Uncertainty analysis and mode recognition method and system for single high-fluctuation new energy station

Abstract

The invention relates to the technical field of new energy power, and discloses an uncertainty analysis and mode recognition method and system for a single high-fluctuation new energy station, and the method comprises the steps: collecting the data of the new energy station, and carrying out the data preprocessing; constructing a probability distribution model of power and voltage fluctuation, extracting time sequence features of the new energy station by adopting a fluctuation mode recognition algorithm, and recognizing an abnormal fluctuation mode; uncertainty quantification is carried out based on the power and voltage fluctuation characteristics of the new energy station; through multi-level sensitivity quantitative analysis of the target voltage on a single observation point variable, the influence level of the target voltage is evaluated, and a new energy station scheduling and control strategy is optimized. According to the method, by fusing uncertainty quantification, deep learning and dynamic optimization technologies, systematic innovation is formed in the aspects of prediction precision, risk control, mode recognition, decision support and the like, and key technical support is provided for grid-connected safety and efficient scheduling of a high-fluctuation new energy station.

Description

Technical Field

[0001] The present invention relates to the field of new energy power technology, and particularly to an uncertainty analysis and pattern recognition method and system for a single high-fluctuation new energy power station. Background Art

[0002] In the current field of new energy power generation, new energy power stations such as wind power and solar power have large power output and voltage fluctuations due to factors such as climate and environment, and have strong uncertainty. This high volatility not only affects the stable operation of the power grid, but also brings huge challenges to the grid connection control and scheduling of new energy.

[0003] Currently, for the prediction of power and voltage fluctuations in new energy power stations, it mainly relies on statistical methods and machine learning algorithms, such as time series analysis, deep learning prediction models, etc. However, these methods often have the following problems when dealing with high-volatility scenarios. For example, the prediction accuracy is limited, and existing prediction models are difficult to fully capture the uncertainty of power and voltage in new energy power stations, resulting in a decrease in prediction accuracy; the pattern recognition ability is insufficient, and existing methods are difficult to effectively identify the patterns of fluctuations in new energy power stations, affecting the in-depth analysis and understanding of fluctuation characteristics; there is a lack of an effective uncertainty analysis framework, and existing research often focuses on power prediction itself, lacking a systematic uncertainty analysis framework and making it difficult to quantitatively study the fluctuation characteristics of new energy power stations; the variable influence relationship is complex, and how the variables at a single observation point affect different levels of the target voltage has not been fully quantified and theoretically analyzed, affecting subsequent decision-making and control optimization. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station, which can improve the prediction accuracy of power and voltage in new energy power stations, and provide theoretical support and decision-making basis for the scheduling and control optimization of new energy power grids.

[0006] To solve the above technical problems, the present invention provides the following technical solution. An uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station includes: collecting data of the new energy power station and performing data preprocessing; constructing a probability distribution model of power and voltage fluctuations, using a fluctuation pattern recognition algorithm to extract time series features of the new energy power station and identify abnormal fluctuation patterns; performing uncertainty quantification based on the power and voltage fluctuation characteristics of the new energy power station; and evaluating the influence level of the target voltage through multi-level sensitivity quantification analysis of the target voltage to variables at a single observation point, and optimizing the scheduling and control strategies of the new energy power station.

[0007] As a preferred solution of the uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station according to the present invention, wherein: the data of the new energy power station includes the historical power, meteorological data and voltage fluctuation data of the new energy power station;

[0008] The data preprocessing includes: dividing the historical data set into a training set and a test set, and constructing model input data through a time series sample generation method.

[0009] As a preferred solution of the uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station according to the present invention, wherein: the construction of the probability distribution model of power and voltage fluctuations includes using a generative probability model to model the potential distribution of power and voltage fluctuations, and its loss function includes a data reconstruction error and a distribution regularization term;

[0010] The extraction of the time series features of the new energy power station includes extracting the local patterns of the input features through a convolutional neural network and capturing the time series dynamic characteristics in combination with a recurrent neural network.

[0011] As a preferred solution of the uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station according to the present invention, wherein: the uncertainty quantification includes using a non-parametric regression method to construct a prediction model and defining a similarity measure of the input data through a kernel function;

[0012] Classify the fluctuation features based on a density clustering algorithm or a graph theory clustering algorithm, wherein the clustering parameters are adaptively adjusted according to the data distribution.

[0013] As a preferred solution of the uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station according to the present invention, wherein: the identification of abnormal fluctuation patterns includes generating a data segmentation path through a recursive segmentation algorithm and calculating an abnormal score according to the depth of the segmentation path.

[0014] As a preferred solution of the uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station according to the present invention, wherein: the multi-level sensitivity quantification analysis includes calculating the contribution rate of each input variable to the target parameter based on a variance decomposition method and evaluating the joint influence of the variables in combination with a game theory method;

[0015] When the prediction error exceeds a preset threshold, retrain the deep learning model and adjust the noise parameters of the Bayesian inference model.

[0016] As a preferred solution of the uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station according to the present invention, wherein: the optimization of the new energy power station scheduling and control strategy includes adjusting the power output strategy or voltage regulation parameters of the new energy power station based on the sensitivity analysis results to achieve the dynamic stable control of the power grid.

[0017] As a preferred solution of the uncertainty analysis and pattern recognition system for a single high-fluctuation new energy power station according to the present invention, it includes: a data processing module, a fluctuation pattern recognition module, an uncertainty quantification module, a multi-level sensitivity analysis module, and a scheduling optimization module;

[0018] The data processing module is responsible for collecting the historical power, meteorological data and voltage fluctuation data of the new energy power station, and processing the original data, including dividing the data into a training set and a test set;

[0019] The fluctuation pattern recognition module uses a generative probability model and convolutional and recurrent neural networks to extract the time series features of power and voltage fluctuations, and identify abnormal fluctuation patterns;

[0020] The uncertainty quantification module classifies and predicts the fluctuation characteristics based on a clustering algorithm and a non-parametric regression method, quantifies the uncertainty and evaluates its impact;

[0021] The multi-level sensitivity analysis module uses variance decomposition and game theory methods to evaluate the contribution rate of each input variable to the target parameter, and analyzes the joint influence of variables;

[0022] The scheduling optimization module adjusts the power output strategy or voltage regulation parameters of the new energy power station according to the sensitivity analysis results to achieve the dynamic stable control of the power grid.

[0023] A computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of the uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station are implemented.

[0024] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station are implemented.

[0025] The beneficial effects of the present invention: improve the prediction accuracy of power and voltage, the prediction error is lower than 3.5%, significantly better than the existing methods; identify the fluctuation patterns of new energy power stations, optimize the power grid scheduling, and improve the stability of the power grid and the new energy consumption capacity; provide the uncertainty quantification analysis of the power and voltage of new energy power stations, and improve the credibility of the prediction results; reveal the complex relationship between the target voltage and the variables of a single observation point, and improve the regulation accuracy of the new energy power grid; widely applicable to high-fluctuation new energy power stations such as wind power and photovoltaic, and improve the stability of new energy grid connection. Description of the Drawings

[0026] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0027] Figure 1 Schematic flow diagram of the uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station provided by an embodiment of the present invention.

[0028] Figure 2 Schematic diagram of the working modules of the uncertainty analysis and pattern recognition system for a single high-fluctuation new energy power station provided by an embodiment of the present invention. Detailed implementation manners

[0029] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific implementation manners of the present invention in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0030] In the following description, many specific details are set forth to facilitate a thorough understanding of the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0031] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selectively exclusive embodiment from other embodiments.

[0032] The present invention is described in detail in conjunction with the schematic diagrams. When detailing the embodiments of the present invention, for the convenience of explanation, the cross-sectional views showing the device structures will be enlarged locally not in accordance with the general scale, and the schematic diagrams are only examples and should not limit the protection scope of the present invention herein. In addition, in actual production, three-dimensional spatial dimensions including length, width, and depth should be included.

[0033] In the description of the present invention, it should be noted that the terms "upper, lower, inner, and outer" and other references to orientations or positional relationships are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first, second, or third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0034] In this disclosure, unless otherwise specified or limited, the terms "mounted," "connected," and "connected" should be interpreted broadly. For example, they may refer to fixed, removable, or integral connections. They may also refer to mechanical, electrical, or direct connections, indirect connections through an intermediary, or internal communication between two components. Those skilled in the art will understand the specific meanings of these terms in this disclosure.

[0035] Example 1, reference Figure 1 , which is the first embodiment of the present invention, provides an uncertainty analysis and pattern recognition method for a single high-fluctuation new energy station, including:

[0036] S1: Collect data from new energy stations and perform data preprocessing.

[0037] Furthermore, historical power, meteorological data, and voltage fluctuation data of new energy stations are collected, and data cleaning, noise reduction, and normalization are performed.

[0038] It should be noted that the historical observation data set of the new energy station is:

[0039] X={x t |t=1,2,...,T}

[0040] Among them, x t Represents the input feature vector at time t, including power, meteorological data, voltage fluctuation data, etc.

[0041] S2: Construct a probability distribution model for power and voltage fluctuations, use a fluctuation pattern recognition algorithm to extract the time series characteristics of new energy stations, and identify abnormal fluctuation patterns.

[0042] Furthermore, CNN performs feature extraction using a one-dimensional convolutional neural network (1DCNN) for temporal feature extraction:

[0043] h t =f(W c *X t +b c )

[0044] Where: Wc is the weight matrix of the CNN convolution kernel, * represents the convolution operation, b c is the bias term, and f(·) is the ReLU activation function.

[0045] It should be noted that LSTM performs temporal modeling and CNN extracts the feature sequence H = {h t Input to the Long Short-Term Memory Network (LSTM):

[0046] i t =σ(W i h t +U i c t-1 +b i )

[0047] f t =σ(W f h t +U f c t-1 +b f )

[0048] o t =σ(W o h t +U o c t-1 +b o )

[0049] c t =f t Θc t-1 +i t Θtanh(W c h t +U c c t-1 +b c )

[0050] h t =o t Θtanh(c t )

[0051] Where: i t ,f t ,o t are the activation values of the input gate, forget gate, and output gate respectively, c t is the memory cell state, W i ,W f ,W o ,W c ,U i ,U f ,U o ,U c is the network weight matrix, b i ,bf , b o , b c is the bias term, σ(·) represents the Sigmoid activation function, and Θ represents the Hadamard multiplication.

[0052] The final time series feature h is obtained after LSTM processing T , which is used for power and voltage prediction.

[0053] Furthermore, GPR uses Bayesian inference for prediction and provides a confidence interval. Let the input of the model be X and the output be y. Assume that y follows a Gaussian process:

[0054] y|X ∼ N(μ(X), K(X, X))

[0055] where: μ(X) is the mean function, usually assumed to be zero mean, i.e., μ(X) = 0, and K(X, X) is the covariance function (kernel function), and the commonly used RBF kernel is:

[0056]

[0057] where: σ f is the signal standard deviation, and l is the length scale parameter.

[0058] GPR prediction result:

[0059]

[0060] where: X * is the new input data, is the noise variance, and I is the identity matrix.

[0061] The uncertainty estimate is given by the prediction variance:

[0062]

[0063] S3: Based on the power and voltage fluctuation characteristics of the new energy power station, uncertainty quantification is performed.

[0064] Furthermore, deep learning modeling (LSTM / VAE / Transformer)

[0065] In order to extract the time series characteristics of the power and voltage fluctuations of the new energy power station, this application uses long short-term memory network (LSTM), variational autoencoder (VAE) and Transformer for time series modeling.

[0066] LSTM is suitable for processing time series data and can capture the time-dependent relationship of the power and voltage fluctuations of the new energy power station. The calculation method is as follows:

[0067] Input gate: it = σ(W i x t + U i h t-1 + b i )

[0068] Forget gate: f t = σ(W f x t + U f h t-1 + b f )

[0069] Memory cell update: c t = f t c t-1 + i t tanh(W c x t + U c h t-1 + b c )

[0070] Output gate: o t = σ(W o x t + U o h t-1 + b o )

[0071] Final hidden state: h t = o t tanhc(t)

[0072] The variational autoencoder (VAE) is used to learn the fluctuation feature distribution of new energy power stations, so as to achieve pattern recognition. Its modeling method is as follows:

[0073] Encoder: z = μ + σ·ε, ε ~ N(0,1)

[0074] Decoder: x' = f(z)

[0075] Loss function: L = ||x - x'|| 2 + D KL (q(z|x)||p(z))

[0076] where D KL is the Kullback-Leibler divergence, which is used to ensure the regularization of latent variables.

[0077] Transformer uses the self-attention mechanism (Self-Attention) to calculate the fluctuation pattern of new energy power stations:

[0078] Attention calculation:

[0079] Output calculation: H = LayerNorm(MultiHead(X) + X)

[0080] It should be noted that K-Means is mainly used for the classification of fluctuation patterns in new energy power stations:

[0081] Objective function:

[0082] where C i is the i-th clustering cluster, and μ i is the cluster center.

[0083] DBSCAN identifies fluctuation patterns through core points, border points, and noise points: A core point is a point with at least MinPts neighbors; A border point is a point with fewer than MinPts but belonging to the neighborhood of a core point; A noise point is a point that does not belong to the neighborhood of any core point.

[0084] DBSCAN adopts the principle of density reachability. If the Euclidean distance d(p, q) between point p and point q ≤ ε and p is a core point, then q belongs to the cluster of p.

[0085] Spectral clustering performs pattern classification by calculating the Laplacian matrix and constructs a similarity matrix:

[0086]

[0087] Calculate the Laplacian matrix: L = D - W, and take the first k eigenvectors of the Laplacian matrix for K-Means clustering.

[0088] Furthermore, Isolation Forest detects anomalies by recursively randomly partitioning the data, sets a random threshold, and randomly partitions the dataset X: Left subtree, Right subtree, and calculate the anomaly score:

[0089]

[0090] where E(h(x)) is the average isolation depth of the data point, and E(h(x)) is the normalization factor.

[0091] LOF calculates the local density of the data point, local reachability density:

[0092]

[0093] Anomaly factor calculation:

[0094]

[0095] If LOFk If (p) >> 1, then point p may be an outlier.

[0096] In another alternative embodiment of the present invention, based on the power and voltage fluctuation characteristics of new energy power stations, a clustering analysis method is used for pattern classification. Specifically, K-means clustering, DBSCAN (density-based clustering), and spectral clustering are mainly used for pattern recognition. The following is the specific mathematical derivation.

[0097] Let the dataset of the new energy power station be:

[0098] X = {x t} where t = 1, 2,..., T

[0099] where x t is the input feature vector at time t, including: power volatility

[0100]

[0101] voltage volatility wind speed change rate photovoltaic irradiance change rate

[0102] where P(t) is the power at time t, V(t) is the voltage, W(t) is the wind speed, and S(t) is the photovoltaic irradiance. Finally, the feature matrix is obtained:

[0103]

[0104] Furthermore, K-means clustering classifies based on minimizing the within-cluster variance. Given K classes, the cluster centers are defined as:

[0105] C = {c1, c2,..., c K}

[0106] Optimization objective:

[0107]

[0108] where: C i is the i-th cluster, x j is the data point belonging to this cluster, and c i is the center point of the cluster. Iterative updates are performed:

[0109]

[0110] DBSCAN identifies fluctuation patterns through density-connected methods, setting: neighborhood radius ε and minimum number of points MinPts; defining core points where N(x)≥MinPts, border points where N(x)<MinPts, and noise points that do not belong to any cluster.

[0111] The clustering rule is to select an unvisited core point x i as the center of a new cluster. Add all points within the neighborhood radius ε to this cluster. If a point is also a core point, continue to expand the cluster until no new points can be added.

[0112] It should be noted that spectral clustering is based on graph theory methods and uses the Laplacian matrix for dimensionality reduction. Given a similarity matrix:

[0113]

[0114] Define the degree matrix: D ii =ΣW ij ; Define the Laplacian matrix: L = D - W

[0115] Calculate the first K eigenvectors of L as the dimensionality-reduced data, and then use K-means for classification.

[0116] The interaction process of the present invention is mainly reflected in the dynamic closed-loop among new energy power station data collection, modeling analysis, prediction feedback, and control optimization. K-means or DBSCAN is used for fluctuation pattern classification to identify high-volatility periods. CNN-LSTM and Transformer are combined with Gaussian process regression (GPR) for power and voltage prediction. The confidence interval is calculated through the Monte Carlo method to identify the prediction error range. If the prediction error exceeds the preset threshold, the prediction model parameters are adjusted to improve prediction stability. Combining sensitivity analysis, the operation strategies of new energy power stations are adjusted (such as adjusting the energy storage system, adjusting the fan angle, etc.). Based on the optimized prediction results, the power dispatching and grid control strategies of new energy power stations are guided.

[0117] This interaction process forms a closed-loop optimization, continuously improving the accuracy of power and voltage prediction of new energy power stations. S4: Through the multi-level sensitivity quantification analysis of the target voltage on the single observation point variable, evaluate the influence level of the target voltage, and optimize the dispatching and control strategies of new energy power stations.

[0118] Furthermore, the uncertainty analysis and evaluation process mainly uses Gaussian process regression (GPR) and sensitivity analysis (Sobol index, Shapley value, variance decomposition), and the specific calculation process is as follows:

[0119] Gaussian process regression (GPR) uncertainty modeling, assuming the model input is X and the output is y, and assuming y follows a Gaussian process:

[0120] y(x) ~ GP(m(x), K(X, X'))

[0121] Among them, m(X) is the mean function, and K(X, X') is the covariance function. The covariance is commonly calculated using the radial basis kernel function (RBF kernel):

[0122]

[0123] Among them: σ f is the signal standard deviation; l is the scale parameter of the kernel function.

[0124] The GPR prediction value is The prediction uncertainty is estimated by the variance:

[0125]

[0126] Among them: X * is the point to be predicted; is the noise variance; I is the identity matrix.

[0127] The Sobol index, Shapley value, and variance decomposition method are used to quantify the influence of each variable on power and voltage fluctuations: to measure the contribution rate of uncertainty of each input variable to the output. Let the total variance be V(Y), and the Sobol index of a certain variable is calculated as follows:

[0128]

[0129] The Shapley value calculation is used to evaluate the marginal contribution of each variable:

[0130]

[0131] Calculate the variance contribution rate of different variables to determine the main influencing factors.

[0132] It should be noted that the uncertainty assessment in the first part is mainly used for historical data analysis. By using methods such as Bayesian networks and VAEs, a probability distribution model of power and voltage is constructed to quantify the inherent uncertainty of new energy power stations. The subsequent uncertainty prediction focuses on future trend prediction. Combining Gaussian process regression (GPR) or Bayesian neural network (BNN), the power and voltage changes at future moments are predicted, and a confidence interval is provided. The front and back are associated to form a complete uncertainty analysis system.

[0133] Example 2, which is an example of the present invention, provides an uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0134] This experiment uses the power, meteorological, and voltage data of a certain new energy power station in the past two years for training and testing. The data sampling period is 5 minutes, including: wind speed (m / s), photovoltaic irradiance (W / m2), power output (MW), and voltage (V); the dataset contains more than 20,000 groups of data, which are divided into a training set (80%) and a test set (20%). The experimental results are shown in Table 1.

[0135] The mean squared error (MSE), root mean squared error (RMSE), and mean squared logarithmic error (MSLE) are used to evaluate the prediction accuracy:

[0136]

[0137] Table 1: Data Table of Experimental Results

[0138]

[0139] This method (CNN-LSTM+GPR) is superior to other models in all evaluation indicators, indicating that the prediction error is reduced (RMSE drops from 2.26 to 1.79) and the uncertainty assessment is more stable (the lowest MSLE).

[0140] The confidence interval (95% CI) is used to analyze the prediction credibility:

[0141]

[0142] where: is the GPR predicted value, and σ y is the prediction standard deviation. The experimental results are shown in Table 2.

[0143] Table 2: Data Table of Experimental Results

[0144]

[0145] The uncertainty interval provided by GPR reasonably covers the true value, indicating that the uncertainty assessment of the model is reliable.

[0146] This invention uses a CNN-LSTM combined model + GPR for power and voltage prediction of new energy power stations. Mathematical derivations and experimental data show that: the prediction error is as low as 3.21 (MSE), significantly superior to traditional methods; GPR provides reasonable uncertainty quantification, enhancing prediction credibility; the confidence interval can cover the true value, indicating that the uncertainty prediction is effective. The experimental results are shown in Table 3

[0147] K-means, DBSCAN, and spectral clustering are used to classify the fluctuation patterns of new energy power stations. For K-means, K = 3 is selected, and for DBSCAN, ε = 0.1 and MinPts = 10 are set.

[0148] Table 3: Experimental result data table

[0149] Clustering method Cluster 1 Cluster 2 Cluster 3 K-means Low volatility mode Medium volatility mode High volatility mode DBSCAN Normal volatility Abnormal volatility Noise point Spectral clustering Short-term sharp fluctuations Medium- and long-term fluctuations Steady state

[0150] The proportion of outliers found by DBSCAN is 5.3%, indicating that there are sudden fluctuations in new energy power stations. The silhouette coefficient is used to evaluate the classification effect:

[0151]

[0152] Where: a i is the average distance of sample x i within the same cluster, and b i is the average distance from x i to the nearest cluster. The experimental results are shown in Table 4.

[0153] Table 4: Experimental result data table

[0154] Method Silhouette coefficient K-means 0.72 DBSCAN 0.81 Spectral clustering 0.76

[0155] The DBSCAN classification effect is the best, indicating that the fluctuation mode of new energy power stations has density characteristics. The present invention uses K-means, DBSCAN, and spectral clustering for fluctuation mode classification. Mathematical derivation shows that DBSCAN is suitable for abnormal fluctuation detection, and K-means is suitable for stable mode classification. The experimental results show that the DBSCAN silhouette coefficient is the highest (0.81), which can effectively identify abnormal fluctuations in new energy power stations.

[0156] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not restrictive. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

[0157] Example 3, the third example of the present invention, which is different from the previous two examples in that:

[0158] When the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0159] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in combination with an instruction execution system, apparatus, or device.

[0160] More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection parts with one or more wirings (electronic devices), portable computer disk cartridges (magnetic devices), random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memories), optical fiber devices, and portable compact disc read-only memories (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting, or processing it in other suitable ways if necessary, and then storing it in a computer memory.

[0161] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc.

[0162] Example 4, referring to Figure 2 , which is an embodiment of the present invention, provides an uncertainty analysis and pattern recognition system for a single high-fluctuation new energy power station, including a data processing module, a fluctuation pattern recognition module, an uncertainty quantification module, a multi-level sensitivity analysis module, and a scheduling optimization module;

[0163] The data processing module is responsible for collecting historical power, meteorological data, and voltage fluctuation data of the new energy power station, and processing the original data, including dividing the data into a training set and a test set;

[0164] The fluctuation pattern recognition module uses generative probability models and convolutional and recurrent neural networks to extract time series features of power and voltage fluctuations, and identify abnormal fluctuation patterns;

[0165] The uncertainty quantification module classifies and predicts the fluctuation characteristics based on clustering algorithms and non-parametric regression methods, quantifies the uncertainty and evaluates its impact;

[0166] The multi-level sensitivity analysis module uses variance decomposition and game theory methods to evaluate the contribution rate of each input variable to the target parameter, and analyzes the joint influence of the variables;

[0167] The scheduling optimization module adjusts the power output strategy or voltage regulation parameters of the new energy power station according to the sensitivity analysis results to achieve the dynamic stable control of the power grid.

[0168] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. Uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station, characterized in that: Including, Collecting data from new energy power stations and performing data preprocessing; Constructing a probability distribution model for power and voltage fluctuations, adopting a fluctuation pattern recognition algorithm, extracting the time series characteristics of new energy power stations, and identifying abnormal fluctuation patterns; Quantifying uncertainty based on the power and voltage fluctuation characteristics of new energy power stations; Evaluating the influence level of the target voltage through multi-level sensitivity quantification analysis of the single observation point variable with respect to the target voltage, and optimizing the dispatching and control strategies of new energy power stations.

2. The uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station according to claim 1, characterized in that: The data of the new energy power station includes historical power, meteorological data, and voltage fluctuation data of the new energy power station; The data preprocessing includes: dividing the historical data set into a training set and a test set, and constructing model input data through a time series sample generation method.

3. The uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station according to claim 2, wherein: The constructing of the probability distribution model for power and voltage fluctuations includes using a generative probability model to model the potential distribution of power and voltage fluctuations, and its loss function includes a data reconstruction error and a distribution regularization term; The extracting of the time series characteristics of new energy power stations includes extracting the local patterns of input features through a convolutional neural network and capturing the time series dynamic characteristics in combination with a recurrent neural network.

4. The uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station according to claim 3, characterized in that: The uncertainty quantification includes using a non-parametric regression method to construct a prediction model and defining a similarity measure of input data through a kernel function; Classifying the fluctuation characteristics based on a density clustering algorithm or a graph theory clustering algorithm, where the clustering parameters are adaptively adjusted according to the data distribution.

5. The uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station according to claim 4, wherein: The identifying of abnormal fluctuation patterns includes generating a data segmentation path through a recursive segmentation algorithm and calculating an abnormal score based on the depth of the segmentation path.

6. The uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station according to claim 5, characterized in that: The multi-level sensitivity quantification analysis includes calculating the contribution rate of each input variable to the target parameter based on a variance decomposition method and evaluating the joint influence of variables in combination with a game theory method; When the prediction error exceeds a preset threshold, retraining the deep learning model and adjusting the noise parameters of the Bayesian inference model.

7. The uncertainty analysis and pattern recognition method for a single high-fluctuation new energy power station according to claim 6, characterized in that: The optimizing of the dispatching and control strategies of new energy power stations includes adjusting the power output strategy or voltage regulation parameters of new energy power stations based on the sensitivity analysis results to achieve the dynamic stable control of the power grid.

8. A system adopting the uncertainty analysis and pattern recognition method of a single high-fluctuation new energy power station as described in any one of claims 1 to 7, characterized in that: Including a data processing module, a fluctuation pattern recognition module, an uncertainty quantification module, a multi-level sensitivity analysis module, and a dispatching optimization module; The data processing module is responsible for collecting historical power, meteorological data, and voltage fluctuation data of new energy power stations and processing the original data, including dividing the data into a training set and a test set; The fluctuation pattern recognition module uses a generative probability model and convolutional and recurrent neural networks to extract the time series characteristics of power and voltage fluctuations and identify abnormal fluctuation patterns; The uncertainty quantification module classifies and predicts the fluctuation characteristics based on a clustering algorithm and a non-parametric regression method, quantifies the uncertainty and evaluates its influence; The multi-level sensitivity analysis module uses a variance decomposition and a game theory method to evaluate the contribution rate of each input variable to the target parameter and analyze the joint influence of variables; The dispatching optimization module adjusts the power output strategy or voltage regulation parameters of new energy power stations according to the sensitivity analysis results to achieve the dynamic stable control of the power grid.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, 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 7 are implemented.

Images