New energy prediction power time sequence curve inflection point identification method, device and equipment

Abstract

The invention relates to an inflection point identification method, device and equipment for a new energy predicted power time sequence curve. The method comprises the following steps: acquiring a plurality of set inflection point numbers and a time sequence of new energy predicted power in a time window; performing function construction by adopting a continuous piecewise linear fitting method according to each set inflection point number and the sequential sequence to obtain an objective function containing a fitting degree evaluation function and a trend difference evaluation function; analyzing the time sequence according to each set inflection point number and the target function to obtain an inflection point time set which corresponds to each set inflection point number and aims at the minimum target value of the target function; according to the predicted power data of the segment where two adjacent inflection points are located in the inflection point time set corresponding to each set inflection point number, calculating to obtain a corresponding identification evaluation index; and screening each set inflection point number according to all the identification evaluation indexes to obtain an optimal set inflection point number, and taking an inflection point time set corresponding to the optimal set inflection point number as an inflection point for identifying the new energy prediction power time sequence curve.

Description

Technical Field

[0001] This application relates to the field of power inflection point identification technology, and in particular to a method, apparatus and equipment for identifying the inflection point of a new energy predicted power time series curve. Background Technology

[0002] For power forecasting of renewable energy power systems, identifying the inflection points of the predicted power curve is crucial for providing key decision support for the safe and efficient operation of the power grid. Identifying the inflection points of the power curve can not only detect sudden increases or decreases in power in advance, helping the power system dispatch center reserve backup power or adjust unit output to avoid grid frequency and voltage instability and ensure operational safety; it can also create space for renewable energy output in advance by predicting power change trends, reducing wind and solar curtailment and improving energy utilization efficiency; and it can also allow grid energy storage systems and backup power to be dispatched on demand, avoiding excessive resource investment and reducing the overall operating cost of the grid.

[0003] Existing methods for identifying energy power inflection points in power systems mainly include thresholding based on time-series data, machine learning clustering, and meteorological factor correlation methods. These methods primarily focus on solving three key problems: The first is determining the number of inflection points: for a given time-series curve, how to determine the appropriate number of inflection points? The second is trend fitting: how to fit a function to a given time-series curve? The third is inflection point identification: given the number of inflection points, how to identify inflection points on a given time-series curve, ensuring sufficient difference between the sub-curves before and after the inflection point?

[0004] As mentioned above, existing methods for identifying inflection points in renewable energy power rely heavily on manually defined inflection point numbers or curve change thresholds to determine the number of inflection points. For example, thresholding methods based on time-series data identify inflection points by setting thresholds such as the rate of power change. Their advantages are simple principles, low computational cost, and the ability to quickly provide preliminary judgments. However, their disadvantages include reliance on manual threshold setting, lack of adaptability, difficulty in handling complex and changing real-world situations, and a tendency to misjudgment and missed judgments. Existing energy power inflection point identification methods involve significant computational demands in trend fitting and inflection point identification, making them difficult to apply to engineering scenarios such as renewable energy power prediction. For instance, machine learning-based inflection point identification methods can automatically learn data features and have good recognition capabilities for complex patterns. However, their disadvantage is the need for large amounts of high-quality data for training, resulting in high model training costs. Summary of the Invention

[0005] This application provides a method, apparatus, and device for identifying inflection points in the time series curve of predicted power of new energy sources, which solves the technical problems of large workload, misjudgment, or omission in existing methods for identifying inflection points in new energy power curves.

[0006] To achieve the above objectives, this application provides the following technical solution:

[0007] On the one hand, a method for identifying the inflection point of the predicted power time series curve of new energy sources is provided, including the following steps:

[0008] Obtain time series sequences of several inflection point numbers and predicted power of new energy sources within a time window, wherein each set inflection point number is less than the maximum inflection point number;

[0009] Based on each set inflection point number and the time series, a continuous piecewise linear fitting method is used to construct a function, resulting in an objective function that includes a goodness-of-fit evaluation function and a trend difference evaluation function.

[0010] The time series is analyzed based on each set of inflection point number and the objective function to obtain a set of inflection point moments corresponding to each set of inflection point number with the objective function aiming at the minimum objective value;

[0011] The identification evaluation index corresponding to each set inflection point number is obtained by calculating the predicted power data of the segment containing two adjacent inflection points in the inflection point time set corresponding to each set inflection point number.

[0012] Based on all the identification and evaluation indicators, the number of each set of inflection points is filtered to obtain the optimal number of set inflection points, and the set of inflection point times corresponding to the optimal number of set inflection points is used as the inflection point of the time series curve of the predicted power of new energy.

[0013] Preferably, the objective function is:

[0014]

[0015]

[0016]

[0017]

[0018] In the formula, K represents the number of inflection points, k∈K, t c，k Let h be the inflection point of the k-th inflection point, h be the segment, and H be [t]. c，k , ..., t c，(k+1) The number of segments in x] tc，k For the tth c，k The predicted power of new energy sources at each inflection point, x t Let be the predicted power of the new energy source at time t. Let b be the time series estimate of the predicted new energy power at time t, which is the goodness-of-fit evaluation function. k Let β be the linear fitting intercept of the predicted power of all new energy sources in the segment containing the k-th inflection point.k，h Let q() be the linear fitting slope of the predicted power of all new energy sources in the h-th segment, and let q() be the trend function. tc，k , ..., x tc，(k+1) ] represents the sub-time series of predicted new energy power between the k-th inflection point and the (k+1)-th inflection point, g[q(x tc，k , ..., x tc，(k+1) )] is the trend function q(x) tc，k , ..., x tc，(k+1) The goodness-of-fit evaluation function, f[q(x)], is used to calculate the goodness of fit. tc，k , ..., x tc，(k+1) ), q(x) tc，k+1 , ..., x tc，(k+1)+1 )] is a trend function difference evaluation function used to evaluate the similarity of sub-time series sequences between two adjacent inflection points, β k，H Let β be the slope of the linear fit of the predicted power of all new energy sources in the H-th segment at the k-th inflection point. k+1，1 Let V(Z) be the linear fitting slope of the predicted power of all new energy sources in the first segment at the (k+1)th inflection point, V(Z) be the objective value of the objective function, and Z be the set of inflection point times.

[0019] Preferably, the time series is analyzed based on each set of inflection points and the objective function to obtain a set of inflection point moments corresponding to each set of inflection points with the objective function aimed at minimizing the objective value, including:

[0020] The time series is segmented according to the set inflection point number to obtain the sub-time series of each segment and the number of segments;

[0021] Obtain initialization parameters, which include setting the number of inflection points, setting the value of the last inflection point as a time window, and setting the value of the kth inflection point as the number of inflection points;

[0022] Based on the initialization parameters and the goodness-of-fit evaluation function of the objective function, the time series is iteratively calculated using dynamic programming recursive backtracking rules to obtain K+1 inflection point times;

[0023] By deleting the last inflection point from the K+1 inflection point times, we obtain a set of K inflection point times.

[0024] Preferably, the time series is iteratively calculated using dynamic programming recursive backtracking rules based on the initialization parameters and the goodness-of-fit evaluation function of the objective function to obtain K+1 inflection point times, including:

[0025] Based on whether the value of the kth inflection point is greater than 1, determine whether to perform backtracking to solve the inflection point time;

[0026] When performing backtracking inflection point time solution, a first time and a second time are set; the inflection point time corresponding to the inflection point time is obtained by using the inflection point time formula based on the goodness of fit evaluation function between the first time and the inflection point time to be solved, and the goodness of fit evaluation function between the next time of the inflection point time to be solved and the second time.

[0027] Update the value of the last inflection point based on the inflection point time to obtain the updated value of the last inflection point time; update the value of the kth inflection point based on the value of the kth inflection point to obtain the updated value of the kth inflection point.

[0028] Based on the updated value of the kth inflection point, re-determine whether to perform backtracking inflection point time calculation until the updated value of the kth inflection point is equal to 1, and calculate the K+1 inflection point times.

[0029] Specifically, the inflection point time corresponding to the inflection point time is calculated using the inflection point time formula based on the goodness-of-fit evaluation function between the first time point and the inflection point time to be solved, and the goodness-of-fit evaluation function between the next time point of the inflection point time to be solved and the second time point; the inflection point time formula is: In the formula, t * For the inflection point time corresponding to the inflection point time t to be solved, g k-1 (u, t) represents the goodness of fit between the first time point and the inflection point to be solved, where u is the first time point, k is the kth inflection point, v is the second time point, and g1(t+1, v) represents the goodness of fit between the next time point after the inflection point to be solved and the second time point.

[0030] Preferably, the inflection point identification method for the new energy predicted power time series curve further includes: calculating, using the root mean square error index formula, an identification evaluation index corresponding to each set inflection point number based on the segmented predicted power data of each inflection point in the inflection point time set corresponding to each set inflection point number; the root mean square error index formula is:

[0031]

[0032]

[0033] In the formula, x tc，k For the tth c，k The predicted power of new energy sources at each inflection point, where T is the time window, and x t Let be the predicted power of the new energy source at time t. Let x be the mean of the predicted power of all new energy sources in the j-th segment, and h be the predicted power of new energy sources. t The segment to which it belongs, RMSE is the identification evaluation index.

[0034] Preferably, the inflection point identification method for the new energy predicted power time series curve further includes: using the argmin function to filter each set inflection point number according to all the identification evaluation indicators to obtain the optimal set inflection point number; the argmin function is:

[0035]

[0036] In the formula, RMSE K The evaluation metric for setting the Kth inflection point is K, where K is the number of inflection points. max K represents the maximum number of inflection points. * Set the number of inflection points to be optimal.

[0037] On the other hand, a device for identifying inflection points of the predicted power time series curve of new energy sources is provided, including a data acquisition module, an objective function construction module, an inflection point list calculation module, an identification index calculation module, and a screening and identification module;

[0038] The data acquisition module is used to acquire time series of several set inflection point numbers and time windows for predicted power of new energy sources, wherein each set inflection point number is less than the maximum inflection point number;

[0039] The objective function construction module is used to construct a function based on each set inflection point number and the time series using a continuous piecewise linear fitting method, to obtain an objective function that includes a fit evaluation function and a trend difference evaluation function.

[0040] The inflection point list calculation module is used to analyze the time series based on each set inflection point number and the objective function to obtain a set of inflection point moments corresponding to each set inflection point number with the objective function aiming at the minimum objective value.

[0041] The identification index calculation module is used to calculate the identification evaluation index corresponding to each set inflection point number based on the predicted power data of the segment containing two adjacent inflection points in the inflection point time set corresponding to each set inflection point number.

[0042] The filtering and identification module is used to filter each of the set inflection point numbers according to all the identification evaluation indicators to obtain the optimal set inflection point number, and to use the set of inflection point times corresponding to the optimal set inflection point number as the inflection point of the time series curve of the predicted power of new energy.

[0043] Preferably, the objective function is:

[0044]

[0045]

[0046]

[0047]

[0048] In the formula, K represents the number of inflection points, k∈K, t c，k Let h be the inflection point of the k-th inflection point, h be the segment, and H be [t]. c，k , ..., t c，(k+1) The number of segments in x] tc，k For the tth c，k The predicted power of new energy sources at each inflection point, x t Let be the predicted power of the new energy source at time t. Let b be the time series estimate of the predicted new energy power at time t, which is the goodness-of-fit evaluation function. k Let β be the linear fitting intercept of the predicted power of all new energy sources in the segment containing the k-th inflection point. k，h Let q() be the linear fitting slope of the predicted power of all new energy sources in the h-th segment, and let q() be the trend function. tc，k , ..., x tc，(k+1) ] represents the sub-time series of predicted new energy power between the k-th inflection point and the (k+1)-th inflection point, g[q(x tc，k , ..., x tc，(k+1) )] is the trend function q(x) tc，k , ..., x tc，(k+1) The goodness-of-fit evaluation function, f[q(x)], is used to calculate the goodness of fit. tc，k , ..., x tc，(k+1) ), q(x) tc，k+1 , ..., x tc，(k+1)+1 )] is a trend function difference evaluation function used to evaluate the similarity of sub-time series sequences between two adjacent inflection points, β k，H Let β be the slope of the linear fit of the predicted power of all new energy sources in the H-th segment at the k-th inflection point. k+1，1 Let V(Z) be the linear fitting slope of the predicted power of all new energy sources in the first segment at the (k+1)th inflection point, V(Z) be the objective value of the objective function, and Z be the set of inflection point times.

[0049] Preferably, the inflection point list calculation module includes a segmentation submodule, a parameter initialization submodule, a loop calculation submodule, and an inflection point list composition submodule;

[0050] The segmentation submodule is used to segment the time series according to the set inflection point number, so as to obtain the sub-time series of each segment and the number of segments;

[0051] The parameter initialization submodule is used to obtain initialization parameters, which include setting the number of inflection points, setting the value at the last inflection point as a time window, and setting the value at the kth inflection point as the number of inflection points.

[0052] The cyclic calculation submodule is used to perform cyclic calculations on the time series according to the initialization parameters and the fit evaluation function of the objective function using dynamic programming recursive backtracking rules to obtain K+1 inflection point times;

[0053] The inflection point list constitutes a submodule, which is used to delete the last inflection point time from the K+1 inflection point times to obtain an inflection point time set consisting of K inflection point times;

[0054] Based on the initialization parameters and the goodness-of-fit evaluation function of the objective function, and using dynamic programming recursive backtracking rules, the time series is iteratively calculated to obtain K+1 inflection point moments, including:

[0055] Based on whether the value of the kth inflection point is greater than 1, determine whether to perform backtracking to solve the inflection point time;

[0056] When performing backtracking inflection point time solution, a first time and a second time are set; the inflection point time corresponding to the inflection point time is obtained by using the inflection point time formula based on the goodness of fit evaluation function between the first time and the inflection point time to be solved, and the goodness of fit evaluation function between the next time of the inflection point time to be solved and the second time.

[0057] Update the value of the last inflection point based on the inflection point time to obtain the updated value of the last inflection point time; update the value of the kth inflection point based on the value of the kth inflection point to obtain the updated value of the kth inflection point.

[0058] Based on the updated value of the kth inflection point, re-determine whether to perform backtracking inflection point time calculation until the updated value of the kth inflection point is equal to 1, and calculate the K+1 inflection point times.

[0059] The formula for the inflection point time is as follows: In the formula, t * For the inflection point time corresponding to the inflection point time t to be solved, g k-1 (u, t) represents the goodness of fit between the first time point and the inflection point to be solved, where u is the first time point, k is the kth inflection point, v is the second time point, and g1(t+1, v) represents the goodness of fit between the next time point after the inflection point to be solved and the second time point.

[0060] On the other hand, a terminal device is provided, including a processor and a memory;

[0061] The memory is used to store program code and transmit the program code to the processor;

[0062] The processor is used to execute the inflection point identification method for the predicted power time series curve of new energy sources as described above, according to the instructions in the program code.

[0063] This invention discloses a method, apparatus, and equipment for identifying inflection points in the time series curve of predicted renewable energy power. The method includes acquiring a time series sequence of predicted renewable energy power within a set number of inflection points and a time window, where each set number of inflection points is less than the maximum number of inflection points. Based on each set number of inflection points and the time series sequence, a continuous piecewise linear fitting method is used to construct a function, resulting in an objective function that includes a goodness-of-fit evaluation function and a trend difference evaluation function. The time series sequence is analyzed based on each set number of inflection points and the objective function to obtain a set of inflection point moments corresponding to each set number of inflection points, with the objective function aiming to minimize its target value. Based on the predicted power data of the segments containing two adjacent inflection points in the inflection point moment set corresponding to each set number of inflection points, an identification evaluation index corresponding to each set number of inflection points is calculated. Finally, each set number of inflection points is filtered based on all identification evaluation indices to obtain an optimal set number of inflection points, and the set of inflection point moments corresponding to the optimal set number of inflection points is used as the inflection point of the predicted renewable energy power time series curve.

[0064] As can be seen from the above technical solutions, this application has the following advantages: The inflection point identification method of the new energy power prediction time series curve first constructs an objective function by obtaining several set inflection point numbers and time series sequences using a continuous piecewise linear fitting method. Based on the inflection point time set of each inflection point identified by the objective function, the corresponding identification evaluation index is calculated according to the inflection point time set of each set inflection point number to obtain the optimal set inflection point number. The inflection point time set corresponding to the optimal set inflection point number is used as the inflection point of the new energy power prediction time series curve. The power inflection point identification is simple and improves the accuracy of power inflection point identification. It solves the technical problems of large workload, misjudgment or omission in the existing new energy power curve inflection point identification methods.

[0065] This inflection point identification device for the time series curve of predicted power from new energy sources transforms the problem of identifying inflection points in predicted power from new energy sources into a digital form by constructing an objective function through a data acquisition module, an objective function construction module, an inflection point list calculation module, an identification index calculation module, and a screening and identification module. This avoids the problem of excessive training volume in machine learning-based methods and facilitates engineering applications. The objective function is constructed using a continuous piecewise linear fitting method, which is not only simple to calculate but also able to fit the nonlinear characteristics of new energy output, improving the identification accuracy. The optimal number of inflection points is determined by various identification evaluation indicators, thus avoiding the problem of other methods relying on manually defined inflection point numbers or curve change thresholds. Attached Figure Description

[0066] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, 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 this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 This is a flowchart illustrating the steps of the inflection point identification method for the predicted power time series curve of new energy sources described in this application embodiment;

[0068] Figure 2 This is a schematic diagram of the inflection point identification device for the new energy predicted power time series curve described in this application embodiment;

[0069] Figure 3 This is a schematic diagram of the terminal device described in an embodiment of this application. Detailed Implementation

[0070] To make the inventive objectives, features, and advantages of this application more apparent and understandable, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0071] In the description of the embodiments of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0072] In the embodiments of this application, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in the embodiments of this application according to the specific circumstances.

[0073] This application provides a method, apparatus, and device for identifying inflection points in the time series curve of predicted power of new energy sources, which solves the technical problems of large workload, misjudgment, or omission in existing methods for identifying inflection points in the power curve of new energy sources.

[0074] Example 1:

[0075] Figure 1 This is a flowchart illustrating the steps of the inflection point identification method for the predicted power time series curve of new energy sources described in this application embodiment.

[0076] like Figure 1 As shown in the figure, this application provides a method for identifying the inflection point of a new energy power forecast time series curve, including the following steps:

[0077] S1. Obtain time series of several inflection point numbers and predicted power of new energy sources within a time window, where each inflection point number is less than the maximum inflection point number.

[0078] It should be noted that in step S1, firstly, several different values ​​of the set inflection point number are obtained; secondly, the time series sequence of the predicted power of new energy sources [x1, ..., x] within the time window [1, 2, ..., T] is obtained. t , ..., x T Each of the aforementioned inflection point numbers is less than the maximum inflection point number K. max This provides data for subsequent steps. In this embodiment, the number of inflection points can be set to K inflection points. t The new energy power at time t is denoted as t_t. The inflection point is denoted as t_t. c The kth inflection point is denoted as t. c , k The sub-time series of predicted new energy power in the segment between the k-th inflection point and the (k+1)-th inflection point is denoted as [x]. tc,k , ..., x tc,(k+1) ].

[0079] S2. Based on each set inflection point number and time series, a continuous piecewise linear fitting method is used to construct the function, resulting in an objective function that includes a goodness-of-fit evaluation function and a trend difference evaluation function.

[0080] It should be noted that, in order to transform the problem of inflection point detection in the time series of new energy power prediction into digital modeling, step S2 constructs an objective function based on the number of inflection points and the time series obtained in step S1, providing a solution for the time series [x1, ..., x...]. t , ..., x TThe optimal set of inflection points in the [reference] provides the objective function. In this embodiment, the function construction can obtain the objective function from the set number of inflection points and time series obtained in step S1, and mathematize the problem of inflection point detection of new energy power prediction time series through the objective function, so as to effectively detect the dynamic changes and trends of new energy power prediction time series through the form of a function.

[0081] Understandably, the objective function is calculated from data such as the Euclidean distance between the estimated fit values ​​of the sub-time series corresponding to multiple inflection points with set inflection point numbers and the actual sample values, and the slope difference between two adjacent segments of the two sub-time series curves. This results in a set of inflection point times, with each inflection point set recorded in time series form, thus reflecting the optimal inflection point time for each set of inflection point numbers. Furthermore, the inflection point times calculated through the objective function can characterize the time-varying trend of the inflection points of the new energy predicted power curve. Therefore, by setting the number of inflection points and using a continuous piecewise linear fitting method to construct the function, this application not only preserves the continuity and trend in the time dimension but also reveals the potential nonlinear characteristics, coupling relationships, and causal influences between different sub-time series.

[0082] In this embodiment of the application, the objective function is:

[0083]

[0084]

[0085]

[0086]

[0087] In the formula, K represents the number of inflection points, k∈K, t c，k Let h be the inflection point of the k-th inflection point, h be the segment, and H be [t]. c，k , ..., t c，(k+1) The number of segments in x] tc，k For the tth c，k The predicted power of new energy sources at each inflection point, x t Let be the predicted power of the new energy source at time t. Let b be the time series estimate of the predicted new energy power at time t, which is the goodness-of-fit evaluation function. k Let β be the linear fitting intercept of the predicted power of all new energy sources in the segment containing the k-th inflection point. k，h Let q() be the linear fitting slope of the predicted power of all new energy sources in the h-th segment, and let q() be the trend function. tc，k , ..., x tc，(k+1)] represents the sub-time series of predicted new energy power between the k-th inflection point and the (k+1)-th inflection point, g[q(x tc，k , ..., x tc，(k+1) )] is the trend function q(x) tc，k , ..., x tc，(k+1) The goodness-of-fit evaluation function, f[q(x)], is used to calculate the goodness of fit. tc，k , ..., x tc，(k+1) ), q(x) tc，k+1 , ..., x tc，(k+1)+1 )] is a trend function difference evaluation function used to evaluate the similarity of sub-time series sequences between two adjacent inflection points, β k，H Let β be the slope of the linear fit of the predicted power of all new energy sources in the H-th segment at the k-th inflection point. k+1，1 Let V(Z) be the linear fitting slope of the predicted power of all new energy sources in the first segment at the (k+1)th inflection point, V(Z) be the objective value of the objective function, and Z be the set of inflection point times.

[0088] It should be noted that in the objective function, g[q(x)] tc，k , ..., x tc，(k+1) )] is the trend function q(x) tc，k , ..., x tc，(k+1) The goodness-of-fit evaluation function for x is q(x). tc，k , ..., x tc，(k+1) The better the fit [x] tc，k , ..., x tc，(k+1) If the trend is as follows, then g[q(x)] tc，k , ..., x tc，(k+1) The smaller the value of f[q(x)], the better; tc，k , ..., x tc，(k+1) ), q(x) tc，k+1 , ..., x tc，(k+1)+1) The trend difference evaluation function is used to assess the similarity between the curves corresponding to two sub-time series before and after an inflection point. In general, the objective function requires that for each time series segment, the goodness of fit should be as high as possible, i.e., g[q(x...] tc，k , ..., x tc，(k+1) The smaller the value of f[q(x)], the larger the difference between the corresponding curves of the two sub-time series before and after the inflection point, i.e., f[q(x)] should be. tc , k , ..., x tc , (k+1) ), q(x) tc，k+1 , ..., x tc，(k+1)+1 [Make it as large as possible]

[0089] In this embodiment, the inflection point identification method for the time series curve of predicted new energy power uses a continuous piecewise linear fitting method to fit the trend of predicted new energy power in the time series. The continuous piecewise linear fitting method uses the trend function q(x) tc，k , ..., x tc，(k+1) The trend function q() is defined as a piecewise linear function, and this trend function q() satisfies the following: in any sub-window [t c，k , ..., t c，(k+1) The points [t] are all piecewise linear, and two adjacent sub-windows [t] are all piecewise linear. c，（k-1） , ..., t c，k ]、[t c，k , ..., t c，(k+1) The value at the [] is continuous. Based on the linear function fitting, the goodness-of-fit evaluation function g[q(x] can be used. tc，k , ..., x tc，(k+1) The value is defined as the Euclidean distance between the estimated fit and the true sample value. The trend function and difference evaluation function are f[q(x)]. tc，k , ..., x tc，(k+1) ), q(x) tc，k+1 , ..., x tc，(k+1)+1) The slope difference between two adjacent segments of the curves corresponding to two sub-time series is calculated.

[0090] S3. Analyze the time series based on each set inflection point number and objective function to obtain the set of inflection point times corresponding to each set inflection point number, with the goal of minimizing the objective function value.

[0091] It should be noted that in step S3, the time series obtained in step S1 is analyzed based on the objective function constructed in step S2 and the set number of each inflection point, to obtain the set of inflection point times Z={t} corresponding to each set number of inflection points, with the goal of minimizing the objective function. c，1 , ..., t c，k , ..., t c，K The analysis yields a set of inflection points that fully considers the complex relationship between linear and nonlinear characteristic data in the predicted power of new energy sources in the time series. It evaluates the results of linear fitting and trend differences based on the minimum objective value of the objective function, and analyzes the optimal set of inflection points for each setting with high accuracy, providing data for subsequent steps.

[0092] Understandably, the process involves calculating the sum of Euclidean distances for all sub-time series at each inflection point within a given set of inflection points, based on the objective function, to obtain the total distance; and calculating the sum of differences in the linear fitting slopes of all adjacent sub-time series, to obtain the sum of differences; then calculating the difference between the total distance and the sum of differences as the objective value; using the minimum objective value as the evaluation metric, the objective function is iteratively calculated with the goal of maximizing the fit (minimizing the Euclidean distance) and maximizing the difference in curves between the two sub-time series before and after the inflection point (maximizing the linear fitting slope), until the objective value is minimized, thus obtaining the inflection point time set. Solving for the inflection point time set not only considers the curve trend fitting of the time series but also the difference in slopes between adjacent sub-time series curves, ensuring the accuracy of the solved inflection point time set.

[0093] S4. Calculate the identification evaluation index corresponding to each set inflection point number based on the predicted power data of the segments containing the two adjacent inflection points in the inflection point time set corresponding to each set inflection point number.

[0094] It should be noted that in step S4, each sub-time series sequence is determined based on the inflection point time set corresponding to each set inflection point number obtained in step S3. The identification and evaluation index of the set inflection point number is calculated based on each sub-time series sequence to provide data for subsequent steps to determine the optimal set inflection point number.

[0095] Understandably, since the optimal inflection point number is unknown, this method for identifying the inflection point of the predicted power time series curve for new energy sources employs a traversal approach to determine the optimal inflection point number. Specifically, a maximum inflection point number K is initially given. max Step S3 is used to process 0~K max The inflection point numbers were tested separately for each setting, resulting in an inflection point time set corresponding to each setting. Sub-time series sequences for each inflection point time set were obtained, and the root mean square error (RMSE) was used to calculate the corresponding identification and evaluation index for each inflection point time set. For each set of inflection point numbers K, inflection point identification was performed, and the RMSE index was used for evaluation. Finally, the optimal inflection point number and corresponding inflection point time set were selected. The RMSE index is one of the core indicators for measuring the accuracy of a prediction model, used to quantify the average difference between the model's predicted values ​​and the actual values, and is particularly suitable for continuous value prediction (regression problems). Here, the optimal inflection point time set Z, after segmenting the original time series, can measure the predicted power x of new energy sources based on the RMSE index. t With x tThe smaller the residual between the average values ​​of the segments, the greater the similarity between the segments of the original time series after segmentation based on the optimal inflection point time set Z, and the more effective the segmentation (inflection point setting) is. Therefore, the inflection point identification method for the new energy predicted power time series curve also includes: calculating the identification evaluation index corresponding to each set inflection point number using the root mean square error index formula based on the segmented predicted power data of each inflection point in the inflection point time set corresponding to each set inflection point number; the root mean square error index formula is:

[0096]

[0097]

[0098] In the formula, x tc，k For the tth c，k The predicted power of new energy sources at each inflection point, where T is the time window, and x t Let be the predicted power of the new energy source at time t. Let x be the mean of the predicted power of all new energy sources in the j-th segment, and h be the predicted power of new energy sources. t The segment to which it belongs, RMSE is the identification evaluation index.

[0099] S5. Based on all the identification and evaluation indicators, each set of inflection point numbers is filtered to obtain the optimal set of inflection point numbers, and the set of inflection point times corresponding to the optimal set of inflection point numbers is used as the inflection point of the time series curve of the predicted power of new energy.

[0100] It should be noted that in step S5, the optimal inflection point number K is obtained by using the argmin function to filter the inflection point numbers based on the identification and evaluation indicators of each setting inflection point number calculated in step S4. * The optimal inflection point number K is set. * The set of inflection points is used to identify the inflection points of the predicted power time series curve of new energy sources. The argmin function is:

[0101]

[0102] In the formula, RMSE K The evaluation metric for setting the Kth inflection point is K, where K is the number of inflection points. max K represents the maximum number of inflection points. * Set the number of inflection points to be optimal.

[0103] In the embodiments of this application, as shown in Table 1, the inflection point identification method of the new energy predicted power time series curve calculates the inflection point time set of each set inflection point number in steps S3 and S4, and uses the argmin function to filter the identification evaluation index to obtain the inflection point of the new energy predicted power time series curve.

[0104] Table 1 shows the flowchart for determining the optimal inflection point number.

[0105]

[0106] In this embodiment, the inflection point identification method for the predicted power time series curve of new energy sources transforms the problem of identifying the inflection point of predicted power into a digital form by constructing an objective function, thus avoiding the problem of excessive training volume in machine learning and facilitating engineering applications. Furthermore, the objective function constructed by this method not only requires improving the fitting degree of the power curve but also requires improving the difference between the sub-curves before and after the inflection point.

[0107] It should be noted that the inflection point identification method for the predicted power time series curve of this new energy source adopts a continuous piecewise linear fitting method to construct the objective function. This method is not only computationally simple but also able to fit the nonlinear characteristics of new energy output, thus improving the identification accuracy. Furthermore, this method determines the optimal number of inflection points by using various evaluation indicators for setting the number of inflection points, thereby avoiding the problem of other methods relying on manually defined inflection point numbers or curve change thresholds.

[0108] This application provides a method for identifying inflection points in a time series curve of predicted new energy power. The method includes acquiring several set inflection point numbers and time series sequences of predicted new energy power within a time window, where each set inflection point number is less than the maximum inflection point number; constructing a function using a continuous piecewise linear fitting method based on each set inflection point number and the time series sequence, obtaining an objective function that includes a goodness-of-fit evaluation function and a trend difference evaluation function; analyzing the time series sequence based on each set inflection point number and the objective function to obtain a set of inflection point times corresponding to each set inflection point number, with the goal of minimizing the objective function's target value; calculating an identification evaluation index corresponding to each set inflection point number based on the predicted power data of the segments containing two adjacent inflection points in the inflection point time set corresponding to each set inflection point number; filtering each set inflection point number based on all identification evaluation indices to obtain the optimal set inflection point number, and using the set of inflection point times corresponding to the optimal set inflection point number as the inflection point of the predicted new energy power time series curve. This method for identifying inflection points in the predicted power time series curve of new energy sources first constructs an objective function by obtaining several set inflection point numbers and time series sequences using a continuous piecewise linear fitting method. Based on the inflection point time set identified by the objective function, the corresponding identification evaluation index is calculated according to the inflection point time set of each set inflection point number to obtain the optimal set inflection point number. The inflection point time set corresponding to the optimal set inflection point number is then used as the inflection point of the predicted power time series curve of new energy sources. The power inflection point identification is simple and improves the accuracy of power inflection point identification, solving the technical problems of large workload, misjudgment or omission in existing methods for identifying inflection points of new energy power curves.

[0109] In one embodiment of this application, the time series is analyzed based on each set inflection point number and objective function to obtain a set of inflection point moments corresponding to each set inflection point number with the objective function aiming at minimizing the objective value.

[0110] The time series is segmented according to the set inflection point number to obtain the sub-time series of each segment and the number of segments;

[0111] Get the initialization parameters, which include setting the number of inflection points, setting the value at the last inflection point to the time window, and setting the value at the kth inflection point to the number of inflection points.

[0112] Based on the initialization parameters and the goodness-of-fit evaluation function of the objective function, the time series is iteratively calculated using dynamic programming recursive backtracking rules to obtain K+1 inflection point times;

[0113] By removing the last inflection point from the K+1 inflection point times, we obtain a set of K inflection point times.

[0114] Based on the initialization parameters and the goodness-of-fit evaluation function of the objective function, and using dynamic programming recursive backtracking rules to iteratively calculate the time series, we obtain K+1 inflection points, including:

[0115] Determine whether to perform backtracking to solve the inflection point based on whether the value of the kth inflection point is greater than 1;

[0116] When performing backtracking inflection point time calculation, a first time and a second time are set; the inflection point time corresponding to the inflection point time is obtained by using the inflection point time formula based on the goodness of fit evaluation function between the first time and the inflection point time to be solved and the goodness of fit evaluation function between the next time of the inflection point time to be solved and the second time.

[0117] Update the value at the last inflection point based on the inflection point time to obtain the updated value at the last inflection point time. Update the value at the kth inflection point based on the value at the kth inflection point to obtain the updated value at the kth inflection point.

[0118] Based on the updated value of the kth inflection point, re-determine whether to perform backtracking inflection point time calculation, until the updated value of the kth inflection point is equal to 1, and obtain the K+1 inflection point time.

[0119] The formula for the inflection point is as follows: In the formula, t * For the inflection point time corresponding to the inflection point time t to be solved, g k-1 (u, t) represents the goodness of fit between the first time point and the inflection point to be solved, where u is the first time point, k is the kth inflection point, v is the second time point, and g1(t+1, v) represents the goodness of fit between the next time point after the inflection point to be solved and the second time point.

[0120] It should be noted that the inflection point identification method of the new energy predicted power time series curve first obtains the initial parameters by setting the number of inflection points, the goodness-of-fit evaluation function of the objective function, and the time series. Then, the dynamic programming recursive backtracking rule is used to iteratively calculate the time series to determine the K inflection point times corresponding to the set number of inflection points.

[0121] Example 2:

[0122] Figure 2 This is a schematic diagram of the inflection point identification device for the new energy predicted power time series curve described in this application embodiment.

[0123] like Figure 2 As shown in the figure, this application provides an inflection point identification device for the time series curve of predicted power of new energy, including a data acquisition module 10, an objective function construction module 20, an inflection point list calculation module 30, an identification index calculation module 40, and a screening and identification module 50.

[0124] Data acquisition module 10 is used to acquire time series of several set inflection point numbers and time windows of new energy predicted power, where each set inflection point number is less than the maximum inflection point number;

[0125] The objective function construction module 20 is used to construct a function based on each set inflection point number and time series using a continuous piecewise linear fitting method, to obtain an objective function that includes a fit evaluation function and a trend difference evaluation function.

[0126] The inflection point list calculation module 30 is used to analyze the time series based on each set inflection point number and objective function to obtain a set of inflection point moments corresponding to each set inflection point number with the objective function aiming at the minimum objective value.

[0127] The identification index calculation module 40 is used to calculate the identification evaluation index corresponding to each set inflection point number based on the predicted power data of the segment containing two adjacent inflection points in the inflection point time set corresponding to each set inflection point number.

[0128] The filtering and identification module 50 is used to filter each set of inflection point numbers according to all identification and evaluation indicators, obtain the optimal set of inflection point numbers, and use the set of inflection point times corresponding to the optimal set of inflection point numbers as the inflection points of the time series curve of the predicted power of new energy.

[0129] It should be noted that the content of the modules in the device of Embodiment 2 has been described in the steps of the method of Embodiment 1, and the content of the inflection point identification device module for the new energy predicted power time series curve will not be described again in this embodiment. In this embodiment, the inflection point identification device for the new energy predicted power time series curve transforms the problem of identifying the inflection point of new energy predicted power into digital form through the objective function constructed by the data acquisition module, the objective function construction module, the inflection point list calculation module, the identification index calculation module, and the screening and identification module. This avoids the problem of excessive training volume based on machine learning and facilitates engineering applications. The objective function is constructed by using a continuous piecewise linear fitting method, which is not only simple to calculate, but also can fit the nonlinear characteristics of new energy output and improve the identification accuracy. The optimal number of inflection points is determined by the identification evaluation index of each set number of inflection points, thus avoiding the problem of other methods relying on manually defined number of inflection points or curve change thresholds.

[0130] In this embodiment of the application, the objective function is:

[0131]

[0132]

[0133]

[0134]

[0135] In the formula, K represents the number of inflection points, k∈K, t c，k Let h be the inflection point of the k-th inflection point, h be the segment, and H be [t]. c，k , ..., t c，(k+1) The number of segments in x] tc，k For the tth c，k The predicted power of new energy sources at each inflection point, x t Let be the predicted power of the new energy source at time t. Let b be the time series estimate of the predicted new energy power at time t, which is the goodness-of-fit evaluation function. k Let β be the linear fitting intercept of the predicted power of all new energy sources in the segment containing the k-th inflection point. k，h Let q() be the linear fitting slope of the predicted power of all new energy sources in the h-th segment, and let q() be the trend function. tc，k , ..., x tc，(k+1) ] represents the sub-time series of predicted new energy power between the k-th inflection point and the (k+1)-th inflection point, g[q(x tc，k , ..., x tc，(k+1) )] is the trend function q(x) tc，k , ..., x tc，(k+1) The goodness-of-fit evaluation function, f[q(x)], is used to calculate the goodness of fit.tc，k , ..., x tc，(k+1) ), q(x) tc，k+1 , ..., x tc，(k+1)+1 )] is a trend function difference evaluation function used to evaluate the similarity of sub-time series sequences between two adjacent inflection points, β k，H Let β be the slope of the linear fit of the predicted power of all new energy sources in the H-th segment at the k-th inflection point. k+1，1 Let V(Z) be the linear fitting slope of the predicted power of all new energy sources in the first segment at the (k+1)th inflection point, V(Z) be the objective value of the objective function, and Z be the set of inflection point times.

[0136] In this embodiment of the application, the inflection point list calculation module 30 includes a segmentation submodule, a parameter initialization submodule, a loop calculation submodule, and an inflection point list composition submodule;

[0137] The segmentation submodule is used to segment the time series according to the set number of inflection points, and obtain the sub-time series of each segment and the number of segments;

[0138] The parameter initialization submodule is used to obtain initialization parameters, which include setting the number of inflection points, setting the value at the last inflection point to the time window, and setting the value at the kth inflection point to the number of inflection points.

[0139] The iterative calculation submodule is used to perform iterative calculations on the time series using dynamic programming recursive backtracking rules based on the initialization parameters and the goodness-of-fit evaluation function of the objective function, to obtain K+1 inflection point times;

[0140] The inflection point list constitutes a submodule, which is used to delete the last inflection point time from the K+1 inflection point times to obtain a set of inflection point times consisting of K inflection point times;

[0141] The dynamic programming recursive backtracking rules include:

[0142] Determine whether to perform backtracking to solve the inflection point based on whether the value of the kth inflection point is greater than 1;

[0143] When performing backtracking inflection point time calculation, a first time and a second time are set; the inflection point time corresponding to the inflection point time is obtained by using the inflection point time formula based on the goodness of fit evaluation function between the first time and the inflection point time to be solved and the goodness of fit evaluation function between the next time of the inflection point time to be solved and the second time.

[0144] Update the value at the last inflection point based on the inflection point time to obtain the updated value at the last inflection point time. Update the value at the kth inflection point based on the value at the kth inflection point to obtain the updated value at the kth inflection point.

[0145] Based on the updated value of the kth inflection point, re-determine whether to perform backtracking inflection point time calculation, until the updated value of the kth inflection point is equal to 1, and obtain the K+1 inflection point time.

[0146] The formula for the inflection point is as follows: In the formula, t * For the inflection point time corresponding to the inflection point time t to be solved, g k-1 (u, t) represents the goodness of fit between the first time point and the inflection point to be solved, where u is the first time point, k is the kth inflection point, v is the second time point, and g1(t+1, v) represents the goodness of fit between the next time point after the inflection point to be solved and the second time point.

[0147] Example 3:

[0148] Figure 3 This is a schematic diagram of the terminal device described in an embodiment of this application.

[0149] like Figure 3 As shown, this application provides a terminal device, including a processor and a memory;

[0150] Memory is used to store program code and transfer the program code to the processor;

[0151] The processor is used to execute the aforementioned inflection point identification method for the predicted power time series curve of new energy sources according to the instructions in the program code.

[0152] It should be noted that the processor is used to execute the steps in the above-described embodiment of a method for identifying inflection points in the power time series curve of a new energy source, according to the instructions in the program code. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above-described system / device embodiments.

[0153] For example, a computer program can be divided into one or more modules / units, one or more of which are stored in memory and executed by a processor to complete this application. One or more modules / units can be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in a terminal device.

[0154] Terminal devices can be computing devices such as desktop computers, laptops, handheld computers, and cloud servers. Terminal devices may include, but are not limited to, processors and memory. Those skilled in the art will understand that this does not constitute a limitation on the terminal device, which may include more or fewer components than illustrated, or combinations of certain components, or different components. For example, a terminal device may also include input / output devices, network access devices, buses, etc.

[0155] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0156] Memory can be an internal storage unit of a terminal device, such as a hard drive or RAM. Memory can also be an external storage device, such as a plug-in hard drive, SmartMedia Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, memory can include both internal and external storage units. Memory is used to store computer programs and other programs and data required by the terminal device. Memory can also be used to temporarily store data that has been output or will be output.

[0157] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0158] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of 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 coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0159] 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 according to actual needs.

[0160] 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.

[0161] If the integrated 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, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0162] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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 this application.

Claims

1. A method for identifying the inflection point of a new energy power forecast time series curve, characterized in that, Includes the following steps: Obtain time series sequences of several inflection point numbers and predicted power of new energy sources within a time window, wherein each set inflection point number is less than the maximum inflection point number; Based on each set inflection point number and the time series, a continuous piecewise linear fitting method is used to construct a function, resulting in an objective function that includes a goodness-of-fit evaluation function and a trend difference evaluation function. The time series is analyzed based on each set of inflection points and the objective function to obtain a set of inflection point moments corresponding to each set of inflection points with the objective function aiming at the minimum objective value. The identification evaluation index corresponding to each set inflection point number is obtained by calculating the predicted power data of the segment containing two adjacent inflection points in the inflection point time set corresponding to each set inflection point number. Based on all the identification and evaluation indicators, the number of each set of inflection points is filtered to obtain the optimal number of set inflection points, and the set of inflection point times corresponding to the optimal number of set inflection points is used as the inflection point of the time series curve of the predicted power of new energy.

2. The method for identifying the inflection point of the predicted power time series curve of new energy sources according to claim 1, characterized in that, The objective function is: In the formula, K represents the number of inflection points, k∈K, t c，k Let h be the inflection point of the k-th inflection point, h be the segment, and H be [t]. c，k , ..., t c，(k+1) The number of segments in x] tc，k For the tth c，k The predicted power of new energy sources at each inflection point, x t Let be the predicted power of the new energy source at time t. Let b be the time series estimate of the predicted new energy power at time t, which is the goodness-of-fit evaluation function. k Let β be the linear fitting intercept of the predicted power of all new energy sources in the segment containing the k-th inflection point. k，h Let q() be the linear fitting slope of the predicted power of all new energy sources in the h-th segment, and let q() be the trend function. tc，k , ..., x tc，(k+1) ] represents the sub-time series of predicted new energy power between the k-th inflection point and the (k+1)-th inflection point, g[q(x tc，k , ..., x tc，(k+1) )] is the trend function q(x) tc，k , ..., x tc，(k+1) The goodness-of-fit evaluation function, f[q(x)], is used to calculate the goodness of fit. tc，k , ..., x tc，(k+1) ), q(x) tc，k+1 , ..., x tc，(k+1)+1 )] is a trend function difference evaluation function used to evaluate the similarity of sub-time series sequences between two adjacent inflection points, β k，H Let β be the slope of the linear fit of the predicted power of all new energy sources in the H-th segment at the k-th inflection point. k+1，1 Let V(Z) be the linear fitting slope of the predicted power of all new energy sources in the first segment at the (k+1)th inflection point, V(Z) be the objective value of the objective function, and Z be the set of inflection point times.

3. The method for identifying the inflection point of the predicted power time series curve of new energy sources according to claim 1, characterized in that, The time series is analyzed based on each set of inflection point numbers and the objective function to obtain a set of inflection point moments corresponding to each set of inflection point numbers, with the objective function aiming to minimize the objective value. This set includes: The time series is segmented according to the set inflection point number to obtain the sub-time series of each segment and the number of segments; Obtain initialization parameters, which include setting the number of inflection points, setting the value of the last inflection point as a time window, and setting the value of the kth inflection point as the number of inflection points; Based on the initialization parameters and the goodness-of-fit evaluation function of the objective function, the time series is iteratively calculated using dynamic programming recursive backtracking rules to obtain K+1 inflection point times; By deleting the last inflection point from the K+1 inflection point times, we obtain a set of K inflection point times.

4. The inflection point identification method for the time series curve of predicted power of new energy sources according to claim 3, characterized in that, Based on the initialization parameters and the goodness-of-fit evaluation function of the objective function, and using dynamic programming recursive backtracking rules, the time series is iteratively calculated to obtain K+1 inflection point moments, including: Based on whether the value of the kth inflection point is greater than 1, determine whether to perform backtracking to solve the inflection point time; When performing backtracking inflection point time solution, a first time and a second time are set; the inflection point time corresponding to the inflection point time is obtained by using the inflection point time formula based on the goodness of fit evaluation function between the first time and the inflection point time to be solved, and the goodness of fit evaluation function between the next time of the inflection point time to be solved and the second time. Update the value of the last inflection point based on the inflection point time to obtain the updated value of the last inflection point time; update the value of the kth inflection point based on the value of the kth inflection point to obtain the updated value of the kth inflection point. Based on the updated value of the kth inflection point, re-determine whether to perform backtracking inflection point time calculation until the updated value of the kth inflection point is equal to 1, and calculate the K+1 inflection point times. The formula for the inflection point time is as follows: In the formula, t * For the inflection point time corresponding to the inflection point time t to be solved, g k-1 (u, t) represents the goodness of fit between the first time point and the inflection point to be solved, where u is the first time point, k is the kth inflection point, v is the second time point, and g1(t+1, v) represents the goodness of fit between the next time point after the inflection point to be solved and the second time point.

5. The method for identifying the inflection point of the predicted power time series curve of new energy sources according to any one of claims 1-4, characterized in that, Also includes: The root mean square error index is calculated using the segmented predicted power data of each inflection point in the inflection point time set corresponding to each set inflection point number, and the identification evaluation index corresponding to each set inflection point number is obtained; the root mean square error index formula is: In the formula, x tc，k For the tth c，k The predicted power of new energy sources at each inflection point, where T is the time window, and x t Let be the predicted power of the new energy source at time t. Let x be the mean of the predicted power of all new energy sources in the j-th segment, and h be the predicted power of new energy sources. t The segment to which it belongs, RMSE is the identification evaluation index.

6. The method for identifying the inflection point of the predicted power time series curve of new energy sources according to any one of claims 1-4, characterized in that, Also includes: Based on all the aforementioned identification and evaluation indicators, the argmin function is used to filter the number of each set inflection point to obtain the optimal number of set inflection points; the argmin function is: In the formula, RMSE K The evaluation metric for setting the Kth inflection point is K, where K is the number of inflection points. max K represents the maximum number of inflection points. * Set the number of inflection points to be optimal.

7. A device for identifying inflection points in the time-series curve of predicted power from new energy sources, characterized in that, include: Data acquisition module, objective function construction module, inflection point list calculation module, identification indicator calculation module, and filtering and identification module; The data acquisition module is used to acquire time series of several set inflection point numbers and time windows for predicted power of new energy sources, wherein each set inflection point number is less than the maximum inflection point number; The objective function construction module is used to construct a function based on each set inflection point number and the time series using a continuous piecewise linear fitting method, to obtain an objective function that includes a fit evaluation function and a trend difference evaluation function. The inflection point list calculation module is used to analyze the time series based on each set inflection point number and the objective function to obtain a set of inflection point moments corresponding to each set inflection point number with the objective function aiming at the minimum objective value. The identification index calculation module is used to calculate the identification evaluation index corresponding to each set inflection point number based on the predicted power data of the segment containing two adjacent inflection points in the inflection point time set corresponding to each set inflection point number. The filtering and identification module is used to filter each of the set inflection point numbers according to all the identification evaluation indicators to obtain the optimal set inflection point number, and to use the set of inflection point times corresponding to the optimal set inflection point number as the inflection point of the new energy predicted power time series curve.

8. The inflection point identification device for the predicted power time series curve of new energy sources according to claim 7, characterized in that, The objective function is: In the formula, K represents the number of inflection points, k∈K, t c，k Let h be the inflection point of the k-th inflection point, h be the segment, and H be [t]. c，k , ..., t c，(k+1) The number of segments in x] tc，k For the tth c，k The predicted power of new energy sources at each inflection point, x t Let be the predicted power of the new energy source at time t. Let b be the time series estimate of the predicted new energy power at time t, which is the goodness-of-fit evaluation function. k Let β be the linear fitting intercept of the predicted power of all new energy sources in the segment containing the k-th inflection point. k，h Let q() be the linear fitting slope of the predicted power of all new energy sources in the h-th segment, and let q() be the trend function. tc，k , ..., x tc，(k+1) ] represents the sub-time series of predicted new energy power between the k-th inflection point and the (k+1)-th inflection point, g[q(x tc，k , ..., x tc，(k+1) )] is the trend function q(x) tc，k , ..., x tc，(k+1) The goodness-of-fit evaluation function, f[q(x)], is used to calculate the goodness of fit. tc，k , ..., x tc，(k+1) ), q(x) tc，k+1 , ..., x tc，(k+1)+1 )] is a trend function difference evaluation function used to evaluate the similarity of sub-time series sequences between two adjacent inflection points, β k，H Let β be the slope of the linear fit of the predicted power of all new energy sources in the H-th segment at the k-th inflection point. k+1，1 Let V(Z) be the linear fitting slope of the predicted power of all new energy sources in the first segment at the (k+1)th inflection point, V(Z) be the objective value of the objective function, and Z be the set of inflection point times.

9. The inflection point identification device for the predicted power time series curve of new energy sources according to claim 7, characterized in that, The inflection point list calculation module includes a segmentation submodule, a parameter initialization submodule, a loop calculation submodule, and an inflection point list composition submodule. The segmentation submodule is used to segment the time series according to the set inflection point number, so as to obtain the sub-time series of each segment and the number of segments; The parameter initialization submodule is used to obtain initialization parameters, which include setting the number of inflection points, setting the value at the last inflection point as a time window, and setting the value at the kth inflection point as the number of inflection points. The cyclic calculation submodule is used to perform cyclic calculations on the time series according to the initialization parameters and the fit evaluation function of the objective function using dynamic programming recursive backtracking rules to obtain K+1 inflection point times; The inflection point list constitutes a submodule, which is used to delete the last inflection point time from the K+1 inflection point times to obtain an inflection point time set consisting of K inflection point times; Based on the initialization parameters and the goodness-of-fit evaluation function of the objective function, and using dynamic programming recursive backtracking rules, the time series is iteratively calculated to obtain K+1 inflection point moments, including: Based on whether the value of the kth inflection point is greater than 1, determine whether to perform backtracking to solve the inflection point time; When performing backtracking inflection point time solution, a first time and a second time are set; the inflection point time corresponding to the inflection point time is calculated by using the inflection point time formula based on the goodness of fit evaluation function between the first time and the inflection point time to be solved and the goodness of fit evaluation function between the next time of the inflection point time to be solved and the second time. Update the value of the last inflection point based on the inflection point time to obtain the updated value of the last inflection point time; update the value of the kth inflection point based on the value of the kth inflection point to obtain the updated value of the kth inflection point. Based on the updated value of the kth inflection point, re-determine whether to perform backtracking inflection point time calculation until the updated value of the kth inflection point is equal to 1, and calculate the K+1 inflection point times. The formula for the inflection point time is as follows: In the formula, t * For the inflection point time corresponding to the inflection point time t to be solved, g k-1 (u, t) represents the goodness of fit between the first time point and the inflection point to be solved, where u is the first time point, k is the kth inflection point, v is the second time point, and g1(t+1, v) represents the goodness of fit between the next time point after the inflection point to be solved and the second time point.

10. A terminal device, characterized in that, Including the processor and memory; The memory is used to store program code and transmit the program code to the processor; The processor is configured to execute the inflection point identification method for the new energy predicted power time series curve as described in any one of claims 1-6, according to the instructions in the program code.

Images