This paper presents a flight-control self-destruction system to mitigate sensitive data leakage when unmanned aerial vehicles are captured. The system employs an over-current-driven physical damage mechanism and an adaptive sliding-window dynamic decision algorithm that integrates statistical features with first-order difference trend analysis. Multi-stage triggering criteria, false-decision tolerance, and power-consumption optimization are combined with displacement monitoring via a linear Hall-effect sensor and encrypted command verification to ensure accurate detection of unauthorized disassembly. A TL494-based over-current module delivers a 40 V/20 A high-energy pulse, irreversibly damaging the flight controller′s core circuits. The experimental results show that the proposed method is effective in complex environments such as strong electromagnetic interference and temperature changes. Its execution time still reaches about 18.2 ms response speed and 2% false trigger rate. Although response speed is 7 to 12 ms slower than fixed-threshold and moving-average algorithms, the false-trigger rate improves by 31% and 21.5%. Compared with existing schemes, the proposed system offers low cost, high reliability, and practical protection for UAV data security.