摘要
The probe card is an essential tool for measuring the yield of semiconductor wafers and is indispensable in semiconductor testing. To ensure that the probe card functions properly after multiple uses, it is crucial to maintain and verify the performance and accuracy of each probe to avoid impacting product quality and delivery time. This study employs network reliability characteristics to convert the probe card's circuit diagram into a general network for analysis, resulting in the reliability values for the normal operation of each probe within the probe card, thus facilitating the monitoring of its operational reliability. As the scale of the network diagram increases, the complexity of solving it also grows. Determining precise reliability values using traditional mathematical methods is an NP-hard problem. Therefore, this study proposes an Optimal Computing Budget Allocation and Monte Carlo Simulation based on the Binary-Addition-Tree algorithm (BAT-MCS-OCBA) to quickly obtain approximate reliability values. BAT-MCS-OCBA combines the accuracy of the Binary-Addition-Tree algorithm (BAT) with the concept of super vectors, which simplifies the extensive time spent on enumerating variable combinations in BAT. It also utilizes Monte Carlo Simulation (MCS) to simulate subsequent scenarios not considered by the super vector, enhancing accuracy. Finally, Optimal Computing Budget Allocation (OCBA) is applied to allocate resources effectively during simulations, enabling the algorithm to achieve faster approximate reliability values with improved solution quality. © 2025 IEEE.