Fault Diagnosis Of Analog Filter Circuit Based On Complex Model

The fault diagnosis of analog circuits is still a hot topic currently.Although there have been deep researches on this topic,there are few researches on the influence of component tolerance.The complex domain modeling method in this thesis can effectively model the hard faults and soft faults of analog circuit.A deeper research on the influence of tolerance on the fault diagnosis of the circuit is done and a countermeasure to this problem is proposed here.This thesis mainly studies the following issues:1.Research on frequency domain fault diagnosis under complex model is conducted.The frequency response with multiple frequency components excitation is studied,combined with the complex model,to deal with the tolerance problem,and improve the robustness of diagnosis.With the excitation from a single frequency to the circuit under test,two different faults in the complex plane may have more than one intersection points in the modeling curve,and thus there may be a problem namely aliasing.The inherent nature of the circuit above causes the failure of isolation among different faults.The excitation containing more than one frequency components is used to circuit expands the circuit,so that better "fault isolation" can be achieved.2.In an analog circuit,the change of the parameters of a circuit component within a certain range of its nominal value is regarded as tolerance instead of a fault.Fault diagnosis of analog circuit is approximately equivalent to pattern recognition problems and classification problems.Using pattern recognition and classification tools,faults to be diagnosed are classified into corresponding categories,so intelligent fault diagnosis can be realized.If two components that are not in the same fuzzy group have intersecting non-failure regions,these components can't be distinguished in the actual circuit due to the influence of tolerance.Therefore,it is necessary to study the tolerance of circuit components to improve the fault isolation rate and robustness.In this paper,Support Vector Machine(SVM)tool is used to study the influence of tolerance on analog circuit fault diagnosis.3.Combined with the oscillation principle,fault diagnosis under complex domain without external excitation is realized.Self-oscillating have been the focus of research in recent years due to their characteristics of generating circuit output without external excitation.Self-oscillating in the original circuit under test can be achieved in two ways.The first one is to add a feedback circuit.The second one is the self-renovation of the circuit.Oscillation can be achieved through either of the two ways or through the combination.After the oscillation of CUT starts,if there is a failure,the oscillation of the circuit after reconstruction will stop,or the frequency,amplitude will deviate from the normal value.Thus,measuring the parameters of the oscillating signal and comparing them with the normal value make it possible to detect the faults in the analog circuit.Parameters of the circuit can be combined with the existing complex model to realize the fault diagnosis of analog circuit based on complex model.