A power system stability analysis method and terminal

Abstract

The application discloses a power system stability analysis method and a terminal, and comprises the following steps: obtaining a standard value and an actual value of a power system characteristic; establishing a power system stability analysis model based on a Hilbert space origin according to the standard value; training the power system stability analysis model by using a hybrid machine learning algorithm and frequency spectrum data based on Hilbert transform; inputting the actual value into the trained power system stability analysis model to obtain a power system stability value; establishing a visualization model based on the power system stability value; and displaying the visualization model on a display screen. The application can effectively improve the accuracy of power system stability analysis, intuitively view the analysis result, and solve the problem of low accuracy in existing power system stability analysis.

Description

Technical Field

[0001] This invention relates to the field of power system stability analysis technology, and in particular to a power system stability analysis method and terminal. Background Technology

[0002] Transient stability of power systems is a crucial component of power quality and a vital factor for the normal and safe operation of power systems. When power systems experience fluctuations, various accidents such as power outages and overvoltages can easily occur, causing significant adverse effects and damage, threatening personal safety and economic development.

[0003] Existing power system stability analysis techniques are qualitative, relying on stability criteria and waveform comparisons to determine the stability of the power system. Most existing methods perform qualitative analysis on transient stability without considering random factors, thus reducing the accuracy of the analysis. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and terminal for power system stability analysis, which can effectively improve the accuracy of power system stability analysis.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method for power system stability analysis includes the following steps: S1. Obtain the standard and actual values ​​of power system characteristics; S2. Establish a power system stability analysis model based on the Hilbert space origin according to the standard values; S3. Train the power system stability analysis model using a hybrid machine learning algorithm and spectrum data based on Hilbert transform; S4. Input the actual value into the trained power system stability analysis model to obtain the power system stability value; S5. Establish a visualization model based on the power system stability values; S6. Display the visualization model on the display screen.

[0006] To solve the above-mentioned technical problems, another technical solution adopted by the present invention is as follows: A power system stability analysis terminal includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps in the power system stability analysis method described above.

[0007] The beneficial effects of this invention are as follows: First, standard and actual values ​​of power system characteristics are obtained. Then, a power system stability analysis model based on the Hilbert space origin is established according to the standard values. A hybrid machine learning algorithm is introduced to train the established model, improving its accuracy. Then, the actual values ​​are input into the trained power system stability analysis model to obtain more comprehensive stability values. Finally, a visualization model is established and displayed on a screen. This improves the accuracy of the analysis while enabling intuitive viewing of the analysis results, solving the problem of low accuracy in existing power system stability analysis. Attached Figure Description

[0008] Figure 1 This is an overall flowchart of a power system stability analysis method according to an embodiment of the present invention; Figure 2 A flowchart for establishing a power system stability analysis model based on the Hilbert space origin according to standard values; Figure 3 A flowchart illustrating power system stability analysis using Hilbert spectrum data as input for a hybrid machine learning algorithm; Figure 4 A flowchart for displaying a visual model on a screen; Figure 5 This is a flowchart illustrating the adjustment of the display brightness and height of the screen based on current environmental data. Figure 6 This is a schematic diagram of the structure of a power system stability analysis terminal according to an embodiment of the present invention.

[0009] Label Explanation: 1. A power system stability analysis terminal; 2. Memory; 3. Processor. Detailed Implementation

[0010] To explain in detail the technical content, objectives, and effects of the present invention, the following description is provided in conjunction with the embodiments and accompanying drawings.

[0011] Please refer to Figures 1 to 5 A method for power system stability analysis, comprising the following steps: S1. Obtain the standard and actual values ​​of power system characteristics; S2. Establish a power system stability analysis model based on the Hilbert space origin according to the standard values; S3. Train the power system stability analysis model using a hybrid machine learning algorithm and spectrum data based on Hilbert transform; S4. Input the actual value into the trained power system stability analysis model to obtain the power system stability value; S5. Establish a visualization model based on the power system stability values; S6. Display the visualization model on the display screen.

[0012] As described above, the beneficial effects of this invention are as follows: First, standard and actual values ​​of power system characteristics are obtained. Then, a power system stability analysis model based on the Hilbert space origin is established according to the standard values. A hybrid machine learning algorithm is introduced to train the established model, improving its accuracy. Then, the actual values ​​are input into the trained power system stability analysis model to obtain more comprehensive stability values. Finally, a visualization model is established and displayed on a screen. This improves the accuracy of the analysis while enabling intuitive viewing of the analysis results, solving the problem of low accuracy in existing power system stability analysis.

[0013] Furthermore, the power system characteristics include active power, reactive power, three-phase voltage, three-phase current, power angle, harmonic distortion rate, and frequency.

[0014] As described above, power system characteristics include active power, reactive power, three-phase voltage, three-phase current, power angle, harmonic distortion rate, and frequency. In other words, considering multiple power system characteristics allows for more comprehensive stability values ​​to be obtained later.

[0015] Further, step S2 specifically includes: S21. Perform Hilbert transform on the standard values ​​to obtain the Hilbert spectrum; S22. Establish a power system stability analysis model based on the Hilbert space origin according to the Hilbert spectrum.

[0016] Further, step S21 specifically includes: S211, The data to be analyzed x ( t The empirical mode decomposition method is used to decompose the data into intrinsic mode functions. The result of the decomposition is shown in the following formula (1): (1); in c j ( t ) is the first j Components of the eigenmode function n It is the number of components of the intrinsic mode function. r n ( t The ) represents the residual. Each component of the intrinsic mode function represents a time-varying oscillation with a specific amplitude and frequency. Singular value decomposition orthogonals the components of each intrinsic mode function. jAs the value increases, the frequency of each component of the intrinsic mode function decreases; S212. Based on step S211, the result of the empirical mode decomposition method is rewritten as the following formula (2): (2); in, x j ( t ) represents the eigenmode function of the eigenmode function. j One portion, c j ( t )middle j =1,..., p , c j ( t The current component is the sinusoidal component to be studied, with a total of... p Item, the rest c k ( t ) represents an unrelated nonsinusoidal quantity, totaling n - p The Hilbert spectrum is obtained by performing a Hilbert transform on the components of the intrinsic mode functions. S213. Perform Hilbert transformation on the result of the empirical mode decomposition method in S212. The approximate result after transformation is as follows (3): (3); in, x Hj ( t () is the sine component c j ( t Hilbert transform of ) i is the imaginary unit. , Let Hilbert's transformation result be: (4); (5); Based on the transformed result, we can obtain: (6); in ω j ( t )= dφ j ( t ) / dt , representing the instantaneous frequency, can be obtained by combining the above formulas (3) and (6): (7); Therefore, we can conclude that: (8); Similarly, we can obtain the first... j The complex signal is ,in z j ( t )= x j ( t )+ ix Hj ( t The original signal is represented as a real number: (9); Formula (9) is the generalized form of Hilbert's spectrum, if p = n Then formula (9) is ( x ( t )- r n ( t The precise value of ).

[0017] As described above, the power system stability analysis model based on the origin of Hilbert space can obtain the numerical value of power system stability. The changes in the obtained numerical value can be used as a reference for power system stability regulation, thereby realizing the analysis of power system stability. The Hilbert space is not limited to finite dimensions, and it visualizes the changes of multiple characteristic indicators, making it suitable for the analysis of multidimensional time-varying problems.

[0018] Further, step S3 specifically includes: S31. Use a data acquisition and monitoring control system to collect raw data on the characteristics of the power system; S32. After processing the collected raw data using the empirical mode decomposition method and the Hilbert transform algorithm, a dataset including intrinsic mode functions and instantaneous frequencies is obtained, which serves as the input values ​​for feature selection and dimensionality reduction of the training module. S33. The training module uses a machine learning algorithm based on hybrid random forest and boosting tree to evaluate the importance of variables, and uses decision tree technology to rank the importance of variables. Each binary decision tree consists of a tree structure. T This indicates that the input vector is ( X 1, X 2,..., X p The range of values ​​is χ 1* χ 2*...* χ p = χ Output variables y value range Y ; Each node t Representation space χ A subset whose root node is χ Based on the top-down construction process of the decision tree, a variable is selected at each step and subjected to a binary search test. s t =( X m < c To split the dataset; Internal nodes t The dataset is divided into two subsets, each corresponding to one of two child nodes. t L and t R For a new data point to be predicted, the predicted value is... The labels obtained when the tree propagates to the leaf nodes are used by the algorithm to construct the decision tree to determine the labels of each node. t Splitting rules S t = s * Among them, N t Node samples are divided into t L and t R Two subsets to minimize the loss function Δi ( s , t )= i ( t )- p L i ( t L )- p R i ( t R The value of ); The specific steps to assess the importance of random forest variables are to... X m All nodes t Weighted loss p ( t ) i ( s t , t Add them together, and apply the results to all the forests. N T Trees are averaged: (10); in, p ( tTo reach the node t The proportion of the sample N t / N , v ( s t ) is used for segmentation s t Variables; S34. The optimization module uses support vector machines and artificial neural network prediction models, as well as nonlinear optimization algorithms, to select the optimal parameters. S35. Use the test module to test and compare the obtained prediction models.

[0019] As described above, the spectrum data obtained by transforming the original data of power system characteristics using the Hilbert algorithm is used as the training and testing sets. This data is then input into the established power system stability analysis model and trained using a hybrid machine learning algorithm that includes mixed random forests and boosting trees. This improves prediction efficiency when faced with limited original data and drastically changing time series parameters, and further enhances the accuracy of the model in analyzing power system stability.

[0020] Further, step S5 specifically includes: S51. Visualization of line power data: Transmission system parameters are represented by a single-line diagram, and the active power, reactive power and reactive capacity data on the transmission line or transformer are displayed in the form of digital text; on the single-line diagram, arrows are used to indicate the direction and magnitude of power flow, and the size of the arrow is proportional to the power on the line; S52. Contour plotting bus data: Using mapping to associate numerical values ​​with colors on the screen; S53. Pie chart to represent line data: The size and color of the pie chart are dynamically adjusted according to the percentage of line load.

[0021] As described above, visualizing line power data, depicting bus data with contour lines, and representing line data with pie charts to establish a visual model of power system stability values ​​allows users to quickly understand the system's flow situation in static diagrams, clearly see the voltage distribution of the entire area, and intuitively display the load distribution and proportion of the lines, with lines with higher loads being more prominent.

[0022] Further, step S6 specifically includes: S61. Adjust the display brightness and display height of the display screen based on the current environmental data, wherein the current environmental data includes the current light intensity and head feature data; S62. Display the visualization model using the adjusted display screen.

[0023] Further, step S61 includes: S611. Measure the current light intensity within the display range of the screen; S612. Take a picture of the scene within the display area of ​​the display screen; S613. Extract features from the scene image to obtain head feature data; S614. Adjust the display brightness of the display screen based on the current light intensity; S615. Adjust the display height of the display screen based on the head feature data.

[0024] Further, the step of extracting features from the scene image to obtain head feature data includes: Build neural network models using hybrid machine learning algorithms; The neural network model is trained using the training dataset to obtain a feature extraction model; The scene image is input into the feature extraction model to obtain head feature data.

[0025] As described above, when displaying a visualization model on a screen, the strategy involves adjusting the display intensity of the screen based on the current light intensity within the display range, thereby ensuring the clarity of the visualization model displayed on the screen. Simultaneously, by constructing a neural network model to extract head feature data, the eye position point is kept horizontal with the height of the screen, improving the user's viewing experience.

[0026] Please refer to Figure 6 A power system stability analysis terminal includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps in the power system stability analysis method described above.

[0027] As described above, the beneficial effects of this invention are as follows: Based on the same technical concept, and in conjunction with the aforementioned power system stability analysis method, a power system stability analysis terminal is provided. First, standard and actual values ​​of power system characteristics are obtained. Then, a power system stability analysis model based on the Hilbert space origin is established according to the standard values. A hybrid machine learning algorithm is introduced to train the established model, improving its accuracy. Then, the actual values ​​are input into the trained power system stability analysis model to obtain more comprehensive stability values. Finally, a visualization model is established and displayed on a screen. This improves the accuracy of the analysis while enabling intuitive viewing of the analysis results, solving the problem of low accuracy in existing power system stability analyses.

[0028] The present invention provides a power system stability analysis method, which is applicable to scenarios where stability analysis of power system characteristics is required. The following is a detailed description with reference to specific embodiments.

[0029] Please refer to Figure 1 Embodiment 1 of the present invention is as follows: A method for power system stability analysis, such as Figure 1 As shown, the steps include: S1. Obtain the standard and actual values ​​of the power system characteristics.

[0030] Among them, the characteristics of the power system include active power, reactive power, three-phase voltage, three-phase current, power angle, harmonic distortion rate and frequency. Considering multiple power system characteristics can enable more comprehensive stability values ​​to be obtained later.

[0031] S2. Establish a power system stability analysis model based on the Hilbert space origin according to the standard values.

[0032] S3. A power system stability analysis model is trained using a hybrid machine learning algorithm and spectrum data based on Hilbert transform.

[0033] S4. Input the actual values ​​into the trained power system stability analysis model to obtain the power system stability values.

[0034] S5. Establish a visualization model based on power system stability numerical values.

[0035] S6. Display the visualization model on the screen.

[0036] In this embodiment, standard and actual values ​​of power system characteristics are first obtained. Then, a power system stability analysis model based on the Hilbert space origin is established based on the standard values. A hybrid machine learning algorithm is introduced to train the established model to improve its accuracy. Then, the actual values ​​are input into the trained power system stability analysis model to obtain more comprehensive stability values. Finally, a visualization model is established and displayed on a screen. This improves the accuracy of the analysis while enabling intuitive viewing of the analysis results, thus solving the problem of low accuracy in existing power system stability analysis.

[0037] Please refer to Figure 2 Embodiment two of the present invention is as follows: A power system stability analysis method, based on the above embodiment one, in this embodiment, as follows: Figure 2 As shown, step S2 specifically involves: S21. Perform a Hilbert transform on the standard values ​​to obtain the Hilbert spectrum, which includes the following steps: S211, The data to be analyzedx ( t The empirical mode decomposition (EMD) method is used to decompose the data into intrinsic mode functions (IMFs). The decomposition result is shown in the following formula (1): (1); in c j ( t ) is the first j The components of the eigenmode functions, n It is the number of components of the intrinsic mode function. r n ( t The ) represents the residual. Each component of the intrinsic mode function represents a time-varying oscillation with a specific amplitude and frequency. Singular value decomposition orthogonals the components of each intrinsic mode function. j As the value increases, the frequency of each component of the intrinsic mode function decreases.

[0038] S212. Based on step S211, the result of the empirical mode decomposition method is rewritten as the following formula (2): (2); in, x j ( t ) represents the eigenmode function of the eigenmode function. j One portion, c j ( t )middle j =1,..., p , c j ( t The current component is the sinusoidal component to be studied, with a total of... p Item, the rest c k ( t ) represents an unrelated nonsinusoidal quantity, totaling n - p Item, that is k from p +1、 p +2, ... keep adding until... n There are a total of n -( p +1)+1= n - p The Hilbert spectrum is obtained by performing a Hilbert transform on the components of the intrinsic mode functions.

[0039] S213. Perform Hilbert transformation on the result of the empirical mode decomposition method in S212. The approximate result after transformation is as follows (3): (3); in, x Hj ( t ) is the sine component c j ( t Hilbert transform of ) i is the imaginary unit. , Let Hilbert's transformation result be: (4); (5); Based on the transformed result, we can obtain: (6).

[0040] in ω j ( t )= dφ j ( t ) / dt , representing the instantaneous frequency, can be obtained by combining the above formulas (3) and (6): (7); Therefore, we can conclude that: (8).

[0041] Similarly, we can obtain the first... j The complex signal is ,in z j ( t )= x j ( t )+ ix Hj ( t The original signal is represented as a real number: (9); Formula (9) is the generalized form of Hilbert's spectrum, if p = n Then formula (9) is ( x ( t )- r n ( t The precise value of ).

[0042] S22. Establish a power system stability analysis model based on the Hilbert space origin according to the Hilbert spectrum.

[0043] In this embodiment, the power system stability analysis model based on the origin of Hilbert space can obtain power system stability values. The changes in the obtained values ​​can be used as a reference for power system stability regulation, thereby realizing the analysis of power system stability. The Hilbert space is not limited to finite dimensions, and it visualizes the changes of multiple characteristic indicators, making it suitable for the analysis of multidimensional time-varying problems.

[0044] Please refer to Figure 3 Embodiment 3 of the present invention is as follows: A power system stability analysis method, based on the above-described Embodiment 1 or Embodiment 2, in this embodiment, step S3 uses a hybrid machine learning algorithm to input Hilbert spectrum data into a preliminarily established power system stability analysis model to analyze the power system stability results. The main objective of the hybrid learning algorithm in training the prediction model is to improve prediction efficiency when faced with limited raw power system characteristic data (i.e., historical data) and drastically changing time series parameters. Therefore, in this embodiment, as... Figure 3 As shown, step S3 specifically involves: S31. Use a Supervisory Control and Data Acquisition (SCADA) system to collect raw data on the characteristics of the power system.

[0045] S32. After processing the collected raw data using Empirical Mode Decomposition (EMD) and Hilbert-Huang Transform (HT), a dataset including intrinsic mode functions and instantaneous frequencies is obtained, which serves as the input values ​​for feature selection and dimensionality reduction of the training module.

[0046] S33. The training module uses a machine learning algorithm based on a hybrid random forest and boosting tree to evaluate the importance of variables, and uses decision tree technology to rank the importance of variables. Each binary decision tree consists of a tree structure. T This indicates that the input vector is ( X 1, X 2,..., X p The range of values ​​is χ 1* χ 2*...* χ p = χ Output variables y value range Y .

[0047] Then each node t Representation space χ A subset whose root node is χ Based on the top-down construction process of the decision tree, a variable is selected at each step and subjected to a binary search test. s t =( X m < c The dataset is divided into internal nodes. t The dataset is divided into two subsets, each corresponding to one of two child nodes. t L and t R For a new data point to be predicted, the predicted value is... The labels obtained when the tree propagates to the leaf nodes are used by the algorithm to construct the decision tree to determine the labels of each node. t Splitting rules S t = s * Among them, N t Node samples are divided into t L and t R Two subsets to minimize the loss function Δi ( s , t )= i ( t )- p L i ( t L )- p R i ( t R The value of ).

[0048] Finally, the importance of the random forest variables was evaluated by specifically... X m All nodes t Weighted loss p ( t ) i ( s t , t Add them together, and apply the results to all the forests. N T Trees are averaged: (10); in,p ( t To reach the node t The proportion of the sample N t / N , v ( s t ) is used for segmentation s t Variables.

[0049] S34. The optimization module uses support vector machines (SVM), artificial neural network prediction models (ANN), and nonlinear optimization algorithms to select the optimal parameters; S35. Use the test module to test and compare the obtained prediction models.

[0050] In this embodiment, the spectrum data obtained by transforming the original data of power system characteristics using the Hilbert algorithm is used as the training and testing sets. These sets are then input into the established power system stability analysis model, which is trained using a hybrid machine learning algorithm that includes mixed random forests and boosting trees. This improves prediction efficiency when faced with limited original data and drastically changing time series parameters, and further enhances the accuracy of the model in analyzing power system stability.

[0051] Please refer to Figure 4 and Figure 5 Embodiment four of the invention is as follows: A power system stability analysis method, based on any one of the above embodiments one to three, specifically includes step S5 in this embodiment: S51. Line Power Data Visualization: Transmission system parameters are represented using single-line diagrams, displaying active power, reactive power, and reactive capacity data on transmission lines or transformers in digital text format. Arrows can be used on the single-line diagram to indicate the direction and magnitude of power flow; the arrow size is proportional to the power on the line. This method allows users to quickly understand the system's flow situation in a static diagram. Furthermore, animation techniques can be used to make the system visualization more vivid.

[0052] S52. Contour lines depict bus data: Contour lines use mapping to associate numerical values ​​with colors on the screen, which can be used to display continuous data in a spatial distribution. For example, blue is used to represent lower values, and red to represent higher values. The advantage of contour lines is that they provide a clear overview of the voltage distribution over the entire area.

[0053] S53. Pie Chart Representation of Line Data: Pie charts can be used to represent line data. By dynamically adjusting the size and color of the pie chart, the load distribution and proportion of the lines can be visually displayed. The size and color of the pie chart can be adjusted according to the percentage of line load, making lines with higher loads stand out more.

[0054] like Figure 4 As shown, step S6 specifically involves: S61. Adjust the display brightness and display height of the display screen based on the current environmental data, which includes the current light intensity and head feature data.

[0055] In this embodiment, such as Figure 5 As shown, step S61 specifically includes: S611. Measure the current light intensity within the display range of the screen.

[0056] S612. Capture scene images within the display area of ​​the screen; specifically, capture scene images using a camera.

[0057] S613. Perform feature extraction on the scene image to obtain head feature data, specifically including: Construct a neural network model, train the neural network model using the training dataset to obtain a feature extraction model; obtain the training dataset, filter, label and divide the training dataset in sequence to obtain a training set and a validation set; train the neural network model using the training set to obtain a pre-trained model, and then use the validation set to validate and optimize the pre-trained model to obtain the feature extraction model with the highest accuracy.

[0058] Then, the scene images are input into the feature extraction model for training to obtain head feature data, specifically: The scene image is formatted according to the input format of the feature extraction model to obtain input data. The input data is then fed into the feature extraction model for training to obtain head feature data.

[0059] S614. Adjust the display brightness of the screen based on the current light intensity, specifically as follows: If the current light intensity is higher than the display brightness, it indicates that the display is in a strong light environment. To increase the clarity of the displayed model, the display brightness needs to be increased. If the current light intensity is lower than the display brightness, it indicates that the display is in a weak light environment. To protect the user's eyesight and save energy, the display brightness needs to be decreased.

[0060] S615. Adjust the display height of the screen based on head feature data, specifically as follows: Capture the eye position points in the head feature data. If the height of the eye position point is higher than the height of the display screen, adjust the height of the display screen. If the height of the eye position point is lower than the height of the display screen, adjust the height of the display screen so that the eye position point is level with the height of the display screen.

[0061] S62. Display the visualization model using the adjusted display screen.

[0062] In addition, in this embodiment, after displaying the visualization model using the adjusted display screen, the following steps are also included: If the feature extraction model fails to extract head feature data within a preset time period, the display screen will turn off. When the feature extraction model re-extracts the head feature data, the display screen is activated to show the visualization model.

[0063] In this embodiment, when displaying the visualization model on the display screen, the current light intensity within the display range of the strategy display screen is used to adjust the display intensity of the display screen based on the light intensity, ensuring the clarity of the display screen displaying the visualization model; at the same time, a neural network model is constructed to train the extraction of head feature data, so that the eye position point is kept horizontal with the height of the display screen, improving the user's viewing experience.

[0064] Please refer to Figure 6 Embodiment five of the present invention is as follows: A power system stability analysis terminal 1, such as Figure 6 As shown, it includes a memory 2, a processor 3, and a computer program stored on the memory 2 and executable on the processor 3. When the processor 3 executes the computer program, it implements the steps in any of the embodiments of the above embodiments one to four.

[0065] In summary, the power system stability analysis method and terminal provided by this invention can not only effectively improve the accuracy of power system stability analysis, but also enable intuitive viewing of analysis results, thus solving the problem of low accuracy in existing power system stability analysis.

[0066] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent modifications made based on the content of the present invention specification and drawings, or direct or indirect applications in related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for power system stability analysis, characterized in that, Including the following steps: S1. Obtain the standard and actual values ​​of power system characteristics; S2. Establish a power system stability analysis model based on the Hilbert space origin according to the standard values; S3. Train the power system stability analysis model using a hybrid machine learning algorithm and spectrum data based on Hilbert transform; S4. Input the actual value into the trained power system stability analysis model to obtain the power system stability value; S5. Establish a visualization model based on the power system stability values; S6. Display the visualization model on the display screen; Step S3 specifically involves: S31. Use a data acquisition and monitoring control system to collect raw data on the characteristics of the power system; S32. After processing the collected raw data using the empirical mode decomposition method and the Hilbert transform algorithm, a dataset including intrinsic mode functions and instantaneous frequencies is obtained, which serves as the input values ​​for feature selection and dimensionality reduction of the training module. S33. The training module uses a machine learning algorithm based on hybrid random forest and boosting tree to evaluate the importance of variables, and uses decision tree technology to rank the importance of variables. Each binary decision tree consists of a tree structure. T This indicates that the input vector is ( X 1, X 2,..., X p The range of values ​​is χ 1* χ 2*...* χ p = χ Output variables y value range Y ; Each node t Representation space χ A subset whose root node is χ Based on the top-down construction process of the decision tree, a variable is selected at each step and subjected to a binary search test. s t =( X m < c To split the dataset; Internal nodes t The dataset is divided into two subsets, each corresponding to one of two child nodes. t L and t R For a new data point to be predicted, the predicted value is... The labels obtained when the tree propagates to the leaf nodes are used by the algorithm to construct the decision tree to determine the labels of each node. t Splitting rules S t = s * Among them, N t Node samples are divided into t L and t R Two subsets to minimize the loss function Δi ( s , t )= i ( t )- p L i ( t L )- p R i ( t R The value of ); The specific steps to assess the importance of random forest variables are to... X m All nodes t Weighted loss p ( t ) i ( s t , t Add them together, and apply the results to all the forests. N T Average the trees: （10）； in, p ( t To reach the node t proportion of samples N t / N , v ( s t ) is used for segmentation s t Variables; S34. The optimization module uses support vector machines and artificial neural network prediction models, as well as nonlinear optimization algorithms, to select the optimal parameters. S35. Use the test module to test and compare the obtained prediction models.

2. The power system stability analysis method according to claim 1, characterized in that, The power system characteristics include active power, reactive power, three-phase voltage, three-phase current, power angle, harmonic distortion rate, and frequency.

3. The power system stability analysis method according to claim 1, characterized in that, Step S2 specifically involves: S21. Perform Hilbert transform on the standard values ​​to obtain the Hilbert spectrum; S22. Establish a power system stability analysis model based on the Hilbert space origin according to the Hilbert spectrum.

4. The power system stability analysis method according to claim 3, characterized in that, Step S21 specifically involves: S211, The data to be analyzed x ( t The empirical mode decomposition method is used to decompose the data into intrinsic mode functions. The result of the decomposition is shown in the following formula (1): （1）； in c j ( t ) is the first j The components of the eigenmode functions, n It is the number of components of the intrinsic mode function. r n ( t The ) represents the residual. Each component of the intrinsic mode function represents a time-varying oscillation with a specific amplitude and frequency. Singular value decomposition orthogonals the components of each intrinsic mode function. j As the value increases, the frequency of each component of the intrinsic mode function decreases; S212. Based on step S211, the result of the empirical mode decomposition method is rewritten as the following formula (2): （2）； in, x j ( t ) represents the j-th component of the intrinsic mode function. c j ( t )middle j =1,..., p , c j ( t The current component is the sinusoidal component to be studied, with a total of... p Item, the rest c k ( t ) represents an unrelated nonsinusoidal quantity, totaling n - p The Hilbert spectrum is obtained by performing a Hilbert transform on the components of the intrinsic mode functions. S213. Perform Hilbert transformation on the result of the empirical mode decomposition method in S212. The approximate result after transformation is as follows (3): （3）； in, x Hj ( t ) is the sine component c j ( t Hilbert transform of ) i The imaginary unit, Let Hilbert's transformation result be: （4）； （5）； Based on the transformed result, we can obtain: （6）； in ω j ( t )= dφ j ( t ) / dt , representing the instantaneous frequency, can be obtained by combining the above formulas (3) and (6): （7）； Therefore, we can conclude that: （8）； Similarly, we can obtain the first... j The complex signal is ,in z j ( t )= x j ( t )+ ix Hj ( t The original signal is represented as a real number: （9）； Formula (9) is the generalized form of Hilbert's spectrum, if p = n Then formula (9) is ( x ( t )- r n ( t The precise value of ).

5. The power system stability analysis method according to claim 1, characterized in that, Step S5 specifically involves: S51. Visualization of line power data: Transmission system parameters are represented by a single-line diagram, and the active power, reactive power and reactive capacity data on the transmission line or transformer are displayed in the form of digital text; on the single-line diagram, arrows are used to indicate the direction and magnitude of power flow, and the size of the arrow is proportional to the power on the line; S52, Contour Line Drawing Bus Data: Using mapping to associate numerical values ​​with colors on the screen; S53. Pie charts represent line data: The size and color of the pie chart are dynamically adjusted based on the percentage of line load.

6. The power system stability analysis method according to claim 1, characterized in that, Step S6 specifically involves: S61. Adjust the display brightness and display height of the display screen based on the current environmental data, wherein the current environmental data includes the current light intensity and head feature data; S62. Display the visualization model using the adjusted display screen.

7. The power system stability analysis method according to claim 6, characterized in that, Step S61 includes: S611. Measure the current light intensity within the display range of the screen; S612. Take a picture of the scene within the display area of ​​the display screen; S613. Extract features from the scene image to obtain head feature data; S614. Adjust the display brightness of the display screen based on the current light intensity; S615. Adjust the display height of the display screen based on the head feature data.

8. The power system stability analysis method according to claim 7, characterized in that, The step of extracting features from the scene image to obtain head feature data includes: Build neural network models using hybrid machine learning algorithms; The neural network model is trained using the training dataset to obtain a feature extraction model; The scene image is input into the feature extraction model to obtain head feature data.

9. A power system stability analysis terminal, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps in a power system stability analysis method as described in any one of claims 1 to 8.

Images