Design And Implementation Of Fault Diagnosis System Software For Radio Frequency Front-end

With the development of communication and electronic technology,the radio frequency(RF)circuit,as an important component of electronic communication systems,has become increasingly complex.Therefore,research on RF circuit fault diagnosis methods is crucial for ensuring the reliability of communication systems.However,unlike low-frequency analog circuits,RF circuits exhibit highly complex fault characteristics,making it difficult to collect fault data,and adding fault diagnosis test points to the circuit can adversely affect its performance.Hence,RF circuit fault diagnosis is extremely challenging.This thesis focuses on the development of an RF front-end circuit fault diagnosis system based on the "XXX Equipment Communication System Fault Diagnosis System Development" project of a certain research institute.A one-click fault diagnosis software for the RF front-end circuit without additional test points was designed and developed.In response to the challenge of diagnosing faults in RF circuits without test points,this thesis conducted research and development in three aspects: fault mode analysis,embedded data acquisition software,and upper computer algorithm software.The main research content and work include:1.Failure mode and effect analysis was carried out on the tested RF front-end circuit.According to the technical specifications of the tested RF front-end circuit,a corresponding simulation circuit was built in the ADS simulation software.Through simulation fault injection,fault mode analysis was performed on the circuit to provide data support and theoretical basis for the development of the fault diagnosis software algorithm module.2.A set of RF front-end circuit data acquisition software was designed and implemented on an embedded platform.In order to achieve the best fault diagnosis effect,the most suitable excitation signal was selected to stimulate faults according to the circuit characteristics of the tested object,and GHz-level excitation signal generation and corresponding response signal data acquisition were implemented on the embedded data acquisition platform.3.The feature extraction of signal data was performed using the Improved Empirical Wavelet Transform Based on PSD and i Forest(IEWTPF)algorithm.Under the condition of no additional test points in the tested circuit,fault diagnosis was carried out only by analyzing the response signal of the circuit.In this case,a good signal feature extraction algorithm can greatly improve the accuracy of the software fault diagnosis.The IEWTPF algorithm implemented in this thesis is an improved empirical wavelet transform algorithm based on PSD theory and Isolation Forest algorithm,which has good feature extraction effect.4.A fault classification model was built based on the residual network theory.Based on the characteristics of multiple sets of AM-FM component signals output by the IEWTPF algorithm,a five-branch Res Net model was built for fault classification.It was verified that the fault isolation rate of the algorithm model can reach 90% for the measured RF front-end circuit.The developed software was applied to the RF front-end circuit fault diagnosis work of a certain research institute,and six types of faulty circuit boards were diagnosed.All faults on the circuit boards were successfully detected and isolated,and the software performed well in actual engineering applications.