The Design and Implementation of an ANN-based Non-linearity Compensator of LVDT Sensor

TL;DR

This paper presents a neural network-based model to correct non-linearities in LVDT sensors, validated through simulation and FPGA hardware implementation, enhancing sensor accuracy and reliability.

Contribution

It introduces a FLANN-based non-linearity compensator for LVDT sensors, validated via simulation and FPGA implementation, improving linearity and sensor performance.

Findings

Nearly 100% linearity range achieved in simulation

FPGA implementation results align with simulation data

Feasibility demonstrated for manufacturing integration

Abstract

Linear variable differential transformer (LVDT) sensors are used in engineering applications due to their fine-grained measurements. However, these sensors exhibit non-linear input-output characteristics, which decrease the reliability of the sensing system. The contribution of this article is three-fold. First, it provides an experimental study of the non-linearity problem of the LVDT. Second, it proposes the design of a functional link artificial neural network (FLANN) based non-linearity compensator model for overcoming it. Finally, it validates the feasibility of the solution in simulation, and presents a proof-of-concept hardware implementation on a SPARTAN-II (PQ208)FPGA using VHDL in Xilinx. The model has been mathematically derived, and its simulation study has been presented that achieves nearly 100% linearity range. The result obtained from the FPGA implementation is in good agreement with the simulation result, which establishes its actualization as part of a general manufacturing process for linearity compensated LVDT sensors.