This paper introduced a passive wireless temperature sensor utilizing RFID technology, designed to overcome limitations of traditional wired and battery-dependent sensors in industrial environments. By exploiting temperature-sensitive dielectric substrate permittivity and thermo-mechanical deformation in RFID antennas, the sensor converts thermal variations into resonant frequency shifts. A multiphysics coupling model (electromagnetic-thermal–mechanical) was established, revealing that 82.4% of frequency shifts originate from dielectric constant changes, while 17.6% stem from thermal expansion. Simulations demonstrated a linear sensitivity of 0.0707 MHz/℃ within an industrial temperature range (− 5 to 55 ℃). The study identifies adaptive mesh refinement as critical for large-temperature-range simulations and emphasizes substrate material selection criteria for industrial applications. Results provide theoretical foundations for embedded temperature-stress monitoring in extreme conditions, with future work focused on experimental validation and addressing material nonlinearity at elevated temperatures. The proposed sensor offers a robust, maintenance-free solution for IoT applications.