Research On Key Technologies Of Design For Testability For Health State Evaluation

As sharp rise of complexity and integration of equipments, requirements of maintenance and logistic support for systems are confronting with many challenges. In order to decrease the maintenance costs and improve the availability and battle effectiveness of equipments, the maintenance and logistic pattern is undergoing a transition from the corrective maintenance, condition-based maintenance(CBM), to the predictive CBM(PCBM) and Autonomic Logistics(ALS). Health State Evaluation(HSE) is a critical technology of ALS, which can drive the decision-making mechanism of ALS. Therefore, HSE is the premise and foundation of realization the function of ALS.Relevant researches has shown that the HSE performance mainly relies on the ability of equipments’ testing and sensing for fault evolution process. Obviously, the Design for Testability(DFT) and product functionality developed in parallel can cope with the problems above effectively. Therefore, this dissertation was supported by the National Foundation Project “The DFT theory and technology oriented for equipment health management” and “The method of sensor selection and optimization based on the testing and timing for fault evolution process of PHM systems and testability mechanism”. The purpose of this paper is to cope with the key problems in the DFT considering the essential requirements of HSE, such as the testability modeling and analysis, test point & sensor selection and optimization, the HSE reasoning model with uncertainty test outcomes and the solving method for it and so on. The main contents of the dissertation are as follows:1. The testability modeling and analysis for HSETo address the problem that the testability model for HSE need to construct the essential relationships between the extent of faults(or fault growth) in the fault evolution process and available tests of equipments, the dissertation develops the contents of traditional Failure Mode, Effect and Critical Analysis(FMECA) by adding the fault evolution mechanism, and a fault analysis method called Failure Mode,Evolution Mechanism,Effect and Critical Analysis(FMEMECA) is proposed. Thus, this dissertation proposes a novel testability modeling called Fault Evolution Testability Modeling(FETM) and introduces its basic theory, flow and modeling method. First, the fault progression-related information of each Unit Under Test(UUT) in equipments is obtained by FMEMECA, and the boolean Failure-Symptom(FS) dependency matrix can be generated; Second, the symptom-test(ST) dependency matrix can be obtained by using bond graph methodology which describes the responses of the systems’ test point due to different fault progressions. The FS, together with ST, describe the relationships between fault evolution and tests. In addition, according to the essential requirements of HSE, besides Fault Detection Rate(FDR) and Fault Isolation Rate(FIR), Fault Track Rate(FTR) and Fault Prediction Rate(FPR) are proposed. The four testability indices(such as FDR, FIR, FTR, FPR) mentioned above are used to evaluate the fault detectability, isolability, trackability and predictability of equipments, respectively. Finally, a strategy and method of testability analysis for HSE are proposed, which establish the foundation for testability scheme optimization and the HSE reasoning based on FETM.2. The testability scheme optimization for HSETo solve the two key problems in the testability scheme optimization for HSE, a Test Selection and Optimization(TSO) method based on FETM and a Sensor Selection and Optimization(SSO) method based on the timing and sensitivity of sensors are proposed. First, a FETM based TSO model, which is used to minimize the test number, is built according to the FETM and testability analysis results, and the different constraints of testability indices(such as FDR, FIR, FTR, FPR) during in the internal and external test design process are taken into consideration respectively. Afterwards, an Adaptive Genetic Algorithm and Simulated Annealing(AGASA) is introduced to settle the above model by integrating the major advantages of the GA(such as broad applicability, flexibility, ease of implementation and the potential of finding near-optimal solutions) and the advantages of Simulated Annealing(SA) which is more effective in finding the global minima. Second, in order to select a more effective sensor set in the optimized tests, the attributes of each available sensor in the optimized tests are analyzed, and the trackability of sensors for fault evolution process and the detectability of sensors for incipient fault are quantified. Furthermore, by analyzing different optimal objects and constraints for internal and external tests and considering two instances that a test is realized by several sensors and several tests are realized by a sensor, the SSO model considered timing and sensitivity of sensors is built. Finally, the AGASA is introduced to solve the SSO model.3. The HSE reasoning based on FETMIn order to cope with the HSE problems with uncertainty tests and static/dynamic reasoning, this dissertation describes the extent of a fault in its fault evolution process by defining several discrete health states and discretizes the test ranges by several general tests. Afterwards, a Health State-General Test(HSGT) matrix can be generated by FETM which is updated according to the testability scheme optimization results. Thus, the static/dynamic HSE reasoning models with uncertainty tests are formulated based on HSGT by using the Bayes rules. To address the HSE reasoning model with the complexity of constraints and non-linear objectives, the dissertation introduces the Lagrangian Relaxation and Adaptive Genetic Algorithm(LRAGA). The solution scheme can be viewed as a two-level coordinated solution framework for the HSE problem. At the top level, the Lagrange multipliers are updated by using adaptive genetic algorithm(AGA). At the bottom level, each of the sub-problems is solved by using AGA. The primary advantages of this method are that it provides a more precise solution and converges to the optimal solution at a faster rate than those of conventional methods by adaptive selection of the chromes.4. Software platform development and study on the case verificationBased on the results of the above theoretical research, a software tool for DFT of HSE systems is designed and developed. The dissertation uses an electromechanical actuator as an example to validate the efficiency of the proposed techniques, such as the modeling approach based on FETM and testability analysis for HSE, the TSO based on FETM, the SSO based on timing and sensitivity of sensors, and the static/dynamic HSE reasoning with uncertainty tests. This experimental results show that the theories and methods proposed in this dissertation provide a powerful way for DFT of HSE systems, and the researches has a good theory foundation and applied values of engineering.