A Virtual Instrumentation Approach to Neural Network-Based Thermistor Linearization on Field Programmable Gate Array

摘要: Temperature measurement is an important industrial requirement in several applications. Thermistor, particular, used to a great extent for this purpose many applications as it cost effective, relatively small size, and has better sensitivity compared its counterparts. It offers moderate range of temperature sensing typically from −55°C 125°C. On the other hand, thermistor highly nonlinear sensor characterized by exponential dependency resistance on temperature. Effective usage thus requires some mechanism linearization. This paper presents simple step-by-step, practically implementable artificial neural network (ANN)-based linearization method characteristic using two-layer having two neurons each layer. The trained feed-forward implemented field programmable gate array (FPGA) NI-PXI platform real-time measurement. Validation proposed technique was carried out experimentally comparative study. A precise thermocouple-based system utilized purpose. readings were recorded after allowing both sensors settle, maximum error ±0.9°C obtained experimental 5°C–65°C.