We investigated the thermomechanical coupling and stress–strain distribution of ultrahigh‑strength 22MnB5 steel during hot stamping using a non‑isothermal simulation method that accounts for dynamic heat generation from plastic work (Taylor–Quinney effect) and friction. Simulation results of the Livermore Software Technology Corporation–Dynamic Nonlinear Analysis revealed that increasing the stamping speed from 30 to 200 mm/s elevated the minimum sheet temperature from 389.8 to 621.3 °C, reducing the maximum stress from 383 to 152 MPa through thermal softening. At a constant speed of 100 mm/s, raising the initial forming temperature from 700 to 900 °C lowered stress from 262 to 155 MPa but enhanced thinning to 1.527 mm, approaching the 25% limit. Quenching analysis results showed an average cooling rate of 60 °C/s, exceeding the critical 25 °C/s, with the final temperature of 262 °C—below the martensite completion point of 280 °C—ensuring full transformation. Beyond materials processing, the developed method predicts internal stress and temperature gradients that physical sensors cannot capture in closed dies. The results provide calibration data for infrared thermography, acoustic emission sensors, and embedded strain gauges, enabling real‑time monitoring and adaptive control in industrial hot stamping. This soft‑sensing concept of the developed method complements hard‑sensor metrology, advancing intelligent manufacturing. Challenges, including sensor durability under rapid thermal cycles and the integration of virtual and physical sensor feedback loops, can be addressed using the simulation method developed in this study for predictive quality assurance.