Method for training and prediction of plasma oscillation risk model for electric thrusters

Abstract

The application provides a training and prediction method of an electric thruster plasma oscillation risk prediction model, and belongs to the technical field of data processing. The method obtains historical oscillation data sets under various working conditions of an electric thruster; extracts target oscillation components in the historical oscillation data sets under various working conditions; determines coupling characteristic data between the target oscillation components under various working conditions based on a bispectrum algorithm; inputs the historical oscillation data sets under various working conditions and the corresponding coupling characteristic data into a pre-constructed oscillation risk prediction model for iterative training, and predicts the oscillation risk based on the trained model. The application trains an oscillation risk prediction model, which takes the coupling characteristic data between oscillation components as training samples, can not only identify the oscillation mode between main oscillation components, but also further identify the oscillation intensity of the mode and the most serious oscillation area, and improves the precision of oscillation risk prediction.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and more specifically, to a training and prediction method for a plasma oscillation risk prediction model for electric thrusters. Background Technology

[0002] An electric propulsion (EP) is a type of aerospace propulsion system. EP uses electrical energy to ionize propellant and then accelerates these charged particles through electromagnetic or electrostatic fields, expelling them at extremely high speeds. The resulting reaction force is the thrust. The plasma signal within an EP often oscillates during operation. If the oscillation amplitude is too large or the frequency is inappropriate, it can affect the EP's efficiency or even lead to equipment malfunction. Therefore, the analysis of the oscillation signal is crucial.

[0003] The plasma generated by the electric thruster exhibits discharge oscillations at different frequencies (low-frequency oscillations below 100 kHz and high-frequency oscillations above 100 kHz), in different directions (axial breathing oscillations and circumferential spoke oscillations), and based on different principles (ionization oscillations and gradient drift-induced oscillations). These oscillation components often coexist when the electric thruster is operating.

[0004] Existing methods for processing oscillation parameters in electric propulsion plasmas often only extract different oscillation components independently when analyzing oscillation signals. However, they cannot identify and extract the spatiotemporal competition and coupling relationships between various oscillation components, resulting in low accuracy in oscillation signal analysis. In addition, there are many types of oscillation components, making manual processing extremely inefficient. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a training and prediction method for an electric thruster plasma oscillation risk prediction model, so as to improve the efficiency and accuracy of electric thruster plasma oscillation risk prediction.

[0006] Firstly, a training method for a plasma oscillation risk prediction model for electric thrusters is provided, the method comprising:

[0007] The historical oscillation dataset of the electric thruster under various operating conditions is obtained. The historical oscillation dataset consists of spatial coordinate data of each measurement point and time-series oscillation signals.

[0008] Extract the target oscillation component from the historical oscillation dataset under various operating conditions; the target oscillation component is the oscillation component within a preset frequency range;

[0009] Based on the bispectral algorithm, the coupling characteristic data between the target oscillation components under various working conditions are determined. The coupling characteristic data includes the coupling strength, the spatial distribution pattern of the coupling strength, and the peak value of the coupling strength.

[0010] Historical oscillation datasets under various operating conditions and their corresponding coupling feature data are input into a pre-built oscillation risk prediction model for iterative training until the preset training stopping condition is reached. Each iteration of training outputs the oscillation mode, the oscillation intensity of the oscillation mode, and the spatial location information of the maximum oscillation intensity.

[0011] Optionally, before extracting the target oscillation component from the historical oscillation dataset under various operating conditions, the method further includes:

[0012] The spatial coordinate data in the historical oscillation dataset is reduced in dimensionality to obtain one-dimensional spatial coordinate data. The two-dimensional historical oscillation dataset is composed of one-dimensional spatial coordinate data and time-series oscillation signals.

[0013] Correspondingly, the target oscillation components extracted from historical oscillation datasets under various operating conditions include:

[0014] Extract the target oscillation component from the two-dimensional historical oscillation dataset under various operating conditions.

[0015] Optionally, the target oscillation components extracted from the two-dimensional historical oscillation dataset under various operating conditions include:

[0016] The target oscillation component is obtained by inputting a two-dimensional historical oscillation dataset under various operating conditions into a pre-trained neural network model. The neural network model is at least one of a convolutional neural network model and a deep neural network model.

[0017] Optionally, the training process of a neural network model includes:

[0018] Acquire data on multiple historical oscillation components;

[0019] The spatiotemporal modes of each oscillation component are extracted using the intrinsic orthogonal decomposition algorithm or the dynamic mode decomposition algorithm, and each spatiotemporal mode represents a type of oscillation component.

[0020] Historical oscillation component data and corresponding spatiotemporal modes are input into the neural network model for iterative training until the preset iteration conditions are met.

[0021] Optionally, the coupling characteristic data between target oscillation components under various operating conditions determined based on the bispectral algorithm include:

[0022] Based on the Fast Fourier Transform (FFT) formula, the time-series oscillation signals in historical oscillation datasets under various operating conditions are converted into frequency-domain oscillation signals, yielding the spectral distribution of the oscillation signals. The FFT formula is as follows:

[0023]

[0024] in, Representing a spatial point The spectral distribution of the oscillation signal at that location; Indicates measurement point The timing oscillation signal at the location; Indicates frequency; Indicates time; Indicates the number of points in space; Indicates time Quantity; Indicates an exponential factor;

[0025] The spectral distribution of the target oscillation component is determined based on the spectral distribution of the oscillation signal and the frequency range of the target oscillation component.

[0026] The bispectral values ​​among multiple target oscillation components are calculated based on the bispectral calculation formula and the spectral distribution of the target oscillation components; the bispectral calculation formula is as follows:

[0027]

[0028] in, It is a bispectral value; and It is an oscillating signal at a frequency and The frequency domain components; It is an oscillating signal at a frequency and The conjugate components; Indicates time Quantity;

[0029] Coupling characteristic data are determined based on the bispectral values ​​between the target oscillation components.

[0030] Optionally, determining coupling feature data based on bispectral values ​​between target oscillation components includes:

[0031] Calculate the magnitude of the bispectral values ​​between the target oscillation components to obtain the coupling strength between them;

[0032] The spatial distribution map of the coupling strength between the target oscillation components is obtained by performing an inverse spatial dimension transformation based on the coupling strength between the target oscillation components.

[0033] The coupling feature data is extracted from the spatial distribution map based on the pre-trained convolutional neural network model. The coupling feature data includes at least the spatial distribution pattern of coupling strength and the peak value of coupling strength.

[0034] Secondly, a method for predicting plasma oscillation risk in electric thrusters is provided, the method comprising:

[0035] Acquire spatial coordinate data and timing oscillation signals at various measurement points in the electric thruster under the current operating conditions;

[0036] Spatial coordinate data and time-series oscillation signals are input into a pre-trained oscillation risk prediction model for prediction, and the oscillation mode, the intensity of the oscillation mode, and the spatial location information of the maximum oscillation intensity are obtained.

[0037] Thirdly, a device for predicting the risk of plasma oscillation in an electric thruster is provided, the device comprising:

[0038] The acquisition unit is used to acquire the spatial coordinate data and timing oscillation signal of each measurement point in the electric thruster under the current operating conditions.

[0039] The prediction unit is used to input spatial coordinate data and time-series oscillation signals into a pre-trained oscillation risk prediction model for prediction, and to obtain the oscillation mode, the intensity of the oscillation mode, and the spatial location information of the maximum oscillation intensity.

[0040] Fourthly, an electronic device is provided, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

[0041] Memory, used to store computer programs;

[0042] A processor, when executing a program stored in memory, implements the steps of the method described in either the first or second aspect.

[0043] Fifthly, a computer-readable storage medium is provided, wherein a computer program is stored therein, and when executed by a processor, the computer program implements the steps of the method described in either the first or second aspect.

[0044] This invention provides a training and prediction method for an electric thruster plasma oscillation risk prediction model. The method involves acquiring historical oscillation datasets under various operating conditions of the electric thruster; extracting target oscillation components from these datasets; determining coupling characteristic data between the target oscillation components under various operating conditions based on a bispectral algorithm; inputting the historical oscillation datasets and their corresponding coupling characteristic data under various operating conditions into a pre-constructed oscillation risk prediction model for iterative training; and predicting oscillation risk based on the trained model. This invention trains an oscillation risk prediction model that uses the coupling characteristic data between various oscillation components as training samples. This model can not only identify the oscillation modes between the main oscillation components but also further identify the oscillation intensity of these modes and the region with the most severe oscillation, improving the accuracy of oscillation component identification and thus improving the accuracy of oscillation risk prediction.

[0045] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 A flowchart is shown below illustrating a training method for an electric thruster plasma oscillation risk prediction model provided in an embodiment of the present invention.

[0048] Figure 2 A flowchart of a method for predicting the risk of plasma oscillation in an electric thruster provided by an embodiment of the present invention is shown;

[0049] Figure 3 This diagram illustrates the structure of an electric thruster plasma oscillation risk prediction device provided in an embodiment of the present invention.

[0050] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present invention is shown. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0052] Existing methods for processing oscillation parameters in electric propulsion plasma often only extract different oscillation components independently when analyzing oscillation signals, but fail to identify and extract the spatiotemporal competition and coupling relationships between various oscillation components, resulting in low accuracy in oscillation signal analysis. Furthermore, the variety of oscillation components makes manual processing extremely inefficient. Therefore, this invention provides a training method and apparatus for an oscillation risk prediction model, which is described below through embodiments.

[0053] This invention provides a training method for a plasma oscillation risk prediction model for electric thrusters, such as... Figure 1 As shown, the method includes the following steps:

[0054] Step S101: Obtain historical oscillation datasets of the electric thruster under various operating conditions.

[0055] In this step, the operating condition refers to the operating conditions of the electric thruster, such as anode voltage, anode current, anode working fluid flow rate, excitation current, cathode heating current, cathode contact electrode voltage, and cathode contact electrode current.

[0056] The historical oscillation dataset consists of spatial coordinate data of each measurement point and time-series oscillation signals.

[0057] In one example, timing oscillation signals can be acquired by probes at each measurement point, and the spatial coordinate data of each measurement point is based on spatial coordinate data in the world coordinate system, and the spatial coordinate data of each measurement point is pre-configured. The spatial coordinate data of each measurement point is bound to the timing oscillation data.

[0058] To facilitate subsequent data processing, the spatial coordinate data is spatially reduced in dimensionality to simplify the data structure.

[0059] In one feasible implementation, the spatial coordinate data in the historical oscillation dataset is dimensionality-reduced to obtain one-dimensional spatial coordinate data, and the one-dimensional spatial coordinate data and the time-series oscillation signal constitute a two-dimensional historical oscillation dataset.

[0060] Specifically, dimensionality reduction can be achieved using the following mapping formula:

[0061] (1);

[0062] In the formula, Q represents the measured oscillation signal; These are the spatial three-dimensional coordinates of the measurement point, where... ; ; ; , , These represent the number of measurement points in the three dimensions. It is a time-series oscillating signal sequence, where, .

[0063] Therefore, the spatiotemporal sequence of the plasma oscillation parameters of the electric thruster can be transformed into a two-dimensional oscillation array:

[0064] (2);

[0065] In the formula, Represents a two-dimensional oscillation array; , is a matrix number of rows; The number of columns; Represented as an arbitrary matrix The total number of spatial points; for Oscillation parameter frames at all spatial points at any given time.

[0066] Using the above method, a two-dimensional historical oscillation dataset of electric thruster plasma can be obtained.

[0067] Step S102: Extract the target oscillation component from the historical oscillation dataset under various operating conditions.

[0068] In this step, the target oscillation component is the main oscillation component, which is generally the oscillation component with obvious oscillation. Some oscillation components with little oscillation or little impact on the operation of electric thruster can be filtered out. This can reduce the amount and complexity of data processing, and help to quickly capture the periodic nonlinear behavior of the oscillation signal, thereby enhancing the efficiency of identifying the main oscillation component.

[0069] In one feasible implementation, the target oscillation component can be determined by setting a preset frequency range.

[0070] Furthermore, continuing from the previous example, the target oscillation component is extracted based on the two-dimensional historical oscillation dataset after spatial dimensionality reduction.

[0071] Step S103: Determine the coupling characteristic data between target oscillation components under various operating conditions based on the bispectral algorithm.

[0072] In this step, the coupling characteristic data includes coupling strength, spatial distribution pattern of coupling strength, and peak value of coupling strength.

[0073] The bispectral algorithm is an advanced signal processing technique for nonlinear system analysis that reveals the interaction between linear and nonlinear components in a signal.

[0074] Step S104: Input the historical oscillation datasets under various operating conditions and their corresponding coupling feature data into the pre-built oscillation risk prediction model for iterative training until the preset training stopping condition is reached; each iteration of training will output the oscillation mode, the oscillation intensity of the oscillation mode, and the spatial location information of the maximum oscillation intensity.

[0075] In this step, a preset training stopping condition is established, such as reaching a preset number of iterations or the loss function reaching a local minimum. The loss function for this model can be the cross-entropy function.

[0076] The oscillation risk prediction model can be trained using traditional machine learning methods such as support vector machines, decision trees, and random forests, or it can be trained using deep learning methods for classification.

[0077] Among them, there are positive and negative coupling relationships between the two oscillating signals. Positive coupling means that the frequencies and phases of the two oscillating signals are synchronized, which belongs to the cooperative oscillation mode; negative coupling is the opposite, where one oscillating signal will suppress the other oscillating signal, which belongs to the competitive oscillation mode.

[0078] The oscillation mode can be analyzed by the coupling strength of the oscillation signal, and then the oscillation mode of the oscillation signal can be labeled. The intensity of the oscillation mode can also be labeled, and the labeling category can be such as strong oscillation coupling, strong oscillation competition, stationary, etc. The spatial location information of the maximum oscillation intensity can also be labeled, thus obtaining the label dataset of each oscillation signal in the historical oscillation dataset. The label dataset and the historical oscillation dataset are input together into the oscillation risk prediction model for training.

[0079] Based on feedback from actual oscillation identification, the model can be further optimized. For example, if coupling problems in certain regions are not accurately identified, the amount of training data can be increased, or the feature extraction method can be improved.

[0080] After training, the model can automatically classify oscillation patterns under different operating conditions and identify the most severe oscillation regions.

[0081] This invention trains an oscillation risk prediction model that uses the coupling feature data between various oscillation components as training samples. This model can not only identify the oscillation patterns between the main oscillation components, but also further identify the oscillation intensity of the pattern and the region with the most severe oscillation, thereby improving the accuracy of oscillation component identification and thus improving the accuracy of oscillation risk prediction.

[0082] Traditional methods for extracting the main oscillatory components typically involve modal analysis using Proper Orthogonal Decomposition (POD) or Dynamic Mode Decomposition (DMD). However, these two methods require about one or two days to obtain analysis results, making them very inefficient and inaccurate.

[0083] Therefore, based on the above embodiments, the target oscillation components extracted from the two-dimensional historical oscillation dataset under various operating conditions include:

[0084] Step S1021: Input the two-dimensional historical oscillation dataset under various working conditions into the pre-trained neural network model to identify the target oscillation component.

[0085] In this embodiment of the invention, the neural network model is at least one of a convolutional neural network model and a deep neural network model.

[0086] Deep neural networks consist of encoders and decoders. The encoder is combined with the spatiotemporal mode coefficients extracted by POD or DMD to input into the encoder for noise reduction, and the decoder reconstructs the target oscillation components.

[0087] The embodiments of the present invention employ a neural network model to extract the target oscillation component, which greatly improves the efficiency and accuracy of extracting the target oscillation component.

[0088] Based on the above embodiments, the training process of the neural network model includes:

[0089] Step S1021A: Obtain data on various historical oscillation components.

[0090] Step S1021B: Extract the spatiotemporal modes of each oscillation component using the intrinsic orthogonal decomposition algorithm or the dynamic mode decomposition algorithm. Each spatiotemporal mode represents a type of oscillation component.

[0091] Step S1021C: Input the historical oscillation component data and the corresponding spatiotemporal modes into the neural network model for iterative training until the preset iteration conditions are met.

[0092] In this step, spatiotemporal modes are used as labels for the oscillatory component data during training. The preset iteration condition is reaching a preset number of iterations. The loss function of this model can be the cross-entropy function.

[0093] This invention, by introducing a neural network model, can automatically identify and classify different oscillation components. Compared with manual methods, it reduces the need for human intervention, improves the efficiency and accuracy of identification, and can handle large-scale datasets, avoiding the subjective bias and inconsistency of manual analysis.

[0094] Based on the above embodiments, the coupling characteristic data between target oscillation components under various operating conditions determined by the bispectral algorithm include:

[0095] Step S1031: Based on the Fast Fourier Transform formula, convert the time-series oscillation signals in the historical oscillation dataset under various operating conditions into frequency-domain oscillation signals to obtain the spectral distribution of the oscillation signals.

[0096] The formula for Fast Fourier Transform is:

[0097] (3);

[0098] in, Representing a spatial point The spectral distribution of the oscillation signal at that location; Indicates measurement point The timing oscillation signal at the location; Indicates frequency; Indicates time; Indicates the number of points in space; Indicates time Quantity; This represents the exponential factor.

[0099] Step S1032: Determine the spectral distribution of the target oscillation component based on the spectral distribution of the oscillation signal and the frequency range of the target oscillation component.

[0100] In this step, the spectrum distribution refers to the change of parameters such as the energy of the oscillation signal with frequency. In the example above, when extracting the target oscillation component, the frequency range of the target oscillation component can be known. Based on the frequency range, the spectrum distribution of the target oscillation component can be obtained from the spectrum distribution obtained in step S1031.

[0101] Because of the large frequency range, analyzing the coupling relationship of each frequency band individually would be extremely labor-intensive. Therefore, by selecting the frequency band of the target oscillation component for subsequent coupling relationship analysis, the workload is greatly reduced and the efficiency of oscillation component identification and analysis is improved.

[0102] Step S1033: Calculate the bispectral values ​​among multiple target oscillation components based on the bispectral calculation formula and the spectral distribution of the target oscillation components; the bispectral calculation formula is:

[0103] (4);

[0104] in, It is a bispectral value; and It is an oscillating signal at a frequency and The frequency domain components; It is an oscillating signal at a frequency and The conjugate components; Indicates time Quantity;

[0105] Step S1034: Determine coupling feature data based on the bispectral values ​​between the target oscillation components.

[0106] In one feasible implementation, determining coupling characteristic data based on bispectral values ​​between target oscillatory components includes:

[0107] Step S1034A: Calculate the modulus of the bispectral values ​​between the target oscillation components to obtain the coupling strength between the target oscillation components.

[0108] Continuing from the previous example, the modulus of this bispectral value is expressed as: The magnitude of the bispectral value represents the coupling strength.

[0109] Step S1034B: Perform inverse spatial dimension transformation based on the coupling strength between the target oscillation components to obtain the spatial distribution map of the coupling strength between the target oscillation components.

[0110] In one example, the coupling strength at the same dual frequency at all spatial points is restored to the spatial distribution of the coupling relationships of each target oscillation component by the inverse spatial dimension reduction transform in step S101:

[0111] (5);

[0112] In the formula, Spatial distribution data representing the coupling relationships of the oscillation components of each target; These are the spatial three-dimensional coordinates of the measurement point, where... ; ; ; , , These represent the number of measurement points in each of the three dimensions.

[0113] Step S1034C: Extract coupling feature data from the spatial distribution map based on the pre-trained convolutional neural network model. The coupling feature data includes at least the spatial distribution pattern of coupling strength and the peak value of coupling strength.

[0114] This invention automatically extracts coupling features between vibrational components by using spatial distribution maps combined with neural network models, significantly reducing the complexity of manual analysis. In traditional analysis methods, manually interpreting complex spatial distribution maps requires significant experience and time, resulting in low efficiency. This invention, however, can automatically extract important coupling features from large-scale data, improving the efficiency and accuracy of analyzing the relationships between oscillating components.

[0115] It should also be noted that when analyzing a small amount of oscillation data, steps S101-S103 can be used to manually analyze the coupling relationship and other data between the oscillation components.

[0116] Based on the oscillation risk prediction model trained in the above embodiments, this invention provides a method for predicting the plasma oscillation risk of electric thrusters, such as... Figure 2 As shown, the method includes the following steps:

[0117] Step S201: Obtain the spatial coordinate data and timing oscillation signal of each measurement point in the electric thruster under the current operating conditions.

[0118] Step S202: Input the spatial coordinate data and the time-series oscillation signal into the pre-trained oscillation risk prediction model for prediction, and obtain the oscillation mode, the intensity of the oscillation mode, and the spatial location information of the maximum oscillation intensity.

[0119] This invention can quickly predict the oscillation mode, oscillation intensity, and most severe oscillation region under different operating conditions, and provides timely early warning and optimization suggestions in actual operation, helping to better adjust the operating conditions and avoid potential system instability.

[0120] Based on the same inventive concept, a device for predicting the risk of plasma oscillation in electric thrusters is provided, such as... Figure 3 As shown, the device includes:

[0121] The acquisition unit 301 is used to acquire the spatial coordinate data and timing oscillation signal of each measurement point in the electric thruster under the current operating conditions.

[0122] The prediction unit 302 is used to input spatial coordinate data and time-series oscillation signals into a pre-trained oscillation risk prediction model for prediction, and obtain the oscillation mode, the intensity of the oscillation mode, and the spatial location information of the maximum oscillation intensity.

[0123] Based on the same technical concept, embodiments of the present invention also provide an electronic device, such as... Figure 4 As shown, it includes a processor 401, a communication interface 402, a memory 403, and a communication bus 404, wherein the processor 401, the communication interface 402, and the memory 403 communicate with each other through the communication bus 404.

[0124] Memory 403 is used to store computer programs;

[0125] The processor 401, when executing the program stored in the memory 403, implements the steps of the training method for the electric thruster plasma oscillation risk model and the prediction method for the electric thruster plasma oscillation risk.

[0126] The communication bus mentioned in the above electronic devices can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0127] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0128] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0129] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0130] The computer program product for predicting the risk of plasma oscillation in electric thrusters provided in this embodiment of the invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0131] The device for predicting plasma oscillation risks in electric thrusters provided in this embodiment of the invention can be specific hardware on the device or software or firmware installed on the device. The implementation principle and technical effects of the device provided in this embodiment of the invention are the same as those in the foregoing method embodiments. For the sake of brevity, any parts not mentioned in the device embodiments can be referred to the corresponding content in the foregoing method embodiments. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can all be referred to the corresponding processes in the above method embodiments, and will not be repeated here.

[0132] In the embodiments provided by this invention, it should be understood that the disclosed apparatus and method can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0133] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0134] In addition, the functional units in the embodiments provided by the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0135] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0136] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0137] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. All should be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A training method for a plasma oscillation risk prediction model for electric thrusters, characterized in that, The method includes: Acquire historical oscillation datasets of electric thrusters under various operating conditions. The historical oscillation datasets consist of spatial coordinate data of each measurement point and time-series oscillation signals. Extract the target oscillation component from the historical oscillation dataset under various operating conditions; the target oscillation component is the oscillation component within a preset frequency range; Based on the Fast Fourier Transform formula, the time-series oscillation signals in the historical oscillation dataset under various operating conditions are converted into frequency-domain oscillation signals to obtain the spectral distribution of the oscillation signals; The spectral distribution of the target oscillation component is determined based on the spectral distribution of the oscillation signal and the frequency range of the target oscillation component; The bispectral values ​​among multiple target oscillation components are calculated based on the bispectral calculation formula and the spectral distribution of the target oscillation components. Calculate the magnitude of the bispectral values ​​between the target oscillation components to obtain the coupling strength between the target oscillation components; Based on the coupling strength between the target oscillation components, an inverse spatial dimension transformation is performed to obtain a spatial distribution map of the coupling strength between the target oscillation components. The coupling feature data in the spatial distribution map is extracted based on a pre-trained convolutional neural network model. The coupling feature data includes at least the spatial distribution pattern of coupling strength and the peak value of coupling strength. The historical oscillation datasets under various operating conditions and their corresponding coupling feature data are input into the pre-constructed oscillation risk prediction model for iterative training until the preset training stopping condition is reached; each iteration of training will output the oscillation mode, the oscillation intensity of the oscillation mode, and the spatial location information of the maximum oscillation intensity.

2. The method according to claim 1, characterized in that, Before extracting the target oscillation component from the historical oscillation dataset under various operating conditions, the method further includes: The spatial coordinate data in the historical oscillation dataset is reduced in dimensionality to obtain one-dimensional spatial coordinate data, and the one-dimensional spatial coordinate data and the time-series oscillation signal form a two-dimensional historical oscillation dataset. Correspondingly, the extraction of target oscillation components from historical oscillation datasets under various operating conditions includes: Extract the target oscillation component from the two-dimensional historical oscillation dataset under various operating conditions.

3. The method according to claim 2, characterized in that, The target oscillation components extracted from the two-dimensional historical oscillation dataset under various operating conditions include: The target oscillation component is obtained by inputting a two-dimensional historical oscillation dataset under various operating conditions into a pre-trained neural network model. The neural network model is at least one of a convolutional neural network model and a deep neural network model.

4. The method according to claim 3, characterized in that, The training process of the neural network model includes: Acquire data on multiple historical oscillation components; The spatiotemporal modes of each oscillation component are extracted using the intrinsic orthogonal decomposition algorithm or the dynamic mode decomposition algorithm, and each spatiotemporal mode represents a type of oscillation component. The historical oscillation component data and the corresponding spatiotemporal modes are input into the neural network model for iterative training until the preset iteration conditions are met.

5. The method according to claim 2, characterized in that, The Fast Fourier Transform formula is as follows: in, Representing a spatial point The spectral distribution of the oscillation signal at that location; Indicates measurement point The timing oscillation signal at the location; Indicates frequency; Indicates time; Indicates the number of points in space; Indicates time Quantity; Indicates an exponential factor; The formula for calculating the bispectrum is: in, It is a bispectral value; and It is an oscillating signal at a frequency and The frequency domain components; It is an oscillating signal at a frequency and The conjugate components; Indicates time The quantity.

6. A method for predicting the risk of plasma oscillation in an electric thruster, characterized in that, The method includes: Acquire spatial coordinate data and timing oscillation signals at various measurement points in the electric thruster under the current operating conditions; The spatial coordinate data and the time-series oscillation signal are input into the oscillation risk prediction model trained by any one of the methods described in claims 1-5 for prediction, so as to obtain the oscillation mode, the intensity of the oscillation mode, and the spatial location information of the maximum oscillation intensity.

7. A device for predicting the risk of plasma oscillation in an electric thruster, characterized in that, The device includes: The acquisition unit is used to acquire the spatial coordinate data and timing oscillation signal of each measurement point in the electric thruster under the current operating conditions. The prediction unit is used to input the spatial coordinate data and the time-series oscillation signal into the oscillation risk prediction model pre-trained by any one of the methods described in claims 1-5 for prediction, so as to obtain the oscillation mode, the intensity of the oscillation mode, and the spatial location information of the maximum oscillation intensity.

8. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the method of any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in any one of claims 1-6.

Images