Generated power prediction method for photovoltaic power generation

Abstract

The invention relates to the technical field of photovoltaic power generation power prediction, and discloses a photovoltaic power generation power prediction method, which comprises the following steps: step S101, collecting characteristic parameters of a photovoltaic generator set, and dividing the characteristic parameters into M clusters through clustering analysis; step S102, constructing a feature sequence of M clusters; s103, constructing time sequence diagram structure data as sample data of a training sample; step S104, collecting the sum value of the output power of the photovoltaic generator set in the second preset time period T2 as a sample label of the training sample; step S105, repeatedly executing the steps S101 to S104 until U training samples are obtained, and training the generated power prediction model through the training samples; according to the invention, the characteristic parameters of the plurality of photovoltaic generator sets are constructed into the time sequence diagram structure data, and the information aggregation is carried out on the time sequence diagram structure data through the generated power prediction model, so that the prediction precision of the generated power of the whole photovoltaic power station is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power generation prediction, and more specifically, to a photovoltaic power generation prediction method. Background Art

[0002] As the global demand for renewable energy continues to grow, photovoltaic power generation has been widely used as a clean and sustainable form of energy. However, the output power of photovoltaic power generation is greatly affected by weather conditions (such as solar radiation, temperature, cloud cover, etc.), and has strong volatility and uncertainty, which brings great challenges to grid dispatching and energy management. Therefore, accurately predicting the power generation of photovoltaic power stations is of great significance for optimizing power dispatching and improving photovoltaic utilization.

[0003] A photovoltaic power station is usually composed of multiple photovoltaic power generation units, which use the photovoltaic effect to directly convert solar energy into electrical energy. The historical operating data of the photovoltaic power station is currently correlated and classified using the grey correlation analysis method to obtain similar day data sets under different weather types. The time series prediction model (such as LSTM, GRU, etc.) is trained using the similar day data sets to eliminate the impact of operating data fluctuations caused by weather factors. The parameters of the time series prediction model are then updated using an optimization algorithm (such as a particle swarm optimization algorithm, a whale optimization algorithm, etc.) to avoid it falling into a local optimal solution, thereby improving the prediction accuracy of the time series prediction model.

[0004] However, the above scheme does not take into account the impact of individual differences between different photovoltaic power generation groups on the power generation of the entire photovoltaic power station. For example, the geographical location, azimuth, shielding conditions and degree of aging of each photovoltaic power generation group are different, resulting in different responses of different photovoltaic power generation groups to meteorological conditions at the same time, resulting in low prediction accuracy of the power generation of the entire photovoltaic power station. Summary of the invention

[0005] The present invention provides a method for predicting power generation of photovoltaic power generation, which solves the technical problems in the above-mentioned background technology.

[0006] The present invention provides a method for predicting power generation of photovoltaic power generation, comprising the following steps: Step S101, within a first preset time period T1, characteristic parameters of K photovoltaic power generation groups in a photovoltaic power station are collected according to a preset time interval t, and the K photovoltaic power generation groups are divided into M clusters by cluster analysis; Characteristic parameters include: rated power, photoelectric conversion efficiency, installation inclination, azimuth, output power, solar radiation intensity, temperature, rainfall, effective illumination area, cumulative operating time, number of faults and aging degree; Step S102, constructing feature sequences of M clusters through a sliding window; The characteristic sequence includes N sequence units, each of which is represented by a characteristic vector, which is obtained by preprocessing the characteristic parameters of all photovoltaic generators in the cluster through a sliding window; Step S103, constructing time series graph structure data as sample data of training samples according to the feature sequences of the M clusters; The time series graph structure data includes: M nodes, node features and edges between nodes; The M nodes establish mapping relationships with the M clusters respectively; The features of a node are represented by the feature sequence of the cluster that establishes a mapping relationship with the node; Step S104, after the first preset time period T1, collecting the sum of the output powers of the K photovoltaic power generation groups in the photovoltaic power station in the second preset time period T2 as the sample label of the training sample; Step S105, repeatedly executing steps S101 to S104 until U training samples are obtained, and training the power generation prediction model through the training samples; The input of the power generation prediction model corresponds to the sample data of the training sample; The output of the power generation prediction model corresponds to the sample label of the training sample; Step S106, executing steps S101 to S103 to obtain the time series diagram structure data, inputting it into the trained power generation prediction model, and outputting the predicted value of the power generation of the photovoltaic power station in the second preset time period T2.

[0007] Further, , where length represents the size of the sliding window, step represents the step size of the sliding window, floor represents rounding down, the size and step size of the sliding window are both custom parameters, and the first preset time period T1, the preset time interval t, the number of photovoltaic power generation groups K, the number of clusters M, the second preset time period T2, and the number of training samples U are all custom parameters.

[0008] Furthermore, the effective illumination area of ​​the photovoltaic generator set The calculation formula is as follows: ; in represents the total area of ​​photovoltaic panels, Represents the output power of the photovoltaic generator set, It represents the photoelectric conversion efficiency of the photovoltaic generator set. Indicates the intensity of solar radiation; The calculation formula for the aging degree of photovoltaic generator sets is as follows: ; ; in and They represent the output power and rated power of the photovoltaic generator set respectively, e represents the natural constant, It represents the time attenuation coefficient, and its value range is between 0.01 and 0.03. Time represents the cumulative running time, which is expressed in years. It indicates the fault impact factor, and its value range is between 0.9 and 1. Freq indicates the number of fault occurrences. represents the environmental adaptability coefficient, It represents the rainfall influence coefficient, and its value range is between 0.01 and 0.02. Rain represents the rainfall.

[0009] Furthermore, the K photovoltaic power generation groups are divided into M clusters through cluster analysis, including the following steps: Step S201, randomly select one photovoltaic power generation group from the K photovoltaic power generation groups as the central node, then select M-1 photovoltaic power generation groups with the greatest similarity to the characteristic parameters of the central node as the central nodes, and initialize the current iteration number to 1; The calculation formula of similarity Sim is as follows: ; Where T = T1 / t, 1≤a≤K, 1≤b≤K, a≠b, and They represent the characteristic parameters of the ath and bth photovoltaic generators at the i-th time point, Dis represents the Euclidean distance, cosine represents the cosine similarity, and Represent the first weight coefficient and the second weight coefficient respectively, both are custom parameters, and + =1.

[0010] Step S202, calculating the similarity between the characteristic parameters of the remaining photovoltaic power generation groups and the characteristic parameters of the central node, and adding the photovoltaic power generation group with the greatest similarity to the central node to form a cluster; The number of photovoltaic power generation groups in each cluster must be greater than or equal to K / 4M; Step S203, respectively calculating the average values ​​of the characteristic parameters of all photovoltaic power generation groups in each cluster, and selecting the photovoltaic power generation group in each cluster with the greatest similarity to the average value as the central node, and the current number of iterations is accumulated by 1; Step S204, repeating steps S202 to S203 until the iteration termination condition is met to obtain M clusters; The iteration termination conditions include: the current iteration number is greater than or equal to the maximum iteration number or the similarity between the feature parameters of the central node of the current iteration number and the feature parameters of the central node of the previous iteration number is less than the similarity threshold, where the maximum iteration number and the similarity threshold are both custom parameters.

[0011] Furthermore, the characteristic parameters of all photovoltaic generators in the cluster are preprocessed through a sliding window to obtain a characteristic vector, including the following steps: Step S301, filling the missing value by taking the average value of the characteristic parameters of two adjacent time points; Step S302, obtaining a first vector by taking the average value of each characteristic parameter of all photovoltaic power generation groups in the sliding window; Step S303: normalize the first vector using a Min-Max normalization method to obtain a feature vector.

[0012] Furthermore, if the correlation between any two clusters is greater than or equal to the correlation threshold, an edge is constructed between the nodes that establish a mapping relationship with the clusters, where the correlation threshold is a custom parameter; The calculation formula of the correlation Rel between two clusters is as follows: ; Where 1≤c≤M, 1≤d≤M, c≠d, and They represent the nth sequence unit of the feature sequence of the cth and dth clusters respectively, Dis represents the Euclidean distance, min represents the minimum value, and max represents the maximum value. Represents the resolution coefficient, which is assigned a value of 0.5.

[0013] Further, the power generation prediction model includes: M first hidden layers, M extraction layers, 1 second hidden layer, 1 third hidden layer and 1 first classifier; The M first hidden layers share weight parameters and bias parameters; Each first hidden layer includes N hidden units, the nth hidden unit of the mth first hidden layer inputs the nth sequence unit of the feature sequence corresponding to the mth node of the time series graph structure data, and outputs a first update vector, where 1≤m≤M, 1≤n≤N; The m-th extraction layer is used to extract the first update vector output by the N-th hidden unit of the m-th first hidden layer as the feature of the m-th node of the time series graph structure data; The second hidden layer inputs the time-series graph structure data and outputs the graph update matrix; The graph update matrix includes M row vectors, each row vector corresponds to the second update vector of a node; The third hidden layer inputs the graph update matrix and outputs the third update vector; The third update vector is input into the first classifier, and the classification space of the first classifier represents the power generation of the photovoltaic power station in the second preset time period T2.

[0014] Furthermore, the first hidden layer is constructed based on a gated recurrent neural network.

[0015] Furthermore, the calculation formula of the second hidden layer includes: ; ; ; Where 1≤i≤M, 1≤j≤M, i≠j, matrix represents the graph update matrix output by the second hidden layer, represents the second update vector of the i-th node, represents the set of nodes that have edges connected to the i-th node, and denote the first update vectors of the i-th and j-th nodes respectively, represents the attention score between the i-th node and the j-th node, and is a real value between 0 and 1. and Respectively represent the first weight parameter and the second weight parameter, MLP represents multi-layer perceptron, concat represents concatenation operation, represents a stacking operation on the second update vectors of M nodes, and Swish represents the Swish activation function; The calculation formula for the third hidden layer is as follows: ; in represents the third update vector of the third hidden layer output, represents the third weight parameter, and b represents the bias parameter.

[0016] Furthermore, before training the power generation prediction model, it is pre-trained. During the pre-training of the power generation prediction model, the first update vector output by the Nth hidden unit of the first hidden layer is input into the second classifier. The classification space of the second classifier represents the sum of the output powers of all photovoltaic power generation groups in the cluster corresponding to the next time interval. During the training of the power generation prediction model, the maximum number of iterations of the power generation prediction model is set, and the difference between the predicted value output by the power generation prediction model and the sample label of the training sample in each iteration is specified as the loss function.

[0017] The beneficial effect of the present invention is that the present invention constructs the characteristic parameters of multiple photovoltaic power generation groups into time-series diagram structure data, and aggregates the information of the time-series diagram structure data through the power generation prediction model, thereby improving the prediction accuracy of the power generation of the entire photovoltaic power station. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flow chart of a photovoltaic power generation power prediction method of the present invention; Figure 2 is a flow chart of dividing K photovoltaic power generation groups into M clusters through cluster analysis of the present invention; Figure 3 It is a flow chart of the present invention for preprocessing characteristic parameters of all photovoltaic power generation groups in a cluster through a sliding window to obtain characteristic vectors; Figure 4 It is a comparison chart of the verification results of the present invention. DETAILED DESCRIPTION

[0019] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Each example may omit, replace or add various processes or components as needed. In addition, the features described relative to some examples may also be combined in other examples.

[0020] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in one or more embodiments of the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0021] like Figure 1 to Figure 4 As shown, a method for predicting power generation of photovoltaic power generation includes the following steps: Step S101, within a first preset time period T1, characteristic parameters of K photovoltaic power generation groups in a photovoltaic power station are collected according to a preset time interval t, and the K photovoltaic power generation groups are divided into M clusters by cluster analysis; Characteristic parameters include: rated power, photoelectric conversion efficiency, installation inclination, azimuth, output power, solar radiation intensity, temperature, rainfall, effective illumination area, cumulative operating time, number of faults and aging degree; Step S102, constructing feature sequences of M clusters through a sliding window; The characteristic sequence includes N sequence units, each of which is represented by a characteristic vector, which is obtained by preprocessing the characteristic parameters of all photovoltaic generators in the cluster through a sliding window; Step S103, constructing time series graph structure data as sample data of training samples according to the feature sequences of the M clusters; The time series graph structure data includes: M nodes, node features and edges between nodes; The M nodes establish mapping relationships with the M clusters respectively; The features of a node are represented by the feature sequence of the cluster that establishes a mapping relationship with the node; Step S104, after the first preset time period T1, collecting the sum of the output powers of the K photovoltaic power generation groups in the photovoltaic power station in the second preset time period T2 as the sample label of the training sample; Step S105, repeatedly executing steps S101 to S104 until U training samples are obtained, and training the power generation prediction model through the training samples; The input of the power generation prediction model corresponds to the sample data of the training sample; The output of the power generation prediction model corresponds to the sample label of the training sample; Step S106, executing steps S101 to S103 to obtain the time series diagram structure data, inputting it into the trained power generation prediction model, and outputting the predicted value of the power generation of the photovoltaic power station in the second preset time period T2.

[0022] It should be noted that for medium and large photovoltaic power stations, which may be composed of tens of thousands of photovoltaic power generation units, if each photovoltaic power generation unit is used as a node of the timing graph structure data, the scale of the timing graph structure data is relatively large, resulting in a slow calculation speed of the power generation prediction model. Therefore, the present invention classifies photovoltaic power generation units with similar characteristics into the same cluster through cluster analysis, which greatly reduces the scale of the timing graph structure data, thereby retaining the data characteristics while greatly improving the training speed and calculation efficiency of the power generation prediction model.

[0023] In one embodiment of the present invention, the first preset time period T1, the preset time interval t, the number of photovoltaic power generation groups K, the number of clusters M, the second preset time period T2, and the number of training samples U are all custom parameters. Preferably, the default value of T1 is 24 hours, the default value of t is 15 minutes, K is set according to the actual number of photovoltaic power generation groups in the photovoltaic power station, M should be much smaller than K, the default value of M is 20, the default value of T2 is 24 hours, and the default value of U is 2000. , where length represents the size of the sliding window, step represents the step length of the sliding window, and floor represents rounding down. The size and step length of the sliding window are both custom parameters. Preferably, length = 4 × t, step = length / 2, then according to the default values ​​of T1 and t above, N = 47 can be calculated.

[0024] It should be noted that the first preset time period T1 and the second preset time period T2 can be set according to the requirements of the prediction time. For example, in short-term prediction, T1 and T2 can be set to 1 hour, and in long-term prediction, T1 and T2 can be set to 30 days. In addition, the sliding window can reduce the number of time points for collecting feature parameters, that is, reduce the length of the feature sequence. Similarly, while retaining data features, it can improve the training speed and computing efficiency of the power generation prediction model, which will not be elaborated here.

[0025] In one embodiment of the present invention, the effective illumination area of ​​the photovoltaic power generation group The calculation formula is as follows: ; in represents the total area of ​​photovoltaic panels, Represents the output power of the photovoltaic generator set, It represents the photoelectric conversion efficiency of the photovoltaic generator set. Indicates the intensity of solar radiation; The calculation formula for the aging degree of photovoltaic generator sets is as follows: ; ; in and They represent the output power and rated power of the photovoltaic generator set respectively, e represents the natural constant, It represents the time attenuation coefficient, and its value range is between 0.01 and 0.03. Time represents the cumulative running time, which is expressed in years. It indicates the fault impact factor, and its value range is between 0.9 and 1. Freq indicates the number of fault occurrences. represents the environmental adaptability coefficient, It represents the rainfall influence coefficient, and its value range is between 0.01 and 0.02. Rain represents the rainfall.

[0026] It should be noted that the photoelectric conversion efficiency refers to the efficiency of a photovoltaic power generation unit in converting the received solar radiation energy into electrical energy. It is related to the type of photovoltaic modules (such as monocrystalline silicon, polycrystalline silicon, thin film, etc.) and is provided by the photovoltaic panel manufacturer; the installation inclination angle refers to the inclination angle of the photovoltaic panel relative to the horizontal plane; the azimuth angle refers to the orientation angle of the photovoltaic panel; the solar radiation intensity, temperature and rainfall are all expressed by the solar radiation intensity, temperature and rainfall per unit time, such as a preset time interval t; the effective illumination area refers to the area on the photovoltaic panel that can effectively absorb sunlight.

[0027] In one embodiment of the present invention, the output power of the photovoltaic power generation group can be directly measured by a monitoring device such as an inverter or a power meter installed in the photovoltaic power generation group, or can be calculated by the following calculation formula: ; in It indicates the output power of the photovoltaic generator set under standard test conditions, that is, the output power of the photovoltaic generator set under a solar radiation intensity of 1000W / ㎡ and a temperature of 25℃. represents the solar radiation intensity, T represents the actual temperature of the photovoltaic panel, Indicates the temperature coefficient, provided by the manufacturer, which indicates the degree to which the output power of the photovoltaic panel changes with temperature. The default value is 0.3%. The above calculation formula will have a certain error compared to the output power measured directly by the monitoring equipment. It is suitable for situations where the monitoring equipment fails or the cost is limited.

[0028] In one embodiment of the present invention, Figure 2 As shown, K photovoltaic power generation groups are divided into M clusters through cluster analysis, including the following steps: Step S201, randomly select one photovoltaic power generation group from the K photovoltaic power generation groups as the central node, then select M-1 photovoltaic power generation groups with the greatest similarity to the characteristic parameters of the central node as the central nodes, and initialize the current iteration number to 1; The calculation formula of similarity Sim is as follows: ; Where T = T1 / t, 1≤a≤K, 1≤b≤K, a≠b, and They represent the characteristic parameters of the ath and bth photovoltaic generators at the i-th time point, Dis represents the Euclidean distance, cosine represents the cosine similarity, and Represent the first weight coefficient and the second weight coefficient respectively, both are custom parameters, and + =1, preferred, and Set to 0.3 and 0.7 respectively.

[0029] Step S202, calculating the similarity between the characteristic parameters of the remaining photovoltaic power generation groups and the characteristic parameters of the central node, and adding the photovoltaic power generation group with the greatest similarity to the central node to form a cluster; The number of photovoltaic power generation groups in each cluster must be greater than or equal to K / 4M; Step S203, respectively calculating the average values ​​of the characteristic parameters of all photovoltaic power generation groups in each cluster, and selecting the photovoltaic power generation group in each cluster with the greatest similarity to the average value as the central node, and the current number of iterations is accumulated by 1; Step S204, repeating steps S202 to S203 until the iteration termination condition is met to obtain M clusters; The iteration termination conditions include: the current number of iterations is greater than or equal to the maximum number of iterations or the similarity between the characteristic parameters of the central node of the current iteration and the characteristic parameters of the central node of the previous iteration is less than the similarity threshold, where the maximum number of iterations and the similarity threshold are both custom parameters. Preferably, the maximum number of iterations is set to 10 and the similarity threshold is set to 0.05.

[0030] It should be noted that the present invention selects M central nodes by the maximum distance principle, which can reduce randomness and improve the separation of clusters. M photovoltaic power generation groups can also be directly randomly selected as central nodes.

[0031] In one embodiment of the present invention, Figure 3 As shown, the characteristic parameters of all photovoltaic generators in the cluster are preprocessed through a sliding window to obtain a characteristic vector, including the following steps: Step S301, filling the missing value by taking the average value of the characteristic parameters of two adjacent time points; Step S302, obtaining a first vector by taking the average value of each characteristic parameter of all photovoltaic power generation groups in the sliding window; That is, the size of the first vector is 1×12, and the 12 dimensions represent the average values ​​of the rated power, photoelectric conversion efficiency, installation inclination, azimuth, output power, solar radiation intensity, temperature, rainfall, effective illumination area, cumulative operating time, number of faults, and aging degree of all photovoltaic generators. Step S303: normalize the first vector using a Min-Max normalization method to obtain a feature vector.

[0032] In one embodiment of the present invention, if the correlation between any two clusters is greater than or equal to a correlation threshold, an edge is constructed between the nodes that establish a mapping relationship with the clusters, wherein the correlation threshold is a custom parameter, and preferably, the correlation threshold is set to 0.5; The calculation formula of the correlation Rel between two clusters is as follows: ; Where 1≤c≤M, 1≤d≤M, c≠d, and They represent the nth sequence unit of the feature sequence of the cth and dth clusters respectively, Dis represents the Euclidean distance, min represents the minimum value, and max represents the maximum value. Represents the resolution coefficient, which is assigned a value of 0.5.

[0033] In one embodiment of the present invention, the power generation prediction model includes: M first hidden layers, M extraction layers, 1 second hidden layer, 1 third hidden layer and 1 first classifier; The M first hidden layers share weight parameters and bias parameters; Each first hidden layer includes N hidden units, the nth hidden unit of the mth first hidden layer inputs the nth sequence unit of the feature sequence corresponding to the mth node of the time series graph structure data, and outputs a first update vector, where 1≤m≤M, 1≤n≤N; The m-th extraction layer is used to extract the first update vector output by the N-th hidden unit of the m-th first hidden layer as the feature of the m-th node of the time series graph structure data; The second hidden layer inputs the time-series graph structure data and outputs the graph update matrix; The graph update matrix includes M row vectors, each row vector corresponds to the second update vector of a node; The third hidden layer inputs the graph update matrix and outputs the third update vector; The third update vector is input into the first classifier, and the classification space of the first classifier represents the power generation of the photovoltaic power station in the second preset time period T2.

[0034] In one embodiment of the present invention, the first hidden layer is constructed based on GRU (gated recurrent neural network) and can also be constructed based on LSTM (long short-term recurrent neural network model). In addition, the number of first hidden layers is determined by the number of clusters M. This parallel design can greatly improve the calculation speed of the power generation prediction model.

[0035] In one embodiment of the present invention, the calculation formula of the second hidden layer includes: ; ; ; Where 1≤i≤M, 1≤j≤M, i≠j, matrix represents the graph update matrix output by the second hidden layer, represents the second update vector of the i-th node, represents the set of nodes that have edges connected to the i-th node, and denote the first update vectors of the i-th and j-th nodes respectively, represents the attention score between the i-th node and the j-th node, and is a real value between 0 and 1. and Respectively represent the first weight parameter and the second weight parameter, MLP represents multi-layer perceptron, concat represents concatenation operation, represents a stacking operation on the second update vectors of M nodes, and Swish represents the Swish activation function; The calculation formula for the third hidden layer is as follows: ; in represents the third update vector of the third hidden layer output, represents the third weight parameter, and b represents the bias parameter.

[0036] It should be noted that the weight parameters and bias parameters in the power generation prediction model are all learnable parameters. For example, the size of the first update vector is 1×Q1, then the first weight parameter and the second weight parameter can be designed as a matrix of size Q1×Q2, then the size of the second update vector is 1×Q2, then the size of the update matrix is ​​M×Q2, and the third weight parameter can be designed as a vector of size Q2×1, then the size of the third update vector is M×1, where Q1 and Q2 are both custom parameters, for example, Q1 is set to 16 and Q2 is set to 8.

[0037] In one embodiment of the present invention, before training the power generation prediction model, it is pre-trained. During the pre-training of the power generation prediction model, the first update vector output by the Nth hidden unit of the first hidden layer is input into the second classifier. The classification space of the second classifier represents the sum of the output powers of all photovoltaic power generation groups in the cluster corresponding to the next time interval. During the training of the power generation prediction model, the maximum number of iterations of the power generation prediction model is set, and the difference between the predicted value output by the power generation prediction model and the sample label of the training sample in each iteration is specified as the loss function.

[0038] It should be noted that pre-training can speed up the training of the power generation prediction model, reduce the number of training samples during training, and speed up the convergence of the model. During the training process, the weight parameters and bias parameters in the power generation prediction model are updated through the gradient descent algorithm and the chain rule. The training of the neural network model is a conventional technical means and will not be elaborated here.

[0039] like Figure 4 As shown, 10 verification data are tested using the power generation prediction model provided by the present invention and the time series prediction model (built based on LSTM) of the prior art respectively. The verification data is the power generation of 1000 photovoltaic power generation groups for one day. It is found through observation that the power generation prediction model provided by the present invention has higher prediction accuracy.

[0040] The above describes an embodiment of the present embodiment, but the present embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present embodiment, ordinary technicians in this field can also make many forms, all of which are within the protection of the present embodiment.

Claims

1. A method for predicting power generation of photovoltaic power generation, characterized in that: The following steps are involved: Step S101, within a first preset time period T1, characteristic parameters of K photovoltaic power generation groups in a photovoltaic power station are collected according to a preset time interval t, and the K photovoltaic power generation groups are divided into M clusters by cluster analysis; Characteristic parameters include: rated power, photoelectric conversion efficiency, installation inclination, azimuth, output power, solar radiation intensity, temperature, rainfall, effective illumination area, cumulative operating time, number of faults and aging degree; Step S102, constructing feature sequences of M clusters through a sliding window; The characteristic sequence includes N sequence units, each of which is represented by a characteristic vector, which is obtained by preprocessing the characteristic parameters of all photovoltaic generators in the cluster through a sliding window; Step S103, constructing time series graph structure data as sample data of training samples according to the feature sequences of the M clusters; The time series graph structure data includes: M nodes, node features and edges between nodes; The M nodes establish mapping relationships with the M clusters respectively; The features of a node are represented by the feature sequence of the cluster that establishes a mapping relationship with the node; Step S104, after the first preset time period T1, collecting the sum of the output powers of the K photovoltaic power generation groups in the photovoltaic power station in the second preset time period T2 as the sample label of the training sample; Step S105, repeatedly executing steps S101 to S104 until U training samples are obtained, and training the power generation prediction model through the training samples; The input of the power generation prediction model corresponds to the sample data of the training sample; The output of the power generation prediction model corresponds to the sample label of the training sample; Step S106, executing steps S101 to S103 to obtain the time series diagram structure data, inputting it into the trained power generation prediction model, and outputting the predicted value of the power generation of the photovoltaic power station in the second preset time period T2.

2. A photovoltaic power generation power prediction method according to claim 1, characterized in that: , where length represents the size of the sliding window, step represents the step size of the sliding window, floor represents rounding down, the size and step size of the sliding window are both custom parameters, and the first preset time period T1, the preset time interval t, the number of photovoltaic power generation groups K, the number of clusters M, the second preset time period T2, and the number of training samples U are all custom parameters.

3. A photovoltaic power generation power prediction method according to claim 1, characterized in that: Effective illumination area of ​​photovoltaic generator set The calculation formula is as follows: ； in represents the total area of ​​photovoltaic panels, Represents the output power of the photovoltaic generator set, It represents the photoelectric conversion efficiency of the photovoltaic generator set. Indicates the intensity of solar radiation; The calculation formula for the aging degree of photovoltaic generator sets is as follows: ； ； in and They represent the output power and rated power of the photovoltaic generator set respectively, e represents the natural constant, It represents the time attenuation coefficient, and its value range is between 0.01 and 0.

03. Time represents the cumulative running time, which is expressed in years. It indicates the fault impact factor, and its value range is between 0.9 and 1. Freq indicates the number of fault occurrences. represents the environmental adaptability coefficient, It represents the rainfall influence coefficient, and its value range is between 0.01 and 0.

02. Rain represents the rainfall.

4. A photovoltaic power generation power prediction method according to claim 1, characterized in that: The K photovoltaic power generation groups are divided into M clusters through cluster analysis, including the following steps: Step S201, randomly select one photovoltaic power generation group from the K photovoltaic power generation groups as the central node, then select M-1 photovoltaic power generation groups with the greatest similarity to the characteristic parameters of the central node as the central nodes, and initialize the current iteration number to 1; The calculation formula of similarity Sim is as follows: ； Where T = T1 / t, 1≤a≤K, 1≤b≤K, a≠b, and They represent the characteristic parameters of the ath and bth photovoltaic generators at the i-th time point, Dis represents the Euclidean distance, cosine represents the cosine similarity, and Represent the first weight coefficient and the second weight coefficient respectively, both are custom parameters, and + =1. Step S202, calculating the similarity between the characteristic parameters of the remaining photovoltaic power generation groups and the characteristic parameters of the central node, and adding the photovoltaic power generation group with the greatest similarity to the central node to form a cluster; The number of photovoltaic power generation groups in each cluster must be greater than or equal to K / 4M; Step S203, respectively calculating the average values ​​of the characteristic parameters of all photovoltaic power generation groups in each cluster, and selecting the photovoltaic power generation group in each cluster with the greatest similarity to the average value as the central node, and the current number of iterations is accumulated by 1; Step S204, repeating steps S202 to S203 until the iteration termination condition is met to obtain M clusters; The iteration termination conditions include: the current iteration number is greater than or equal to the maximum iteration number or the similarity between the feature parameters of the central node of the current iteration number and the feature parameters of the central node of the previous iteration number is less than the similarity threshold, where the maximum iteration number and the similarity threshold are both custom parameters.

5. The photovoltaic power generation prediction method according to claim 1, characterized in that: The characteristic parameters of all photovoltaic generators in the cluster are preprocessed through a sliding window to obtain a characteristic vector, including the following steps: Step S301, filling the missing value by taking the average value of the characteristic parameters of two adjacent time points; Step S302, obtaining a first vector by taking the average value of each characteristic parameter of all photovoltaic power generation groups in the sliding window; Step S303: normalize the first vector using a Min-Max normalization method to obtain a feature vector.

6. A photovoltaic power generation power prediction method according to claim 1, characterized in that: If the correlation between any two clusters is greater than or equal to the correlation threshold, an edge is built between the nodes that establish a mapping relationship with the clusters, where the correlation threshold is a custom parameter; The calculation formula of the correlation Rel between two clusters is as follows: ； Where 1≤c≤M, 1≤d≤M, c≠d, and They represent the nth sequence unit of the feature sequence of the cth and dth clusters respectively, Dis represents the Euclidean distance, min represents the minimum value, and max represents the maximum value. Represents the resolution coefficient, which is assigned a value of 0.

5.

7. A photovoltaic power generation power prediction method according to claim 1, characterized in that: The power generation prediction model includes: M first hidden layers, M extraction layers, 1 second hidden layer, 1 third hidden layer and 1 first classifier; The M first hidden layers share weight parameters and bias parameters; Each first hidden layer includes N hidden units, the nth hidden unit of the mth first hidden layer inputs the nth sequence unit of the feature sequence corresponding to the mth node of the time series graph structure data, and outputs a first update vector, where 1≤m≤M, 1≤n≤N; The m-th extraction layer is used to extract the first update vector output by the N-th hidden unit of the m-th first hidden layer as the feature of the m-th node of the time series graph structure data; The second hidden layer inputs the time-series graph structure data and outputs the graph update matrix; The graph update matrix includes M row vectors, each row vector corresponds to the second update vector of a node; The third hidden layer inputs the graph update matrix and outputs the third update vector; The third update vector is input into the first classifier, and the classification space of the first classifier represents the power generation of the photovoltaic power station in the second preset time period T2.

8. A photovoltaic power generation power prediction method according to claim 7, characterized in that: The first hidden layer is built based on a gated recurrent neural network.

9. A photovoltaic power generation power prediction method according to claim 7, characterized in that: The calculation formula for the second hidden layer includes: ； ； ； Where 1≤i≤M, 1≤j≤M, i≠j, matrix represents the graph update matrix output by the second hidden layer, represents the second update vector of the i-th node, represents the set of nodes that have edges connected to the i-th node, and denote the first update vectors of the i-th and j-th nodes respectively, represents the attention score between the i-th node and the j-th node, and is a real value between 0 and 1. and Represent the first weight parameter and the second weight parameter respectively, MLP represents multi-layer perceptron, concat represents concatenation operation, represents a stacking operation on the second update vectors of M nodes, and Swish represents the Swish activation function; The calculation formula for the third hidden layer is as follows: ； in represents the third update vector of the third hidden layer output, represents the third weight parameter, and b represents the bias parameter.

10. The photovoltaic power generation power prediction method according to claim 1, characterized in that: Before training the power generation prediction model, it is pre-trained. During the pre-training of the power generation prediction model, the first update vector output by the Nth hidden unit of the first hidden layer is input into the second classifier. The classification space of the second classifier represents the sum of the output powers of all photovoltaic power generation groups in the cluster corresponding to the next time interval. During the training of the power generation prediction model, the maximum number of iterations of the power generation prediction model is set, and the difference between the predicted value output by the power generation prediction model and the sample label of the training sample in each iteration is specified as the loss function.

Images