Wideband oscillation impedance sub-band identification method and system based on ASSA method

Abstract

The invention discloses a broadband oscillation impedance sub-band identification method based on an ASSA method, and the method comprises the steps: obtaining a broadband impedance different-band identification task of a new energy system, and obtaining an input feature from the identification task; using a sharing layer to extract global features of all frequency band tasks from the input features; the task dedicated layer generates a dedicated feature of each frequency band task according to the global feature, outputs an impedance curve corresponding to each frequency band according to different frequency bands, and obtains a dynamic response of the system under the corresponding frequency; an attention mechanism is adopted, and according to different effects of stability and efficiency of each frequency band in the system, impedance characteristic concern points of different frequency bands are dynamically adjusted by utilizing attention weighted characteristics; and inputting the attention weighted features into a task dedicated layer to extract the most relevant features in the current frequency band task, and outputting an impedance curve corresponding to the current frequency band task. The high-precision and real-time online identification of the broadband impedance of the wind power plant is realized, and the adaptability, the robustness and the precision of the method are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of grid-connected new energy power stations, and in particular to a method and system for identifying broadband oscillating impedance frequency bands based on an ASSA method. Background Art

[0002] In recent years, as wind farms have been gradually connected to the power grid, the problem of broadband oscillation has become a focus of attention. During the grid connection process, the unique characteristics of power electronic equipment often lead to frequent broadband oscillations. These phenomena are different from oscillations in traditional power systems and are unique to new energy systems. They may cause equipment to be disconnected from the grid, damaged, or even seriously threaten the stability of the power grid. Therefore, assessing the risk of broadband oscillations in wind farms is crucial to ensure the safe and stable operation of the power grid.

[0003] Impedance-based analysis methods have become a key means of evaluating the broadband oscillation stability of grid-connected renewable energy systems. This method is favored because of its clear physical interpretation, ease of use, and ability to quantify system stability. The core challenge of establishing an impedance network model for a renewable energy power station is to obtain the impedance model of each unit in the station. Current impedance modeling methods are divided into two categories: white box mechanism modeling and black / gray box measurement modeling. The white box method is widely used because of its wide applicability, but it can only capture the impedance characteristics of the unit under specific working conditions. Several studies have developed multi-machine models using impedance sensitivity and genetic algorithms, and established a subsynchronous oscillation impedance model for directly driving wind farms. Other studies aggregate wind turbines under the same feeder into one unit to construct a sequential impedance model. Another method is to simplify it into an aggregate impedance model by obtaining impedance data from all units, collector lines, and transformers, and connecting them according to the actual topology. However, these methods face huge challenges in practical applications, especially when dealing with renewable energy units with commercial confidential designs and parameters, which limits the use of white box models. The black / grey box approach does not rely on the internal information of the unit, but calculates the impedance by injecting voltage or current perturbations at the grid connection point and recording the corresponding voltage and current responses. Some studies use this approach to identify decoupled model parameters under operating conditions, while others verify the stability of multi-unit systems. Although this approach avoids the limitations of white box modeling, it requires continuous harmonic perturbations, which may interfere with grid operation and increase harmonic pollution.

[0004] For new energy units with known control structures and parameters, theoretical impedance models can be established. However, as the scale of power plants increases, the complexity of the aggregate impedance model grows exponentially, leading to dimensionality problems. For units with "black / gray box" characteristics, the impedance model needs to be obtained through frequency sweep measurements, which is a time-consuming and complex process. With the development of big data technology, data-driven methods are increasingly used in the power field, providing new opportunities for impedance identification. A small number of studies have applied artificial neural networks (ANNs) to impedance identification of power electronic systems. Some studies use recurrent neural networks (RNNs) to identify impedance under specific operating conditions, while others use traditional artificial neural networks to develop multi-point impedance models for grid-connected inverters. In addition, there are studies that use artificial neural networks to model the impedance of doubly fed wind power systems. Physical information neural networks and recurrent neural networks have been used for impedance modeling under multiple operating conditions, but these methods usually treat the impedance as a whole and fail to consider changes across frequency bands. Although these methods are effective in some cases, they fail to address the complexity of broadband impedance identification, especially when the impedance characteristics between high and low frequencies differ significantly.

[0005] Current impedance modeling methods face the following technical challenges: the confidentiality of the internal control structure and parameters of renewable energy units limits the application of traditional white box models; black box or gray box methods bypass these limitations, but require the introduction of continuous harmonic disturbances, which may interfere with grid operation. As the scale of power stations expands, the complexity of aggregate impedance models also increases significantly, making it difficult for existing methods to adapt to different operating conditions while ensuring accuracy and efficiency. Summary of the invention

[0006] The purpose of the present invention is to provide a broadband oscillating impedance frequency band identification method, system, terminal and medium based on the ASSA method, which realizes high-precision, real-time online identification of broadband impedance of wind farms and significantly improves the adaptability, robustness and accuracy of the method.

[0007] The present invention is achieved through the following technical solutions:

[0008] In a first aspect, an embodiment of the present invention provides a method for identifying broadband oscillating impedance frequency bands based on an ASSA method, comprising:

[0009] Obtain the broadband impedance recognition task of the new energy system in different frequency bands, and obtain input features from the recognition task;

[0010] A shared layer is used to extract global features for all frequency band tasks from the input features;

[0011] The task-specific layer generates the specific features of each frequency band task based on the global features, outputs the impedance curve corresponding to each frequency band according to different frequency bands, and obtains the dynamic response of the system at the corresponding frequency;

[0012] The attention mechanism is used to dynamically adjust the focus of impedance characteristics of different frequency bands according to the different roles of each frequency band in the stability and efficiency of the system using the attention weighted features;

[0013] The attention-weighted features are input into the task-specific layer to extract the most relevant features of the current frequency band task, and the impedance curve corresponding to the current frequency band task is output.

[0014] In a second aspect, another embodiment of the present invention provides a broadband oscillating impedance frequency band identification system based on the ASSA method, comprising: a data acquisition module, a shared dedicated module and an attention mechanism module;

[0015] The data acquisition module is used to obtain input features from the recognition task according to different frequency bands of broadband impedance of the new energy system;

[0016] The shared dedicated module includes a shared layer unit and a task-specific layer unit, wherein the shared layer unit is used to extract global features of all frequency band tasks from input features, and the task-specific layer unit is used to generate dedicated features of each frequency band task according to the global features, and output the impedance curve corresponding to each frequency band according to different frequency bands to obtain the dynamic response of the system at the corresponding frequency;

[0017] The attention mechanism module adopts the attention mechanism according to the different roles of each frequency band in the stability and efficiency of the system, and uses the attention weighted features to dynamically adjust the focus of the impedance characteristics of different frequency bands. The attention weighted features are input into the task-specific layer to extract the most relevant features of the current frequency band task, and the impedance curve corresponding to the current frequency band task is output.

[0018] In a third aspect, an embodiment of the present invention provides an intelligent terminal, comprising a processor, an input device, an output device and a memory, wherein the processor is connected to the input device, the output device and the memory respectively, the memory is used to store a computer program, the computer program comprises program instructions, and the processor is configured to call the program instructions to execute the method described in the above embodiment.

[0019] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor executes the method described in the above embodiment.

[0020] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0021] The embodiment of the present invention provides a method, system, terminal and medium for identifying broadband oscillating impedance frequency bands based on the ASSA method, which organically combines the shared and dedicated structures with the attention mechanism. The shared layer is responsible for extracting the global features in the input features, and the task-specific layer optimizes the specific features of each frequency band. The attention mechanism dynamically adjusts the focus of the impedance characteristics of different frequency bands according to the different roles of each frequency band in the stability and efficiency of the system. The attention-weighted features are input into the task-specific layer to extract the most relevant features of the current frequency band task, so that the focus of the input features can be dynamically adjusted, thereby improving the accuracy and efficiency of the broadband impedance curve identification of the new energy system. By introducing the ASSA algorithm, high-precision, real-time online identification of broadband impedance of wind farms is achieved, which significantly improves the adaptability, robustness and accuracy of the system, and solves the problems of large errors, poor robustness, and inability to measure online in real time in traditional methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without creative work. In the drawings:

[0023] Figure 1 A flowchart of a broadband oscillating impedance frequency band identification method based on the ASSA method provided in the first embodiment of the present invention;

[0024] Figure 2 Schematic diagram of the attention mechanism structure;

[0025] Figure 3 This is the overall architecture diagram of the ASSA model;

[0026] Figure 4 Another embodiment of the present invention provides a structural block diagram of a broadband oscillating impedance frequency band identification system based on the ASSA method;

[0027] Figure 5 A structural block diagram of a smart terminal provided by another embodiment of the present invention. DETAILED DESCRIPTION

[0028] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0029] Example 1

[0030] like Figure 1 As shown, a method for identifying broadband oscillating impedance frequency bands based on the ASSA method is provided in the first embodiment of the present invention, comprising the following steps:

[0031] Obtain the broadband impedance recognition task of the new energy system in different frequency bands, and obtain input features from the recognition task;

[0032] A shared layer is used to extract global features for all frequency band tasks from the input features;

[0033] The task-specific layer generates the specific features of each frequency band task based on the global features, outputs the impedance curve corresponding to each frequency band according to different frequency bands, and obtains the dynamic response of the system at the corresponding frequency;

[0034] The attention mechanism is used to dynamically adjust the focus of impedance characteristics of different frequency bands according to the different roles of each frequency band in the stability and efficiency of the system using the attention weighted features;

[0035] The attention-weighted features are input into the task-specific layer to extract the most relevant features of the current frequency band task, and the impedance curve corresponding to the current frequency band task is output.

[0036] ASSA is an adaptive sparse self-attention mechanism. The core idea of ​​the attention mechanism is to enable the model to focus on certain key parts of the input data. In the broadband impedance identification task of the new energy system, the input features usually include the AC and DC components of the voltage and current, which reflect the response characteristics of the system in different frequency bands. However, different frequency band tasks focus on different features, and their sensitivity may vary from task to task. In this embodiment, a shared-dedicated structure is used for feature extraction and task optimization, which can effectively cope with complex tasks. Ensure that the impedance curve prediction of each frequency band is accurate.

[0037] First, the shared layer extracts global features for all frequency band tasks from the input features. This shared layer is responsible for capturing the common characteristics of the system in multiple frequency bands, which is crucial for overall stability analysis, especially when facing complex new energy systems with fluctuating impedance responses in different frequency bands.

[0038] 1) Shared Layer

[0039] The shared layer is derived from the original input data (e.g., the voltage U of the dq component) d ,U q and current I d ,I q ) to extract global features, reflecting the response of the system in different frequency bands. Its mathematical expression is:

[0040] H s =f(W s X+b s ) (1),

[0041] In formula (1), is the weight matrix of the shared layer, is the bias term, f is the activation function, H s Represents the global features extracted from the input data, providing a basis for subsequent task-specific layers.

[0042] 2) Task-specific layer

[0043] For each frequency band task, the response characteristics of the system are significantly different. Therefore, the task-specific layer generates a dedicated feature representation for each frequency band task based on the shared features. The task-specific layer optimizes the shared global features according to the needs of each frequency band task to ensure accurate processing of the impedance curve prediction.

[0044] The mathematical expression of the task-specific layer is:

[0045] H i =f(W i H s +b i ) (2),

[0046] In formula (2), is the weight matrix of task i, is the bias term, Represents the dedicated features for task i. This layer ensures that the model can be optimized based on the impedance characteristics of different frequency bands.

[0047] Finally, the output impedance curve Z of task i i (f) is expressed as:

[0048] Z i (f) = W o H i +b o (3)

[0049] In formula (3), W0 is the weight matrix of the impedance curve output layer, and b0 is the bias term of the impedance curve output layer. By optimizing the task-specific layer, the model can generate an accurate impedance curve for each frequency band, reflecting the dynamic response of the system at a specific frequency.

[0050] In practical applications, the importance of each frequency band feature in the new energy system varies. Different frequency bands play different roles in the stability and efficiency of the system, and the model must dynamically adjust its focus according to the importance of the features of different frequency bands. The attention mechanism improves the accuracy of impedance curve identification by dynamically adjusting the model's focus on the most critical input features of the current task. The structure of the attention mechanism is shown in Figure 2:

[0051] 1) Generation of query vector, key vector and value vector

[0052] The attention mechanism generates query vectors, key vectors, and value vectors in the task-specific layer to measure the relevance of different input features to each specific task. Its mathematical expression is:

[0053] Q i =W q H s (4)

[0054] K i =W k H s (5)

[0055] V i =W v H s (6)

[0056] Among them, Q i is the query vector, K i is the key vector, V i is a value vector, is the weight matrix of the query vector, is the weight matrix of the key vector, is a weight matrix of value vectors. These vectors help quantify the relevance of the input features to the current task.

[0057] 2) Calculation of attention weight

[0058] The relevance between each input feature and the task is calculated by the dot product of the query vector and the key vector, which reflects the importance of a particular input feature to the current task:

[0059]

[0060] Among them, A i is the attention weight, d k is the dimension of the key vector. Next, the softmax function is used to normalize the attention score to obtain the attention weight α i :

[0061] α i =softmax(A i ) (8),

[0062] These attention weights α i Used to weight different input features and guide the model to focus on the input information most relevant to the task.

[0063] 3) Generation of attention-weighted features

[0064] Attention weight α iUsed to perform weighted summation on the value vector to generate the attention weighted feature T of task i i :

[0065]

[0066] Among them, T i Represents the combination of the most important input features, ensuring that the model can dynamically adjust its focus according to the specific needs of each frequency band task; α i,j V represents the attention weight of the jth feature to task i, which is used to measure the importance of this dimension or feature to the task; i,j Represents the attention value of the jth feature for task i.

[0067] In the broadband impedance frequency division identification task, the impedance curves of different frequency bands are crucial to the stability and efficiency of the system. In order to improve the accuracy and efficiency of identification, the task-specific layer in the ASSA method does not directly use the output H of the shared layer. s , but the attention weighted feature T i As input. The overall architecture of the ASSA model is as follows Figure 3 This design enables the task-specific layer to dynamically focus on the most relevant features of the current band task and optimize the output results based on the system’s response.

[0068] Specifically, in the ASSA model, the input of the task-specific layer is no longer the direct output H of the shared layer s , but the weighted feature T generated by the attention mechanism i . Its mathematical expression is:

[0069] H i =f(W i T i +b i ) (10),

[0070] Based on this, the output impedance curve Z of task i i (f) can be expressed as:

[0071] Z i (f) = W o H i +b o (11)

[0072] This combination ensures the flexibility of the ASSA model in handling tasks in different frequency bands, enabling it to dynamically adjust its focus on input features, thereby improving the accuracy and efficiency of broadband impedance curve identification of new energy systems. By combining the shared-dedicated structure with the attention mechanism, the ASSA method can effectively cope with complex broadband impedance identification tasks and ensure the stability and security of the system in various frequency bands.

[0073] The beneficial effects of the present invention are as follows, focusing on the technical effects directly produced by the innovative points, and comparing with the prior art through experimental data and theoretical analysis:

[0074] 1.Significantly improved accuracy

[0075] The ASSA algorithm (shared and dedicated structure based on attention mechanism) proposed in this paper shows higher accuracy in broadband impedance identification tasks compared with traditional regression algorithms (such as linear regression, decision tree, random forest, Bayesian regression). It can be seen from the experimental data that the mean square error (MSE) of the ASSA algorithm is only 0.0098, the root mean square error (RMSE) is 0.0989, the mean absolute error (MAE) is as low as 0.0105, and the coefficient of determination (R 2 ) reaches 0.9412, as shown in Table 1, which is better than other comparison algorithms. This shows that the algorithm proposed in the embodiment of the present invention has obvious advantages in error control and data fitting ability, and can effectively improve the accuracy of impedance identification, especially showing excellent performance when processing wind farm data under complex working conditions.

[0076] The following are the calculation formulas for each error indicator mentioned:

[0077] Mean Squared Error (MSE):

[0078]

[0079] Root Mean Squared Error (RMSE):

[0080]

[0081] Mean Absolute Error (MAE):

[0082]

[0083] Coefficient of determination (R 2 ,Coefficient of Determination):

[0084]

[0085] Table 1 Algorithm error index comparison table

[0086] algorithm Mean Square Error (MSE) Root mean square error (RMSE) Mean Absolute Error (MAE) <![CDATA[Coefficient of determination (R 2 )]]> ASSA Algorithm 0.0098 0.0989 0.0105 0.9412 Linear Regression 0.0567 0.2381 0.0589 0.8235 Decision Tree 0.0345 0.1856 0.0367 0.8812 Random Forest 0.0212 0.1453 0.0234 0.9056 Bayesian Regression 0.0456 0.2134 0.0478 0.8567

[0087] 2. Enhanced robustness and generalization capabilities

[0088] Under various wind farm conditions, the ASSA algorithm ensures the wide coverage of the data set and the adaptability of the model by training various random factors such as wind speed, number of wind turbines, and series compensation. In the regression task, the ASSA algorithm demonstrates strong robustness and generalization ability, and can maintain low error and high prediction accuracy in complex practical application scenarios. Compared with traditional algorithms such as linear regression and random forest, ASSA has significantly reduced errors in multi-condition tests, and R 2 The value is always close to 1, indicating that the algorithm can stably capture the broadband impedance characteristics of the system under different conditions and has stronger prediction ability.

[0089] 3. Dynamically adjust feature weights to improve system performance

[0090] The present invention combines shared and dedicated structures and attention mechanisms, so that the ASSA model can dynamically adjust the focus on input features according to the needs of tasks in different frequency bands. This design greatly improves the impedance identification accuracy of the model in different frequency bands, especially in scenarios with a wide frequency range and significant characteristic differences. Experimental results show that the fitting error between the predicted value and the true value of the ASSA algorithm in impedance identification in high-frequency (1500Hz to 3000Hz) and low-frequency (1Hz to 50Hz) segments is significantly lower than that of traditional methods, ensuring stability and accuracy over a wide frequency range.

[0091] 4. Real-time online recognition capability

[0092] The present invention uses the impedance identification model built by TensorFlow 2.8 and the ASSA algorithm to realize real-time online impedance identification on the server. Compared with the traditional frequency sweep measurement method, the algorithm proposed in the present invention can avoid interference and harmonic pollution to the operation of the power grid during the frequency sweep process, and realize efficient online identification. By comparing the frequency sweep measurement results in the frequency range of 1Hz to 1000Hz, it is verified that the predicted value of the ASSA algorithm is highly consistent with the true value, indicating that it has strong real-time and practicality, and can be reliably applied in the actual operation of wind farms.

[0093] In summary, the present invention introduces the ASSA algorithm to identify the impedance characteristics of different frequency bands in parallel, without the need for frequency scanning equipment, thereby realizing online monitoring of the impedance of new energy power stations in actual operation, reducing harmonic pollution, improving the safety and operational stability of the power grid, achieving high-precision broadband impedance of wind farms, significantly improving the adaptability, robustness and accuracy of the system, and solving the problems of large errors, poor robustness, and inability to perform real-time online measurements in traditional methods.

[0094] Example 2

[0095] like Figure 4As shown, another embodiment of the present invention provides a broadband oscillating impedance frequency band identification system based on the ASSA method, including: a data acquisition module, a shared dedicated module and an attention mechanism module; the data acquisition module is used to obtain different frequency band identification tasks of the broadband impedance of the new energy system, and obtain input features from the identification tasks; the shared dedicated module includes a shared layer unit and a task-specific layer unit, the shared layer unit is used to extract global features of all frequency band tasks from the input features, the task-specific layer unit is used to generate specific features of each frequency band task according to the global features, and output the impedance curve corresponding to each frequency band according to different frequency bands to obtain the dynamic response of the system at the corresponding frequency; the attention mechanism module adopts the attention mechanism according to the different roles of each frequency band in the stability and efficiency of the system, and uses the attention weighted features to dynamically adjust the focus of the impedance characteristics of different frequency bands, inputs the attention weighted features into the task-specific layer to extract the most relevant features of the current frequency band task, and outputs the impedance curve corresponding to the current frequency band task.

[0096] The input features include the DC and AC components of voltage and current. The attention mechanism module includes a relevance calculation unit, an attention weight calculation unit, and a weighted combination unit; the relevance calculation unit uses the attention mechanism to generate query vectors, key vectors, and value vectors in the task-specific layer to measure the relevance of different input features and each specific task; the attention weight calculation unit calculates the relevance between each input feature and the task through the dot product of the query vector and the key vector, and uses the softmax function to normalize the attention score to obtain the attention weight; the weighted combination unit uses the attention weight to perform weighted summation on the value vector to generate the attention weighted feature of the task.

[0097] An embodiment of the present invention provides a broadband oscillating impedance frequency band identification system based on the ASSA method, which organically combines shared and dedicated structures with attention mechanisms. The shared layer unit is responsible for extracting the global features of the system, the task-specific layer unit optimizes the specific characteristics of each frequency band, and the attention mechanism unit dynamically adjusts the model's focus on the input features to ensure accurate identification of impedance characteristics in different frequency bands. This significantly improves the accuracy and efficiency of broadband impedance identification and solves the difficulties faced by traditional models when dealing with complex characteristics of multiple frequency bands. The ASSA model is used to identify impedance characteristics of different frequency bands in parallel without the need for frequency scanning equipment, thereby realizing online impedance monitoring of new energy power stations in actual operation, reducing harmonic pollution, and improving the safety and operational stability of the power grid. Compared with traditional white box and black box methods, this system can be more efficiently and accurately applied to multi-operating scenarios of complex new energy systems.

[0098] Example 3

[0099] like Figure 5As shown, a structural block diagram of an intelligent terminal according to another embodiment of the present invention is shown, the terminal includes a processor, an input device, an output device and a memory, the processor is connected to the input device, the output device and the memory respectively, the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the method described in the first embodiment above.

[0100] It should be understood that in the embodiments of the present invention, the processor referred to may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0101] Input devices may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint direction information), a microphone, etc., and output devices may include a display (LCD, etc.), a speaker, etc.

[0102] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0103] In a specific implementation, the processor, input device, and output device described in the embodiments of the present invention may execute the implementation described in the method embodiment provided in the embodiments of the present invention, or may execute the implementation described in the system embodiment described in the embodiments of the present invention, which will not be described in detail here.

[0104] Example 4

[0105] In another embodiment of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program. The computer program includes program instructions. When the program instructions are executed by a processor, the processor executes the method described in the first embodiment.

[0106] The computer-readable storage medium may be an internal storage unit of the terminal described in the foregoing embodiments, such as a hard disk or memory of the terminal. The computer-readable storage medium may also be an external storage device of the terminal, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the terminal. Further, the computer-readable storage medium may also include both an internal storage unit of the terminal and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by the terminal. The computer-readable storage medium may also be used to temporarily store data that has been output or is to be output.

[0107] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0108] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the terminals and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here.

[0109] In the several embodiments provided in the present application, it should be understood that the disclosed terminal and method can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or it can be an electrical, mechanical or other form of connection.

[0110] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for identifying broadband oscillating impedance frequency bands based on the ASSA method, characterized in that: include: Obtain the broadband impedance recognition task of the new energy system in different frequency bands, and obtain input features from the recognition task; A shared layer is used to extract global features for all frequency band tasks from the input features; The task-specific layer generates the specific features of each frequency band task based on the global features, outputs the impedance curve corresponding to each frequency band according to different frequency bands, and obtains the dynamic response of the system at the corresponding frequency; The attention mechanism is used to dynamically adjust the focus of impedance characteristics of different frequency bands according to the different roles of each frequency band in the stability and efficiency of the system using the attention weighted features; The attention-weighted features are input into the task-specific layer to extract the most relevant features of the current frequency band task, and the impedance curve corresponding to the current frequency band task is output.

2. The method according to claim 1, characterized in that The input characteristics include DC and AC components of voltage and current.

3. The method according to claim 2, characterized in that The formula for the shared layer to extract the global features of all frequency band tasks from the input features is: H s =f(W s X+b s ) In the formula, is the weight matrix of the shared layer, is the bias term, f is the activation function, H s Represents the global features extracted from the input data.

4. The method according to claim 3, characterized in that The specific method of using the attention mechanism to dynamically adjust the focus points of impedance characteristics of different frequency bands by using the attention weighted features according to the different effects of each frequency band in the stability and efficiency of the system includes: An attention mechanism is used to generate query vectors, key vectors, and value vectors within the task-specific layer to measure the relevance of different input features to each specific task. The relevance between each input feature and the task is calculated by the dot product of the query vector and the key vector, and the attention score is normalized using the softmax function to obtain the attention weight; The attention weights are used to perform weighted summation on the value vector to generate the attention-weighted features of the task.

5. A broadband oscillating impedance frequency band identification system based on the ASSA method, characterized in that: include: Data acquisition module, shared dedicated module and attention mechanism module; The data acquisition module is used to obtain input features from the recognition task according to different frequency bands of broadband impedance of the new energy system; The shared dedicated module includes a shared layer unit and a task-specific layer unit, wherein the shared layer unit is used to extract global features of all frequency band tasks from input features, and the task-specific layer unit is used to generate dedicated features of each frequency band task according to the global features, and output the impedance curve corresponding to each frequency band according to different frequency bands to obtain the dynamic response of the system at the corresponding frequency; The attention mechanism module adopts the attention mechanism according to the different roles of each frequency band in the stability and efficiency of the system, and uses the attention weighted features to dynamically adjust the focus of the impedance characteristics of different frequency bands. The attention weighted features are input into the task-specific layer to extract the most relevant features of the current frequency band task, and the impedance curve corresponding to the current frequency band task is output.

6. The system according to claim 5, characterized in that The input characteristics include DC and AC components of voltage and current.

7. The system according to claim 6, characterized in that The formula for extracting the global features of all frequency band tasks from the input features by the shared layer unit is: H s =f(W s X+b s ) In the formula, is the weight matrix of the shared layer, is the bias term, f is the activation function, G s Represents the global features extracted from the input data.

8. The system according to claim 7, characterized in that The attention mechanism module includes a relevance calculation unit, an attention weight calculation unit and a weighted combination unit; The relevance calculation unit uses an attention mechanism to generate a query vector, a key vector, and a value vector in a task-specific layer to measure the relevance of different input features to each specific task; The attention weight calculation unit calculates the correlation between each input feature and the task by the dot product of the query vector and the key vector, and normalizes the attention score using the softmax function to obtain the attention weight; The weighted combination unit performs weighted summation on the value vector using the attention weight to generate an attention weighted feature of the task.

9. An intelligent terminal, comprising a processor, an input device, an output device and a memory, wherein the processor is connected to the input device, the output device and the memory respectively, and the memory is used to store a computer program, wherein the computer program comprises program instructions, and wherein: The processor is configured to call the program instructions to execute the method according to any one of claims 1-4.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor is caused to perform the method according to any one of claims 1 to 4.

Images