Research On Testability Growth Test Theory And Method

Testability growth is a process that aims to improve the testability level of the equipment via finding and fixing the defects of design for testability.The testability design requirement listed in the design contract is the objective of testability growth.Testability growth test(TGT)is the most powerful way to enhance and reach the testability design requirement during the development process of equipment.Contrasting with reliability growth test which is already applied in engineering successfully,although there are substantial studies on testability design and update,testability growth is never treated as a system engineering problem,and the attentions paid to the TGT is neither not too much.The reason for this matter is the scarcity of the scientific and effective management model and strategy.The focus of this thesis is to solve the problems in TGT both theoretically and practicably.Specifically,the theory frame,technical model and test management methods for TGT are proposed,and a particular testability design and update issue in testability growth is studied.The contributions of this thesis are valuable for the scientific implementation of TGT.The main studies and contributions of the thesis are as follows.(1)The theory framework,definitions,technological process and engineering management of TGT are proposed and shaped.Firstly,the defects fixing methods and testability growth efficiency in each phrase of the equipment life cycle are analyzed separately.The comparisons reveal that TGT is the most effective mean to achieve the testability requirement among the different testability growth progressions.Secondly,the technological process and the corresponding definitions in TGT are given and explained in detail.At last,the framework of the TGT management theory is studied especially.The concepts of test planning,tracking and projecting in TGT management are reminded,and the important roles of TGT management are highlighted.All the above analysis and introductions put a foundation for the remainder of this thesis.(2)The TGT planning models and methods are proposed and studied for different TGT scenarios considering the change laws of testability metric in the growth process.All the proposed planning methods consider the factors that influence the testability growth speed and the components of the testability growth cost.Firstly,for the in-time correction based TGT,the dynamic planning model is given to specify the tolerable number of failed fault detection/isolation tests in each stage,so as to minimize the test cost.Secondly,for the delay correction based TGT,the test resources allocating problem arising in the growth process is studied and modeled,and the model is solved by Local Search and Lagrangian Relaxation Algorithm.At last,for the multi-stages testability growth process that is normal in the development period,the multi-stages growth object decision model and method are proposed,in which the optimal objective function is the total test cost,and constrained by the required testability metric.The case studies and simulations illustrated the validity and effectiveness of the given models.(3)The TGT tracking and projecting models and methods are proposed,including the testability metric evaluation method for TGT tracking based on the analysis of TGT data characteristics,and a testability growth model,using which the testability growth tracking and projecting curve can be drawn.Firstly,two Bayes based testability metric estimations are proposed for in-time correction based TGT tracking,considering the test planning information and metric increase constrain respectively.The merits and demerits are discussed by some numerical simulations.For the delay correction based TGT tracking,a Bayes estimation framework considering the multi-sources information and the existing study results is given.Secondly,in consideration of the role of testability growth model(TGM)to the TGT tracking and projecting,a parametric TGM is developed to depict the testability metric growth process considering the imperfect design update.At last,taken the TGT tracking results as model’s inputs,a genetic algorithm and particle swarm optimization hybrid method is utilized to estimate the TGM parameters,so as to draw the TGT tracking and projecting curves.Simulations show that the proposed TGM is convenient for the tracking and projecting curve drawing,and can reduce the tracking errors from Bayes estimation,and forecast the testability growth development trend conveniently.(4)The testability defects fixing problems encountered in TGT are solved,including the testability design update methods selection and the data based fault diagnosis update problem.Firstly,in order to define the testability defects uniformly,the imperfect test definition is extended,and the quantitative measure methods for testability defects are proposed considering the test data.According to the abominable degree of the exposed design defects,the possible defects fixing methods are selected.Secondly,a scheme consisted by the density based cluster,artificial immune system based data compression and hybrid learning is applied to handle data based fault diagnosis update problems.Hereby,the problems include the samples imbalance problem between the normal and fault data,the classifier update problem and Built-in Test Equipment(BITE)hardware limitation.Studies show that TGT data can support testability defects fixing and design update.The proposed method in this thesis can improve the diagnostic accuracy effectively when the data based fault diagnosis is utilized in the testability design.