A regional wind, solar and storage development capacity adjustment method and system

Abstract

The present invention proposes a method and system for adjusting the development capacity of regional wind, solar and storage, obtains the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set and performs quantitative analysis, extracts the spatiotemporal characteristics of the wind, solar and storage project development capacity in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area; constructs a small sample probability correlation model between the wind, solar and storage policy library and the wind, solar and storage project development capacity, performs feature mapping processing on the correlation between the target area wind, solar and storage policy and the target area wind, solar and storage project development capacity through manifold learning and BiLSTM to obtain the correlation result; adjusts the development capacity of wind, solar and storage in the target area. The problem of inaccurate adjustment of wind, solar and storage development capacity caused by the high dimension of training data in the prior art model is solved. The processing method of BiLSTM is combined with manifold learning to reduce the dimension of sample data, better achieve the feature mapping effect, and improve the accuracy of development capacity adjustment.

Description

Technical Field

[0001] The present invention relates to the field of new energy policies and project development, and in particular to a method and system for adjusting the capacity of regional wind, solar and storage development. Background Art

[0002] With the intensive introduction of wind, solar and storage-related policies and the continuous adjustment of national subsidy policies, and the influence of economic factors such as the in-depth promotion of the power market, the rapid development of the carbon market and the price fluctuations of the wind, solar and storage-related industrial chain, it is necessary to study the relationship between wind, solar and storage project policies and project development capacity, and propose a wind, solar and storage development plan suitable for the regional resource characteristics of Guangdong. Compared with technological progress, policy support has a stronger incentive effect on the development of wind, solar and storage and green certificate market transactions, and has a significant role in promoting the long-term and stable development of wind, solar and storage. Therefore, by studying the deep mining of related information and the construction of wind, solar and storage calculation models, we can lay a good foundation for the policy recommendations on wind, solar and storage.

[0003] At present, the existing technology mainly uses the LSTM model method to pre-process the adjustment of wind, solar and storage development capacity. However, the single LSTM model has the problem of high dimension, which will cause problems in model training and lead to inaccurate adjustment of wind, solar and storage development capacity. Summary of the invention

[0004] Based on the above problems, the present invention proposes a regional wind, solar and storage development capacity adjustment method and system to solve the problem of inaccurate wind, solar and storage development capacity adjustment caused by high dimension of model training data in the prior art.

[0005] To achieve the above-mentioned purpose, an embodiment of the present invention provides a method for adjusting the development capacity of regional wind, solar and storage, including: obtaining a target regional wind, solar and storage policy set and a target regional wind, solar and storage project development capacity data set;

[0006] Quantitatively analyze the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set to obtain several data indicators corresponding to the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set;

[0007] Extracting the spatiotemporal characteristics of the wind, solar, and storage project development capacity in the target area and the spatiotemporal characteristics of the wind, solar, and storage policy set in the target area based on the data indicators to obtain small sample parameters;

[0008] According to the small sample parameters, a small sample probability correlation model between the wind, solar and storage policy database and the wind, solar and storage project development capacity is constructed to obtain different probability correlation characteristics;

[0009] According to the different probability association features, feature mapping is performed on the correlation between the wind, solar and storage policies in the target area and the development capacity of the wind, solar and storage projects in the target area through manifold learning and BiLSTM to obtain the correlation result;

[0010] According to the correlation results, the development capacity of wind, solar and storage in the target area is adjusted.

[0011] The embodiment of the present invention proposes a regional wind, solar and storage development capacity adjustment method. A large number of samples are collected for quantitative analysis, and finally the samples are converted into small sample parameters with spatiotemporal characteristics. The small sample parameters are then used to construct a small sample probability correlation model between a wind, solar and storage policy library and the development capacity of wind, solar and storage projects to obtain different probability correlation characteristics. Finally, the data is feature mapped through BiLSTM. The BiLSTM processing method is combined with manifold learning to achieve dimensionality reduction of sample data, and the feature mapping effect can be better achieved, thereby greatly improving the accuracy of development capacity adjustment.

[0012] Furthermore, the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set are quantitatively analyzed to obtain several data indicators corresponding to the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set, specifically:

[0013] Quantitatively process the target area wind-solar-storage policy set and the target area wind-solar-storage project development capacity data set to obtain a first quantitative result and a second quantitative result;

[0014] Analyze and process the first quantization result corresponding to the second quantization result to obtain a number of data indicators;

[0015] Among them, the several data indicators include: electricity price subsidy data, installed capacity data during the target processing period, and the development amount of wind, solar and storage projects corresponding to different regions.

[0016] Furthermore, according to the several data indicators, the spatiotemporal characteristics of the development capacity of the wind, solar and storage projects in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area are extracted to obtain small sample parameters, which are specifically:

[0017] According to the several data indicators, respectively extracting the spatiotemporal characteristics of the development capacity of the wind, solar and storage projects in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area;

[0018] According to the matrix probability association model, the spatiotemporal characteristics of the development capacity of the wind, solar and storage projects in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area are converted into small sample parameters; wherein the small sample parameters include: reference sequence parameters and comparison sequence parameters.

[0019] Furthermore, according to the small sample parameters, a small sample probability correlation model of the wind-solar-storage policy library and the development capacity of the wind-solar-storage project is constructed to obtain different probability correlation features, specifically:

[0020] Preprocessing the small sample parameters so that the small sample parameters meet the model construction requirements; wherein the small sample parameters also include: the reference sequence parameters are the spatiotemporal characteristics of the wind, solar and storage policy set extracted in the target area and the comparison sequence parameters are the spatiotemporal characteristics of the wind, solar and storage project development capacity extracted in the target area;

[0021] Determining the sample capacity of the comparison sequence parameters to obtain a determination result;

[0022] According to the judgment result, the sample capacity of the reference sequence parameters is processed to obtain different probability association features.

[0023] Further, the sample capacity of the reference sequence parameter is processed according to the judgment result to obtain different probability association features, specifically:

[0024] When it is determined that the sample size of the comparison sequence parameter is less than a preset value, an empirical distribution function is constructed according to the sample order ratio statistic, and the sample size of the reference sequence parameter is collected to obtain a first probability association feature;

[0025] When it is determined that the sample size of the comparison sequence parameter is greater than or equal to a preset value, a resampled sample is obtained by resampling, an empirical distribution function is constructed according to the resampled sample, the sample size of the reference sequence parameter is collected, and a second probability association feature is obtained.

[0026] Furthermore, according to the different probability association features, feature mapping processing is performed on the association between the wind, solar and storage policies in the target area and the development capacity of the wind, solar and storage projects in the target area through manifold learning and BiLSTM to obtain the association result, which is specifically:

[0027] According to the first probability association feature and the second probability association feature, forward and reverse LSTM feature mapping associations are performed to obtain a correlation model between the wind, solar, and storage policies in the target area and the development capacity of the wind, solar, and storage projects in the target area;

[0028] According to the correlation model between the wind, solar and storage policy in the target area and the development capacity of the wind, solar and storage projects in the target area, recursive feedback processing is performed on the reference sequence parameters and the comparison sequence parameters, and the hidden layer state of the BiLSTM network is output;

[0029] Manifold learning dimensionality reduction is performed on the hidden layer state of the BiLSTM network to obtain the association result.

[0030] The present invention also provides a regional wind, solar and storage development capacity adjustment system, including: a data acquisition module, a quantitative analysis module, a feature extraction module, a small probability correlation model construction module, a feature mapping processing module and a development capacity adjustment module;

[0031] The data acquisition module is used to obtain the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set;

[0032] The quantitative analysis module is used to perform quantitative analysis on the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set to obtain a number of data indicators corresponding to the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set;

[0033] The feature extraction module is used to extract the spatiotemporal characteristics of the wind, solar and storage project development capacity and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area according to the data indicators to obtain small sample parameters;

[0034] The small probability correlation model construction module is used to construct a small sample probability correlation model between the wind, solar and storage policy library and the wind, solar and storage project development capacity according to the small sample parameters to obtain different probability correlation features;

[0035] The feature mapping processing module is used to perform feature mapping processing on the correlation between the wind, solar and storage policies in the target area and the development capacity of the wind, solar and storage projects in the target area through manifold learning and BiLSTM according to the different probability association features, so as to obtain the correlation result;

[0036] The development capacity adjustment module is used to adjust the development capacity of wind, solar and storage in the target area according to the correlation result.

[0037] A regional wind, solar and storage development capacity adjustment system proposed in an embodiment of the present invention performs quantitative analysis on a large amount of data through a data acquisition module and a quantitative analysis module, and then refines the samples into data labels with different probability spatiotemporal characteristics through a feature extraction module and a small probability association model construction module. Finally, the samples are subjected to BiLSTM processing and popular learning processing through a feature mapping processing module, thereby achieving dimensionality reduction of data samples, reducing abnormal problems in model processing data, and greatly improving the accuracy of development capacity adjustment.

[0038] Further, the quantitative analysis module is used to perform quantitative analysis on the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set to obtain several data indicators corresponding to the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set, and also includes:

[0039] A quantitative processing unit and an analytical processing unit;

[0040] The quantification processing unit is used to quantify the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set respectively to obtain a first quantification result and a second quantification result;

[0041] The analysis and processing unit is used to analyze and process the first quantification result corresponding to the second quantification result to obtain a number of data indicators; wherein the several data indicators include: electricity price subsidy data, installed capacity data and development amounts of wind, solar and storage projects corresponding to different regions within the target processing time period.

[0042] Furthermore, the feature extraction module is used to extract the spatiotemporal characteristics of the wind, solar and storage project development capacity and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area according to the data indicators to obtain small sample parameters, and also includes:

[0043] Feature extraction unit and parameter conversion unit;

[0044] The feature extraction unit is used to extract the spatiotemporal characteristics of the wind, solar and storage project development capacity and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area according to the plurality of data indicators;

[0045] The parameter conversion unit is used to convert the spatiotemporal characteristics of the development capacity of the wind, solar and storage projects in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area into small sample parameters according to a matrix probability association model; wherein the small sample parameters include: reference sequence parameters and comparison sequence parameters.

[0046] Furthermore, the small probability correlation model construction module is used to construct a small sample probability correlation model of the wind, solar and storage policy library and the wind, solar and storage project development capacity according to the small sample parameters to obtain different probability correlation features, and also includes:

[0047] Preprocessing unit, result judgment unit and feature output unit;

[0048] The preprocessing unit is used to preprocess the small sample parameters so that the small sample parameters meet the model construction requirements; wherein the small sample parameters also include: the reference sequence parameters are the spatiotemporal characteristics of the wind, solar and storage policy set extracted in the target area and the comparison sequence parameters are the spatiotemporal characteristics of the wind, solar and storage project development capacity extracted in the target area;

[0049] The result judgment unit is used to judge the sample capacity of the comparison sequence parameters to obtain a judgment result;

[0050] The feature output unit is used to process the sample capacity of the reference sequence parameters according to the judgment result to obtain different probability association features.

[0051] Furthermore, the feature output unit is used to process the sample capacity of the reference sequence parameter according to the judgment result to obtain different probability association features, and also includes:

[0052] a first probability feature acquisition subunit and a second probability feature acquisition subunit;

[0053] The first probability feature acquisition subunit is used to, when it is determined that the sample size of the comparison sequence parameter is less than a preset value, construct an empirical distribution function according to the sample order ratio statistic, collect the sample size of the reference sequence parameter, and obtain a first probability association feature;

[0054] The second probability feature acquisition subunit is used to obtain a resampled sample by resampling when it is determined that the sample size of the comparison sequence parameter is greater than or equal to a preset value, construct an empirical distribution function according to the resampled sample, collect the sample size of the reference sequence parameter, and obtain a second probability association feature.

[0055] Further, the feature mapping processing module is used to perform feature mapping processing on the correlation between the wind, solar and storage policies in the target area and the development capacity of the wind, solar and storage projects in the target area through manifold learning and BiLSTM according to the different probability association features to obtain the correlation result, and also includes:

[0056] Model building unit, recursive feedback processing unit and result output unit;

[0057] The model building unit is used to perform forward and reverse LSTM feature mapping association according to the first probability association feature and the second probability association feature to obtain a correlation model between the wind, solar and storage policy in the target area and the development capacity of the wind, solar and storage project in the target area;

[0058] The recursive feedback processing unit is used to perform recursive feedback processing on the reference sequence parameters and the comparison sequence parameters according to the correlation model between the wind, solar and storage policy in the target area and the development capacity of the wind, solar and storage projects in the target area, and output the hidden layer state of the BiLSTM network;

[0059] The result output unit is used to perform manifold learning dimensionality reduction on the hidden layer state of the BiLSTM network to obtain the association result. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 A schematic flow chart of the steps of a method for adjusting the regional wind, solar and storage development capacity provided by an embodiment of the present invention;

[0061] Figure 2 A schematic diagram of the module structure of a regional wind, solar and storage development capacity adjustment system provided in an embodiment of the present invention;

[0062] Figure 3 A schematic diagram of the structure of a quantitative analysis module of a regional wind, solar, and storage development capacity adjustment system provided by a certain embodiment of the present invention;

[0063] Figure 4 A schematic diagram of the structure of a feature extraction module of a regional wind, solar, and storage development capacity adjustment system provided by an embodiment of the present invention;

[0064] Figure 5 A schematic diagram of a small probability correlation model building module structure of a regional wind, solar and storage development capacity adjustment system provided by an embodiment of the present invention;

[0065] Figure 6 A schematic diagram of the structure of a characteristic output unit of a regional wind, solar, storage and development capacity adjustment system provided by an embodiment of the present invention;

[0066] Figure 7 A schematic diagram of the structure of a feature mapping processing module of a regional wind, solar, and storage development capacity adjustment system provided by an embodiment of the present invention;

[0067] Figure 8 A schematic diagram of the LSTM minimum model unit structure of a regional wind, solar and storage development capacity adjustment method provided in an embodiment of the present invention;

[0068] Fig. 9 A schematic diagram of the BiLSTM network structure of a regional wind, solar and storage development capacity adjustment method provided for a certain embodiment of the present invention. DETAILED DESCRIPTION

[0069] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0070] In the embodiment of the present invention, the wind-solar-storage regulation scenario in Guangdong is taken as an example to explain the feasibility of the present invention. This is only an example and does not limit the present invention in any way, and will not be repeated below.

[0071] Example 1

[0072] See also Figure 1 , Figure 1 A schematic flow chart of the steps of a method for adjusting the regional wind, solar and storage development capacity provided by an embodiment of the present invention. Figure 1As shown, the present invention proposes a method for adjusting the regional wind, solar and storage development capacity, including steps 101 to 106, each of which is specifically as follows:

[0073] Step 101, obtaining a target area wind, solar and storage policy set and a target area wind, solar and storage project development capacity data set;

[0074] As an example of this embodiment, based on a typical regional wind, solar and storage policy database, policies of provinces and regions similar to Guangdong (Zhejiang, Jiangsu and Shandong) are collected, such as electricity price subsidy policies and installed capacity policies over a period of time; through in-depth research, a wind, solar and storage project capacity database for provinces and regions similar to Guangdong and other regions is established.

[0075] Step 102, quantitatively analyzing the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set to obtain a number of data indicators corresponding to the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set;

[0076] As an example of this embodiment, the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set are quantified respectively to obtain a first quantification result and a second quantification result; the first quantification result is analyzed and processed corresponding to the second quantification result to obtain a number of data indicators; wherein the several data indicators include: electricity price subsidy data, installed capacity data and wind, solar and storage project development amounts corresponding to different regions during the target processing time period.

[0077] Step 103, extracting the spatiotemporal characteristics of the wind, solar and storage project development capacity and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area according to the plurality of data indicators, and obtaining small sample parameters;

[0078] As an example of this embodiment, according to the several data indicators, the spatiotemporal characteristics of the development capacity of the wind, solar and storage projects in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area are extracted respectively;

[0079] According to the matrix probability association model, the spatiotemporal characteristics of the development capacity of the wind, solar and storage projects in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area are converted into small sample parameters; wherein the small sample parameters include: reference sequence parameters and comparison sequence parameters.

[0080] As another example of this embodiment, under the same initial conditions, a reference sequence x and a comparison sequence y are obtained respectively:

[0081]

[0082]

[0083] Where: m is the reference sequence x 1:m (k) Sample size, n is the comparison sequence y 1:n (k) Sample size: a matrix probability association model can be used here.

[0084] Step 104, constructing a small sample probability correlation model between the wind, solar and storage policy library and the wind, solar and storage project development capacity according to the small sample parameters, and obtaining different probability correlation features;

[0085] As an example of this embodiment, the small sample parameters are preprocessed so that the small sample parameters meet the model construction requirements; wherein the small sample parameters also include: the reference sequence parameters are the extracted spatiotemporal characteristics of the wind, solar and storage policy set in the target area and the comparison sequence parameters are the extracted spatiotemporal characteristics of the wind, solar and storage project development capacity in the target area; the sample capacity of the comparison sequence parameters is judged to obtain a judgment result; according to the judgment result, the sample capacity of the reference sequence parameters is processed to obtain different probability association features.

[0086] As another example of this embodiment, when it is determined that the sample size of the comparison sequence parameters is less than a preset value, an empirical distribution function is constructed according to the sample order ratio statistic, and the sample size of the reference sequence parameters is collected to obtain a first probability association feature; when it is determined that the sample size of the comparison sequence parameters is greater than or equal to the preset value, a resampled sample is obtained by resampling, an empirical distribution function is constructed according to the resampled sample, and the sample size of the reference sequence parameters is collected to obtain a second probability association feature.

[0087] As another example of the present invention, the reference sequence x and the comparison sequence y are preprocessed to satisfy the data sequence requirements of equal step length and equal length.

[0088] Calculate the probability correlation coefficient p at time k operator (k):

[0089] (1) When y 1:n (k) Sample size n<5:

[0090] 1) Construct the sample order ratio statistic K ijk =(x (j) -x (i) ) / (x (k) -x (i) ),1≤i <j<k≤n+1,0≤K ijk ≤1, and estimate the empirical distribution function of K based on the sample

[0091] 2) Substitute the reference sample x(k) into the empirical distribution function Calculate the cumulative distribution function value and get the correlation coefficient poperator (k).

[0092] (2) When y 1:n (k) When the sample size n ≥ 5:

[0093] 1) For comparison sample y 1:n (k) Use the Bootstrap method to resample and obtain the resampled sample y' 1:n (k), according to the resampled sample y' 1:n (k) Estimate the empirical distribution function of y

[0094] 2) Substitute the reference sample x(k) into the empirical distribution function Calculate the cumulative distribution function value and get the relevant relationship p operator (k).

[0095] Test p operator Under a certain confidence level α, whether it obeys the uniform distribution on [0,1]. According to the probability correlation coefficient, evaluate the expected effect of each decision option, and use the probability correlation coefficient to perform sensitivity analysis to understand the impact of different probability scenarios on the decision results. If the test passes, it means that the correlation analysis has passed. Otherwise, the probability correlation analysis has not passed. Finally, based on the probability correlation coefficient and the evaluation results, formulate the best decision strategy.

[0096] Step 105, according to the different probability association features, feature mapping processing is performed on the association between the wind, solar and storage policies in the target area and the development capacity of the wind, solar and storage projects in the target area through manifold learning and BiLSTM to obtain an association result;

[0097] As an example of this embodiment, forward and reverse LSTM feature mapping associations are performed according to the first probability association feature and the second probability association feature to obtain a correlation model between the wind, solar and storage policy in the target area and the development capacity of the wind, solar and storage projects in the target area; according to the correlation model between the wind, solar and storage policy in the target area and the development capacity of the wind, solar and storage projects in the target area, recursive feedback processing is performed on the reference sequence parameters and the comparison sequence parameters, and the hidden layer state of the BiLSTM network is output; manifold learning dimensionality reduction is performed on the hidden layer state of the BiLSTM network to obtain the correlation result.

[0098] As another example of this embodiment, specifically, the policy characteristics may include cultivating the wind, solar and energy storage market, regulating the order of the wind, solar and energy storage market, promoting the upgrading of the wind, solar and energy storage industry, and promoting local and regional economic development. Cultivating the wind, solar and energy storage market means that the government may promote the development of the wind, solar and energy storage market through investment, incentives or other means; regulating the order of the wind, solar and energy storage market means that the government will take measures to ensure the healthy development of the wind, solar and energy storage market, including regulatory and management measures to prevent potential market chaos and improper behavior; promoting the upgrading of the wind, solar and energy storage industry means promoting the improvement of the technology and economic level of the wind, solar and energy storage industry through technological innovation, industrial policies and other means; promoting local and regional economic development, that is, the construction and development of wind, solar and energy storage projects will contribute to local and regional economic growth. Project characteristics may include installed capacity, electricity price information, etc. By extracting the above policy characteristics and project characteristics from the collected wind, solar and energy storage policies and projects, it is helpful to better quantify the wind, solar and energy storage policies and projects.

[0099] As another example of this embodiment, LSTM can only learn information forward along the time axis. Considering the temporal and spatial relationship between wind, solar and storage policies and project capacity, BiLSTM can more deeply explore the characteristics of wind, solar and storage policies and project capacity time series. BiLSTM consists of two layers of LSTM in opposite directions. It combines the data features in the forward and backward directions to associate the data at the current moment. While using historical policy and project data information, it can learn future historical policies and project data, and perform recursion and feedback on them, explore the connection between past and future data in the time series, improve data utilization, and better utilize the time characteristics of the time series, thereby establishing an association model between the policy library and the development capacity of wind, solar and storage projects. See Figure 8 , Figure 8 A schematic diagram of the LSTM minimum model unit structure of a regional wind, solar and storage development capacity adjustment method provided by an embodiment of the present invention is as follows: Figure 8 As shown, assuming is the hidden layer state of the forward LSTM network at time t, and its calculation formula is shown in the following formula. It can be regarded as a single-layer LSTM network, which consists of the state at time t-1 The process of calculating the state at time t, x t is the input at time t.

[0100]

[0101] in: is the hidden layer state of the forward LSTM network at time t; LSTM is the LSTM unit; x t is the input at time t; is the hidden layer state of the forward LSTM network at time t-1.

[0102] akin is the hidden layer state of the reverse LSTM network at time t, and its calculation formula is as follows:

[0103]

[0104] in: is the hidden layer state of the forward LSTM network at time t; LSTM is the LSTM unit; x t is the input at time t; is the hidden layer state of the forward LSTM network at time t-1. The output of the BiLSTM network is the hidden layer state of the two parts. and Combined together, they form the overall hidden state h of the network t The reference sequence parameter x is the extracted data feature of the wind, solar and energy storage policy, which is the input x of the BiLSTM model. t ; The comparison sequence parameter y is the spatial feature of the wind, solar and energy storage project at the corresponding time, which is the hidden state h of the BiLSTM model t For details, see Fig. 9 , Fig. 9 A schematic diagram of the BiLSTM network structure of a regional wind, solar and storage development capacity adjustment method provided for a certain embodiment of the present invention.

[0105] As another example of this embodiment, the LSTM unit includes three parts: forgetting, input, and output, and a state unit is introduced to coordinate the operation of the entire network.

[0106] (1) Forget Gate

[0107] The function of the forget gate is to determine the degree of retention of the information passed in at the previous moment. t is the input x at time t t and the hidden layer output h at time t-1 t-1 It is obtained by linear transformation and then applying the activation function σ. The calculation method is shown in the following formula.

[0108] f t =σ(W f x t +U f h t-1 +b f )

[0109] Where: f t is the forget gate; σ is the activation function; W f is the forget gate parameter; x t is the input at time t; U f is the forget gate parameter; h t-1 is the hidden layer output at time t-1; b fis the forget gate parameter.

[0110] (2) Input Gate

[0111] The work of the input gate mainly determines the degree of retention of input information at time t. t , is calculated in the same way as the forget gate f t Similar, as shown below.

[0112] i t =σ(W i x t +U i h t-1 +b i )

[0113] Where: i t is the input gate; σ is the activation function; W i is the input gate parameter; x t is the input at time t; U i is the input gate parameter; h t-1 is the hidden layer output at time t-1; b i are the input gate parameters.

[0114] It is used to describe the input state at time t, and the hidden layer output h at time t-1 t-1 and the input x at time t t , obtained by linear transformation and then applying tanh, as shown in the following formula, This is equivalent to inputting x t , the hidden layer state h at time t-1 t-1 The contained state information is integrated to form a new state quantity.

[0115]

[0116] in: is the input state at time t; tanh is the activation function; W c is the input state parameter; x t is the input of the moment; U c is the input state parameter; h t-1 is the hidden layer state b at time t-1 c is the input state parameter.

[0117] (3) Unit status

[0118] The cell state is the information conveyor belt throughout the LSTM network, allowing historical policy information to flow downward in an unchanged manner.

[0119] The main function of the state unit update is to update the internal state of LSTM and update the internal state C of the previous moment.t-1 Update to the internal state C at that moment t . C t The calculation formula is as shown below. First, C t-1 f method to determine the retention of information at time t-1, and then use the retained information Add it to the state information at that moment and calculate the output C at time t t , as shown below.

[0120]

[0121] Where: C t is the internal state at time t; C t-1 is the internal state at time t-1; f t C is the forget gate; t is the input state at time t; i t is the input gate.

[0122] (4) Output gate

[0123] The output of the LSTM output gate at the moment of control depends on the state unit C t The output gate O t The method is also similar to f t 、i t Similar to the following formula.

[0124] o t =σ(W o x t +U o h t-1 +b o )

[0125] Among them: t is the output gate; σ is the activation function; W o is the output gate parameter; x t is the input at time t; U o is the output gate parameter; h t-1 is the hidden layer output at time t-1; b o is the output gate parameter. Finally, the hidden state output at time t is h t , is the internal state C at time t t AND output gate O t The calculation formula is shown below.

[0126] h t =o t ·Tanh (c t )

[0127] Where: h t is the hidden layer output at time t; Ot is the output gate; Tanh is the activation function C t is the internal state at time t.

[0128] As another example of this embodiment, manifold learning is a new dimensionality reduction method that can preserve the intrinsic manifold structure of high-dimensional data sets. This project uses a supervised manifold learning algorithm. Supervised learning can achieve better feature mapping and feature dimensionality reduction effects by increasing the distance between different types of data and reducing the distance between similar data. This project uses the label information of the project capacity sample to define the distance between sample points by the following formula.

[0129]

[0130] σ depends on the density of the data and is set to the average Euclidean distance between data points. μ can increase the distance between data of different classes. i is the data characteristics of wind, solar and energy storage policies, y i is the spatial feature of the wind, solar and storage project at the corresponding time. The distance range between samples of the same type is [0,1], and the distance range between samples of different types is [μ,∞]. i -x j || 2 As increases, the distance between samples of the same type increases more slowly than the distance between different samples. i ,y i ) can increase the clustering of samples of the same type and the differences between samples of different types, and effectively achieve the feature mapping and feature dimensionality reduction effects of project capacity.

[0131] Step 106: Adjust the development capacity of wind, solar and energy storage in the target area according to the correlation result.

[0132] As an example of this embodiment, the association result is used to provide the development capacity of wind turbines, photovoltaic power stations and energy storage under a specific wind, photovoltaic and energy storage policy, and to adaptively adjust the development capacity of wind, photovoltaic and energy storage in the target area.

[0133] The embodiment of the present invention proposes a regional wind, solar and storage development capacity adjustment method. A large number of samples are collected for quantitative analysis, and finally the samples are converted into small sample parameters with spatiotemporal characteristics. The small sample parameters are then used to construct a small sample probability correlation model between a wind, solar and storage policy library and the development capacity of wind, solar and storage projects to obtain different probability correlation characteristics. Finally, the data is feature mapped through BiLSTM. The BiLSTM processing method is combined with manifold learning to achieve dimensionality reduction of sample data, and the feature mapping effect can be better achieved, thereby greatly improving the accuracy of development capacity adjustment.

[0134] Example 2

[0135] See also Figure 2 , Figure 2 A schematic diagram of the module structure of a regional wind, solar and storage development capacity adjustment system provided in an embodiment of the present invention; Figure 2 As shown, the regional wind, solar and storage development capacity adjustment system includes: a data acquisition module 201, a quantitative analysis module 202, a feature extraction module 203, a small probability correlation model construction module 204, a feature mapping processing module 205 and a development capacity adjustment module 206;

[0136] The data acquisition module 201 is used to obtain a target area wind, solar and storage policy set and a target area wind, solar and storage project development capacity data set;

[0137] The quantitative analysis module 202 is used to perform quantitative analysis on the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set to obtain a number of data indicators corresponding to the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set;

[0138] As an example of this embodiment, see Figure 3 , Figure 3 A schematic diagram of the structure of a quantitative analysis module of a regional wind, solar, and storage capacity adjustment system provided by an embodiment of the present invention. Figure 3 As shown, the quantitative analysis module 202 is used to perform quantitative analysis on the target area wind-solar-storage policy set and the target area wind-solar-storage project development capacity data set to obtain several data indicators corresponding to the target area wind-solar-storage policy set and the target area wind-solar-storage project development capacity data set, and also includes:

[0139] Quantification processing unit 301 and analysis processing unit 302;

[0140] The quantification processing unit 301 is used to quantify the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set to obtain a first quantification result and a second quantification result;

[0141] The analysis and processing unit 302 is used to analyze and process the first quantification result corresponding to the second quantification result to obtain a number of data indicators; wherein the several data indicators include: electricity price subsidy data, installed capacity data and development amounts of wind, solar and storage projects corresponding to different regions within the target processing time period.

[0142] The feature extraction module 203 is used to extract the spatiotemporal characteristics of the wind, solar and storage project development capacity and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area according to the data indicators to obtain small sample parameters;

[0143] As an example of this embodiment, see Figure 4 , Figure 4 A schematic diagram of the structure of a feature extraction module of a regional wind, solar, and storage capacity adjustment system provided by an embodiment of the present invention. Figure 4 As shown, the feature extraction module 203 is used to extract the spatiotemporal characteristics of the development capacity of the wind, solar and storage projects in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area according to the several data indicators to obtain small sample parameters, and also includes:

[0144] Feature extraction unit 401 and parameter conversion unit 402;

[0145] The feature extraction unit 401 is used to extract the spatiotemporal characteristics of the wind, solar and storage project development capacity and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area according to the plurality of data indicators;

[0146] The parameter conversion unit 402 is used to convert the spatiotemporal characteristics of the development capacity of the wind, solar and storage projects in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area into small sample parameters according to a matrix probability association model; wherein the small sample parameters include: reference sequence parameters and comparison sequence parameters.

[0147] The small probability correlation model construction module 204 is used to construct a small sample probability correlation model of the wind, solar and storage policy library and the wind, solar and storage project development capacity according to the small sample parameters to obtain different probability correlation features;

[0148] As an example of this embodiment, see Figure 5 , Figure 5 A schematic diagram of a small probability correlation model building module structure of a regional wind, solar, and storage development capacity adjustment system provided by an embodiment of the present invention. Figure 5 As shown, the small probability correlation model construction module 204 is used to construct a small sample probability correlation model of the wind, solar and storage policy library and the wind, solar and storage project development capacity according to the small sample parameters to obtain different probability correlation features, and also includes:

[0149] Preprocessing unit 501, result judgment unit 502 and feature output unit 503;

[0150] The preprocessing unit 501 is used to preprocess the small sample parameters so that the small sample parameters meet the model construction requirements; wherein the small sample parameters also include: the reference sequence parameters are the spatiotemporal characteristics of the wind, solar and storage policy set extracted in the target area and the comparison sequence parameters are the spatiotemporal characteristics of the wind, solar and storage project development capacity extracted in the target area;

[0151] The result judgment unit 502 is used to judge the sample capacity of the comparison sequence parameter and obtain a judgment result;

[0152] The feature output unit 503 is used to process the sample capacity of the reference sequence parameters according to the judgment result to obtain different probability association features.

[0153] As another example of this embodiment, see Figure 6 , Figure 6 A schematic diagram of a characteristic output unit structure of a regional wind, solar, storage and development capacity adjustment system provided by an embodiment of the present invention; Figure 6 As shown, the feature output unit 503 is used to process the sample capacity of the reference sequence parameter according to the judgment result to obtain different probability association features, and also includes:

[0154] A first probability feature acquisition subunit 601 and a second probability feature acquisition subunit 602;

[0155] The first probability feature acquisition subunit 601 is used to construct an empirical distribution function according to the sample order ratio statistic when it is determined that the sample size of the comparison sequence parameter is less than a preset value, collect the sample size of the reference sequence parameter, and obtain a first probability association feature;

[0156] The second probability feature acquisition subunit 602 is used to obtain a resampled sample by resampling when it is determined that the sample size of the comparison sequence parameter is greater than or equal to a preset value, construct an empirical distribution function according to the resampled sample, collect the sample size of the reference sequence parameter, and obtain a second probability association feature.

[0157] The feature mapping processing module 205 is used to perform feature mapping processing on the correlation between the wind, solar and storage policies in the target area and the development capacity of the wind, solar and storage projects in the target area through manifold learning and BiLSTM according to the different probability association features, so as to obtain the correlation result;

[0158] As an example of this embodiment, see Figure 7 , Figure 7 A structural schematic diagram of a feature mapping processing module of a regional wind, solar, and storage development capacity adjustment system provided by an embodiment of the present invention; Figure 7 As shown, the feature mapping processing module 205 is used to perform feature mapping processing on the correlation between the wind, solar and storage policies in the target area and the development capacity of the wind, solar and storage projects in the target area according to the different probability association features through manifold learning and BiLSTM to obtain the correlation result, and also includes:

[0159] Model building unit 701, recursive feedback processing unit 702 and result output unit 703;

[0160] The model building unit 701 is used to perform forward and reverse LSTM feature mapping association according to the first probability association feature and the second probability association feature to obtain a correlation model between the wind, solar and storage policy in the target area and the development capacity of the wind, solar and storage project in the target area;

[0161] The recursive feedback processing unit 702 is used to perform recursive feedback processing on the reference sequence parameters and the comparison sequence parameters according to the correlation model between the wind, solar and storage policy in the target area and the development capacity of the wind, solar and storage projects in the target area, and output the hidden layer state of the BiLSTM network;

[0162] The result output unit 703 is used to perform manifold learning dimensionality reduction on the hidden layer state of the BiLSTM network to obtain the association result.

[0163] The development capacity adjustment module 206 is used to adjust the development capacity of wind, solar and storage in the target area according to the correlation result.

[0164] A regional wind, solar and storage development capacity adjustment system proposed in an embodiment of the present invention performs quantitative analysis on a large amount of data through a data acquisition module and a quantitative analysis module, and then refines the samples into data labels with different probability spatiotemporal characteristics through a feature extraction module and a small probability association model construction module. Finally, the samples are subjected to BiLSTM processing and popular learning processing through a feature mapping processing module, thereby achieving dimensionality reduction of data samples, reducing abnormal problems in model processing data, and greatly improving the accuracy of development capacity adjustment.

[0165] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

[0166] In the description of this specification, 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. 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, unless they are contradictory.

[0167] 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 technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

Claims

1. A regional wind, solar, storage and development capacity adjustment method, characterized in that: include: Obtain the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set; Quantitatively analyze the target area wind-solar-storage policy set and the target area wind-solar-storage project development capacity data set to obtain a number of data indicators corresponding to the target area wind-solar-storage policy set and the target area wind-solar-storage project development capacity data set; specifically: quantify the target area wind-solar-storage policy set and the target area wind-solar-storage project development capacity data set respectively to obtain a first quantification result and a second quantification result; Analyze and process the first quantification result corresponding to the second quantification result to obtain a number of data indicators; wherein the number of data indicators include: electricity price subsidy data, installed capacity data and development amounts of wind, solar and storage projects corresponding to different regions within the target processing time period; According to the several data indicators, the spatiotemporal characteristics of the development capacity of the wind, solar and storage projects in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area are extracted to obtain small sample parameters; specifically: according to the several data indicators, the spatiotemporal characteristics of the development capacity of the wind, solar and storage projects in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area are extracted respectively; according to the matrix probability association model, the spatiotemporal characteristics of the development capacity of the wind, solar and storage projects in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area are converted into small sample parameters; wherein the small sample parameters include: reference sequence parameters and comparison sequence parameters; According to the small sample parameters, a small sample probability correlation model of the wind, solar and storage policy library and the wind, solar and storage project development capacity is constructed to obtain different probability correlation features; specifically: preprocessing the small sample parameters so that the small sample parameters meet the model construction requirements; wherein the small sample parameters also include: the reference sequence parameters are the spatiotemporal characteristics of the wind, solar and storage policy set extracted in the target area and the comparison sequence parameters are the spatiotemporal characteristics of the wind, solar and storage project development capacity extracted in the target area; judging the sample capacity of the comparison sequence parameters to obtain a judgment result; according to the judgment result, processing the sample capacity of the reference sequence parameters to obtain different probability correlation features; According to the different probability association features, feature mapping processing is performed on the association between the wind, solar and storage policies in the target area and the development capacity of the wind, solar and storage projects in the target area through manifold learning and BiLSTM to obtain the association result; specifically, when it is determined that the sample capacity of the comparison sequence parameters is less than the preset value, an empirical distribution function is constructed according to the sample order ratio statistic, and the sample capacity of the reference sequence parameters is collected to obtain the first probability association feature; when it is determined that the sample capacity of the comparison sequence parameters is greater than or equal to the preset value, a resampled sample is obtained by resampling, an empirical distribution function is constructed according to the resampled sample, and the sample capacity of the reference sequence parameters is collected to obtain the second probability association feature; According to the correlation result, the development capacity of wind, solar and storage in the target area is adjusted; specifically: according to the first probability association feature and the second probability association feature, forward and reverse LSTM feature mapping associations are performed to obtain a correlation model between the wind, solar and storage policy in the target area and the development capacity of the wind, solar and storage projects in the target area; according to the correlation model between the wind, solar and storage policy in the target area and the development capacity of the wind, solar and storage projects in the target area, recursive feedback processing is performed on the reference sequence parameters and the comparison sequence parameters, and the hidden layer state of the BiLSTM network is output; manifold learning dimensionality reduction is performed on the hidden layer state of the BiLSTM network to obtain the correlation result.

2. A regional wind, solar, storage and development capacity adjustment system, characterized in that: include: Data acquisition module, quantitative analysis module, feature extraction module, small probability association model building module, feature mapping processing module and development capacity adjustment module; The data acquisition module is used to obtain the target area wind, solar and storage policy set and the target area wind, solar and storage project development capacity data set; The quantitative analysis module is used to perform quantitative analysis on the target area wind, solar, and storage policy set and the target area wind, solar, and storage project development capacity data set to obtain a number of data indicators corresponding to the target area wind, solar, and storage policy set and the target area wind, solar, and storage project development capacity data set; it also includes: a quantitative processing unit and an analytical processing unit; the quantitative processing unit is used to perform quantitative processing on the target area wind, solar, and storage policy set and the target area wind, solar, and storage project development capacity data set respectively to obtain a first quantitative result and a second quantitative result; the analytical processing unit is used to perform analytical processing on the first quantitative result corresponding to the second quantitative result to obtain a number of data indicators; wherein the several data indicators include: electricity price subsidy data, installed capacity data, and wind, solar, and storage project development amounts corresponding to different regions within the target processing time period; The feature extraction module is used to extract the spatiotemporal characteristics of the development capacity of the wind, solar and storage projects in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area according to the several data indicators to obtain small sample parameters; it also includes: a feature extraction unit and a parameter conversion unit; the feature extraction unit is used to extract the spatiotemporal characteristics of the development capacity of the wind, solar and storage projects in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area according to the several data indicators; the parameter conversion unit is used to convert the spatiotemporal characteristics of the development capacity of the wind, solar and storage projects in the target area and the spatiotemporal characteristics of the wind, solar and storage policy set in the target area into small sample parameters according to a matrix probability association model; wherein the small sample parameters include: reference sequence parameters and comparison sequence parameters; The small probability association model construction module is used to construct a small sample probability association model between the wind, solar and storage policy library and the wind, solar and storage project development capacity according to the small sample parameters to obtain different probability association features; it also includes: a preprocessing unit, a result judgment unit and a feature output unit; the preprocessing unit is used to preprocess the small sample parameters so that the small sample parameters meet the model construction requirements; wherein the small sample parameters also include: the reference sequence parameters are the spatiotemporal characteristics of the wind, solar and storage policy set extracted in the target area and the comparison sequence parameters are the spatiotemporal characteristics of the wind, solar and storage project development capacity extracted in the target area; the result judgment unit is used to judge the sample capacity of the comparison sequence parameters to obtain a judgment result; the feature output unit is used to process the sample capacity of the reference sequence parameters according to the judgment result to obtain different probability association features; The feature mapping processing module is used to perform feature mapping processing on the correlation between the wind, solar and storage policies in the target area and the development capacity of the wind, solar and storage projects in the target area through manifold learning and BiLSTM according to the different probability association features, and obtain the correlation result; it also includes: a first probability feature acquisition subunit and a second probability feature acquisition subunit; the first probability feature acquisition subunit is used to, when it is determined that the sample capacity of the comparison sequence parameters is less than a preset value, construct an empirical distribution function according to the sample order ratio statistic, collect the sample capacity of the reference sequence parameters, and obtain the first probability association feature; the second probability feature acquisition subunit is used to, when it is determined that the sample capacity of the comparison sequence parameters is greater than or equal to the preset value, obtain a resampled sample by resampling, construct an empirical distribution function according to the resampled sample, collect the sample capacity of the reference sequence parameters, and obtain the second probability association feature; The development capacity adjustment module is used to adjust the development capacity of wind, solar and storage in the target area according to the correlation result; it also includes: a model construction unit, a recursive feedback processing unit and a result output unit; the model construction unit is used to perform forward and reverse LSTM feature mapping associations according to the first probability association feature and the second probability association feature to obtain a correlation model between the wind, solar and storage policy in the target area and the development capacity of the wind, solar and storage projects in the target area; the recursive feedback processing unit is used to perform recursive feedback processing on the reference sequence parameters and the comparison sequence parameters according to the correlation model between the wind, solar and storage policy in the target area and the development capacity of the wind, solar and storage projects in the target area, and output the hidden layer state of the BiLSTM network; the result output unit is used to perform manifold learning dimensionality reduction on the hidden layer state of the BiLSTM network to obtain the correlation result.

Images