A broadband oscillation suppression method, medium, and system for shore-based power supply systems

By combining a random forest classifier and an RBF neural network, the broadband oscillation suppression parameter is adaptively selected, which solves the oscillation problem of the shore-based power supply system and improves the system's stability and renewable energy absorption capacity.

CN116505552BActive Publication Date: 2026-11-17STATE GRID SHANGHAI MUNICIPAL ELECTRIC POWER CO +1
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202310497410.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-05-05
Publication Date
2026-11-17
Estimated Expiration
2043-05-05

AI Technical Summary

Technical Problem

After a large-capacity load is connected to a ship, the power electronic devices of the shore-based power supply system introduce multi-timescale and multi-physical field coupling, resulting in weak disturbance immunity and wide-frequency oscillations, which affect the steady-state characteristics and stable operation of the system.

Method used

By combining a random forest classifier model and an RBF neural network, frequency and characteristic data of the shore-based power supply system are acquired to generate a decision tree classification sub-model, output a broadband oscillation suppression strategy, and actively control the broadband oscillation by adaptively selecting suppression parameters through a neural network.

Benefits of technology

It improved the stability of the shore-based power supply system and the level of new energy absorption, eliminated oscillation problems, reduced system delay, and enhanced the stability and reliability of the system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116505552B_ABST
    Figure CN116505552B_ABST
Patent Text Reader

Abstract

This invention discloses a method, medium, and system for suppressing broadband oscillations in shore-based power supply systems. The method includes: acquiring the frequency and corresponding characteristic data of broadband oscillations generated by the shore-based power supply system, and inputting this data into a classifier model as a training set to generate a decision tree classification sub-model; inputting the actual frequency and amplitude of the broadband oscillations generated by the shore-based power supply system, and the equivalent external characteristics of the power supply system, into the classifier model after generating the decision tree classification sub-model, so that the classifier model outputs a broadband oscillation suppression strategy based on classification conditions; inputting the broadband oscillation suppression strategy into a first neural network, outputting broadband oscillation suppression parameters, and suppressing the broadband oscillations of the shore-based power supply system through these parameters. This invention can eliminate oscillation problems in shore-based power supply systems and improve the effective absorption level of new energy sources.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of broadband oscillation suppression technology, and specifically to a broadband oscillation suppression method, medium, and system for shore-based power supply systems. Background Technology

[0002] With the development of inland waterway shipping, the number and density of ships have increased significantly, causing serious pollution to river and coastal areas. In order to effectively curb pollution from ships to ports, efforts have been made to promote shore-based power supply systems, using power electronic devices to achieve electric propulsion for ships instead of traditional diesel engines, thus realizing green shipping.

[0003] However, with the implementation of shore power strategies, the stability and reliability of shore-based power supply systems are difficult to guarantee. On the one hand, the application of a large number of power electronic devices introduces multi-timescale and multi-physical field coupling into shore-based power supply systems, resulting in weak disturbance immunity. On the other hand, the connection of large-capacity loads on ships leads to large fluctuations in the output of power electronics, which seriously affects the steady-state characteristics of shore-based power supply systems. When the output is large, it will threaten the stable operation of the system and cause wide-frequency oscillations. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the purpose of this invention is to provide a broadband oscillation suppression method, medium and system for shore power supply systems, so as to solve the problems mentioned in the background art.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A broadband oscillation suppression method for shore-based power supply systems includes the following steps:

[0007] The frequency and corresponding characteristic data of the broadband oscillation generated by the shore-based power supply system are obtained and input into the classifier model as a training set so that the classifier model can generate a decision tree classification sub-model.

[0008] Input the frequency and amplitude of the broadband oscillations generated by the actual shore-based power supply system, as well as the equivalent external characteristics of the power supply system, into the classifier model after generating the decision tree classification sub-model, so that the classifier model outputs a broadband oscillation suppression strategy based on the classification conditions;

[0009] The broadband oscillation suppression strategy is input into the first neural network, which outputs broadband oscillation suppression parameters. The broadband oscillation of the shore-based power supply system is suppressed by the broadband oscillation suppression parameters.

[0010] A computer-readable storage medium storing computer program instructions that, when executed by a processor, implement the broadband oscillation suppression method.

[0011] A broadband oscillation suppression system for shore-based power supply systems, comprising:

[0012] The training module is used to train the classifier model and generate decision tree classification sub-models;

[0013] The acquisition module is used to acquire the frequency and amplitude of the broadband oscillation generated by the actual shore-based power supply system, the equivalent external characteristics of the power supply system, and input them into the trained classifier model to obtain the broadband oscillation suppression strategy, and then input it into the first neural network.

[0014] The neural network module stores the parameters for running the first neural network and outputting wideband oscillation suppression parameters.

[0015] The suppression module is used to suppress broadband oscillations in the shore-based power supply system through broadband oscillation suppression parameters.

[0016] The beneficial effects of this invention are:

[0017] The method of this invention inputs the measured relevant data and actual monitoring data into the classifier model for clustering operation to obtain a broadband suppression strategy. Then, the most suitable broadband oscillation suppression parameters are adaptively selected through the first neural network to achieve the best broadband oscillation suppression effect. This can eliminate the oscillation problem in the shore-based power supply system and improve the effective absorption level of new energy. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of the broadband oscillation suppression method in an embodiment of the present invention;

[0020] Figure 2 This is a schematic diagram of the random forest algorithm in an embodiment of the present invention;

[0021] Figure 3 This is a schematic diagram of the first neural network structure in an embodiment of the present invention;

[0022] Figure 4 This is a schematic diagram of the controller structure combining the first neural network and the complex frequency domain impedance stabilizer in an embodiment of the present invention;

[0023] Figure 5 This is a schematic diagram of the ultra-high ratio shore-based power supply system platform structure in an embodiment of the present invention;

[0024] Figure 6This is a block diagram of the broadband oscillation suppression system in an embodiment of the present invention;

[0025] Figure 7 This is a schematic diagram illustrating the effect of existing broadband suppression strategies;

[0026] Figure 8 This is a schematic diagram illustrating the effect of the broadband suppression strategy in an embodiment of the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] This embodiment provides a broadband oscillation suppression method for shore-based power supply systems. It addresses the problem that in practice, the suppression effect is constrained by multiple uncertainties in the complex frequency domain impedance, making the selection of suppression strategies difficult and inaccurate. This method utilizes system oscillation monitoring technology, combining historical oscillation information and actual observation results to guide the equipment to adaptively suppress oscillations, thereby achieving a strategy of proactively controlling the system before oscillations occur.

[0029] Please see Figure 1 As shown, a broadband oscillation suppression method for shore-based power supply systems includes the following steps:

[0030] Step S101: First, obtain the frequency and corresponding characteristic data of the broadband oscillation generated by the shore-based power supply system, and then input them into the classifier model as a training set so that the classifier model can generate a decision tree classification sub-model.

[0031] The steps to obtain the frequency and corresponding characteristic data of the broadband oscillation generated by the shore-based power supply system are as follows:

[0032] 1) Inject current into the shore-based power supply system and collect the current response signal of the system. The specific operation can be carried out by using a direct-connected oscillation active prediction device for current injection, such as the current injection device described in "Integrated Governance Technology and Application of Direct-Connected High Voltage Distribution Network Power Quality" [J] (Zhang Dinghua et al., Journal of Electrical Engineering, 2013, 28(7):243-251.).

[0033] 2) Analyze the collected current response signals to obtain the characteristic data of the shore power supply system. The characteristic data includes the port external characteristics of the shore power supply system and the complex frequency domain characteristics of the external equivalent power grid. The port external characteristics represent the equivalent impedance values ​​of the shore power supply system in different frequency domains. The complex frequency domain characteristics of the external equivalent power grid include the voltage, current and power readings of each node of the shore power supply system. The method for processing the current response signal already exists in the existing technology, such as the processing methods mentioned in "A Novel Online System Impedance Measurement Method" [J] (Tong Liqing et al., Journal of Electrical Engineering, 2008, 23(1):137-142.) and "Piecewise Binary Tree Method for Impedance Measurement of Power Electronic Systems" [J] (Yue Xiaolong et al., Journal of Electrical Engineering, 2015, 30(24):76-83.).

[0034] Step S102: Input the frequency and amplitude of the broadband oscillation generated by the actual shore-based power supply system and the equivalent external characteristics of the power supply system into the classifier model, so that the classifier model outputs a broadband oscillation suppression strategy based on the classification conditions.

[0035] A windowed Fourier transform (ZoomFFT)-based detection method was used to rapidly acquire the broadband oscillation frequency and amplitude of a practical shore-based power supply system. The detailed steps are as follows:

[0036] 1) Collect electrical parameters of each node in the shore-based power supply system and convert the collected electrical parameters into frequency domain data; electrical parameters include current, voltage and power, etc.; usually, a daily electrical parameter dataset is formed; since the collected electrical parameters are time domain data, in order to use windowed Fourier transform for processing, the time domain data needs to be converted into frequency domain data;

[0037] 2) Divide the frequency domain data into different frequency bands and decompose them using the windowed Fourier method to obtain the amplitude of different frequency bands; different frequency bands include low frequency, fundamental frequency (including fundamental frequency and interharmonics near it) and high frequency (including higher harmonics and higher frequency interharmonics). The windowed Fourier method is an existing technology and will not be described in detail here, such as the harmonic detection algorithm based on windowed interpolation FFT [J] (Luo Zhaotong, Chen Kailun, Chen Qiong. Information Technology, 2018, 42(12):65-69.), or the wind farm adaptive oscillation suppression strategy based on windowed FFT [J] (Miao Shuo, Li Qinan, Zha Kunpeng, Li Lanfang, Cao Jianchun, Zhang Fan. China Electric Power, 2022, 55(10):112-123.).

[0038] It is worth noting that the collected electrical parameter signals are usually affected by interference signals. To eliminate interference signals, different digital filters and dedicated algorithms are used in different frequency bands. The denoising method is an existing technology, such as the method disclosed in "Application of Variable Step Size Adaptive Algorithm in Harmonic Detection of Active Filter" [J] (Li Hui et al., Proceedings of the CSEE, 2006, 26(9): 99-103.), which will not be elaborated further.

[0039] 3) Determine whether the amplitude of a frequency band exceeds its threshold based on the amplitude threshold of different frequency bands. If it does, determine that the shore-based power supply system is generating broadband oscillation, and record the frequency and amplitude of the corresponding frequency band.

[0040] The above-mentioned broadband power grid signal measurement method based on windowed interpolation and ZoomFFT can adaptively measure high, medium and low frequency power grid electrical parameter signals, improve the processing efficiency of broadband measurement equipment, and achieve adaptive, fast and accurate measurement of harmonics and interharmonics.

[0041] The classifier model used in steps S101 and S102 is the random forest classifier model.

[0042] The principle of the Bagging classification algorithm is to improve the independence of each base model by randomly constructing training samples and randomly selecting features. Random sampling with replacement is performed on the original training set to obtain M smaller training sets, and M models are trained on these sets. Then, model ensemble is completed through voting.

[0043] The key to the Random Forest algorithm is how to select the number of sample attributes and the attribute split points in the decision trees of the forest. The Random Forest algorithm trains by randomly selecting n samples from N sample sets, choosing m attributes from M attributes of each sample, and calculating their Gini coefficients to obtain branch nodes with the highest possible purity. After generating a random forest containing a specific number of classification trees, the final decision is obtained by voting based on the judgment results of each classification tree.

[0044] The structure of the random forest algorithm is as follows: Figure 2 As shown, the production of a classification tree includes the following steps:

[0045] Use the Gini index to select the optimal feature. For a given sample set D, its Gini index is:

[0046]

[0047] In the formula, C k It is the subset of samples in D that belong to the k-th class, where K is the number of classes.

[0048] Given characteristic A, the Gini index of set D is defined as:

[0049]

[0050] Input: Training dataset D, and the condition for stopping computation;

[0051] Output: Decision tree;

[0052] Based on the training dataset, starting from the root node, recursively perform the following operations on each node to construct a binary decision tree;

[0053] 1) Let the training dataset of the nodes be D. Calculate the Gini index of the existing features on this dataset. For each feature A, for each possible value a, divide D into two parts, D1 and D2, based on whether the test of A = a for the sample point is "yes" or "no". Calculate the Gini index when A = a.

[0054] 2) Among all possible features A and their possible split points a, select the feature with the smallest Gini index and its corresponding split point as the optimal feature and optimal split point. Based on the optimal feature and optimal split point, generate two child nodes from the current node, and distribute the training dataset to the two child nodes according to the features.

[0055] 3) Recursively call (1) and (2) on the two child nodes until the stopping condition is met.

[0056] 4) Generate a decision tree.

[0057] The algorithm stops calculating when the number of samples in a node is less than a predetermined threshold, or the Gini index of the sample set is less than a predetermined threshold (the samples are basically of the same class), or there are no more features.

[0058] Random forest is a classifier algorithm that uses a large number of decision trees trained to segment samples based on different features to classify a dataset, and finally votes the classification results to form the final judgment.

[0059] The random forest classifier model ultimately outputs a broadband oscillation suppression strategy, including the amplitude of the injected current, controller parameters, damping of the shore-based power supply system, and the resistance and reactance values ​​of the complex frequency domain impedance, etc.

[0060] In summary, through the above steps, the random forest classifier model screens and determines the zero-pole cases of the complex frequency domain impedance involved in suppression.

[0061] Step S103: Input the broadband oscillation suppression strategy and oscillation parameters into the first neural network, and output the broadband oscillation suppression parameters to suppress the broadband oscillation of the shore-based power supply system.

[0062] The structure of the first neural network is as follows: Figure 3As shown, the first neural network is an RBF neural network consisting of an input layer i, a hidden layer j, and an output layer l connected in sequence. It should be understood that the first neural network is pre-trained.

[0063] This embodiment uses a first neural network to adaptively adjust the control parameters of the device's complex frequency domain impedance stabilizer, namely the specific resistance and reactance of the complex frequency domain impedance. The training process of the RBF neural network is as follows:

[0064] 1) Determine the parameters

[0065] 1. Determine the input vector X

[0066] X = [x1, x2, ..., x n ] T , where n is the number of input layer units.

[0067] 2. Determine the output vector Y and the desired output vector O.

[0068] Y = [y1, y2, ... y q ] T q is the number of output layer units.

[0069] O = [o1, o2, ..., o q ] T

[0070] 3. Initialize the connection weights from the hidden layer to the output layer.

[0071] W k =[w k1 ,w k2 ,...,w kp ] T ,(k=1,2,...,q), where p is the number of hidden layer units and q is the number of output layer units.

[0072] 4. Initialize the central parameters of each neuron in the hidden layer.

[0073] C j =[c j1 ,c j2 ,...,c jn ] T , where j is the hidden layer number and n is the input layer number.

[0074] The specific formula for initializing the central parameters is as follows:

[0075] Where mini is the minimum value of all input information of the i-th feature in the training set, and maxi is the maximum value of all input information of the i-th feature in the training set.

[0076] 5. Initialize the width vector

[0077] D j =[d j1 ,d j2 ,...d jn The width vector affects the range of a neuron's action on input information: the smaller the width, the narrower the shape of the action function of the corresponding hidden layer neuron, and the smaller the response of information near the center of other neurons at that neuron. The specific initialization formula is:

[0078] Where d f This is the width adjustment coefficient, and its value is less than 1.

[0079] 2) Calculate the output value z of the j-th neuron in the hidden layer. j

[0080]

[0081] 3) Calculate the output of the output layer neurons

[0082] Y = [y1, y2, ..., y q ] T ,in

[0083] 4) Iterative calculation of weight parameters

[0084] The evaluation function for the RBF neural network is:

[0085] Among them, o lk y represents the expected output value of the k-th output neuron with the l-th input sample. lk This represents the network output value of the k-th output neuron when it receives the l-th input sample.

[0086] Supervised modification of the hidden layer's center parameters, width vector, and weights from the hidden layer to the output layer is achieved using gradient descent.

[0087]

[0088]

[0089]

[0090] In the formula, η is the step size, and an inertia term is added to enable the algorithm to converge quickly.

[0091] The workflow of an RBF neural network is as follows:

[0092] The input parameters of the input layer are denoted as x1, x2, and x3, and the j-th output value of the n-th layer is denoted as O. jIf (n), then the specific expression is as follows:

[0093]

[0094] The activation function of the second hidden layer is a Gaussian function of the distance from the input value to the center parameter of the hidden layer, as follows:

[0095]

[0096]

[0097]

[0098] In the formula, j is the hidden layer number.

[0099] The inputs and outputs of the third-layer network output layer are as follows. The two nodes of the output layer correspond to the equivalent resistance and reactance R of the complex frequency domain impedance controller, respectively. i X i .

[0100]

[0101] The controller structure using a combination of a first neural network and a complex frequency domain impedance stabilizer is as follows: Figure 4 As shown, r(k) is the system input, y(k) is the system output, e(k) and e(k-1) are the system error and the rate of change of the error, respectively. The specific resistance and reactance value of the complex frequency domain impedance is output as u(k), and the change in the output of the complex frequency domain impedance stabilizer is Δu(k).

[0102] The relationship between system input and output is as follows:

[0103]

[0104] The error function can be expressed as:

[0105]

[0106] This step determines whether to engage the adaptive oscillation suppression stage, allowing for online updating of filter parameters based on frequency identification results. It also extracts the corresponding oscillation components and inputs them into the damping control stage, adaptively calculating the system's equivalent impedance to meet oscillation suppression requirements. This generates a current reference signal based on the oscillation components, controlling the converter's injected current, enhancing system damping, and achieving adaptive suppression of wideband oscillations.

[0107] In summary, the method of this invention inputs the historically obtained port external characteristics, the equivalent complex frequency domain characteristics of the external power grid, and the real-time monitoring broadband oscillation frequency and amplitude obtained by low-frequency, fundamental frequency, and high-frequency Fourier decomposition under these characteristics into a random forest classification system to train the decision tree sub-classifier in the classifier model, so as to classify the specific system external characteristics and obtain the oscillation suppression strategy applicable to the equivalent complex frequency domain impedance condition of the system.

[0108] This invention also relates to a computer-readable storage medium storing computer program instructions. When the computer program instructions are executed by a processor, the broadband oscillation suppression method described in the above embodiments is implemented.

[0109] Furthermore, embodiments of the present invention also provide a wideband oscillation suppression system, comprising: a computer-readable storage medium as described in the above embodiments.

[0110] like Figure 5 As shown, this is an ultra-high ratio shore-based power supply system platform applying the broadband oscillation suppression system of this invention, such as... Figure 6 The diagram shown is a structural block diagram of a broadband oscillation suppression system according to an embodiment of the present invention.

[0111] like Figure 7 As shown, the system becomes unstable when a 20A current is injected into the shore-based power supply system. At 0.2s, the traditional stabilization control strategy is insufficient to stabilize the system. Figure 8 As shown in the figure, the oscillation suppression after instability is performed using the technology of this embodiment 0.2s after the system becomes unstable. The results shown in the figure can be seen that the technology of this embodiment can achieve oscillation suppression that traditional stability control strategies cannot achieve, which verifies that the embodiment of this invention has the superiority over the prior art.

[0112] In summary, this embodiment of the invention inputs actual monitoring data into a random forest classifier model for clustering calculations. Based on the classification results, the number of zeros and poles of the complex frequency domain impedance involved in suppression is adaptively selected, resulting in a suppression control strategy suitable for the current broadband oscillation situation. Compared with traditional broadband suppression strategies, this method can more accurately adopt suppression strategies for oscillations in different frequency bands and adaptively select the most suitable controller suppression parameters to achieve the best broadband oscillation suppression effect.

[0113] By eliminating oscillation problems in shore power supply systems, the embodiments of the present invention can improve the effective absorption level of shore power generation system projects, significantly reduce the project delays caused by system oscillations, and improve the stability and reliability of shore power supply systems based on new energy sources.

[0114] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0115] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0116] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0117] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0118] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0119] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A broadband oscillation suppression method for shore-based power supply systems, characterized in that, Includes the following steps: The frequency and corresponding characteristic data of the broadband oscillation generated by the shore-based power supply system are obtained and input into the classifier model as a training set so that the classifier model can generate a decision tree classification sub-model. Input the frequency and amplitude of the broadband oscillations generated by the actual shore-based power supply system, as well as the equivalent external characteristics of the power supply system, into the classifier model after generating the decision tree classification sub-model, so that the classifier model outputs a broadband oscillation suppression strategy based on the classification conditions; The broadband oscillation suppression strategy is input into the first neural network, which outputs broadband oscillation suppression parameters. The broadband oscillation of the shore-based power supply system is suppressed by the broadband oscillation suppression parameters.

2. The broadband oscillation suppression method for shore-based power supply systems according to claim 1, characterized in that, The steps for acquiring the frequency and corresponding characteristic data of the broadband oscillation generated by the shore-based power supply system include: Inject current into the shore-based power supply system and collect the current response signal of the shore-based power supply system; The current response signal is processed to obtain the characteristic data of the shore power supply system. The characteristic data includes the port external characteristics of the shore power supply system and the complex frequency domain characteristics of the external equivalent power grid.

3. The broadband oscillation suppression method for a shore-based power supply system according to claim 1, characterized in that, The steps for obtaining the frequency and amplitude of the broadband oscillation generated by the actual shore-based power supply system include: Collect electrical quantities at each node of the shore-based power supply system and convert the collected electrical quantities into frequency domain quantities; After dividing the frequency domain quantity into different frequency bands, the amplitude of each frequency band is obtained by decomposing it using the windowed Fourier method. According to the amplitude threshold of different frequency bands, it is determined whether the amplitude of the frequency band exceeds the amplitude threshold of the frequency band. If it does, it is determined that the shore power supply system generates broadband oscillation, and the frequency and amplitude corresponding to the frequency band are recorded.

4. The broadband oscillation suppression method for a shore-based power supply system according to claim 1, characterized in that, The classifier model is a random forest classifier model.

5. A broadband oscillation suppression method for a shore-based power supply system according to claim 1, characterized in that, The first neural network is an RBF neural network consisting of an input layer, a hidden layer, and an output layer connected in sequence.

6. A broadband oscillation suppression method for a shore-based power supply system according to claim 1, characterized in that, The wideband oscillation suppression parameters are the equivalent resistance and reactance values ​​of the complex frequency domain impedance controller.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, implement the broadband oscillation suppression method as described in any one of claims 1 to 6.

8. A broadband oscillation suppression system for shore-based power supply systems, characterized in that, include: The training module is used to train the classifier model and generate decision tree classification sub-models; The acquisition module is used to acquire the frequency and amplitude of the broadband oscillation generated by the actual shore-based power supply system, the equivalent external characteristics of the power supply system, and input them into the trained classifier model to obtain the broadband oscillation suppression strategy, and then input it into the first neural network. The neural network module stores the parameters for running the first neural network and outputting wideband oscillation suppression parameters. The suppression module is used to suppress broadband oscillations in the shore-based power supply system through broadband oscillation suppression parameters.

9. A broadband oscillation suppressor for a shore-based power supply system, storing a program for running the broadband oscillation suppression system for a shore-based power supply system as described in claim 8.

Citation Information

Patent Citations

  • Broadband oscillation suppression method, medium and system

    CN116599033A