CN117748501B - A wind power prediction method and system for energy storage assisted black start - Google Patents

Abstract

The present application belongs to the field of wind power technology, and specifically proposes a wind power prediction method and system for energy storage-assisted black start, the method comprising obtaining a wind speed sequence; decomposing the wind speed sequence using the CEEMD algorithm to obtain multiple modal components; obtaining the corresponding wind speed component prediction value based on each modal component using the first recurrent neural network model, and obtaining the error coefficient corresponding to each modal component based on each wind speed component prediction value and each modal component; reconstructing all modal components based on the positive and negative properties of each error coefficient to obtain multiple complementary modal components; obtaining the corresponding target wind speed component prediction value based on each complementary modal component using the second recurrent neural network model; obtaining the wind speed prediction value of the next sampling point of the current sampling point based on each target wind speed component prediction value, and then using the wind speed prediction value to obtain the wind power prediction value so as to use the wind power prediction value to participate in energy storage-assisted black start in the event of a power outage. The method of the present application can improve the prediction accuracy of wind power.

Description

A wind power prediction method and system for energy storage assisted black start

Technical Field

The present application relates to the field of wind power technology, and in particular to a method and system for predicting wind power during energy storage-assisted black start.

Background technique

As the scale of the power grid increases, the consequences of large-scale power outages are becoming more and more serious; due to natural resource constraints, traditional black start power sources in areas with abundant wind and solar power are insufficient. Using wind, solar and storage systems as black start power sources can improve the black start capability of regional power grids. When the wind, solar and storage power generation system is used as a black start power source, the charge/discharge power constraints and power constraints of the energy storage device are considered. During the black start process, when the output of wind farms and photovoltaic power stations is insufficient or fluctuates violently, the energy storage may be overcharged or over-discharged, resulting in the inability to continue to use the energy storage, causing the black start to fail. Therefore, in order to better use the wind, solar and storage power generation system for black start after a power outage, it is necessary to evaluate the wind power of the wind farm's continuous effective output probability based on historical wind speed data. However, traditional wind power forecasting has the problem of poor prediction accuracy.

Summary of the invention

The present application aims to solve one of the technical problems in the related art at least to some extent.

To this end, the first objective of the present application is to propose a wind power prediction method for energy storage-assisted black start to improve the prediction accuracy of wind power.

The second objective of the present application is to propose a wind power prediction system for energy storage-assisted black start.

The third objective of the present application is to provide an electronic device.

A fourth objective of the present application is to provide a computer-readable storage medium.

To achieve the above-mentioned purpose, the first embodiment of the present application proposes a wind power prediction method for energy storage-assisted black start, comprising:

Acquire a wind speed sequence, wherein the wind speed sequence includes wind speed measurement values of a current sampling point and multiple historical sampling points;

Decomposing the wind speed sequence by using CEEMD algorithm to obtain multiple modal components;

Based on each modal component, a corresponding wind speed component prediction value is obtained using a first recurrent neural network model, and based on each wind speed component prediction value and each modal component, an error coefficient corresponding to each modal component is obtained;

Based on the positive and negative properties of each error coefficient, all modal components are reconstructed to obtain multiple complementary modal components;

Based on each complementary modal component, a second recurrent neural network model is used to obtain a corresponding target wind speed component prediction value;

The wind speed prediction value of the next sampling point of the current sampling point is obtained based on the prediction value of each target wind speed component, and the wind power prediction value is then used to obtain the wind speed prediction value, so as to participate in energy storage-assisted black start when a power outage occurs.

In the method of the first aspect of the present application, each modal component is composed of wind speed components obtained by decomposing the wind speed measurement values of all sampling points of the wind speed sequence, and the number of wind speed components is equal to the number of sampling points of the wind speed sequence; the parameters of the first recurrent neural network model are set so that the first recurrent neural network model outputs a preset number of wind speed component prediction values, and the actual values corresponding to the preset number of wind speed component prediction values output by the model are the same number of wind speed components at the current sampling point and before in the input modal component; the method of obtaining the corresponding wind speed component prediction value based on each modal component using the first recurrent neural network model, and obtaining the error coefficient corresponding to each modal component based on each wind speed component prediction value and each modal component, includes: for any modal component, inputting the modal component into the first recurrent neural network model to obtain the corresponding preset number of wind speed component prediction values; obtaining the error coefficient of the modal component based on the preset number of wind speed component prediction values and the corresponding actual values, and then obtaining the error coefficient of each modal component.

In the method of the first aspect of the present application, all modal components are reconstructed based on the positive or negative nature of each error coefficient to obtain a plurality of complementary modal components, including: based on the positive or negative nature of the error coefficient of each modal component, all modal components are divided into a non-negative array and a negative array; according to the error coefficient, the modal components in the non-negative array are arranged from large to small to obtain a target non-negative array, and the modal components in the negative array are arranged from small to large to obtain a target negative array; and the modal components at corresponding positions in the target non-negative array and the target negative array are added to obtain a plurality of complementary modal components.

In the method of the first aspect of the present application, the parameters of the second recurrent neural network model are set so that the second recurrent neural network model outputs a target wind speed component prediction value for the next sampling point of the current sampling point; the method of obtaining the corresponding target wind speed component prediction value based on each complementary modal component using the second recurrent neural network model includes: for any complementary modal component, inputting the complementary modal component into the second recurrent neural network model to obtain the corresponding target wind speed component prediction value for the next sampling point of the current sampling point, and then obtaining the target wind speed component prediction value corresponding to each complementary modal component.

In the method of the first aspect of the present application, obtaining the wind speed prediction value of the next sampling point of the current sampling point based on each target wind speed component prediction value includes: summing the target wind speed component prediction values to obtain the wind speed prediction value of the next sampling point of the current sampling point.

In the method of the first aspect of the present application, the first recurrent neural network model and the second recurrent neural network model respectively adopt GRU models.

In the method of the first aspect of the present application, the error coefficient of each modal component satisfies: , where/> is the error coefficient of the kth modal component, N is the preset number of model outputs corresponding to the kth modal component, yk _,n is the predicted value of the nth wind speed component output by the model corresponding to the kth modal component, and _sk,n is the actual value corresponding to yk _,n .

To achieve the above-mentioned purpose, the second embodiment of the present application proposes a wind power prediction system for energy storage-assisted black start, comprising:

An acquisition module, used to acquire a wind speed sequence, wherein the wind speed sequence includes wind speed measurement values of a current sampling point and multiple historical sampling points;

A decomposition module, used for decomposing the wind speed sequence by using a CEEMD algorithm to obtain multiple modal components;

An error calculation module is used to obtain a corresponding wind speed component prediction value based on each modal component using the first recurrent neural network model, and obtain an error coefficient corresponding to each modal component based on each wind speed component prediction value and each modal component;

A reconstruction module, used for reconstructing all modal components to obtain multiple complementary modal components based on the positive and negative properties of each error coefficient;

A prediction module, used for obtaining a corresponding target wind speed component prediction value using a second recurrent neural network model based on each complementary modal component;

The control module is used to obtain the wind speed prediction value of the next sampling point of the current sampling point based on the prediction value of each target wind speed component, and then use the wind speed prediction value to obtain the wind power prediction value, so as to use the wind power prediction value to participate in energy storage assisted black start when a power outage occurs.

To achieve the above-mentioned purpose, the third aspect embodiment of the present application proposes an electronic device, comprising: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method proposed in the first aspect of the present application.

To achieve the above-mentioned purpose, the fourth aspect embodiment of the present application proposes a computer-readable storage medium, in which computer execution instructions are stored. When the computer execution instructions are executed by a processor, they are used to implement the method proposed in the first aspect of the present application.

The wind power prediction method, system, electronic device and storage medium for energy storage assisted black start provided in the present application obtain a wind speed sequence, which includes wind speed measurement values of a current sampling point and multiple historical sampling points; decompose the wind speed sequence using a CEEMD algorithm to obtain multiple modal components; obtain a corresponding wind speed component prediction value based on each modal component using a first recurrent neural network model, and obtain an error coefficient corresponding to each modal component based on each wind speed component prediction value and each modal component; reconstruct all modal components based on the positive and negative properties of each error coefficient to obtain multiple complementary modal components; obtain a corresponding target wind speed component prediction value based on each complementary modal component using a second recurrent neural network model; obtain a wind speed prediction value of the next sampling point of the current sampling point based on each target wind speed component prediction value, and then use the wind speed prediction value to obtain a wind power prediction value, so that the wind power prediction value can be used to participate in the energy storage assisted black start when a power outage occurs. In this case, after using the CEEMD algorithm to obtain multiple modal components, the recurrent neural network model is also used to obtain the predicted value of the wind speed component and then obtain the error coefficient corresponding to each modal component. Through the positivity and negativity of each error coefficient, all modal components are reconstructed to obtain multiple complementary modal components. The complementary modal components are used to obtain the predicted wind speed value of the next sampling point of the current sampling point, avoiding the use of only the traditional CEEMD algorithm to decompose and obtain components for prediction as in the prior art, reducing the errors of each component, improving the accuracy of wind speed prediction, and thus improving the prediction accuracy of wind power.

Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application.

BRIEF DESCRIPTION OF THE DRAWINGS

The above and/or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

FIG1 is a schematic flow chart of a method for predicting wind power during energy storage-assisted black start provided in an embodiment of the present application;

FIG2 is a schematic diagram of a specific flow chart of a wind speed prediction process provided in an embodiment of the present application;

FIG3 is a block diagram of a wind power prediction system for energy storage-assisted black start provided in an embodiment of the present application.

Detailed ways

The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

The following describes the wind power prediction method and system for energy storage-assisted black start in embodiments of the present application with reference to the accompanying drawings.

The embodiment of the present application provides a wind power prediction method for energy storage-assisted black start to improve the prediction accuracy of wind power. The embodiment of the present application first predicts the wind speed, and then uses the predicted wind speed to obtain the required wind power prediction value. The specific process is as follows.

Fig. 1 is a schematic flow chart of a method for predicting wind power during energy storage-assisted black start provided in an embodiment of the present application. Fig. 2 is a schematic flow chart of a specific wind speed prediction process provided in an embodiment of the present application.

As shown in FIG1 , the energy storage-assisted black start wind power prediction method includes the following steps:

Step S101, obtaining a wind speed sequence, where the wind speed sequence includes wind speed measurement values of a current sampling point and multiple historical sampling points.

Specifically, in step S101, the number of sampling points is set to M, including the current sampling point and M-1 historical sampling points, and the wind speed measurement values collected at the M sampling points are obtained to obtain the required wind speed sequence (also called the original wind speed sequence). In other words, the number of wind speed measurement values in the wind speed sequence is equal to M.

Step S102: Decompose the wind speed sequence using the CEEMD algorithm to obtain multiple modal components.

It is easy to understand that the CEEMD (Complementary Ensemble Empirical Mode Decomposition) algorithm is an adaptive signal processing method. It uses a group of auxiliary white noises with opposite signs and the original signal to obtain a group of new positive and negative mixed sequences, and then performs EMD (Empirical Mode Decomposition) decomposition on each sequence. Through multiple decompositions, the original signal (such as the wind speed sequence) is finally decomposed into a finite number of modal components (Intrinsic Mode Function, IMF). In step S102, the number of modal components obtained by decomposing the wind speed sequence using the CEEMD algorithm is K, where the kth modal component is represented as IMF _k , and k is 1~K.

Since the wind speed sequence includes M wind speed measurement values, each modal component decomposed in step S102 can be regarded as a wind speed component decomposed from the wind speed measurement values of all sampling points of the wind speed sequence. The number of wind speed components is equal to the number of sampling points of the wind speed sequence. That is, the sequence length of each modal component is equal to M, each subsequence in the modal component is a wind speed component, the sampling point corresponding to each modal component is consistent with the acquired wind speed sequence, and each wind speed component is decomposed from the wind speed measurement value at the corresponding sampling point.

Step S103, based on each modal component, using the first recurrent neural network model to obtain the corresponding wind speed component prediction value, and based on each wind speed component prediction value and each modal component, obtaining the error coefficient corresponding to each modal component.

Specifically, in step S103, the parameters of the first recurrent neural network model are set so that the first recurrent neural network model outputs a preset number of wind speed component prediction values, and the actual values corresponding to the preset number of wind speed component prediction values output by the model are the same number of wind speed components in the input modal components at the current sampling point and before; based on each modal component, the corresponding wind speed component prediction value is obtained using the first recurrent neural network model, and based on each wind speed component prediction value and each modal component, the error coefficient corresponding to each modal component is obtained, including: for any modal component, the modal component is input into the first recurrent neural network model to obtain the corresponding preset number of wind speed component prediction values; based on the preset number of wind speed component prediction values and the corresponding actual values, the error coefficient of the modal component is obtained, and then the error coefficient of each modal component is obtained.

Taking the first modal component IMF ₁ , the preset number N, and the sequence length of the modal component equal to M, where M＞N, as an example, the model input is the first modal component IMF ₁ , and the first modal component IMF ₁ includes M wind speed components at the current sampling point and M-1 historical sampling points. The modal component is input into the first recurrent neural network model to obtain N wind speed component prediction values. The actual values corresponding to the N wind speed component prediction values are the wind speed components at the current sampling point and the previous N-1 historical sampling points in the first modal component IMF _1. Based on the N wind speed component prediction values and the corresponding N wind speed components, the error coefficient of the first modal component IMF ₁ is obtained. .

In step S103, the error coefficient of each modal component satisfies: , where/> is the error coefficient of the kth modal component, N is the preset number of model outputs corresponding to the kth modal component, y _k,n is the predicted value of the nth wind speed component output by the model corresponding to the kth modal component, and n ranges from 1 to N. s _k,n is the actual value corresponding to y _k,n .

In step S103, the first recurrent neural network model adopts a gated recurrent unit (GRU) model. It is easy to understand that the GRU network is an improved model of the LSTM (Long Short-Term Memory). By integrating the forget gate and the input gate into an update gate, the training parameters of the network are reduced to a certain extent, while ensuring the memory of effective information. The update gate and reset gate states are z _t and r _t respectively, the input vector is x _t , the hidden layer state is h _t , and the mathematical expression is:

In the formula, is the input state at time t and the hidden state at the previous time/> The process vector of Represents the sigmoid function; /> 、/> Represent the weight matrices of the input vector to the update gate and the reset gate, respectively./> 、/> Respectively represent the weight matrices of the hidden layer state to the update gate and the reset gate;/> represents the weight matrix of hidden layer state to process vector; W represents the weight matrix of input vector to process vector;/> represents the output weight matrix, b is the bias; /> is the output at time t.

As shown in Figure 2, in the stage of establishing the complementary sequence, the original wind speed sequence is decomposed by CEEMD to obtain K modal components IMF (IMF ₁ ,...,IMF _K ). The K IMF components are respectively put into the GRU model for prediction, and the error coefficient of each modal component is obtained by using the predicted value of the wind speed component and the corresponding actual value (i.e. ,...,/> ).

Step S104: reconstruct all modal components based on the positive and negative properties of each error coefficient to obtain a plurality of complementary modal components.

Specifically, in step S104, based on the positive or negative property of each error coefficient, all modal components are reconstructed to obtain a plurality of complementary modal components, including: based on the positive or negative property of the error coefficient of each modal component, all modal components are divided into a non-negative array and a negative array; according to the error coefficient, the modal components in the non-negative array are arranged from large to small to obtain a target non-negative array, and the modal components in the negative array are arranged from small to large to obtain a target negative array; and the modal components at corresponding positions in the target non-negative array and the target negative array are added to obtain a plurality of complementary modal components.

Taking the case where the number of modal components in the non-negative array is L and the number of modal components in the negative array is KL as an example, the K error coefficients are divided into two groups according to their positive and negative signs, where L error coefficients are greater than or equal to 0 and KL error coefficients are less than 0. Then, the modal components corresponding to the error coefficients greater than or equal to 0 are arranged from large to small according to the error coefficients to obtain the target non-negative array [IMF _max-L+1 ,…,IMF _max-2 ,IMF _max-1 ,IMF _max ]. The modal components corresponding to the error coefficients less than 0 are arranged from small to large according to the error coefficients to obtain the target negative array [IMF _min-K+L+1 ,…,IMF _min-2 ,IMF _min-1 ,IMF _min ]. Then, the IMFs at corresponding positions in these two groups are added in pairs until the group with the smaller sequence length is added, so as to form multiple complementary modal components [ _H1 , _H2 ,…, _Hmax(L,KL) ] (see Figure 2). When adding two by two, the addition starts from the starting position of each group (that is, starting from IMF _max-L+1 +IMF _min-K+L+1 ) until the group with a smaller sequence length is added, and the remaining modal components in the group with a larger sequence length are directly retained.

Step S105, obtaining the corresponding target wind speed component prediction value using the second recurrent neural network model based on each complementary modal component.

Specifically, in step S105, the parameters of the second recurrent neural network model are set so that the second recurrent neural network model outputs a target wind speed component prediction value of the next sampling point of the current sampling point; based on each complementary modal component, the second recurrent neural network model is used to obtain the corresponding target wind speed component prediction value of the next sampling point of the current sampling point, including: for any complementary modal component, the complementary modal component is input into the second recurrent neural network model to obtain the corresponding target wind speed component prediction value of the next sampling point of the current sampling point, and then the target wind speed component prediction value corresponding to each complementary modal component is obtained.

In step S105, the second recurrent neural network model may adopt a GRU model. As shown in FIG2, in the prediction stage, the multiple complementary modal components [ _H1 , _H2 , ..., _{Hmax (L, KL)} ] obtained in step S104 are respectively input into the GRU model to obtain the target wind speed component prediction values (i.e., _Y1 ^' , _Y2 ^' , ..., _{Ymax (L, KL)} ^' ) of the corresponding complementary modal components.

Step S106, based on the predicted values of each target wind speed component, the predicted wind speed value of the next sampling point of the current sampling point is obtained, and then the predicted wind speed value is used to obtain the predicted wind power value, so as to use the predicted wind power value to participate in energy storage assisted black start when a power outage occurs.

Specifically, in step S106, the wind speed prediction value of the next sampling point of the current sampling point is obtained based on each target wind speed component prediction value, including: summing each target wind speed component prediction value to obtain the wind speed prediction value of the next sampling point of the current sampling point. As shown in FIG2, all target wind speed component prediction values (i.e., Y ₁ ^' , Y ₂ ^' , ..., Y _max(L,KL) ^' ) are summed to obtain the final result of wind speed prediction, which is the wind speed prediction value of the next sampling point of the current sampling point.

In step S106, using the wind speed prediction value to obtain the wind power prediction value includes: obtaining the wind power prediction value based on the wind speed prediction value, the rated output power of the wind turbine, the rated wind speed, the cut-in wind speed and the cut-out wind speed. The relationship between wind speed and wind power (i.e., the power generation power of the wind turbine) satisfies:

Wherein, P _r is the rated output power of the wind turbine (kW); v is the wind speed at the height of the wind turbine hub (m/s), v _r is the rated wind speed (m/s); v _ci is the cut-in wind speed (m/s), and v _co is the cut-out wind speed (m/s).

In step S106, the predicted wind speed value is regarded as the wind speed v at the height of the wind turbine hub, and the predicted wind power value P _w is obtained according to the relationship between wind speed and wind power.

In order to verify the effect of the method of the present application, a verification is performed. Among them, the experimental data obtained (i.e., the original wind speed sequence) comes from a wind farm with energy storage assisted black start capability. From September 1, 2022 to September 14, 2016, sampled wind speed data is obtained every 15 minutes. For each group of cases, the wind speed sequence is divided into a training set and a test set. Therefore, 7 days of data are selected, and a total of 672 data samples at intervals of 15 minutes are provided to train the prediction model; the next 96 data (corresponding to 1 day of data) are used to test the performance of the proposed model.

In order to verify the performance of the model proposed in this application, experiments are conducted to further evaluate the proposed model. Among them, GRU and CEEMD-GRU are called benchmark models, which are used to compare with the model proposed in this application. In the experiments of this application, mean absolute percentage error (MAPE) and root mean square error (RMSE) are selected as the evaluation criteria for each model. In the experiments of this application, the experimental results are shown in Table 1.

Table 1 Evaluation index table of each model

As shown in Table 1, when compared with all other benchmark models, the model proposed in this application has the smallest error coefficient. Compared with the CEEMD-GRU model, the error coefficients of the two evaluation indicators of the model of this application have been greatly reduced, proving that the method proposed in this application can effectively improve the prediction performance. Compared with the GRU model of this application, the error coefficients of the two evaluation indicators have been greatly reduced, proving that data preprocessing can improve the prediction accuracy.

In order to implement the above embodiments, the present application also proposes a wind power prediction system for energy storage-assisted black start.

FIG3 is a block diagram of a wind power prediction system for energy storage-assisted black start provided in an embodiment of the present application.

As shown in FIG3 , the energy storage assisted black start wind power prediction system includes an acquisition module 11, a decomposition module 12, an error calculation module 13, a reconstruction module 14, a prediction module 15 and a control module 16, wherein:

An acquisition module 11 is used to acquire a wind speed sequence, where the wind speed sequence includes wind speed measurement values of a current sampling point and multiple historical sampling points;

A decomposition module 12 is used to decompose the wind speed sequence using a CEEMD algorithm to obtain multiple modal components;

The error calculation module 13 is used to obtain the corresponding wind speed component prediction value based on each modal component using the first recurrent neural network model, and obtain the error coefficient corresponding to each modal component based on each wind speed component prediction value and each modal component;

A reconstruction module 14, configured to reconstruct all modal components based on the positive and negative properties of each error coefficient to obtain a plurality of complementary modal components;

A prediction module 15, configured to obtain a corresponding target wind speed component prediction value based on each complementary modal component using a second recurrent neural network model;

The control module 16 is used to obtain the wind speed prediction value of the next sampling point of the current sampling point based on the predicted values of each target wind speed component, and then use the wind speed prediction value to obtain the wind power prediction value, so as to use the wind power prediction value to participate in energy storage assisted black start when a power outage occurs.

Furthermore, in a possible implementation of an embodiment of the present application, each modal component is composed of wind speed components obtained by decomposing the wind speed measurement values of all sampling points of the wind speed sequence, the number of wind speed components is equal to the number of sampling points of the wind speed sequence, and the parameters of the first recurrent neural network model are set so that the first recurrent neural network model outputs a preset number of wind speed component prediction values, and the actual values corresponding to the preset number of wind speed component prediction values output by the model are the same number of wind speed components in the input modal components at the current sampling point and before; the error calculation module 13 is specifically used to: for any modal component, input the modal component into the first recurrent neural network model to obtain the corresponding preset number of wind speed component prediction values; obtain the error coefficient of the modal component based on the preset number of wind speed component prediction values and the corresponding actual values, and then obtain the error coefficient of each modal component.

Furthermore, in a possible implementation of an embodiment of the present application, the first recurrent neural network model and the second recurrent neural network model respectively adopt GRU models.

Further, in a possible implementation of the embodiment of the present application, in the error calculation module 13, the error coefficient of each modal component satisfies: , where/> is the error coefficient of the kth modal component, N is the preset number of model outputs corresponding to the kth modal component, yk _,n is the predicted value of the nth wind speed component output by the model corresponding to the kth modal component, and _sk,n is the actual value corresponding to yk _,n .

Furthermore, in a possible implementation of an embodiment of the present application, the reconstruction module 14 is specifically used to: divide all modal components into a non-negative array and a negative array based on the positive or negative nature of the error coefficient of each modal component; arrange the modal components in the non-negative array from large to small according to the error coefficient to obtain a target non-negative array, and arrange the modal components in the negative array from small to large to obtain a target negative array; add the modal components at corresponding positions in the target non-negative array and the target negative array to obtain multiple complementary modal components.

Furthermore, in a possible implementation of an embodiment of the present application, the parameters of the second recurrent neural network model are set so that the second recurrent neural network model outputs a predicted value of the target wind speed component of the next sampling point of the current sampling point; the prediction module 15 is specifically used to: for any complementary modal component, input the complementary modal component into the second recurrent neural network model to obtain the corresponding predicted value of the target wind speed component of the next sampling point of the current sampling point, and then obtain the predicted value of the target wind speed component corresponding to each complementary modal component.

Furthermore, in a possible implementation of the embodiment of the present application, the control module 16 is specifically configured to: sum the predicted values of each target wind speed component to obtain a predicted wind speed value of a sampling point next to the current sampling point.

It should be noted that the aforementioned explanation of the embodiment of the wind power prediction method for energy storage-assisted black start is also applicable to the wind power prediction system for energy storage-assisted black start of this embodiment, and will not be repeated here.

In an embodiment of the present application, a wind speed sequence is obtained, and the wind speed sequence includes wind speed measurement values of a current sampling point and multiple historical sampling points; the wind speed sequence is decomposed using a CEEMD algorithm to obtain multiple modal components; a corresponding wind speed component prediction value is obtained based on each modal component using a first recurrent neural network model, and an error coefficient corresponding to each modal component is obtained based on each wind speed component prediction value and each modal component; based on the positive and negative properties of each error coefficient, all modal components are reconstructed to obtain multiple complementary modal components; based on each complementary modal component, a corresponding target wind speed component prediction value is obtained using a second recurrent neural network model; based on each target wind speed component prediction value, a wind speed prediction value of the next sampling point of the current sampling point is obtained, and then a wind power prediction value is obtained using the wind speed prediction value, so that the wind power prediction value can be used to participate in energy storage-assisted black start when a power outage occurs. In this case, after using the CEEMD algorithm to obtain multiple modal components, the recurrent neural network model is also used to obtain the wind speed component prediction value and then obtain the error coefficient corresponding to each modal component. Through the positive and negative properties of each error coefficient, all modal components are reconstructed to obtain multiple complementary modal components. The wind speed prediction value of the next sampling point of the current sampling point is obtained by using the complementary modal components, avoiding the use of only the traditional CEEMD algorithm to decompose and obtain components for prediction as in the prior art, reducing the error of each component, improving the accuracy of wind speed prediction, and thus improving the prediction accuracy of wind power. The method of the present application solves the problem that the subsequence decomposed by the traditional data decomposition technology (CEEMD) has large errors, which will affect the prediction, and the use of a combined model prediction will increase the running time, thereby improving the prediction accuracy of wind power.

In order to implement the above embodiments, the present application also proposes an electronic device, comprising: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method provided by the above embodiments.

In order to implement the above embodiments, the present application also proposes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the methods provided by the above embodiments.

In order to implement the above embodiments, the present application also proposes a computer program product, including a computer program, which implements the methods provided by the above embodiments when executed by a processor.

In the description of the aforementioned embodiments, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

The logic and/or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

The storage medium mentioned above may be a read-only memory, a disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims (9)

1. The wind power prediction method for energy storage auxiliary black start is characterized by comprising the following steps of:

Acquiring a wind speed sequence, wherein the wind speed sequence comprises wind speed measurement values of a current sampling point and a plurality of historical sampling points;

decomposing the wind speed sequence by using a CEEMD algorithm to obtain a plurality of modal components;

Obtaining a corresponding wind speed component predicted value by using a first cyclic neural network model based on each modal component, and obtaining an error coefficient corresponding to each modal component based on each wind speed component predicted value and each modal component;

Reconstructing all modal components based on the positive and negative of each error coefficient to obtain a plurality of complementary modal components;

Obtaining a corresponding target wind speed component predicted value by using a second cyclic neural network model based on each complementary modal component;

Obtaining a wind speed predicted value of a next sampling point of the current sampling point based on each target wind speed component predicted value, and further obtaining a wind power predicted value by utilizing the wind speed predicted value so as to participate in energy storage auxiliary black start when a power failure occurs;

wherein reconstructing all modal components based on the positive and negative of each error coefficient to obtain a plurality of complementary modal components comprises:

dividing all modal components into a non-negative array and a negative array based on the positive and negative of error coefficients of each modal component;

According to the error coefficient, arranging the modal components in the non-negative array from large to small to obtain a target non-negative array, and arranging the modal components in the negative array from small to large to obtain a target negative array;

And adding the modal components of the corresponding positions in the target non-negative array and the target negative array to obtain a plurality of complementary modal components.

2. The energy storage assisted black start wind power prediction method according to claim 1, wherein each modal component consists of wind speed components obtained by decomposing wind speed measurement values of all sampling points of the wind speed sequence, and the number of wind speed components is equal to the number of sampling points of the wind speed sequence;

Setting parameters of a first cyclic neural network model, so that the first cyclic neural network model outputs a preset number of wind speed component predicted values, and the actual values corresponding to the preset number of wind speed component predicted values output by the model are the current sampling points in the input modal components and the same number of wind speed components before the current sampling points;

The method for obtaining the corresponding wind speed component predicted value based on each modal component by using the first cyclic neural network model, obtaining the error coefficient corresponding to each modal component based on each wind speed component predicted value and each modal component comprises the following steps:

inputting any modal component into a first cyclic neural network model to obtain a corresponding preset number of wind speed component predicted values; and obtaining error coefficients of the modal components based on the predicted values of the wind speed components of the preset quantity and the corresponding actual values, and further obtaining the error coefficients of the modal components.

3. The energy storage-assisted black-start wind power prediction method according to claim 1, wherein parameters of the second cyclic neural network model are set so that the second cyclic neural network model outputs a target wind speed component predicted value of a sampling point next to the current sampling point;

The obtaining a corresponding target wind speed component predicted value based on each complementary modal component by using a second cyclic neural network model comprises the following steps:

And inputting the complementary modal component into a second cyclic neural network model for any complementary modal component to obtain a target wind speed component predicted value of a next sampling point of the corresponding current sampling point, and further obtaining the target wind speed component predicted value corresponding to each complementary modal component.

4. A method of energy storage assisted black start wind power prediction according to claim 3, wherein the obtaining a wind speed prediction value for a next sampling point to a current sampling point based on each target wind speed component prediction value comprises:

and summing the predicted values of the target wind speed components to obtain the predicted value of the wind speed of the next sampling point of the current sampling point.

5. The energy storage assisted black start wind power prediction method of claim 1, wherein the first and second recurrent neural network models each employ a GRU model.

6. The energy storage assisted black start wind power prediction method according to claim 2, wherein error coefficients of each modal component satisfy: wherein/> For the error coefficient of the kth modal component, N is the preset number of model outputs corresponding to the kth modal component, y _k,n is the nth wind speed component predicted value of the model outputs corresponding to the kth modal component, and s _k,n is the actual value corresponding to y _k,n.

7. An energy storage assisted black start wind power prediction system, comprising:

the system comprises an acquisition module, a control module and a control module, wherein the acquisition module is used for acquiring a wind speed sequence, and the wind speed sequence comprises wind speed measurement values of a current sampling point and a plurality of historical sampling points;

The decomposition module is used for decomposing the wind speed sequence by using a CEEMD algorithm to obtain a plurality of modal components;

the error calculation module is used for obtaining a corresponding wind speed component predicted value by utilizing the first cyclic neural network model based on each modal component and obtaining an error coefficient corresponding to each modal component based on each wind speed component predicted value and each modal component;

The reconstruction module is used for reconstructing all the modal components based on the positive and negative of each error coefficient to obtain a plurality of complementary modal components;

The prediction module is used for obtaining a corresponding target wind speed component predicted value by using a second cyclic neural network model based on each complementary modal component;

the control module is used for obtaining a wind speed predicted value of a next sampling point of the current sampling point based on each target wind speed component predicted value, and further obtaining a wind power predicted value by utilizing the wind speed predicted value so as to participate in energy storage auxiliary black start when a power failure occurs;

wherein reconstructing all modal components based on the positive and negative of each error coefficient to obtain a plurality of complementary modal components comprises:

dividing all modal components into a non-negative array and a negative array based on the positive and negative of error coefficients of each modal component;

According to the error coefficient, arranging the modal components in the non-negative array from large to small to obtain a target non-negative array, and arranging the modal components in the negative array from small to large to obtain a target negative array;

And adding the modal components of the corresponding positions in the target non-negative array and the target negative array to obtain a plurality of complementary modal components.

8. An electronic device, comprising: a processor, and a memory communicatively coupled to the processor;

The memory stores computer-executable instructions;

the processor executes computer-executable instructions stored in the memory to implement the method of any one of claims 1-6.

9. A computer readable storage medium having stored therein computer executable instructions which when executed by a processor are adapted to carry out the method of any one of claims 1-6.