Research On The Test Optimization Method Of Large-Scale Analog Circuit Based On Genetic Algorithm

As science and technology advance,electronic products are becoming more and more functional,resulting in an increase in the size and complexity of the corresponding circuits,which brings challenges to the testing and diagnosis of circuits.Due to the advantages of transfer function and simple operation,small linear circuits have achieved considerable results in fault diagnosis,but the large-scale circuit has a large number of devices,test nodes and a large amount of data,which increases the difficulty of research.There are few studies on large-scale nonlinear circuits.Therefore,this thesis mainly studies the test optimization scheme of large-scale analog circuits.By incorporating fault diagnosis theory,this problem is converted into an optimization problem,where a genetic algorithm is employed to generate an optimal test plan for the circuit,and the results are subsequently verified.The specific work is as follows :1.Establish a defect detection process for large-scale analog circuits.According to the characteristics of large-scale analog circuits,Hspice simulation software is used,and then the fault model of the circuit is established.The corresponding fault model is used to inject defects into the circuit,and the algorithm of flattening the hierarchical network table is designed.Finally,the simulation file is processed accordingly,and the feature information is extracted to facilitate the subsequent defect detection.2.Research on defect detection method based on simulation.According to the established process of large-scale circuit defect detection,the corresponding simulation method is set up.Subsequently,a defect detection algorithm is developed based on the file characteristics generated by different simulation methods.To address the distinct characteristics of transient simulation sampling points,a curve-fitting-based area defect detection approach is proposed.3.The test optimization scheme for large-scale analog circuits is designed based on single-objective genetic algorithm.By using the feature vector obtained from different simulation methods,a threshold value is then established for defect detection,allowing for the construction of a fault test matrix for the circuit.A genetic algorithm is employed to optimize the measurement points within the circuit.Taking the bandgap circuit and LDO circuit in the IEEE standard circuit as the experimental object,the proposed algorithm undergoes verification,with its superiority over traditional algorithms being demonstrated through comparative analysis.4.A multi-objective evolutionary algorithm was utilized in the design of a test scheme for large-scale analog circuits.Since the circuit testability index(FDR,FIR)and the minimum test cost we want to obtain are a mutually restrictive problem.To generate a test optimization scheme,a multi-objective evolutionary algorithm is employed.Using the fault test matrix,a multi-objective fitness function is constructed with the optimization objectives being the minimum test cost and the highest testability index.The nondominated solution set is then searched using this function as a guide.The multi-objective evolutionary algorithm is utilized to resolve conflicts between the two objectives..Then the designed algorithm is experimentally verified on the bandgap circuit and LDO circuit.Finally,the advantages of the multi-objective evolutionary algorithm compared with the single objective are compared.