Energy storage method and device for output fluctuation suppression, electronic equipment and medium

Abstract

This invention relates to the field of active power distribution network energy storage strategy optimization technology, and particularly to an energy storage method, device, electronic device, and medium for suppressing power output fluctuations. The method first acquires a historical power generation data queue and multiple historical factor data queues; then, it transforms a first power generation characteristic matrix using a dimension transformation vector, and performs characteristic matching based on the transformed first characteristic vector and multiple power generation characteristic classes to obtain a target class; next, it inputs multiple factor prediction data queues into the power generation characteristic model corresponding to the target class to obtain a power generation prediction data queue; finally, based on the power generation prediction data queue and the load prediction data queue, it constructs an energy storage strategy aimed at mitigating power output fluctuations and minimizing the peak-valley difference in the power grid. This invention achieves coordinated matching between the energy storage system, distributed power sources, and grid loads, effectively mitigating power output fluctuations from distributed power sources, ensuring grid operational stability, and reducing the peak-valley difference in the power grid.

Description

Technical Field

[0001] This invention relates to the field of active power distribution network energy storage strategy optimization technology, and in particular to an energy storage method, device, electronic device and medium for suppressing power output fluctuations. Background Technology

[0002] Active distribution networks refer to distribution networks that are connected to renewable power sources such as distributed photovoltaic and wind power and have bidirectional power flow regulation capabilities. Their core disadvantages are caused by the inherent characteristics of strong randomness, intermittency and volatility in wind and solar power output.

[0003] Energy storage is a core technology for mitigating fluctuations in wind and solar power output. Existing strategies can be categorized into three main types based on their control logic: passive mitigation strategies (filtering methods, moving average methods), prediction-driven optimization strategies (model predictive control, MPC, robust / stochastic optimization), and data-driven intelligent strategies. Their commonalities and unique drawbacks are as follows: Almost all optimized energy storage strategies rely on ultra-short-term / short-term power output forecasts for photovoltaic / wind power. However, existing wind and solar power output forecasts are affected by sudden weather changes, with a normal error of 10%-30%, and even exceeding 50% under extreme weather conditions.

[0004] Prediction errors can directly lead to energy storage charging and discharging plans deviating significantly from actual needs, either resulting in inadequate mitigation and excessive power fluctuations, or causing overcharging and over-discharging of energy storage and triggering protection shutdowns. Passive strategies that do not rely on predictions are completely unforeseen and can only compensate after the fact. They are slow to respond to sudden increases and decreases in wind and solar power output and cannot mitigate large-scale shock fluctuations.

[0005] Therefore, it is necessary to develop an energy storage method for suppressing power output fluctuations. Summary of the Invention

[0006] The present invention provides an energy storage method, device, electronic device and medium for suppressing power output fluctuations, which solves the problem that existing energy storage strategy optimization methods often result in large power output fluctuations due to inaccurate power output prediction.

[0007] In a first aspect, embodiments of the present invention provide an energy storage method for suppressing power output fluctuations, comprising: The system acquires historical power generation data queues and multiple historical factor data queues. The historical factor data queues are obtained by sampling factors that affect the power generation of distributed power sources. The first power generation characteristic matrix is ​​transformed by a dimension transformation vector, and the target class is obtained by characteristic matching based on the first characteristic vector obtained by the transformation and multiple power generation characteristic classes. The first power generation characteristic matrix is ​​constructed based on the historical power generation data queue and the multiple historical factor data queues. Multiple factor prediction data queues are input into the power generation characteristic model corresponding to the target class to obtain the power generation prediction data queue; Based on the aforementioned power generation forecast data queue and load forecast data queue, an energy storage strategy is constructed with the goal of smoothing out power output fluctuations and minimizing the peak-valley difference in the power grid.

[0008] In one possible implementation, the transformation of the first power generation characteristic matrix using a dimension transformation vector, and the characteristic matching based on the transformed first characteristic vector and multiple power generation characteristic classes to obtain the target class, includes: Obtain the dimension transformation vector; The historical power generation data queue and the multiple historical factor data queues are respectively used as column vectors and arranged in a predetermined order to construct the first power generation characteristic matrix; The dimension transformation vector is multiplied by the first power generation characteristic matrix as a row vector to obtain the first characteristic vector; Based on the first characteristic vector, multiple reference vectors are found from the multiple power generation characteristic classes, wherein the distance between the reference vector and the first characteristic vector is less than a first distance threshold; The power generation characteristic class containing the most reference vectors is selected as the target class.

[0009] In one possible implementation, the dimension transformation vector is constructed based on a plurality of second power generation characteristic matrices, including: Obtain multiple second power generation characteristic matrices, multiple power generation characteristic models, and a first vector; Each second characteristic matrix is ​​split into a second factor data matrix representing the fluctuation characteristics of multiple factors and a second power generation data queue representing the fluctuation characteristics of historical power generation. The second factor data matrix is ​​then input into each power generation characteristic model. Based on the output of the model and the second power generation data queue, an applicable model for the second characteristic matrix is ​​selected from the multiple power generation characteristic models. Grouping the second characteristic matrices applicable to the same model into the same class yields multiple reference classes; The first vector is used as a row vector and multiplied by each second power generation characteristic matrix to obtain multiple second characteristic vectors, where each second characteristic vector corresponds to a second power generation characteristic matrix; Cluster the multiple second characteristic vectors according to the number of reference classes to obtain multiple first process classes; Based on the plurality of first process classes and the plurality of reference classes, a clustering bias is determined, wherein the clustering bias characterizes the degree of difference between the second power generation characteristic matrix associated with the plurality of first process classes and the second power generation characteristic matrix contained in the plurality of reference classes; If the clustering bias is greater than the bias threshold, then the first vector is adjusted according to the clustering bias; Otherwise, the first vector is used as the dimension transformation vector.

[0010] In one possible implementation, determining the clustering bias based on the plurality of first process classes and the plurality of reference classes includes: Based on the correspondence between the second characteristic vector and the second power generation characteristic matrix, multiple second process classes are constructed, wherein each second process class corresponds to a first process class, and multiple second characteristic vectors in the first process class correspond to multiple second power generation characteristic matrices in the corresponding second process class. For each second process class, the closest reference class is found from the plurality of second reference classes as the approximate class, and the total number is calculated. The approximate class contains the largest number of second power generation characteristic matrices belonging to the second process class, and the total number is the number of second power generation characteristic matrices belonging to the second process class in the approximate class. Sum the multiple shared quantities to get the total total quantity. The difference between the number of the plurality of second power generation characteristic matrices and the total number is taken as the clustering difference; The ratio of the difference in the number of clusters to the number of the plurality of second power generation characteristic matrices is used as the clustering bias.

[0011] In one possible implementation, the power generation characteristic model is constructed based on a plurality of second power generation characteristic matrices, including: Multiple second factor data matrices and multiple second power generation data queues are obtained. Each second factor data matrix corresponds to a second power generation data queue. The second factor data matrix is ​​constructed based on multiple second factor data queues that characterize the fluctuation characteristics of factors. The second power generation data queues characterize the historical power generation fluctuation characteristics. The multiple second factor data matrices are respectively input into the first model to obtain multiple model output queues, wherein each model output queue corresponds to a second factor data matrix; The prediction bias of the first model is determined based on the multiple model output queues and the multiple second power generation data queues; If the prediction deviation is less than the deviation threshold, then the first model is used as the power generation characteristic model; Otherwise, based on the prediction deviation, multiple parameters of the first model are adjusted using an optimization method, and the process jumps to the step of inputting the multiple second factor data matrices into the first model respectively to obtain multiple model output queues.

[0012] In one possible implementation, the first model is:

[0013] In the formula, For the first model output queue One element, For the exponent, To integrate independent variables, The number of rows in the input matrix. The number of columns in the input matrix. For the first The first coefficient, For the first One bias coefficient For the input matrix, the first Line number The elements of the column.

[0014] In one possible implementation, the construction of an energy storage strategy based on the power generation forecast data queue and the load forecast data queue, aimed at smoothing power output fluctuations and minimizing grid peak-to-valley differences, includes: Acquire and initialize the energy storage strategy data queue; Construct an energy consumption balance equation, wherein the energy consumption balance equation is as follows:

[0015] In the formula, In order to be in The load of time, In order to be in Power generation at any given moment In order to be in The grid output power at any given time, In order to be in Energy storage output power at any given moment; Based on the time nodes, data are extracted from the power generation forecast data queue and the load forecast data queue and substituted into the energy consumption balance equation to obtain a set of energy consumption balance equations. Based on the time node, data is extracted from the energy storage strategy data queue and substituted into the energy consumption balance equation set to construct the multiple grid output powers into a grid output data queue. Based on the power grid output data queue and the energy storage strategy data queue, determine the output fluctuation characteristic value and the power grid peak-valley difference value; If the output fluctuation characteristic value exceeds the fluctuation threshold, and / or the grid peak-valley difference value exceeds the peak-valley difference threshold, then the energy storage strategy data queue is adjusted, and the process jumps to the step of extracting data from the energy storage strategy data queue according to the time node and substituting it into the energy consumption balance equation set to construct the multiple grid output powers as the grid output data queue. Otherwise, the energy storage strategy data queue is used as the energy storage strategy.

[0016] Secondly, embodiments of the present invention provide an energy storage device for suppressing power output fluctuations, used to implement the energy storage method for suppressing power output fluctuations as described in the first aspect or any possible implementation thereof, wherein the energy storage device for suppressing power output fluctuations includes: The historical data acquisition module is used to acquire historical power generation data queues and multiple historical factor data queues. The historical factor data queues are obtained by sampling based on factors that affect the power generation of distributed power sources. The power generation characteristic matching module is used to transform the first power generation characteristic matrix through a dimension transformation vector, and perform characteristic matching based on the first characteristic vector obtained by the transformation and multiple power generation characteristic classes to obtain the target class. The first power generation characteristic matrix is ​​constructed based on the historical power generation data queue and the multiple historical factor data queues. The power generation prediction module is used to input multiple factor prediction data queues into the power generation characteristic model corresponding to the target class to obtain the power generation prediction data queue. as well as, The energy storage strategy determination module is used to construct an energy storage strategy with the goal of smoothing out power output fluctuations and minimizing the peak-valley difference in the power grid, based on the power generation forecast data queue and the load forecast data queue.

[0017] Thirdly, embodiments of the present invention provide an electronic device, including a memory and a processor, wherein the memory stores a computer program executable on the processor, and the processor executes the computer program to implement the steps of the method as described in the first aspect or any possible implementation of the first aspect.

[0018] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method as described in the first aspect or any possible implementation thereof.

[0019] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows: This invention discloses an energy storage method for suppressing power output fluctuations. First, it acquires a historical power generation data queue and multiple historical factor data queues, wherein the historical factor data queues are obtained by sampling factors affecting distributed power generation. Then, it transforms a first power generation characteristic matrix using a dimension transformation vector, and performs characteristic matching based on the transformed first characteristic vector and multiple power generation characteristic classes to obtain a target class. The first power generation characteristic matrix is ​​constructed based on the historical power generation data queues and the multiple historical factor data queues. Next, it inputs multiple factor prediction data queues into the power generation characteristic model corresponding to the target class to obtain a power generation prediction data queue. Finally, based on the power generation prediction data queue and the load prediction data queue, it constructs an energy storage strategy aimed at smoothing power output fluctuations and minimizing the peak-valley difference in the power grid.

[0020] Through iterative optimization, the construction process of this energy storage strategy has achieved coordinated matching between the energy storage system, distributed power sources, and grid loads. It can effectively smooth out the output fluctuations of distributed power sources and ensure the stability of grid operation, while also reducing the peak-valley difference of the grid and lowering grid operating costs. It is suitable for energy storage scheduling and management in various distributed power source grid connection scenarios.

[0021] Through the above steps and implementation methods, the specific operations of data acquisition, matrix construction, dimension transformation, characteristic matching and vector optimization in the process of distributed power generation prediction can be clearly defined, providing detailed guidance for technology implementation and improving the accuracy and reliability of the prediction model. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart of an energy storage method for suppressing power output fluctuations provided by an embodiment of the present invention; Figure 2 This is a schematic diagram of the dimension transformation vector construction process provided by the embodiments of the present invention; Figure 3 This is a functional block diagram of an energy storage device for suppressing power output fluctuations provided by an embodiment of the present invention; Figure 4 This is a functional block diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0024] In the following description, specific details such as particular system structures and techniques are set forth for illustrative purposes and not for limitation, so as to provide a thorough understanding of embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0025] To make the objectives, technical solutions, and advantages of the present invention clearer, specific embodiments will be described below in conjunction with the accompanying drawings.

[0026] The embodiments of the present invention will be described in detail below. This example is implemented based on the technical solution of the present invention, and provides detailed implementation methods and specific operation processes. However, the protection scope of the present invention is not limited to the following embodiments.

[0027] Figure 1 A flowchart of an energy storage method for suppressing power output fluctuations provided in an embodiment of the present invention.

[0028] like Figure 1 As shown, a flowchart illustrating the implementation of the energy storage method for suppressing power output fluctuations provided by an embodiment of the present invention is illustrated below: In step 101, a historical power generation data queue and multiple historical factor data queues are obtained. The historical factor data queues are obtained by sampling based on factors that affect the power generation of distributed power sources.

[0029] In step 102, the first power generation characteristic matrix is ​​transformed by a dimension transformation vector, and the target class is obtained by characteristic matching based on the first characteristic vector obtained by the transformation and multiple power generation characteristic classes. The first power generation characteristic matrix is ​​constructed based on the historical power generation data queue and the multiple historical factor data queues.

[0030] In some implementations, the step of transforming the first power generation characteristic matrix using a dimension transformation vector, and then performing characteristic matching based on the transformed first characteristic vector and multiple power generation characteristic classes to obtain a target class, includes: Obtain the dimension transformation vector; The historical power generation data queue and the multiple historical factor data queues are respectively used as column vectors and arranged in a predetermined order to construct the first power generation characteristic matrix; The dimension transformation vector is multiplied by the first power generation characteristic matrix as a row vector to obtain the first characteristic vector; Based on the first characteristic vector, multiple reference vectors are found from the multiple power generation characteristic classes, wherein the distance between the reference vector and the first characteristic vector is less than a first distance threshold; The power generation characteristic class containing the most reference vectors is selected as the target class.

[0031] In some implementations, the dimension transformation vector is constructed based on a plurality of second power generation characteristic matrices, including: Obtain multiple second power generation characteristic matrices, multiple power generation characteristic models, and a first vector; Each second characteristic matrix is ​​split into a second factor data matrix representing the fluctuation characteristics of multiple factors and a second power generation data queue representing the fluctuation characteristics of historical power generation. The second factor data matrix is ​​then input into each power generation characteristic model. Based on the output of the model and the second power generation data queue, an applicable model for the second characteristic matrix is ​​selected from the multiple power generation characteristic models. Grouping the second characteristic matrices applicable to the same model into the same class yields multiple reference classes; The first vector is used as a row vector and multiplied by each second power generation characteristic matrix to obtain multiple second characteristic vectors, where each second characteristic vector corresponds to a second power generation characteristic matrix; Cluster the multiple second characteristic vectors according to the number of reference classes to obtain multiple first process classes; Based on the plurality of first process classes and the plurality of reference classes, a clustering bias is determined, wherein the clustering bias characterizes the degree of difference between the second power generation characteristic matrix associated with the plurality of first process classes and the second power generation characteristic matrix contained in the plurality of reference classes; If the clustering bias is greater than the bias threshold, then the first vector is adjusted according to the clustering bias; Otherwise, the first vector is used as the dimension transformation vector.

[0032] In some implementations, determining the clustering bias based on the plurality of first process classes and the plurality of reference classes includes: Based on the correspondence between the second characteristic vector and the second power generation characteristic matrix, multiple second process classes are constructed, wherein each second process class corresponds to a first process class, and multiple second characteristic vectors in the first process class correspond to multiple second power generation characteristic matrices in the corresponding second process class. For each second process class, the closest reference class is found from the plurality of second reference classes as the approximate class, and the total number is calculated. The approximate class contains the largest number of second power generation characteristic matrices belonging to the second process class, and the total number is the number of second power generation characteristic matrices belonging to the second process class in the approximate class. Sum the multiple shared quantities to get the total total quantity. The difference between the number of the plurality of second power generation characteristic matrices and the total number is taken as the clustering difference; The ratio of the difference in the number of clusters to the number of the plurality of second power generation characteristic matrices is used as the clustering bias.

[0033] For example, as mentioned above, for a distribution network connected to distributed power sources, such as a distribution network connected to photovoltaic equipment, it mainly suppresses the output fluctuations of photovoltaic equipment through energy storage devices, which obviously improves the energy utilization rate compared to the method of curtailment.

[0034] However, current technologies rely heavily on forecasting renewable energy generation, and these technologies suffer from several shortcomings. A significant drawback is their inability to construct large-scale forecasting models for small distribution networks, hindering accurate predictions. Therefore, this invention attempts to extract the current generation characteristics of distributed power sources from historical generation data. Specifically, a pre-constructed dimensional transformation vector is used to transform the obtained historical generation data and the factors influencing it. The results are then matched with generation characteristic classes. A prediction model is selected based on this matching result. This prediction model is relatively simple and easy to implement, and through the aforementioned matching process, a good prediction effect can be achieved.

[0035] To achieve the above objectives, the present invention provides a detailed description of steps 101 to 104.

[0036] In step 101, a historical power generation data queue and multiple historical factor data queues are acquired. The historical power generation data queue covers the power generation monitoring data of the distributed power source within a preset time period in the past, including but not limited to the actual power generation records for each hour and day. The data sampling interval can be flexibly set according to the accuracy requirements of the actual application scenario. The multiple historical factor data queues are obtained by sampling various key factors that affect the power generation of the distributed power source. The influencing factors include natural environmental factors and equipment operation factors. Natural environmental factors can be specifically such as light intensity, wind speed, ambient temperature, precipitation, etc., while equipment operation factors can be specifically such as the operating load of the distributed power source, equipment wear and tear, maintenance frequency, etc. Each influencing factor corresponds to an independent historical data queue to ensure the relevance and completeness of the data.

[0037] Step 102 transforms the first power generation characteristic matrix using a dimension transformation vector, and then performs characteristic matching based on the transformed first characteristic vector and multiple preset power generation characteristic classes to accurately locate the target class that best matches the current power generation scenario. The first power generation characteristic matrix is ​​constructed based on the historical power generation data queues obtained in step 101 and multiple historical factor data queues.

[0038] This invention involves transforming the first power generation characteristic matrix using a dimension transformation vector, and then performing characteristic matching based on the first characteristic vector obtained from the transformation and multiple power generation characteristic classes to obtain the target class. First, obtain the dimension transformation vector, which is used to realize the dimension transformation of the first power generation characteristic matrix. Its dimension matches the number of columns of the first power generation characteristic matrix. The construction process will be explained in detail later. Its main purpose is to realize the process of generating matching feature data from the operating data.

[0039] Then, the historical power generation data queue and the multiple historical factor data queues are each treated as independent column vectors and arranged in a preset order to construct the first power generation characteristic matrix. The arrangement order is strict and predetermined. This order is typically set according to the influence weight of the factors on power generation; for example, the historical power generation data queue is placed as the first column, and the queues corresponding to factors with higher influence weights, such as sunlight intensity and wind speed, are arranged sequentially to ensure that the matrix clearly reflects the correlation between various types of data.

[0040] Next, the dimension transformation vector is used as a row vector and multiplied with the first power generation characteristic matrix. A linear transformation is used to convert the high-dimensional matrix into a low-dimensional vector, ultimately obtaining the first characteristic vector. This vector retains the core feature information of the first power generation characteristic matrix while effectively reducing the data dimensionality, facilitating subsequent characteristic matching operations.

[0041] Then, based on the first characteristic vector, multiple reference vectors are found from the multiple power generation characteristic classes, wherein the distance between the reference vector and the first characteristic vector must be less than a preset first distance threshold. The distance can be calculated using common vector distance methods such as Euclidean distance and Manhattan distance. The first distance threshold can be calibrated according to the accuracy requirements of the actual application scenario to ensure that the selected reference vectors have a high degree of similarity to the first characteristic vector.

[0042] Finally, the number of reference vectors contained in each power generation characteristic class is counted, and the power generation characteristic class with the most reference vectors is selected as the target class.

[0043] In some implementations, the dimension transformation vector is constructed based on multiple second power generation characteristic matrices to ensure that the dimension transformation vector can accurately adapt to the power generation characteristics of distributed power sources. The specific construction process includes the following detailed steps: like Figure 2As shown, firstly, multiple second power generation characteristic matrices, multiple preset power generation characteristic models 203, and an initial first vector 205 are obtained. The multiple second power generation characteristic matrices are constructed in the same way as the first power generation characteristic matrices, both consisting of historical power generation data queues and multiple historical factor data queues, to provide sufficient data. The multiple power generation characteristic models 203 are used to analyze the characteristics of the second power generation characteristic matrices; different models correspond to different power generation scenarios and characteristic analysis logics. The first vector 205 serves as the initial value for the dimension transformation vector; its dimension matches the number of columns in the second power generation characteristic matrix and can be set using random initialization.

[0044] Then, each second characteristic matrix is ​​split into a second factor data matrix 201 representing the fluctuation characteristics of multiple factors and a second power generation data queue 202 representing the fluctuation characteristics of historical power generation. The second factor data matrix 201 is constructed from multiple historical factor data queues as column vectors, reflecting the fluctuation patterns of various influencing factors; the second power generation data queue 202 directly corresponds to the fluctuation of historical power generation data. Each second factor data matrix 201 is then input into each power generation characteristic model 203, and the output results are obtained through model calculations. The model output results are then compared and analyzed with the corresponding second power generation data queue 202. Based on the magnitude of the comparison error, the applicable model for the second characteristic matrix is ​​selected from the multiple power generation characteristic models 203, that is, the model with the smallest error between its output result and the second power generation data queue 202 is selected as the matching model for the second characteristic matrix.

[0045] Next, the second characteristic matrices applicable to the same model are grouped into the same class, thus obtaining multiple reference classes 204. Each reference class 204 corresponds to a second characteristic matrix with similar power generation characteristics and fluctuation patterns of influencing factors, realizing the classification and organization of data and providing a foundation for subsequent cluster analysis.

[0046] Next, the first vector 205 is used as a row vector and multiplied with each second power generation characteristic matrix to obtain multiple second characteristic vectors 206. Each second characteristic vector 206 corresponds to a second power generation characteristic matrix and retains the core feature information of the corresponding second characteristic matrix.

[0047] Subsequently, the multiple second characteristic vectors 206 are clustered according to the number of reference classes 204 to obtain multiple first process classes 207. The clustering process can use common clustering algorithms such as K-means and hierarchical clustering to ensure that the clustering result is consistent with the number of reference classes 204, so that each first process class 207 corresponds to a class of second characteristic vectors 206 with similar features; Subsequently, based on multiple first process classes 207 and multiple reference classes 204, the clustering bias is determined. The clustering bias characterizes the difference between the second power generation characteristic matrix associated with multiple first process classes 207 and the second power generation characteristic matrix contained in multiple reference classes 204. The larger the bias value, the worse the conversion effect of the current first vector 205 is, and further optimization is needed. Finally, it is determined whether the clustering deviation is greater than a preset deviation threshold.

[0048] If the clustering deviation is greater than the deviation threshold, the first vector 205 is iteratively adjusted according to the magnitude and direction of the clustering deviation. After adjustment, the above clustering and deviation calculation steps are repeated until the clustering deviation meets the requirements. If the clustering deviation is not greater than the deviation threshold, it means that the transformation effect of the current first vector 205 can meet the requirements, and the first vector 205 is used as the final dimension transformation vector.

[0049] Based on multiple first process classes 207 and multiple reference classes 204, the specific implementation process of determining clustering bias is as follows: First, based on the one-to-one correspondence between the second characteristic vector 206 and the second power generation characteristic matrix, multiple second process classes are constructed. The multiple second power generation characteristic matrices corresponding to the multiple second characteristic vectors 206 in the first process class 207 together constitute this second process class, realizing the association mapping between the first process class 207 and the second power generation characteristic matrix.

[0050] Then, for each second process class, the closest reference class 204 is found from multiple reference classes 204 as the approximate class of that second process class. The criterion for judging the closeness between the reference class 204 and the second process class is the number of second power generation characteristic matrices belonging to that second process class contained in the reference class 204. That is, the reference class 204 containing the most of these matrices is selected as the approximate class of that second process class. At the same time, the number of second power generation characteristic matrices belonging to that second process class in the approximate class is counted as the common number.

[0051] Next, the total number of common values ​​corresponding to all second process classes is summed to obtain the total number of common values. This value is used to characterize the degree of fit between the clustering results and the classification results of the reference class 204. The larger the total number of common values, the better the clustering effect.

[0052] Next, the difference between the total number of multiple second power generation characteristic matrices and the total number is calculated to obtain the clustering difference. This difference is used to characterize the degree of deviation between the clustering result and the reference class 204 classification result. The larger the difference, the more obvious the deviation.

[0053] Finally, the ratio of the difference in the number of clusters to the total number of multiple second power generation characteristic matrices is calculated, and the resulting ratio is taken as the clustering bias. This ratio can intuitively reflect the relative magnitude of the clustering bias, which facilitates subsequent adjustment and optimization of the first vector 205 and ensures the accuracy and applicability of the dimension transformation vector.

[0054] In step 103, multiple factor prediction data queues are input into the power generation characteristic model corresponding to the target class to obtain the power generation prediction data queue.

[0055] In some implementations, the power generation characteristic model is constructed based on a plurality of second power generation characteristic matrices, including: Multiple second factor data matrices and multiple second power generation data queues are obtained. Each second factor data matrix corresponds to a second power generation data queue. The second factor data matrix is ​​constructed based on multiple second factor data queues that characterize the fluctuation characteristics of factors. The second power generation data queues characterize the historical power generation fluctuation characteristics. The multiple second factor data matrices are respectively input into the first model to obtain multiple model output queues, wherein each model output queue corresponds to a second factor data matrix; The prediction bias of the first model is determined based on the multiple model output queues and the multiple second power generation data queues; If the prediction deviation is less than the deviation threshold, then the first model is used as the power generation characteristic model; Otherwise, based on the prediction deviation, multiple parameters of the first model are adjusted using an optimization method, and the process jumps to the step of inputting the multiple second factor data matrices into the first model respectively to obtain multiple model output queues.

[0056] In some implementations, the first model is:

[0057] In the formula, For the first model output queue One element, For the exponent, To integrate independent variables, The number of rows in the input matrix. The number of columns in the input matrix. For the first The first coefficient, For the first One bias coefficient For the input matrix, the first Line number The elements of the column.

[0058] For example, step 103 inputs multiple factor prediction data queues into the power generation characteristic model corresponding to the target class, and obtains the power generation prediction data queue through accurate calculation of the model, providing a forward-looking basis for subsequent power output adjustment of the energy storage system. The factor prediction data queue is a set of prediction data for various key factors affecting the power generation of distributed power sources within a preset time period. Its types are consistent with the historical factor data queue in step 101, covering prediction data for natural environmental factors such as light intensity, wind speed, and ambient temperature, as well as equipment operation factors such as equipment operating load and maintenance plans. Each factor corresponds to an independent prediction data queue, and the prediction duration and sampling interval of the data match the power generation prediction requirements, ensuring the relevance and practicality of the prediction data. The power generation characteristic model corresponding to the target class is a model that has been trained and optimized in the early stage and can accurately adapt to the power generation characteristics of the target class. It can effectively explore the correlation between factor prediction data and power generation, improving the accuracy of power generation prediction.

[0059] The power generation characteristic model is constructed through iterative optimization based on multiple second power generation characteristic matrices. This ensures that the model can accurately adapt to the power generation characteristics of distributed power sources and improve prediction accuracy. The specific construction process includes the following detailed steps: First, multiple second-factor data matrices and multiple second-generation data queues are acquired to provide sufficient sample data for model training. Each second-factor data matrix corresponds to one second-generation data queue, and the two are collected synchronously in a one-to-one correspondence. The second-factor data matrices are constructed based on multiple second-factor data queues representing the fluctuation characteristics of various factors. Their construction method is consistent with the construction logic of the first generation characteristic matrix in step 101, using various second-factor data queues as column vectors, arranged according to a preset weight order, to centrally represent the fluctuation patterns of various influencing factors. The second-generation data queues directly correspond to the historical generation fluctuation characteristics of distributed power sources within the corresponding time period, recording the actual generation data under different combinations of factors, serving as label data for model training.

[0060] Then, multiple second-factor data matrices are input into a pre-defined first model. Through model processing, multiple model output queues are obtained. After each second-factor data matrix is ​​input, the model outputs a corresponding power generation prediction queue. This output queue is used to compare with the actual second-factor power generation data queue to verify the model's prediction performance. The first model can be a prediction model built based on regression principles. Its structure and initial parameter values ​​can be preset according to the power generation characteristics of distributed power sources to ensure that the model has basic prediction capabilities.

[0061] Next, based on the multiple model output queues and multiple second power generation data queues, the prediction deviation of the first model is determined. The prediction deviation can be calculated using common error calculation methods such as mean squared error and mean absolute error. By performing difference calculations on the values ​​at corresponding times in each model output queue and the corresponding second power generation data queue, and then performing statistical processing, the overall prediction deviation is obtained. This deviation value directly reflects the prediction accuracy of the first model; the smaller the deviation value, the better the model's prediction effect.

[0062] Next, it is determined whether the prediction deviation is less than a preset deviation threshold. This deviation threshold is calibrated according to the accuracy requirements of actual power generation prediction and is the core standard for measuring whether the model is qualified. If the prediction deviation is less than the deviation threshold, it means that the prediction accuracy of the first model has met the actual application requirements and no further optimization is needed. The first model is then used as the final power generation characteristic model.

[0063] If the prediction deviation is greater than or equal to the deviation threshold, it indicates that the prediction accuracy of the first model has not met the requirements and parameter optimization is necessary. At this point, based on the magnitude and distribution of the prediction deviation, a suitable optimization method is used to adjust multiple parameters of the first model. Common optimization methods include gradient descent and genetic algorithms. By iteratively adjusting the model parameters, the prediction deviation is reduced. After the parameter adjustment is completed, the process jumps to the step of inputting the multiple second-factor data matrices into the first model to obtain multiple model output queues. The model calculation and deviation calculation are then repeated until the prediction deviation is less than the deviation threshold, thus completing the construction of the power generation characteristic model.

[0064] In some implementations, the first model is constructed using the following mathematical expression, the structure of which can accurately fit the nonlinear correlation between factor data and power generation, thereby improving prediction accuracy:

[0065] The specific meanings of each parameter in the formula are as follows: For the first model output queue Each element corresponds to a predicted power generation value at a certain moment; It serves as a time series index, and its values ​​are matched with the data sampling interval and the prediction duration. The exponential order is preset based on the degree of nonlinearity of the actual power generation characteristics. The larger the value, the stronger the model's ability to fit nonlinear relationships. The optimal value can be determined through preliminary experimental calibration. As a comprehensive independent variable, it is a comprehensive representation of the data of various influencing factors, integrating the influence weights of different factors on power generation; The number of rows in the input matrix (i.e., the second factor data matrix) corresponds to the number of times the factor data was sampled. The number of columns in the input matrix corresponds to the types of factors that affect power generation; For the first The first coefficient is used to characterize the first... Time, input matrix Line number The column elements (i.e., the values ​​of a certain influencing factor at a certain moment) are relative to the comprehensive independent variable. The influence weights are determined through model training and optimization. For the first One bias coefficient is used to correct the model prediction bias and adapt to the fluctuations in power generation characteristics at different times; For the input matrix, the first Line number The elements of the column are the specific values ​​of a certain influencing factor at a certain sampling time.

[0066] This model structure enables the precise integration of various influencing factors and accurate prediction of power generation.

[0067] In step 104, an energy storage strategy is constructed based on the power generation forecast data queue and the load forecast data queue, with the goal of smoothing out power output fluctuations and minimizing the peak-valley difference in the power grid.

[0068] In some implementations, the step of constructing an energy storage strategy based on the power generation forecast data queue and the load forecast data queue, with the aim of smoothing out power output fluctuations and minimizing the peak-to-valley difference in the power grid, includes: Acquire and initialize the energy storage strategy data queue; Construct an energy consumption balance equation, wherein the energy consumption balance equation is as follows:

[0069] In the formula, In order to be in The load of time, In order to be in Power generation at any given moment In order to be in The grid output power at any given time, In order to be in Energy storage output power at any given moment; Based on the time nodes, data are extracted from the power generation forecast data queue and the load forecast data queue and substituted into the energy consumption balance equation to obtain a set of energy consumption balance equations. Based on the time node, data is extracted from the energy storage strategy data queue and substituted into the energy consumption balance equation set to construct the multiple grid output powers into a grid output data queue. Based on the power grid output data queue and the energy storage strategy data queue, determine the output fluctuation characteristic value and the power grid peak-valley difference value; If the output fluctuation characteristic value exceeds the fluctuation threshold, and / or the grid peak-valley difference value exceeds the peak-valley difference threshold, then the energy storage strategy data queue is adjusted, and the process jumps to the step of extracting data from the energy storage strategy data queue according to the time node and substituting it into the energy consumption balance equation set to construct the multiple grid output powers as the grid output data queue. Otherwise, the energy storage strategy data queue is used as the energy storage strategy.

[0070] For example, step 104 constructs an energy storage strategy with the dual objectives of smoothing power output fluctuations and minimizing the peak-to-valley difference in the power grid, based on the power generation forecast data queue obtained in step 103 and the pre-acquired load forecast data queue. This achieves dynamic matching between distributed power generation output and grid load, improving the stability and economy of power grid operation. The objective of smoothing power output fluctuations aims to reduce the instantaneous fluctuation amplitude of distributed power generation, avoid power surges from impacting the power grid, and ensure power quality. The objective of minimizing the peak-to-valley difference in the power grid aims to balance the power load during peak and off-peak hours through the charging and discharging regulation of the energy storage system, reducing the pressure on power grid operation and improving energy utilization efficiency.

[0071] Based on the aforementioned power generation forecast data queue and load forecast data queue, an energy storage strategy aimed at smoothing power output fluctuations and minimizing the peak-valley difference in the power grid is constructed. The specific implementation process undergoes multiple iterative optimization steps to ensure the scientific validity and feasibility of the strategy. Detailed steps are as follows: First, the energy storage strategy data queue is acquired and initialized. This data queue is a core set of data characterizing the operating status and control logic of the energy storage system, covering key parameters such as charging / discharging power, remaining capacity, and charging / discharging duration at each predicted time. The initialization process requires combining the energy storage system's rated capacity, maximum charging / discharging power, and charging / discharging efficiency with hardware parameters, and presetting the initial charging / discharging state and capacity thresholds for each time point. This ensures that the initialization data conforms to the actual operating constraints of the energy storage system, providing a reasonable initial foundation for subsequent iterative optimization.

[0072] Then, an energy consumption balance equation is constructed. This equation is the core constraint for ensuring the balance of power supply and demand in the power grid. It is used to relate the generation power, load power, grid output power and energy storage output power. Its specific expression is as follows:

[0073] The specific meanings of each parameter in the formula are as follows: for The power grid load at a given time is the total power demand of all types of electrical loads on the power grid at that time, and its value comes from the load forecast data queue. for The distributed power generation at any given moment is derived from the power generation forecast data queue. for The grid output power at any given time represents the power supplied by the grid to the load. When this value is positive, the grid supplies power to the load; when this value is negative, the excess power from the distributed generation is fed back to the grid. for The energy storage output power at any given time. When this value is positive, the energy storage system discharges to the grid or load. When this value is negative, the energy storage system charges from the grid or distributed power sources. For the current decision-making moment, To predict the time step, For the specific prediction time, its value range is consistent with the duration of power generation prediction and load prediction.

[0074] Next, based on preset time nodes, the corresponding values ​​are extracted from the power generation forecast data queue and the load forecast data queue, and substituted into the aforementioned energy consumption balance equation to construct a set of energy consumption balance equations. The time nodes must be consistent with the sampling interval of power generation forecast and load forecast, for example, one time node per hour, and each time node corresponds to one energy consumption balance equation. The set of equations covers the power supply and demand balance constraints at all forecast times, ensuring that the power supply and demand at each time can meet the balance requirements.

[0075] Next, based on the same time points, the corresponding energy storage output power data is extracted from the energy storage strategy data queue and substituted into the energy balance equations. The grid output power at each time point is obtained by solving the equations. Then, the grid output power at all times is arranged in chronological order to construct a grid output data queue. This queue intuitively reflects the grid output state at each predicted time point under the current energy storage strategy and is the core basis for subsequent evaluation of the effectiveness of the energy storage strategy.

[0076] Subsequently, based on the grid output data queue and the energy storage strategy data queue, the output fluctuation characteristic value and the grid peak-valley difference value are determined to assess whether the current energy storage strategy meets the preset objectives. The output fluctuation characteristic value quantifies the fluctuation degree of distributed power generation after load matching. It can be obtained by calculating the standard deviation of the difference between power generation and load power at adjacent times, the maximum fluctuation amplitude, etc. The smaller the value, the smoother the output fluctuation. The grid peak-valley difference value is obtained by calculating the difference between the maximum and minimum values ​​in the grid output data queue. The smaller the difference, the less significant the peak-valley difference in the grid, and the less pressure on grid operation. Simultaneously, preset output fluctuation thresholds and grid peak-valley difference thresholds are calibrated according to grid operation specifications, the energy storage system's regulation capability, and actual power supply quality requirements. These are the core standards for judging whether the energy storage strategy is qualified.

[0077] Subsequently, it is determined whether the output fluctuation characteristic value exceeds the fluctuation threshold, and / or whether the grid peak-valley difference exceeds the peak-valley difference threshold. If either condition is met, it indicates that the current energy storage strategy has not achieved the preset target and needs to be optimized and adjusted. At this time, the energy storage strategy data queue is adjusted according to the deviation of the output fluctuation characteristic value, the grid peak-valley difference, and the corresponding threshold. The adjustment direction includes optimizing the energy storage charging and discharging power at each time, adjusting the charging and discharging timing, and correcting the energy storage remaining capacity threshold. For example, when the output fluctuation is too large, the charging and discharging regulation amplitude of the energy storage system can be increased to smooth out power surges; when the grid peak-valley difference is too large, the energy storage discharge during peak hours and charging during off-peak hours can be optimized to balance the grid load. After the adjustment is completed, the process jumps to the step of extracting data from the energy storage strategy data queue according to the time node and substituting it into the energy consumption balance equation set to construct the grid output data queue from the obtained multiple grid output powers. The grid output data is recalculated, the target parameters are evaluated, and the strategy is adjusted until the target requirements are met.

[0078] If the output fluctuation characteristic value does not exceed the fluctuation threshold and the peak-valley difference value of the power grid does not exceed the peak-valley difference threshold, it indicates that the current energy storage strategy can simultaneously meet the dual objectives of smoothing out output fluctuations and minimizing the peak-valley difference of the power grid, and meets the operating constraints of the energy storage system and the operating requirements of the power grid. In this case, the energy storage strategy data queue is used as the final energy storage strategy to guide the actual operation of the energy storage system.

[0079] This invention provides an energy storage method implementation for suppressing power output fluctuations. First, it acquires a historical power generation data queue and multiple historical factor data queues, where the historical factor data queues are obtained by sampling factors affecting distributed power generation. Then, it transforms a first power generation characteristic matrix using a dimension transformation vector, and performs characteristic matching based on the transformed first characteristic vector and multiple power generation characteristic classes to obtain a target class. The first power generation characteristic matrix is ​​constructed based on the historical power generation data queues and the multiple historical factor data queues. Next, it inputs multiple factor prediction data queues into the power generation characteristic model corresponding to the target class to obtain a power generation prediction data queue. Finally, based on the power generation prediction data queue and the load prediction data queue, it constructs an energy storage strategy aimed at smoothing power output fluctuations and minimizing the peak-valley difference in the power grid.

[0080] Through iterative optimization, the construction process of this energy storage strategy has achieved coordinated matching between the energy storage system, distributed power sources, and grid loads. It can effectively smooth out the output fluctuations of distributed power sources and ensure the stability of grid operation, while also reducing the peak-valley difference of the grid and lowering grid operating costs. It is suitable for energy storage scheduling and management in various distributed power source grid connection scenarios.

[0081] Through the above steps and implementation methods, the specific operations of data acquisition, matrix construction, dimension transformation, characteristic matching and vector optimization in the process of distributed power generation prediction can be clearly defined, providing detailed guidance for technology implementation, while ensuring the logical coherence and rigor of each link, and improving the accuracy and reliability of the prediction model.

[0082] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0083] The following are embodiments of the apparatus of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0084] Figure 3 This is a functional block diagram of an energy storage device for suppressing power output fluctuations provided by an embodiment of the present invention, referred to as follows. Figure 3 The energy storage device for suppressing power output fluctuations includes: a historical data acquisition module 301, a power generation characteristic matching module 302, a power generation prediction module 303, and a power generation prediction module 304, wherein: The historical data acquisition module 301 is used to acquire historical power generation data queues and multiple historical factor data queues, wherein the historical factor data queues are obtained by sampling based on factors affecting the power generation of distributed power sources; The power generation characteristic matching module 302 is used to transform the first power generation characteristic matrix through a dimension transformation vector, and perform characteristic matching based on the first characteristic vector obtained by the transformation and multiple power generation characteristic classes to obtain a target class, wherein the first power generation characteristic matrix is ​​constructed based on the historical power generation data queue and the multiple historical factor data queues; The power generation prediction module 303 is used to input multiple factor prediction data queues into the power generation characteristic model corresponding to the target class to obtain the power generation prediction data queue. as well as, The power generation prediction module 304 is used to construct an energy storage strategy based on the power generation prediction data queue and the load prediction data queue, with the goal of smoothing out power output fluctuations and minimizing the peak-valley difference of the power grid.

[0085] Figure 4 This is a functional block diagram of the electronic device provided in an embodiment of the present invention. For example... Figure 4 As shown, the electronic device 4 of this embodiment includes a processor 400 and a memory 401, wherein the memory 401 stores a computer program 402 that can run on the processor 400. When the processor 400 executes the computer program 402, it implements the steps of the various energy storage methods and embodiments for suppressing power output fluctuations described above, for example... Figure 1 Steps 101 to 104 are shown.

[0086] For example, the computer program 402 may be divided into one or more modules / units, which are stored in the memory 401 and executed by the processor 400 to complete the present invention.

[0087] The electronic device 4 can be a desktop computer, laptop, handheld computer, cloud server, or other computing device. The electronic device 4 may include, but is not limited to, a processor 400 and a memory 401. Those skilled in the art will understand that... Figure 4 This is merely an example of electronic device 4 and does not constitute a limitation on electronic device 4. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device 4 may also include input / output devices, network access devices, buses, etc.

[0088] The processor 400 may 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 may be a microprocessor or any conventional processor.

[0089] The memory 401 can be an internal storage unit of the electronic device 4, such as a hard disk or memory. The memory 401 can also be an external storage device of the electronic device 4, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory 401 can include both internal and external storage units of the electronic device 4. The memory 401 is used to store the computer program 402 and other programs and data required by the electronic device 4. The memory 401 can also be used to temporarily store data that has been output or will be output.

[0090] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the aforementioned method embodiments, and will not be repeated here.

[0091] In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0092] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0093] In the embodiments provided by this invention, it should be understood that the disclosed devices / electronic devices and methods can be implemented in other ways. For example, the device / electronic device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0094] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0095] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0096] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above-described embodiments 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 methods and apparatus 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.

[0097] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. An energy storage method for suppressing power output fluctuations, characterized in that, include: The system acquires historical power generation data queues and multiple historical factor data queues. The historical factor data queues are obtained by sampling factors that affect the power generation of distributed power sources. The first power generation characteristic matrix is ​​transformed by a dimension transformation vector, and the target class is obtained by characteristic matching based on the first characteristic vector obtained by the transformation and multiple power generation characteristic classes. The first power generation characteristic matrix is ​​constructed based on the historical power generation data queue and the multiple historical factor data queues. Multiple factor prediction data queues are input into the power generation characteristic model corresponding to the target class to obtain the power generation prediction data queue; Based on the aforementioned power generation forecast data queue and load forecast data queue, an energy storage strategy is constructed with the goal of smoothing out power output fluctuations and minimizing the peak-valley difference in the power grid.

2. The energy storage method for suppressing output fluctuations according to claim 1, characterized in that, The process involves transforming the first power generation characteristic matrix using a dimension transformation vector, and then performing characteristic matching based on the transformed first characteristic vector and multiple power generation characteristic classes to obtain the target class, including: Obtain the dimension transformation vector; The historical power generation data queue and the multiple historical factor data queues are respectively used as column vectors and arranged in a predetermined order to construct the first power generation characteristic matrix; The dimension transformation vector is multiplied by the first power generation characteristic matrix as a row vector to obtain the first characteristic vector; Based on the first characteristic vector, multiple reference vectors are found from the multiple power generation characteristic classes, wherein the distance between the reference vector and the first characteristic vector is less than a first distance threshold; The power generation characteristic class containing the most reference vectors is selected as the target class.

3. The energy storage method for suppressing output fluctuations according to claim 1, characterized in that, The dimension transformation vector is constructed based on multiple second power generation characteristic matrices, including: Obtain multiple second power generation characteristic matrices, multiple power generation characteristic models, and a first vector; Each second characteristic matrix is ​​split into a second factor data matrix representing the fluctuation characteristics of multiple factors and a second power generation data queue representing the fluctuation characteristics of historical power generation. The second factor data matrix is ​​then input into each power generation characteristic model. Based on the output of the model and the second power generation data queue, an applicable model for the second characteristic matrix is ​​selected from the multiple power generation characteristic models. Grouping the second characteristic matrices applicable to the same model into the same class yields multiple reference classes; The first vector is used as a row vector and multiplied by each second power generation characteristic matrix to obtain multiple second characteristic vectors, where each second characteristic vector corresponds to a second power generation characteristic matrix; Cluster the multiple second characteristic vectors according to the number of reference classes to obtain multiple first process classes; Based on the plurality of first process classes and the plurality of reference classes, a clustering bias is determined, wherein the clustering bias characterizes the degree of difference between the second power generation characteristic matrix associated with the plurality of first process classes and the second power generation characteristic matrix contained in the plurality of reference classes; If the clustering bias is greater than the bias threshold, then the first vector is adjusted according to the clustering bias; Otherwise, the first vector is used as the dimension transformation vector.

4. The energy storage method for suppressing output fluctuations according to claim 3, characterized in that, The step of determining clustering bias based on the plurality of first process classes and the plurality of reference classes includes: Based on the correspondence between the second characteristic vector and the second power generation characteristic matrix, multiple second process classes are constructed, wherein each second process class corresponds to a first process class, and multiple second characteristic vectors in the first process class correspond to multiple second power generation characteristic matrices in the corresponding second process class. For each second process class, the closest reference class is found from the plurality of second reference classes as the approximate class, and the total number is calculated. The approximate class contains the largest number of second power generation characteristic matrices belonging to the second process class, and the total number is the number of second power generation characteristic matrices belonging to the second process class in the approximate class. Sum the multiple shared quantities to get the total total quantity. The difference between the number of the plurality of second power generation characteristic matrices and the total number is taken as the clustering difference; The ratio of the difference in the number of clusters to the number of the plurality of second power generation characteristic matrices is used as the clustering bias.

5. The energy storage method for suppressing output fluctuations according to claim 1, characterized in that, The power generation characteristic model is constructed based on multiple second power generation characteristic matrices, including: Multiple second factor data matrices and multiple second power generation data queues are obtained. Each second factor data matrix corresponds to a second power generation data queue. The second factor data matrix is ​​constructed based on multiple second factor data queues that characterize the fluctuation characteristics of factors. The second power generation data queues characterize the historical power generation fluctuation characteristics. The multiple second factor data matrices are respectively input into the first model to obtain multiple model output queues, wherein each model output queue corresponds to a second factor data matrix; The prediction bias of the first model is determined based on the multiple model output queues and the multiple second power generation data queues; If the prediction deviation is less than the deviation threshold, then the first model is used as the power generation characteristic model; Otherwise, based on the prediction deviation, multiple parameters of the first model are adjusted using an optimization method, and the process jumps to the step of inputting the multiple second factor data matrices into the first model respectively to obtain multiple model output queues.

6. The energy storage method for suppressing output fluctuations according to claim 5, characterized in that, The first model is: In the formula, For the first model output queue One element, For the exponent, To integrate independent variables, The number of rows in the input matrix. The number of columns in the input matrix. For the first The first coefficient, For the first One bias coefficient For the input matrix, the first Line number The elements of the column.

7. The energy storage method for suppressing output fluctuations according to any one of claims 1-6, characterized in that, The energy storage strategy, constructed based on the power generation forecast data queue and the load forecast data queue, with the goal of smoothing power output fluctuations and minimizing the peak-to-valley difference in the power grid, includes: Acquire and initialize the energy storage strategy data queue; Construct an energy consumption balance equation, wherein the energy consumption balance equation is as follows: In the formula, In order to be in The load of time, In order to be in Power generation at any given moment In order to be in The grid output power at any given time, In order to be in Energy storage output power at any given moment; Based on the time nodes, data are extracted from the power generation forecast data queue and the load forecast data queue and substituted into the energy consumption balance equation to obtain a set of energy consumption balance equations. Based on the time node, data is extracted from the energy storage strategy data queue and substituted into the energy consumption balance equation set to construct the multiple grid output powers into a grid output data queue. Based on the power grid output data queue and the energy storage strategy data queue, determine the output fluctuation characteristic value and the power grid peak-valley difference value; If the output fluctuation characteristic value exceeds the fluctuation threshold, and / or the grid peak-valley difference value exceeds the peak-valley difference threshold, then the energy storage strategy data queue is adjusted, and the process jumps to the step of extracting data from the energy storage strategy data queue according to the time node and substituting it into the energy consumption balance equation set to construct the multiple grid output powers as the grid output data queue. Otherwise, the energy storage strategy data queue is used as the energy storage strategy.

8. An energy storage device for suppressing power output fluctuations, characterized in that, An energy storage method for implementing the power output fluctuation suppression method as described in any one of claims 1-7, wherein the energy storage device for power output fluctuation suppression comprises: The historical data acquisition module is used to acquire historical power generation data queues and multiple historical factor data queues. The historical factor data queues are obtained by sampling based on factors that affect the power generation of distributed power sources. The power generation characteristic matching module is used to transform the first power generation characteristic matrix through a dimension transformation vector, and perform characteristic matching based on the first characteristic vector obtained by the transformation and multiple power generation characteristic classes to obtain the target class. The first power generation characteristic matrix is ​​constructed based on the historical power generation data queue and the multiple historical factor data queues. The power generation prediction module is used to input multiple factor prediction data queues into the power generation characteristic model corresponding to the target class to obtain the power generation prediction data queue. as well as, The energy storage strategy determination module is used to construct an energy storage strategy with the goal of smoothing out power output fluctuations and minimizing the peak-valley difference in the power grid, based on the power generation forecast data queue and the load forecast data queue.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7 above.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7 above.

Images