Method, device and computer equipment for generating new energy output scene
By preprocessing the new energy historical power data and applying the LSTM model, a more accurate new energy output scenario prediction curve is generated, which solves the problem of low reliability of prediction scenarios in traditional methods and improves the accuracy of prediction.
Patent Information
- Application Number
- CN202310580416.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-22
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2043-05-22
AI Technical Summary
In the prior art, when traditional probability statistical methods generate new energy output scenarios, there is a large error between the prediction scenario and the real scenario, the predicted scenario is low, and it is difficult to capture the nonlinear relationship between big data.
By obtaining the historical power data of new energy, preprocessing is performed to extract the low-frequency and high-frequency partial curves, the prediction curve is generated using the long and short-term memory network LSTM (LSTM), and the high-frequency partial curves are superimposed to represent the multiple output scenarios of new energy sources. target prediction curve.
It improves the reliability of new energy output scenario prediction, can better capture the timing relationship between data, and reduce prediction errors.
Smart Images

Figure CN116805782B_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the technical field of power systems, and more specifically, to a method, device, computer equipment and readable storage medium for generating a new energy output scenario. Background Art
[0002] As global energy demand continues to grow, the large-scale development of traditional energy has led to increasingly prominent problems such as resource shortages, environmental pollution, and climate change. Therefore, the proportion of new energy has gradually increased. The intermittent and random nature of new energy output seriously affects the stable operation of the power system, and has an important impact on the future planning of the power system, the preparation of annual plans, and the calculation of the flexibility demand of the power system. How to describe the uncertainty of renewable energy output is the key to overcoming the above problems.
[0003] Existing methods mainly use traditional probability and statistics methods to generate multiple typical scenarios that may occur in the future, analyze each scenario, and then carry out planning and demand estimation; however, the applicant recognizes that the essence of traditional probability and statistics methods is to model the joint probability density distribution in random processes, which can only process data with smaller dimensions and have weak processing capabilities for nonlinear mapping problems. It is difficult to capture the nonlinear relationship between big data, resulting in a large error between the generated prediction scenarios and the real scenarios, and the reliability of the prediction scenarios is low. Summary of the invention
[0004] In view of this, the present invention provides a method, device, computer equipment and readable storage medium for generating new energy output scenarios, the main purpose of which is to solve the problem that there is a large error between the predicted scenarios generated by traditional probability statistics methods and the actual scenarios, and the reliability of the predicted scenarios is low.
[0005] According to the first aspect of the present application, a method for generating a new energy output scenario is provided, comprising:
[0006] Acquire historical power data of new energy, preprocess the historical power data, and extract a curve of a target low-frequency part of the power generated by new energy varying over time and a curve of a target high-frequency part of the power generated by new energy varying over time;
[0007] Based on the curve of the target low-frequency part of the new energy power generation changing over time, a plurality of prediction curves are obtained by using a long short-term memory network LSTM, wherein the prediction curve is a curve of the new energy power generation predicted to change over time;
[0008] The multiple prediction curves are superimposed using the curve of the target high-frequency part changing with time to obtain multiple target prediction curves for representing the new energy output scenario.
[0009] Optionally, the preprocessing of the historical power data to extract a curve of a target low-frequency part of the renewable energy power generation power varying over time and a curve of a target high-frequency part of the renewable energy power generation power varying over time includes:
[0010] Determine a first expression of a discrete wavelet sequence and a second expression of discrete wavelet transform coefficients, wherein the first expression and the second expression are respectively:
[0011] Formula 1:
[0012] Among them, j represents the jth decomposition level, k represents the parameter used to complete the translation operation in wavelet transform, and a 0 is the lowest resolution low-frequency information, b 0 is the lowest resolution detail information, ψ is the wavelet mother function, t is the time, ψ j,k (t) is the discrete wavelet sequence;
[0013] Formula 2:
[0014] Among them, C j,k is the discrete wavelet transform coefficient, and f(t) is the signal to be analyzed;
[0015] A reconstruction formula is constructed based on the first expression and the second expression, and the reconstruction formula is expressed as:
[0016] Formula 3:
[0017] Where C is a constant and f'(t) is a discrete signal;
[0018] The historical power data is subjected to wavelet decomposition to obtain a decomposition formula, which is:
[0019] Formula 4:
[0020] Among them, A n is the low-frequency part of the nth decomposition layer, D l is the high frequency part of the lth sub-level, l = 1, 2, ..., n, S is the original discrete signal;
[0021] Based on the decomposition formula, the reconstruction formula is subjected to discrete wavelet transform to obtain a discrete wavelet reconstruction function, which is expressed as:
[0022] Formula 5:
[0023] Wherein, f”(t) is the discrete wavelet reconstruction function, d j,k = <f(t),ψ j,k(t)> is the wavelet coefficient, φ j,k (t) is the scaling function;
[0024] The number of decomposition layers is determined, and based on the number of decomposition layers and the discrete wavelet reconstruction function, a curve of a target low-frequency part of the new energy power generation varying with time and a curve of a target high-frequency part of the new energy power generation varying with time are determined.
[0025] Optionally, performing discrete wavelet transform on the reconstruction formula based on the decomposition formula to obtain a discrete wavelet reconstruction function includes:
[0026] The reconstruction function is projected onto the low-frequency part of the decomposition layer j to obtain the overview signal at the j scale. The overview signal at the j scale is expressed as:
[0027] Formula 6:
[0028] Among them, c j,k is the scale factor, and c j,k = <f(t),φ j,k (t)>, is the overview signal at the j scale;
[0029] The reconstruction function is projected onto the high frequency part of the decomposition layer j to obtain the detail signal at the j scale, wherein the detail signal at the j scale is expressed as:
[0030] Formula 7:
[0031] in, is the detail signal at the j scale;
[0032] The number of decomposition levels is determined, and the discrete wavelet reconstruction function is determined based on the number of decomposition levels, the overview signal at the j scale, and the detail signal at the j scale.
[0033] Optionally, the curve based on the change of the target low-frequency part of the renewable energy power generation power over time uses a long short-term memory network LSTM to obtain multiple prediction curves, including:
[0034] Using the long short-term memory network LSTM, predicting the time-varying curve of the target low-frequency part of the renewable energy power generation power, and obtaining a probability distribution curve of the predicted power error;
[0035] Based on the curve of the target low-frequency part of the new energy power generation power changing with time, the probability distribution curve of the predicted power is determined, and the long short-term memory network LSTM is used to generate the multiple prediction curves according to the probability distribution curve of the predicted power and the probability distribution curve of the predicted power error.
[0036] Optionally, the using the long short-term memory network LSTM to predict the curve of the target low-frequency part of the renewable energy power generation power changing over time to obtain a probability distribution curve of the predicted power error includes:
[0037] Extracting a plurality of points from a curve of the target low-frequency part of the renewable energy power generation changing with time, and determining a characteristic value of each of the plurality of points, wherein the characteristic value is one of a valley value, a peak value and a slope of the curve;
[0038] The characteristic value of each point in the multiple points is input into the long short-term memory network LSTM to obtain the probability distribution curve of the predicted power error.
[0039] Optionally, the determining of a probability distribution curve of predicted power based on a curve of a target low-frequency part of the renewable energy power generation power varying with time comprises:
[0040] Extracting a plurality of points from a curve of the target low-frequency part of the renewable energy power generation changing with time, and determining a characteristic value of each of the plurality of points, wherein the characteristic value is one of a valley value, a peak value, and a line slope;
[0041] The eigenvalue of each point in the plurality of points is fitted using the matrix laboratory MATLAB tool to obtain a probability distribution curve of the predicted power.
[0042] Optionally, the generating the multiple prediction curves using the long short-term memory network LSTM according to the probability distribution curve of the predicted power and the probability distribution curve of the predicted power error includes:
[0043] Randomly sampling the probability distribution curve of the predicted power to obtain the probability of the predicted power corresponding to each first input point among the plurality of first input points;
[0044] Randomly sampling the probability distribution curve of the predicted power error to obtain the probability of the predicted power error corresponding to each second input point among the plurality of second input points;
[0045] The probability of the predicted power corresponding to each first input point in the multiple first input points and the probability of the predicted power error corresponding to each second input point in the multiple second input points are input into a long short-term memory network LSTM to generate the multiple prediction curves.
[0046] According to the second aspect of the present application, a device for generating a new energy output scene is provided, comprising:
[0047] A processing module, used to obtain historical power data of new energy, pre-process the historical power data, and extract a curve of a target low-frequency part of the power generated by the new energy over time and a curve of a target high-frequency part of the power generated by the new energy over time;
[0048] An acquisition module is used to obtain multiple prediction curves using a long short-term memory network (LSTM) based on a curve of a target low-frequency part of the power generation of the new energy source changing over time, wherein the prediction curve is a curve of the power generation of the new energy source changing over time;
[0049] The superposition module is used to superimpose the multiple prediction curves using the curve of the target high-frequency part changing with time, so as to obtain multiple superimposed curves as multiple target prediction curves for representing the new energy output scenario.
[0050] According to a third aspect of the present application, a computer device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the methods described in the first aspect when executing the computer program.
[0051] According to a fourth aspect of the present application, a readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods described in the first aspect are implemented.
[0052] By means of the above technical scheme, the present application provides a method for generating a new energy output scenario, pre-processes the acquired historical power data of new energy, extracts the curve of the target low-frequency part of the new energy power generation changing with time and the curve of the target high-frequency part changing with time, and then uses the long short-term memory network LSTM to obtain multiple prediction curves based on the curve of the target low-frequency part of the new energy power generation changing with time, and finally uses the curve of the target high-frequency part changing with time to superimpose the multiple prediction curves to obtain multiple target prediction curves for representing the new energy output scenario; the long short-term memory network LSTM can effectively record data from a period of time before, and can capture the temporal relationship between the output data of new energy, so that the generated scenario effect is better and the reliability of the prediction scenario is improved.
[0053] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0054] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:
[0055] Figure 1 A flow chart of a method for generating a new energy output scenario provided in an embodiment of the present application is shown.
[0056] Figure 2 A parsing diagram of a long short-term memory network LSTM of a method for generating a new energy output scenario provided in an embodiment of the present application is shown.
[0057] Figure 3 A schematic diagram of a wind power generation power prediction curve for a method for generating a new energy output scenario provided in an embodiment of the present application is shown.
[0058] Figure 4 A schematic diagram of multiple wind power output curves of a method for generating a new energy output scenario provided in an embodiment of the present application is shown.
[0059] Figure 5 A schematic diagram of the structure of a device for generating a new energy output scenario provided in an embodiment of the present application is shown.
[0060] Figure 6 A schematic diagram of the device structure of a computer device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION
[0061] Various aspects and features of the present application are described herein with reference to the accompanying drawings.
[0062] It should be understood that various modifications may be made to the embodiments of the present application. Therefore, the above description should not be considered as limiting, but only as an example of an embodiment. Other modifications within the scope and spirit of the present application will occur to those skilled in the art.
[0063] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments of the present application and, together with the general description of the present application given above and the detailed description of the embodiments given below, serve to explain the principles of the present application.
[0064] These and other characteristics of the present application will become apparent from the following description of a preferred form of embodiment given as a non-limiting example with reference to the accompanying drawings.
[0065] It should also be understood that although the present application has been described with reference to some specific examples, those skilled in the art will be able to readily implement many other equivalent forms of the present application.
[0066] The above and other aspects, features and advantages of the present application will become more apparent in view of the following detailed description when taken in conjunction with the accompanying drawings.
[0067] Specific embodiments of the present application are described hereinafter with reference to the accompanying drawings; however, it should be understood that the embodiments applied for are merely examples of the present application, which may be implemented in a variety of ways. Well-known and / or repeated functions and structures are not described in detail to avoid unnecessary or redundant details that obscure the present application. Therefore, the specific structural and functional details applied for herein are not intended to be limiting, but merely serve as a basis and representative basis for the claims to teach those skilled in the art to use the present application in a variety of ways with substantially any suitable detailed structure.
[0068] This specification may use the phrases "in one embodiment," "in another embodiment," "in yet another embodiment," or "in other embodiments," all of which may refer to one or more of the same or different embodiments according to the present application.
[0069] The present application embodiment provides a method for generating a new energy output scenario, such as Figure 1 As shown, including:
[0070] 101. Obtain historical power data of new energy, pre-process the historical power data, and extract the curve of the target low-frequency part of the new energy power generation power changing with time and the curve of the target high-frequency part changing with time.
[0071] 102. Based on the curve of the target low-frequency part of the renewable energy power generation changing with time, a long short-term memory network (LSTM) is used to obtain multiple prediction curves.
[0072] In the embodiment of the present application, the prediction curve is a curve showing the change of the power generation of the new energy over time.
[0073] 103. Multiple prediction curves are superimposed using the curve of the target high-frequency part changing with time to obtain multiple target prediction curves for representing the new energy output scenario.
[0074] The method provided in the embodiment of the present application pre-processes the historical power data of the acquired new energy, extracts the curve of the target low-frequency part of the new energy power generation changing with time and the curve of the target high-frequency part changing with time, then uses the long short-term memory network LSTM to obtain multiple prediction curves based on the curve of the target low-frequency part of the new energy power generation changing with time, and finally uses the curve of the target high-frequency part changing with time to superimpose the multiple prediction curves to obtain multiple target prediction curves for representing the new energy output scenario; the long short-term memory network LSTM can effectively record the data of a previous period of time, and can capture the temporal relationship between the output data of the new energy, so that the generated scenario effect is better and the reliability of the predicted scenario is improved.
[0075] Further, as a refinement and extension of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, the embodiment of the present application provides another method for generating a new energy output scenario, the method comprising:
[0076] 201. Obtain historical power data of new energy, pre-process the historical power data, and extract a curve of a target low-frequency part of the power generation of new energy varying with time and a curve of a target high-frequency part of the power generation of new energy varying with time.
[0077] In the embodiment of the present application, after acquiring the historical power data of the new energy, a first expression of the discrete wavelet sequence and a second expression of the discrete wavelet transform coefficient are determined, and the first expression and the second expression are respectively:
[0078] Formula 1:
[0079] Among them, j represents the jth decomposition level, k represents the parameter used to complete the translation operation in wavelet transform, and a 0 is the lowest resolution low-frequency information, b 0 is the lowest resolution detail information, ψ is the wavelet mother function, t is the time, ψ j,k (t) is a discrete wavelet sequence;
[0080] Formula 2:
[0081] Among them, C j,k is the discrete wavelet transform coefficient, and f(t) is the signal to be analyzed;
[0082] A reconstruction formula is constructed based on the first expression and the second expression. The reconstruction formula is expressed as:
[0083] Formula 3:
[0084] Where C is a constant and f'(t) is a discrete signal;
[0085] Perform wavelet decomposition on the historical power data and obtain the decomposition formula:
[0086] Formula 4:
[0087] Among them, A n is the low-frequency part of the nth decomposition layer, D l is the high frequency part of the lth sub-level, l = 1, 2, ..., n, S is the original discrete signal; it should be noted that, if Figure 2 As shown, after discrete wavelet transform, the original discrete signal will be decomposed into the sum of multiple D signals and A signals.
[0088] Based on the decomposition formula, the reconstruction formula is subjected to discrete wavelet transform to obtain a discrete wavelet reconstruction function, which is expressed as:
[0089] Formula 5:
[0090] Among them, f”(t) is the discrete wavelet reconstruction function, d j,k = <f(t),ψ j,k (t)> is the wavelet coefficient, φ j,k (t) is the scaling function;
[0091] The number of decomposition layers is determined, and based on the number of decomposition layers and the discrete wavelet reconstruction function, a curve of a target low-frequency part of the renewable energy power generation varying with time and a curve of a target high-frequency part varying with time are determined.
[0092] Furthermore, based on the decomposition formula, the reconstruction formula is subjected to discrete wavelet transform to obtain a discrete wavelet reconstruction function. The specific steps are as follows:
[0093] Project the reconstruction function to the low-frequency part of the decomposition layer j to obtain the overview signal at scale j. The overview signal at scale j is expressed as:
[0094] Formula 6:
[0095] Among them, c j,k is the scale factor, and c j,k = <f(t),φ j,k (t)>, is the general signal at scale j;
[0096] Project the reconstruction function to the high-frequency part of the decomposition layer j to obtain the detail signal at the j scale, where the detail signal at the j scale is expressed as:
[0097] Formula 7:
[0098] in, is the detail signal at scale j;
[0099] The number of decomposition levels is determined, and a discrete wavelet reconstruction function is determined based on the number of decomposition levels, the overview signal at the j scale, and the detail signal at the j scale.
[0100] 202. Based on the curve of the target low-frequency part of the renewable energy power generation power changing with time, a long short-term memory network (LSTM) is used to obtain multiple prediction curves.
[0101] In the embodiment of the present application, the curve of the target low-frequency part changing with time is predicted using the long short-term memory network LSTM, that is, firstly, multiple inputs are determined, for example, 10 consecutive days are one input. The long short-term memory network LSTM is a time recurrent neural network, and its analysis diagram is as follows: Figure 3 As shown in Figure 1, three types of gates are introduced, namely input gate, forget gate and output gate, as well as memory cells with the same shape as the hidden state. Specifically, assuming that the number of hidden units is h, given a small batch input Xt∈Rn×d (the number of samples is n, the number of inputs is d) at time step t and the hidden state Ht-1∈Rn×h at the previous time step. The input gate It∈Rn×h, forget gate Ft∈Rn×h and output gate Ot∈Rn×h at time step t are calculated as follows:
[0102] It=σ(XtWxi+Ht-1Whi+bi),
[0103] Ft=σ(XtWxf+Ht-1Whf+bf),
[0104] Ot=σ(XtWxo+Ht-1Who+bo),
[0105] Among them, Wxi∈Rd×h, Wxf∈Rd×h, Wxo∈Rd×h, Whi∈Rh×h, Whf∈Rh×h, Who∈Rh×h, Wxi, Wxf, Wxo as well as Whi, Whf, Who are all weight parameters, bi∈R1×h, bf∈R1×h, bo∈R1×h, bi, bf, bo are all bias parameters.
[0106] Specifically, firstly, the curve of the target low-frequency part of the renewable energy power generation changing with time is predicted by using the long short-term memory network LSTM to obtain the probability distribution curve of the predicted power error. Then, the probability distribution curve of the predicted power is determined based on the curve of the target low-frequency part of the renewable energy power generation changing with time. And according to the probability distribution curve of the predicted power and the probability distribution curve of the predicted power error, the long short-term memory network LSTM is used to generate multiple prediction curves, wherein the prediction curve is the curve of the predicted renewable energy power generation changing with time. For example, the prediction curve of wind power generation is as follows: Figure 3As shown, the ordinate is the predicted power of wind power generation, and the abscissa is time.
[0107] Furthermore, the long short-term memory network LSTM is used to predict the curve of the target low-frequency part of the renewable energy power generation power changing with time, and obtain the probability distribution curve of the predicted power error. The specific steps are: first, extract multiple points from the curve of the target low-frequency part of the renewable energy power generation power changing with time, and determine the characteristic value of each of the multiple points, where the characteristic value is one of the valley value, the peak value and the slope of the curve; then input the characteristic value of each of the multiple points into the long short-term memory network LSTM to obtain the probability distribution curve of the predicted power error.
[0108] Furthermore, based on the curve of the target low-frequency part of the renewable energy power generation power changing with time, the probability distribution curve of the predicted power is determined. The specific steps are: first, multiple points are extracted from the curve of the target low-frequency part of the renewable energy power generation power changing with time, and the eigenvalue of each of the multiple points is determined, and the eigenvalue is one of the valley value, the peak value and the slope of the line. Then, the Matrix Laboratory MATLAB tool is used to fit the eigenvalue of each of the multiple points to obtain the probability distribution curve of the predicted power.
[0109] Furthermore, a long short-term memory network LSTM is used to generate multiple prediction curves based on the probability distribution curve of the predicted power and the probability distribution curve of the predicted power error. The specific steps are: randomly sampling the probability distribution curve of the predicted power to obtain the probability of the predicted power corresponding to each first input point among the multiple first input points, then randomly sampling the probability distribution curve of the predicted power error to obtain the probability of the predicted power error corresponding to each second input point among the multiple second input points, and then inputting the probability of the predicted power corresponding to each first input point among the multiple first input points and the probability of the predicted power error corresponding to each second input point among the multiple second input points into the long short-term memory network LSTM to generate multiple prediction curves.
[0110] 203. Multiple prediction curves are superimposed using the curve of the target high-frequency part changing with time to obtain multiple target prediction curves for representing the new energy output scenario.
[0111] In the embodiment of the present application, after superimposing multiple prediction curves using the curve of the target high-frequency part changing with time, multiple superimposed prediction curves are obtained. The superimposed prediction curves are the curves of the target new energy predicted power generation changing with time, and the superimposed prediction curves are used as the target prediction curves for representing the new energy output scenario. For example, Figure 4 As shown, a plurality of wind power output curves are generated, that is, curves showing the predicted wind power generation changing with time, where the ordinate is power and the abscissa is time.
[0112] The method provided in the embodiment of the present application introduces wavelet transform, first extracts the main trend of the data, and then generates the scene. At the same time, the embodiment of the present application uses the long short-term memory network LSTM in deep learning. The long short-term memory network LSTM can effectively record data from a period of time before, and can capture the temporal relationship between the output data of new energy sources, so that the generated scene effect is better and the reliability of the predicted scene is improved.
[0113] Further, as Figure 1 The specific implementation of the method is as follows: Figure 5 As shown, an embodiment of the present application provides a device for generating a new energy output scenario, including: a processing module 301, an acquisition module 302 and a superposition module 303.
[0114] The processing module 301 is used to obtain historical power data of new energy, pre-process the historical power data, and extract the curve of the target low-frequency part of the new energy power generation over time and the curve of the target high-frequency part of the new energy power generation over time;
[0115] The acquisition module 302 is used to obtain multiple prediction curves based on the curve of the target low-frequency part of the new energy power generation changing with time using a long short-term memory network LSTM, where the prediction curve is a curve of the new energy power generation predicted to change with time;
[0116] The superposition module 303 is used to superimpose multiple prediction curves using the curve of the target high-frequency part changing with time, so as to obtain multiple superimposed curves as multiple target prediction curves for representing the new energy output scenario.
[0117] In a specific application scenario, the processing module 301 is also used to:
[0118] Determine the first expression of the discrete wavelet sequence and the second expression of the discrete wavelet transform coefficient, the first expression and the second expression are respectively:
[0119] Formula 1:
[0120] Among them, j represents the jth decomposition level, k represents the parameter used to complete the translation operation in wavelet transform, and a 0 is the lowest resolution low-frequency information, b 0 is the lowest resolution detail information, ψ is the wavelet mother function, t is the time, ψ j,k (t) is a discrete wavelet sequence;
[0121] Formula 2:
[0122] Among them, C j,k is the discrete wavelet transform coefficient, f(t) is the signal to be analyzed;
[0123] A reconstruction formula is constructed based on the first expression and the second expression. The reconstruction formula is expressed as:
[0124] Formula 3:
[0125] Where C is a constant and f'(t) is a discrete signal;
[0126] Perform wavelet decomposition on the historical power data and obtain the decomposition formula:
[0127] Formula 4:
[0128] Among them, A n is the low-frequency part of the nth decomposition layer, D l is the high frequency part of the lth sub-level, l = 1, 2, ..., n, S is the original discrete signal;
[0129] Based on the decomposition formula, the reconstruction formula is subjected to discrete wavelet transform to obtain a discrete wavelet reconstruction function, which is expressed as:
[0130] Formula 5:
[0131] Where, f”(t) is the discrete wavelet reconstruction function, d j,k = <f(t),ψ j,k (t)> is the wavelet coefficient, φ j,k (t) is the scaling function;
[0132] The number of decomposition layers is determined, and based on the number of decomposition layers and the discrete wavelet reconstruction function, a curve of a target low-frequency part of the renewable energy power generation varying with time and a curve of a target high-frequency part varying with time are determined.
[0133] In a specific application scenario, the processing module 301 is also used to:
[0134] Project the reconstruction function to the low-frequency part of the decomposition layer j to obtain the overview signal at scale j. The overview signal at scale j is expressed as:
[0135] Formula 6:
[0136] Among them, c j,k is the scale factor, and c j,k = <f(t),φ j,k (t)>, is the general signal at scale j;
[0137] Project the reconstruction function to the high-frequency part of the decomposition layer j to obtain the detail signal at the j scale, where the detail signal at the j scale is expressed as:
[0138] Formula 7:
[0139] in, is the detail signal at scale j;
[0140] The number of decomposition levels is determined, and a discrete wavelet reconstruction function is determined based on the number of decomposition levels, the overview signal at the j scale, and the detail signal at the j scale.
[0141] In a specific application scenario, the acquisition module 302 is also used to: use the long short-term memory network LSTM to predict the curve of the target low-frequency part of the renewable energy power generation power changing with time, and obtain the probability distribution curve of the predicted power error; determine the probability distribution curve of the predicted power based on the curve of the target low-frequency part of the renewable energy power generation power changing with time, and use the long short-term memory network LSTM to generate the multiple prediction curves based on the probability distribution curve of the predicted power and the probability distribution curve of the predicted power error.
[0142] In a specific application scenario, the acquisition module 302 is also used to: extract multiple points from a curve of the target low-frequency part of the renewable energy power generation power changing with time, determine the characteristic value of each of the multiple points, and the characteristic value is one of the valley value, the peak value and the slope of the curve; input the characteristic value of each of the multiple points into the long short-term memory network LSTM to obtain the probability distribution curve of the predicted power error.
[0143] In a specific application scenario, the acquisition module 302 is also used to: extract multiple points from a curve showing that the target low-frequency part of the renewable energy power generation power changes with time, and determine the characteristic value of each of the multiple points, where the characteristic value is one of a valley value, a peak value, and a line slope; use the matrix laboratory MATLAB tool to fit the characteristic value of each of the multiple points to obtain a probability distribution curve for the predicted power.
[0144] In a specific application scenario, the acquisition module 302 is also used to: randomly sample the probability distribution curve of the predicted power to obtain the probability of the predicted power corresponding to each first input point among multiple first input points; randomly sample the probability distribution curve of the predicted power error to obtain the probability of the predicted power error corresponding to each second input point among multiple second input points; input the probability of the predicted power corresponding to each first input point among multiple first input points and the probability of the predicted power error corresponding to each second input point among the multiple second input points into the long short-term memory network LSTM to generate multiple prediction curves.
[0145] The device provided in the embodiment of the present application pre-processes the historical power data of the acquired new energy through a processing module, extracts the curve of the target low-frequency part of the new energy power generation power changing with time and the curve of the target high-frequency part changing with time, and then obtains multiple prediction curves based on the curve of the target low-frequency part of the new energy power generation power changing with time through the acquisition module using the long short-term memory network LSTM, and finally superimposes the multiple prediction curves using the curve of the target high-frequency part changing with time through the superposition module to obtain multiple target prediction curves for representing the new energy output scenario; the long short-term memory network LSTM can effectively record the data of a previous period of time, and can capture the temporal relationship between the output data of the new energy, so that the generated scenario effect is better and the reliability of the predicted scenario is improved.
[0146] In an exemplary embodiment, see Figure 6 , and also provides a computer device, which includes a bus, a processor, a memory and a communication interface, and may also include an input and output interface and a display device, wherein each functional unit can communicate with each other through the bus. The memory stores a computer program, and the processor is used to execute the program stored in the memory and execute the method for generating the new energy output scene in the above embodiment.
[0147] A readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for generating a new energy output scenario.
[0148] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by hardware, or by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each implementation scenario of the present application.
[0149] Those skilled in the art will appreciate that the accompanying drawings are merely schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily required for implementing the present application.
[0150] Those skilled in the art will appreciate that the modules in the device in the implementation scenario can be distributed in the device in the implementation scenario according to the implementation scenario description, or can be changed accordingly and located in one or more devices different from the implementation scenario. The modules in the above implementation scenario can be combined into one module, or can be further split into multiple sub-modules.
[0151] The above application serial numbers are for description only and do not represent the advantages or disadvantages of the implementation scenarios.
[0152] The above embodiments are only exemplary embodiments of the present application and are not intended to limit the present application. The protection scope of the present application is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present application within the essence and protection scope of the present application, and such modifications or equivalent substitutions shall also be deemed to fall within the protection scope of the present application.
Claims
1. A method for generating a new energy output scenario, It is characterized in that include: Acquire historical power data of new energy, preprocess the historical power data, and extract a curve of a target low-frequency part of the power generated by new energy varying over time and a curve of a target high-frequency part of the power generated by new energy varying over time; Based on the curve of the target low-frequency part of the new energy power generation changing over time, a plurality of prediction curves are obtained by using a long short-term memory network LSTM, wherein the prediction curve is a curve of the new energy power generation predicted to change over time; Using the curve of the target high-frequency part changing with time to superimpose the multiple prediction curves, a plurality of target prediction curves for representing the new energy output scenario are obtained; The preprocessing of the historical power data to extract the curve of the target low-frequency part of the new energy power generation changing with time and the curve of the target high-frequency part changing with time include: Determine a first expression of a discrete wavelet sequence and a second expression of discrete wavelet transform coefficients, wherein the first expression and the second expression are respectively: Formula 1: Among them, j represents the jth decomposition level, k represents the parameter used to complete the translation operation in wavelet transform, and a 0 is the lowest resolution low-frequency information, b 0 is the lowest resolution detail information, ψ is the wavelet mother function, t is the time, ψ j,k (t) is the discrete wavelet sequence; Formula 2: Among them, C j,k is the discrete wavelet transform coefficient, and f(t) is the signal to be analyzed; A reconstruction formula is constructed based on the first expression and the second expression, and the reconstruction formula is expressed as: Formula 3: Where C is a constant and f'(t) is a discrete signal; The historical power data is subjected to wavelet decomposition to obtain a decomposition formula, which is: Formula 4: Among them, A n is the low-frequency part of the nth decomposition layer, D l is the high frequency part of the lth sub-level, l = 1, 2, ..., n, S is the original discrete signal; Based on the decomposition formula, the reconstruction formula is subjected to discrete wavelet transform to obtain a discrete wavelet reconstruction function, which is expressed as: Formula 5: Wherein, f”(t) is the discrete wavelet reconstruction function, d j,k = <f(t),ψ j,k (t)> is the wavelet coefficient, φ j,k (t) is the scaling function; Determine the number of decomposition layers, and determine a curve of a target low-frequency part of the new energy power generation changing over time and a curve of a target high-frequency part of the new energy power generation changing over time based on the number of decomposition layers and the discrete wavelet reconstruction function; The step of performing discrete wavelet transform on the reconstruction formula based on the decomposition formula to obtain a discrete wavelet reconstruction function includes: The reconstruction function is projected onto the low-frequency part of the decomposition layer j to obtain the overview signal at the j scale. The overview signal at the j scale is expressed as: Formula 6: Among them, c j,k is the scale factor, and c j,k = <f(t),φ j,k (t)>, is the overview signal at the j scale; The reconstruction function is projected onto the high frequency part of the decomposition layer j to obtain the detail signal at the j scale, wherein the detail signal at the j scale is expressed as: Formula 7: in, is the detail signal at the j scale; Determine the number of decomposition levels, and determine the discrete wavelet reconstruction function based on the number of decomposition levels, the overview signal at the j scale, and the detail signal at the j scale; The curve of the target low-frequency part of the new energy power generation power changing over time uses a long short-term memory network LSTM to obtain multiple prediction curves, including: Using the long short-term memory network LSTM, predicting the time-varying curve of the target low-frequency part of the renewable energy power generation power, and obtaining a probability distribution curve of the predicted power error; Determine a probability distribution curve of predicted power based on a curve of a target low-frequency part of the renewable energy power generation power varying with time, and generate the plurality of prediction curves using the long short-term memory network (LSTM) according to the probability distribution curve of predicted power and the probability distribution curve of predicted power error; The method of using the long short-term memory network LSTM to predict the curve of the target low-frequency part of the renewable energy power generation power changing over time to obtain a probability distribution curve of the predicted power error includes: Extracting a plurality of points from a curve of the target low-frequency part of the renewable energy power generation changing with time, and determining a characteristic value of each of the plurality of points, wherein the characteristic value is one of a valley value, a peak value and a slope of the curve; Inputting the characteristic value of each point in the plurality of points into the long short-term memory network LSTM to obtain a probability distribution curve of the predicted power error; The method of determining a probability distribution curve of predicted power based on a curve of a target low-frequency part of the renewable energy power generation power varying with time comprises: Extracting a plurality of points from a curve of the target low-frequency part of the renewable energy power generation changing with time, and determining a characteristic value of each of the plurality of points, wherein the characteristic value is one of a valley value, a peak value, and a line slope; The eigenvalue of each point in the plurality of points is fitted using the matrix laboratory MATLAB tool to obtain a probability distribution curve of the predicted power.
2. The method for generating a new energy output scenario according to claim 1, It is characterized in that The method of generating the multiple prediction curves using the long short-term memory network LSTM according to the probability distribution curve of the predicted power and the probability distribution curve of the predicted power error includes: Randomly sampling the probability distribution curve of the predicted power to obtain the probability of the predicted power corresponding to each first input point among the plurality of first input points; Randomly sampling the probability distribution curve of the predicted power error to obtain the probability of the predicted power error corresponding to each second input point among the plurality of second input points; The probability of the predicted power corresponding to each first input point in the multiple first input points and the probability of the predicted power error corresponding to each second input point in the multiple second input points are input into a long short-term memory network LSTM to generate the multiple prediction curves.
3. A device for generating a new energy output scene, It is characterized in that include: A processing module, used to obtain historical power data of new energy, pre-process the historical power data, and extract a curve of a target low-frequency part of the power generated by the new energy over time and a curve of a target high-frequency part of the power generated by the new energy over time; An acquisition module is used to obtain multiple prediction curves using a long short-term memory network (LSTM) based on a curve of a target low-frequency part of the power generation of the new energy source changing over time, wherein the prediction curve is a curve of the power generation of the new energy source changing over time; A superposition module, used to superimpose the multiple prediction curves using the curve of the target high-frequency part changing with time, to obtain multiple superimposed curves as multiple target prediction curves for representing the new energy output scenario; The preprocessing of the historical power data to extract the curve of the target low-frequency part of the new energy power generation changing with time and the curve of the target high-frequency part changing with time include: Determine a first expression of a discrete wavelet sequence and a second expression of discrete wavelet transform coefficients, wherein the first expression and the second expression are respectively: Formula 1: Among them, j represents the jth decomposition level, k represents the parameter used to complete the translation operation in wavelet transform, and a 0 is the lowest resolution low-frequency information, b 0 is the lowest resolution detail information, ψ is the wavelet mother function, t is the time, ψ j,k (t) is the discrete wavelet sequence; Formula 2: Among them, C j,k is the discrete wavelet transform coefficient, and f(t) is the signal to be analyzed; A reconstruction formula is constructed based on the first expression and the second expression, and the reconstruction formula is expressed as: Formula 3: Where C is a constant and f'(t) is a discrete signal; The historical power data is subjected to wavelet decomposition to obtain a decomposition formula, which is: Formula 4: Among them, A n is the low-frequency part of the nth decomposition layer, D l is the high frequency part of the lth sub-level, l = 1, 2, ..., n, S is the original discrete signal; Based on the decomposition formula, the reconstruction formula is subjected to discrete wavelet transform to obtain a discrete wavelet reconstruction function, which is expressed as: Formula 5: Wherein, f”(t) is the discrete wavelet reconstruction function, d j,k = <f(t),ψ j,k (t)> is the wavelet coefficient, φ j,k (t) is the scaling function; Determine the number of decomposition layers, and determine a curve of a target low-frequency part of the new energy power generation changing over time and a curve of a target high-frequency part of the new energy power generation changing over time based on the number of decomposition layers and the discrete wavelet reconstruction function; The step of performing discrete wavelet transform on the reconstruction formula based on the decomposition formula to obtain a discrete wavelet reconstruction function includes: The reconstruction function is projected onto the low-frequency part of the decomposition layer j to obtain the overview signal at the j scale. The overview signal at the j scale is expressed as: Formula 6: Among them, c j,k is the scale factor, and c j,k = <f(t),φ j,k (t)>, is the overview signal at the j scale; The reconstruction function is projected onto the high frequency part of the decomposition layer j to obtain the detail signal at the j scale, wherein the detail signal at the j scale is expressed as: Formula 7: in, is the detail signal at the j scale; Determine the number of decomposition levels, and determine the discrete wavelet reconstruction function based on the number of decomposition levels, the overview signal at the j scale, and the detail signal at the j scale; The curve of the target low-frequency part of the new energy power generation power changing over time uses a long short-term memory network LSTM to obtain multiple prediction curves, including: Using the long short-term memory network LSTM, predicting the time-varying curve of the target low-frequency part of the renewable energy power generation power, and obtaining a probability distribution curve of the predicted power error; Determine a probability distribution curve of predicted power based on a curve of a target low-frequency part of the renewable energy power generation power varying with time, and generate the plurality of prediction curves using the long short-term memory network (LSTM) according to the probability distribution curve of predicted power and the probability distribution curve of predicted power error; The method of using the long short-term memory network LSTM to predict the curve of the target low-frequency part of the renewable energy power generation power changing over time to obtain a probability distribution curve of the predicted power error includes: Extracting a plurality of points from a curve of the target low-frequency part of the renewable energy power generation changing with time, and determining a characteristic value of each of the plurality of points, wherein the characteristic value is one of a valley value, a peak value and a slope of the curve; Inputting the characteristic value of each point in the plurality of points into the long short-term memory network LSTM to obtain a probability distribution curve of the predicted power error; The method of determining a probability distribution curve of predicted power based on a curve of a target low-frequency part of the renewable energy power generation power varying with time comprises: Extracting a plurality of points from a curve of the target low-frequency part of the renewable energy power generation changing with time, and determining a characteristic value of each of the plurality of points, wherein the characteristic value is one of a valley value, a peak value, and a line slope; The eigenvalue of each point in the plurality of points is fitted using the matrix laboratory MATLAB tool to obtain a probability distribution curve of the predicted power.
4. A computer device comprising a memory and a processor, wherein the memory stores a computer program, It is characterized in that When the processor executes the computer program, the steps of the method according to any one of claims 1 to 2 are implemented.
5. A readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 2 are implemented.
