A Novel Exclusive Group Testing Strategy towards Large-Scale Electromagnetic Susceptibility Testing with Improved Efficiency

Electromagnetic susceptibility testing often involves a large number of repeated tests across various excitation input combinations (e.g., frequencies, waveforms) to identify sparsely distributed characteristics in the excitation domain that contribute to product failure. Group testing provides a cost-effective approach for large-scale fault detection in such sparsely distributed scenarios. However, obtaining an optimal grouping in multi-stage group testing remains challenging due to the transcendental nature of the equations involved in existing analytical solutions. This paper introduces a novel multi-stage group testing strategy that maximizes the number of groups by assigning each fault sample to as many distinct groups as possible. At each stage, the strategy estimates the maximum number of fault groups based on known prior probabilities, enabling the calculation of optimal groupings even in worst-case scenarios where the number of groups exceeds the number of fault samples. Through theoretical analysis and simulation, we demonstrate the strategy significantly reduces costs compared to traditional group testing, particularly with low prior probabilities and sufficient sample sizes. Our selection framework guides practitioners in choosing efficient testing strategies based on probability and sample size. Validation through simulations and practical implementation confirms substantial efficiency improvements over conventional methods.