Research On Key Technology Of Intermittent Faults Diagnosis In Extreme Temperature Environment

Testability and Built-In Test(BIT) have been widely used in a variety of weapons and equipment. In their development and applications, false alarm(FA) is one of the most prominent problems, which can seriously affect the availability and readiness of equipment and present a serious challenge to the maintenance support. It is shown from the failure statistics and damage mechanism analysis that the environmental stress is one of the most important causes of equipment failure, especially for the causes of intermittent faults(IFs) and FAs. In Extreme Environment(EE), as the stress is more severe and stochastic, the IFs and FAs problems will become more prominent. Therefore, recognition of IFs and FAs caused by environmental stress is one of the key and most difficult issures in the research field of testability.In this thesis, in order to diagnose faults caused by temperature especially by extreme temperature, based on the analysis of failure mechanism, fault event evaluation(FEE) based on environmental stress and IFs diagnosis approach are studied.The main contents and conclusions are summarized as follows:1. Thermal failure mechanism analysis for Multi-Chip Module(MCM) in Extreme Temperature Environment(ETE)Aiming at answering the questions such as: where the electronic components prone to failure, what temperature characteristic parameters is sensitive to, what kind of failure will lead to, MCM is taken as an example for analysis where numerical simulation, theoretical analysis and experimental investigation are used. Firstly, the coupling of multi-field in MCM is analyzed and a decoupling strategy using finite element analysis is carried out. Secondly, the ETE concept is defined, the finite element model of MCM is constructed, and then the thermal and thermal stress analyses are used to study the thermal and thermal stress distribution in order to find the weakness of thermal stress failure. Thereafter, the interfacial model of the weakness is built and temperature characteristic parameters are analyzed. Finally, extreme temperature experiments are conducted. The results show that MCM is most sensitizing to high temperature, and intermittent thermal effect failure appears in extreme high temperature, which validates the correctness of simulation and theoretical analysis.2. FEE based on failure mechanism and grey relation analysis(GRA)Fault events are usually not directly observable. In order to diagnose the faults induced by environmental stress with bound delay, environmental stress is treated as fault event, then the fault events can be observed by evaluating the relative environmental stress. Firstly, according to the severity, the stress is divided into three intervals, respectively corresponding to normal, intermittent and permanent faults(PFs) events, and an associational model between environmental stress(key characteristic parameter) and fault events is built based on damage mechanism. Secondly, considering the faults induced by environmental stress are caused mainly by several major environmental stresses, in order to improve the veracity and the efficiency of FEE, the selection flow of the main environmental stress based on failure mode, effect and environmental analysis(FMEEA) are given and the key characteristic parameter selection approach based on interval grey relation analysis(IGRA) and program preference is proposed. Finally, threshold selection method based on support vector machine(SVM) and FEE based on IGRA and entropy are proposed, which can make the fault event ―observable‖ and lay the foundation for fault diagnosis.3. IFs diagnosis based on automata model and FEEThe model-based diagnosis of discrete-event systems(DESs) usually potentially presumes faults are permanent and do not take IFs into account. It thus can’t identify IFs. In this thesis, the diagnosis model is extended to include IFs and PFs, and then, a novel IFs diagnosis approach based on automata model and FEE is derived.(1) Firstly, logical automata which represent system normal, IFs and PFs are built, and the regulations of system state transition evoluting through fault and reset events along the traces of system are studied. Secondly, the label concept which carries system state and fault information is introduced and the label transfer function which denotes the system state transition regulations is derived. Thereafter, the construction of diagnoser is introduced and the diagnosability concepts of PFs and IFs are defined and analyzed. Finally, aeronautic gyroscope is given to illustrate the diagnosis process.(2) Aiming at the rigour of diagnoabilities of logical automata and the indeterminacy of system state transition, considering the probability of events or states occur, an IFs diagnosis approach based on stochastic automata is derived by appending each state an transition probability matrix which is used to update the system state assessment. This approach relaxes the system diagnosabilities and greatly enhances the ability of fault diagnosis.In summary, the key technical problems of feature parameter selection and FEE based on environmental stress in ETE are studied, and then an IFs diagnosis approach based on automata model and FEE is derived. MCM and aeronautic gyroscope are chosen as examples to verify and validate the proposed models and approaches, the result show that the proposed models and approaches are feasible and reasonable, therefore are of great engineering significance.