System design method and system supporting automatic testing of multiple radio frequency parameters

Abstract

The invention provides a system design method and system supporting automatic testing of multiple radio frequency parameters, and relates to the technical field of testing, and the method comprises the steps: receiving and analyzing a test task, generating a test configuration file according to the information of radio frequency equipment to be tested and multiple radio frequency parameter test items, and using a self-adaptive optimization algorithm to optimize the test configuration file. Dynamically adjusting the test configuration file based on historical test data to generate an optimized test configuration file; controlling a radio frequency signal generator to generate a test signal, inputting the test signal into the radio frequency equipment to be tested, starting a real-time monitoring module, performing multi-dimensional analysis on a digital signal processed by a radio frequency signal analyzer, performing anomaly detection on an analysis result by using a deep learning algorithm, identifying and marking an abnormal data point; and generating a test report according to the analysis result and the anomaly detection result.

Description

Technical Field

[0001] The present invention relates to testing technology, and in particular to a system design method and system supporting automatic testing of multiple radio frequency parameters. Background Art

[0002] With the rapid development of wireless communication technology, the performance test of RF equipment and systems has become increasingly important. Traditional RF parameter test methods usually require professionals to manually operate various test instruments to test different RF parameters one by one. This method is not only time-consuming and labor-intensive, but also easily affected by human factors, making it difficult to ensure the consistency and accuracy of the test results.

[0003] Currently, there are some automated test systems on the market, but most of them can only test specific RF parameters or specific types of equipment. For example, some systems focus on testing the output power and frequency stability of transmitters, while others are mainly used for receiver sensitivity and selectivity testing. Such single-function test systems are difficult to meet the needs of comprehensive performance evaluation of modern RF devices, and also increase testing costs and complexity.

[0004] In addition, existing automated test systems often lack flexibility and scalability. As new communication standards and technologies continue to emerge, test requirements are also constantly changing. However, the software and hardware architectures of many test systems are relatively closed, making it difficult to quickly adapt to new test requirements or integrate new test methods.

[0005] In terms of test efficiency, existing systems usually use fixed test sequences and cannot dynamically adjust test strategies based on the actual performance characteristics of the equipment. This approach may lead to unnecessary waste of test time, especially in mass production testing, where efficiency issues are particularly prominent.

[0006] Data processing and result analysis is another challenge faced by existing systems. Most systems only provide basic data logging functions and lack in-depth data analysis and visualization capabilities. This makes it difficult for testers to quickly identify potential problems or conduct trend analysis, which affects the interpretation and application of test results.

[0007] In terms of test accuracy, although some high-end systems can provide higher measurement accuracy, they are usually expensive and not suitable for widespread application. Relatively low-priced systems often have deficiencies in measurement accuracy and stability, making it difficult to meet the needs of high-precision testing.

[0008] In addition, the existing automatic test systems also have room for improvement in terms of user interface and operating experience. Many systems have complex operating interfaces and require professional training to be proficient in use, which increases labor costs and the risk of operational errors.

[0009] In summary, the existing technology has many limitations and deficiencies in supporting the automatic testing of multiple RF parameters. Developing a system that can support the automatic testing of multiple RF parameters, has high flexibility and scalability, can improve the test efficiency and accuracy, and is easy to operate is of great significance to improving the test level of RF equipment and systems. Summary of the invention

[0010] The embodiments of the present invention provide a system design method and system that support automatic testing of multiple radio frequency parameters, which can solve the problems in the prior art.

[0011] According to a first aspect of the embodiments of the present invention,

[0012] Provides a system design method that supports automatic testing of multiple RF parameters, including:

[0013] Receive and parse a test task, wherein the test task includes information of a radio frequency device to be tested and multiple radio frequency parameter test items; generate a test configuration file according to the information of the radio frequency device to be tested and the multiple radio frequency parameter test items, wherein the test configuration file includes a radio frequency frequency range, a power level, a modulation mode, a bandwidth, and a test duration; and dynamically adjust the test configuration file based on historical test data using an adaptive optimization algorithm to generate an optimized test configuration file;

[0014] According to the optimized test configuration file, the RF signal generator is controlled to generate a test signal, and the parameters of the test signal match the parameters in the optimized test configuration file; the test signal is input into the RF device to be tested, and a real-time monitoring module is started at the same time, and the real-time monitoring module is used to monitor the temperature, voltage and current of the RF device to be tested; when the monitored temperature, voltage or current exceeds a preset threshold, the power level of the test signal is automatically adjusted or the test is suspended; the RF signal analyzer is controlled to receive the output signal of the RF device to be tested, and the output signal is sampled and digitized;

[0015] Perform multi-dimensional analysis on the digital signal processed by the RF signal analyzer, including spectrum analysis, modulation quality analysis and power measurement; use deep learning algorithms to detect anomalies in the analysis results, identify and mark abnormal data points; generate a test report based on the analysis results and anomaly detection results, the test report includes the test results of various RF parameters, performance curves and abnormal data analysis; compare the test report with historical data, and generate a performance trend chart through time series analysis; based on the test results and performance trends, use the expert system to generate optimization suggestions and predictive maintenance plans for the RF equipment to be tested.

[0016] In an optional embodiment,

[0017] Using an adaptive optimization algorithm, dynamically adjusting the test configuration file based on historical test data to generate an optimized test configuration file includes:

[0018] Constructing a historical test database, wherein the historical test database contains test configurations of radio frequency devices and corresponding test results, wherein the test configurations include frequency range, power level, and modulation mode, and the test results include test duration, test result accuracy, and abnormal conditions during the test process;

[0019] According to the basic information of the radio frequency device to be tested, the test configuration of similar devices is screened out from the historical test database as an initial particle group;

[0020] Calculating a fitness value for each particle in the initial particle swarm, wherein the fitness value is calculated by a fitness function, wherein the fitness function considers three factors, namely, the inverse of the test duration, the result accuracy, and the coverage, and assigns different weights to each factor, and obtains a comprehensive score by weighted summation;

[0021] The particle swarm optimization algorithm is used to update the position and velocity of the particles, wherein the position of the particles represents the specific value of the test parameter;

[0022] Introduce the crossover and mutation operations of the genetic algorithm to perform crossover and mutation on the updated particle swarm to generate a new particle swarm;

[0023] According to the results of each iteration, the inertia weight of the particle swarm optimization algorithm and the crossover and mutation probability of the genetic algorithm are dynamically adjusted;

[0024] Repeat the fitness evaluation, particle position update, crossover mutation and dynamic adjustment steps until the preset number of iterations is reached or the improvement of the fitness value for multiple consecutive iterations is less than the preset threshold;

[0025] Select the particle with the highest fitness in the final particle swarm as the optimized test configuration;

[0026] Integrate the optimized test configuration and its corresponding expected test duration and possible abnormal conditions into an optimized test configuration file;

[0027] Executing an online learning mechanism to add newly completed test data to the historical test database and trigger an incremental learning process to update the parameters of the optimization algorithm;

[0028] The fitness function is calculated by respectively assigning weights to the inverse of the test duration, the result accuracy and the coverage and summing them up.

[0029] In an optional embodiment,

[0030] According to the optimized test configuration file, controlling the radio frequency signal generator to generate a test signal, wherein the parameters of the test signal match the parameters in the optimized test configuration file, comprises:

[0031] Parsing the optimized test configuration file to extract key parameters of the test signal, the key parameters including frequency range, power level, modulation mode and bandwidth;

[0032] Converting the key parameters into low-level control instructions of the radio frequency signal generator through a parameter mapping engine, wherein the parameter mapping engine uses a pre-established parameter mapping table for conversion;

[0033] Based on software-defined radio technology, field programmable gate arrays are used to implement digital signal processing, and digital-to-analog converters are used to generate radio frequency signals;

[0034] Execute digital pre-distortion algorithms to compensate for nonlinear distortion of the DAC and RF front end at different frequencies and powers;

[0035] Utilize the parallel processing capability of field programmable gate arrays to pre-calculate and cache multiple sets of signal parameters to achieve fast parameter switching;

[0036] The power and frequency of the output signal are continuously monitored by a high-precision power detector and frequency counter;

[0037] When it is detected that the power or frequency deviation of the output signal exceeds a preset threshold, an automatic calibration procedure is triggered to adjust the internal parameters of the RF signal generator;

[0038] According to the modulation parameters in the test configuration file, select or dynamically generate a baseband signal that meets the requirements from the modulation signal library;

[0039] Use digital up-conversion technology to convert the baseband signal into the required RF signal;

[0040] According to the channel model specified in the test configuration file, the generated RF signal is processed in real time to simulate the interference and attenuation effects in the real transmission environment;

[0041] The entire test process is coordinated through an intelligent sequence controller, which automatically schedules the parameter setting, switching and calibration operations of the RF signal generator;

[0042] The intelligent sequence controller is also responsible for synchronous control with other test equipment to ensure the consistency of the test process;

[0043] During the test, the output of the RF signal generator is continuously monitored, and the signal parameters are dynamically adjusted based on the real-time monitoring results to keep the test signal continuously matching the parameters in the optimized test configuration file.

[0044] In an optional embodiment,

[0045] Perform multi-dimensional analysis on the digital signal processed by the RF signal analyzer, including spectrum analysis, modulation quality analysis and power measurement, including:

[0046] Performing an improved fast Fourier transform algorithm on the digital signal to convert the digital signal into the frequency domain;

[0047] Fine interpolation is performed between adjacent points of the fast Fourier transform result to improve the frequency resolution of the spectrum;

[0048] According to the characteristics of the digital signal, the most suitable window function is automatically selected for fast Fourier transform, wherein the window function includes Hanning window and Blackman window;

[0049] executing an intelligent peak detection algorithm to identify the main signal, harmonic components and interference signals, wherein the intelligent peak detection algorithm distinguishes between true signal peaks and noise peaks by setting dynamic thresholds and applying pattern recognition techniques;

[0050] Perform digital down-conversion and symbol timing recovery on digital signals to extract in-phase and quadrature components of baseband signals;

[0051] The extracted in-phase component and quadrature component are processed by an adaptive equalization algorithm to compensate for the linear distortion introduced by the channel;

[0052] Calculate error vector magnitude, amplitude error and phase error, wherein the error vector magnitude is calculated using a weighting factor to reflect the modulation quality of different symbol positions;

[0053] Draw and analyze constellation diagrams in real time, and quantitatively evaluate the modulation accuracy of signals by counting the distribution characteristics of constellation points;

[0054] For FM signals, the modulation index and frequency deviation are calculated by the instantaneous frequency detection algorithm; for AM signals, the modulation depth and distortion are analyzed by the envelope detection technology;

[0055] Adopt digital integration technology, combined with real-time power calculation algorithm of sliding window, to measure average power and peak power;

[0056] Regularly inject a calibration signal of known power to compensate for gain drift and nonlinear errors in the measurement link in real time;

[0057] Monitor ambient temperature in real time and apply pre-established temperature-error models to eliminate the impact of temperature changes on power measurement;

[0058] Utilize a multi-dimensional data fusion engine to comprehensively process the results of spectrum analysis, modulation quality analysis, and power measurement;

[0059] The multi-dimensional data fusion engine uses machine learning algorithms to analyze the correlation between various indicators and automatically identify potential signal anomalies and performance issues;

[0060] Compare the analysis results with the preset standards, automatically generate a test report, and provide targeted optimization suggestions.

[0061] In an optional embodiment,

[0062] Using deep learning algorithms to perform anomaly detection on analysis results, identify and mark abnormal data points including:

[0063] Perform multi-level data preprocessing on the original analysis results, including: filling missing values ​​based on a time series interpolation algorithm that takes into account the time correlation of the RF signal; using an improved Z-score method to initially screen outliers, dynamically adjusting the threshold to adapt to the characteristics of different types of signals; performing adaptive normalization to automatically select the most suitable normalization method based on the distribution characteristics of different parameters; extracting time-frequency domain features through wavelet transform and Hilbert-Huang transform;

[0064] Inputting the preprocessed data into a hybrid deep learning model, wherein the hybrid deep learning model includes a long short-term memory network layer and a convolutional neural network layer;

[0065] A semi-supervised learning method is used to train the hybrid deep learning model, which includes using a large amount of normal data to train the model to learn normal patterns, and introducing a small amount of known abnormal samples and artificially synthesized abnormal samples to fine-tune the model;

[0066] Integrate attention mechanisms in hybrid deep learning models and analyze attention weights to improve the interpretability of anomaly detection;

[0067] Execute an adaptive threshold adjustment algorithm, analyze historical detection results and expert feedback, and use the Bayesian optimization method to dynamically adjust the threshold for abnormality determination;

[0068] Build a correlation anomaly detection module based on graph neural network, construct the analysis results of different dimensions into a relationship graph, and learn the complex dependency relationships between nodes through graph convolution operations;

[0069] Use clustering algorithms to classify detected anomalies and generate a preliminary analysis report on the causes of anomalies in combination with expert knowledge base;

[0070] Establish a mapping relationship between abnormal patterns and possible causes to provide technicians with targeted troubleshooting suggestions;

[0071] Continuously update and optimize the anomaly detection system, including updating training data, adjusting model parameters, and optimizing anomaly classification rules.

[0072] According to a second aspect of the embodiments of the present invention,

[0073] Provides a system design system that supports automatic testing of multiple RF parameters, including:

[0074] The first unit is used to receive and parse a test task, wherein the test task includes information of a radio frequency device to be tested and multiple radio frequency parameter test items; generate a test configuration file according to the information of the radio frequency device to be tested and the multiple radio frequency parameter test items, wherein the test configuration file includes a radio frequency range, a power level, a modulation mode, a bandwidth, and a test duration; and dynamically adjust the test configuration file based on historical test data using an adaptive optimization algorithm to generate an optimized test configuration file;

[0075] The second unit is used to control the RF signal generator to generate a test signal according to the optimized test configuration file, and the parameters of the test signal match the parameters in the optimized test configuration file; input the test signal into the RF device to be tested, and start the real-time monitoring module at the same time, and the real-time monitoring module is used to monitor the temperature, voltage and current of the RF device to be tested; when the monitored temperature, voltage or current exceeds a preset threshold, automatically adjust the power level of the test signal or suspend the test; control the RF signal analyzer to receive the output signal of the RF device to be tested, and sample and digitize the output signal;

[0076] The third unit is used to perform multi-dimensional analysis on the digital signal processed by the RF signal analyzer, including spectrum analysis, modulation quality analysis and power measurement; use deep learning algorithms to detect anomalies in the analysis results, identify and mark abnormal data points; generate a test report based on the analysis results and anomaly detection results, and the test report includes test results, performance curves and abnormal data analysis of various RF parameters; compare the test report with historical data, and generate a performance trend chart through time series analysis; based on the test results and performance trends, use the expert system to generate optimization suggestions and predictive maintenance plans for the RF equipment to be tested.

[0077] According to a third aspect of the embodiments of the present invention,

[0078] An electronic device is provided, comprising:

[0079] processor;

[0080] a memory for storing processor-executable instructions;

[0081] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0082] According to a fourth aspect of the embodiments of the present invention,

[0083] A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Figure 1 A schematic diagram of a flow chart of a system design method supporting automatic testing of multiple radio frequency parameters according to an embodiment of the present invention;

[0085] Figure 2 A schematic diagram of the structure of a system design system supporting automatic testing of multiple radio frequency parameters according to an embodiment of the present invention. DETAILED DESCRIPTION

[0086] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0087] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0088] Figure 1 FIG. 1 is a flow chart of a system design method for supporting automatic testing of multiple radio frequency parameters according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0089] S101. Receive and parse a test task, wherein the test task includes information of a radio frequency device to be tested and multiple radio frequency parameter test items; generate a test configuration file according to the information of the radio frequency device to be tested and the multiple radio frequency parameter test items, wherein the test configuration file includes a radio frequency range, a power level, a modulation mode, a bandwidth, and a test duration; and dynamically adjust the test configuration file based on historical test data using an adaptive optimization algorithm to generate an optimized test configuration file;

[0090] S102. According to the optimized test configuration file, control the RF signal generator to generate a test signal, the parameters of the test signal match the parameters in the optimized test configuration file; input the test signal into the RF device to be tested, and start the real-time monitoring module at the same time, the real-time monitoring module is used to monitor the temperature, voltage and current of the RF device to be tested; when the monitored temperature, voltage or current exceeds a preset threshold, automatically adjust the power level of the test signal or suspend the test; control the RF signal analyzer to receive the output signal of the RF device to be tested, and sample and digitize the output signal;

[0091] S103. Perform multi-dimensional analysis on the digital signal processed by the RF signal analyzer, including spectrum analysis, modulation quality analysis and power measurement; use deep learning algorithms to detect anomalies in the analysis results, identify and mark abnormal data points; generate a test report based on the analysis results and anomaly detection results, the test report including the test results of various RF parameters, performance curves and abnormal data analysis; compare the test report with historical data, and generate a performance trend chart through time series analysis; based on the test results and performance trends, use an expert system to generate optimization suggestions and predictive maintenance plans for the RF equipment to be tested.

[0092] In an optional embodiment,

[0093] Using an adaptive optimization algorithm, dynamically adjusting the test configuration file based on historical test data to generate an optimized test configuration file includes:

[0094] Constructing a historical test database, wherein the historical test database contains test configurations of radio frequency devices and corresponding test results, wherein the test configurations include frequency range, power level, and modulation mode, and the test results include test duration, test result accuracy, and abnormal conditions during the test process;

[0095] According to the basic information of the radio frequency device to be tested, the test configuration of similar devices is screened out from the historical test database as an initial particle group;

[0096] Calculating a fitness value for each particle in the initial particle swarm, wherein the fitness value is calculated by a fitness function, wherein the fitness function considers three factors, namely, the inverse of the test duration, the result accuracy, and the coverage, and assigns different weights to each factor, and obtains a comprehensive score by weighted summation;

[0097] The particle swarm optimization algorithm is used to update the position and velocity of the particles, wherein the position of the particles represents the specific value of the test parameter;

[0098] Introduce the crossover and mutation operations of the genetic algorithm to perform crossover and mutation on the updated particle swarm to generate a new particle swarm;

[0099] According to the results of each iteration, the inertia weight of the particle swarm optimization algorithm and the crossover and mutation probability of the genetic algorithm are dynamically adjusted;

[0100] Repeat the fitness evaluation, particle position update, crossover mutation and dynamic adjustment steps until the preset number of iterations is reached or the improvement of the fitness value for multiple consecutive iterations is less than the preset threshold;

[0101] Select the particle with the highest fitness in the final particle swarm as the optimized test configuration;

[0102] Integrate the optimized test configuration and its corresponding expected test duration and possible abnormal conditions into an optimized test configuration file;

[0103] Executing an online learning mechanism to add newly completed test data to the historical test database and trigger an incremental learning process to update the parameters of the optimization algorithm;

[0104] The fitness function is calculated by respectively assigning weights to the inverse of the test duration, the result accuracy and the coverage and summing them up.

[0105] For example, we first built a historical test database that contains the test configurations and corresponding test results of different types of RF devices. This database not only stores test parameters such as frequency range, power level, and modulation, but also records test duration, test result accuracy, and abnormal conditions that occurred during the test.

[0106] Based on this database, we designed an adaptive optimization algorithm based on the combination of particle swarm optimization (PSO) and genetic algorithm (GA). The core idea of ​​this algorithm is to regard each set of test configurations as a particle in multidimensional space and find the best test configuration through iterative optimization.

[0107] The specific steps of the algorithm are as follows:

[0108] Initialization: Based on the basic information of the RF device to be tested, the test configurations of similar devices are selected from the historical database as the initial particle group.

[0109] Fitness evaluation: Calculate the fitness value for each particle (i.e., test configuration). The fitness function takes into account three factors: test duration, result accuracy, and coverage, and obtains a comprehensive score through weighted summation.

[0110] Update particle position: Use the PSO algorithm to update the position and velocity of the particle. Here, the position of the particle represents the specific value of the test parameters, such as the start and end values ​​of the frequency range, power level, etc.

[0111] Crossover and mutation: Introduce crossover and mutation operations of GA to increase population diversity. The crossover operation can combine the excellent features of different configurations, while the mutation operation can explore new parameter combinations.

[0112] Adaptive adjustment: Dynamically adjust the inertia weight of PSO and the crossover probability of GA according to the results of each iteration. For example, when the population tends to converge, increase the probability of mutation to increase exploration.

[0113] Termination condition: The algorithm terminates when the preset number of iterations is reached or the improvement of the fitness value after multiple consecutive iterations is less than the threshold.

[0114] Output optimization results: Select the particle with the highest fitness in the final population as the optimized test configuration.

[0115] To verify the effectiveness of the algorithm, we selected three different types of RF devices for testing: low noise amplifiers, power amplifiers, and RF transceivers. The experimental results show that compared with the traditional fixed configuration, the test configuration optimized by this algorithm can shorten the test time by an average of 23.5%, while improving the accuracy of the test results by 8.7%.

[0116] In addition, we have implemented an online learning mechanism. After each test is completed, new test data is added to the historical database and triggers the algorithm's incremental learning process. This allows the optimization algorithm to continuously adapt to new device types and test requirements, maintaining its long-term effectiveness.

[0117] In this way, we have achieved intelligent and personalized test configuration, laying the foundation for subsequent automatic test execution. The optimized test configuration file not only contains the best parameter settings, but also comes with the expected test duration and possible abnormal situations, which will play an important role in the next step of test execution and real-time monitoring.

[0118] In an optional embodiment,

[0119] According to the optimized test configuration file, controlling the radio frequency signal generator to generate a test signal, wherein the parameters of the test signal match the parameters in the optimized test configuration file, comprises:

[0120] Parsing the optimized test configuration file to extract key parameters of the test signal, the key parameters including frequency range, power level, modulation mode and bandwidth;

[0121] Converting the key parameters into low-level control instructions of the radio frequency signal generator through a parameter mapping engine, wherein the parameter mapping engine uses a pre-established parameter mapping table for conversion;

[0122] Based on software-defined radio technology, field programmable gate arrays are used to implement digital signal processing, and digital-to-analog converters are used to generate radio frequency signals;

[0123] Execute digital pre-distortion algorithms to compensate for nonlinear distortion of the DAC and RF front end at different frequencies and powers;

[0124] Utilize the parallel processing capability of field programmable gate arrays to pre-calculate and cache multiple sets of signal parameters to achieve fast parameter switching;

[0125] The power and frequency of the output signal are continuously monitored by a high-precision power detector and frequency counter;

[0126] When it is detected that the power or frequency deviation of the output signal exceeds a preset threshold, an automatic calibration procedure is triggered to adjust the internal parameters of the RF signal generator;

[0127] According to the modulation parameters in the test configuration file, select or dynamically generate a baseband signal that meets the requirements from the modulation signal library;

[0128] Use digital up-conversion technology to convert the baseband signal into the required RF signal;

[0129] According to the channel model specified in the test configuration file, the generated RF signal is processed in real time to simulate the interference and attenuation effects in the real transmission environment;

[0130] The entire test process is coordinated through an intelligent sequence controller, which automatically schedules the parameter setting, switching and calibration operations of the RF signal generator;

[0131] The intelligent sequence controller is also responsible for synchronous control with other test equipment to ensure the consistency of the test process;

[0132] During the test, the output of the RF signal generator is continuously monitored, and the signal parameters are dynamically adjusted based on the real-time monitoring results to keep the test signal continuously matching the parameters in the optimized test configuration file.

[0133] For example, after obtaining the optimized test configuration file, the RF parameter automatic test system needs to accurately control the RF signal generator to generate a test signal that matches the configuration file parameters. This process involves multiple key technical points, and needs to consider the accuracy and stability of the signal as well as the automation of the test process.

[0134] First, a high-precision RF signal generator control module based on software-defined radio (SDR) technology was designed. This module uses FPGA to implement digital signal processing and cooperates with high-performance DAC to complete the generation of RF signals. By implementing a complex digital pre-distortion algorithm, the nonlinear distortion of DAC and RF front end at different frequencies and powers is effectively compensated, ensuring the high fidelity of the output signal.

[0135] In order to achieve accurate matching of test signal parameters with configuration files, a parameter mapping engine was developed. The engine first parses the optimized test configuration file to extract key parameters such as frequency range, power level, modulation mode, and bandwidth. Then, by looking up the pre-established parameter mapping table, these high-level parameters are converted into low-level control instructions for the RF signal generator. This approach not only improves the accuracy of parameter settings, but also greatly enhances the scalability of the system, enabling it to adapt to different models of signal generators.

[0136] Considering the need to frequently switch parameters in RF testing, a fast parameter switching technology is implemented. This technology uses the parallel processing capability of FPGA to pre-calculate and cache multiple sets of signal parameters. When parameters need to be switched, the cached data is quickly switched to achieve a sub-microsecond parameter switching speed, significantly improving test efficiency.

[0137] In order to ensure the stability of the signal during long-term testing, a real-time calibration subsystem is designed. This subsystem continuously monitors the power and frequency of the output signal through a high-precision power detector and a frequency counter. Once a deviation exceeding the preset threshold is detected, the automatic calibration procedure is immediately triggered to adjust the internal parameters of the signal generator to ensure that the output signal always remains within the specified range.

[0138] In terms of modulation signal generation, a flexible modulation signal library has been developed. This library not only contains common digital modulation methods (such as QPSK, QAM, etc.), but also supports custom modulation schemes. Through the modulation parameters in the configuration file, the system can dynamically generate baseband signals that meet the requirements and convert them into the required RF signals through digital up-conversion technology.

[0139] In order to improve the authenticity of the test, a channel simulator module is implemented. This module can process the generated signal in real time according to the channel model specified in the configuration file (such as AWGN, Rayleigh fading, etc.), simulating various interference and attenuation effects in the real transmission environment. This function makes the test results closer to the actual application scenario.

[0140] Finally, an intelligent sequence controller is designed to coordinate the entire test process. This controller automatically schedules the parameter setting, switching, and calibration operations of the signal generator according to the order of the test items in the configuration file. At the same time, it is also responsible for the synchronous control with other test equipment (such as spectrum analyzers, vector network analyzers, etc.) to ensure the smoothness and consistency of the entire test process.

[0141] Through the comprehensive application of the above technologies, the test signal is accurately matched with the optimized configuration file, laying a solid foundation for the subsequent RF parameter measurement. This highly automated and precisely controlled method not only improves the repeatability and reliability of the test, but also significantly improves the overall test efficiency.

[0142] In an optional embodiment,

[0143] Perform multi-dimensional analysis on the digital signal processed by the RF signal analyzer, including spectrum analysis, modulation quality analysis and power measurement, including:

[0144] Performing an improved fast Fourier transform algorithm on the digital signal to convert the digital signal into the frequency domain;

[0145] Fine interpolation is performed between adjacent points of the fast Fourier transform result to improve the frequency resolution of the spectrum;

[0146] According to the characteristics of the digital signal, the most suitable window function is automatically selected for fast Fourier transform, wherein the window function includes Hanning window and Blackman window;

[0147] executing an intelligent peak detection algorithm to identify the main signal, harmonic components and interference signals, wherein the intelligent peak detection algorithm distinguishes between true signal peaks and noise peaks by setting dynamic thresholds and applying pattern recognition techniques;

[0148] Perform digital down-conversion and symbol timing recovery on digital signals to extract in-phase and quadrature components of baseband signals;

[0149] The extracted in-phase component and quadrature component are processed by an adaptive equalization algorithm to compensate for the linear distortion introduced by the channel;

[0150] Calculate error vector magnitude, amplitude error and phase error, wherein the error vector magnitude is calculated using a weighting factor to reflect the modulation quality of different symbol positions;

[0151] Draw and analyze constellation diagrams in real time, and quantitatively evaluate the modulation accuracy of signals by counting the distribution characteristics of constellation points;

[0152] For FM signals, the modulation index and frequency deviation are calculated by the instantaneous frequency detection algorithm; for AM signals, the modulation depth and distortion are analyzed by the envelope detection technology;

[0153] Adopt digital integration technology, combined with real-time power calculation algorithm of sliding window, to measure average power and peak power;

[0154] Regularly inject a calibration signal of known power to compensate for gain drift and nonlinear errors in the measurement link in real time;

[0155] Monitor ambient temperature in real time and apply pre-established temperature-error models to eliminate the impact of temperature changes on power measurement;

[0156] Utilize a multi-dimensional data fusion engine to comprehensively process the results of spectrum analysis, modulation quality analysis, and power measurement;

[0157] The multi-dimensional data fusion engine uses machine learning algorithms to analyze the correlation between various indicators and automatically identify potential signal anomalies and performance issues;

[0158] Compare the analysis results with the preset standards, automatically generate a test report, and provide targeted optimization suggestions.

[0159] For example, the digital signal processed by the RF signal analyzer contains rich information, and multi-dimensional analysis is required to fully evaluate the quality and performance of the signal. To this end, a comprehensive analysis system has been developed that integrates three core modules: spectrum analysis, modulation quality analysis, and power measurement, to achieve in-depth analysis of the signal.

[0160] First, the spectrum analysis module uses an improved fast Fourier transform (FFT) algorithm to perform frequency domain conversion on digital signals. In order to improve the accuracy and resolution of spectrum analysis, an interpolation-based spectrum refinement technology is implemented. This technology significantly improves the frequency resolution of the spectrum by finely interpolating between adjacent points of the FFT results. At the same time, an adaptive window function selection mechanism is introduced to automatically select the most suitable window function according to the characteristics of the signal, such as Hanning window, Blackman window, etc., which effectively reduces the spectrum leakage phenomenon.

[0161] Based on spectrum analysis, an intelligent peak detection algorithm was developed. This algorithm can not only accurately identify the main signal and harmonic components, but also automatically detect and classify various interference signals. By setting dynamic thresholds and applying pattern recognition technology, the algorithm can effectively distinguish between real signal peaks and noise peaks, greatly improving the reliability of spectrum analysis.

[0162] The modulation quality analysis module focuses on the modulation characteristics of the signal. A general modulation quality assessment framework is designed for different modulation modes. The framework first extracts the in-phase (I) and quadrature (Q) components of the baseband signal through digital down-conversion and symbol timing recovery technology. Then, the extracted IQ signals are processed using an adaptive equalization algorithm to compensate for the linear distortion introduced by the channel.

[0163] Based on IQ signal processing, the calculation of multiple modulation quality indicators is realized. For digital modulation signals, the error vector magnitude (EVM), amplitude error and phase error are calculated. The calculation of EVM adopts an improved algorithm, which introduces weighting factors to more accurately reflect the modulation quality of different symbol positions. In addition, the real-time drawing and analysis function of the constellation diagram is realized, and the modulation accuracy of the signal is quantitatively evaluated by statistically analyzing the distribution characteristics of the constellation points.

[0164] For analog modulation signals, a set of special analysis methods has been developed. For example, for frequency modulation (FM) signals, the modulation index and frequency deviation are calculated through the instantaneous frequency detection algorithm; for amplitude modulation (AM) signals, the modulation depth and distortion are analyzed through envelope detection technology. These analysis results provide a comprehensive basis for evaluating the quality of analog modulation signals.

[0165] The power measurement module uses digital integration technology to achieve high-precision average power and peak power measurement. In order to accurately capture the instantaneous power changes of the signal, a real-time power calculation algorithm based on a sliding window is designed. This algorithm can track power changes with extremely high time resolution and is particularly suitable for the analysis of pulse signals and time division multiple access signals.

[0166] In order to improve the accuracy of power measurement, a self-calibration mechanism is introduced. By regularly injecting a calibration signal of known power, the system can compensate for gain drift and nonlinear errors in the measurement link in real time. In addition, a temperature compensation algorithm is implemented, which effectively eliminates the impact of temperature changes on power measurement by monitoring the ambient temperature in real time and applying a pre-established temperature-error model.

[0167] Finally, a multi-dimensional data fusion engine was designed to comprehensively process the results of spectrum analysis, modulation quality analysis, and power measurement. The engine uses machine learning algorithms to automatically identify potential signal anomalies and performance issues by analyzing the correlation between various indicators. For example, by comparing spectrum characteristics and EVM changes, it is possible to quickly locate whether the signal quality degradation is caused by frequency offset or phase noise.

[0168] This multi-dimensional analysis method not only provides a comprehensive signal quality assessment, but also lays the foundation for subsequent automated testing and intelligent diagnosis. By comparing the analysis results with the preset standards, the system can automatically generate a test report and give targeted optimization suggestions, greatly improving the efficiency and depth of RF testing.

[0169] In an optional embodiment,

[0170] Using deep learning algorithms to perform anomaly detection on analysis results, identify and mark abnormal data points including:

[0171] Perform multi-level data preprocessing on the original analysis results, including: filling missing values ​​based on a time series interpolation algorithm that takes into account the time correlation of the RF signal; using an improved Z-score method to initially screen outliers, dynamically adjusting the threshold to adapt to the characteristics of different types of signals; performing adaptive normalization to automatically select the most suitable normalization method based on the distribution characteristics of different parameters; extracting time-frequency domain features through wavelet transform and Hilbert-Huang transform;

[0172] Inputting the preprocessed data into a hybrid deep learning model, wherein the hybrid deep learning model includes a long short-term memory network layer and a convolutional neural network layer;

[0173] A semi-supervised learning method is used to train the hybrid deep learning model, which includes using a large amount of normal data to train the model to learn normal patterns, and introducing a small amount of known abnormal samples and artificially synthesized abnormal samples to fine-tune the model;

[0174] Integrate attention mechanisms in hybrid deep learning models and analyze attention weights to improve the interpretability of anomaly detection;

[0175] Execute an adaptive threshold adjustment algorithm, analyze historical detection results and expert feedback, and use the Bayesian optimization method to dynamically adjust the threshold for abnormality determination;

[0176] Build a correlation anomaly detection module based on graph neural network, construct the analysis results of different dimensions into a relationship graph, and learn the complex dependency relationships between nodes through graph convolution operations;

[0177] Use clustering algorithms to classify detected anomalies and generate a preliminary analysis report on the causes of anomalies in combination with expert knowledge base;

[0178] Establish a mapping relationship between abnormal patterns and possible causes to provide technicians with targeted troubleshooting suggestions;

[0179] Continuously update and optimize the anomaly detection system, including updating training data, adjusting model parameters, and optimizing anomaly classification rules.

[0180] For example, in the process of RF signal analysis, accurately identifying and marking abnormal data points is crucial to ensure the reliability of test results. To this end, a deep learning-based anomaly detection system has been developed, which can automatically identify and mark abnormal data points in the analysis results, greatly improving the accuracy and efficiency of anomaly detection.

[0181] First, a multi-level data preprocessing pipeline was constructed to clean and standardize the original analysis results. The pipeline includes steps such as missing value filling, outlier screening, data standardization, and feature engineering. In the missing value filling stage, a time series-based interpolation algorithm was used to consider the time correlation of the RF signal to estimate the missing data more accurately. The outlier screening adopted an improved Z-score method, which dynamically adjusted the threshold to adapt to the characteristics of different types of signals.

[0182] Adaptive normalization technology is introduced in the data standardization stage, which automatically selects the most suitable normalization method according to the distribution characteristics of different parameters, such as Min-Max normalization or Z-score normalization. The feature engineering stage focuses on the time-frequency characteristics of RF signals, and extracts a series of time-frequency domain features through methods such as wavelet transform and Hilbert-Huang transform, enriching the input dimension of anomaly detection.

[0183] The preprocessed data is fed into a carefully designed deep learning model. The model uses a hybrid architecture based on long short-term memory network (LSTM) and convolutional neural network (CNN). The LSTM layer is responsible for capturing the long-term temporal dependencies in the RF signal analysis results, while the CNN layer focuses on extracting local feature patterns. This hybrid architecture can consider both the temporal characteristics of the signal and the local abnormal patterns, significantly improving the accuracy of anomaly detection.

[0184] The model is trained using an innovative semi-supervised learning method. First, a large amount of normal data is used to train the model to learn normal patterns. Then, a small amount of known abnormal samples and artificially synthesized abnormal samples are introduced to fine-tune the model's sensitivity to abnormalities. This method effectively solves the problem of scarcity of abnormal samples in practical applications, while improving the model's generalization ability to new types of abnormalities.

[0185] In order to further improve the interpretability of anomaly detection, an attention mechanism is integrated into the model. By analyzing the attention weights, we can intuitively understand which features or time points the model focuses on when making anomaly judgments. This not only improves the credibility of the detection results, but also provides valuable clues for subsequent fault diagnosis.

[0186] Based on anomaly detection, an adaptive threshold adjustment algorithm was developed. The algorithm dynamically adjusts the threshold for anomaly determination by analyzing historical detection results and expert feedback. The algorithm uses the Bayesian optimization method to effectively reduce the false alarm rate while maintaining a high detection rate.

[0187] In order to deal with the possible correlation anomalies in the multi-dimensional analysis results, a correlation anomaly detection module based on graph neural network (GNN) is implemented. This module constructs the analysis results of different dimensions into a relationship graph, learns the complex dependencies between nodes through graph convolution operations, and thus identifies correlation anomalies that are difficult to find in a single dimension.

[0188] Finally, an anomaly classification and tracing system was designed. The system uses clustering algorithms to classify detected anomalies and automatically generates a preliminary analysis report on the causes of anomalies in combination with an expert knowledge base. By establishing a mapping relationship between anomaly patterns and possible causes, the system can provide technicians with targeted troubleshooting suggestions, greatly shortening the fault diagnosis time.

[0189] This deep learning-based anomaly detection system can not only accurately identify and mark abnormal data points, but also provide possible causes of anomalies and treatment suggestions. Through continuous learning and updating, the system's detection capabilities are constantly improving, providing strong support for the quality control of RF signal analysis.

[0190] Figure 2 A schematic diagram of the structure of a system design system supporting automatic testing of multiple radio frequency parameters according to an embodiment of the present invention is shown as follows: Figure 2 As shown, the system comprises:

[0191] The first unit is used to receive and parse a test task, wherein the test task includes information of a radio frequency device to be tested and multiple radio frequency parameter test items; generate a test configuration file according to the information of the radio frequency device to be tested and the multiple radio frequency parameter test items, wherein the test configuration file includes a radio frequency range, a power level, a modulation mode, a bandwidth, and a test duration; and dynamically adjust the test configuration file based on historical test data using an adaptive optimization algorithm to generate an optimized test configuration file;

[0192] The second unit is used to control the RF signal generator to generate a test signal according to the optimized test configuration file, and the parameters of the test signal match the parameters in the optimized test configuration file; input the test signal into the RF device to be tested, and start the real-time monitoring module at the same time, and the real-time monitoring module is used to monitor the temperature, voltage and current of the RF device to be tested; when the monitored temperature, voltage or current exceeds a preset threshold, automatically adjust the power level of the test signal or suspend the test; control the RF signal analyzer to receive the output signal of the RF device to be tested, and sample and digitize the output signal;

[0193] The third unit is used to perform multi-dimensional analysis on the digital signal processed by the RF signal analyzer, including spectrum analysis, modulation quality analysis and power measurement; use deep learning algorithms to detect anomalies in the analysis results, identify and mark abnormal data points; generate a test report based on the analysis results and anomaly detection results, and the test report includes test results, performance curves and abnormal data analysis of various RF parameters; compare the test report with historical data, and generate a performance trend chart through time series analysis; based on the test results and performance trends, use the expert system to generate optimization suggestions and predictive maintenance plans for the RF equipment to be tested.

[0194] According to a third aspect of the embodiments of the present invention,

[0195] An electronic device is provided, comprising:

[0196] processor;

[0197] a memory for storing processor-executable instructions;

[0198] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0199] According to a fourth aspect of the embodiments of the present invention,

[0200] A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.

[0201] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0202] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A system design method supporting automatic testing of multiple radio frequency parameters, characterized in that: include: Receiving and parsing a test task, wherein the test task includes information of a radio frequency device to be tested and a plurality of radio frequency parameter test items; Generate a test configuration file according to the RF device information to be tested and multiple RF parameter test items, wherein the test configuration file includes a RF frequency range, a power level, a modulation mode, a bandwidth, and a test duration; dynamically adjust the test configuration file based on historical test data using an adaptive optimization algorithm to generate an optimized test configuration file; According to the optimized test configuration file, controlling the radio frequency signal generator to generate a test signal, wherein the parameters of the test signal match the parameters in the optimized test configuration file; The test signal is input into the radio frequency device to be tested, and a real-time monitoring module is started at the same time, wherein the real-time monitoring module is used to monitor the temperature, voltage and current of the radio frequency device to be tested; when the monitored temperature, voltage or current exceeds a preset threshold, the power level of the test signal is automatically adjusted or the test is suspended; the radio frequency signal analyzer is controlled to receive the output signal of the radio frequency device to be tested, and the output signal is sampled and digitized; Perform multi-dimensional analysis on the digital signal processed by the RF signal analyzer, including spectrum analysis, modulation quality analysis, and power measurement; use deep learning algorithms to detect anomalies in the analysis results, identify and mark abnormal data points; generate a test report based on the analysis results and anomaly detection results, the test report includes the test results of various RF parameters, performance curves, and abnormal data analysis; compare the test report with historical data, and generate a performance trend chart through time series analysis; Based on the test results and performance trends, the expert system generates optimization recommendations and predictive maintenance plans for the RF devices under test.

2. The method according to claim 1, characterized in that: Using an adaptive optimization algorithm, dynamically adjusting the test configuration file based on historical test data to generate an optimized test configuration file includes: Constructing a historical test database, wherein the historical test database contains test configurations of radio frequency devices and corresponding test results, wherein the test configurations include frequency range, power level, and modulation mode, and the test results include test duration, test result accuracy, and abnormal conditions during the test process; According to the basic information of the radio frequency device to be tested, the test configuration of similar devices is selected from the historical test database as the initial particle group; Calculating a fitness value for each particle in the initial particle swarm, wherein the fitness value is calculated by a fitness function, wherein the fitness function considers three factors, namely, the inverse of the test duration, the result accuracy, and the coverage, and assigns different weights to each factor, and obtains a comprehensive score by weighted summation; Using a particle swarm optimization algorithm to update the position and velocity of particles, the position of the particles represents the specific value of the test parameter; Introduce the crossover and mutation operations of the genetic algorithm to perform crossover and mutation on the updated particle swarm to generate a new particle swarm; According to the results of each iteration, the inertia weight of the particle swarm optimization algorithm and the crossover and mutation probability of the genetic algorithm are dynamically adjusted; Repeat the fitness evaluation, particle position update, crossover mutation and dynamic adjustment steps until the preset number of iterations is reached or the improvement of the fitness value for multiple consecutive iterations is less than the preset threshold; Select the particle with the highest fitness in the final particle swarm as the optimized test configuration; Integrate the optimized test configuration and its corresponding expected test duration and possible abnormal conditions into an optimized test configuration file; Executing an online learning mechanism to add newly completed test data to the historical test database and trigger an incremental learning process to update the parameters of the optimization algorithm; The fitness function is calculated by respectively assigning weights to the inverse of the test duration, the result accuracy and the coverage and summing them up.

3. The method according to claim 1, characterized in that According to the optimized test configuration file, controlling the radio frequency signal generator to generate a test signal, wherein the parameters of the test signal match the parameters in the optimized test configuration file, comprises: Parsing the optimized test configuration file to extract key parameters of the test signal, the key parameters including frequency range, power level, modulation mode and bandwidth; Converting the key parameters into low-level control instructions of the radio frequency signal generator through a parameter mapping engine, wherein the parameter mapping engine uses a pre-established parameter mapping table for conversion; Based on software-defined radio technology, field programmable gate arrays are used to implement digital signal processing, and digital-to-analog converters are used to generate radio frequency signals; Execute digital pre-distortion algorithms to compensate for nonlinear distortion of the DAC and RF front end at different frequencies and powers; Utilize the parallel processing capability of field programmable gate arrays to pre-calculate and cache multiple sets of signal parameters to achieve fast parameter switching; The power and frequency of the output signal are continuously monitored by a high-precision power detector and frequency counter; When it is detected that the power or frequency deviation of the output signal exceeds a preset threshold, an automatic calibration procedure is triggered to adjust the internal parameters of the RF signal generator; According to the modulation parameters in the test configuration file, select or dynamically generate a baseband signal that meets the requirements from the modulation signal library; Use digital up-conversion technology to convert the baseband signal into the required RF signal; According to the channel model specified in the test configuration file, the generated RF signal is processed in real time to simulate the interference and attenuation effects in the real transmission environment; The entire test process is coordinated through an intelligent sequence controller, which automatically schedules the parameter setting, switching and calibration operations of the RF signal generator; The intelligent sequence controller is also responsible for synchronous control with other test equipment to ensure the consistency of the test process; During the test, the output of the RF signal generator is continuously monitored, and the signal parameters are dynamically adjusted based on the real-time monitoring results to keep the test signal continuously matching the parameters in the optimized test configuration file.

4. The method according to claim 1, characterized in that Perform multi-dimensional analysis on the digital signal processed by the RF signal analyzer, including spectrum analysis, modulation quality analysis and power measurement, including: Performing an improved fast Fourier transform algorithm on the digital signal to convert the digital signal into the frequency domain; Fine interpolation is performed between adjacent points of the fast Fourier transform result to improve the frequency resolution of the spectrum; According to the characteristics of the digital signal, the most suitable window function is automatically selected for fast Fourier transform, wherein the window function includes Hanning window and Blackman window; executing an intelligent peak detection algorithm to identify the main signal, harmonic components and interference signals, wherein the intelligent peak detection algorithm distinguishes between true signal peaks and noise peaks by setting dynamic thresholds and applying pattern recognition techniques; Perform digital down-conversion and symbol timing recovery on digital signals to extract in-phase and quadrature components of baseband signals; The extracted in-phase component and quadrature component are processed by an adaptive equalization algorithm to compensate for the linear distortion introduced by the channel; Calculate error vector magnitude, amplitude error and phase error, wherein the error vector magnitude is calculated using a weighting factor to reflect the modulation quality of different symbol positions; Draw and analyze constellation diagrams in real time, and quantitatively evaluate the modulation accuracy of signals by counting the distribution characteristics of constellation points; For FM signals, the modulation index and frequency deviation are calculated by the instantaneous frequency detection algorithm; for AM signals, the modulation depth and distortion are analyzed by the envelope detection technology; Adopt digital integration technology, combined with real-time power calculation algorithm of sliding window, to measure average power and peak power; Regularly inject a calibration signal of known power to compensate for gain drift and nonlinear errors in the measurement link in real time; Monitor ambient temperature in real time and apply pre-established temperature-error models to eliminate the impact of temperature changes on power measurement; Utilize a multi-dimensional data fusion engine to comprehensively process the results of spectrum analysis, modulation quality analysis, and power measurement; The multi-dimensional data fusion engine uses machine learning algorithms to analyze the correlation between various indicators and automatically identify potential signal anomalies and performance issues; Compare the analysis results with the preset standards, automatically generate a test report, and provide targeted optimization suggestions.

5. The method according to claim 1, characterized in that Using deep learning algorithms to perform anomaly detection on analysis results, identify and mark abnormal data points including: Perform multi-level data preprocessing on the original analysis results, including: filling missing values ​​based on a time series interpolation algorithm that takes into account the time correlation of the RF signal; using an improved Z-score method to initially screen outliers, dynamically adjusting the threshold to adapt to the characteristics of different types of signals; performing adaptive normalization to automatically select the most suitable normalization method based on the distribution characteristics of different parameters; extracting time-frequency domain features through wavelet transform and Hilbert-Huang transform; Inputting the preprocessed data into a hybrid deep learning model, wherein the hybrid deep learning model includes a long short-term memory network layer and a convolutional neural network layer; A semi-supervised learning method is used to train the hybrid deep learning model, which includes using a large amount of normal data to train the model to learn normal patterns, and introducing a small amount of known abnormal samples and artificially synthesized abnormal samples to fine-tune the model; Integrate attention mechanisms in hybrid deep learning models and analyze attention weights to improve the interpretability of anomaly detection; Execute an adaptive threshold adjustment algorithm, analyze historical detection results and expert feedback, and use the Bayesian optimization method to dynamically adjust the threshold for abnormality determination; Build a correlation anomaly detection module based on graph neural network, construct the analysis results of different dimensions into a relationship graph, and learn the complex dependency relationships between nodes through graph convolution operations; Use clustering algorithms to classify detected anomalies and generate a preliminary analysis report on the causes of anomalies in combination with expert knowledge base; Establish a mapping relationship between abnormal patterns and possible causes to provide technicians with targeted troubleshooting suggestions; Continuously update and optimize the anomaly detection system, including updating training data, adjusting model parameters, and optimizing anomaly classification rules.

6. A system design system supporting automatic testing of multiple radio frequency parameters, used to implement the method described in any one of claims 1 to 5, characterized in that: include: The first unit is used to receive and parse a test task, wherein the test task includes information of a radio frequency device to be tested and a plurality of radio frequency parameter test items; Generate a test configuration file according to the RF device information to be tested and multiple RF parameter test items, wherein the test configuration file includes a RF frequency range, a power level, a modulation mode, a bandwidth, and a test duration; dynamically adjust the test configuration file based on historical test data using an adaptive optimization algorithm to generate an optimized test configuration file; A second unit is used to control a radio frequency signal generator to generate a test signal according to the optimized test configuration file, wherein parameters of the test signal match parameters in the optimized test configuration file; The test signal is input into the radio frequency device to be tested, and a real-time monitoring module is started at the same time, wherein the real-time monitoring module is used to monitor the temperature, voltage and current of the radio frequency device to be tested; when the monitored temperature, voltage or current exceeds a preset threshold, the power level of the test signal is automatically adjusted or the test is suspended; the radio frequency signal analyzer is controlled to receive the output signal of the radio frequency device to be tested, and the output signal is sampled and digitized; The third unit is used to perform multi-dimensional analysis on the digital signal processed by the RF signal analyzer, including spectrum analysis, modulation quality analysis and power measurement; use a deep learning algorithm to perform anomaly detection on the analysis results, identify and mark abnormal data points; generate a test report based on the analysis results and the anomaly detection results, the test report including the test results of various RF parameters, performance curves and abnormal data analysis; compare the test report with historical data, and generate a performance trend chart through time series analysis; Based on the test results and performance trends, the expert system generates optimization recommendations and predictive maintenance plans for the RF devices under test.

7. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method described in any one of claims 1 to 5.

8. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images