High-Precision Measurement of High Temperature in Dual-Junction Thin-Film Thermocouples Under Small Temperature Gradients

The thin-film thermocouples (TFTCs) are widely utilized in high-temperature measurements due to their superior capabilities. However, under extreme high-temperature conditions and complex heat transfer scenarios, the cold junction temperature can rise uncontrollably, reducing the temperature gradient between junctions, presenting challenges in achieving high-precision measurements. This study introduces a novel dual-junction TFTC design, combining three selected sensitive materials using a multilayer perceptron (MLP) approach, fabricated through a screen-printing method. The calibration of the dual-junction TFTC was accomplished using an integrated learning model, achieving a measurement range of $900~^{\circ }$ C. For cold junction temperatures from $0~^{\circ }$ C to $500~^{\circ }$ C, we predicted the hot junction temperature based on electromotive force alone. Compared to a K-type thermocouple, the average relative error in the predicted hot junction temperature across 29 experimental samples with small gradients was 1.07%. Additionally, we predicted the real-time cold junction temperature, with an average relative error of 2.91%. This work achieves accurate measurement of both junction temperatures under unstable cold junction conditions, enhancing the measurement precision of TFTCs and eliminating the need for compensation wires and water-cooling devices.