A New Measurement Circuit of Multipoint Thermocouple with Error Correction Using the Decision Tree Algorithm

Multipoint thermoelectric converters (thermocouples) are widely used in the metallurgical, chemical, and food industries, as well as in laboratory research for studying material properties and physicochemical processes. This article proposes a new measurement circuit design for multipoint thermoelectric converters. The circuit enables temperature measurement at three zones using individual measurement junctions at each zone. The proposed multipoint thermocouple is experimentally evaluated for chromel-copel and chromel-alumel thermocouple materials. The results show an increase in the mean temperature error of the multipoint thermocouple up to 24 °C, due to variations in temperature differences along thermocouple wires. To address this issue, the article suggests a machine learning correction model based on a decision tree algorithm. The mean temperature error was less than 3 °C after correction.