Test Selection And Veritication Of Analog Circuits Based On Multi-feature Model

In the electronic system, there exists digital and simulated sections. But the majority of failures appear in the analog circuit, so we need test and analysis the analog circuits, but with the increasing complexity of electronic systems, it is becoming more and more difficult for the test selcetion of analog circuits. Now the existing modeling methods are imprecise and relatively complex, and the methods of test selection do a low searching efficiency and their accuracies are not very tall. For those issues metioned right row, in this article we present a multi-feature-signal model to perform analysis for the simulated circuits and due to this model we use a novel algorithm to do test selection. The main research work are as bellow:1. multi-feature model and its automatic modeling method. We raise a number of testing characteristic signals, such as timed features, frequencied features,statistical features and wavelet features to compose feature vetors based on the response waveform on test nodes of analog circuits combined multi-signal model with multi-profiling techniques. Then we can generate a dependent matrix include tests and faults to perform testability analysis, and achieve the automatic modeling of multi-feature model of analog circuits on the software level.2. To achieve the test selection based on the multi-feature model. We applied the algorithm to the issue of test selection based on dependent matrix. To further make the better optimization and accuracy of the algorithm,we decide to optimize the initial population by introducing the group reverse learning strategy and the make the memory of particle swarm optimization algorithm into the algorithm. Thus we can prevents the algorithm staying near the local optimal solution, and also speed up the search then we can show that this algorithm has better performances.3. Testing and validation. Through contrasting genetic algorithms, particle swarm algorithm, standard gravitational algorithm with each other, we discover the modified algorithm gravity research has faster speed and accuracy in searching optimal test set. From the view of diagnostic we analysis the test signals of choosing test, and the results show it has some better advantages. Using the two-stage four-opamp biquad low-pass filter as the instance for testing and verifying, through a multi-feautre modeling analysis, we will obtain the multi-feature model of this analog circuit. Then we use the metioned algorithm for the test selecting and validating.4. To achieve the development of multi-feature analog circuit modeling software platform. The platform is based on multi-feature signal model and it has interactive interface and is consisit of clien-side and server-side. Using this platform we can simulate and analysis the simulated circuits and extracts the features, then it will generate the dependent matrix. This platform includes user logining module, reading the file and seting the simulation properties module, model displaying and testabiligy analysis module, test selection module and the compared results of algorithms. And we test this software platform in functional aspect and reliability aspect.