Researches On Test Nodes Optimal Selection And Test Generation Of Analog Circuit

Both test nodes optimal selection and test stimuli generation are important parts of analog circuit fault diagnosis and design for testability. The study of these two technologies has great significance on reducing the circuit test complexity and design cost and enhancing the circuit fault diagnosis accuracy.In this paper, the basic principles of optimal for analog circuit test nodes and test stimuli are introduced, and a new method which is effective and robust for the optimization of analog test nodes and test-stimuli is proposed. The main research works of this dissertation can be summarized as follows:(1) A new method for analog circuit test nodes optimal selection which is based on inner-and inter-class sensitivity factor and fault isolation degree is proposed. The inner-and inter-class dispersion of sampling data which is used to define the test nodes'sensitivity factor is calculated first, reordering the test points according to the value of the sensitivity factor, and then computing the isolatable fault group and degree of the reordered test nodes by KNN network, finally the optimal test nodes set can be selected with guidance of the proposed algorithm. Experimental analysis proved that, compared with other related literatures, the size of the optimal test nodes set which was derived by the proposed method is much smaller, and when select a same size test nodes set ,the proposed method has the ability of giving a better result.(2) A new method for test-stimuli optimization which is based on genetic algorithm is proposed. The features of genetic algorithm such as probabilistic global optimization and parallel search capability are considered in the proposed method, and the method takes the maximum of different fault circuit response distribution as the criterion to optimize the test-stimuli. The circuit AC sweep simulation and modeling analysis are executed first ,then the objective function is defined based on the calculation of inter-class and inner-class distance for different fault circuit response, finally the test-stimuli corresponding sensitive factor would be calculated, through the above operations, the optimal test-stimuli can be generated successfully by genetic algorithm. Experimental result shows that the optimal test-stimuli generated by the proposed method can reduce the ambiguity in the distribution of fault features of the circuit greatly, and achieve a satisfactory diagnostic result.The research of this paper is funded by National Natural Science Foundation of China (60871009) and Aeronautical Science Foundation of China (2009ZD52045).