Control sensor linearization using a microcontroller-based neural network

Control sensor and actuator linearization is a promising application area for artificial neural networks (ANNs). These applications require a relatively small number of neurons which can be implemented in microcontroller software or "fixed" analog hardware. We show how a microcontroller based ANN can be used to extend the linear region of operation of a nonlinear sensor. The results were applied to the design of a discrete component phase locked loop (PLL). Experimental results showed that the neural approach doubled the sensor's linear range and thus the PLL tracking range.