Photovoltaic maximum output power prediction method and system, electronic equipment and storage medium

Abstract

The invention relates to a photovoltaic maximum output power prediction method, and the method comprises the steps: obtaining the weather data, the operation data of a photovoltaic power generation system, and the parameters of a photovoltaic module, and obtaining the photovoltaic full-aperture maximum output based on the operation data and the parameters of the photovoltaic module, and combining a quantile regression algorithm, a bidirectional time convolution network, a bidirectional gating circulation network and an attention mechanism to construct a maximum output prediction model, and inputting weather data and photovoltaic full-aperture maximum output into the trained maximum output prediction model to obtain a prediction result of the maximum output power of the photovoltaic scheduling day. According to the invention, the problem of low prediction precision of the photovoltaic maximum output power is solved. Based on photovoltaic full-aperture maximum output predicted by a mechanism model, a maximum output prediction model constructed by combining a quantile regression algorithm, a bidirectional time convolution network, a bidirectional gating circulation network and an attention mechanism is combined, and finally accurate prediction of photovoltaic maximum output during scheduling is realized.

Description

Technical Field

[0001] The present application relates to the field of output power prediction, and in particular to a photovoltaic maximum output power prediction method, system, electronic device and storage medium. Background Art

[0002] With the advancement of renewable energy policies and the continuous progress of technology, photovoltaics have penetrated into many households and businesses. Compared with traditional power sources, photovoltaic output is more complexly affected by external factors of the system and shows obvious intermittency and randomness.

[0003] In the relevant technologies, photovoltaic output prediction is mainly divided into physical model prediction method and data-driven prediction method based on the model principle. The mainstream models currently used by artificial intelligence, such as BP neural network and LSTM neural network, have problems such as slow learning speed, sensitivity to initialization and adjustment parameters, and cannot fully capture the connection between features, resulting in low prediction accuracy.

[0004] Currently, no effective solution has been proposed for the problem of low prediction accuracy of photovoltaic maximum output power in related technologies. Summary of the invention

[0005] The embodiments of the present application provide a photovoltaic maximum output power prediction method, system, electronic device and storage medium to solve the problem of low photovoltaic maximum output power prediction accuracy in related technologies.

[0006] In a first aspect, an embodiment of the present application provides a photovoltaic maximum output power prediction method, the method comprising: Obtain weather data, as well as photovoltaic power generation system operation data and photovoltaic module parameters; Based on the operating data and the photovoltaic module parameters, obtaining the photovoltaic full-caliber maximum output; The maximum output prediction model is constructed by combining the quantile regression algorithm, bidirectional temporal convolutional network, bidirectional gated recurrent network and attention mechanism; The weather data and the photovoltaic full-caliber maximum output are input into the trained maximum output prediction model to obtain the prediction result of the maximum output power of the photovoltaic scheduling day.

[0007] In some embodiments, the photovoltaic module parameters include the rated capacity, incident light intensity and photovoltaic array surface temperature of the photovoltaic array under standard test conditions, and the operating data include the current surface temperature of the photovoltaic array and the received solar radiation intensity; the full-aperture maximum photovoltaic output obtained based on the operating data and the photovoltaic module parameters includes: Based on a preset ideal photovoltaic maximum output model, the photovoltaic full-aperture maximum output is determined according to the rated capacity, incident light intensity and photovoltaic array surface temperature under the standard test conditions, as well as the surface temperature and solar radiation intensity of the current photovoltaic array, wherein the photovoltaic full-aperture maximum output represents the maximum power that the photovoltaic power generation system can generate under ideal conditions.

[0008] In some embodiments, the preset ideal photovoltaic maximum output model includes:

[0009] Among them, p pv Indicates the maximum output of the photovoltaic full-caliber, Y pv Indicates the rated capacity of the photovoltaic array under standard test conditions, in KW; f pv Represents the photovoltaic derating factor, G T Indicates the solar radiation intensity currently received by the photovoltaic array, in KW / ㎡; G T,STC It indicates the incident light intensity under standard test conditions, in KW / ㎡; S pv Represents the photovoltaic panel area; α p Indicates the temperature correction coefficient; T c Indicates the current surface temperature of the photovoltaic array, in °C; T c，STC Indicates the surface temperature of the photovoltaic array under standard test conditions, in °C.

[0010] In some embodiments, the weather data and the photovoltaic full-caliber maximum output are input into the trained maximum output prediction model to obtain the prediction result of the maximum output power of the photovoltaic scheduling day, including: Arranging and preprocessing the weather data to obtain daily maximum temperature, daily minimum temperature, weather characteristics and seasonal characteristics; Encoding the weather characteristics and the season characteristics of each day to obtain weather labels and season labels; The weather label, the season label, the daily maximum temperature, the daily minimum temperature and the photovoltaic full-caliber maximum output are input into the trained maximum output prediction model to obtain the prediction result of the maximum output power of the photovoltaic scheduling day.

[0011] In some embodiments, encoding the weather characteristics of each day to obtain a weather tag includes: Based on the weather data, determining weather characteristics involved in each day from the weather characteristics; The weather labels of the weather features involved in one day are set to 1, and the weather labels of other weather features are set to 0, wherein the other weather features represent all weather features except the weather features involved.

[0012] In some embodiments, the maximum output prediction model includes: a bidirectional temporal convolution layer, a bidirectional gated recurrent layer, a self-attention layer, and a quantile regression layer. The bidirectional temporal convolution layer is used to capture forward hidden features and reverse hidden features according to the forward sequence and reverse sequence of the input data, and fuse the forward hidden features and the reverse hidden features to obtain fused features; The bidirectional gated recurrent layer is used to capture long-term dependencies and local patterns in the input data; The self-attention layer is used to filter out unimportant information in the input data and dynamically weight the filtered input data; The quantile regression layer is used to perform regression prediction based on the quantile regression algorithm.

[0013] In some embodiments, the weather data and the photovoltaic full-caliber maximum output are input into the trained maximum output prediction model to obtain the prediction result of the maximum output power of the photovoltaic scheduling day, including: Input the weather data and the photovoltaic full-caliber maximum output into the trained maximum output prediction model, the photovoltaic scheduling day maximum output prediction value and upper and lower ranges; A regression graph is drawn according to the predicted value and the true value of the maximum output on the scheduling day, and an error index of the prediction result is determined based on the regression graph.

[0014] In a second aspect, an embodiment of the present application provides a photovoltaic maximum output power prediction system, the system comprising: a data acquisition module, an analysis module, a model building module and a prediction module, wherein: The data acquisition module is used to acquire weather data, operation data of the photovoltaic power generation system and photovoltaic module parameters; The analysis module is used to obtain the maximum photovoltaic full-caliber output based on the operating data and the photovoltaic component parameters; The model building module is used to build a maximum output prediction model by combining a quantile regression algorithm, a bidirectional temporal convolutional network, a bidirectional gated recurrent network, and an attention mechanism; The prediction module is used to input the weather data and the photovoltaic full-caliber maximum output into the trained maximum output prediction model to obtain the prediction result of the maximum output power of the photovoltaic scheduling day.

[0015] In a third aspect, an embodiment of the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the photovoltaic maximum output power prediction method as described in the first aspect above is implemented.

[0016] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the photovoltaic maximum output power prediction method as described in the first aspect above.

[0017] Compared with the related art, the photovoltaic maximum output power prediction method provided in the embodiment of the present application obtains weather data and the operating data and photovoltaic module parameters of the photovoltaic power generation system, obtains the photovoltaic full-caliber maximum output based on the operating data and photovoltaic module parameters, and constructs a maximum output prediction model by combining the quantile regression algorithm, bidirectional time convolution network, bidirectional gated recurrent network and attention mechanism. The weather data and the photovoltaic full-caliber maximum output are input into the trained maximum output prediction model to obtain the prediction result of the maximum output power of the photovoltaic scheduling day, solving the problem of low accuracy of the photovoltaic maximum output power prediction. Based on the full-caliber maximum output of photovoltaic predicted by the mechanism model, the maximum output prediction model constructed by combining the quantile regression algorithm, bidirectional time convolution network, bidirectional gated recurrent network and attention mechanism finally realizes the accurate prediction of the maximum output of photovoltaic on the scheduling day. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 is a flow chart of a photovoltaic maximum output power prediction method according to an embodiment of the present application; Figure 2 is a structural diagram of a bidirectional temporal convolutional network according to an embodiment of the present application; Figure 3 is a structural diagram of a bidirectional gated recurrent network according to an embodiment of the present application; Figure 4 is a flow chart of a photovoltaic maximum output power prediction method according to an embodiment of the present application; Figure 5 is a schematic diagram of a training set interval prediction result according to an embodiment of the present application; Figure 6 is a schematic diagram of a test set interval prediction result according to an embodiment of the present application; Figure 7 is a prediction result regression diagram according to an embodiment of the present application; Figure 8 is a structural block diagram of a photovoltaic maximum output power prediction system according to an embodiment of the present application; Fig. 9 It is a schematic diagram of the internal structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0020] Obviously, the drawings described below are only some examples or embodiments of the present application. For ordinary technicians in this field, the present application can also be applied to other similar scenarios based on these drawings without creative work. In addition, it can also be understood that although the efforts made in this development process may be complicated and lengthy, for ordinary technicians in this field related to the content disclosed in this application, some changes in design, manufacturing or production based on the technical content disclosed in this application are just conventional technical means, and should not be understood as insufficient content disclosed in this application.

[0021] Reference to "embodiments" in this application means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those of ordinary skill in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0022] Unless otherwise defined, the technical terms or scientific terms involved in this application should be understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the" and the like involved in this application do not indicate a quantity limitation, and may indicate the singular or plural. The terms "include", "comprise", "have" and any of their variations involved in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there may be three relationships, for example, "A and / or B" can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects before and after are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0023] This embodiment provides a method for predicting maximum photovoltaic output power. Figure 1 is a flow chart of a photovoltaic maximum output power prediction method according to an embodiment of the present application, such as Figure 1 As shown, the process includes the following steps: Step S101, obtaining weather data, operation data of the photovoltaic power generation system and photovoltaic module parameters.

[0024] Weather data includes but is not limited to climate characteristics, temperature and solar radiation intensity corresponding to different seasons and dates. PV module parameters include but are not limited to PV rated capacity, PV derating factor and temperature correction factor. The operation data of the PV power generation system and PV module parameters collected from weather data are preprocessed, including but not limited to removing redundant data, identifying abnormal data, filling missing values ​​and data integration.

[0025] Step S102, based on the operation data and the photovoltaic module parameters, the maximum photovoltaic output of the entire aperture is obtained.

[0026] In some embodiments, the photovoltaic module parameters include the rated capacity of the photovoltaic array under standard test conditions, the incident light intensity under standard conditions, and the surface temperature of the photovoltaic array under standard test conditions, and the operation data include the current surface temperature of the photovoltaic array and the solar radiation intensity received by the photovoltaic array; step S102 includes: Based on the preset ideal photovoltaic maximum output model, the full-aperture maximum photovoltaic output is determined according to the rated capacity, incident light intensity and photovoltaic array surface temperature under standard test conditions, as well as the current photovoltaic array surface temperature and solar radiation intensity. The full-aperture maximum photovoltaic output represents the maximum power that the photovoltaic power generation system can generate under ideal conditions.

[0027] In order to predict the maximum output of photovoltaic power on the day of dispatch, it is necessary to first calculate the maximum output of photovoltaic power. The maximum output of photovoltaic power refers to the maximum power that the photovoltaic power generation system can generate under ideal conditions. Photovoltaic power generation is affected by many factors, including weather, equipment parameters, installation conditions, etc. A preset ideal photovoltaic maximum output model is established based on the light energy transfer of sunlight and the internal mechanism of the photovoltaic system:

[0028] Among them, p pv Indicates the maximum output of photovoltaic power. pv Indicates the rated capacity of the photovoltaic array under standard test conditions, in KW; f pv Represents the photovoltaic derating factor, G T Indicates the solar radiation intensity currently received by the photovoltaic array, in KW / ㎡; G T,STC It indicates the incident light intensity under standard test conditions, in KW / ㎡; S pv Represents the photovoltaic panel area; α p Indicates the temperature correction coefficient; T c Indicates the current surface temperature of the photovoltaic array, in °C; T c，STC Indicates the surface temperature of the photovoltaic array under standard test conditions, in °C.

[0029] Step S103, combining the quantile regression algorithm, the bidirectional temporal convolutional network, the bidirectional gated recurrent network and the attention mechanism to build a maximum output prediction model.

[0030] In some of the embodiments, the maximum output prediction model includes: a bidirectional temporal convolution layer, a bidirectional gated recurrent layer, a self-attention layer, and a quantile regression layer.

[0031] Combining the advantages of quantile regression (QR), bidirectional temporal convolutional network (BiTCN), bidirectional gated recurrent network (BiGRU) and attention mechanism (Attention), an interval prediction model (maximum output prediction model) based on QRBiTCN-BiGRU-Attention is constructed.

[0032] The bidirectional temporal convolution layer is used to capture the forward hidden features and the reverse hidden features according to the forward sequence and the reverse sequence of the input data, and fuse the forward hidden features and the reverse hidden features to obtain the fused features.

[0033] Traditional temporal convolutional neural networks (TCNs) only perform forward convolution calculations on the input sequence to extract forward data features, while ignoring implicit information in the reverse direction. In order to improve the feature fusion effect of the TCN model on the bidirectional time series information of photovoltaic output data, based on the structural foundation of the TCN model, taking into account the high-dimensional and bidirectional time series correlation characteristics of photovoltaic output data, TCN is constructed into a forward network and a reverse network, namely a bidirectional temporal convolutional neural network (BiTCN). Figure 2 It is a structural diagram of a bidirectional temporal convolutional network according to an embodiment of the present application. After each layer of convolution of the input sequence, the dilation factor grows exponentially, and BiTCN obtains a larger receptive field. The increase in the receptive field of BiTCN brings about problems such as gradient vanishing and slow convergence. The residual block is introduced to avoid these problems while achieving efficient feature extraction of the sequence. Therefore, a bidirectional temporal convolution structure BiTCN is used to capture hidden features in the forward and backward directions to better obtain the long-term dependency of the sequence.

[0034] Bidirectional gated recurrent layers to capture long-term dependencies and local patterns in the input data.

[0035] Bidirectional Gated Recurrent Network (BiGRU) combines the advantages of recurrent neural networks and gating mechanisms, and can effectively capture long-term dependencies and local patterns in sequence data. Figure 3 is a structural diagram of a bidirectional gated recurrent network according to an embodiment of the present application, such as Figure 3 As shown in the figure, BiGRU contains two-directional GRU units, which are used for the forward and reverse data sequences processed by BiTCN respectively. It connects the GRU layers in the forward and reverse directions, one processes the input sequence from front to back (forward), and the other processes it in reverse. In this way, BiGRU can more completely explore the dependency of features and obtain contextual associations.

[0036] The self-attention layer is used to filter out unimportant information in the input data and dynamically weight the filtered input data.

[0037] The self-attention mechanism is added to the constructed BiTCN-BiGRU model. As an important variant of the attention mechanism, the self-attention model has the unique ability to explore the comprehensive correlation between input and output, and can also identify the interaction between elements that are far apart in the sequence. After adding the self-attention mechanism to the BiTCN-BiGRU layer, the global information feature extraction ability of the attention mechanism at different positions before and after the BiTCN-BiGRU is explored, unimportant information is filtered out, and the focus is on the most important part of the sequence, which improves the efficiency and accuracy of model training.

[0038] In order to improve the computing power of the model, preferably, the self-attention mechanism adopts three parameters, namely, the query vector Q (query), the key vector K (key), and the value vector V (value), where the key vector K and the value vector V are one-to-one corresponding, that is, the relationship between key-value pairs. The key vector K and the value vector V perform a dot product operation to obtain the correlation between the input and the output, and the value vector V is used to calculate the output based on the correlation between the input and the key. The core formula of the self-attention mechanism is:

[0039] Where X represents the input sequence, W Q Represents the weight matrix of the query vector, W K represents the weight matrix of the key vector, W V Represents the weight matrix of the value vector, d k The dimension of the representation vector. Both the query vector Q and the key vector K are obtained by multiplying the input sequence X by the corresponding matrix weights, which means that the representation of each position will be used to calculate its own attention weight and the attention weights of other positions at the same time.

[0040] The quantile regression layer is used for regression prediction based on the quantile regression algorithm.

[0041] Traditional interval prediction is mostly implemented by probability density estimation, which often requires estimation error and assumption of error distribution. In this embodiment, the quantile regression method is used, and the weighted absolute deviation is used as the objective function to be optimized, without giving the distribution function of the random variable. Quantile regression is used for regression prediction, and the prediction result and its upper and lower interval range are obtained according to the confidence level. The interval prediction is plotted and the error index is calculated.

[0042] The conditional quantile regression model can be expressed as:

[0043] Where x is a p+1 dimensional vector, Indicates the sample condition of y under the explanatory variable x Quantile, It can be obtained by optimizing the following objective function:

[0044] In the above formula, , For The regression coefficient at the quantile can also be written as:

[0045] is the tilt absolute value function:

[0046] Among them, u can be expressed as The above formula can be further rewritten as

[0047] Step S104, inputting weather data and the full-caliber maximum output of photovoltaic power into the trained maximum output prediction model to obtain the prediction result of the maximum output power of photovoltaic power scheduling day.

[0048] In some embodiments, step S104 specifically includes: Step S1041, sorting and preprocessing the weather data to obtain the daily maximum temperature, daily minimum temperature, weather characteristics and seasonal characteristics.

[0049] It should be noted that weather characteristics include but are not limited to sunny, cloudy, overcast, light rain, sunny to cloudy, light rain to moderate rain, and overcast to light rain.

[0050] Step S1042, encode the weather characteristics of each day and the seasonal characteristics to obtain weather labels and season labels.

[0051] In some embodiments, step S1042 specifically includes: Step S201, based on weather data, determining weather characteristics involved in each day from weather characteristics.

[0052] Step S202, setting the weather tags of the weather features involved in a day to 1, and setting the weather tags of other weather features to 0, wherein other weather features represent all weather features except the weather features involved.

[0053] Considering the obvious differences in light duration and light intensity in different seasons, it is necessary to recode the time variables by quarter and convert text data into numerical data.

[0054] On the other hand, since weather indicators are relatively complex, such as sunny to cloudy, light rain to moderate rain, and cloudy to light rain, multiple weather characteristics are involved, so the weather characteristics need to be preprocessed: The weather characteristics involved are sorted out separately, including but not limited to sunny, cloudy, overcast, light rain, and heavy rain, and then the indicators are coded according to the daily weather forecast. For example, if the weather is sunny on a certain day, the indicator sunny is 1, and other weather indicators are 0; if the weather is sunny and then cloudy, the indicator sunny is 1, cloudy is also 1, and other weather indicators are 0. That is, the main weather characteristics of the day are set to 1, and other weather characteristics are set to 0 to improve the efficiency of data processing.

[0055] Step S1043, input the weather label, season label, daily maximum temperature, daily minimum temperature and photovoltaic full-caliber maximum output into the trained maximum output prediction model to obtain the prediction result of the maximum output power of the photovoltaic scheduling day.

[0056] In some embodiments, weather data and the full-caliber maximum output of photovoltaic power are input into the trained maximum output prediction model to obtain the prediction results of the maximum output power of photovoltaic scheduling day, including: The weather data and the full-caliber maximum output of photovoltaic power are input into the trained maximum output prediction model to obtain the maximum output power prediction value and upper and lower ranges of the photovoltaic scheduling day.

[0057] A regression graph is drawn based on the predicted value and true value of the maximum output on the dispatching day, and the error index of the prediction result is determined based on the regression graph.

[0058] Figure 4 is a flow chart of a photovoltaic maximum output power prediction method according to an embodiment of the present application, such as Figure 4 As shown, data is collected, including meteorological data, equipment parameters and time information; based on the mechanism model, the full-caliber maximum output power of photovoltaic power is obtained according to meteorological data and equipment parameters, and the meteorological data and time data are classified and derived to obtain meteorological characteristics and time characteristics; after normalizing the full-caliber maximum output power of photovoltaic power, meteorological characteristics and time characteristics, they are input into the maximum output prediction model to obtain the prediction result of the maximum output on the dispatching day. Among them, the model input parameters mainly include season, minimum temperature, maximum temperature, weather mark, full-caliber maximum output data, and the prediction parameter is the maximum output on the dispatching day.

[0059] In order to verify the accuracy of the prediction results based on the maximum output prediction model (QRBiTCN-BiGRU-Attention model), the sample data obtained above was used to divide the training set and the test set into 8:2, and 80% of the sample data was randomly selected and input into the model to obtain the prediction results. Figure 5 is a schematic diagram of a training set interval prediction result according to an embodiment of the present application, Figure 6 This is a schematic diagram of the test set interval prediction result according to the embodiment of the present application. From the training set prediction result diagram and the test set prediction result diagram, it can be seen that the error between the predicted value of the maximum output of the photovoltaic scheduling day output by the maximum output prediction model and the actual value is small, and the data change trend can be accurately tracked. The result is subjected to regression prediction, Figure 7 This is a regression graph of the prediction results according to an embodiment of the present application. It can be seen from the regression graph that the data fitting effect is good and the R value can reach 0.99.

[0060] Through the above steps, weather data, operation data of photovoltaic power generation system and parameters of photovoltaic modules are obtained. Based on the operation data and parameters of photovoltaic modules, the maximum output of photovoltaic power generation is obtained. The maximum output prediction model is constructed by combining quantile regression algorithm, bidirectional temporal convolutional network, bidirectional gated recurrent network and attention mechanism. The weather data and the maximum output of photovoltaic power generation are input into the trained maximum output prediction model to obtain the prediction result of the maximum output power of photovoltaic power generation on the day of scheduling, which solves the problem of low prediction accuracy of photovoltaic power generation on the day of scheduling. Based on the prediction of the maximum output of photovoltaic power generation on the day of scheduling, the maximum output of photovoltaic power generation on the day of scheduling is finally realized by combining the maximum output prediction model constructed by quantile regression algorithm, bidirectional temporal convolutional network, bidirectional gated recurrent network and attention mechanism.

[0061] The present invention classifies and derives time data and weather characteristics, combines the full-caliber maximum output of photovoltaic power predicted by the mechanism model and the QRBiTCN-BiGRU-Attention model, and finally realizes the accurate prediction of the maximum output of photovoltaic power on the dispatching day. Through example verification, the prediction results of the above hybrid model are compared with the single LSTM data model, QRBiTCN hybrid data model, and QRBiTCN-BiGRU hybrid data model. The RMSE is reduced by 80%, 69% and 10% respectively, and the prediction accuracy is greatly improved.

[0062] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0063] This embodiment also provides a photovoltaic maximum output power prediction system, which is used to implement the above embodiments and preferred implementations, and will not be repeated here. As used below, the terms "module", "unit", "subunit", etc. can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0064] Figure 8 is a structural block diagram of a photovoltaic maximum output power prediction system according to an embodiment of the present application, such as Figure 8 As shown, the system includes: a data acquisition module 81, an analysis module 82, a model building module 83 and a prediction module 84.

[0065] The data acquisition module 81 is used to acquire weather data, operating data of the photovoltaic power generation system and photovoltaic component parameters.

[0066] The analysis module 82 is used to obtain the maximum photovoltaic full-caliber output based on the operating data and photovoltaic component parameters.

[0067] The model building module 83 is used to build a maximum output prediction model by combining a quantile regression algorithm, a bidirectional temporal convolutional network, a bidirectional gated recurrent network, and an attention mechanism.

[0068] The prediction module 84 is used to input the weather data and the full-caliber maximum output of photovoltaic power into the trained maximum output prediction model to obtain the prediction result of the maximum output power of photovoltaic power scheduling day.

[0069] In some of the embodiments, the photovoltaic component parameters include the rated capacity of the photovoltaic array under standard test conditions, the incident light intensity and the surface temperature of the photovoltaic array, and the operating data include the current surface temperature of the photovoltaic array and the received solar radiation intensity; the analysis module 82 includes: a full-caliber maximum output module.

[0070] The full-aperture maximum output module is used to determine the full-aperture maximum output of photovoltaics based on a preset ideal photovoltaic maximum output model, according to the rated capacity, incident light intensity and photovoltaic array surface temperature under standard test conditions, as well as the current photovoltaic array surface temperature and solar radiation intensity. The full-aperture maximum output of photovoltaics represents the maximum power that a photovoltaic power generation system can generate under ideal conditions.

[0071] In some embodiments, the preset ideal photovoltaic maximum output model includes:

[0072] Among them, p pv Indicates the maximum output of photovoltaic power. pv Indicates the rated capacity of the photovoltaic array under standard test conditions, in KW; f pv Represents the photovoltaic derating factor, G T Indicates the solar radiation intensity currently received by the photovoltaic array, in KW / ㎡; G T,STC It indicates the incident light intensity under standard test conditions, in KW / ㎡; S pv Represents the photovoltaic panel area; α p Indicates the temperature correction coefficient; T c Indicates the current surface temperature of the photovoltaic array, in °C; T c，STC Indicates the surface temperature of the photovoltaic array under standard test conditions, in °C.

[0073] In some of the embodiments, the prediction module 84 includes: a pre-processing module, a coding module and an output prediction module.

[0074] The preprocessing module is used to organize and preprocess the weather data to obtain the daily maximum temperature, daily minimum temperature, weather characteristics and seasonal characteristics.

[0075] The encoding module is used to encode the weather characteristics and seasonal characteristics of each day to obtain weather labels and seasonal labels; The output prediction module is used to input the weather label, season label, daily maximum temperature, daily minimum temperature and photovoltaic full-caliber maximum output into the trained maximum output prediction model to obtain the prediction result of the maximum output power of the photovoltaic scheduling day.

[0076] In some of the embodiments, the encoding module includes: a feature extraction module and a label setting module.

[0077] A feature extraction module is used to determine the weather features involved in each day from the weather features based on the weather data; The label setting module is used to set the weather labels of the weather features involved in a day to 1, and set the weather labels of other weather features to 0, wherein other weather features represent all weather features except the weather features involved.

[0078] In some of the embodiments, the maximum output prediction model includes: a bidirectional temporal convolution layer, a bidirectional gated recurrent layer, a self-attention layer, and a quantile regression layer.

[0079] The bidirectional temporal convolution layer is used to capture the forward hidden features and the reverse hidden features according to the forward sequence and the reverse sequence of the input data, and fuse the forward hidden features and the reverse hidden features to obtain the fused features; Bidirectional gated recurrent layers to capture long-term dependencies and local patterns in the input data; The self-attention layer is used to filter out unimportant information in the input data and dynamically weight the filtered input data; The quantile regression layer is used for regression prediction based on the quantile regression algorithm.

[0080] In some embodiments, the prediction module 84 includes: a range determination module and an error calculation module.

[0081] The range determination module is used to input weather data and the maximum output of photovoltaic power into the trained maximum output prediction model to obtain the maximum output power prediction value and upper and lower ranges of photovoltaic scheduling day; The error calculation module is used to draw a regression graph based on the predicted value and true value of the maximum output on the scheduling day, and determine the error index of the prediction result based on the regression graph.

[0082] Through the above system, the data acquisition module 81 obtains weather data, as well as the operating data and photovoltaic component parameters of the photovoltaic power generation system. The analysis module 82 obtains the photovoltaic full-caliber maximum output based on the operating data and photovoltaic component parameters. The model construction module 83 combines the quantile regression algorithm, the bidirectional time convolution network, the bidirectional gated recurrent network and the attention mechanism to construct a maximum output prediction model. The prediction module 84 inputs the weather data and the photovoltaic full-caliber maximum output into the trained maximum output prediction model to obtain the prediction result of the maximum output power of the photovoltaic scheduling day, solving the problem of low prediction accuracy of the photovoltaic maximum output power. Based on the full-caliber maximum output of the photovoltaic predicted by the mechanism model, the maximum output prediction model constructed by combining the quantile regression algorithm, the bidirectional time convolution network, the bidirectional gated recurrent network and the attention mechanism finally realizes the accurate prediction of the maximum output of the photovoltaic on the scheduling day.

[0083] It should be noted that the above modules can be functional modules or program modules, and can be implemented by software or hardware. For modules implemented by hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0084] This embodiment further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0085] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0086] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program: S1, obtain weather data, as well as operation data of the photovoltaic power generation system and photovoltaic component parameters.

[0087] S2, based on the operating data and PV module parameters, obtains the maximum PV output of the entire system.

[0088] S3, combines the quantile regression algorithm, bidirectional temporal convolutional network, bidirectional gated recurrent network and attention mechanism to build a maximum output prediction model.

[0089] S4, inputting weather data and the full-caliber maximum output of photovoltaic power into the trained maximum output prediction model to obtain the prediction result of the maximum output power of photovoltaic power dispatching day.

[0090] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0091] In one embodiment, Fig. 9 is a schematic diagram of the internal structure of an electronic device according to an embodiment of the present application, such as Fig. 9 As shown, an electronic device is provided, which may be a server, and its internal structure diagram may be as shown in Fig. 9 As shown. The electronic device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the electronic device is used to store data. The network interface of the electronic device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a photovoltaic maximum output power prediction method is implemented.

[0092] Those skilled in the art will understand that Fig. 9 The structure shown in the figure is merely a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the electronic device to which the scheme of the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0093] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0094] Those skilled in the art should understand that the technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0095] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A photovoltaic maximum output power prediction method, characterized in that: The method comprises: Obtain weather data, as well as photovoltaic power generation system operation data and photovoltaic module parameters; Based on the operating data and the photovoltaic module parameters, obtaining the photovoltaic full-caliber maximum output; The maximum output prediction model is constructed by combining the quantile regression algorithm, bidirectional temporal convolutional network, bidirectional gated recurrent network and attention mechanism; The weather data and the photovoltaic full-caliber maximum output are input into the trained maximum output prediction model to obtain the prediction result of the maximum output power of the photovoltaic scheduling day.

2. The method according to claim 1, characterized in that: The photovoltaic module parameters include the rated capacity, incident light intensity and photovoltaic array surface temperature of the photovoltaic array under standard test conditions, and the operating data include the current surface temperature of the photovoltaic array and the received solar radiation intensity; the photovoltaic full-caliber maximum output obtained based on the operating data and the photovoltaic module parameters includes: Based on a preset ideal photovoltaic maximum output model, the photovoltaic full-aperture maximum output is determined according to the rated capacity, incident light intensity and photovoltaic array surface temperature under the standard test conditions, as well as the surface temperature and solar radiation intensity of the current photovoltaic array, wherein the photovoltaic full-aperture maximum output represents the maximum power that the photovoltaic power generation system can generate under ideal conditions.

3. The method according to claim 2, characterized in that The preset ideal photovoltaic maximum output model includes: Among them, p pv Indicates the maximum output of the photovoltaic full-caliber, Y pv Indicates the rated capacity of the photovoltaic array under standard test conditions, in KW; f pv Represents the photovoltaic derating factor; G T Indicates the solar radiation intensity currently received by the photovoltaic array, in KW / ㎡; G T,STC It indicates the incident light intensity under standard test conditions, in KW / ㎡; S pv Represents the photovoltaic panel area; α p Indicates the temperature correction coefficient; T c Indicates the current surface temperature of the photovoltaic array, in °C; T c，STC Indicates the surface temperature of the photovoltaic array under standard test conditions, in °C.

4. The method according to claim 1, characterized in that: The step of inputting the weather data and the photovoltaic full-caliber maximum output into the trained maximum output prediction model to obtain the prediction result of the maximum output power of the photovoltaic scheduling day includes: Arranging and preprocessing the weather data to obtain daily maximum temperature, daily minimum temperature, weather characteristics and seasonal characteristics; Encoding the weather characteristics and the season characteristics of each day to obtain weather labels and season labels; The weather label, the season label, the daily maximum temperature, the daily minimum temperature and the photovoltaic full-caliber maximum output are input into the trained maximum output prediction model to obtain the prediction result of the maximum output power of the photovoltaic scheduling day.

5. The method according to claim 4, characterized in that The weather characteristics of each day are encoded to obtain weather labels including: Based on the weather data, determining weather characteristics involved in each day from the weather characteristics; The weather labels of the weather features involved in one day are all set to 1, and the weather labels of other weather features are set to 0, wherein the other weather features represent all weather features except the weather features involved.

6. The method according to claim 1, characterized in that The maximum output prediction model includes: a bidirectional temporal convolution layer, a bidirectional gated recurrent layer, a self-attention layer and a quantile regression layer. The bidirectional temporal convolution layer is used to capture forward hidden features and reverse hidden features according to the forward sequence and reverse sequence of the input data, and fuse the forward hidden features and the reverse hidden features to obtain fused features; The bidirectional gated recurrent layer is used to capture long-term dependencies and local patterns in the input data; The self-attention layer is used to filter out unimportant information in the input data and dynamically weight the filtered input data; The quantile regression layer is used to perform regression prediction based on the quantile regression algorithm.

7. The method according to claim 1, characterized in that The step of inputting the weather data and the photovoltaic full-caliber maximum output into the trained maximum output prediction model to obtain the prediction result of the maximum output power of the photovoltaic scheduling day includes: Input the weather data and the photovoltaic full-caliber maximum output into the trained maximum output prediction model to obtain the maximum output prediction value and upper and lower ranges of the photovoltaic scheduling day; A regression graph is drawn according to the predicted value and the actual value of the maximum output on the scheduling day, and an error index of the prediction result is determined based on the regression graph.

8. A photovoltaic maximum output power prediction system, characterized in that: The system includes: a data acquisition module, an analysis module, a model building module and a prediction module, wherein: The data acquisition module is used to acquire weather data, operation data of the photovoltaic power generation system and photovoltaic module parameters; The analysis module is used to obtain the maximum photovoltaic full-caliber output based on the operating data and the photovoltaic component parameters; The model building module is used to build a maximum output prediction model by combining a quantile regression algorithm, a bidirectional temporal convolutional network, a bidirectional gated recurrent network, and an attention mechanism; The prediction module is used to input the weather data and the photovoltaic full-caliber maximum output into the trained maximum output prediction model to obtain the prediction result of the maximum output power of the photovoltaic scheduling day.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the photovoltaic maximum output power prediction method according to any one of claims 1 to 7 is implemented.

10. A storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the photovoltaic maximum output power prediction method according to any one of claims 1 to 7 is implemented.

Images