This paper presents the development of an intelligent temperature transducer to measure temperature in the range of 0 °C-100 °C using a negative temperature coefficient (NTC) thermistor. The NTC thermistor is connected in a timer circuit to convert the temperature change into frequency. The timer circuit acts as a signal conditioning circuit (SCC) for the NTC thermistor and exhibits a stable temperature-frequency characteristic with a reasonable error. The Levenberg-Marquardt training algorithm is used in a multilayer perceptron neural network to further reduce the nonlinearity error of the SCC. The trained artificial neural network (ANN) improved the linearity, sensitivity, and precision of the SCC to an appreciable range. A linearity of approximately ±0.8% and the sensitivity of about 5 kHz/°C are achieved. The intelligence of the trained ANN is embedded in a microcontroller unit, and the performance of the developed transducer is experimentally studied on a prototype board.

This paper presents the development of an artificial neural network (ANN)-based linearization technique for the voltage controlled oscillator (VCO) thermistor circuit. VCO thermistor circuit exhibits a linear analog temperature-frequency relation over a temperature range of 0 °C-100 °C with good response and reasonable error. An ANN-based direct modeling technique is used for a nonlinearity estimation of VCO thermistor circuit to further improve the linearity. The performance of the developed technique has been experimentally verified. A linearity of approximately ±0.2% and sensitivity of 5 kHz/°C are achieved over a temperature range of 0 °C-100 °C, with a high thermal stability. A notable feature of the developed technique is linearity and sensitivity of the temperature measurement, which are quite high. The efficacy of the technique is established through simulation studies and its practicality is demonstrated on a prototype unit.

Thermistor is most widely used sensor in the temperature measurement due to its high sensitivity and fast response. The non‐linearity of the thermistor gives rise to several difficulties for on‐chip interface, direct digital readout, wireless transmission and so on. Hence, an effective lineariser is needed to overcome the difficulties. In this study, an artificial neural network‐based lineariser has been developed for the thermistor connected in operational amplifier circuit. Operational amplifier‐based thermistor signal conditioning circuit exhibits a stable temperature–voltage relation over a range of 0–100°C with low linearity. A multilayer perceptron feed‐forward neural network is used for non‐linearity compensation of thermistor circuit to further improve the linearity. A linearity of ±0.3% is achieved over 0–100°C with high temperature stability. A notable feature of the proposed method is the non‐linearity error remains low over the entire dynamic range of the thermistor. The efficacy of the method is established through simulation studies and its practicality demonstrated with experimental results obtained on a prototype unit built and tested.