Reseach On Analog Circuit Diagnosis And Testability Based On Amplitude And Phase Charactorstics

Fault diagnosis of analog circuit is a complex and difficult work, however, it is also an important work. According to statistics,in mixed signal circuits,80% faults occur in the analog circuit sections,and its corresponding test time assumption might be as high as 80% of total test time.At present, due to the influence of circuit parameter continuity and components’ tolerance, the analog circuit fault diagnosis research progress does not go well.In this paper,one puts forward a novel fault modeling primarily based on amplitude and phase information from circuit voltage responses for analog circuit fault diagnosis. Besides, this paper also promotes an involved discussion of tolerance dealing and testability analysis problem. The author’s work include following aspects:1. Content 1st: Based on the phasor analysis of linear analog circuits, it is proposed that a new method for fault voltage modeling. Thereto the hard fault voltage response could be estimated through adding auxiliary excitation or branch while measuring the output voltage before its actual failure occurs, while the continuous soft fault voltage responses expression in analog circuit could be established with these hard fault voltages. This method has the advantages of simple deduction and convenient application.2. Content 2nd: Furthermore, through the establishment and observation of the single fault or double-faults model, there are two novel fault diagnosis algorithms proposed At last, this paper gives general idea of K- fault modeling in linear analog circuit(k = 3, such as triple-faults). The experiment studies show that the fault models and diagnosis algorithms are used under the environment of tolerance. Its fault component location is accurate, while the deviated value of fault parameter is small.3.Content 3rd: Based on the idea of fault voltage modeling in content 1stand Content 2nd,this paper proposes fault voltage modeling of nonlinear resistance and nonlinear controlled source in turn.Here, such these two non-linear component fault modeling is of simple and efficient characteristics. Through nonlinear components fault modeling and the minimum distance method,it is to solve the fault diagnosis problem under DC test conditions. The practical research shows that its diagnosis application is simple and effective.4. Content 4th Based on the idea of fault voltage modeling in content 1stand Content 2nd,this paper obtains in the first place corresponding fault current modeling results based on the circuit voltage fault modeling presented in content 1st, with wich a new circuit testability matrix can be proposed. Practical examples show that it obtain the consistent conclusions as what the traditional method gets, utilizing this testability matrix for testability analysis research in typical circuits, and analysis is very simple.5. Content 5th :this paper presents a network modeling expression for fault diagnosis(detection and isolation) information. And according to its network parameters, it designs new test-points selection algorithm. Statistical test and specific example verification are to show that the results of these new test-points optimization algorithms could get better optimization results, compared with what entropy-based algorithm or entropy–plus-exclusion-based test-points selection algorithm can do.At last,based on the fault modeling in content 1st,it proposes a new test-frequencies selection method, which means that an optimization of frequencies combination is carried out, with the aim of reducing the number of common fuzzy fault-pairs. This requires the optimization of the numerical column rank of the corresponding matrix or its minimum non-zero singular value.6. Content 6th: Based on slope fault model and the voltage fault model proposed in content 1st, the fifth chapter proposes a method of dealing tolerance, in which both amplitude and phase characteristics are estimated with statistical theories at the same time. The respective numerical ranges of estimated signal modulus increment ratio and the phase angle difference are given. These numerical change intervals are the fault diagnosis feature in the fault dictionary, which are used for the fault detection and fault isolation. Analysis in examples shows that using different signal modulus increment ratio features and its phase angle D-value feature for statistical estimation, sometimes a better est-point selection results could be obtained, while it provides more reliable fault detection and isolation result to us.