Plasma parameter diagnosis method based on Bi-LSTM model and related equipment

Abstract

The invention discloses a plasma parameter diagnosis method based on a Bi-LSTM model and related equipment, and the method comprises the steps: obtaining I-V characteristic data of double probes in a DC glow discharge plasma device under different potential deviation angles and different discharge voltage-air pressure combinations, plasma parameters corresponding to the I-V characteristic data at the datum point are recorded; taking the plasma parameters as labels of the I-V characteristic data to construct a data set, dividing the data set into a training set and a verification set, and training a preset Bi-LSTM model by taking minimization of a mean absolute error as a target to obtain a target Bi-LSTM model; and inputting the I-V characteristic data of the double probes to be tested into the target Bi-LSTM model to obtain plasma parameters corresponding to the I-V characteristic data of the double probes to be tested.

Description

Technical Field

[0001] This invention relates to the field of plasma technology, and in particular to a plasma parameter diagnosis method and related equipment based on a Bi-LSTM model. Background Technology

[0002] Langmuir probe diagnostics can be categorized into single-probe, dual-probe, and triple-probe systems based on the number of probes. Single-probe systems, using the discharge electrode as a reference, are prone to problems such as excessive probe current, damage, and interference with the plasma environment. In contrast, the dual-probe system proposed by Johnson in 1950 effectively overcomes the shortcomings of single-probe systems because both probes are suspended and the probe current is limited to a smaller ion current. Triple-probe systems, by adding auxiliary electrodes, can simultaneously measure electron temperature and density, but this increases system complexity.

[0003] Traditional Langmuir dual-probe theory, when deriving the current-voltage characteristic relationship to solve plasma parameters, typically assumes that the two probes are identical and that they are at equal plasma potentials. Therefore, the ideal dual-probe current-voltage characteristic curve conforms to a hyperbolic tangent function relationship, meaning the curve is symmetrical about the origin and passes through it. However, experimentally acquired dual-probe current-voltage characteristic curves are usually deviated from the origin and are asymmetrical. Since the Langmuir probe method for diagnosing plasma parameters is highly dependent on the analysis of the current-voltage characteristic curve, this potential asymmetry causing curve deviation inevitably affects the accuracy of the dual-probe diagnostic results. The factors causing this curve deviation can be summarized into two categories: first, significant interfering factors, including probe surface contamination and secondary electron emission; second, inherent systematic errors, including differences in the absorption areas of the two probes and potential asymmetry. After corrections through probe structure optimization and theoretical adjustments, the calculation error of the IV characteristic slope of the dual probe in the delay region can usually be controlled within the range of 10%-25%, a level of error widely accepted in engineering applications. However, due to the difficulty in completely eliminating factors such as differences in probe manufacturing processes and the complexity of experimental environments, it is difficult to completely solve the diagnostic error problem caused by the asymmetry of absorption area and potential of dual probes by relying solely on experimental improvements or theoretical corrections. Summary of the Invention

[0004] In view of this, the present invention provides a plasma parameter diagnosis method and related equipment based on a Bi-LSTM model, which addresses the problem in existing technologies where differences in probe manufacturing processes and the complexity of experimental environments make it difficult to completely eliminate diagnostic errors caused by plasma potential asymmetry in dual probes, and where experimental improvements or theoretical corrections alone are insufficient to completely resolve these errors. To achieve one, some, or all of the above objectives, or other objectives, the present invention proposes a plasma parameter diagnosis method based on a Bi-LSTM model, comprising: Acquire IV characteristic data of dual probes in a DC glow discharge plasma device under different potential deviation angles and different discharge voltage-pressure combinations, and record the plasma parameters corresponding to the IV characteristic data at the reference point, including electron temperature and electron density. The plasma parameters are used as labels for the IV characteristic data to construct a dataset, which is then divided into a training set and a validation set. The preset Bi-LSTM model is trained with the goal of minimizing the mean absolute error to obtain the target Bi-LSTM model. The dual-probe IV characteristic data to be tested are input into the target Bi-LSTM model to obtain the plasma parameters corresponding to the dual-probe IV characteristic data to be tested.

[0005] Optionally, before the step of recording the plasma parameters corresponding to the IV characteristic data at the reference point, the method further includes: The dual probes were fixed to the rotating platform of the DC glow discharge plasma device to obtain the IV characteristic curve dataset of the dual probes at different deflection angles. In the IV characteristic curve dataset, the IV characteristic curve with the smallest distance from the origin and the highest left-right symmetry is selected as the target curve, and the dual probe position corresponding to the target curve is recorded as the reference point.

[0006] Optionally, the step of acquiring IV characteristic data of dual probes under different potential deviation angles and different discharge voltage-pressure combinations in a DC glow discharge plasma device includes: Centered on the reference point, with the clockwise direction as the positive direction, the rotating platform is adjusted so that the dual probes acquire the first IV characteristic data of the dual probes in the DC glow discharge plasma device under different potential deviation angles and different discharge voltage-pressure combinations according to the preset first sampling conditions. The first sampling conditions include at least: rotating in the positive direction within the rated range, with a sampling interval of 100V within the discharge voltage range that can generate glow discharge, with a sampling interval of 2Pa within the discharge pressure range that can generate glow discharge, and with the dual probe bias voltage range between -30V and 30V, with a sampling interval of 0.5V. Centered on the reference point, with the counterclockwise direction as the negative direction, the rotating platform is adjusted so that the dual probes acquire the second IV characteristic data of the dual probes in the DC glow discharge plasma device under different potential deviation angles and different discharge voltage-pressure combinations according to the preset second sampling conditions. The second sampling conditions include at least: rotating in the negative direction within the rated range, with a sampling interval of 100V within the discharge voltage range that can generate glow discharge, with a sampling interval of 2Pa within the discharge pressure range that can generate glow discharge, and with the dual probe bias voltage range between -30V and 30V, with a sampling interval of 0.5V. Both the first IV characteristic data and the second IV characteristic data are used as IV characteristic data.

[0007] Optionally, the preset Bi-LSTM model includes a two-layer Bi-LSTM network, 16 hidden neurons, and 32 fully connected layer neurons.

[0008] Optionally, the preset Bi-LSTM model further includes at least two identical Dropout layers, one Dropout layer is set between two Bi-LSTM network layers, and the other Dropout layer is set between the fully connected layer of the preset Bi-LSTM model and the output layer of the preset Bi-LSTM model, with the Dropout rate of the Dropout layer being 0.2.

[0009] Optionally, the step of dividing the dataset into a training set and a validation set, and training a preset Bi-LSTM model with the objective of minimizing the mean absolute error to obtain a target Bi-LSTM model includes: The Adam optimizer is used, with the loss function being the mean squared error function. The goal is to minimize the mean absolute error. Based on the training set and validation set, the preset Bi-LSTM model is trained until the model converges and there is no overfitting, thus obtaining the target Bi-LSTM model.

[0010] Optionally, the optimal learning rate of the preset Bi-LSTM model is 0.0005.

[0011] Secondly, this application provides a plasma parameter diagnostic device based on a Bi-LSTM model, comprising: The data processing module is used to acquire IV characteristic data of the dual probes in a DC glow discharge plasma device under different potential deviation angles and different discharge voltage-gas pressure combinations, and to record the plasma parameters corresponding to the IV characteristic data at the reference point, including electron temperature and electron density. The training module is used to use the plasma parameters as labels for the IV characteristic data to construct a dataset, and to divide the dataset into a training set and a validation set. With the goal of minimizing the mean absolute error, the preset Bi-LSTM model is trained to obtain the target Bi-LSTM model. The diagnostic module is used to input the dual-probe IV characteristic data to be tested into the target Bi-LSTM model to obtain the plasma parameters corresponding to the dual-probe IV characteristic data to be tested.

[0012] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the plasma parameter diagnosis method based on the Bi-LSTM model as described above.

[0013] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the plasma parameter diagnosis method based on the Bi-LSTM model as described above.

[0014] Implementing the embodiments of the present invention will have the following beneficial effects: When the plasma potential of the dual probes is asymmetrical, the Bi-LSTM model exhibits significant stability and accuracy. For electron density calculation, the Bi-LSTM model can further improve data accuracy beyond the maximum error of traditional fitting methods, especially at high deflection angles, significantly reducing errors. Furthermore, the model can stably correct errors at different deflection angles, further confirming its effectiveness and robustness in electron density prediction. Combining the analysis results of electron temperature and density, the Bi-LSTM model demonstrates comprehensive and excellent performance in plasma diagnostics. Moreover, after model training, it can output correction results in real time, adapting to the rapid diagnostic needs of industrial scenarios. While significantly improving diagnostic efficiency, it solves the problem of low measurement accuracy in existing technologies. In addition to DC glow discharge plasma, after parameter adjustment, it can be used for radio frequency plasma. For example, it can solve the problem of IV curve distortion caused by radio frequency self-bias, providing a general method for high-precision diagnostics in fields such as etching, spraying, and nuclear fusion. Attached Figure Description

[0015] 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, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] in: Figure 1 This is a schematic flowchart of a plasma parameter diagnosis method based on a Bi-LSTM model provided in an embodiment of this application; Figure 2 This is an application flowchart of a Bi-LSTM model provided in an embodiment of this application; Figure 3 This is a schematic diagram of the DC glow discharge plasma device provided in the embodiments of this application; Figure 4 This is a schematic diagram of the structure of a plasma parameter diagnostic device based on a Bi-LSTM model provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of a computer-readable storage medium provided in an embodiment of this application. Detailed Implementation

[0017] 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.

[0018] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. In the description of this invention, it should be noted that unless otherwise explicitly specified and limited, the terms "installed," "connected," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0019] like Figure 1 As shown in the figure, this application provides a plasma parameter diagnosis method based on a Bi-LSTM model, including: S110. Obtain IV characteristic data of dual probes in a DC glow discharge plasma device under different potential deviation angles and different discharge voltage-pressure combinations, and record the plasma parameters corresponding to the IV characteristic data at the reference point. The plasma parameters include electron temperature and electron density. S120. The plasma parameters are used as labels for the IV characteristic data to construct a dataset, and the dataset is divided into a training set and a validation set. The preset Bi-LSTM model is trained with the goal of minimizing the mean absolute error to obtain the target Bi-LSTM model. For example, the dual-probe IV curve current data is selected as the model input, and the electron temperature and electron density of the corresponding reference point are used as labels to form a dataset of (1061, 62), which is divided into 849 training sets, 106 validation sets, and 106 test sets in an 8:1:1 ratio.

[0020] S130. Input the dual-probe IV characteristic data to be tested into the target Bi-LSTM model to obtain the plasma parameters corresponding to the dual-probe IV characteristic data to be tested.

[0021] When the plasma potential of the dual probes is asymmetrical, the Bi-LSTM model exhibits significant stability and accuracy. For electron density calculation, the Bi-LSTM model can further improve data accuracy beyond the maximum error of traditional fitting methods, especially at high deflection angles, significantly reducing errors. Furthermore, the model can stably correct errors at different deflection angles, further confirming its effectiveness and robustness in electron density prediction. Combining the analysis results of electron temperature and density, the Bi-LSTM model demonstrates comprehensive and excellent performance in plasma diagnostics. Moreover, after model training, it can output correction results in real time, adapting to the rapid diagnostic needs of industrial scenarios. While significantly improving diagnostic efficiency, it solves the problem of low measurement accuracy in existing technologies. In addition to DC glow discharge plasma, after parameter adjustment, it can be used for radio frequency plasma. For example, it can solve the problem of IV curve distortion caused by radio frequency self-bias, providing a general method for high-precision diagnostics in fields such as etching, spraying, and nuclear fusion.

[0022] In one possible implementation, prior to the step of recording the plasma parameters corresponding to the IV characteristic data at the reference point, the method further includes: The dual probes were fixed to the rotating platform of the DC glow discharge plasma device to obtain the IV characteristic curve dataset of the dual probes at different deflection angles. In the IV characteristic curve dataset, the IV characteristic curve with the smallest distance from the origin and the highest left-right symmetry is selected as the target curve, and the dual probe position corresponding to the target curve is recorded as the reference point.

[0023] For example, two probes are fixed on a rotating platform, and their positions are adjusted to the center of the positive column region of the discharge tube. By precisely adjusting the deflection angle of the two probes, the shape of the IV characteristic curve is monitored. A reference point is determined according to the rule of minimizing the deviation of the curve from the origin and maximizing the left-right symmetry. The probe position corresponding to the reference point is recorded as the reference point with no potential difference, denoted as 0°, at which point the potentials of the two probes are nearly equal.

[0024] In one possible implementation, the step of acquiring IV characteristic data of the dual probes in a DC glow discharge plasma device under different potential deviation angles and different discharge voltage-pressure combinations includes: Centered on the reference point, with the clockwise direction as the positive direction, the rotating platform is adjusted so that the dual probes acquire the first IV characteristic data of the dual probes in the DC glow discharge plasma device under different potential deviation angles and different discharge voltage-pressure combinations according to the preset first sampling conditions. The first sampling conditions include at least: rotating in the positive direction within the rated range, with a sampling interval of 100V within the discharge voltage range that can generate glow discharge, with a sampling interval of 2Pa within the discharge pressure range that can generate glow discharge, and with the dual probe bias voltage range between -30V and 30V, with a sampling interval of 0.5V. Centered on the reference point, with the counterclockwise direction as the negative direction, the rotating platform is adjusted so that the dual probes acquire the second IV characteristic data of the dual probes in the DC glow discharge plasma device under different potential deviation angles and different discharge voltage-pressure combinations according to the preset second sampling conditions. The second sampling conditions include at least: rotating in the negative direction within the rated range, with a sampling interval of 100V within the discharge voltage range that can generate glow discharge, with a sampling interval of 2Pa within the discharge pressure range that can generate glow discharge, and with the dual probe bias voltage range between -30V and 30V, with a sampling interval of 0.5V. Both the first IV characteristic data and the second IV characteristic data are used as IV characteristic data.

[0025] For example, with a reference point as the center, and clockwise as the positive direction, the rotating platform is adjusted to rotate the two probes to the left and right respectively, simulating different potential deviation conditions, rotating 20°, 40°, and 60° in the positive direction. Counterclockwise is taken as the negative direction, and rotations of 20°, 40°, and 60° in the negative direction are recorded as negative values. During the experiment, the discharge voltage range is 800-1900V, with intervals of 100V; the discharge pressure range is 10-34Pa, with intervals of 2Pa. For each potential deviation angle and each discharge voltage-pressure combination, dual-probe IV characteristic data are collected: the bias voltage range is between -30V and 30V, with intervals of 0.5V; simultaneously with each scan and data collection, the electron temperature and electron density at the reference point with the same discharge parameters are recorded as true labels.

[0026] In one possible implementation, the preset Bi-LSTM model includes a two-layer Bi-LSTM network, 16 hidden neurons, and 32 fully connected layer neurons.

[0027] In one possible implementation, the preset Bi-LSTM model further includes at least two identical Dropout layers, one Dropout layer is set between two Bi-LSTM network layers, and the other Dropout layer is set between the fully connected layer of the preset Bi-LSTM model and the output layer of the preset Bi-LSTM model, with the Dropout rate of the Dropout layer being 0.2.

[0028] In one possible implementation, the step of dividing the dataset into a training set and a validation set, training a preset Bi-LSTM model with the objective of minimizing the mean absolute error, and obtaining a target Bi-LSTM model includes: The Adam optimizer is used, with the loss function being the mean squared error function. The goal is to minimize the mean absolute error. Based on the training set and validation set, the preset Bi-LSTM model is trained until the model converges and there is no overfitting, thus obtaining the target Bi-LSTM model.

[0029] For example, by conducting comparative experiments with the goal of minimizing MAE and ensuring that the performance on the training and validation sets is as close as possible, the optimal model hyperparameters can be selected to complete the model training process. Figure 2As shown, a two-layer Bi-LSTM structure is used, with 16 neurons in each hidden layer and 32 neurons in each fully connected layer. Too many layers can lead to overfitting, while too few neurons cannot fully extract features. The optimal learning rate is 0.0005; a learning rate that is too high can cause model oscillations and non-convergence, while a learning rate that is too low will slow down the training speed. The batch size should be 64; small batches provide stable gradient estimation but require more iterations, while large batches are prone to overfitting. The window size should be 5; a window size that is too small will not utilize enough information, while a window size that is too large will introduce noise. Regularization is achieved by adding Dropout layers between the two LSTM layers and between the fully connected layer and the output layer, with a Dropout rate of 0.2 to suppress overfitting. The optimizer is Adam, and the loss function is mean squared error, which meets the optimization requirements of regression problems.

[0030] In one possible implementation, the optimal learning rate of the preset Bi-LSTM model is 0.0005.

[0031] For example, the preset Bi-LSTM model has 200 training epochs. In the initial 1-50 epochs of training and validation, the MSE of the training and validation sets decreases rapidly. From 50-150 epochs, the MSE decreases more slowly, and after 150 epochs, the MSE stabilizes at 10. -3 The magnitude of the MSE values ​​and the near overlap between the training and validation sets indicate that the model has converged and is not overfitting.

[0032] For example, such as Figure 3 As shown, the DC glow discharge plasma device mainly consists of three parts: a discharge system, a probe measurement system, and a gas delivery and control system. The discharge system comprises a DC power supply, a cylindrical discharge tube, and a circulating cooling water circuit. By applying a DC power supply between two parallel metal electrodes, a stable glow discharge mode is generated after the gas is broken down. Simultaneously, circulating water cooling effectively reduces the cathode temperature, preventing electrode overheating due to prolonged discharge. The gas delivery and control system consists of a mechanical pump, an argon cylinder, and a vacuum gauge. Through precise coordination of a baffle valve and a mass flow meter, the pumping and intake rates can be dynamically adjusted, thereby controlling and maintaining the required discharge gas pressure environment for the experiment. To reduce probe disturbance to the plasma and eliminate edge effects, the probe measurement system uses a cylindrical tungsten wire probe with a length of 8 mm and a diameter of 0.8 mm (length-to-diameter ratio of 10:1). The system provides bias voltage to the two probes through an adjustable DC power supply and uses an automatic data acquisition device to record probe voltage and current data in real time for subsequent processing.

[0033] In one possible implementation, such as Figure 4 As shown, this application provides a plasma parameter diagnostic device based on a Bi-LSTM model, comprising: The data processing module 201 is used to acquire IV characteristic data of the dual probes in a DC glow discharge plasma device under different potential deviation angles and different discharge voltage-gas pressure combinations, and to record the plasma parameters corresponding to the IV characteristic data at the reference point, including electron temperature and electron density. Training module 202 is used to use the plasma parameters as labels for the IV characteristic data to construct a dataset, and to divide the dataset into a training set and a validation set. With the goal of minimizing the mean absolute error, a preset Bi-LSTM model is trained to obtain a target Bi-LSTM model. The diagnostic module 203 is used to input the dual-probe IV characteristic data to be tested into the target Bi-LSTM model to obtain the plasma parameters corresponding to the dual-probe IV characteristic data to be tested.

[0034] In one possible implementation, such as Figure 5 As shown, this application embodiment provides a terminal device 300, including: a memory 310, a processor 320, and a first computer program 311 stored in the memory 310 and executable on the processor 320. When the processor 320 executes the first computer program 311, it performs the following steps: acquiring IV characteristic data of dual probes in a DC glow discharge plasma device under different potential deviation angles and different discharge voltage-pressure combinations, and recording the plasma parameters corresponding to the IV characteristic data at a reference point, wherein the plasma parameters include electron temperature and electron density; using the plasma parameters as labels for the IV characteristic data to construct a dataset, and dividing the dataset into a training set and a validation set, training a preset Bi-LSTM model with the goal of minimizing the mean absolute error to obtain a target Bi-LSTM model; and inputting the dual probe IV characteristic data to be tested into the target Bi-LSTM model to obtain the plasma parameters corresponding to the dual probe IV characteristic data to be tested.

[0035] In one possible implementation, such as Figure 6As shown, this application embodiment provides a computer-readable storage medium 400, on which a second computer program 411 is stored. When the second computer program 411 is executed by a processor, it acquires IV characteristic data of dual probes in a DC glow discharge plasma device under different potential deviation angles and different discharge voltage-pressure combinations, and records the plasma parameters corresponding to the IV characteristic data at a reference point. The plasma parameters include electron temperature and electron density. The plasma parameters are used as labels for the IV characteristic data to construct a dataset, and the dataset is divided into a training set and a validation set. A preset Bi-LSTM model is trained with the goal of minimizing the mean absolute error to obtain a target Bi-LSTM model. The steps of inputting the dual probe IV characteristic data to be tested into the target Bi-LSTM model to obtain the plasma parameters corresponding to the dual probe IV characteristic data to be tested are as follows.

[0036] Note that the above description is merely a preferred embodiment of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, the scope of which is determined by the scope of the appended claims.

[0037] The above description discloses only preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A plasma parameter diagnosis method based on a Bi-LSTM model, characterized in that, include: Acquire IV characteristic data of dual probes in a DC glow discharge plasma device under different potential deviation angles and different discharge voltage-pressure combinations, and record the plasma parameters corresponding to the IV characteristic data at the reference point, including electron temperature and electron density. The plasma parameters are used as labels for the IV characteristic data to construct a dataset, which is then divided into a training set and a validation set. The preset Bi-LSTM model is trained with the goal of minimizing the mean absolute error to obtain the target Bi-LSTM model. The dual-probe IV characteristic data to be tested are input into the target Bi-LSTM model to obtain the plasma parameters corresponding to the dual-probe IV characteristic data to be tested.

2. The plasma parameter diagnosis method based on a Bi-LSTM model as described in claim 1, characterized in that, Before the step of recording the plasma parameters corresponding to the IV characteristic data at the reference point, the method further includes: The dual probes were fixed to the rotating platform of the DC glow discharge plasma device to obtain the IV characteristic curve dataset of the dual probes at different deflection angles. In the IV characteristic curve dataset, the IV characteristic curve with the smallest distance from the origin and the highest left-right symmetry is selected as the target curve, and the dual probe position corresponding to the target curve is recorded as the reference point.

3. The plasma parameter diagnosis method based on a Bi-LSTM model as described in claim 2, characterized in that, The steps for obtaining IV characteristic data of dual probes under different potential deviation angles and different discharge voltage-pressure combinations in a DC glow discharge plasma device include: Centered on the reference point, with the clockwise direction as the positive direction, the rotating platform is adjusted so that the dual probes acquire the first IV characteristic data of the dual probes in the DC glow discharge plasma device under different potential deviation angles and different discharge voltage-pressure combinations according to the preset first sampling conditions. The first sampling conditions include at least: rotating in the positive direction within the rated range, with a sampling interval of 100V within the discharge voltage range that can generate glow discharge, with a sampling interval of 2Pa within the discharge pressure range that can generate glow discharge, and with the dual probe bias voltage range between -30V and 30V, with a sampling interval of 0.5V. Centered on the reference point, with the counterclockwise direction as the negative direction, the rotating platform is adjusted so that the dual probes acquire the second IV characteristic data of the dual probes in the DC glow discharge plasma device under different potential deviation angles and different discharge voltage-pressure combinations according to the preset second sampling conditions. The second sampling conditions include at least: rotating in the negative direction within the rated range, with a sampling interval of 100V within the discharge voltage range that can generate glow discharge, with a sampling interval of 2Pa within the discharge pressure range that can generate glow discharge, and with the dual probe bias voltage range between -30V and 30V, with a sampling interval of 0.5V. Both the first IV characteristic data and the second IV characteristic data are used as IV characteristic data.

4. The plasma parameter diagnosis method based on a Bi-LSTM model as described in claim 1, characterized in that, The preset Bi-LSTM model includes a two-layer Bi-LSTM network, 16 hidden neurons, and 32 fully connected neurons.

5. The plasma parameter diagnosis method based on a Bi-LSTM model as described in claim 1, characterized in that, The preset Bi-LSTM model also includes at least two identical Dropout layers. One Dropout layer is set between two Bi-LSTM network layers, and the other Dropout layer is set between the fully connected layer of the preset Bi-LSTM model and the output layer of the preset Bi-LSTM model. The Dropout rate of the Dropout layer is 0.

2.

6. The plasma parameter diagnosis method based on a Bi-LSTM model as described in claim 1, characterized in that, The step of dividing the dataset into a training set and a validation set, training a preset Bi-LSTM model with the objective of minimizing the mean absolute error, and obtaining the target Bi-LSTM model includes: The Adam optimizer is used, with the loss function being the mean squared error function. The goal is to minimize the mean absolute error. Based on the training set and validation set, the preset Bi-LSTM model is trained until the model converges and there is no overfitting, thus obtaining the target Bi-LSTM model.

7. The plasma parameter diagnosis method based on a Bi-LSTM model as described in claim 1, characterized in that, The optimal learning rate of the preset Bi-LSTM model is 0.0005.

8. A plasma parameter diagnostic device based on a Bi-LSTM model, characterized in that, include: The data processing module is used to acquire IV characteristic data of the dual probes in a DC glow discharge plasma device under different potential deviation angles and different discharge voltage-gas pressure combinations, and to record the plasma parameters corresponding to the IV characteristic data at the reference point, including electron temperature and electron density. The training module is used to use the plasma parameters as labels for the IV characteristic data to construct a dataset, and to divide the dataset into a training set and a validation set. With the goal of minimizing the mean absolute error, the preset Bi-LSTM model is trained to obtain the target Bi-LSTM model. The diagnostic module is used to input the dual-probe IV characteristic data to be tested into the target Bi-LSTM model to obtain the plasma parameters corresponding to the dual-probe IV characteristic data to be tested.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the plasma parameter diagnosis method based on the Bi-LSTM model as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the plasma parameter diagnosis method based on the Bi-LSTM model as described in any one of claims 1 to 7.

Images