New energy consumption boundary identification method and system based on time sequence characteristics

Abstract

The invention discloses a new energy consumption boundary identification method and system based on time sequence characteristics, and the method comprises the steps: obtaining a data set of a source domain and a data set of a target domain, and enabling the data set of the source domain to comprise the power generation data and meteorological data of the source domain in a set time period, the data set of the target domain comprises meteorological data of the target domain in a set time period and does not have power generation data of the target domain in the set time period; determining a source domain common feature corresponding to the data set of the source domain and a target domain common feature corresponding to the data set of the target domain; generating a new energy output time sequence curve matched with the target domain based on the source domain public features and the target domain public features; determining an optimization model based on the new energy output time sequence curve, the power generation data of the source domain, the meteorological data of the source domain and the meteorological data of the target domain; and the new energy consumption boundary corresponding to the target domain is identified based on the meteorological features extracted from the meteorological data of the target domain by using the optimization model, so that the new energy consumption boundary of the to-be-identified region can be accurately identified.

Description

Technical Field

[0001] The present application relates to the technical field of power systems, and in particular to a method and system for identifying boundaries of renewable energy consumption based on time series characteristics. Background Art

[0002] New energy consumption refers to the process of effectively connecting the electricity generated by renewable energy (such as wind energy, solar energy, hydropower, etc.) to the power grid and reasonably distributing and utilizing it in the power grid. Generally speaking, the power generation of new energy is greatly affected by meteorological conditions and has significant volatility and randomness. Therefore, as the proportion of new energy connected to the power grid continues to increase, it will bring great uncertainty to the balanced dispatching, power flow distribution and system security of the power grid.

[0003] In the existing technology, the new energy consumption technology mainly relies on the prediction method based on the physical model. By conducting historical regression analysis on meteorological variables such as wind speed and irradiance in the prediction area, combined with the historical power generation data of the prediction area, the short-term and medium-term output of wind power and photovoltaic power generation is predicted, thereby determining the new energy consumption boundary of the prediction area. This technology performs well when the meteorological conditions are stable and the amount of historical meteorological and power generation data is large, but when the amount of power generation data in the prediction area is insufficient (for example, a newly built new energy site), the accuracy of the identified consumption boundary is low due to the lack of power generation data in the area, which will affect the safety of the power system and cannot meet the power system's demand for high-precision scheduling. Summary of the invention

[0004] In order to solve the above technical problems, the embodiment of the present application proposes a method and system for identifying new energy consumption boundaries based on time series characteristics, which can accurately identify the new energy consumption boundaries of the area to be identified without relying on the power generation data of the area to be identified, thereby improving the safety and scheduling accuracy of the power system.

[0005] In a first aspect, an embodiment of the present application provides a method for identifying a new energy consumption boundary based on time series characteristics, including: Acquire a data set of a source domain and a data set of a target domain, wherein the data set of the source domain includes power generation data and meteorological data of the source domain within a set period, and the data set of the target domain includes meteorological data of the target domain within the set period, and does not include power generation data of the target domain within the set period; Determining source domain common features corresponding to the data set of the source domain, and target domain common features corresponding to the data set of the target domain; Based on the common characteristics of the source domain and the common characteristics of the target domain, generating a new energy output timing curve matching the target domain; Determining an optimization model based on the new energy output time series curve, the power generation data of the source domain, the meteorological data of the source domain, and the meteorological data of the target domain; The optimization model is used to identify the new energy consumption boundary corresponding to the target domain based on the meteorological features extracted from the meteorological data of the target domain.

[0006] Optionally, the determining the source domain common features corresponding to the data set of the source domain and the target domain common features corresponding to the data set of the target domain includes: Using a graph-based feature extractor, feature extraction is performed on the meteorological data in the source domain to generate the source domain common features, and feature extraction is performed on the meteorological data in the target domain to generate the target domain common features.

[0007] Optionally, the using a graph-based feature extractor to perform feature extraction on the meteorological data in the source domain to generate the source domain common features, and the performing feature extraction on the meteorological data in the target domain to generate the target domain common features, includes: Inputting the meteorological data of the source domain into the feature extractor to obtain the initial source domain common features output by the feature extractor; Inputting the meteorological data of the target domain into the feature extractor to obtain the initial target domain common features output by the feature extractor; Mapping the initial source domain common features into the target space to obtain the source domain common features, and mapping the initial target domain common features into the target space to obtain the target domain common features; Among them, the feature extractor is configured to perform feature extraction through empirical mode decomposition and multi-head attention mechanism.

[0008] Optionally, the target space includes a reproducing kernel Hilbert space, and mapping the initial source domain common features into the target space to obtain the source domain common features, and mapping the initial target domain common features into the target space to obtain the target domain common features, comprises: Mapping the initial source domain common features to the reproducing kernel Hilbert space through a maximum mean difference function to obtain the source domain common features; The initial target domain common features are mapped into the reproducing kernel Hilbert space through a maximum mean difference function to obtain the target domain common features.

[0009] Optionally, determining the optimization model based on the new energy output time series curve, the power generation data of the source domain, the meteorological data of the source domain, and the meteorological data of the target domain includes: Based on the new energy output time series curve and the power generation data of the source domain, a difference calculation is performed using a mean square error loss function to obtain a difference calculation result; Based on the pre-trained target domain classifier, the difference calculation result, the meteorological data of the source domain and the meteorological data of the target domain, a preset multi-objective optimization function is configured to obtain the optimization model; in, The target domain classifier is constructed based on a domain classifier adversarial learning network, and the domain classifier adversarial learning network includes a gradient reversal layer; The training process of the target domain classifier includes: optimizing and antagonizing the gradient reversal layer according to the data set of the source domain and the data set of the target domain to train the target domain classifier.

[0010] Optionally, the using the optimization model to identify the new energy consumption boundary corresponding to the target domain based on meteorological features extracted from meteorological data of the target domain includes: Based on the meteorological features extracted from the meteorological data of the target domain, using the optimization model to generate the corresponding new energy power generation information of the target domain within the set time period; Based on the new energy power generation information, a new energy consumption boundary corresponding to the target domain is identified.

[0011] Optionally, the identifying, based on the new energy power generation information, a new energy consumption boundary corresponding to the target domain includes: Inputting the renewable energy generation information into a preset power system dispatch model, so that the power system dispatch model performs simulation according to the renewable energy generation information to generate optimized power resource configuration data corresponding to the target domain, wherein the optimized power resource configuration data at least includes renewable energy output data; Extracting time similarity from the renewable energy output data, and determining spatial boundary nodes of the renewable energy output data; Based on the time similarity and the spatial boundary nodes, a new energy consumption boundary corresponding to the target domain is determined.

[0012] Optionally, extracting time similarity from the renewable energy output data and determining spatial boundary nodes of the renewable energy output data includes: Using the t-SNE algorithm, extracting the time similarity from the renewable energy output data; Using a feature extractor based on a graph structure, extracting features from the renewable energy output data to generate features of the renewable energy output data represented by an adjacency matrix of the graph structure, wherein the graph structure includes a plurality of nodes, and the adjacency matrix is ​​determined by weights between the plurality of nodes; According to the renewable energy output data characteristics, the spatial boundary node is determined among the multiple nodes using key node analysis.

[0013] Optionally, the new energy consumption boundary is used to determine the consumption space under the boundary scenario, wherein the consumption space is determined based on the difference between the load active power at the boundary moment corresponding to the new energy consumption boundary and the minimum technical output value of the power system.

[0014] In a second aspect, the embodiment of the present application provides a new energy consumption boundary identification system based on time series characteristics, including: A data acquisition module, used to acquire a data set of a source domain and a data set of a target domain, wherein the data set of the source domain includes power generation data and meteorological data of the source domain within a set period, and the data set of the target domain includes meteorological data of the target domain within the set period, and does not include power generation data of the target domain within the set period; A common feature extraction module, used to determine a source domain common feature corresponding to the data set of the source domain, and a target domain common feature corresponding to the data set of the target domain; An output curve determination module, used to generate a new energy output timing curve matching the target domain based on the common characteristics of the source domain and the common characteristics of the target domain; An optimization module, configured to determine an optimization model based on the new energy output time series curve, the power generation data of the source domain, the meteorological data of the source domain, and the meteorological data of the target domain; The identification module is used to use the optimization model to identify the new energy consumption boundary corresponding to the target domain based on the meteorological features extracted from the meteorological data of the target domain.

[0015] In summary, the embodiments of the present application have at least the following beneficial effects: According to an embodiment of the present application, a data set of a source domain and a data set of a target domain are obtained, wherein the data set of the source domain includes power generation data and meteorological data of the source domain within a set period, and the data set of the target domain includes meteorological data of the target domain within the set period, and there is no power generation data of the target domain within the set period; a source domain common feature corresponding to the data set of the source domain and a target domain common feature corresponding to the data set of the target domain are determined; based on the source domain common feature and the target domain common feature, a new energy output timing curve matching the target domain is generated; based on the new energy output timing curve, the power generation data of the source domain, the meteorological data of the source domain and the meteorological data of the target domain, an optimization model is determined; using the optimization model, based on the meteorological features extracted from the meteorological data of the target domain, the new energy consumption boundary corresponding to the target domain is identified, so that the new energy consumption boundary of the area to be identified can be accurately identified without relying on the power generation data of the area to be identified, using sufficient and accurate data from other areas as auxiliary conditions, thereby improving the safety and scheduling accuracy of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of a method for identifying new energy consumption boundaries based on time series characteristics provided in an embodiment of the present application; Figure 2 is a schematic diagram of the analysis results on time similarity provided in an embodiment of the present application; Figure 3 is a schematic diagram of the analysis results on spatial similarity provided in an embodiment of the present application; Figure 4 It is a structural diagram of a new energy consumption boundary identification system based on time series characteristics provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0018] In the description of the present application, the terms "first", "second", "third", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, features defined as "first", "second", "third", etc. may explicitly or implicitly include one or more of the features. In the description of the present application, unless otherwise specified, the meaning of "multiple" is two or more. In the description of the present application, the term "including" and its variations are open inclusions, i.e., "including but not limited to". The term "based on" means "at least partially based on". The term "according to" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments".

[0019] In the description of this application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0020] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this application have the same meaning as those commonly understood by those skilled in the art. The terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood by specific circumstances.

[0021] First, see Figure 1 , shows a schematic flow chart of a method for identifying a new energy consumption boundary based on time series characteristics provided in an embodiment of the present application, the method comprising steps S101-S105, which are specifically as follows: S101, obtaining a data set of a source domain and a data set of a target domain, wherein the data set of the source domain includes power generation data and meteorological data of the source domain within a set time period, and the data set of the target domain includes meteorological data of the target domain within the set time period, and there is no power generation data of the target domain within the set time period; wherein the source domain and the target domain in this embodiment may represent different areas.

[0022] In an example, the dataset of the source domain can be represented as , the dataset of the target domain can be expressed as ,in, represents the i-th meteorological data in the source domain, represents the i-th type of power generation data corresponding to the i-th type of meteorological data in the source domain, represents the jth meteorological data in the target domain, represents the number of samples in the dataset of the source domain, represents the number of samples in the dataset of the target domain, , ,Here, different kinds of data can represent differences in time or space.

[0023] In one example, meteorological data may include data collected by ground meteorological stations, data collected by remote sensing meteorological satellites, etc., ensuring the diversity and comprehensiveness of the data.

[0024] S102, determining a source domain common feature corresponding to the data set of the source domain, and a target domain common feature corresponding to the data set of the target domain; S103, generating a new energy output timing curve matching the target domain based on the common characteristics of the source domain and the common characteristics of the target domain; S104, determining an optimization model based on the new energy output time series curve, the power generation data of the source domain, the meteorological data of the source domain, and the meteorological data of the target domain; S105, using the optimization model, based on meteorological features extracted from the meteorological data of the target domain, identifying a new energy consumption boundary corresponding to the target domain.

[0025] In this embodiment, the source domain data includes not only historical meteorological data, but also actual output data related to power generation, while the target domain mainly includes current meteorological data. In the absence of labeled data in the target domain, the differences in geographical location, meteorological conditions, and data annotation between the source and target domains are effectively processed, and by extracting domain-independent common features, the prediction model can adapt to the meteorological conditions of the target domain, thereby improving the accuracy of power generation prediction.

[0026] In an optional implementation, determining the source domain common features corresponding to the data set in the source domain and the target domain common features corresponding to the data set in the target domain includes: Using a graph-based feature extractor, feature extraction is performed on the meteorological data in the source domain to generate the source domain common features, and feature extraction is performed on the meteorological data in the target domain to generate the target domain common features.

[0027] In an optional implementation, the using a graph-based feature extractor to perform feature extraction on the meteorological data in the source domain to generate the source domain common features, and the performing feature extraction on the meteorological data in the target domain to generate the target domain common features, includes: Inputting the meteorological data of the source domain into the feature extractor to obtain the initial source domain common features output by the feature extractor; Inputting the meteorological data of the target domain into the feature extractor to obtain the initial target domain common features output by the feature extractor; The initial source domain common features are mapped into the target space to obtain the source domain common features, and the initial target domain common features are mapped into the target space to obtain the target domain common features.

[0028] Among them, the feature extractor is configured to perform feature extraction through empirical mode decomposition and multi-head attention mechanism.

[0029] In one example, the feature extractor can be configured to: perform empirical mode decomposition in time series on the input data to obtain multiple nodes of a graph structure, and use a multi-head attention mechanism to calculate the weights between the multiple nodes to generate an adjacency matrix of the graph structure based on the weights for characterizing common features output in response to the input data.

[0030] In this embodiment, empirical mode decomposition can be performed on the meteorological time series of meteorological data, each subsequence in the meteorological time series is regarded as a node of a graph, and the weights between the nodes are calculated using a multi-head attention mechanism to generate an adjacency matrix of the graph. The determination of the adjacency matrix can include the following formula: in, represents the i-th adjacency matrix, represents the first node feature matrix, represents the second node feature matrix, represents the first transformation matrix corresponding to the i-th adjacency matrix, The second transformation matrix corresponding to the i-th adjacency matrix, N is the number of nodes.

[0031] As mentioned above, this embodiment can dynamically adjust the structure of the graph to capture the complex dependencies between meteorological variables, thereby improving the accuracy of power generation forecasts. In order to solve the spatiotemporal dependency problem in meteorological data, this embodiment adopts a feature extraction method based on graph structure. By performing empirical mode decomposition on meteorological time series data, data on different time scales are regarded as nodes of the graph, and the weights between nodes are calculated using a multi-head attention mechanism to dynamically generate an adjacency matrix. This embodiment can accurately capture the complex spatiotemporal dependencies between meteorological variables, ensuring that the model can identify the potential interactions between different meteorological variables, thereby enhancing the adaptability of the model to variable meteorological conditions.

[0032] In an optional implementation, the target space includes a reproducing kernel Hilbert space, and mapping the initial source domain common features into the target space to obtain the source domain common features, and mapping the initial target domain common features into the target space to obtain the target domain common features, comprises: Mapping the initial source domain common features to the reproducing kernel Hilbert space through a maximum mean difference function to obtain the source domain common features; The initial target domain common features are mapped into the reproducing kernel Hilbert space through a maximum mean difference function to obtain the target domain common features.

[0033] In this embodiment, a feature extractor is used to extract domain-independent common features from the meteorological data of the source domain and the target domain, and the extracted features are mapped to the reproducing kernel Hilbert space through the Maximum Mean Discrepancy (MMD) function to align the feature distributions of the source domain and the target domain. In this way, this embodiment can ensure that the corresponding features of the source domain and the target domain can maintain consistency in the shared space, thereby reducing the differences between domains.

[0034] This embodiment maps the meteorological data of the source domain and the target domain to the reproducing kernel Hilbert space through the maximum mean difference function, and realizes the alignment of the feature distribution of the two. This embodiment ensures that the data difference between the source domain and the target domain is effectively eliminated, so that even if the target domain lacks labeled data, the model can still use the features of the source domain data to accurately predict power generation. The introduction of the feature alignment process improves the domain adaptability of the model, making its model performance more stable under different meteorological conditions.

[0035] In an optional implementation, determining the optimization model based on the new energy output time series curve, the power generation data of the source domain, the meteorological data of the source domain, and the meteorological data of the target domain includes: Based on the new energy output time series curve and the power generation data of the source domain, a difference calculation is performed using a mean square error loss function to obtain a difference calculation result.

[0036] Based on the pre-trained target domain classifier, the difference calculation result, the meteorological data of the source domain and the meteorological data of the target domain, a preset multi-objective optimization function is configured to obtain the optimization model; in, The target domain classifier is constructed based on a domain classifier adversarial learning network, and the domain classifier adversarial learning network includes a gradient reversal layer; The training process of the target domain classifier includes: optimizing and antagonizing the gradient reversal layer according to the data set of the source domain and the data set of the target domain to train the target domain classifier.

[0037] In this embodiment, the generated new energy output time series curve of the target domain is compared with the actual power generation data of the source domain, and the mean square error loss function is minimized. , to ensure that the generated new energy output curve is highly consistent with the actual power generation data, thereby improving the accuracy of output prediction.

[0038] In one example, the mean squared error loss function It can be expressed by the following formula: in, represents the number of samples in the dataset of the source domain, represents the i-th type of power generation data in the source domain, Represents the data value corresponding to the i-th power generation data in the new energy output timing curve. In this embodiment, a wind and solar power generation prediction generator can be used to generate a new energy output timing curve based on the common characteristics of the source domain and the common characteristics of the target domain. Among them, the generator can dynamically generate a new energy output timing curve (including the respective power curves of wind power generation and photovoltaic power generation) according to the input characteristics, and optimize it through the mean square error loss function to ensure that the generated new energy output timing curve is highly consistent with the power generation data in the source domain. This embodiment not only improves the prediction accuracy, but also ensures that the prediction curve is closely matched with the actual meteorological conditions, providing a reliable numerical basis for the stable operation of the power system.

[0039] In one example, the loss function for the target domain classifier is It can be expressed by the following formula: in, represents the parameters of the feature extractor, represents the parameters of the domain classifier, represents the number of samples in the dataset of the source domain, , represents the prediction of the domain classifier for the i-th source domain data sample, Represents the output probability value of the domain classifier for the i-th source domain data sample. In this embodiment, by optimizing the loss function, the model can better distinguish the source domain and the target domain, thereby enhancing the domain adaptability of the model. In this embodiment, the adversarial learning process between the source domain and the target domain is optimized by combining the domain classifier with the gradient reversal layer. By maximizing the ability of the domain classifier to distinguish the data of the source domain and the target domain, the model can better identify the differences between the source domain and the target domain, thereby enhancing the generalization ability and adaptability of the model, so that it can still accurately estimate the power output under different meteorological conditions.

[0040] In one example, the optimization model can be expressed by the following formula: in, represents the parameters of the feature extractor, represents the parameters of the domain classifier, represents the mean square error loss function, It represents the process of generating the common features of the source domain and the common features of the target domain according to the meteorological data of the source domain and the target domain respectively through the maximum mean difference function. represents the loss function of the target domain classifier, is the weight used to control the maximum mean difference function, It is the weight used to control the loss of the target domain classifier. In this embodiment, these weights can be adjusted dynamically to gradually optimize the alignment ability and prediction accuracy of the model during the training process. This embodiment introduces a multi-objective optimization strategy. By dynamically adjusting the weights of the maximum mean difference loss function and the domain classification loss function, it is ensured that the model improves the accuracy of power generation prediction while aligning the data distribution of the source domain and the target domain. This multi-objective optimization strategy further optimizes the overall performance of the model by balancing the conflict between feature alignment and power generation prediction, enabling it to maintain high robustness under complex meteorological conditions.

[0041] In an optional implementation, the using of the optimization model to identify the new energy consumption boundary corresponding to the target domain based on meteorological features extracted from meteorological data of the target domain includes: Based on the meteorological features extracted from the meteorological data of the target domain, using the optimization model to generate the corresponding new energy power generation information of the target domain within the set time period; Based on the new energy power generation information, a new energy consumption boundary corresponding to the target domain is identified.

[0042] In an optional implementation, the identifying the new energy consumption boundary corresponding to the target domain based on the new energy power generation information includes: Inputting the renewable energy generation information into a preset power system dispatch model, so that the power system dispatch model performs simulation according to the renewable energy generation information to generate optimized power resource configuration data corresponding to the target domain, wherein the optimized power resource configuration data at least includes renewable energy output data; Extracting time similarity from the renewable energy output data, and determining spatial boundary nodes of the renewable energy output data; Based on the time similarity and the spatial boundary nodes, a new energy consumption boundary corresponding to the target domain is determined.

[0043] In one example, the power system dispatch model can be used to perform a real-time sequential production operation simulation to generate an optimized power resource configuration, wherein the power system dispatch model It can be expressed by the following formula: in, Indicates the total running time of the time series production simulation, usually set to 8760 hours; Represents wind power at time The maximum accessible capacity within Represents photovoltaic at time The maximum accessible capacity within the range. At least one of the following constraints can be considered during the optimization simulation: power balance constraint, spinning reserve constraint, conventional unit output constraint, ramp rate constraint, start and stop constraint, renewable energy output constraint, and power flow constraints such as node power balance and line capacity.

[0044] In this embodiment, the time-series output scenario of new energy can be generated based on the optimized power generation prediction model, and applied to the dispatching and resource allocation of the power system. During the simulation process, the system not only considers the changes in the power generation curve, but also comprehensively considers various constraints such as power balance, rotating reserve, and conventional unit output, thereby optimizing the operation efficiency of the power grid. By accurately simulating the power generation under different meteorological conditions, the power dispatching system can respond to changing meteorological conditions more flexibly and ensure the safety and stability of the power grid.

[0045] In an optional implementation, extracting time similarity from the renewable energy output data and determining spatial boundary nodes of the renewable energy output data includes: Using a t-SNE (t-Distributed Stochastic Neighbor Embedding) algorithm, extracting the time similarity from the renewable energy output data; Using a feature extractor based on a graph structure, extracting features from the renewable energy output data to generate features of the renewable energy output data represented by an adjacency matrix of the graph structure, wherein the graph structure includes a plurality of nodes, and the adjacency matrix is ​​determined by weights between the plurality of nodes; According to the renewable energy output data characteristics, the spatial boundary node is determined among the multiple nodes using key node analysis.

[0046] In one example, in a high-dimensional space, the t-SNE algorithm first calculates the similarity between each pair of data points. and , whose similarity is expressed by conditional probability, reflecting the Select point Probability : in, is the standard deviation parameter, which controls the distance weight distribution between data points.

[0047] Then define the joint probability for: In low-dimensional space, a point and The similarity probability between pass Distributed computing: The goal of t-SNE is to minimize the Kullback-Leibler divergence (KL divergence) to make the similarity in a low-dimensional space As close as possible to similarity in high-dimensional space The formula is as follows: The loss function is minimized by gradient descent. t-SNE optimizes the embedding by updating the position of the points in the low-dimensional space so that the similarity distribution between the high-dimensional and low-dimensional spaces is as consistent as possible.

[0048] In one example, based on the renewable energy output data characteristics, using key node analysis to determine the spatial boundary node among the multiple nodes can include: in the spatial dimension, according to the power flow characteristics of the main grid represented by the renewable energy output data characteristics, sorting and reducing the dimensions of the multiple nodes, and using key node analysis to determine the spatial boundary node among the sorted and reduced nodes.

[0049] In this embodiment, the spatiotemporal similarity extraction technology is used to identify the absorption boundary of renewable energy and calculate the absorption capacity using the t-SNE algorithm. In this process, by analyzing the similarity between different nodes, the maximum and minimum boundaries of renewable energy absorption capacity are dynamically identified, and the corresponding absorption scenarios are generated. This method not only improves the absorption capacity of the power system under the access of a high proportion of renewable energy, but also provides a basis for the optimal configuration of power resources, making the system more flexible and efficient when dealing with complex meteorological conditions and load fluctuations.

[0050] See also Figure 2 , shows the analysis results obtained by analyzing the time similarity after adopting this embodiment. In this embodiment, all 8760 hours of operation data are divided into five clusters (i.e., corresponding to Figure 2 In the five coordinate systems from top to bottom in the figure, the number of instances in each cluster are 1442, 1474, 1995, 1818 and 2051 respectively. Figure 2 The results show that: 1) spatial nodes 16 and 40 are more likely to experience significant wind power abandonment; 2) in the fourth cluster (i.e., corresponding to Figure 2 In the fourth coordinate system from top to bottom in the figure, all nodes show obvious wind power abandonment phenomenon.

[0051] See also Figure 3 , shows the analysis results obtained from the spatial similarity analysis after adopting this embodiment. In this embodiment, the whole year is divided into 8760 hours, and the clustering curve shows significant wind power abandonment peaks at the 852nd and 3685th time steps, corresponding to 12:00 on February 5 and 01:00 on June 3, respectively, indicating that these time points are low load demand periods. In addition, it can be observed that the fourth category (i.e. corresponding to Figure 2 The fourth coordinate system from top to bottom) and the fifth type (corresponding to Figure 2 The clusters (the fifth coordinate system from top to bottom) generally show a high level of wind power curtailment, indicating that other factors may limit the absorption of renewable energy at these nodes. Figure 2 The analysis results in are similar to those in , where it can be observed that node 16 and node 40 are classified into the 4th and 5th clusters respectively.

[0052] In an optional implementation, the new energy consumption boundary is used to determine the consumption space under the boundary scenario, wherein the consumption space is determined based on the difference between the load active power at the boundary moment corresponding to the new energy consumption boundary and the minimum technical output value of the power system.

[0053] In one example, the absorption space It can be determined by the following formula: in, is the load active power at the boundary moment corresponding to the new energy consumption boundary, It indicates the minimum technical output value of the power system corresponding to the new energy consumption boundary.

[0054] On the second aspect, accordingly, the embodiments of the present application also provide a new energy consumption boundary identification system based on time series characteristics, which can implement all the processes of the new energy consumption boundary identification method based on time series characteristics provided in the above embodiments.

[0055] See also Figure 4 , shows a schematic diagram of the structure of a new energy consumption boundary identification system based on time series characteristics provided in an embodiment of the present application, the new energy consumption boundary identification system includes: The data acquisition module 401 is used to acquire a data set of a source domain and a data set of a target domain, wherein the data set of the source domain includes power generation data and meteorological data of the source domain within a set period, and the data set of the target domain includes meteorological data of the target domain within the set period, and does not include power generation data of the target domain within the set period; A common feature extraction module 402, configured to determine a source domain common feature corresponding to the data set of the source domain, and a target domain common feature corresponding to the data set of the target domain; An output curve determination module 403 is used to generate a new energy output timing curve matching the target domain based on the common characteristics of the source domain and the common characteristics of the target domain; An optimization module 404, configured to determine an optimization model based on the new energy output time series curve, the power generation data of the source domain, the meteorological data of the source domain, and the meteorological data of the target domain; The identification module 405 is used to identify the new energy consumption boundary corresponding to the target domain based on the meteorological features extracted from the meteorological data of the target domain by using the optimization model.

[0056] In an optional implementation, determining the source domain common features corresponding to the data set in the source domain and the target domain common features corresponding to the data set in the target domain includes: Using a graph-based feature extractor, feature extraction is performed on the meteorological data in the source domain to generate the source domain common features, and feature extraction is performed on the meteorological data in the target domain to generate the target domain common features.

[0057] In an optional implementation, the using a graph-based feature extractor to perform feature extraction on the meteorological data in the source domain to generate the source domain common features, and the performing feature extraction on the meteorological data in the target domain to generate the target domain common features, includes: Inputting the meteorological data of the source domain into the feature extractor to obtain the initial source domain common features output by the feature extractor; Inputting the meteorological data of the target domain into the feature extractor to obtain the initial target domain common features output by the feature extractor; Mapping the initial source domain common features into the target space to obtain the source domain common features, and mapping the initial target domain common features into the target space to obtain the target domain common features; Among them, the feature extractor is configured to perform feature extraction through empirical mode decomposition and multi-head attention mechanism.

[0058] In an optional implementation, the target space includes a reproducing kernel Hilbert space, and mapping the initial source domain common features into the target space to obtain the source domain common features, and mapping the initial target domain common features into the target space to obtain the target domain common features, comprises: Mapping the initial source domain common features to the reproducing kernel Hilbert space through a maximum mean difference function to obtain the source domain common features; The initial target domain common features are mapped into the reproducing kernel Hilbert space through a maximum mean difference function to obtain the target domain common features.

[0059] In an optional implementation, determining the optimization model based on the new energy output time series curve, the power generation data of the source domain, the meteorological data of the source domain, and the meteorological data of the target domain includes: Based on the new energy output time series curve and the power generation data of the source domain, a difference calculation is performed using a mean square error loss function to obtain a difference calculation result; Based on the pre-trained target domain classifier, the difference calculation result, the meteorological data of the source domain and the meteorological data of the target domain, a preset multi-objective optimization function is configured to obtain the optimization model; in, The target domain classifier is constructed based on a domain classifier adversarial learning network, and the domain classifier adversarial learning network includes a gradient reversal layer; The training process of the target domain classifier includes: optimizing and antagonizing the gradient reversal layer according to the data set of the source domain and the data set of the target domain to train the target domain classifier.

[0060] In an optional implementation, the using of the optimization model to identify the new energy consumption boundary corresponding to the target domain based on meteorological features extracted from meteorological data of the target domain includes: Based on the meteorological features extracted from the meteorological data of the target domain, using the optimization model to generate the corresponding new energy power generation information of the target domain within the set time period; Based on the new energy power generation information, a new energy consumption boundary corresponding to the target domain is identified.

[0061] In an optional implementation, the identifying the new energy consumption boundary corresponding to the target domain based on the new energy power generation information includes: Inputting the renewable energy generation information into a preset power system dispatch model, so that the power system dispatch model performs simulation according to the renewable energy generation information to generate optimized power resource configuration data corresponding to the target domain, wherein the optimized power resource configuration data at least includes renewable energy output data; Extracting time similarity from the renewable energy output data, and determining spatial boundary nodes of the renewable energy output data; Based on the time similarity and the spatial boundary nodes, a new energy consumption boundary corresponding to the target domain is determined.

[0062] In an optional implementation, extracting time similarity from the renewable energy output data and determining spatial boundary nodes of the renewable energy output data includes: Using the t-SNE algorithm, extracting the time similarity from the renewable energy output data; Using a feature extractor based on a graph structure, extracting features from the renewable energy output data to generate features of the renewable energy output data represented by an adjacency matrix of the graph structure, wherein the graph structure includes a plurality of nodes, and the adjacency matrix is ​​determined by weights between the plurality of nodes; According to the renewable energy output data characteristics, the spatial boundary node is determined among the multiple nodes using key node analysis.

[0063] In an optional implementation, the new energy consumption boundary is used to determine the consumption space under the boundary scenario, wherein the consumption space is determined based on the difference between the load active power at the boundary moment corresponding to the new energy consumption boundary and the minimum technical output value of the power system.

[0064] In summary, the embodiments of the present application have at least the following beneficial effects: According to an embodiment of the present application, a data set of a source domain and a data set of a target domain are obtained, wherein the data set of the source domain includes power generation data and meteorological data of the source domain within a set period, and the data set of the target domain includes meteorological data of the target domain within the set period, and there is no power generation data of the target domain within the set period; a source domain common feature corresponding to the data set of the source domain and a target domain common feature corresponding to the data set of the target domain are determined; based on the source domain common feature and the target domain common feature, a new energy output timing curve matching the target domain is generated; based on the new energy output timing curve, the power generation data of the source domain, the meteorological data of the source domain and the meteorological data of the target domain, an optimization model is determined; using the optimization model, based on the meteorological features extracted from the meteorological data of the target domain, the new energy consumption boundary corresponding to the target domain is identified, so that the new energy consumption boundary of the area to be identified can be accurately identified without relying on the power generation data of the area to be identified, using sufficient and accurate data from other areas as auxiliary conditions, thereby improving the safety and scheduling accuracy of the power system.

[0065] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary hardware platform, and of course, it can also be implemented entirely by hardware. Based on such an understanding, all or part of the contribution of the technical solution of the present application to the background technology can be embodied in the form of a software product, and the computer software product can be stored in a storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present application or some parts of the embodiments.

[0066] The above is a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications are also considered to be within the scope of protection of the present application.

Claims

1. A method for identifying the boundary of new energy consumption based on time series characteristics, characterized in that: include: Acquire a data set of a source domain and a data set of a target domain, wherein the data set of the source domain includes power generation data and meteorological data of the source domain within a set period, and the data set of the target domain includes meteorological data of the target domain within the set period, and does not include power generation data of the target domain within the set period; Determining source domain common features corresponding to the data set of the source domain, and target domain common features corresponding to the data set of the target domain; Based on the common characteristics of the source domain and the common characteristics of the target domain, generating a new energy output timing curve matching the target domain; Determining an optimization model based on the new energy output time series curve, the power generation data of the source domain, the meteorological data of the source domain, and the meteorological data of the target domain; The optimization model is used to identify the new energy consumption boundary corresponding to the target domain based on the meteorological features extracted from the meteorological data of the target domain.

2. The method for identifying the boundary of new energy consumption based on time series characteristics according to claim 1, characterized in that: The determining of the source domain common features corresponding to the data set of the source domain and the target domain common features corresponding to the data set of the target domain includes: Using a graph-based feature extractor, feature extraction is performed on the meteorological data in the source domain to generate the source domain common features, and feature extraction is performed on the meteorological data in the target domain to generate the target domain common features.

3. The method for identifying the boundary of new energy consumption based on time series characteristics according to claim 2, characterized in that: The method of using a graph-based feature extractor to extract features from the meteorological data in the source domain to generate the source domain common features, and extracting features from the meteorological data in the target domain to generate the target domain common features, includes: Inputting the meteorological data of the source domain into the feature extractor to obtain the initial source domain common features output by the feature extractor; Inputting the meteorological data of the target domain into the feature extractor to obtain the initial target domain common features output by the feature extractor; Mapping the initial source domain common features into the target space to obtain the source domain common features, and mapping the initial target domain common features into the target space to obtain the target domain common features; Among them, the feature extractor is configured to perform feature extraction through empirical mode decomposition and multi-head attention mechanism.

4. The method for identifying the boundary of new energy consumption based on time series characteristics according to claim 3 is characterized in that: The target space includes a reproducing kernel Hilbert space, and mapping the initial source domain common features to the target space to obtain the source domain common features, and mapping the initial target domain common features to the target space to obtain the target domain common features, comprises: Mapping the initial source domain common features to the reproducing kernel Hilbert space through a maximum mean difference function to obtain the source domain common features; The initial target domain common features are mapped into the reproducing kernel Hilbert space through a maximum mean difference function to obtain the target domain common features.

5. The method for identifying the boundary of new energy consumption based on time series characteristics according to claim 1, characterized in that: The determining of the optimization model based on the new energy output time series curve, the power generation data of the source domain, the meteorological data of the source domain and the meteorological data of the target domain includes: Based on the new energy output time series curve and the power generation data of the source domain, a difference calculation is performed using a mean square error loss function to obtain a difference calculation result; Based on the pre-trained target domain classifier, the difference calculation result, the meteorological data of the source domain and the meteorological data of the target domain, a preset multi-objective optimization function is configured to obtain the optimization model; in, The target domain classifier is constructed based on a domain classifier adversarial learning network, and the domain classifier adversarial learning network includes a gradient reversal layer; The training process of the target domain classifier includes: optimizing and antagonizing the gradient reversal layer according to the data set of the source domain and the data set of the target domain to train the target domain classifier.

6. The method for identifying the boundary of new energy consumption based on time series characteristics according to claim 1, characterized in that: The using of the optimization model to identify the new energy consumption boundary corresponding to the target domain based on the meteorological features extracted from the meteorological data of the target domain includes: Based on the meteorological features extracted from the meteorological data of the target domain, using the optimization model to generate the corresponding new energy power generation information of the target domain within the set time period; Based on the new energy power generation information, a new energy consumption boundary corresponding to the target domain is identified.

7. The method for identifying the boundary of new energy consumption based on time series characteristics according to claim 6, characterized in that: The identifying, based on the new energy power generation information, a new energy consumption boundary corresponding to the target domain includes: Inputting the renewable energy generation information into a preset power system dispatch model, so that the power system dispatch model performs simulation according to the renewable energy generation information to generate optimized power resource configuration data corresponding to the target domain, wherein the optimized power resource configuration data at least includes renewable energy output data; Extracting time similarity from the renewable energy output data, and determining spatial boundary nodes of the renewable energy output data; Based on the time similarity and the spatial boundary nodes, a new energy consumption boundary corresponding to the target domain is determined.

8. The method for identifying the boundary of new energy consumption based on time series characteristics according to claim 7, characterized in that: The extracting time similarity from the renewable energy output data and determining the spatial boundary nodes of the renewable energy output data includes: Using the t-SNE algorithm, extracting the time similarity from the renewable energy output data; Using a feature extractor based on a graph structure, extracting features from the renewable energy output data to generate features of the renewable energy output data represented by an adjacency matrix of the graph structure, wherein the graph structure includes a plurality of nodes, and the adjacency matrix is ​​determined by weights between the plurality of nodes; According to the renewable energy output data characteristics, the spatial boundary node is determined among the multiple nodes using key node analysis.

9. The method for identifying the boundary of new energy consumption based on time series characteristics according to any one of claims 1 to 8, characterized in that: The new energy consumption boundary is used to determine the consumption space under the boundary scenario, wherein the consumption space is determined based on the difference between the load active power at the boundary moment corresponding to the new energy consumption boundary and the minimum technical output value of the power system.

10. A new energy consumption boundary identification system based on time series characteristics, characterized in that: include: A data acquisition module, used to acquire a data set of a source domain and a data set of a target domain, wherein the data set of the source domain includes power generation data and meteorological data of the source domain within a set period, and the data set of the target domain includes meteorological data of the target domain within the set period, and does not include power generation data of the target domain within the set period; A common feature extraction module, used to determine a source domain common feature corresponding to the data set of the source domain, and a target domain common feature corresponding to the data set of the target domain; An output curve determination module, used to generate a new energy output timing curve matching the target domain based on the common characteristics of the source domain and the common characteristics of the target domain; An optimization module, configured to determine an optimization model based on the new energy output time series curve, the power generation data of the source domain, the meteorological data of the source domain, and the meteorological data of the target domain; The identification module is used to use the optimization model to identify the new energy consumption boundary corresponding to the target domain based on the meteorological features extracted from the meteorological data of the target domain.

Images