Radio frequency power supply service life simulation prediction method, device, equipment and medium

Abstract

The invention relates to a radio frequency power supply service life simulation prediction method, device, equipment and medium, and the method comprises the steps: obtaining the observation data of a radio frequency power supply, and extracting the features of the radio frequency power supply in a high-frequency state as a source feature set; expanding the source feature set into an expanded feature set by adopting a radio frequency power supply parameter tree; fusing the source feature set and the extended feature set to form a target feature set; performing model training on a machine learning model by using the target feature set; and predicting the change trend of a preset index along with time by using the trained model, thereby obtaining the service life estimation of the radio frequency power supply. According to the method, the characteristics in the high-frequency state are extracted, the characteristics are expanded, and more characteristics related to the service life of the radio-frequency power supply are extracted, so that the problem of data sparsity is solved, the accuracy and the reliability of estimating the service life of the radio-frequency power supply are improved, and the robustness of the model is effectively improved.

Description

Technical Field

[0001] The present disclosure relates to the field of electrical performance testing, and in particular to a method, device, equipment and medium for simulating and predicting the service life of a radio frequency power supply. Background Art

[0002] RF power supply is a power supply device used to generate RF signals. It is a core component of many key technologies and applications and plays a vital role in engineering and manufacturing. The simulation prediction of the service life of RF power supply is very important in engineering and manufacturing.

[0003] First of all, observation data is very important for the simulation prediction of RF power supply service life. Life simulation requires a large amount of experimental data to verify and calibrate the model. However, due to the changeable working conditions and environment of RF power supply, it is difficult to obtain sufficient and comprehensive experimental data, which will lead to a certain deviation between the simulation results and the actual situation; some data may be very sparse or missing, which will cause the simulation prediction of RF power supply life to fail to work properly. Therefore, data sparsity is one of the challenges faced by the simulation prediction of RF power supply service life.

[0004] In addition, in the prior art, life simulation, such as battery life simulation, often collects many parameters. When too many parameters are introduced, the complexity of the simulation model will increase significantly. Complex models often require more computing resources and may cause the simulation speed to slow down. This method cannot focus on the most critical parameters that affect the RF power supply life simulation.

[0005] Therefore, a method is needed to accurately and reliably simulate and predict the service life of RF power supplies. Summary of the invention

[0006] The present invention provides a method, device, equipment and medium for simulating and predicting the service life of a radio frequency power supply, which are used to accurately and reliably perform a method for simulating and predicting the service life of a radio frequency power supply.

[0007] In a first aspect, the present disclosure provides a method for simulating and predicting the service life of a radio frequency power supply, comprising:

[0008] Obtain observation data of the radio frequency power source, and extract features of the radio frequency power source in a high frequency state as a source feature set;

[0009] The source feature set and the extended feature set are merged to form a target feature set; wherein the extended feature set is generated in the following manner: each feature in the source feature set is used as a tree node to form a radio frequency power supply parameter tree; 2n features in the radio frequency power supply parameter tree are selected from high to low according to the frequency parameters of the features as n node pairs, where n is the feature expansion multiple, and each node pair includes a starting node and a corresponding ending node, and the difference in the frequency parameters of the starting node and the corresponding ending node meets a preset condition; the optimal path between the starting node and the corresponding ending node in each node pair is obtained, the features of the nodes on the optimal path are interpolated, and the features at the maximum curvature are selected as the extended feature set;

[0010] The target feature set is used to train the machine learning model; the trained model is used to predict the changing trend of the preset indicator over time, so as to obtain an estimate of the service life of the radio frequency power supply.

[0011] According to the RF power supply service life simulation prediction method provided by the present disclosure, extracting the characteristics of the RF power supply under high frequency state includes:

[0012] The observation data of the radio frequency power supply is sampled and processed by using a sliding sampling window, and the length of the sliding sampling window is adjusted in real time with the power of the radio frequency power supply;

[0013] Calculating the variance value of the slope of all sampling points in the sliding sampling window, and determining the high-frequency state range according to the variance value of the slope;

[0014] The characteristics of the radio frequency power supply in the high frequency state are extracted according to the high frequency state range.

[0015] According to the RF power supply service life simulation prediction method provided by the present disclosure, each feature in the source feature set is used as a tree node to form a RF power supply parameter tree, including:

[0016] Each feature in the source feature set is sorted according to the sampling time of the feature, the first sampled feature is used as the root node, and a radio frequency power supply parameter tree is constructed according to the sorted features.

[0017] According to the RF power supply service life simulation prediction method provided by the present disclosure, fusing the source feature set and the extended feature set to form a target feature set includes:

[0018] Setting a first weight for each feature in the source feature set according to the magnitude of a frequency parameter;

[0019] Setting a second weight for each feature in the extended feature set according to the size of the curvature, wherein the first weight is greater than the second weight;

[0020] Selecting a feature in the source feature set in which the first weight is less than a first threshold and a feature in the extended feature set in which the second weight is greater than a second threshold as a training set of a target feature set;

[0021] The remaining features in the target feature set are used as a test set.

[0022] According to the RF power supply service life simulation prediction method provided by the present disclosure, the machine learning model includes a convolution layer, a pooling layer and an output layer connected in sequence;

[0023] The convolution layer calculates the correlation between multiple features for the target feature set, and optimizes the weight according to the correlation;

[0024] The pooling layer reduces the feature quantity by reducing the feature dimension;

[0025] The output layer outputs the training result.

[0026] According to the RF power supply service life simulation prediction method provided by the present disclosure, the convolution layer calculates the correlation between multiple features for the target feature set, including:

[0027] Calculate the mean and variance of the two features respectively;

[0028] Multiply the variance of two features;

[0029] The multiplied variance is divided by the sum of the means and the square root is taken to obtain the correlation;

[0030] The pooling layer reduces the feature quantity by reducing the feature dimension, including:

[0031] The maximum value within each pooling window is retained through the maximum pooling operation;

[0032] The average pooling operation retains the average value within each pooling window;

[0033] The two pooled features are weighted and summed according to certain weights to obtain the comprehensive pooled features;

[0034] The sizes of the pooling windows of the maximum pooling operation and the average pooling operation are dynamically adjusted according to the frequency parameters of the input features.

[0035] According to the RF power supply service life simulation prediction method provided by the present disclosure, after obtaining the observation data of the RF power supply, the method further includes:

[0036] analyzing data patterns and structures between different data sources in the observational data;

[0037] Determine the matching rules and mapping relationships between data;

[0038] According to the matching rules and mapping relationships, the multi-source heterogeneous observation data of the RF power supply are integrated into a unified data storage warehouse, and the timestamps are synchronized so that all data have the same scale.

[0039] In a second aspect, the present disclosure further provides a device for simulating and predicting the service life of a radio frequency power supply, comprising:

[0040] The source feature module is used to obtain the observation data of the RF power supply and extract the features of the RF power supply in the high-frequency state as the source feature set;

[0041] An extension module, used for fusing the source feature set and the extended feature set to form a target feature set; wherein the extended feature set is generated in the following manner: taking each feature in the source feature set as a tree node to form a radio frequency power supply parameter tree; selecting 2n features in the radio frequency power supply parameter tree from high to low according to the frequency parameters of the features as n node pairs, where n is the feature expansion multiple, each node pair includes a starting node and a corresponding ending node, and the difference in the frequency parameters of the starting node and the corresponding ending node meets a preset condition; obtaining the optimal path between the starting node and the corresponding ending node in each node pair, interpolating the features of the nodes on the optimal path, and selecting the feature at the maximum curvature as the extended feature set;

[0042] A prediction module is used to use the target feature set to train the machine learning model; use the trained model to predict the changing trend of the preset indicator over time, so as to obtain an estimate of the service life of the RF power supply.

[0043] Compared with the prior art, the RF power supply service life simulation prediction method disclosed in the present invention obtains the observation data of the RF power supply, and extracts the features of the RF power supply in the high-frequency state as the source feature set; uses the RF power supply parameter tree to expand the source feature set into an extended feature set; fuses the source feature set and the extended feature set to form a target feature set; uses the target feature set to train the machine learning model; uses the trained model to predict the changing trend of the preset indicator over time, thereby obtaining the service life estimation of the RF power supply. The present invention solves the data sparsity problem by extracting features in the high-frequency state, performing feature expansion, and extracting more features related to the life of the RF power supply, thereby improving the accuracy and reliability of the RF power supply service life estimation and effectively improving the robustness of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the present disclosure, the drawings required for use in the embodiments or prior art descriptions are briefly introduced below. Obviously, the drawings described below are some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0045] Figure 1 It is a flow chart of a method for simulating and predicting the service life of a radio frequency power supply provided by the present disclosure;

[0046] Figure 2 It is a schematic diagram of the sources of observation data provided by the present disclosure;

[0047] Figure 3 It is a schematic diagram of a radio frequency power supply service life simulation prediction device provided by the present disclosure;

[0048] Figure 4 A schematic diagram of the electronic device provided in the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solutions and advantages of the present disclosure clearer, the technical solutions in the present disclosure will be clearly and completely described below in conjunction with the drawings in the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.

[0050] RF power supply is a device used to generate RF signals, which is widely used in wireless communication, radar, medical diagnosis, scientific research, industrial heating, etc. Its main components include: oscillator, power amplifier, tuning network, control circuit, connector and interface.

[0051] The working principle of RF power supply is based on the oscillation circuit. The oscillator generates a RF signal of the basic frequency, which is then amplified by the power amplifier and adjusted by the tuning network to finally output the RF signal of the required frequency and power. The frequency parameters of RF power supply are usually 3kHz to 300GHz.

[0052] like Figure 1 FIG. 1 is a flowchart of a method for simulating and predicting the service life of a radio frequency power supply provided by the present disclosure, the method comprising:

[0053] Step S110: obtaining observation data of the radio frequency power source, and extracting features of the radio frequency power source in a high frequency state as a source feature set;

[0054] Specifically, it is first necessary to obtain the RF power supply observation data, which mainly includes the frequency parameters, output power, temperature, voltage, current and other parameters of the RF power supply.

[0055] High-frequency working state is the most critical factor affecting the service life of RF power supply, affecting the service life of RF power supply in terms of thermal effect, electromagnetic interference, frequency stability, etc. Focusing on the most critical parameters affecting the simulation of RF power supply life, it becomes particularly important to extract the characteristics of RF power supply in high-frequency state, which can improve the accuracy of model calculation.

[0056] In one embodiment, the collected raw observation data may contain noise, outliers, or missing values, and need to be preprocessed to improve data quality. Preprocessing the observation data of the RF power supply includes:

[0057] Detect and remove missing data points and fill in missing values ​​through spline interpolation method; identify and handle outliers through sliding window algorithm of mean shift clustering method;

[0058] Specifically, the identification and processing of outliers by the sliding window algorithm of the mean shift clustering method includes:

[0059] (1) Determine the size and parameters of the sliding window: The size and shape of the sliding window (e.g., circular or rectangular) will affect the clustering results;

[0060] (2) Initialize the sliding window: select a random point as the center of the sliding window and set the radius or size of the window;

[0061] (3) Calculate the mean of the data in the window: In each iteration, calculate the mean of all data points in the sliding window. This mean will become the new center point;

[0062] (4) Move the sliding window: Move the center of the sliding window to the newly calculated mean point and recalculate the mean of the data in the window. Repeat this process until the center point of the window no longer changes significantly, i.e. convergence;

[0063] (5) Clustering: When multiple sliding windows overlap, the window containing the most points is retained and clustered according to the sliding window in which the data points are located;

[0064] (6) Calculate the statistics of each cluster: For each cluster, calculate its mean, standard deviation and other statistics, which can reflect the distribution characteristics of each cluster;

[0065] (7) Define the threshold of outliers: Define the threshold of outliers based on the statistics of each cluster. For example, data points that are more than a certain standard deviation (such as 2 or 3) away from the cluster mean can be considered outliers;

[0066] (8) Identify and handle outliers: Identify outliers in each cluster based on the defined threshold. These outliers can then be processed according to specific needs, such as deletion, replacement, or labeling.

[0067] Step S120: fusing the source feature set and the extended feature set to form a target feature set; wherein the extended feature set is generated in the following manner: taking each feature in the source feature set as a tree node to form a radio frequency power supply parameter tree; selecting 2n features in the radio frequency power supply parameter tree from high to low according to the frequency parameters of the features as n node pairs, where n is the feature expansion multiple, each node pair includes a starting node and a corresponding ending node, and the difference in frequency parameters between the starting node and the corresponding ending node meets a preset condition; obtaining the optimal path between the starting node and the corresponding ending node in each node pair, interpolating the features of the nodes on the optimal path, and selecting the feature at the maximum curvature as the extended feature set;

[0068] Specifically, each node of the RF power supply parameter tree represents a feature of the RF power supply, including the frequency parameters, output power, temperature, voltage, current and other parameters of the RF power supply. For example, the data format of a feature is [X i，1 , X i，2 , X i，3 , ...], X i，1 is the frequency parameter of the i-th feature, X i，2 is the output power of the i-th feature, X i，3 is the temperature of the i-th feature, and so on. Using the RF power parameter tree to store the features of the RF power supply has many advantages in terms of data management, query efficiency, and feature extraction. The RF power parameter tree provides a hierarchical data management method that can classify and group the features of the RF power supply according to different dimensions and attributes, making the organization and storage of features more organized and convenient for users to query and access features as needed. In addition, through the parent-child relationship between nodes, the required features can be quickly located.

[0069] Different types of RF power parameter trees can be selected according to the specific feature set and user application requirements. For example, if the correlation between features is high, a correlation-based clustering tree can be selected; if fast query is required, a balanced tree can be selected. At the same time, there are many ways to build a tree, such as hierarchical clustering, K-means clustering, etc. This method can build an RF power parameter tree based on the similarity or distance between features.

[0070] Furthermore, when selecting a node pair, the difference in the frequency parameters of the node pair needs to meet the preset conditions, that is, the frequency difference between the two nodes at least reaches a certain threshold to ensure that the extended features can cover different frequency ranges. In addition, it is also necessary to ensure that the positions of the starting node and the corresponding end node in the RF power parameter tree are sufficiently dispersed, that is, the depth difference or path difference of the calculated node in the RF power parameter tree meets the preset conditions. Sufficient dispersion can ensure that more feature space is covered by searching for the optimal path in subsequent steps, thereby improving the accuracy of the prediction. Specifically, first traverse all nodes in the RF power parameter tree, and for each node, regard it as a candidate node; starting from the candidate node, traverse its child nodes to find possible corresponding end nodes; for each candidate node and one of its potential corresponding end nodes, calculate its frequency difference and / or depth difference; if the frequency difference and / or depth difference is greater than or equal to the preset threshold, the node pair (candidate node and potential corresponding end node) is used as the starting node and the corresponding end node.

[0071] Furthermore, in each node pair, the optimal path between the starting node and the corresponding end node is searched. The optimal path between the starting node and the corresponding end node is determined based on the frequency difference, and a path is found in the RF power supply parameter tree so that the sum of the frequency change or frequency difference on the path is maximized. Specifically, starting from the starting node, all possible paths are traversed to reach the corresponding end node. For each path, the variance values ​​of all nodes on the path are calculated, and the path with the largest variance value is selected as the optimal path. Determining the optimal path through frequency difference can expand diversified key points, help obtain more comprehensive data and information, and thus improve the accuracy and reliability of data analysis.

[0072] Furthermore, the features of the nodes on the optimal path are interpolated, and the features at the maximum curvature are selected as the extended feature set. Specifically, feature data is extracted from each node on the optimal path, and the feature data are frequency parameters, output power, temperature, voltage, current and other parameters of the RF power supply; a plurality of curves of different parameters are constructed by a linear interpolation method, such as a frequency parameter interpolation curve and an output power interpolation curve; the maximum curvature of the plurality of parameter curves is obtained, and the value at the maximum curvature is used as the extended feature, such as the frequency parameter Y at the maximum curvature of the frequency parameter interpolation curve t，1 , the output power Y at the point where the curvature of the output power interpolation curve is maximum t，2 As the data of the extended features, the data format of each extended feature is [Y t，1 , Y t，2 , Y t，3 , ...], Y t，1 is the frequency parameter of the tth extended feature, Y t，2 is the output power of the tth extended feature, Yt，3 is the temperature of the tth extended feature, and so on.

[0073] Step S130: using the target feature set to perform model training on the machine learning model; using the trained model to predict the changing trend of the preset indicator over time, thereby obtaining an estimate of the service life of the RF power supply.

[0074] Specifically, first select a suitable machine learning model, such as a neural network, support vector machine, random forest, etc., and use a target feature set with rich information to train the model. During the training process, the machine learning model will learn the relationship between the target feature set and the service life of the RF power supply, and automatically adjust the model parameters to minimize the prediction error. Through multiple iterations and optimizations, a machine learning model that can accurately predict the service life of the RF power supply is finally obtained. At the same time, the model can also dynamically estimate the service life of the RF power supply based on the changing trend of the preset indicators to more accurately reflect the actual operating status of the RF power supply.

[0075] Among them, the preset indicators are key parameters or variables that can reflect the performance, health status or potential problems of the RF power supply, mainly including the output power, efficiency, temperature, working time, load conditions, failure rate and other parameters of the RF power supply.

[0076] Compared with the prior art, the present invention adopts the above scheme to obtain the observation data of the RF power supply, and extracts the features of the RF power supply in the high-frequency state as the source feature set; uses the RF power supply parameter tree to expand the source feature set into an extended feature set; fuses the source feature set and the extended feature set to form a target feature set; uses the target feature set to train the machine learning model; uses the trained model to predict the changing trend of the preset indicators over time, thereby obtaining the service life estimation of the RF power supply. The present invention solves the data sparsity problem by extracting the features in the high-frequency state, performing feature expansion, and extracting more features related to the life of the RF power supply, thereby improving the accuracy and reliability of the RF power supply service life estimation and effectively improving the robustness of the model.

[0077] In one embodiment, the extracting of the features of the RF power supply under the high frequency state in step S110 includes:

[0078] The observation data of the radio frequency power supply is sampled and processed by using a sliding sampling window, and the length of the sliding sampling window is adjusted in real time with the power of the radio frequency power supply;

[0079] Calculating the variance value of the slope of all sampling points in the sliding sampling window, and determining the high-frequency state range according to the variance value of the slope;

[0080] The characteristics of the radio frequency power supply in the high frequency state are extracted according to the high frequency state range.

[0081] Specifically, the sliding sampling window is a dynamic data processing technology that can intercept a fixed or variable length of data in a continuous data stream for processing and analysis. In the RF power system, due to the real-time changes in operating frequency and power, the observed data will also show dynamic changes. The real-time adjustment of the length of the sliding sampling window can adapt to the data sampling requirements under different operating frequencies and powers.

[0082] The observation data of the RF power supply in the high-frequency state will show large fluctuations and rapid changes, and the change in slope can reflect the fluctuation. Calculating the variance of the slope of all sampling points in the sliding sampling window can effectively quantify the size of the fluctuation, thereby accurately determining the range of the high-frequency state. When the variance value exceeds a preset threshold, it can be determined that the RF power supply is in the high-frequency state.

[0083] In one embodiment, the step S120 of using each feature in the source feature set as a tree node to form a radio frequency power supply parameter tree includes:

[0084] Each feature in the source feature set is sorted according to the sampling time of the feature, the first sampled feature is used as the root node, and a radio frequency power supply parameter tree is constructed according to the sorted features.

[0085] In one embodiment, fusing the source feature set and the extended feature set to form a target feature set in step S120 includes:

[0086] Setting a first weight for each feature in the source feature set according to the magnitude of a frequency parameter;

[0087] Setting a second weight for each feature in the extended feature set according to the size of the curvature, wherein the first weight is greater than the second weight;

[0088] Selecting a feature in the source feature set in which the first weight is less than a first threshold and a feature in the extended feature set in which the second weight is greater than a second threshold as a training set of a target feature set;

[0089] The remaining features in the target feature set are used as a test set.

[0090] Specifically, the source feature set is obtained based on direct observation or measurement, is directly related to the actual performance of the RF power supply, can directly reflect the physical characteristics and working state of the RF power supply, and has high information content and reliability. The extended feature set is often obtained by performing some kind of conversion or calculation on the source feature, and has relatively low information content and may contain redundancy or noise, so the first weight is set to be greater than the second weight.

[0091] During the model training process, the data needs to be divided into a training set and a test set. The training set is used to train the model, while the test set is used to evaluate the performance of the model. In order to effectively utilize the limited RF power features, the training set and test set of the target feature set can be selected according to the feature weights.

[0092] First, the feature whose first weight is less than the first threshold in the source feature set is selected. Although the frequency parameter of this feature is not very high, it still contains valuable information. Including it in the training set can help the model learn the basic characteristics of RF power supply.

[0093] Secondly, the feature with the second weight greater than the second threshold in the extended feature set is selected, which provides additional information about the performance of the RF power supply and has a high enough weight, indicating that it is important in the model. Including it in the training set can further improve the performance of the model.

[0094] Finally, the remaining features in the target feature set are used as the test set. These features may include some features that were not fully considered in the training set, or some features used to verify the generalization ability of the model. By using them as the test set, you can fully understand the performance of the model.

[0095] In one embodiment, step S130 further includes: the machine learning model includes a convolutional layer, a pooling layer, and an output layer connected in sequence;

[0096] The convolution layer calculates the correlation between multiple features for the target feature set, and optimizes the weight according to the correlation;

[0097] The pooling layer reduces the feature quantity by reducing the feature dimension;

[0098] The output layer outputs the training result.

[0099] In one embodiment, the convolution layer calculates the correlation between multiple key features for the target feature set, including:

[0100] Calculate the mean and variance of the two key features respectively;

[0101] Multiply the variance of the two key features;

[0102] The multiplied variance is divided by the sum of the means and the square root is taken to obtain the correlation;

[0103] The pooling layer reduces the feature quantity by reducing the feature dimension, including:

[0104] The maximum value in each pooling window is retained through the maximum pooling operation; the maximum pooling operation retains the maximum value in each pooling window, which helps to retain the most significant features in the target feature set;

[0105] The average pooling operation retains the average value within each pooling window, which helps to preserve the overall feature trend in the target feature set.

[0106] The two pooled features are weighted and summed according to certain weights to obtain the comprehensive pooled features;

[0107] Among them, the sizes of the pooling windows of the maximum pooling operation and the average pooling operation are dynamically adjusted according to the size of the input feature.

[0108] In one embodiment, if Figure 2 As shown, the observation data in step S110 is multi-source heterogeneous observation data, including thermal infrared images, network monitoring data, log data and environmental data; after obtaining the observation data of the radio frequency power supply, the method further includes;

[0109] analyzing data patterns and structures between different data sources in the observational data;

[0110] Determine the matching rules and mapping relationships between data;

[0111] According to the matching rules and mapping relationships, the observation data from different data sources of the RF power supply are integrated into a unified data storage warehouse, and the timestamps are synchronized so that all data have the same scale.

[0112] A device for simulating and predicting the service life of a radio frequency power supply provided by the present disclosure is described below. The detection system described below and the detection method described above can be referenced to each other.

[0113] like Figure 3 As shown, a device for simulating and predicting the service life of a radio frequency power supply comprises:

[0114] The source feature module is used to obtain the observation data of the RF power supply and extract the features of the RF power supply in the high-frequency state as the source feature set;

[0115] An extension module, used for fusing the source feature set and the extended feature set to form a target feature set; wherein the extended feature set is generated in the following manner: taking each feature in the source feature set as a tree node to form a radio frequency power supply parameter tree; selecting 2n features in the radio frequency power supply parameter tree from high to low according to the frequency parameters of the features as n node pairs, where n is the feature expansion multiple, each node pair includes a starting node and a corresponding ending node, and the difference in the frequency parameters of the starting node and the corresponding ending node meets a preset condition; obtaining the optimal path between the starting node and the corresponding ending node in each node pair, interpolating the features of the nodes on the optimal path, and selecting the feature at the maximum curvature as the extended feature set;

[0116] A prediction module is used to use the target feature set to train the machine learning model; use the trained model to predict the changing trend of the preset indicator over time, so as to obtain an estimate of the service life of the RF power supply.

[0117] See also Figure 4 The electronic device further includes: a bus 403 and a communication interface 404, a processor 402, a communication interface 404 and a memory 401 are connected via the bus 403; the processor 402 is used to execute an executable module stored in the memory 401, for example, a computer program.

[0118] The memory 401 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk storage. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 404 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used.

[0119] The bus 403 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0120] Among them, the memory 401 is used to store programs, and the processor 402 executes the program after receiving the execution instruction. The method executed by the device defined by the process disclosed in any embodiment of the present application can be applied to the processor 402 or implemented by the processor 402.

[0121] The processor 402 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 402. The above processor 402 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiments of the present application can be directly embodied as a hardware decoding processor to execute, or the hardware and software modules in the decoding processor can be combined to execute. The software module may be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 401, and the processor 402 reads the information in the memory 401 and completes the steps of the above method in combination with its hardware.

[0122] Corresponding to the above-mentioned RF power supply service life simulation prediction method, an embodiment of the present application also provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to execute the steps of the above-mentioned RF power supply service life simulation prediction method.

[0123] The RF power supply service life simulation prediction device provided in the embodiment of the present application 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 the embodiment of the present application are the same as those of the aforementioned method embodiment. For the sake of brief description, the parts not mentioned in the device embodiment can refer to the corresponding contents in the aforementioned method embodiment. Technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices and units described above can all refer to the corresponding processes in the aforementioned method embodiment, and will not be repeated here.

[0124] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0125] For another example, the flowchart and block diagram in the accompanying drawings show the possible architecture, function and operation of the device, method and computer program product according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or the flowchart, and the combination of the boxes in the block diagram and / or the flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0126] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0127] In addition, each functional unit in the embodiments provided in the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0128] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the RF power supply service life simulation prediction method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store program codes.

[0129] It should be noted that similar numbers and letters represent similar items in the following figures. 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 only used to distinguish the description and are not to be understood as indicating or implying relative importance.

[0130] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solution of the present application, rather than to limit it. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the aforementioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solution recorded in the aforementioned embodiments within the technical scope disclosed in the present application, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solution deviate from the scope of the technical solution of the embodiment of the present application. They should all be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for simulating and predicting the service life of a radio frequency power supply, characterized in that: The method comprises the following steps: Obtain observation data of the radio frequency power source, and extract features of the radio frequency power source in a high frequency state as a source feature set; The source feature set and the extended feature set are merged to form a target feature set; wherein the extended feature set is generated in the following manner: each feature in the source feature set is used as a tree node to form a radio frequency power supply parameter tree; 2n features in the radio frequency power supply parameter tree are selected from high to low according to the frequency parameters of the features as n node pairs, where n is the feature expansion multiple, and each node pair includes a starting node and a corresponding ending node, and the difference in the frequency parameters of the starting node and the corresponding ending node meets a preset condition; the optimal path between the starting node and the corresponding ending node in each node pair is obtained, the features of the nodes on the optimal path are interpolated, and the features at the maximum curvature are selected as the extended feature set; The target feature set is used to train the machine learning model; the trained model is used to predict the changing trend of the preset indicator over time, so as to obtain an estimate of the service life of the radio frequency power supply.

2. The method for simulating and predicting the service life of a radio frequency power supply according to claim 1, characterized in that: The feature of extracting the radio frequency power supply in a high frequency state comprises: The observation data of the radio frequency power supply is sampled and processed by using a sliding sampling window, and the length of the sliding sampling window is adjusted in real time with the power of the radio frequency power supply; Calculating the variance value of the slope of all sampling points in the sliding sampling window, and determining the high-frequency state range according to the variance value of the slope; The characteristics of the radio frequency power supply in the high frequency state are extracted according to the high frequency state range.

3. The method for simulating and predicting the service life of a radio frequency power supply according to claim 1, characterized in that: The forming of a radio frequency power supply parameter tree by taking each feature in the source feature set as a tree node comprises: Each feature in the source feature set is sorted according to the sampling time of the feature, the first sampled feature is used as the root node, and a radio frequency power supply parameter tree is constructed according to the sorted features.

4. The method for simulating and predicting the service life of a radio frequency power supply according to claim 1, characterized in that: The fusing the source feature set and the extended feature set to form a target feature set comprises: Setting a first weight for each feature in the source feature set according to the magnitude of a frequency parameter; Setting a second weight for each feature in the extended feature set according to the size of the curvature, wherein the first weight is greater than the second weight; Selecting a feature in the source feature set in which the first weight is less than a first threshold and a feature in the extended feature set in which the second weight is greater than a second threshold as a training set of a target feature set; The remaining features in the target feature set are used as a test set.

5. The method for simulating and predicting the service life of a radio frequency power supply according to claim 4, characterized in that: The machine learning model includes a convolutional layer, a pooling layer, and an output layer connected sequentially; The convolution layer calculates the correlation between multiple features for the target feature set, and optimizes the weight according to the correlation; The pooling layer reduces the feature quantity by reducing the feature dimension; The output layer outputs the training result.

6. The method for simulating and predicting the service life of a radio frequency power supply according to claim 5, characterized in that: The convolution layer calculates the correlation between multiple features for the target feature set, including: Calculate the mean and variance of the two features respectively; Multiply the variance of two features; The multiplied variance is divided by the sum of the means and the square root is taken to obtain the correlation; The pooling layer reduces the feature quantity by reducing the feature dimension, including: The maximum value within each pooling window is retained through the maximum pooling operation; The average pooling operation retains the average value within each pooling window; The two pooled features are weighted and summed according to certain weights to obtain the comprehensive pooled features; The sizes of the pooling windows of the maximum pooling operation and the average pooling operation are dynamically adjusted according to the frequency parameters of the input features.

7. The method for simulating and predicting the service life of a radio frequency power supply according to claim 1, characterized in that: After obtaining the observation data of the radio frequency power source, the method further includes: analyzing data patterns and structures between different data sources in the observational data; Determine the matching rules and mapping relationships between data; According to the matching rules and mapping relationships, the observation data from different data sources of the RF power supply are integrated into a unified data storage warehouse, and the timestamps are synchronized so that all data have the same scale.

8. A device for simulating and predicting the service life of a radio frequency power supply, characterized in that: include: The source feature module is used to obtain the observation data of the RF power supply and extract the features of the RF power supply in the high-frequency state as the source feature set; An extension module, used for fusing the source feature set and the extended feature set to form a target feature set; wherein the extended feature set is generated in the following manner: taking each feature in the source feature set as a tree node to form a radio frequency power supply parameter tree; selecting 2n features in the radio frequency power supply parameter tree from high to low according to the frequency parameters of the features as n node pairs, where n is the feature expansion multiple, each node pair includes a starting node and a corresponding ending node, and the difference in the frequency parameters of the starting node and the corresponding ending node meets a preset condition; obtaining the optimal path between the starting node and the corresponding ending node in each node pair, interpolating the features of the nodes on the optimal path, and selecting the feature at the maximum curvature as the extended feature set; A prediction module is used to use the target feature set to train the machine learning model; use the trained model to predict the changing trend of the preset indicator over time, so as to obtain an estimate of the service life of the RF power supply.

9. An electronic device, comprising: processor; A memory storing a program, wherein the program comprises instructions which, when executed by the processor, cause the processor to perform the method according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer-readable storage medium stores instructions or computer programs, and when the instructions or computer programs are executed on a device, the device executes the method according to any one of claims 1 to 7.

Images