

[0071] It should be noted that, based on the first training sample candidate set, n training sample sets with the same weather type in the first output measured period are selected, and the first data-driven model is trained to obtain the output in the first output measured period. For example, according to the first output measured period of the day before the time period to be predicted, a training sample set with the same weather type as that in this output measured period is selected from the total training sample candidate set, that is, several training sample sets are selected, and the first data-driven model is trained to obtain the output in the first output measured period. Therefore, based on the first training sample candidate set, n training sample sets with the same weather type in the same output measured period are selected, and the first data-driven model is trained to obtain n output prediction sequences {p q [1],p q [2],…,p q [n]}.

[0072] S103, performing mutual information processing between output prediction sequences based on the first model output prediction sequence, and selecting a better output sequence and a prediction input feature vector to perform model training on a second data-driven model to obtain a second model output prediction sequence;

[0073] The first data-driven model is trained based on the total set of candidate training samples to obtain the first model predicted output sequence, the mutual information between output prediction sequences is processed based on the first model predicted output sequence, and based on the mutual information processing results between the output prediction sequences, a better output sequence and a predicted input feature vector are selected to train the second data-driven model to obtain the second model output prediction sequence.

[0074] Based on the output sequence predicted by the first model, mutual information processing is performed between the output prediction sequences, and a better output sequence and a prediction input feature vector are selected to perform model training on the second data-driven model, so as to obtain the output prediction sequence of the second model, including the following steps:

[0075] Based on the output sequence predicted by the first model, mutual information processing is performed between n output prediction sequences, and mutual information between the n output prediction sequences and the actual output sequence c is calculated;

[0076] Selecting a better output sequence and a predicted input feature vector to perform model training on the second data-driven model;

[0077] The model training result is performed based on the second data-driven model to obtain a second model output prediction sequence.

[0078] Selecting a better output sequence specifically includes the following steps:

[0079] Let F = {p q [1],p q [2],…,p q [n]}, select the set {I(p q [i],c)|i=1,2,…n}, the p corresponding to the largest element q [i], let S = {p q [i]}, F = F-{p q [i]};

[0080] Select p from F that satisfies the maximum relevance and minimum redundancy criterion (maxΦ(D,R)) q [j], let S = S + {p q [j]}, F = F-{p q [j]}, where Φ(D,R)=DR, {p F [k]|k=1,…,K}=F, K is the number of elements in the set F;

[0081] When the number of elements in the set S reaches m, we get the set {p q [1],p q [2],…,p q [n]}, the set S consisting of the first m best prediction sequences that meet the maximum correlation and minimum redundancy criteria r ={p r [1],p r [2],…,p r [m]}, and record the corresponding fixed training sample collection period set T r ={t1,t2,…,t m} and the mutual information set I(P r ,c)={I(p r [1],c),I(p r [2],c),…,I(p r [m],c)}.

[0082] Selecting a better output sequence and predicting the input feature vector to train the second data-driven model includes the following steps:

[0083] Based on the second training sample candidate set, a plurality of training sample sets with the same weather type in the same output measurement period and a training sample set with a better output sequence are selected to perform model training on the second data-driven model;

[0084] The predicted input feature vector is used as input data of the base predictor of the second data-driven model to obtain the second model output prediction sequence.

[0085] It should be noted that, based on the second training sample candidate set, several training sample sets with the same weather type in the same output measurement period and the training sample set with the better output sequence are selected to train the second data-driven model. Based on the second training sample candidate set, several training sample sets with the same weather type in the same output measurement period are selected to train the second data-driven model to obtain the output of the first output prediction period of the prediction period. The training sample set with the better output sequence is selected to train the second data-driven model, that is, the training sample collection period set T of the better output sequence is selected. r , the second data-driven model is trained. Therefore, based on the second training sample candidate set, several training sample sets with the same weather type in the same output measurement period and a training sample set with a better output sequence are selected to train the second data-driven model, and m output prediction models are obtained.

[0086] The predicted input feature vector is used as the input data of the base predictor of the second data-driven model to obtain the second model output prediction sequence. The predicted input feature vector is used as the input data of m output prediction models to obtain m output prediction sequences, that is, the predicted input feature vector is used as the input data of the base predictor of the second data-driven model to obtain the second model output prediction sequence.

[0087] S104, constructing a weight vector through a relatively optimal output sequence, and determining a final output prediction sequence based on the weight vector and the second model output prediction sequence;

[0088] By performing mutual information processing between output prediction sequences based on the first model predicted output sequence, based on the mutual information processing results between the output prediction sequences, a better output sequence and a predicted input feature vector are selected to perform model training on the second data-driven model to obtain the second model output prediction sequence, a weight vector is constructed through the better output sequence, and a final output prediction sequence is determined based on the weight vector and the second model output prediction sequence.

[0089] Constructing a weight vector through a better output sequence, and determining a final output prediction sequence based on the weight vector and the second model output prediction sequence includes the following steps:

[0090] The weight vector is constructed based on the optimal output sequence. The weight factor calculation expression of the weight vector is as follows:

[0091]

[0092] Among them, ω i is the weight factor, I(p r [i],c) is the mutual information set I(P r ,c) in the i-th mutual information, I(p r [j], c) is the jth mutual information in the mutual information set of the better output sequence, and the weight vector w = [ω1, ω2, …, ω m ] T .

[0093] The final output prediction sequence is calculated based on the weight vector and the second model output prediction sequence to obtain the final output prediction sequence. The expression for the final output prediction sequence calculation is as follows:

[0094] p ω =P×w

[0095] Among them, p ω is the output prediction sequence of the photovoltaic power station, P is the output prediction matrix {p p [1],p p [2],…,p p[m]}, the output prediction matrix is ​​formed by the output prediction sequence of the second model, and w is the weight vector.

[0096] It should be noted that the output prediction matrix is ​​constructed, and the output values ​​in other output prediction cycles of the predicted period are predicted, so as to obtain the training sample collection period set T for the predicted period and the selection of the better output sequence. r The m output prediction sequences {p p [1],p p [2],…,p p [m]}, the output prediction sequences are combined to form the output prediction matrix P = [p p [1],p p [2],…,p p [m]].

[0097] The above is a method for predicting the output of a photovoltaic power station based on a model output prediction sequence provided in an embodiment of the present application. The following is a system for predicting the output of a photovoltaic power station based on a model output prediction sequence provided in an embodiment of the present application.

[0098] A photovoltaic power station output prediction system based on a model output prediction sequence, comprising:

[0099] The local communication subsystem is used to communicate with the centralized ground power station control system through a wired or wireless local area network to obtain the output data information of the photovoltaic power station to be predicted and transmit it to the embedded prediction host subsystem;

[0100] The multi-source wide-area communication subsystem is used to communicate with the meteorological cloud service station, obtain meteorological forecast data information, and transmit it to the embedded prediction host subsystem;

[0101] Embedded prediction host subsystem, including:

[0102] The first processing unit is used to perform preprocessing based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information to obtain a total set of candidate training samples and a prediction model input feature vector;

[0103] A second processing unit is used to perform model training on the first data-driven model based on the total candidate set of training samples to obtain a first model predicted output sequence;

[0104] A third processing unit is used to perform mutual information processing between output prediction sequences based on the output sequence predicted by the first model, and select a better output sequence and a prediction input feature vector to perform model training on the second data-driven model to obtain a second model output prediction sequence;

[0105] The fourth processing unit is used to construct a weight vector through a relatively optimal output sequence, and determine a final output prediction sequence based on the weight vector and the second model output prediction sequence.

[0106] The embedded prediction host subsystem includes a first processing unit, a second processing unit, a third processing unit, and a fourth processing unit.

[0107] The first processing unit is used to perform preprocessing based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information to obtain the total set of candidate training samples and the input feature vector of the prediction model. The preprocessing based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information to obtain the total set of candidate training samples and the input feature vector of the prediction model includes the following steps:

[0108] Constructing a prediction original data set based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information;

[0109] Perform linear interpolation on the original prediction data set;

[0110] Perform feature extraction on the predicted original data set processed by linear interpolation;

[0111] The feature extraction results are normalized, and the normalized expression is as follows:

[0112]

[0113] Among them, x nor are the normalized input and output eigenvalues, x is the original eigenvalue, and x max is the maximum eigenvalue, x min is the minimum eigenvalue;

[0114] Based on the normalized processing results, the output of the photovoltaic power station to be predicted and the astronomical total solar irradiance during the measured period are calculated.

[0115] The second processing unit is used to perform model training on the first data-driven model based on the total set of training sample candidates to obtain a first model predicted output sequence. Performing model training on the first data-driven model based on the total set of training sample candidates to obtain a first model predicted output sequence includes the following steps:

[0116] Based on the first training sample candidate set, n training sample sets with the same weather type within the same output measurement period are selected to perform model training on the first data-driven model;

[0117] Based on the first data-driven model training results, a first model output prediction sequence is obtained.

[0118] The third processing unit is used to perform mutual information processing between output prediction sequences based on the first model predicted output sequence, and select a better output sequence and a predicted input feature vector to perform model training on the second data-driven model to obtain a second model output prediction sequence. Performing mutual information processing between output prediction sequences based on the first model predicted output sequence, and selecting a better output sequence and a predicted input feature vector to perform model training on the second data-driven model to obtain a second model output prediction sequence includes the following steps:

[0119] Based on the output sequence predicted by the first model, mutual information processing is performed between n output prediction sequences, and mutual information between the n output prediction sequences and the actual output sequence c is calculated;

[0120] Selecting a better output sequence and a predicted input feature vector to perform model training on the second data-driven model;

[0121] The model training result is performed based on the second data-driven model to obtain a second model output prediction sequence.

[0122] Selecting a better output sequence specifically includes the following steps:

[0123] Let F = {p q [1],p q [2],…,p q [n]}, select the set {I(p q [i],c)|i=1,2,…n}, the p corresponding to the largest element q [i], let S = {p q [i]}, F = F-{p q [i]};

[0124] Select p from F that satisfies the maximum relevance and minimum redundancy criterion (maxΦ(D,R)) q [j], let S = S + {p q [j]}, F = F-{p q [j]}, where Φ(D,R)=DR, {p F [k]|k=1,…,K}=F, K is the number of elements in the set F;

[0125] When the number of elements in the set S reaches m, we get the set {p q [1],p q [2],…,p q [n]}, the set S consisting of the first m best prediction sequences that meet the maximum correlation and minimum redundancy criteria r ={p r [1],p r [2],…,pr [m]}, and record the corresponding fixed training sample collection period set T r ={t1,t2,…,t m} and the mutual information set I(P r ,c)={I(p r [1],c),I(p r [2],c),…,I(P r [m],c)}.

[0126] Selecting a better output sequence and predicting the input feature vector to train the second data-driven model includes the following steps:

[0127] Based on the second training sample candidate set, a plurality of training sample sets with the same weather type in the same output measurement period and a training sample set with a better output sequence are selected to perform model training on the second data-driven model;

[0128] The predicted input feature vector is used as input data of the base predictor of the second data-driven model to obtain the second model output prediction sequence.

[0129] The fourth processing unit is used to construct a weight vector through a better output sequence, and determine a final output prediction sequence based on the weight vector and the second model output prediction sequence. Constructing a weight vector through a better output sequence, and determining a final output prediction sequence based on the weight vector and the second model output prediction sequence includes the following steps:

[0130] The weight vector is constructed based on the optimal output sequence. The weight factor calculation expression of the weight vector is as follows:

[0131]

[0132] Among them, ω i is the weight factor, I(p r [i],c) is the mutual information set I(P r ,c) in the i-th mutual information, I(p r [j], c) is the jth mutual information in the mutual information set of the better output sequence, and the weight vector w = [ω1, ω2, …, ω m ] T .

[0133] The final output prediction sequence is calculated based on the weight vector and the second model output prediction sequence to obtain the final output prediction sequence. The expression for the final output prediction sequence calculation is as follows:

[0134] p ω =P×w

[0135] Among them, p ω is the output prediction sequence of the photovoltaic power station, P is the output prediction matrix {p p [1],p p [2],…,p p [m]}, the output prediction matrix is ​​formed by the output prediction sequence of the second model, and w is the weight vector.

[0136] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0137] The terms "first", "second" and "third" etc. in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in a sequence other than those illustrated or described herein, for example. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0138] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0139] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0140] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0141] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0142] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (full name in English: Read-Only Memory, English abbreviation: ROM), random access memory (full name in English: Random Access Memory, English abbreviation: RAM), disk or optical disk and other media that can store program codes.

[0143] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for predicting photovoltaic power station output based on a model output prediction sequence, characterized in that: The following steps are involved: S101, preprocessing is performed based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information to obtain a total set of candidate training samples and a prediction model input feature vector; S102, performing model training on a first data-driven model based on the total candidate set of training samples to obtain a first model predicted output sequence; S103, performing mutual information processing between output prediction sequences based on the first model output prediction sequence, and selecting a better output sequence and a prediction input feature vector to perform model training on a second data-driven model to obtain a second model output prediction sequence; S104, constructing a weight vector through a relatively optimal output sequence, and determining a final output prediction sequence based on the weight vector and the second model output prediction sequence; The method of performing mutual information processing between output prediction sequences based on the first model output prediction sequence, and selecting a better output sequence and a prediction input feature vector to perform model training on the second data-driven model to obtain the second model output prediction sequence includes the following steps: Based on the output sequence predicted by the first model, mutual information processing is performed between n output prediction sequences, and mutual information between the n output prediction sequences and the actual output sequence c is calculated; Selecting a better output sequence and a predicted input feature vector to perform model training on the second data-driven model; Perform model training based on the second data-driven model to obtain a second model output prediction sequence; The selection of a better output sequence specifically includes the following steps: Let F = {p q [1],p q [2],…,p q [n]}, select the set {I(p q [i],c)|i=1,2,…n}, the p corresponding to the largest element q [i], let S = {p q [i]}, F=F―{p q [i]}; Select p from F that satisfies the maximum relevance and minimum redundancy criterion (maxΦ(D,R)) q [j], let S = S + {p q [j]}, F=F―{p q [j]}, where Φ(D,R)=D―R, {p F [k]|k=1,…,K}=F, K is the number of elements in the set F; When the number of elements in the set s reaches m, we get the set {p q [1],p q [2],…,p q [n]}, the set S consisting of the first m best prediction sequences that meet the maximum correlation and minimum redundancy criteria r ={p r [1],p r [2],…,p r [m]}, and record the corresponding fixed training sample collection period set T r ={t1,t2,…,t m } and the mutual information set I(P r ,c)={I(p r [1],c),I(p r [2],c),…,I(p r [m],c)}.

2. The method for predicting photovoltaic power station output based on model output prediction sequence according to claim 1 is characterized in that: The preprocessing based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information to obtain the total set of training sample candidates and the prediction model input feature vector includes the following steps: Constructing a prediction original data set based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information; Perform linear interpolation on the original prediction data set; Perform feature extraction on the predicted original data set processed by linear interpolation; Normalize the feature extraction results; Based on the normalized processing results, the output of the photovoltaic power station to be predicted and the astronomical total solar irradiance during the measured period are calculated.

3. The method for predicting photovoltaic power station output based on model output prediction sequence according to claim 2, characterized in that: The feature extraction processing result is normalized, and the expression of the normalization processing is as follows: Among them, x nor are the normalized input and output eigenvalues, x is the original eigenvalue, and x max is the maximum eigenvalue, x min is the minimum eigenvalue.

4. The method for predicting photovoltaic power station output based on model output prediction sequence according to claim 1, characterized in that: The method of performing model training on the first data-driven model based on the total candidate set of training samples to obtain the first model predicted output sequence comprises the following steps: Based on the first training sample candidate set, n training sample sets with the same weather type within the same output measurement period are selected to perform model training on the first data-driven model; Based on the first data-driven model training results, a first model output prediction sequence is obtained.

5. The method for predicting photovoltaic power station output based on model output prediction sequence according to claim 1, characterized in that: The selecting of a better output sequence and a predicted input feature vector to perform model training on the second data driven model comprises the following steps: Based on the second training sample candidate set, a plurality of training sample sets with the same weather type in the same output measurement period and a training sample set with a better output sequence are selected to perform model training on the second data-driven model; The predicted input feature vector is used as input data of the base predictor of the second data-driven model to obtain the second model output prediction sequence.

6. The method for predicting photovoltaic power station output based on model output prediction sequence according to claim 1, characterized in that: The method of constructing a weight vector through a relatively optimal output sequence and determining a final output prediction sequence based on the weight vector and the second model output prediction sequence comprises the following steps: Construct a weight vector based on the optimal output sequence; The final output prediction sequence is calculated based on the weight vector and the second model output prediction sequence to obtain the final output prediction sequence. The expression for the final output prediction sequence calculation is as follows: p ω =P×w Among them, p ω is the output prediction sequence of the photovoltaic power station, P is the output prediction matrix {p p [1],p p [2],…,p p [m]}, the output prediction matrix is ​​formed by the output prediction sequence of the second model, and w is the weight vector.

7. The method for predicting photovoltaic power station output based on model output prediction sequence according to claim 6, characterized in that: The weight vector is constructed based on the optimal output sequence, and the weight factor calculation expression of the weight vector is as follows: Among them, ω i is the weight factor, I(p r [i],c) is the mutual information set I(P r ,c) in the i-th mutual information, I(p r [j], c) is the jth mutual information in the mutual information set of the better output sequence, and the weight vector w = [ω1, ω2, …, ω m ] T .

8. A system for predicting the output of a photovoltaic power station based on a model output prediction sequence, used to implement the method for predicting the output of a photovoltaic power station based on a model output prediction sequence as described in any one of claims 1 to 7, characterized in that: include: The local communication subsystem is used to communicate with the centralized ground power station control system through a wired or wireless local area network to obtain the output data information of the photovoltaic power station to be predicted and transmit it to the embedded prediction host subsystem; The multi-source wide-area communication subsystem is used to communicate with the meteorological cloud service station, obtain meteorological forecast data information, and transmit it to the embedded prediction host subsystem; Embedded prediction host subsystem, including: The first processing unit is used to perform preprocessing based on the output data information of the photovoltaic power station to be predicted and the weather forecast data information to obtain a total set of candidate training samples and a prediction model input feature vector; A second processing unit is used to perform model training on the first data-driven model based on the total candidate set of training samples to obtain a first model predicted output sequence; A third processing unit is used to perform mutual information processing between output prediction sequences based on the output sequence predicted by the first model, and select a better output sequence and a prediction input feature vector to perform model training on the second data-driven model to obtain a second model output prediction sequence; The fourth processing unit is used to construct a weight vector through a relatively optimal output sequence, and determine a final output prediction sequence based on the weight vector and the second model output prediction sequence.

Images