Data-driven wide frequency oscillation risk assessment method for wind power transmission system via flexible direct current
Patent Information
- Application Number
- CN202610951139.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-29
- Publication Date
- 2026-10-30
AI Technical Summary
模拟法一般借助蒙特卡洛法对系统可能的运行状态进行抽样，该方法可以计及多种随机因素，但需要对系统重复进行潮流计算与稳定性分析，计算量大，难以适用于大规模复杂系统
本发明定义了宽频振荡风险系数，可以用于评估系统在计及风电等不确定因素下的宽频振荡风险；基于数据驱动方法建立了风电经柔直送出系统宽频振荡稳定性评估模型，进而进行宽频振荡稳定校验，显著提高了计算效率；基于GAN生成多风电场风速场景，模拟风电经柔直送出系统的运行状态，可以真实反映系统在多种随机因素下的运行状态。
Smart Images
Figure CN123014661A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of broadband oscillation risk assessment technology for wind power grid-connected systems, and particularly to a data-driven method for broadband oscillation risk assessment of wind power systems transmitted via flexible direct current. Background Technology
[0002] Existing broadband oscillation risk assessment methods are mostly designed for specific operating conditions. Due to the randomness of wind speed, the operating conditions of wind power grid-connected systems are highly uncertain, and stability analysis under specific operating conditions is difficult to reflect the oscillation risk of actual wind power grid-connected systems. Therefore, probabilistic stability assessment is needed. Probabilistic stability analysis refers to determining the probabilistic stability index of a system based on the statistical characteristics of random factors affecting system stability.
[0003] Currently, there are two main methods: simulation and analytical methods. Analytical methods represent the system's eigenvalues as functions of random variables, approximating the probability distribution of the eigenvalues from the probability distribution of the random variables. This method typically requires simplification of the system model, thus reducing computational accuracy. Simulation methods generally use the Monte Carlo method to sample the possible operating states of the system. This method can account for various random factors, but it requires repeated power flow calculations and stability analyses, resulting in a large computational load and making it unsuitable for large-scale complex systems. Summary of the Invention
[0004] The main objective of this invention is to provide a data-driven method for assessing the broadband oscillation risk of wind power transmitted via flexible direct transmission systems.
[0005] Another objective of this invention is to provide a data-driven broadband oscillation risk assessment device for wind power transmitted via flexible direct transmission systems.
[0006] The third objective of this invention is to provide an electronic device.
[0007] A fourth objective of this invention is to provide a non-transitory computer-readable storage medium.
[0008] To achieve the above objectives, a first aspect of the present invention proposes a data-driven method for assessing broadband oscillation risk in wind power transmission systems via flexible direct current transmission, comprising:
[0009] S1. Construct a broadband oscillation stability assessment model, taking the combined output active power of each wind turbine generator set in each wind farm as the input feature, and outputting the judgment result of whether the system is oscillating and stable through the trained neural network model. S2: Obtain historical wind speed data from multiple wind farms and train a generative adversarial network. Use the trained generator to generate diverse wind speed scene sequences from multiple wind farms that conform to historical distribution characteristics. S3, based on the relationship between wind speed and active power conversion of wind turbine generators, converts the generated diverse wind speed scene sequences of multiple wind farms into corresponding system operation status sampling data sets. S4. Input each operating state sample from the system operating state sampling data set into the broadband oscillation stability assessment model for stability determination, and count the proportion of samples that do not meet the broadband oscillation stability constraints to the total number of samples to obtain the broadband oscillation risk coefficient.
[0010] Optionally, a broadband oscillation stability assessment model is constructed, using the combined output active power of individual wind turbine generators in each wind farm as input features. The trained neural network model outputs a determination of whether the system is oscillating and stable, including: The active power output of each wind turbine generator in each wind farm in the system is used as input to form a large number of different operating conditions. For each operating condition, its stability is judged according to the broadband oscillation stability analysis method. A dataset containing input vectors and output labels is constructed, where the input vector is a sequence of active power output of each wind turbine generator in each wind farm, and the output label is whether the system is oscillating and stable. A multi-layer feedforward neural network model is built. The number of hidden layers and the number of neurons in each layer are set according to experience. Cross-entropy is used as the loss function. The backpropagation algorithm combined with the gradient descent strategy is used to iteratively optimize the model parameters until the model converges to obtain the trained neural network model. The trained neural network model is used to calculate the input features and outputs the probability that the system is oscillating and stable or unstable in the current state. The class with the higher probability is taken as the final evaluation result.
[0011] Optionally, historical wind speed data from multiple wind farms can be acquired and a generative adversarial network (GAN) can be trained. The trained generator can then be used to generate diverse multi-wind farm wind speed scene sequences that conform to historical distribution characteristics, including: Collect historical wind speed data of multiple wind farms in the system through which wind power is transmitted via flexible direct transmission, and represent each sample as a multi-dimensional time series or a multi-point wind speed matrix, where the multi-dimensional time series is the wind speed sequence of each wind farm in the same time period. The original data is normalized to construct a training sample set for training the generative adversarial network. A generative adversarial network consisting of a generator G and a discriminator D is constructed, where the input of the generator G is a random noise vector and the output is a synthetic wind speed sample, and the input of the discriminator D is a real sample or a generated sample and the output is the probability that the sample is real data. An alternating optimization strategy is used to train the generator and discriminator, enabling the generator to gradually learn the statistical characteristics of wind speed data from multiple wind farms. After training, the generator G is fixed to generate diverse wind speed scenarios that conform to historical distribution characteristics.
[0012] Optionally, based on the conversion relationship between wind speed and active power of wind turbine generators, the generated diverse multi-wind farm wind speed scenario sequences are converted into corresponding system operation status sampling data sets, including: Set the required number of scenes as follows Random sampling from the prior distribution noise vectors ,in Values ; The noise vector Input the trained generator To obtain the corresponding wind speed scene :
[0013] Under the maximum wind energy tracking control strategy, based on the rated wind speed of the wind turbine generator... Cut-in wind speed Cut-off wind speed and rated active power By utilizing the wind speed-power conversion relationship, the wind speed of each wind farm in each wind speed scenario is converted into the active power output of a single wind turbine generator. ; The active power output of each individual wind turbine generator in all wind farms is combined to form a sample data set of the operating status of the wind power transmission system via flexible direct current. .
[0014] Optionally, the operating state samples from the system operating state sampling data set are input into the broadband oscillation stability assessment model for stability determination. The proportion of samples that do not meet the broadband oscillation stability constraints is calculated to obtain the broadband oscillation risk coefficient, including: Traverse the system operation status sampling data set Each running state sample The sample is then input into the broadband oscillation stability evaluation model, and the broadband oscillation stability judgment criterion is used to determine whether the sample meets the broadband oscillation stability constraint. Statistical sets Number of samples with mid-bandwidth oscillation instability and the number of samples with stable broadband oscillations ; Calculate the percentage of wideband oscillatory unstable samples in the total number of samples. The ratio is used to obtain the broadband oscillation risk coefficient. :
[0015] Among them, the larger the broadband oscillation risk coefficient, the higher the broadband oscillation risk faced by the system.
[0016] To achieve the above objectives, a second aspect of the present invention provides a data-driven broadband oscillation risk assessment device for wind power transmitted via flexible direct current systems, comprising: The module is used to build a broadband oscillation stability assessment model. It takes the combined output active power of each wind turbine generator in each wind farm as input features and outputs the judgment result of whether the system is oscillating and stable through the trained neural network model. The training module is used to acquire historical wind speed data from multiple wind farms and train a generative adversarial network. The trained generator is then used to generate diverse wind speed scene sequences from multiple wind farms that conform to historical distribution characteristics. The conversion module is used to convert the generated diverse wind speed scene sequences of multiple wind farms into corresponding system operation status sampling data sets based on the conversion relationship between wind speed and active power of wind turbine generators. The determination module is used to input each operating state sample from the system operating state sampling data set into the broadband oscillation stability assessment model for stability determination, and to count the proportion of samples that do not meet the broadband oscillation stability constraints to the total number of samples, thereby obtaining the broadband oscillation risk coefficient.
[0017] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.
[0018] To achieve the above objectives, a third aspect of this application provides an electronic device, including a processor and a memory; wherein the processor runs a program corresponding to the executable program code stored in the memory, for implementing the data-driven broadband oscillation risk assessment method for wind power transmitted via flexible direct transmission system as described in the first aspect embodiment.
[0019] To achieve the above objectives, a fourth aspect of this application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the data-driven broadband oscillation risk assessment method for wind power transmitted via flexible direct current systems as described in the first aspect embodiment.
[0020] The embodiments of the present invention have the following beneficial effects: This invention defines a broadband oscillation risk coefficient, which can be used to assess the broadband oscillation risk of a system under uncertainties such as wind power. Based on a data-driven method, a broadband oscillation stability assessment model for a wind power transmission system via flexible direct transmission is established, and broadband oscillation stability verification is performed, significantly improving computational efficiency. Based on GAN, multiple wind farm wind speed scenarios are generated to simulate the operating state of the wind power transmission system via flexible direct transmission, which can realistically reflect the operating state of the system under various random factors. Attached Figure Description
[0021] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 A flowchart of a data-driven broadband oscillation risk assessment method for wind power transmitted via flexible direct transmission system is provided for embodiments of the present invention. Figure 2 A flowchart of another data-driven broadband oscillation risk assessment method for wind power transmitted via flexible direct transmission system provided in this embodiment of the invention; Figure 3 This is a structural diagram of a data-driven broadband oscillation risk assessment device for a wind power transmission system via flexible direct transmission, provided as an embodiment of the present invention. Detailed Implementation
[0022] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.
[0023] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0024] The following description, with reference to the accompanying drawings, describes a data-driven broadband oscillation risk assessment method and apparatus for wind power transmitted via flexible direct transmission systems, based on embodiments of the present invention.
[0025] Example 1 This invention provides a data-driven method for assessing broadband oscillation risks in wind power transmission systems via flexible direct current transmission, such as... Figure 1 As shown, the method includes the following steps: S1. Construct a broadband oscillation stability assessment model, taking the combined output active power of each wind turbine generator in each wind farm as the input feature, and outputting the judgment result of whether the system is oscillating and stable through the trained neural network model.
[0026] To achieve accurate assessment of broadband oscillation stability of wind power transmission systems via flexible direct current transmission, this application constructs and trains a neural network-based assessment model, combines dataset construction and stability judgment criteria, to complete the stability assessment of the system's operating status, providing model support for subsequent risk assessment.
[0027] In this embodiment of the application, a stability assessment system based on neural networks is constructed to address the risk assessment requirements of broadband oscillations in wind power transmission systems via flexible direct power transmission. The specific implementation process is as follows: First, in this embodiment, the core input quantity is the active power output of a single wind turbine generator in each wind farm within the flexible direct transmission system. This is artificially combined to form a massive number of differentiated operating conditions. For each condition, existing broadband oscillation stability analysis methods are used to determine its stable state, thereby constructing a standardized dataset. The dataset is expressed in the following form:
[0028] in, and These are the input and output of the dataset, respectively; the input is the feature vector. This is a vector consisting of the active power output of each individual wind turbine generator in each wind farm. , To output active power for each individual wind turbine generator unit in each wind farm. This dataset represents the total number of wind farms within the system, providing reliable data support for subsequent neural network model training.
[0029] In this embodiment, a multi-layer feedforward neural network model is constructed. The number of hidden layers and neurons in each layer are reasonably set based on practical engineering experience. Cross-entropy is selected as the model loss function. The backpropagation algorithm combined with gradient descent is used to iteratively optimize the weights and bias parameters of the neural network, ensuring that the model can accurately learn the system's operating rules. The layer-by-layer mapping relationship of this neural network satisfies:
[0030]
[0031]
[0032]
[0033]
[0034] in, These are the weighting coefficients for each layer. For each layer of bias terms, This represents the total number of hidden layers. The tanh function is used as the activation function for the hidden layer. The softmax function is used as the activation function for the output layer. For model output, For the system's stable prediction probability, This neural network model provides the predicted probability of system instability, enabling accurate prediction of the system's operating state.
[0035] In this embodiment of the application, based on the neural network model trained above, a broadband oscillation stability judgment criterion for wind power transmitted via flexible direct current is established, namely:
[0036] Among them, the stability discriminant function This discriminant function can accurately classify the stable state of the system, providing a core basis for subsequent risk assessment.
[0037] In this embodiment of the application, by constructing a standardized dataset, building and training a multi-layer feedforward neural network model, and establishing a broadband oscillation stability judgment criterion, a precise stability assessment of the operating status of the wind power transmission system via flexible direct transmission is achieved, laying the foundation for the subsequent generation of diverse wind speed scenario sequences that conform to historical distribution characteristics.
[0038] S2: Acquire historical wind speed data from multiple wind farms and train a generative adversarial network. Use the trained generator to generate diverse wind speed scene sequences from multiple wind farms that conform to historical distribution characteristics.
[0039] In order to obtain diverse wind speed scenarios from multiple wind farms that closely reflect actual working conditions, this application collects historical wind speed data and performs normalization processing, builds a generative adversarial network and trains it through alternating adversarial training, thereby achieving the batch generation of massive amounts of real wind speed scenarios.
[0040] In this embodiment, the application first collects historical wind speed measurement data from multiple wind farms under the jurisdiction of the flexible direct transmission system. Each wind speed sample is represented as a multi-dimensional time series or a multi-point wind speed matrix. The multi-dimensional time series mentioned in this application specifically refers to the wind speed sequence data synchronously collected by each wind farm within the same time period. In this embodiment, the application performs normalization preprocessing on the collected raw wind speed data to eliminate the adverse effects of dimensional differences between different wind speed data, thereby constructing a standard sample set that can be directly used for training generative adversarial networks, providing a standardized and reliable data foundation for subsequent network model training.
[0041] In this embodiment, a Generative Adversarial Network (GAN) structure is further constructed. The GAN mainly consists of two core functional parts: a generator G and a discriminator D. In this embodiment, the generator G uses a random noise vector as its input signal and outputs synthetic wind speed data with a dimension completely consistent with the dimension of the real wind speed samples. It can simultaneously generate complete wind speed sequences for multiple wind farms at corresponding times, meeting the batch generation requirements of multiple wind farm scenarios. In this embodiment, the discriminator D uses real historical wind speed samples or synthetic wind speed samples generated by the generator as input. The discriminator D can output the probability value that the current input sample belongs to real historical measured data, thereby realizing the identification and judgment of the authenticity of the sample.
[0042] In this embodiment, a training strategy of alternating optimization of the generator and discriminator is adopted. Relying on the continuous game-like adversarial process between the generator and discriminator, the generator gradually learns and accurately fits the inherent probability distribution characteristics and temporal statistical properties of historical wind speed data from multiple wind farms. In this embodiment, after the overall training of the generative adversarial network reaches convergence, the network parameters of the generator G are fixed. Subsequently, the trained and parameter-fixed generator G can be used to generate batches of wind speed scene sequences from multiple wind farms that conform to the distribution patterns of historical data, have comprehensive coverage of operating conditions, and are sufficiently diverse. This provides sufficient and realistic wind speed scene data sources for subsequent system operation state sampling modeling.
[0043] In this embodiment, by collecting historical wind speed data from multiple wind farms simultaneously and performing normalization preprocessing, a generative adversarial network containing a generator and a discriminator is constructed and trained using an alternating optimization strategy. This generates a diverse sequence of wind speed scenarios that conforms to historical distribution characteristics and comprehensively covers operating conditions, laying the foundation for subsequent conversion into a corresponding system operating status sampling data set.
[0044] S3, based on the relationship between wind speed and active power conversion of wind turbine generators, converts the generated diverse wind speed scenario sequences of multiple wind farms into corresponding system operation status sampling data sets.
[0045] In order to transform the generated multi-wind farm wind speed scenarios into samples of actual system operation conditions, this application generates wind speed sequences by sampling noise vectors, calculates active power based on the wind speed-power conversion relationship, and integrates and constructs a system operation status sampling dataset to provide sample support for subsequent broadband oscillation risk assessment.
[0046] In this embodiment, the total number of wind speed scenes to be generated is first preset to N. In this embodiment, N sets of independent noise vectors are randomly sampled from a prior distribution such as the standard normal distribution, and these noise vectors are denoted as... , where j ranges from 1 to N.
[0047] In this embodiment of the application, the noise vector obtained from each group of samples is... The inputs are sequentially fed into the trained generator G, which then performs calculations to generate the corresponding multi-wind farm wind speed scenarios. The calculation expression is as follows:
[0048] in, Representing the The group generated multiple wind farm wind speed scenarios. Then it represents the first The wind speed values corresponding to each wind farm in this set of scenarios.
[0049] In this embodiment, under the constraint of the maximum wind energy tracking control strategy for wind turbine generators, and combined with the inherent operating characteristics of the wind turbine generators themselves, a quantitative conversion relationship between wind speed and active power is established. This wind speed-active power conversion relationship is defined in segments according to different operating ranges of the wind turbine:
[0050] in, The rated wind speed of the wind turbine generator set. Cut-in wind speed for wind turbine generators Cut off the wind speed for the wind turbine generator. The rated active power of the wind turbine generator set. For the first In the scenario, the first The wind speed value corresponding to each wind farm. This represents the active power output of the wind turbine generator under the corresponding operating conditions.
[0051] In this embodiment, the active power of wind turbines corresponding to all wind farms under the same wind speed scenario is vector-integrated and combined to construct a set of sampling data on the operating status of the wind power transmission system via flexible direct current transmission. The expression for this sampled data set is:
[0052] in, , Each of the data sets Each corresponds to a complete and independent actual operating condition of the wind power transmission system via flexible direct transmission, which can provide sufficient and complete sample data support for subsequent system broadband oscillation stability judgment and broadband oscillation risk quantitative assessment.
[0053] In this embodiment, multiple wind farm wind speed scenarios are generated by sampling noise vectors, the unit output is calculated based on the active power conversion formula of wind turbine wind speed, and all operating conditions are integrated to construct a system operation status sampling dataset, laying the foundation for subsequent broadband oscillation risk assessment.
[0054] S4. Input each operating state sample from the system operating state sampling data set into the broadband oscillation stability assessment model for stability determination, and count the proportion of samples that do not meet the broadband oscillation stability constraints to the total number of samples to obtain the broadband oscillation risk coefficient.
[0055] In order to quantitatively assess the broadband oscillation risk of wind power transmitted via flexible direct transmission, this application traverses the operating condition samples to determine stability, calculates the risk coefficient by statistically analyzing the proportion of unstable samples, and realizes a quantitative evaluation of the broadband oscillation risk of the system.
[0056] In this embodiment of the application, the application sequentially traverses the system operation status sampling data set. Each working condition sample included therein In this embodiment, each operating condition sample is input into the trained broadband oscillation stability evaluation model one by one. This application relies on a pre-set stability discrimination function. The corresponding value judgment rules are used to check and judge each working condition sample one by one to determine whether each group of working condition samples meets the broadband oscillation stability constraint conditions corresponding to the wind power through flexible direct transmission system.
[0057] In this embodiment of the application, the samples within the sampling dataset are classified and statistically analyzed to determine the number of broadband oscillation stable samples. and the number of broadband oscillation instability samples This application defines the ratio of the number of broadband oscillation instability samples to the total number of all sampled scenarios as the broadband oscillation risk coefficient of the wind power-to-flexible direct transmission system. The formula for calculating this risk coefficient is as follows:
[0058] in, The broadband oscillation risk coefficient of the wind power transmission system via flexible direct current. The number of unstable operating condition samples obtained is defined by N, where N is the total number of sampling scenarios. A broadband oscillation risk coefficient is set. The higher the value, the higher the probability that the actual operating conditions of the system will fall into the broadband oscillation instability range, and the higher the overall broadband oscillation risk faced by the system. This quantitative method can accurately and reliably complete the quantitative assessment of the broadband oscillation risk of wind power transmitted via flexible direct transmission.
[0059] In this embodiment, stability is determined by traversing the system operating condition samples one by one, the number of stable and unstable samples is counted, and the proportion of unstable samples is calculated to obtain the risk coefficient, thereby completing the quantitative assessment of the broadband oscillation risk of the wind power transmission system via flexible direct transmission.
[0060] Example 2 This invention provides a data-driven broadband oscillation risk assessment device 10 for wind power transmission via flexible direct power transmission systems, such as... Figure 3 As shown, the device includes: Module 100 is used to build a broadband oscillation stability assessment model. It takes the combined output active power of each wind turbine generator in each wind farm as the input feature and outputs the judgment result of whether the system is oscillating and stable through the trained neural network model. Training module 200 is used to acquire historical wind speed data of multiple wind farms and train a generative adversarial network. The trained generator is used to generate diverse wind speed scene sequences of multiple wind farms that conform to historical distribution characteristics. The conversion module 300 is used to convert the generated diverse wind speed scene sequences of multiple wind farms into corresponding system operation status sampling data sets based on the conversion relationship between wind speed and active power of wind turbine generator sets. The judgment module 400 is used to input each operating state sample in the system operating state sampling data set into the broadband oscillation stability assessment model for stability judgment, and to count the proportion of the number of samples that do not meet the broadband oscillation stability constraints to the total number of samples, thereby obtaining the broadband oscillation risk coefficient.
[0061] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.
[0062] Example 3 To implement the methods of the above embodiments, the present invention also provides an electronic device, which includes a memory and a processor; wherein the processor reads executable program code stored in the memory to run a program corresponding to the executable program code, so as to implement the various steps of the methods described above.
[0063] Example 4 To implement the above embodiments, this application also proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the method described in the foregoing embodiments.
[0064] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
[0065] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.
[0066] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.
Claims
1. A data-driven method for assessing the broadband oscillation risk of wind power transmitted via flexible direct current transmission systems, characterized in that, include: S1. Construct a broadband oscillation stability assessment model, taking the combined output active power of each wind turbine generator set in each wind farm as the input feature, and outputting the judgment result of whether the system is oscillating and stable through the trained neural network model. S2: Obtain historical wind speed data from multiple wind farms and train a generative adversarial network. Use the trained generator to generate diverse wind speed scene sequences from multiple wind farms that conform to historical distribution characteristics. S3, based on the relationship between wind speed and active power conversion of wind turbine generators, converts the generated diverse wind speed scene sequences of multiple wind farms into corresponding system operation status sampling data sets. S4. Input each operating state sample from the system operating state sampling data set into the broadband oscillation stability assessment model for stability determination, and count the proportion of samples that do not meet the broadband oscillation stability constraints to the total number of samples to obtain the broadband oscillation risk coefficient.
2. The method according to claim 1, characterized in that, The constructed broadband oscillation stability assessment model uses the combined output active power of individual wind turbine generators in each wind farm as input features. The trained neural network model outputs a determination of whether the system is oscillating stably, including: The active power output of each wind turbine generator in each wind farm in the system is used as input to form a large number of different operating conditions. For each operating condition, its stability is judged according to the broadband oscillation stability analysis method. A dataset containing input vectors and output labels is constructed, where the input vector is a sequence of active power output of each wind turbine generator in each wind farm, and the output label is whether the system is oscillating and stable. A multi-layer feedforward neural network model is built. The number of hidden layers and the number of neurons in each layer are set according to experience. Cross-entropy is used as the loss function. The backpropagation algorithm combined with the gradient descent strategy is used to iteratively optimize the model parameters until the model converges to obtain the trained neural network model. The trained neural network model is used to calculate the input features and outputs the probability that the system is oscillating and stable or unstable in the current state. The class with the higher probability is taken as the final evaluation result.
3. The method according to claim 1, characterized in that, The process involves acquiring historical wind speed data from multiple wind farms and training a generative adversarial network (GAN). The trained generator is then used to generate diverse multi-wind farm wind speed scene sequences that conform to historical distribution characteristics, including: Collect historical wind speed data of multiple wind farms in the system through which wind power is transmitted via flexible direct transmission, and represent each sample as a multi-dimensional time series or a multi-point wind speed matrix, where the multi-dimensional time series is the wind speed sequence of each wind farm in the same time period. The original data is normalized to construct a training sample set for training the generative adversarial network. A generative adversarial network consisting of a generator G and a discriminator D is constructed, where the input of the generator G is a random noise vector and the output is a synthetic wind speed sample, and the input of the discriminator D is a real sample or a generated sample and the output is the probability that the sample is real data. An alternating optimization strategy is used to train the generator and discriminator, enabling the generator to gradually learn the statistical characteristics of wind speed data from multiple wind farms. After training, the generator G is fixed to generate diverse wind speed scenarios that conform to historical distribution characteristics.
4. The method according to claim 1, characterized in that, The method, based on the conversion relationship between wind speed and active power of wind turbine generators, converts the generated diverse multi-wind farm wind speed scenario sequences into corresponding system operation status sampling data sets, including: Set the required number of scenes as follows Random sampling from the prior distribution noise vectors ,in Values ; The noise vector Input the trained generator To obtain the corresponding wind speed scene : Under the maximum wind energy tracking control strategy, based on the rated wind speed of the wind turbine generator... Cut-in wind speed Cut-off wind speed and rated active power By utilizing the wind speed-power conversion relationship, the wind speed of each wind farm in each wind speed scenario is converted into the active power output of a single wind turbine generator. ; The active power output of each individual wind turbine in all wind farms is combined to form a sample data set of the operating status of the wind power transmission system via flexible direct current. .
5. The method according to claim 1, characterized in that, The process involves inputting each operating state sample from the system operating state sampling data set into a broadband oscillation stability assessment model for stability determination. The proportion of samples that do not meet the broadband oscillation stability constraints is then calculated to obtain the broadband oscillation risk coefficient, including: Traverse the system operation status sampling data set Each running state sample The sample is then input into the broadband oscillation stability evaluation model, and the broadband oscillation stability judgment criterion is used to determine whether the sample meets the broadband oscillation stability constraint. Statistical sets Number of samples with mid-bandwidth oscillation instability and the number of samples with stable broadband oscillations ; Calculate the percentage of wideband oscillatory unstable samples in the total number of samples. The ratio is used to obtain the broadband oscillation risk coefficient. : Among them, the larger the broadband oscillation risk coefficient, the higher the broadband oscillation risk faced by the system.
6. A data-driven broadband oscillation risk assessment device for wind power transmitted via flexible direct current transmission system, characterized in that, include: The module is used to build a broadband oscillation stability assessment model. It takes the combined output active power of each wind turbine generator in each wind farm as input features and outputs the judgment result of whether the system is oscillating and stable through the trained neural network model. The training module is used to acquire historical wind speed data from multiple wind farms and train a generative adversarial network. The trained generator is then used to generate diverse wind speed scene sequences from multiple wind farms that conform to historical distribution characteristics. The conversion module is used to convert the generated diverse wind speed scene sequences of multiple wind farms into corresponding system operation status sampling data sets based on the conversion relationship between wind speed and active power of wind turbine generators. The determination module is used to input each operating state sample from the system operating state sampling data set into the broadband oscillation stability assessment model for stability determination, and to count the proportion of samples that do not meet the broadband oscillation stability constraints to the total number of samples, thereby obtaining the broadband oscillation risk coefficient.
7. An electronic device, characterized in that, Including processor and memory; The processor reads executable program code stored in the memory to run a program corresponding to the executable program code, so as to implement the method as described in any one of claims 1-5.
8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-5.