Photovoltaic power prediction method and device, terminal equipment and storage medium

Abstract

The invention discloses a photovoltaic power prediction method and device, terminal equipment and a storage medium, and belongs to the field of power prediction, and the method comprises the steps: obtaining a plurality of time delay vectors of a power system, a plurality of photovoltaic power sequences and meteorological factor data corresponding to the photovoltaic power sequences; for each time delay vector, determining a time delay embedded data set; aiming at each time delay embedded data set, taking meteorological factor data adjusted by each time delay as a position coordinate of each node, taking a corresponding photovoltaic power sequence as a numerical value label of the corresponding node, determining a plurality of nodes, and forming a local graph; calculating a sorting rank vector of each node in the local graph, and calculating global consistency measurement; and performing data alignment on the plurality of photovoltaic power sequences based on each global consistency measurement to obtain a plurality of aligned photovoltaic power sequences, and then performing photovoltaic power prediction, so that the problem of data deviation in photovoltaic power prediction in the prior art can be solved.

Description

Technical Field

[0001] This invention relates to the field of power prediction, and more particularly to a photovoltaic power prediction method, apparatus, terminal equipment, and storage medium. Background Technology

[0002] Short-term photovoltaic (PV) power forecasting has long been a research hotspot in the PV power generation field. It not only involves the safe and stable operation of PV power systems but is also closely related to the allocation and storage of clean energy, the operational efficiency of the electricity market, the formulation of energy policies, and environmental protection. Traditional short-term PV power forecasting methods focus on the time-series characteristics of historical power data and the mapping relationship between power and factors to build models, such as time series analysis and regression analysis. While these methods can meet the needs to a certain extent, their prediction accuracy and adaptability often fall short of the requirements of modern power systems when faced with massive sample sets and multi-step forecasting.

[0003] In recent years, with the gradual maturation of technologies such as deep learning, short-term photovoltaic power forecasting has become more accurate in capturing the dynamic changes in electricity consumption and better able to cope with various complex situations. However, these methods have not carefully considered the sample alignment problem. For example, due to the electricity consumption collection process, sensor delays, and changes in the meteorological acquisition environment, there is a time-axis alignment deviation between power data and influencing factors. This situation can seriously disrupt the stable quantitative relationship between power data and influencing factors, thus affecting the overall reliability of the model and the accuracy of the prediction results. Therefore, there is an urgent need for a short-term photovoltaic power forecasting method that can solve the above problems. Summary of the Invention

[0004] This invention provides a photovoltaic power prediction method, apparatus, terminal equipment, and storage medium, which can solve the problem of data alignment deviation in photovoltaic power prediction in the prior art.

[0005] The photovoltaic power prediction method provided by this invention includes: Acquire several time delay vectors of the power system, several photovoltaic power sequences, and meteorological factor data corresponding to each photovoltaic power sequence; For each of the aforementioned time delay vectors, based on each time delay value of the time delay vector, the meteorological factor data corresponding to the time delay value is adjusted for delay. Based on each adjusted meteorological factor data and the corresponding photovoltaic power sequence, a time delay embedding dataset is determined; wherein, each time delay vector corresponds to a time delay embedding dataset. For each time-delay embedded dataset, the time-delay adjusted meteorological factor data is used as the location coordinates of each node, and the photovoltaic power sequence corresponding to the time-delay adjusted meteorological factor data is used as the numerical label of the corresponding node. Several nodes are determined, and a local graph is formed based on the several nodes. For each local graph, calculate the sorted rank vector of each node in the local graph, and calculate the global consistency measure based on the sorted rank vector of each node. Based on the global consistency metric corresponding to each time-delay embedded dataset, several photovoltaic power sequences are aligned to obtain several aligned photovoltaic power sequences, and then photovoltaic power prediction is performed based on each aligned photovoltaic power sequence.

[0006] Furthermore, the local graph composed of several nodes includes: Based on the positional distance between each node, determine the edge connecting each pair of nodes; A fully connected graph is determined based on each of the aforementioned nodes and each of the aforementioned edges; Calculate the weight of each side in the fully connected graph and determine the number of neighboring nodes corresponding to each node; Based on several neighboring nodes of each node, determine the local neighborhood to which each node belongs. A local graph is formed based on each of the aforementioned local neighborhoods.

[0007] Further, the sorting rank vector includes: a positional sorting rank vector and a numerical sorting rank vector; calculating the sorting rank vector of each node in the local graph includes: Within each local neighborhood of the local graph, each node is sorted according to its positional distance, and the positional sorting rank vector of each node is determined based on its index. Within each local neighborhood of the local graph, each node is sorted according to its numerical distance, and the numerical sorting rank vector of each node is determined based on its index.

[0008] Furthermore, the calculation of the global consistency metric based on the sorted rank vector of each node includes: Substitute the sorted rank vector of each node into the Spearman correlation coefficient formula to calculate the Spearman correlation coefficient of each node. The global consistency metric is calculated based on the Spearman correlation coefficient of each node.

[0009] Furthermore, the Spearman calculation formula includes: In the formula, Let be the Spearman correlation coefficient of node i. Sort the rank vector of the k-th position of node i. Let the rank vector of the k-th numerical value of node i be sorted. This represents the element difference between the position-sorted rank vector and the numerical-sorted rank vector. This represents the number of nearest neighbor nodes.

[0010] Furthermore, the formula for calculating the global consistency metric includes: In the formula, Represents a global consistency measure. This represents the total number of nodes.

[0011] Furthermore, the step of aligning the photovoltaic power sequences based on the global consistency metric corresponding to each time-delay embedded dataset includes: The optimal global consistency metric is determined based on the global consistency metric corresponding to each latency embedding dataset. Select the time delay vector corresponding to the optimal global consistency metric to determine the time delay value of each photovoltaic power sequence; Based on the time delay values ​​of several photovoltaic power sequences, data alignment is performed on several photovoltaic power sequences to obtain several aligned photovoltaic power sequences.

[0012] Another embodiment of the present invention provides a photovoltaic power prediction device, including: a data acquisition module, a time delay embedding module, a mapping module, a data calculation module, and a result generation module; The data acquisition module is used to acquire several time delay vectors of the power system, several photovoltaic power sequences, and meteorological factor data corresponding to each photovoltaic power sequence. The time delay embedding module is used to perform time delay adjustment on the meteorological factor data corresponding to each time delay value of each time delay vector, based on each time delay value of the time delay vector, and determine the time delay embedding dataset according to each time delay adjusted meteorological factor data and the corresponding photovoltaic power sequence; wherein, each time delay vector corresponds to a time delay embedding dataset. The mapping module is used to determine several nodes for each time-delay embedded dataset, using the time-delay adjusted meteorological factor data as the position coordinates of each node, and using the photovoltaic power sequence corresponding to the time-delay adjusted meteorological factor data as the numerical label of the corresponding node, and to form a local graph based on the several nodes. The data calculation module is used to calculate the sorted rank vector of each node in each local graph, and to calculate the global consistency measure based on the sorted rank vector of each node. The result generation module is used to perform data alignment on several photovoltaic power sequences based on the global consistency metric corresponding to each time-delay embedded dataset, to obtain several aligned photovoltaic power sequences, and then perform photovoltaic power prediction based on each aligned photovoltaic power sequence.

[0013] Another embodiment of the present invention provides a terminal device, including: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, it implements the steps of the photovoltaic power prediction method provided by the present invention.

[0014] Another embodiment of the present invention also provides a computer-readable storage medium item, including: a stored computer program, which, when the computer program is running, controls the device where the computer-readable storage medium is located to perform the steps of the photovoltaic power prediction method provided by the present invention.

[0015] The following benefits can be obtained by implementing the present invention: This invention discloses a photovoltaic (PV) power prediction method. By introducing time delay values ​​into meteorological factor data using several time delay vectors, each delayed meteorological factor data point serves as the position coordinate of each node, and the corresponding PV power sequence is used as the numerical label for that node. Several nodes are constructed to form a local graph. By calculating the global consistency metric of the local graph, the PV power sequence is aligned, resulting in an aligned PV power sequence for PV power prediction. This invention solves the problem of data alignment bias in PV power prediction, thus improving the accuracy of PV power prediction. Attached Figure Description

[0016] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 This is a schematic flowchart of a photovoltaic power prediction method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a photovoltaic power prediction device provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of a partial view provided in an embodiment of the present invention; Figure 4 This is a bar chart showing the improvement effect of each model's MSE index in dataset A provided by an embodiment of the present invention; Figure 5 This is a bar chart showing the improvement effect of each model's MAE index in dataset A provided by an embodiment of the present invention; Figure 6 This is a bar chart showing the improvement effect of each model's RMSE index in dataset A provided by an embodiment of the present invention; Figure 7 This is a comparison chart of the prediction results of the Transformer model provided in an embodiment of the present invention on dataset A before and after alignment; Figure 8 This is a comparison of the prediction results of the LSTM model provided in an embodiment of the present invention on dataset A before and after alignment; Figure 9 This is a comparison chart of the relationship between the prediction step size and the average value of the MSE error index of the Transformer model before and after alignment, provided in an embodiment of the present invention. Figure 10 This is a comparison chart of the relationship between the prediction step size and the average value of the MAE error index of the Transformer model before and after alignment, provided in an embodiment of the present invention. Figure 11 This is a comparison chart of the relationship between the prediction step size and the average RMSE error index of the Transformer model before and after alignment, provided in an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0020] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly defined.

[0021] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0022] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0023] In the description of the embodiments of this application, the term "multiple" refers to two or more (including two), similarly, "multiple sets" refers to two or more (including two sets), and "multiple pieces" refers to two or more (including two pieces).

[0024] In the description of the embodiments of this application, unless otherwise expressly specified and limited, technical terms such as "installation," "connection," "joining," and "fixing" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. For those skilled in the art, the specific meaning of the above terms in the embodiments of this application can be understood according to the specific circumstances.

[0025] See Figure 1 To address the data alignment discrepancy issue in existing photovoltaic power prediction technologies, an embodiment of the present invention provides a photovoltaic power prediction method, comprising: 101. Obtain several time delay vectors of the power system, several photovoltaic power sequences, and meteorological factor data corresponding to each photovoltaic power sequence.

[0026] 102. For each of the aforementioned time delay vectors, based on each time delay value of the time delay vector, the meteorological factor data corresponding to the time delay value is adjusted for delay. Based on each adjusted meteorological factor data and the corresponding photovoltaic power sequence, a time delay embedding dataset is determined; wherein, each time delay vector corresponds to a time delay embedding dataset.

[0027] In a specific embodiment, to address the alignment bias problem in the predicted data, a time-delay embedding model is used to describe the quantitative relationship between future photovoltaic power and historical meteorological factors. The time-delay parameterization is used to describe the common temporal alignment bias between photovoltaic power and related meteorological factors, specifically including: Assuming a stable quantitative relationship exists between future power and historical meteorological factors, a precise mapping from input to output is achieved through data-driven high-dimensional nonlinear function approximation. Describe the quantitative relationship between historical meteorological factors and future power prediction in aligned scenarios, where, Describes the stable quantitative relationship that exists in the time domain between the historical factor sequence and the target sequence. This represents the future power sequence segment, i.e., the target sequence. This represents a fragment of a historical meteorological factor sequence, allowing This defines the input sample.

[0028] The ultimate goal of power prediction is to capture these quantitative relationships, and a common solution for multivariate time series prediction tasks involves training well-designed deep neural networks to approximate functional relationships. That is, minimizing the loss function: In the above formula, This represents the neural network trained based on the aligned samples. Indicates a measure of the gap. This is the parameter set of the neural network. Therefore, it is trained based on the input sample set. The closer it is to the actual functional relationship The higher the accuracy of the final power prediction, the better.

[0029] It should be noted that various external factors cause a certain alignment bias between the factors and the target components in the input sample over the time domain. When this alignment bias exists, the mapping relationship between the factor sequence and the target sequence in the time domain becomes unstable and random, leading to a decrease in prediction accuracy. To quantify this alignment bias, a time-delay embedding model is used to represent the time-domain alignment bias using a time-delay vector. Parameterization. The quantitative relationships after offsetting delay values ​​are embedded into the delay-in-the-model. Characterization, where the time delay vector The parameterized temporal alignment bias is described in vector form, where Indicates the first The time delay between the factors and the target in each sample. The factor portion after the time delay is canceled out can be expressed as: Based on this, it is only necessary to construct a latency metric to determine the correct latency vector. That's all.

[0030] 103. For each time-delay embedded dataset, the time-delay adjusted meteorological factor data is used as the position coordinates of each node, and the photovoltaic power sequence corresponding to the time-delay adjusted meteorological factor data is used as the numerical label of the corresponding node. Several nodes are determined, and a local graph is formed based on the several nodes.

[0031] In this embodiment, the process of forming a local graph based on several nodes includes: Based on the positional distance between each node, determine the edge connecting each pair of nodes; A fully connected graph is determined based on each of the aforementioned nodes and each of the aforementioned edges; Calculate the weight of each side in the fully connected graph and determine the number of neighboring nodes corresponding to each node; Based on several neighboring nodes of each node, determine the local neighborhood to which each node belongs. A local graph is formed based on each of the aforementioned local neighborhoods.

[0032] To better illustrate the construction of the local graph, please refer to the following process: This approach establishes a connection between multivariate time series data and an undirected spatial graph, mapping each sample sequentially to a node in the space and assigning it a number. This is achieved through a defined node position function. make ,in, This represents the mapping from the node set to the number domain, where the factor sequence portion of the sample is used as the spatial coordinates of the nodes in the graph; R is the set of real numbers. By establishing a connection between the factor sequence and the graph, the positional distance relationship between each node can be measured based on the factor vector. This is achieved by defining a node value function. make Establish a connection between the target sequence and the graph, where, The edges of an undirected graph are defined by the node values ​​and the distances between nodes. In high-dimensional space, Mahalanobis distance is used to calculate the weights of the edges. In the formula, Represents the weight of the edge. Represents the position vector of a node. This represents the covariance matrix between vectors.

[0033] The following is an example of replacing fully connected graphs with local graphs. Figure 3As shown, for each node, only the N nodes closest to its position are considered. The nodes are connected by edges, and the positional distance between the nodes is used as the weight value attached to the edge. For each node, according to the weight value attached to its edge, a local neighborhood is constructed by filtering its N nearest nodes according to a preset weight threshold (which can be set by the user). The local graph is determined based on each local neighborhood.

[0034] 104. For each local graph, calculate the sorted rank vector of each node in the local graph, and calculate the global consistency measure based on the sorted rank vector of each node.

[0035] In this embodiment, the sorting rank vector includes: a positional sorting rank vector and a numerical sorting rank vector; calculating the sorting rank vector of each node in the local graph includes: Within each local neighborhood of the local graph, each node is sorted according to its positional distance, and the positional sorting rank vector of each node is determined based on its index. Within each local neighborhood of the local graph, each node is sorted according to its numerical distance, and the numerical sorting rank vector of each node is determined based on its index.

[0036] In one specific embodiment, the sort rank is calculated within each local neighborhood of the local graph.

[0037] Based on the local graph, consider each node and its N nearest neighbors, and sort them according to their positional and numerical information relative to the node under study, constructing two types of sorted rank vectors. The specific construction definition of the sorted rank vector is as follows: Definition 1: For a node The set of its N nearest neighbors (i.e., several neighboring nodes in the local neighborhood described in this invention) is determined by positional distance: Node The corresponding set of N neighboring nodes in, This indicates the first position sorted in ascending order by positional distance. Nearest neighbor nodes, This represents the distance between nodes i and k, which satisfies: , .

[0038] Definition 2: The position rank vector directly records the positions of the N nearest neighbors. Sequential numbering in: ,Right now Indicates the first The positional order of neighboring nodes, for example: if ,but (Corresponding number) (The nodes). The numerical rank vector in a fixed set of N neighboring nodes. In the process of reordering based on numerical distance, the numerical distance of each neighboring node is first calculated. yi: Then to Arrange the nodes in ascending order to obtain a new order. ,in This is for sorting the indexes. Finally, the numerical rank vector is defined as the mapping of the original neighbor node numbers in the new order: , It is in the original set of N neighboring nodes The numbering position.

[0039] Construct a set of position-ordered rank vectors by combining definitions 1 and 2. With numerical sorting rank vector set .

[0040] In this embodiment, the calculation of the global consistency metric based on the sorted rank vector of each node includes: Substitute the sorted rank vector of each node into the Spearman correlation coefficient formula to calculate the Spearman correlation coefficient of each node. The global consistency metric is calculated based on the Spearman correlation coefficient of each node.

[0041] In this embodiment, the Spearman calculation formula includes: In the formula, Let be the Spearman correlation coefficient of node i. Sort the rank vector of the k-th position of node i. Let the rank vector of the k-th numerical value of node i be sorted. This represents the element difference between the position-sorted rank vector and the numerical-sorted rank vector. This represents the number of nearest neighbor nodes.

[0042] Furthermore, the formula for calculating the global consistency metric includes: In the formula, Represents a global consistency measure. This represents the total number of nodes.

[0043] In one specific implementation, the arithmetic mean of the local consistency of all nodes is taken. The global consistency metric characterizes the degree of consistency between the positional ranking rank and the numerical ranking rank in the overall sample.

[0044] 105. Based on the global consistency metric corresponding to each time-delay embedded dataset, perform data alignment on several photovoltaic power sequences to obtain several aligned photovoltaic power sequences, and then perform photovoltaic power prediction based on each aligned photovoltaic power sequence.

[0045] In this embodiment, the step of aligning the photovoltaic power sequences based on the global consistency metric corresponding to each time-delay embedding dataset includes: The optimal global consistency metric is determined based on the global consistency metric corresponding to each latency embedding dataset. Select the time delay vector corresponding to the optimal global consistency metric to determine the time delay value of each photovoltaic power sequence; Based on the time delay values ​​of several photovoltaic power sequences, data alignment is performed on several photovoltaic power sequences to obtain several aligned photovoltaic power sequences.

[0046] In one specific embodiment, the largest global consistency metric is selected as the optimal global consistency metric among the global consistency metrics corresponding to each latency embedding dataset.

[0047] In one specific embodiment, the data studied in this embodiment originates from data collected from a photovoltaic power station in the DKASC photovoltaic power generation project. Data comes from a series of solar power generation facilities operating under arid conditions in Alice Springs, central Australia, and is represented by datasets (Kaneka, 6.0kW, AmorphousSilicon, Fixed, 2008) and (Kyocera, 5.4kW, poly-Si, Dual, 2008) under their respective single-technology directories, denoted as Dataset A and Dataset B.

[0048] The data collection interval for both datasets was 5 minutes. Both dataset A and dataset B contained 12 attribute sequences excluding timestamps. The active power sequence was used as the target sequence for prediction. The meteorological sequence was selected as the factor sequence, and sequences with almost no change in the meteorological sequence were removed as the prediction input, as shown in Table 1.

[0049] Due to the massive amount of data, this paper only extracts 88,129 data points from datasets A and B, covering the period from March 1st to December 31st, 2015, for the photovoltaic power prediction task. Furthermore, because the missing data in the datasets themselves could affect the research content, a nonlinear interpolation method was used to fill in the missing values ​​before the experiment.

[0050] Because the correlation strength between each factor and the target variable varies, in this example, if a factor has a weak correlation with the target value, the alignment deviation between them can be considered to have a negligible impact on the overall prediction task. Therefore, to reduce computational complexity and improve analytical efficiency, the maximum information coefficient (MIC) is used to quantitatively evaluate the correlation between the factor sequences and the target variable in the two datasets. This method allows for the selection of factor sequences with strong correlation to the target sequence, which are then the focus of time delay estimation and correction. Figure 9 , 10 The four meteorological factor sequences shown in Figure 11—Weather_Relative_Humidity (WRH), Wind_Direction (WD), Weather_Temperature_Celsius (WTC), and Wind_Speed ​​(WS)—have negligible correlation with the target sequence, and the impact of their alignment deviation with the target on the prediction task can be ignored.

[0051] For the preprocessed datasets A and B above, a time delay estimation algorithm based on ranking is used to estimate the time delay vectors among the four selected factor sequences. First, all combinations of time delay values ​​are traversed, mapping the photovoltaic power dataset under each time delay combination to nodes in a spatial undirected graph. Edge weights are calculated using Mahalanobis distance, and the N nearest neighbors are found for each node. Based on the local N-connectivity graph, the ranking vector of each node is calculated, and a local consistency metric is computed. All nodes are aggregated, and the global consistency metric is calculated by taking the arithmetic mean. Finally, the correct time delay value is searched by maximizing the global consistency metric.

[0052] The corresponding latency values ​​for the two datasets A and B are shown in Table 2: Based on the aforementioned time delay values, temporal alignment deviations were corrected. To ensure consistency in the prediction process and model parameters, six prediction Transformer extension models (including Transformer and Informer) and classic networks such as LSTM, GRU, and XGboost were employed. A uniform input sequence length of 48 steps (equivalent to 4 hours), a label length of 24 steps, and a prediction length of 48 steps were used as a single input sample. The input dataset was divided into training, validation, and test sets in a 70%:15%:15% ratio. A short-term photovoltaic power prediction task was performed, and the prediction accuracy was evaluated using three error metrics: Mean Squared Error (MSE), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE). Specific results are shown in Tables 3 and 4. The improvement in prediction accuracy after dataset A alignment compared to before alignment is also illustrated in the figure. Figure 4 , 5 6; To highlight the difference in prediction curves before and after alignment, the prediction results of the Transformer model and the LSTM model before and after alignment on dataset A are compared. Figure 7 , 8 In the text, Before Adjustment represents the prediction result before alignment, After Adjustment represents the prediction result after alignment, Ground Truth represents the true result, and Time Steps represents the time step. Figure 9 , 10 In section 11, Before represents the prediction result before alignment, After represents the prediction result after alignment, and Time Step represents the time step. The comparison results are as follows: Figure 7 , 8 As shown in 9, 10, and 11.

[0053] like Figure 2 As shown, based on the above method embodiments, corresponding apparatus embodiments are provided; An embodiment of the present invention provides a photovoltaic power prediction device, including: a data acquisition module 201, a time delay embedding module 202, a mapping module 203, a data calculation module 204, and a result generation module 205; The data acquisition module is used to acquire several time delay vectors of the power system, several photovoltaic power sequences, and meteorological factor data corresponding to each photovoltaic power sequence. The time delay embedding module is used to perform time delay adjustment on the meteorological factor data corresponding to each time delay value of each time delay vector, based on each time delay value of the time delay vector, and determine the time delay embedding dataset according to each time delay adjusted meteorological factor data and the corresponding photovoltaic power sequence; wherein, each time delay vector corresponds to a time delay embedding dataset. The mapping module is used to determine several nodes for each time-delay embedded dataset, using the time-delay adjusted meteorological factor data as the position coordinates of each node, and using the photovoltaic power sequence corresponding to the time-delay adjusted meteorological factor data as the numerical label of the corresponding node, and to form a local graph based on the several nodes. The data calculation module is used to calculate the sorted rank vector of each node in each local graph, and to calculate the global consistency measure based on the sorted rank vector of each node. The result generation module is used to perform data alignment on several photovoltaic power sequences based on the global consistency metric corresponding to each time-delay embedded dataset, to obtain several aligned photovoltaic power sequences, and then perform photovoltaic power prediction based on each aligned photovoltaic power sequence.

[0054] It is understood that the above-described device embodiments correspond to the method embodiments of the present invention, and can implement the photovoltaic power prediction method provided by any of the above-described method embodiments of the present invention.

[0055] Beneficial effects: This embodiment introduces time delay values ​​into meteorological factor data by using several time delay vectors. Each delayed meteorological factor data point serves as the position coordinate of each node, and the photovoltaic power sequence corresponding to the delayed meteorological factor data serves as the numerical label of the corresponding node. Several nodes are constructed to form a local graph. By calculating the global consistency metric of the local graph, the photovoltaic power sequence is aligned, resulting in an aligned photovoltaic power sequence for photovoltaic power prediction. This invention solves the problem of data alignment deviation in photovoltaic power prediction, thus improving the accuracy of photovoltaic power prediction.

[0056] It should be noted that the device embodiments described above are merely illustrative, and some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can specifically be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0057] Based on the above embodiments of the photovoltaic power prediction method, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the photovoltaic power prediction method of any embodiment of the present invention.

[0058] For example, in this embodiment, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the terminal device.

[0059] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0060] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting all parts of the terminal device via various interfaces and lines.

[0061] Based on the above-described method embodiments, another embodiment of the present invention provides a computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the photovoltaic power prediction method described in any of the above-described method embodiments of the present invention.

[0062] The modules / units integrated in the device / terminal equipment, if implemented as software functional units and sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0063] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A photovoltaic power prediction method, characterized in that, include: Acquire several time delay vectors of the power system, several photovoltaic power sequences, and meteorological factor data corresponding to each photovoltaic power sequence; For each of the aforementioned time delay vectors, based on each time delay value of the time delay vector, the meteorological factor data corresponding to the time delay value is adjusted for delay. Based on each adjusted meteorological factor data and the corresponding photovoltaic power sequence, a time delay embedding dataset is determined; wherein, each time delay vector corresponds to a time delay embedding dataset. For each time-delay embedded dataset, the time-delay adjusted meteorological factor data is used as the location coordinates of each node, and the photovoltaic power sequence corresponding to the time-delay adjusted meteorological factor data is used as the numerical label of the corresponding node. Several nodes are determined, and a local graph is formed based on the several nodes. For each local graph, calculate the sorted rank vector of each node in the local graph, and calculate the global consistency measure based on the sorted rank vector of each node. Based on the global consistency metric corresponding to each time-delay embedded dataset, several photovoltaic power sequences are aligned to obtain several aligned photovoltaic power sequences, and then photovoltaic power prediction is performed based on each aligned photovoltaic power sequence.

2. The photovoltaic power prediction method as described in claim 1, characterized in that, The local graph, composed of several nodes, includes: Based on the positional distance between each node, determine the edge connecting each pair of nodes; A fully connected graph is determined based on each of the aforementioned nodes and each of the aforementioned edges; Calculate the weight of each side in the fully connected graph and determine the number of neighboring nodes corresponding to each node; Based on several neighboring nodes of each node, determine the local neighborhood to which each node belongs. A local graph is formed based on each of the aforementioned local neighborhoods.

3. The photovoltaic power prediction method as described in claim 2, characterized in that, The sorting rank vector includes: a positional sorting rank vector and a numerical sorting rank vector; calculating the sorting rank vector of each node in the local graph includes: Within each local neighborhood of the local graph, each node is sorted according to its positional distance, and the positional sorting rank vector of each node is determined based on its index. Within each local neighborhood of the local graph, each node is sorted according to its numerical distance, and the numerical sorting rank vector of each node is determined based on its index.

4. The photovoltaic power prediction method as described in claim 3, characterized in that, The calculation of the global consistency metric based on the sorted rank vector of each node includes: Substitute the sorted rank vector of each node into the Spearman correlation coefficient formula to calculate the Spearman correlation coefficient of each node. The global consistency metric is calculated by substituting the Spearman correlation coefficient of each node into the global consistency metric calculation formula.

5. The photovoltaic power prediction method as described in claim 4, characterized in that, The Spearman calculation formula includes: In the formula, Let be the Spearman correlation coefficient of node i. Sort the rank vector of the k-th position of node i. Let the rank vector of the k-th numerical value of node i be sorted. This represents the element difference between the position-sorted rank vector and the numerical-sorted rank vector. This represents the number of nearest neighbor nodes.

6. The photovoltaic power prediction method as described in claim 5, characterized in that, The formula for calculating the global consistency metric includes: In the formula, Represents a global consistency measure. This represents the total number of nodes.

7. The photovoltaic power prediction method as described in claim 6, characterized in that, The step of aligning several photovoltaic power sequences based on a global consistency metric corresponding to each time-delay embedded dataset includes: The optimal global consistency metric is determined based on the global consistency metric corresponding to each latency embedding dataset. Select the time delay vector corresponding to the optimal global consistency metric to determine the time delay value of each photovoltaic power sequence; Based on the time delay values ​​of several photovoltaic power sequences, data alignment is performed on several photovoltaic power sequences to obtain several aligned photovoltaic power sequences.

8. A photovoltaic power prediction device, characterized in that, include: The system includes a data acquisition module, a latency embedding module, a mapping module, a data calculation module, and a result generation module. The data acquisition module is used to acquire several time delay vectors of the power system, several photovoltaic power sequences, and meteorological factor data corresponding to each photovoltaic power sequence. The time delay embedding module is used to perform time delay adjustment on the meteorological factor data corresponding to each time delay value of each time delay vector, based on each time delay value of the time delay vector, and determine the time delay embedding dataset according to each time delay adjusted meteorological factor data and the corresponding photovoltaic power sequence; wherein, each time delay vector corresponds to a time delay embedding dataset. The mapping module is used to determine several nodes for each time-delay embedded dataset, using the time-delay adjusted meteorological factor data as the position coordinates of each node, and using the photovoltaic power sequence corresponding to the time-delay adjusted meteorological factor data as the numerical label of the corresponding node, and to form a local graph based on the several nodes. The data calculation module is used to calculate the sorted rank vector of each node in each local graph, and to calculate the global consistency measure based on the sorted rank vector of each node. The result generation module is used to perform data alignment on several photovoltaic power sequences based on the global consistency metric corresponding to each time-delay embedded dataset, to obtain several aligned photovoltaic power sequences, and then perform photovoltaic power prediction based on each aligned photovoltaic power sequence.

9. A terminal device, characterized in that, The method includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, it implements the photovoltaic power prediction method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, include: A stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform the photovoltaic power prediction method as described in any one of claims 1-7.

Images