A method and device for adaptive load matching of radio frequency power supply

Abstract

The present invention provides an adaptive load matching method and device for a radio frequency power supply, which relates to the field of power supply management technology, including: establishing a load data set of a load, establishing a load feature extraction result, and building a steady-state prediction network for the load; performing random frame sampling on the load, establishing a frame sampling result, performing steady-state prediction, and generating a steady-state prediction result; performing time interval clustering of the load, establishing a time interval clustering result, and re-collecting data within the time interval clustering result to generate a data collection result; configuring a comprehensive evaluation formula to generate an impedance feature comprehensive evaluation result; establishing an impedance matching model for the radio frequency power supply, and generating matching parameters for the radio frequency power supply; and controlling the radio frequency power supply to perform power supply control management for the load. The present invention solves the technical problem in the prior art that the radio frequency power supply has a slow response speed and cannot adjust the matching parameters in time when facing dynamic load changes, resulting in low energy transmission efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply management, and in particular to an adaptive load matching method and device for a radio frequency power supply. Background Art

[0002] RF power supplies are widely used in many fields. For example, they are used as energy sources for signal transmission in wireless communication systems to ensure that signals can be transmitted efficiently during transmission and reception. However, the existing technology has a slow response speed when facing dynamic load changes and cannot adjust matching parameters in real time, resulting in performance degradation of RF power supplies when load changes are large and low energy transmission efficiency. In addition, the load matching method in the existing technology cannot accurately predict and adjust load changes, resulting in unsatisfactory matching effects of RF power supplies in actual operation, which will cause an increase in reflected power and affect the transmission efficiency and stability of the system. Due to unsatisfactory load matching and insufficient dynamic response, the system loses a lot of energy during operation, and cannot fully utilize the input energy, reducing the overall efficiency of the system. Summary of the invention

[0003] The present application provides an adaptive load matching method and device for an RF power supply, aiming to solve the technical problem in the prior art that the RF power supply has a slow response speed and cannot adjust the matching parameters in time when facing dynamic load changes, resulting in low energy transmission efficiency.

[0004] The first aspect disclosed in the present application provides an adaptive load matching method for a radio frequency power supply, the method comprising: establishing a load data set of the load, extracting load characteristics according to the load data set, establishing a load characteristic extraction result, and building a steady-state prediction network for the load with the load characteristic extraction result; performing random frame sampling on the load to establish a frame sampling result, performing data preprocessing on the frame sampling result, and inputting the result into the steady-state prediction network, performing steady-state prediction, and generating a steady-state prediction result; performing time interval clustering of the load according to the steady-state prediction result, establishing a time interval clustering result, and re-collecting data within the time interval clustering result to generate a data collection result; configuring a comprehensive evaluation formula, performing a comprehensive evaluation of the impedance characteristics of the data collection result based on the comprehensive evaluation formula, and generating an impedance characteristic comprehensive evaluation result; establishing an impedance matching model of the radio frequency power supply, inputting the impedance characteristic comprehensive evaluation result into the impedance matching model, and generating matching parameters of the radio frequency power supply; and controlling the radio frequency power supply according to the matching parameters to perform power supply control and management of the load.

[0005] In a second aspect disclosed in the present application, there is provided an adaptive load matching device for an RF power supply, the device being used for the adaptive load matching method for an RF power supply described above, the device comprising: a feature extraction module, the feature extraction module being used to establish a load data set of the load, extract load features according to the load data set, establish a load feature extraction result, and use the load feature extraction result to build a steady-state prediction network for the load; a steady-state prediction module, the steady-state prediction module being used to perform random frame sampling on the load, establish a frame sampling result, perform data preprocessing on the frame sampling result, and then input it into the steady-state prediction network, perform steady-state prediction, and generate a steady-state prediction result; a data acquisition module, the data acquisition module being used to perform random frame sampling on the load, establish a frame sampling result, perform data preprocessing on the frame sampling result, and then input it into the steady-state prediction network, perform steady-state prediction, and generate a steady-state prediction result; The steady-state prediction results are clustered into time intervals of the load, a time interval clustering result is established, and data collection is performed again within the time interval clustering result to generate a data collection result; a comprehensive evaluation module, the comprehensive evaluation module is used to configure a comprehensive evaluation formula, and based on the comprehensive evaluation formula, an impedance characteristic comprehensive evaluation is performed on the data collection result, and an impedance characteristic comprehensive evaluation result is generated; a matching parameter generation module, the matching parameter generation module is used to establish an impedance matching model of the RF power supply, and the impedance characteristic comprehensive evaluation result is input into the impedance matching model to generate matching parameters of the RF power supply; a power supply control management module, the power supply control management module is used to control the RF power supply according to the matching parameters to perform power supply control management of the load.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] By establishing a load data set and extracting load characteristics, and constructing a steady-state prediction network, the steady-state characteristics under different load conditions can be accurately predicted, thereby improving the matching accuracy; through random frame sampling and steady-state prediction, real-time load monitoring and prediction are performed to achieve rapid response and adjustment to load changes, ensure the stable operation of the system under dynamic load conditions, and improve energy transmission efficiency; by clustering the load time intervals and collecting data within the time interval clustering results, the load data is comprehensively collected and utilized, the data utilization rate is improved, and comprehensive data support is provided for the matching model; by configuring a comprehensive evaluation formula, a comprehensive impedance characteristic comprehensive evaluation is performed on the data collection results, and a comprehensive evaluation result is generated, and the optimized matching parameters are generated in combination with the impedance matching model, which significantly improves the matching effect and improves the overall performance and stability of the system. In general, the adaptive load matching method of the RF power supply solves the problems of inaccurate load matching, insufficient dynamic response, low data utilization, and unsatisfactory matching effect in the existing technology through comprehensive load data analysis, steady-state prediction, and comprehensive evaluation, and improves the overall performance and stability of the RF power supply.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic flow chart of an adaptive load matching method for a radio frequency power supply provided in an embodiment of the present application;

[0010] Figure 2 A schematic diagram of the structure of an adaptive load matching device for a radio frequency power supply provided in an embodiment of the present application.

[0011] Explanation of reference numerals: feature extraction module 10 , steady-state prediction module 20 , data acquisition module 30 , comprehensive evaluation module 40 , matching parameter generation module 50 , power supply control management module 60 . DETAILED DESCRIPTION

[0012] The embodiment of the present application provides an adaptive load matching method for an RF power supply, thereby solving the technical problem in the prior art that the RF power supply has a slow response speed and cannot adjust the matching parameters in time when facing dynamic load changes, resulting in low energy transmission efficiency.

[0013] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically introduced below in conjunction with the accompanying drawings of the specification. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0014] like Figure 1 As shown, an embodiment of the present application provides an adaptive load matching method for a radio frequency power supply, the method comprising:

[0015] A load data set of the load is established, load characteristics are extracted according to the load data set, a load characteristic extraction result is established, and a steady-state prediction network of the load is built with the load characteristic extraction result.

[0016] Collect load data from the data recording system of the RF power supply, including parameters such as voltage, current, power, impedance, reflection coefficient, and standing wave ratio, and establish a load data set. Perform feature extraction on the load data set, for example, using time domain analysis to extract features from time domain data, such as average value, standard deviation, maximum value, minimum value, etc.; use frequency domain analysis, Fourier transform or wavelet transform, to extract features from frequency domain data, such as frequency distribution, frequency component amplitude, etc., and combine the extracted features into feature vectors to form a feature set.

[0017] According to the characteristics of the load data and the prediction requirements, select the appropriate machine learning or deep learning model, such as multi-layer perceptron (MLP), long short-term memory network (LSTM), convolutional neural network (CNN), etc. Divide the data set into training set, validation set and test set, use the training set to train the model, adjust the model parameters and structure, so that it can accurately predict the load steady state; evaluate the model performance on the validation set, adjust the model hyperparameters according to the validation results to prevent overfitting or underfitting; evaluate the performance of the model on the test set, and further optimize the model according to the evaluation results to improve the prediction accuracy and robustness. Finally, a trained load steady-state prediction network is obtained, which is used to input load data in real time for steady-state prediction.

[0018] The load is randomly sampled to establish a frame sampling result, the frame sampling result is subjected to data preprocessing, and then input into the steady-state prediction network to perform steady-state prediction and generate a steady-state prediction result.

[0019] Determine the sampling frequency according to system requirements and data characteristics to ensure that the sampled data can fully represent the load changes. Use sensors and data acquisition equipment, such as oscilloscopes, power meters, vector network analyzers, etc., to randomly sample the load at the sampling frequency to form a frame sampling result.

[0020] Data preprocessing, including processing missing values, noise filtering, and data normalization, is performed to improve data quality and eliminate dimensional differences between different parameters.

[0021] The preprocessed data is input into the steady-state prediction network, and the prediction value is calculated through the forward propagation of the neural network. The neurons in each layer linearly transform the input features through weighting and bias, and then output them through the activation function. The steady-state prediction network outputs the steady-state prediction results. These results reflect the steady-state characteristics of the system under the current load conditions, providing an important basis for subsequent load matching and power supply control management.

[0022] The load is clustered by time interval according to the steady-state prediction result, a time interval clustering result is established, and data collection within the time interval clustering result is re-performed to generate a data collection result.

[0023] Collect all steady-state prediction results, which are the predicted values ​​of the steady-state characteristics of the system at different time points. According to the time distribution of the prediction results, divide the time axis into several intervals, which can be equal intervals or dynamic intervals based on specific events or feature changes.

[0024] Select a clustering algorithm, such as K-means, DBSCAN, hierarchical clustering, etc., and set the parameters of the clustering algorithm according to the data characteristics and clustering requirements, such as the number of clusters in K-means, the minimum number of samples and neighborhood radius in DBSCAN, etc. Use the selected clustering algorithm to cluster the steady-state prediction results into time intervals, obtain the clustering results of each time interval, assign a clustering label to each time interval, identify different time intervals and their corresponding features, and obtain the time interval clustering results.

[0025] According to the time interval clustering results, a new data collection strategy is formulated to determine the data collection frequency and sampling points in each time interval. Data collection is performed according to the new strategy in each time interval to ensure that the data covers different time intervals and their characteristic changes. The collected data is recorded to form data collection results.

[0026] A comprehensive evaluation formula is configured, and a comprehensive evaluation of the impedance characteristics of the data acquisition results is performed based on the comprehensive evaluation formula, and an impedance characteristic comprehensive evaluation result is generated.

[0027] Select key indicators for evaluating impedance characteristics, including reflection coefficient, standing wave ratio, power transmission efficiency and matching loss. Determine the weight coefficient of each indicator according to the degree of influence of each indicator on the comprehensive evaluation result. Establish a comprehensive evaluation formula based on the evaluation indicators and the corresponding weight coefficients.

[0028] The impedance characteristic parameters required by the above evaluation formula are extracted from the data acquisition results, and the extracted impedance characteristic parameters are substituted into the comprehensive evaluation formula. The comprehensive evaluation value of each data sample is calculated according to the formula, and the comprehensive evaluation values ​​of all data samples are summarized to form a complete impedance characteristic comprehensive evaluation result.

[0029] An impedance matching model of the radio frequency power supply is established, and the comprehensive evaluation result of the impedance characteristics is input into the impedance matching model to generate matching parameters of the radio frequency power supply.

[0030] Based on electromagnetic theory and RF circuit design principles, a mathematical model is established to describe the relationship between load impedance and RF power output. Using existing data sets, machine learning or deep learning methods are used to train matching models and learn to predict matching parameters from input features. The impedance feature data set obtained from time interval clustering and comprehensive evaluation is used as training data for model training. For machine learning models, parameter tuning is required, and model performance is verified through cross-validation, holdout method, and other methods to ensure that the model has good generalization ability.

[0031] The comprehensive evaluation results of impedance characteristics are input into the impedance matching model for inference calculation. The model calculates the matching parameters of the RF power supply based on the input characteristics. These matching parameters include tuning capacitance, inductance value, load impedance value, etc. These matching parameters can effectively guide the adjustment of the RF power supply and ensure that the system can achieve optimal performance under various load conditions.

[0032] The RF power supply is controlled according to the matching parameters to perform power supply control and management of the load.

[0033] Configure the hardware part of the RF power control system, including tuning capacitors, inductor adjustment modules, load matching networks, etc., apply matching parameters to the control strategy of the RF power supply, and achieve precise power supply control of the load. Through the above steps, the RF power supply can achieve optimal matching and efficient energy transmission under various load conditions, thereby improving the overall performance and stability of the system.

[0034] Furthermore, the configuration of the comprehensive evaluation formula, performing a comprehensive evaluation of the impedance characteristics of the data acquisition results based on the comprehensive evaluation formula, and generating a comprehensive evaluation result of the impedance characteristics, also includes:

[0035] The comprehensive evaluation formula is as follows:

[0036]

[0037] Among them, E represents the comprehensive evaluation result of impedance characteristics, |Γ| represents the reflection coefficient, which is directly obtained through data acquisition results, and VSWR represents the standing wave ratio. η represents the power transmission efficiency, ML represents the matching loss, w1, w2, w3, w4 are weight coefficients of the characteristics, and w1+w2+w3+w4=1; the impedance characteristics of the data acquisition results are comprehensively evaluated according to the comprehensive evaluation formula.

[0038] Specifically, the comprehensive evaluation formula is as follows:

[0039]

[0040] Among them, E represents the comprehensive evaluation result of impedance characteristics. The higher the value, the better the impedance matching performance of the system. The reflection coefficient |Γ| represents the ratio of the reflected wave to the incident wave of the RF power supply at the load end, reflecting the energy reflection situation. The smaller the value, the less energy reflection and the better the matching. It is expressed in the formula by the term (1-|Γ|). The purpose is to convert the reflection coefficient into a forward evaluation, that is, the smaller the reflection coefficient, the larger the value of this term.

[0041] The standing wave ratio (VSWR) is a function of the reflection coefficient, which indicates the ratio of the standing wave on the transmission line and is defined as Use in formulas To express, the smaller the standing wave ratio, The larger the value, the better the match.

[0042] The power transmission efficiency η represents the ratio of the effective power received by the load to the output power of the RF power supply, reflecting the efficiency of energy transmission. The higher the value, the more efficient the energy transmission. This item is directly used in the formula. The higher the power transmission efficiency, the greater its contribution to the comprehensive evaluation results.

[0043] Matching loss ML is an indicator that reflects the energy loss during transmission and is defined as The smaller the value, the better the match, because the energy loss is smaller. ML is directly used in the formula. The smaller the matching loss, the greater the contribution to the comprehensive evaluation results.

[0044] w1, w2, w3, and w4 are weight coefficients of the features, which are used to adjust the relative importance of each item in the comprehensive evaluation results. Their sum is 1. By adjusting the weight coefficients, the importance of certain features can be highlighted according to actual needs.

[0045] In summary, this comprehensive evaluation formula combines the four characteristics of reflection coefficient, standing wave ratio, power transmission efficiency and matching loss through the weighted average method to obtain a comprehensive evaluation result. This result is used to quantify and evaluate the impedance matching performance of the RF power supply system. The higher the value, the better the matching performance of the system.

[0046] The impedance characteristic parameters required by the above evaluation formula are extracted from the data acquisition results, and the extracted impedance characteristic parameters are substituted into the comprehensive evaluation formula. The comprehensive evaluation value of each data sample is calculated according to the formula, and the comprehensive evaluation values ​​of all data samples are summarized to form a complete impedance characteristic comprehensive evaluation result.

[0047] Furthermore, the impedance characteristic comprehensive evaluation of the data acquisition results based on the comprehensive evaluation formula also includes:

[0048] Configure time zero for the time interval clustering result, perform data collection at the time zero, and establish first data; call the steady-state prediction result, and perform stable value compensation of the first data in the time interval clustering result based on the steady-state prediction result to generate updated data; use the updated data as the data collection result to perform a comprehensive evaluation of impedance characteristics.

[0049] Based on the time interval clustering results, the specific starting point and ending point of each time interval are determined, and the starting point of each time interval is taken as the time zero point. Data collection is performed at the time zero point to collect key parameters such as reflection coefficient, standing wave ratio, power transmission efficiency and matching loss to form the first data.

[0050] The actual collected value in the first data is compared with the steady-state prediction result, and the deviation value of each data point is calculated. The deviation value is the actual value minus the predicted steady-state value. According to the calculated deviation, a compensation algorithm is designed for adjustment. For example, linear compensation, nonlinear compensation or weighted average compensation method can be used. Each data point is compensated based on the compensation algorithm to make it closer to the actual steady-state situation. The compensated data points are combined to form updated data.

[0051] The updated data obtained is used as the final data collection result. The required key parameters are extracted from the updated data set and substituted into the comprehensive evaluation formula to calculate the comprehensive evaluation result of each data point and summarize the comprehensive evaluation result of each data point to form an overall evaluation result set.

[0052] Furthermore, the method further comprises:

[0053] According to the time interval clustering result, steady-state time nodes are configured; tolerant sampling points are configured with the steady-state time nodes, multi-point data collection within the time interval clustering result is performed based on the tolerant sampling points, and second data is constructed according to the verification result of the multi-point data collection result; data authentication of the updated data is performed through the second data, and compensation for the comprehensive evaluation result of the impedance characteristics is performed based on the data authentication result.

[0054] Reviewing the time interval clustering, each clustered time interval represents a different operating state of the system. Identify the stable state period in each time interval. These periods are periods when the system parameters are relatively stable and have little change. Determine which key parameters are in a stable state in each time interval. Based on historical data and system performance, determine the stable value range of each key parameter, that is, the range in which the parameter fluctuation is small. Based on the determined stable state, select time nodes that can represent the steady-state characteristics of the entire time interval. These time nodes can fully reflect the steady-state behavior of the system in the time interval, and mark the selected steady-state time nodes in the time interval.

[0055] Set a tolerance range for each steady-state time node, that is, add a certain time period before and after the time node to cover more data points. For example, you can set a tolerance range of 1 minute or longer before and after, and evenly distribute sampling points within each tolerance range to ensure that the entire tolerance range is covered. The distribution of sampling points should be able to capture the changes around the steady-state time node.

[0056] Configure data acquisition equipment to perform continuous and high-frequency sampling within the set tolerance range, record key parameters including reflection coefficient, standing wave ratio, power transmission efficiency and matching loss, and form multi-point data acquisition results.

[0057] The consistency of multi-point data collection results is verified to ensure that the data is logically and physically reasonable. The verification results of the multi-point data are analyzed to identify the patterns and trends in the data. The verified multi-point data are integrated to obtain the second data, which can fully reflect the steady-state characteristics of the system in each time interval.

[0058] The updated data is compared with the second data, and the deviation of each data point is calculated. If the deviation is within the acceptable range, the data is authenticated. The impedance characteristic comprehensive evaluation results of the updated data are compared with the evaluation results of the second data, and the deviations in the evaluation results are identified. According to the deviation analysis results, a reasonable compensation method is designed, such as linear compensation, weighted average compensation, etc. The compensation method is applied to the preliminary comprehensive evaluation results and adjusted to make it closer to the comprehensive evaluation results of the second data after data authentication.

[0059] Furthermore, the step of constructing the second data according to the verification result of the multi-point data collection result further includes:

[0060] The multi-point data collection results are clustered to establish cluster centers; data distance analysis is performed on the multi-point data collection results using the cluster centers, and data elimination is performed based on the data distance analysis results; the multi-point data collection results after data elimination are used as verification results, mean calculation is performed, and second data is constructed based on the mean calculation results.

[0061] Perform preliminary cleaning on the multi-point data collection results, handle missing values ​​and outliers, and normalize the data to ensure that data of different dimensions are comparable when clustering. Select a suitable clustering algorithm according to the data characteristics, such as K-means clustering, DBSCAN, hierarchical clustering, etc., set appropriate parameters for the selected clustering algorithm, such as the number of clusters K in K-means, the minimum number of samples and neighborhood radius in DBSCAN, etc., use the selected clustering algorithm to cluster the multi-point data collection results, obtain the cluster label and cluster center position of each data point, and form the cluster center.

[0062] Select a suitable distance measurement method, such as Euclidean distance, Manhattan distance, etc., calculate the distance from each data point to the cluster center of the cluster to which it belongs, form a distance distribution, and analyze the distance distribution of all data points. Specifically, set a distance threshold and identify abnormal points whose distance exceeds the threshold. These points may be noise or abnormal data. According to the results of the distance distribution analysis, data elimination is performed, such as retaining data points whose distance is within the threshold and eliminating data points that exceed the threshold.

[0063] The multi-point data collection results after data elimination are used as verification results, and mean calculation is performed to obtain the average value of each key parameter. The calculated mean data are integrated to obtain the second data.

[0064] Furthermore, clustering the load by time intervals according to the steady-state prediction result and establishing the time interval clustering result further includes:

[0065] An abnormal steady state is marked for the time interval clustering result, and a real-time response verification of the time interval clustering result is performed through the abnormal steady state; if the abnormal steady state breaks through the preset steady state threshold, the real-time response verification is passed, and when the time node reaches the corresponding time clustering result, a real-time response control instruction is generated; the real-time response control instruction is used to call the RF power supply to perform real-time response power supply control management.

[0066] According to the load characteristics and system requirements, select key indicators for identifying abnormal steady states, such as reflection coefficient, standing wave ratio, power transmission efficiency, matching loss, etc. Use historical data to analyze the normal fluctuation range and abnormal conditions of each indicator to determine reasonable thresholds.

[0067] The real-time monitoring data is compared with the set threshold to identify abnormal steady states. That is, if an indicator exceeds the threshold range, it is marked as an abnormal steady state. The detected abnormal steady state is verified in real time, and the abnormal situations that pass the verification are marked, and the specific time, duration and related parameters of the abnormality are recorded.

[0068] Continuously monitor key parameters. If a parameter value is detected to exceed the preset threshold, the event is recorded. When a parameter value exceeds the preset threshold and the existence of an abnormal steady state is confirmed through verification, it is determined that the trigger condition is met. According to the type and severity of the abnormality, a corresponding response strategy is formulated, such as adjusting the matching network, adjusting the power supply parameters, etc. Based on the response strategy, specific real-time response control instructions are generated, including which parameters to adjust, what the target value of the adjustment is, etc. According to the time interval clustering results, the specific time node where the control instruction needs to be generated is determined.

[0069] The generated real-time response control instructions are transmitted to the RF power control system. After receiving the instructions, the RF power control system immediately performs corresponding operations, such as adjusting the capacitance value, inductance value, changing the supply voltage or current, etc., to solve the detected abnormal steady-state problems and ensure the stable and efficient operation of the system.

[0070] Furthermore, the method further comprises:

[0071] A monitoring window is configured, and verification monitoring of the load is performed through the monitoring window to construct a verification monitoring data set; verification feedback is constructed through the verification monitoring data set, and impedance matching model optimization is performed through the verification feedback.

[0072] Configure the monitoring window. Specifically, determine the time length of the monitoring window based on system requirements and load characteristics. For example, it can be set to a continuous monitoring interval of several seconds, minutes, or longer, and define the interval time between continuous monitoring windows to ensure that all critical time periods are covered to avoid data omissions.

[0073] Configure the required monitoring equipment, such as voltmeter, ammeter, power meter, network analyzer, etc., collect load data in real time according to the monitoring window, perform data cleaning, and organize the cleaned data into a verification monitoring data set.

[0074] Extract key features such as reflection coefficient, standing wave ratio, power transmission efficiency, etc. from the verification monitoring data set. Use the verification monitoring data set to detect abnormal conditions in system operation, record abnormal events and their related parameters, and generate verification feedback, including verification monitoring results, abnormal conditions, response measures and their effects.

[0075] The verification monitoring data set is used to evaluate the performance indicators of the existing impedance matching model, such as accuracy, stability, response speed, etc. According to the verification feedback, the problems and deficiencies in the impedance matching model are identified. According to the results of the verification feedback, the parameters of the matching model are adjusted, such as adjusting the weight coefficient, learning rate, etc. The optimized impedance matching model is deployed to the RF power control system to replace the old model, thereby improving the performance and stability of the RF power supply.

[0076] In summary, the adaptive load matching method of a radio frequency power supply provided in the embodiment of the present application has the following technical effects:

[0077] By establishing a load data set and extracting load characteristics, and constructing a steady-state prediction network, the steady-state characteristics under different load conditions can be accurately predicted, thereby improving the matching accuracy; through random frame sampling and steady-state prediction, real-time load monitoring and prediction are performed to achieve rapid response and adjustment to load changes, ensure the stable operation of the system under dynamic load conditions, and improve energy transmission efficiency; by clustering the load time intervals and collecting data within the time interval clustering results, the load data is comprehensively collected and utilized, the data utilization rate is improved, and comprehensive data support is provided for the matching model; by configuring a comprehensive evaluation formula, a comprehensive impedance characteristic comprehensive evaluation is performed on the data collection results, and a comprehensive evaluation result is generated, and the optimized matching parameters are generated in combination with the impedance matching model, which significantly improves the matching effect and improves the overall performance and stability of the system. In general, the adaptive load matching method of the RF power supply solves the problems of inaccurate load matching, insufficient dynamic response, low data utilization, and unsatisfactory matching effect in the existing technology through systematic load data analysis, steady-state prediction, and comprehensive evaluation, and improves the overall performance and stability of the RF power supply.

[0078] Based on the same inventive concept as the adaptive load matching method of a radio frequency power supply in the aforementioned embodiment, Figure 2 As shown, an embodiment of the present application provides an adaptive load matching device for a radio frequency power supply, the device comprising:

[0079] A feature extraction module 10 is used to establish a load data set of the load, extract load features according to the load data set, establish a load feature extraction result, and build a steady-state prediction network for the load with the load feature extraction result; a steady-state prediction module 20 is used to perform random frame sampling on the load, establish a frame sampling result, perform data preprocessing on the frame sampling result, and input it into the steady-state prediction network to perform steady-state prediction and generate a steady-state prediction result; a data acquisition module 30 is used to perform time interval clustering of the load according to the steady-state prediction result, establish a time interval clustering result, and re- A new data collection is performed within the time interval clustering result to generate a data collection result; a comprehensive evaluation module 40, the comprehensive evaluation module 400 is used to configure a comprehensive evaluation formula, perform a comprehensive evaluation of the impedance characteristics of the data collection result based on the comprehensive evaluation formula, and generate an impedance characteristic comprehensive evaluation result; a matching parameter generation module 50, the matching parameter generation module 50 is used to establish an impedance matching model of the RF power supply, input the impedance characteristic comprehensive evaluation result into the impedance matching model, and generate matching parameters of the RF power supply; a power supply control management module 60, the power supply control management module 60 is used to control the RF power supply according to the matching parameters to perform power supply control management of the load.

[0080] Furthermore, the device also includes an impedance characteristic comprehensive evaluation module to perform the following operation steps:

[0081] The comprehensive evaluation formula is as follows:

[0082]

[0083] Among them, E represents the comprehensive evaluation result of impedance characteristics, |Γ| represents the reflection coefficient, which is directly obtained through data acquisition results, and VSWR represents the standing wave ratio. η represents the power transmission efficiency, ML represents the matching loss, w1, w2, w3, w4 are weight coefficients of the characteristics, and w1+w2+w3+w4=1; the impedance characteristics of the data acquisition results are comprehensively evaluated according to the comprehensive evaluation formula.

[0084] Furthermore, the device also includes a comprehensive evaluation module to perform the following operation steps:

[0085] Configure time zero for the time interval clustering result, perform data collection at the time zero, and establish first data; call the steady-state prediction result, and perform stable value compensation of the first data in the time interval clustering result based on the steady-state prediction result to generate updated data; use the updated data as the data collection result to perform a comprehensive evaluation of impedance characteristics.

[0086] Furthermore, the device further comprises an evaluation result compensation module to perform the following operation steps:

[0087] According to the time interval clustering result, steady-state time nodes are configured; tolerant sampling points are configured with the steady-state time nodes, multi-point data collection within the time interval clustering result is performed based on the tolerant sampling points, and second data is constructed according to the verification result of the multi-point data collection result; data authentication of the updated data is performed through the second data, and compensation for the comprehensive evaluation result of the impedance characteristics is performed based on the data authentication result.

[0088] Furthermore, the device further includes a second data construction module to perform the following operation steps:

[0089] The multi-point data collection results are clustered to establish cluster centers; data distance analysis is performed on the multi-point data collection results using the cluster centers, and data elimination is performed based on the data distance analysis results; the multi-point data collection results after data elimination are used as verification results, mean calculation is performed, and second data is constructed based on the mean calculation results.

[0090] Furthermore, the device also includes a power supply control management module to perform the following operation steps:

[0091] An abnormal steady state is marked for the time interval clustering result, and a real-time response verification of the time interval clustering result is performed through the abnormal steady state; if the abnormal steady state breaks through the preset steady state threshold, the real-time response verification is passed, and when the time node reaches the corresponding time clustering result, a real-time response control instruction is generated; the real-time response control instruction is used to call the RF power supply to perform real-time response power supply control management.

[0092] Furthermore, the device also includes a model optimization module to perform the following operation steps:

[0093] A monitoring window is configured, and verification monitoring of the load is performed through the monitoring window to construct a verification monitoring data set; verification feedback is constructed through the verification monitoring data set, and impedance matching model optimization is performed through the verification feedback.

[0094] Through the above-mentioned detailed description of an adaptive load matching method of an RF power supply in this specification, those skilled in the art can clearly understand an adaptive load matching device of an RF power supply in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0095] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An adaptive load matching method for a radio frequency power supply, characterized in that: The method comprises: Establishing a load data set of the load, extracting load features according to the load data set, establishing a load feature extraction result, and building a steady-state prediction network of the load with the load feature extraction result; Performing random frame sampling on the load, establishing a frame sampling result, performing data preprocessing on the frame sampling result, and then inputting the result into the steady-state prediction network, performing steady-state prediction, and generating a steady-state prediction result; Clustering the load by time interval according to the steady-state prediction result, establishing the time interval clustering result, and re-collecting the data within the time interval clustering result to generate the data collection result; Configure a comprehensive evaluation formula, perform a comprehensive evaluation of the impedance characteristics of the data acquisition results based on the comprehensive evaluation formula, and generate a comprehensive evaluation result of the impedance characteristics; Establishing an impedance matching model of the radio frequency power supply, inputting the comprehensive evaluation result of the impedance characteristics into the impedance matching model, and generating matching parameters of the radio frequency power supply; Controlling the radio frequency power supply to control and manage the power supply of the load according to the matching parameters; The configuring a comprehensive evaluation formula, performing a comprehensive evaluation of the impedance characteristics of the data acquisition results based on the comprehensive evaluation formula, and generating a comprehensive evaluation result of the impedance characteristics, further includes: The comprehensive evaluation formula is as follows: ； in, Characterize the comprehensive evaluation results of impedance characteristics, Characterize the reflection coefficient and obtain it directly through data acquisition results. Characterize the standing wave ratio, , Characterizes the power transfer efficiency, Characterize the matching loss, , , , , is the weight coefficient of the feature, and ; A comprehensive evaluation of the impedance characteristics of the data acquisition results is performed according to the comprehensive evaluation formula.

2. The adaptive load matching method of a radio frequency power supply according to claim 1, characterized in that: The impedance characteristic comprehensive evaluation of the data acquisition results based on the comprehensive evaluation formula also includes: configuring a time zero point for the time interval clustering result, performing data collection at the time zero point, and establishing first data; Calling the steady-state prediction result, and performing stable value compensation of the clustering result of the first data in the time interval based on the steady-state prediction result to generate updated data; The updated data is used as a data collection result to perform a comprehensive evaluation of impedance characteristics.

3. The adaptive load matching method of a radio frequency power supply according to claim 2, characterized in that: The method further comprises: Configuring steady-state time nodes according to the time interval clustering results; configuring a tolerant sampling point with the steady-state time node, performing multi-point data collection within the time interval clustering result based on the tolerant sampling point, and constructing second data according to a verification result of the multi-point data collection result; The data updating data is authenticated by the second data, and the impedance characteristic comprehensive evaluation result is compensated based on the data authentication result.

4. The adaptive load matching method of a radio frequency power supply according to claim 3, characterized in that: The step of constructing the second data according to the verification result of the multi-point data collection result further includes: Performing data clustering on the multi-point data collection results to establish a cluster center; Performing data distance analysis on multi-point data collection results using the cluster center, and performing data elimination according to the data distance analysis results; The multi-point data collection results after data elimination are used as verification results, mean calculation is performed, and second data is constructed according to the mean calculation results.

5. The adaptive load matching method of a radio frequency power supply according to claim 1, characterized in that: The step of clustering the load by time intervals according to the steady-state prediction result and establishing the time interval clustering result further includes: An abnormal steady state is marked for the time interval clustering result, and a real-time response verification of the time interval clustering result is performed through the abnormal steady state; If the abnormal steady state breaks through the preset steady state threshold, the real-time response verification is passed, and when the time node reaches the corresponding time clustering result, a real-time response control instruction is generated; The real-time response control instruction is used to call the radio frequency power supply to perform real-time response power supply control management.

6. The adaptive load matching method of a radio frequency power supply according to claim 1, characterized in that: The method further comprises: Configure a monitoring window, perform verification monitoring of the load through the monitoring window, and construct a verification monitoring data set; A verification feedback is constructed through the verification monitoring data set, and an impedance matching model is optimized through the verification feedback.

7. An adaptive load matching device for a radio frequency power supply, characterized in that: A method for implementing an adaptive load matching method for a radio frequency power supply according to any one of claims 1 to 6, the device comprising: A feature extraction module, wherein the feature extraction module is used to establish a load data set of the load, extract load features according to the load data set, establish a load feature extraction result, and build a steady-state prediction network of the load with the load feature extraction result; A steady-state prediction module, the steady-state prediction module is used to perform random frame sampling on the load, establish a frame sampling result, perform data preprocessing on the frame sampling result, input it into the steady-state prediction network, perform steady-state prediction, and generate a steady-state prediction result; A data collection module, the data collection module is used to perform time interval clustering of the load according to the steady-state prediction result, establish a time interval clustering result, and re-collect data within the time interval clustering result to generate a data collection result; A comprehensive evaluation module, the comprehensive evaluation module is used to configure a comprehensive evaluation formula, perform a comprehensive evaluation of the impedance characteristics of the data acquisition results based on the comprehensive evaluation formula, and generate a comprehensive evaluation result of the impedance characteristics; A matching parameter generation module, wherein the matching parameter generation module is used to establish an impedance matching model of the radio frequency power supply, input the impedance characteristic comprehensive evaluation result into the impedance matching model, and generate matching parameters of the radio frequency power supply; A power supply control management module is used to control the RF power supply to perform power supply control management of the load according to the matching parameters.

Images