The miniaturization and increasing material complexity of semiconductor packaging necessitate advanced automated systems for sorting abnormal chips. Although sensor networks in the system enable a high‑throughput method for separation, the optimal configuration is a multicriteria engineering challenge owing to trade‑offs among hardware sensitivity, computational latency, and deployment costs. In this study, we present an integrated optimization framework based on the fuzzy analytic hierarchy process (FAHP) to configure a multisensor sorting system for recycling abnormal chips. Supporting green manufacturing through resource-efficient sensor deployment and lower energy consumption. By integrating expert judgments and operational field data in FAHP, the framework balances interdependent parameters across sixteen indicators to identify the optimal sensor network distribution. Validation results of the framework on an operational micro‑semiconductor sorting line showed that the FAHP‑optimized sorting system improved defect classification precision from 89.4 to 98.1%. The system also reduced data‑processing queues and lowered localized edge‑inference latency to 0.18 s per item, maintaining continuous power consumption below 2.0 W. This framework applies symbolic decision analysis to a cyber‑physical sensor network, leading to the development of a reproducible methodology for optimizing resource‑constrained manufacturing environments.