A novel multidimensional fitness function genetic algorithm is proposed to optimize test vectors of R-2R Digital-to-Analog Converters (DAC). The proposed method employs distribution of characteristic vectors and the number of test vectors to formulate a multidimensional fitness function to search a non-dominate (ND) solution set. The searching process is directed by a ND sort method. Each individual in the ND set does not contain redundant test vectors. The test vectors are taken as the input excitation of the circuit under test (CUT) and the fault diagnosis is performed. As the number of test vectors increases, the accuracy of fault diagnosis also increases. The validity of the proposed method is verified by fault diagnosis. The average fault diagnosis rate is more than 85% for the R-2R network under the influence of tolerance. In addition, the parametric faults of \(\pm 50\mathbf{\%}\) deviation from the nominal value in each resistor of the R-2R network can be detected in the proposed method. Finally, the comparative experiments and results are briefly described in this paper.

提出了一种新的多维适应度函数遗传算法来优化 R-2R 数模转换器 (DAC) 的测试向量。所提出的方法利用特征向量的分布和测试向量的数量来制定多维适应度函数来搜索非支配（ND）解集。搜索过程由 ND 排序方法指导。ND 集中的每个个体不包含冗余测试向量。将测试向量作为被测电路（CUT）的输入激励，进行故障诊断。随着测试向量数量的增加，故障诊断的准确性也随之提高。通过故障诊断验证了所提方法的有效性。R-2R网络在容差影响下的平均故障诊断率超过85%。此外，\(\pm 50\mathbf{\%}\)与 R-2R 网络的每个电阻器的标称值的偏差可以在所提出的方法中检测到。最后，本文简要描述了对比实验和结果。