Research On Fault Diagnosis Of Analog Circuits Module Based On SVDD And D-S Evidence Therory

With the rapid development of science and technology,electronic technology continues to break through innovation.Electronic systems have become an indispensable part in many fields such as life,medical care,and equipment.As a result,the safety of electronic systems has come one after another.The technical requirements for the reliability of electronic devices have been examined.Also increasing.Many outstanding scholars have conducted in-depth research and a large number of experiments on analog circuit fault diagnosis methods.They mainly rely on training a large number of fault state samples to achieve analog circuit fault location.However,in practical applications,there are often some negative samples of some electronic devices that are not even available,and it is difficult to establish an effective classification model,resulting in the failure to find and replace the circuit faults in a timely manner.If the failure detection of the circuit is achieved by a single classification and the entire electronic system is replaced,it will result in a huge cost loss.Therefore,using the existing positive samples to complete the fault location is a difficult problem in the field of analog circuit fault diagnosis.To solve this problem,this paper takes the analog circuit module fault as the research object,and deeply studies the method of locating the fault module.For modularized analog circuits,a method based on Support Vector Data Description(SVDD)and D-S(Dempster-Shafer envividence theory)evidence theory for fault diagnosis of analog circuit modules is proposed.First,the circuit is divided according to the functional modules,and there is no feedback between the modules.The principle of non-public nodes exists in the parallel modules.Then the pulse signal and the sweep signal are used to stimulate the circuit,and the time domain and frequency domain voltage signals output by each module are extracted.The extracted features are processed through principal component analysis.SVDD linear core module fault detection model,and analysis of the module detection effect obtained from the time domain and frequency domain.Later,SVDD Gaussian nuclear and polynomial core detection models were proposed to improve the detection rate of module failures.Because the SVDD detection can only give the state information of each module,it needs to combine the location information of the module to find the source of the fault to locate and is more susceptible to noise interference.Therefore,the D-S theory is introduced to integrate the detection results of various models to locate the fault.Firstly,before the fault diagnosis,the above two models are obtained from the time domain and the frequency domain,and the detection results of each module of the circuit are obtained through simulation.The detection results of each module are taken as prior knowledge.During the diagnosis,the actual test results of each module of the circuit are obtained through the two models,and the obtained a priori data is used to configure a certain weight and credibility for the actual measurement result,and the fault module information is given,and the information module is completed through information fusion and decision failure.Fault Diagnosis of Analog Circuit Module Based on SVDD and D-S envividence Theory.In order to verify the effectiveness of the method,a serial band-pass filter circuit and dual bandpass filter circuit with parallel components were selected for simulation verification and physical verification.The results show that the proposed method has a good fault diagnosis effect on analog circuit module faults.