Wind turbine frequency control method and system based on twin network

Abstract

The present application provides a wind turbine frequency control method and system based on a twin network, which relates to the field of frequency control technology, including: based on a predetermined output power factor dimension, multi-dimensional feature collection of a target wind turbine to obtain target output power factor features; obtaining a target output power prediction value, and determining whether it meets a predetermined output power threshold; if it meets, issuing a monitoring instruction to dynamically obtain target operating state information of the target wind turbine; comparing history with the current state through a twin network, obtaining the optimal frequency regulation strategy, and performing dynamic frequency regulation control. This application can solve the problems of the prior art due to the inaccuracy of output power prediction, the lack of real-time and dynamic frequency regulation, and the inability to make timely adjustments according to the actual operating state of the unit, resulting in poor frequency control effects. Dynamic frequency regulation control can be performed on the target wind turbine according to the optimal historical frequency regulation strategy to maintain the stability of the grid frequency and improve the operating efficiency of the wind turbine.

Description

Technical Field

[0001] The present application relates to the technical field of frequency modulation control, and in particular to a frequency modulation control method and system for a wind turbine generator set based on a twin network. Background Art

[0002] With the continuous advancement of technology, wind turbines have gradually developed from megawatts to larger capacities, and wind farms have also formed large-scale construction and operation. As the penetration rate of wind power in the power system continues to increase, the impact of wind power grid connection on the frequency stability of the power grid is becoming increasingly significant. The intermittent and volatile nature of wind power output reduces the frequency stability of the power grid.

[0003] At present, the existing wind turbine frequency control systems mainly rely on fixed rules or experience to formulate frequency regulation strategies, and the output power prediction often uses simple statistical models or empirical formulas.

[0004] In summary, due to the inaccuracy of output power prediction, the existing technology lacks real-time and dynamic frequency regulation and cannot make timely adjustments according to the actual operating status of the unit, resulting in poor frequency control effect, which further affects the operating efficiency of the wind turbine and the stability of the power grid. Summary of the invention

[0005] The purpose of this application is to provide a wind turbine frequency control method and system based on a twin network, so as to solve the problems in the prior art such as inaccurate output power prediction, lack of real-time and dynamic frequency regulation, and inability to make timely adjustments according to the actual operating status of the unit, resulting in poor frequency control effect.

[0006] In view of the above problems, the present application provides a wind turbine frequency control method and system based on a twin network.

[0007] In a first aspect, the present application provides a wind turbine frequency control method based on a twin network, wherein the wind turbine frequency control method based on a twin network is implemented by a wind turbine frequency control system based on a twin network, wherein the wind turbine frequency control method based on a twin network includes: collecting multi-dimensional features of a target wind turbine based on a predetermined output power factor dimension to obtain a target output power factor feature; retrieving an output power prediction model to analyze the target output power factor feature to obtain a target output power prediction value; judging whether the target output power prediction value meets a predetermined output power threshold to obtain a judgment result; according to the judgment result, if the target output power prediction value meets the predetermined output power threshold, issuing a monitoring instruction, and dynamically obtaining target operating status information of the target wind turbine based on the monitoring instruction; obtaining historical operating status information, and inputting the historical operating status information and the target operating status information into the twin network to obtain state similarity; when the state similarity reaches a predetermined similarity threshold, obtaining an optimal historical frequency regulation strategy corresponding to the historical operating status information; and dynamically controlling the frequency of the target wind turbine according to the optimal historical frequency regulation strategy.

[0008] In the second aspect, the present application also provides a wind turbine frequency control system based on a twin network, which is used to execute the wind turbine frequency control method based on a twin network as described in the first aspect, wherein the wind turbine frequency control system based on the twin network includes: a feature collection module, which is used to collect multi-dimensional features of the target wind turbine based on a predetermined output power factor dimension to obtain a target output power factor feature; a prediction value acquisition module, which is used to call the output power prediction model to analyze the target output power factor feature to obtain a target output power prediction value; a threshold judgment module, which is used to judge whether the target output power prediction value meets the predetermined output power threshold to obtain a judgment result; an operation status information acquisition module A block is used to issue a monitoring instruction according to the judgment result, if the target output power prediction value meets the predetermined output power threshold, and dynamically obtain the target operating state information of the target wind turbine group based on the monitoring instruction; a state similarity acquisition module is used to obtain historical operating state information, and input the historical operating state information and the target operating state information into the twin network to obtain state similarity; an optimal historical frequency modulation strategy acquisition module is used to obtain the optimal historical frequency modulation strategy corresponding to the historical operating state information when the state similarity reaches a predetermined similarity threshold; a frequency modulation control module is used to dynamically control the frequency modulation of the target wind turbine group according to the optimal historical frequency modulation strategy.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] The target output power factor feature is obtained by collecting multi-dimensional features of the target wind turbine based on the predetermined output power factor dimension; the output power prediction model is retrieved to analyze the target output power factor feature to obtain the target output power prediction value; it is judged whether the target output power prediction value meets the predetermined output power threshold value to obtain a judgment result; according to the judgment result, if the target output power prediction value meets the predetermined output power threshold value, a monitoring instruction is issued, and the target operating state information of the target wind turbine is dynamically obtained based on the monitoring instruction; historical operating state information is obtained, and the historical operating state information and the target operating state information are input into the twin network to obtain state similarity; when the state similarity reaches the predetermined similarity threshold value, the optimal historical frequency modulation strategy corresponding to the historical operating state information is obtained; the target wind turbine is dynamically controlled according to the optimal historical frequency modulation strategy, which effectively solves the problem that the inaccuracy of output power prediction, the lack of real-time and dynamic frequency modulation, and the inability to make timely adjustments according to the actual operating state of the unit lead to poor frequency modulation control effect. The target wind turbine can be dynamically controlled according to the optimal historical frequency modulation strategy to maintain the stability of the grid frequency and improve the operating efficiency of the wind turbine.

[0011] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0013] Figure 1 This is a flow chart of a wind turbine frequency control method based on a twin network in this application;

[0014] Figure 2 This is a structural schematic diagram of the wind turbine frequency control system based on the twin network in this application.

[0015] Description of reference numerals:

[0016] Feature collection module 11, prediction value acquisition module 12, threshold judgment module 13, operation status information acquisition module 14, status similarity acquisition module 15, optimal historical frequency modulation strategy acquisition module 16, frequency modulation control module 17. DETAILED DESCRIPTION

[0017] The present application provides a wind turbine frequency control method and system based on a twin network, thereby solving the problems of the prior art due to inaccuracy in output power prediction, lack of real-time and dynamic frequency regulation, and inability to make timely adjustments according to the actual operating status of the unit, resulting in poor frequency control effect. The application can dynamically control the frequency of the target wind turbine according to the optimal historical frequency regulation strategy to maintain the stability of the grid frequency, thereby improving the operating efficiency of the wind turbine.

[0018] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0019] For example, please refer to the attached Figure 1 The present application provides a wind turbine frequency control method based on a twin network, wherein the wind turbine frequency control method based on a twin network is applied to a wind turbine frequency control system based on a twin network, and the wind turbine frequency control method based on a twin network specifically includes the following steps:

[0020] S1: Collect multi-dimensional features of the target wind turbine generator system based on a predetermined output power factor dimension to obtain target output power factor features.

[0021] Specifically, the predetermined output power factor dimension is a set of key parameters or indicators defined in advance, including the unit factor dimension and the environmental condition dimension. Multidimensional feature collection refers to the collection of data and information related to the predetermined output power factor dimension from multiple angles and levels. It can be collected from the data acquisition and monitoring control system, meteorological station, unit sensor, etc. of the wind farm. The collected raw data is preprocessed, including data cleaning, noise removal, outlier removal, data conversion, such as unit unification, type conversion, and data compression, such as downsampling, feature selection, etc. Based on the preprocessed data, statistical analysis, signal processing, machine learning and other methods are used to extract key features related to the target output power. These features include statistical features such as the mean, variance, and trend of wind speed, as well as features such as frequency components and energy distribution obtained through signal processing; through the data obtained through multidimensional feature collection, a set of key parameters or indicators that can characterize the output power characteristics of the target wind turbine unit, namely, the target output power factor characteristics, are extracted.

[0022] S2: Retrieving an output power prediction model to analyze the target output power factor characteristics to obtain a target output power prediction value.

[0023] Specifically, the output power prediction model is a model built based on historical data and machine learning algorithms, including linear regression, decision tree, random forest, neural network, etc. According to historical data and prediction requirements, a suitable machine learning algorithm is selected to build the output power prediction model; the model is trained using historical data to learn the relationship between output power and input features; the output power prediction model can predict the output power of the wind turbine according to the given input features, namely the target output power factor features. The target output power factor features are preprocessed features collected from various data sources of the wind turbine, including wind speed, wind direction, temperature, humidity, air pressure, unit state parameters, etc. The target output power factor features are used as input and passed to the output power prediction model, which analyzes and processes these features, and the model outputs a predicted value, namely the output power of the target wind turbine in the future. The predicted value represents the expected output power of the wind turbine in the future, such as a few minutes or hours.

[0024] S3: Determine whether the target output power prediction value meets a predetermined output power threshold value, and obtain a determination result.

[0025] Specifically, the predetermined output power threshold is a power value preset by a person skilled in the art, and is used as a standard for judging whether the target wind turbine output power is within an acceptable range. It can be set based on the operational requirements of the wind farm, the stability requirements of the power grid, or the technical specifications of the wind turbine. The target output power prediction value is the prediction result obtained by analyzing the target output power factor characteristics through the output power prediction model, which represents the expected output power of the wind turbine in the future. The target output power prediction value is compared with the predetermined output power threshold. If the prediction value is within the threshold range, including being equal to the threshold, it is judged to be in compliance; if the prediction value exceeds the threshold range, it is judged to be non-compliant. Based on the comparison result, a clear judgment is made as to whether the target output power prediction value meets the predetermined output power threshold.

[0026] S4: According to the judgment result, if the target output power prediction value meets the predetermined output power threshold, a monitoring instruction is issued, and the target operating state information of the target wind turbine is dynamically acquired based on the monitoring instruction.

[0027] Specifically, when the target output power prediction value is determined to be within the predetermined output power threshold range, it means that the expected output power of the wind turbine meets the requirements. Based on the judgment result, an instruction is issued to require it to start or continue monitoring the target wind turbine. The monitoring instruction includes the monitoring frequency, required data points, data transmission method, etc. Based on the issued monitoring instructions, real-time operating status information is dynamically obtained from various sensors and data sources of the wind turbine, including wind speed, wind direction, unit speed, generator temperature, gearbox oil temperature, vibration level, etc.

[0028] S5: Obtain historical operating status information, and input the historical operating status information and the target operating status information into the twin network to obtain status similarity.

[0029] Specifically, historical operating status information refers to the operating status information of a target wind turbine or similar wind turbine in a certain period of time in the past, including time series data of multiple parameters such as wind speed, wind direction, unit speed, temperature, vibration level, etc. The target operating status information is the target information at the current moment or in the recent period of time, and is used for real-time comparison and analysis with the actual operating status. The twin network is a deep learning model. In this embodiment, the twin network can be used to compare the similarity between historical operating status information and target operating status information. A quantitative indicator obtained after twin network analysis represents the degree of similarity between historical operating status information and target operating status information. The similarity can be used to determine whether the current state of the wind turbine is similar to a good or bad state in history.

[0030] S6: When the state similarity reaches a predetermined similarity threshold, the optimal historical frequency modulation strategy corresponding to the historical operating state information is obtained.

[0031] Specifically, the predetermined similarity threshold is a preset similarity level used to determine whether the current state is sufficiently similar to a state in history. The threshold is set by technical personnel in this field based on experience and historical data. The similarity index is obtained by analyzing the historical operating state information and the target operating state information through the twin network. The similarity is used to quantify the degree of proximity between the two states. The optimal historical frequency regulation strategy is the best frequency regulation strategy corresponding to the historical operating state information. It has been proven in the past to optimize the performance and output power of wind turbines, including adjusting the settings of parameters such as wind speed, wind direction, unit speed, and generator temperature. When the state similarity reaches or exceeds the predetermined threshold, the optimal frequency regulation strategy corresponding to the state most similar to the current state is retrieved from the historical data.

[0032] S7: Performing dynamic frequency regulation control on the target wind turbine generator set according to the optimal historical frequency regulation strategy.

[0033] Specifically, various sensors installed on the wind turbine are used to collect data on key parameters such as wind speed, wind direction, unit speed, generator temperature, and output power in real time, and the collected data is preprocessed, including filtering, denoising, and normalization. According to the real-time monitored state of the wind turbine, it is matched with the optimal historical frequency modulation strategy to find the most suitable frequency modulation strategy under the current state. According to the matched frequency modulation strategy, the control parameters of the wind turbine are adjusted, including changing the set value of the rotor speed, adjusting the pitch angle to change the captured wind energy, or adjusting the output power of the generator, etc. The adjusted control parameters are sent to the control system of the wind turbine to perform dynamic frequency modulation control. The effect of the dynamic frequency modulation control after implementation is monitored in real time, including the stability of the grid frequency, the output power of the wind turbine, the rotor speed and other parameters. According to the monitoring results, the dynamic frequency modulation control strategy is feedback adjusted. If the control effect is not ideal, it is necessary to re-match the optimal historical frequency modulation strategy or adjust the control parameters; if the control effect is good, the current strategy continues to be executed.

[0034] Furthermore, the predetermined output power factor dimension includes a unit factor dimension and an environmental condition dimension, wherein the unit factor dimension includes unit mechanical loss, unit electrical loss, and unit structural design, the unit structural design includes at least blade shape, blade size, blade material, blade angle, and tower height, and the environmental condition dimension includes ambient wind speed, air density, and ambient temperature and humidity.

[0035] Specifically, the dimension of unit factors is the design and performance of the wind turbine itself, which directly affects the output power of the unit. It includes unit mechanical loss, which refers to the power loss caused by friction and wear of mechanical parts during the operation of the wind turbine. Unit electrical loss, electrical loss mainly occurs in the electrical system of the wind turbine, including the resistance loss of components such as cables, transformers, and generators. Unit structural design, the structural design of the unit has a significant impact on the output power. Among them, blade shape, blade size, blade material, blade angle and tower height are important parameters for unit structural design. Blade shape, the shape of the blade directly affects the efficiency of wind energy capture. Reasonable blade shape can maximize the use of wind energy and improve the output power of the unit. Blade size, larger blades can capture more wind energy, but it will also increase the manufacturing cost and operation difficulty of the unit. Blade material, choosing the right material can reduce the weight of the blade, improve the operating efficiency of the unit, and extend the service life of the unit. Blade angle, the angle of the blade determines the interaction between wind energy and the unit. By adjusting the blade angle, the wind capture capacity of the unit can be optimized and the output power can be increased. Tower height: The height of the tower determines the height range of the wind turbine. A higher tower enables the turbine to capture wind energy at a higher altitude, thereby increasing the output power. The environmental condition dimension is the impact of the external environment in which the wind turbine is located on the output power. This includes the ambient wind speed, which is the most direct factor affecting the output power of the wind turbine. The higher the wind speed, the greater the output power of the turbine; but too high a wind speed may also cause the turbine to overload or damage. Air density: Air density also affects the output power of the wind turbine. At the same wind speed, the greater the air density, the more wind energy the turbine captures and the higher the output power. Ambient temperature and humidity: The temperature and humidity of the environment also affect the performance of the wind turbine. For example, a high temperature environment may cause overheating and performance degradation of the turbine components; while excessive humidity may affect the electrical performance and life of the turbine.

[0036] Furthermore, the present application also includes:

[0037] Obtain historical output power monitoring records of similar units of the target wind turbine unit, and extract a first historical monitoring record from the historical output power monitoring records; perform machine learning on a first data group composed of a first historical unit factor characteristic, a first historical environmental factor characteristic, and a first historical output power value in the first historical monitoring record, and train to obtain the output power prediction model.

[0038] Specifically, determine the same type of wind turbines as the target wind turbines. These wind turbines should have similar models, capacities, blade designs, control systems and other characteristics. Retrieve the historical output power monitoring records of these similar wind turbines from the data storage system of the wind farm, including unit factor characteristics, environmental factor characteristics and corresponding output power values. Screen the retrieved data to ensure the integrity and accuracy of the data, and delete records that are missing key information or obviously wrong. Select the first record from the screened historical output power monitoring records as the first historical monitoring record, including the first historical unit factor characteristics, the first historical environmental factor characteristics and the first historical output power value. Perform data preprocessing on the first historical monitoring record, including data cleaning, normalization, etc. Use the first historical unit factor characteristics, the first historical environmental factor characteristics and the first historical output power value to form a first data group. According to the nature of the problem and the characteristics of the data, select a suitable machine learning algorithm for model training, including linear regression, support vector machine, neural network, etc. Use the first data group to train the selected machine learning algorithm to obtain an output power prediction model. During the training process, the prediction accuracy of the model is improved by adjusting the algorithm parameters and optimization methods. Use a portion of historical data as a test set to evaluate the trained output power prediction model. Evaluation indicators include prediction error, accuracy, etc. Optimize the model based on the evaluation results to improve its prediction performance. Optimization methods include increasing the amount of training data, adjusting model parameters, and using more complex algorithms.

[0039] Furthermore, the present application also includes:

[0040] According to the judgment result, if the target output power prediction value does not meet the predetermined output power threshold, a predetermined frequency regulation auxiliary mechanism is activated to perform frequency regulation auxiliary processing on the target wind turbine group.

[0041] Specifically, if the target output power prediction value does not meet the predetermined output power threshold, that is, the prediction value is too high or too low, it means that there may be a problem with the output power of the wind turbine. Activate the predetermined frequency regulation auxiliary mechanism. The frequency regulation auxiliary mechanism is a preset control strategy and adjustment measure that can adjust the output power of the wind turbine to meet the predetermined requirements. After activating the frequency regulation auxiliary mechanism, the target wind turbine is adjusted accordingly according to the control strategy in the mechanism. This includes adjusting the speed, pitch angle, generator output and other parameters of the unit to change the wind capture capacity and output power of the unit. For example, under low wind speed conditions, the output power of the wind turbine is low and the frequency regulation capacity is limited. This can be achieved by optimizing blade design, improving converter control, optimizing converter control strategy, using energy storage system, frequency droop control adjustment, virtual inertia control, reducing unit losses, and limiting non-essential loads.

[0042] Furthermore, the present application also includes:

[0043] Read the predetermined state characteristics; the first sub-network in the twin network performs feature extraction on the historical operation state information based on the predetermined state characteristics to obtain historical operation characteristic parameters; the second sub-network in the twin network performs feature extraction on the target operation state information based on the predetermined state characteristics to obtain target operation characteristic parameters; and the parameter similarity between the historical operation characteristic parameters and the target operation characteristic parameters is used as the state similarity.

[0044] Specifically, the predetermined state feature is a predefined feature parameter used to describe the ideal or standard operating state of the wind turbine, including the mechanical loss, electrical loss, structural design parameters (such as blade shape, size, material, etc.) and environmental conditions (such as wind speed, air density, etc.) of the unit. The twin network consists of two sub-networks, which are used to process historical operating state information and target operating state information respectively. The first sub-network receives the historical operating state information as input, and extracts features based on the predetermined state feature to obtain historical operating feature parameters. The second sub-network receives the target operating state information as input, and also extracts features based on the predetermined state feature to obtain target operating feature parameters. State similarity is an indicator to measure the similarity between the current operating state and the historical operating state of the wind turbine. By comparing the historical operating feature parameters with the target operating feature parameters, the parameter similarity between them can be calculated, and the parameter similarity can be calculated by Euclidean distance, cosine similarity, Hamming distance, etc. The calculated parameter similarity is the state similarity, which reflects the closeness between the current operating state and the historical operating state of the wind turbine.

[0045] Furthermore, the present application also includes:

[0046] Obtain an arbitrary frequency modulation index; extract a first historical frequency modulation record from a historical frequency modulation control record, wherein the first historical frequency modulation record includes a first historical feature parameter corresponding to a first state feature and a first historical frequency modulation strategy; match an arbitrary frequency modulation index parameter corresponding to the arbitrary frequency modulation index in the first historical frequency modulation strategy, and use the arbitrary frequency modulation index parameter as a dependent variable; use the first historical feature parameter as an independent variable to perform a correlation analysis with the dependent variable to obtain a first correlation coefficient; if the first correlation coefficient meets a predetermined correlation constraint, add the first state feature to the predetermined state feature.

[0047] Specifically, any frequency regulation index can be any parameter related to the frequency regulation performance of the wind turbine, such as frequency regulation response time, frequency regulation accuracy, etc. The first record is extracted from the historical frequency regulation control record as the first historical frequency regulation record, which contains the characteristic parameters of the unit in a certain state (i.e., the first historical characteristic parameters corresponding to the first state characteristics) and the frequency regulation strategy adopted at that time (i.e., the first historical frequency regulation strategy). In the first historical frequency regulation strategy, the parameters corresponding to the arbitrary frequency regulation index are found, which reflects the frequency regulation performance of the unit under the frequency regulation strategy. The first historical characteristic parameter is used as an independent variable and a correlation analysis is performed with the dependent variable (i.e., the arbitrary frequency regulation index parameter). A first correlation coefficient is calculated using a suitable correlation analysis method, such as the Pearson correlation coefficient, the Spearman rank correlation coefficient, etc., to reflect the linear relationship strength between the first state characteristics and the frequency regulation performance. If the first correlation coefficient meets the predetermined correlation constraint, that is, the coefficient value is large enough, indicating a strong correlation, it means that the first state characteristics have an important influence on the frequency regulation performance, and then the first state characteristics are added to the predetermined state characteristics set.

[0048] Furthermore, the present application also includes:

[0049] The historical frequency modulation control records are filtered using the historical operating status information as a filtering constraint to obtain a candidate historical frequency modulation record list; multiple historical frequency modulation strategies in multiple historical frequency modulation records in the candidate historical frequency modulation record list are used as an optimization space; a frequency modulation effect evaluation function is introduced as an optimization evaluation index; frequency optimization is performed in the optimization space based on the optimization evaluation index to obtain the optimal historical frequency modulation strategy.

[0050] Specifically, the historical operating status information is used as a screening constraint to screen the historical frequency regulation control records, and the historical frequency regulation records that are similar or related to the current operating status are found as a candidate historical frequency regulation record list. The record list contains multiple historical frequency regulation records, and each record corresponds to a historical frequency regulation strategy. These historical frequency regulation strategies are used as the optimization space, that is, the optimal frequency regulation strategy is searched in this space. A frequency regulation effect evaluation function is introduced and designed according to actual needs. For example, factors such as frequency regulation response time, frequency regulation accuracy, and energy loss during frequency regulation can be considered. In the optimization space, the frequency regulation effect evaluation function is used as an evaluation index to evaluate different frequency regulation strategies. By comparing the evaluation function values ​​under different strategies, the frequency regulation strategy that makes the evaluation function value optimal (that is, the best frequency regulation effect) is found. After frequency optimization, the optimal historical frequency regulation strategy in the candidate historical frequency regulation record list can be obtained. The optimal historical frequency regulation strategy can be used as a recommended frequency regulation strategy under the current operating state to guide the frequency regulation operation of wind turbines.

[0051] Furthermore, the frequency regulation effect evaluation function refers to a weighted evaluation function of power grid stability and frequency regulation energy consumption, and the power grid stability refers to the stability result obtained by weighting the variation of predetermined power grid stability indicators, and the predetermined power grid stability indicators include regulating vibration frequency, power change rate, steady-state error and control robustness.

[0052] Specifically, grid stability refers to the ability of the grid to maintain operation in the event of external disturbances or internal faults, and is an indicator of safe and reliable grid operation. In the frequency modulation effect evaluation function, grid stability is the stability result obtained by weighting the variation of the predetermined grid stability indicators. The predetermined grid stability indicators include but are not limited to regulating vibration frequency, power change rate, steady-state error and control robustness. The vibration frequency is the number of vibration cycles of an object per second. The stability of regulating vibration frequency is to ensure that the grid avoids instability caused by excessive frequency fluctuations during the frequency modulation process. The power change rate indicates how fast the power changes over time. For the grid, fast and stable power regulation is the key to maintaining grid stability. Steady-state error reflects whether the frequency modulation strategy can ultimately stabilize the grid in the desired operating state. Control robustness evaluates the stability and reliability of the frequency modulation strategy under different operating conditions. The various disturbances and faults that the grid may encounter should be considered during the evaluation, and the performance of the frequency modulation strategy under these conditions should be verified. Frequency modulation energy consumption refers to the energy consumed during the frequency modulation process, including mechanical loss and electrical loss of wind turbines due to frequency modulation operations. The frequency regulation effect evaluation function combines the grid stability and frequency regulation energy consumption in a weighted manner to form a comprehensive evaluation index. The weighted coefficient reflects the relative importance of grid stability and frequency regulation energy consumption in the evaluation process.

[0053] In summary, the wind turbine frequency regulation control method based on twin network provided in this application has the following technical effects:

[0054] The target output power factor feature is obtained by collecting multi-dimensional features of the target wind turbine based on the predetermined output power factor dimension; the output power prediction model is retrieved to analyze the target output power factor feature to obtain the target output power prediction value; it is judged whether the target output power prediction value meets the predetermined output power threshold value to obtain a judgment result; according to the judgment result, if the target output power prediction value meets the predetermined output power threshold value, a monitoring instruction is issued, and the target operating state information of the target wind turbine is dynamically obtained based on the monitoring instruction; historical operating state information is obtained, and the historical operating state information and the target operating state information are input into the twin network to obtain state similarity; when the state similarity reaches the predetermined similarity threshold value, the optimal historical frequency modulation strategy corresponding to the historical operating state information is obtained; the target wind turbine is dynamically controlled according to the optimal historical frequency modulation strategy, which effectively solves the problem that the inaccuracy of output power prediction, the lack of real-time and dynamic frequency modulation, and the inability to make timely adjustments according to the actual operating state of the unit lead to poor frequency modulation control effect. The target wind turbine can be dynamically controlled according to the optimal historical frequency modulation strategy to maintain the stability of the grid frequency and improve the operating efficiency of the wind turbine.

[0055] Embodiment 2: Based on the wind turbine frequency control method based on twin network in the above embodiment, the present application also provides a wind turbine frequency control system based on twin network, please refer to the attached Figure 2 , the wind turbine frequency control system based on twin network includes:

[0056] The feature collection module 11 is used to collect multi-dimensional features of the target wind turbine generator system based on a predetermined output power factor dimension to obtain a target output power factor feature.

[0057] The prediction value acquisition module 12 is used to retrieve the output power prediction model to analyze the target output power factor characteristics to obtain the target output power prediction value.

[0058] The threshold judgment module 13 is used to judge whether the target output power prediction value meets the preset output power threshold value and obtain a judgment result.

[0059] The operation status information acquisition module 14 is used to issue a monitoring instruction according to the judgment result if the target output power prediction value meets the predetermined output power threshold, and dynamically acquire the target operation status information of the target wind turbine group based on the monitoring instruction.

[0060] The state similarity acquisition module 15 is used to acquire historical operating state information, and input the historical operating state information and the target operating state information into the twin network to obtain state similarity.

[0061] The optimal historical frequency modulation strategy acquisition module 16 is used to acquire the optimal historical frequency modulation strategy corresponding to the historical operating state information when the state similarity reaches a predetermined similarity threshold.

[0062] The frequency regulation control module 17 is used to perform dynamic frequency regulation control on the target wind turbine generator set according to the optimal historical frequency regulation strategy.

[0063] Furthermore, the predetermined output power factor dimension includes a unit factor dimension and an environmental condition dimension, wherein the unit factor dimension includes unit mechanical loss, unit electrical loss, and unit structural design, the unit structural design includes at least blade shape, blade size, blade material, blade angle, and tower height, and the environmental condition dimension includes ambient wind speed, air density, and ambient temperature and humidity.

[0064] Furthermore, the wind turbine frequency control system based on the twin network also includes a model training module for:

[0065] Obtain historical output power monitoring records of similar units of the target wind turbine unit, and extract a first historical monitoring record from the historical output power monitoring records; perform machine learning on a first data group composed of a first historical unit factor characteristic, a first historical environmental factor characteristic, and a first historical output power value in the first historical monitoring record, and train to obtain the output power prediction model.

[0066] Furthermore, the wind turbine frequency control system based on the twin network also includes a frequency regulation auxiliary module, which is used to:

[0067] According to the judgment result, if the target output power prediction value does not meet the predetermined output power threshold, a predetermined frequency regulation auxiliary mechanism is activated to perform frequency regulation auxiliary processing on the target wind turbine group.

[0068] Furthermore, the wind turbine frequency control system based on the twin network also includes a state similarity acquisition module, which is used to:

[0069] Read the predetermined state characteristics; the first sub-network in the twin network performs feature extraction on the historical operation state information based on the predetermined state characteristics to obtain historical operation characteristic parameters; the second sub-network in the twin network performs feature extraction on the target operation state information based on the predetermined state characteristics to obtain target operation characteristic parameters; and the parameter similarity between the historical operation characteristic parameters and the target operation characteristic parameters is used as the state similarity.

[0070] Furthermore, the wind turbine frequency regulation control system based on the twin network also includes a state feature adding module, which is used to:

[0071] Obtain an arbitrary frequency modulation index; extract a first historical frequency modulation record from a historical frequency modulation control record, wherein the first historical frequency modulation record includes a first historical feature parameter corresponding to a first state feature and a first historical frequency modulation strategy; match an arbitrary frequency modulation index parameter corresponding to the arbitrary frequency modulation index in the first historical frequency modulation strategy, and use the arbitrary frequency modulation index parameter as a dependent variable; use the first historical feature parameter as an independent variable to perform a correlation analysis with the dependent variable to obtain a first correlation coefficient; if the first correlation coefficient meets a predetermined correlation constraint, add the first state feature to the predetermined state feature.

[0072] Furthermore, the wind turbine frequency control system based on the twin network also includes an optimal space acquisition module, which is used to:

[0073] The historical frequency modulation control records are filtered using the historical operating status information as a filtering constraint to obtain a candidate historical frequency modulation record list; multiple historical frequency modulation strategies in multiple historical frequency modulation records in the candidate historical frequency modulation record list are used as an optimization space; a frequency modulation effect evaluation function is introduced as an optimization evaluation index; frequency optimization is performed in the optimization space based on the optimization evaluation index to obtain the optimal historical frequency modulation strategy.

[0074] Furthermore, the frequency regulation effect evaluation function refers to a weighted evaluation function of power grid stability and frequency regulation energy consumption, and the power grid stability refers to the stability result obtained by weighting the variation of predetermined power grid stability indicators, and the predetermined power grid stability indicators include regulating vibration frequency, power change rate, steady-state error and control robustness.

[0075] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. Figure 1 The wind turbine frequency control method based on twin network and the specific examples in the first embodiment are also applicable to the wind turbine frequency control system based on twin network in this embodiment. Through the above detailed description of the wind turbine frequency control method based on twin network, those skilled in the art can clearly know the wind turbine frequency control system based on twin network in this embodiment, so for the sake of brevity of the specification, it will not be described in detail here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method description.

[0076] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0077] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. A wind turbine frequency control method based on twin networks, characterized in that: include: Based on the predetermined output power factor dimension, multi-dimensional features of the target wind turbine are collected to obtain target output power factor features; Retrieving an output power prediction model to analyze the target output power factor characteristics to obtain a target output power prediction value; Determine whether the target output power prediction value meets a predetermined output power threshold value, and obtain a determination result; According to the judgment result, if the target output power prediction value meets the predetermined output power threshold, a monitoring instruction is issued, and target operating state information of the target wind turbine is dynamically acquired based on the monitoring instruction; Acquire historical operating state information, and input the historical operating state information and the target operating state information into the twin network to obtain state similarity; When the state similarity reaches a predetermined similarity threshold, obtaining an optimal historical frequency modulation strategy corresponding to the historical operation state information; Performing dynamic frequency regulation control on the target wind turbine generator set according to the optimal historical frequency regulation strategy; The method further comprises: reading predetermined state characteristics; The first sub-network in the twin network extracts features of the historical operation state information based on the predetermined state features to obtain historical operation feature parameters; The second sub-network in the twin network performs feature extraction on the target operation state information based on the predetermined state feature to obtain a target operation feature parameter; Taking the parameter similarity between the historical operation characteristic parameter and the target operation characteristic parameter as the state similarity; The method further comprises: Get any frequency modulation index; Extracting a first historical frequency modulation record from the historical frequency modulation control record, wherein the first historical frequency modulation record includes a first historical feature parameter corresponding to the first state feature and a first historical frequency modulation strategy; Matching any frequency modulation index parameter corresponding to the arbitrary frequency modulation index in the first historical frequency modulation strategy, and using the arbitrary frequency modulation index parameter as a dependent variable; Performing correlation analysis on the first historical characteristic parameter as an independent variable and the dependent variable to obtain a first correlation coefficient; If the first correlation coefficient meets a predetermined correlation constraint, adding the first state feature to the predetermined state feature; The method further comprises: The historical frequency modulation control records are screened using the historical operation status information as a screening constraint to obtain a candidate historical frequency modulation record list; Using multiple historical frequency modulation strategies in multiple historical frequency modulation records in the candidate historical frequency modulation record list as the optimization space; The frequency modulation effect evaluation function is introduced as the optimization evaluation index; Based on the optimization evaluation index, frequency optimization is performed in the optimization space to obtain the optimal historical frequency optimization strategy; The frequency modulation effect evaluation function refers to a weighted evaluation function of power grid stability and frequency modulation energy consumption. The power grid stability refers to the stability result obtained by weighting the variation of predetermined power grid stability indicators. The predetermined power grid stability indicators include regulating vibration frequency, power change rate, steady-state error and control robustness.

2. The wind turbine frequency control method based on twin network according to claim 1 is characterized in that: The predetermined output power factor dimension includes a unit factor dimension and an environmental condition dimension, wherein the unit factor dimension includes unit mechanical loss, unit electrical loss, and unit structural design, and the unit structural design includes at least blade shape, blade size, blade material, blade angle, and tower height; and the environmental condition dimension includes ambient wind speed, air density, and ambient temperature and humidity.

3. The wind turbine frequency control method based on twin network according to claim 2 is characterized in that: include: Obtaining historical output power monitoring records of similar units of the target wind turbine unit, and extracting a first historical monitoring record from the historical output power monitoring records; Machine learning is performed on a first data group composed of the first historical unit factor characteristics, the first historical environmental factor characteristics and the first historical output power value in the first historical monitoring record, and the output power prediction model is obtained through training.

4. The wind turbine frequency control method based on twin network according to claim 1 is characterized in that: It also includes, based on the judgment result, activating a predetermined frequency regulation auxiliary mechanism to perform frequency regulation auxiliary processing on the target wind turbine group if the target output power prediction value does not meet the predetermined output power threshold.

5. The wind turbine frequency control system based on twin network is characterized by: The steps for implementing the wind turbine frequency control method based on the twin network as claimed in any one of claims 1 to 4, wherein the wind turbine frequency control system based on the twin network comprises: A feature collection module is used to collect multi-dimensional features of a target wind turbine generator set based on a predetermined output power factor dimension to obtain a target output power factor feature; A prediction value acquisition module is used to retrieve the output power prediction model to analyze the target output power factor characteristics to obtain a target output power prediction value; A threshold judgment module is used to judge whether the target output power prediction value meets a predetermined output power threshold value and obtain a judgment result; An operation status information acquisition module, configured to issue a monitoring instruction according to the judgment result, if the target output power prediction value meets the predetermined output power threshold, and dynamically acquire the target operation status information of the target wind turbine generator set based on the monitoring instruction; A state similarity acquisition module is used to acquire historical operation state information, and input the historical operation state information and the target operation state information into the twin network to obtain state similarity; An optimal historical frequency modulation strategy acquisition module, used for acquiring the optimal historical frequency modulation strategy corresponding to the historical operation state information when the state similarity reaches a predetermined similarity threshold; The frequency regulation control module is used to perform dynamic frequency regulation control on the target wind turbine generator set according to the optimal historical frequency regulation strategy.

Images