Implementation and Analysis of an Evolutionary Optimized Non-Linear Function for Linearization of Thermo-Resistive Sensors

The linearity of thermo-resistive sensors is affected by changes in the properties of materials, lead resistances, and self-healing ability. However, the existing methods employed for the linearization of thermo-resistive sensors are ineffective, given their high error rate and prolonged computational time. As a result, this paper discusses using a field-programmable gate array(FPGA) to implement an evolutionary optimized ratiometric and a log-ratiometric function for thermo-resistive sensor linearization. The optimal values of the linearizing parameters in the non-linear function were determined using the differential evolution optimization algorithm. Also, this study introduces the additional usage of a log-ratiometric one-variable function to linearize the sensor characteristics. Furthermore, the results of the non-linear function were compared with those of a neural network-based linearization method. The neural network algorithm based on Bayesian Regularization eliminates errors; however, its implementation using an FPGA is tedious owing to the increased number of neurons. The experimental results revealed that implementing the non-linear function resulted in a better full-scale error reduction capability and reduced computation time than the conventional linearization methods. In the future, owing to their high response speed and low power consumption, current-mode circuits can replace FPGA based voltage-mode circuits.