Gradient Methodology for 3-Axis Accelerometer Static Calibration

Knowledge of real parameters of the sensors' transfer characteristics is very important for a correct system function. The paper presents a novel easy-to-use iterative calibration algorithm for a vector field sensor's accuracy improvement, which can be successfully applied to the estimation of the calibration constants of the 3-axis accelerometers that are commonly used in the role of inertial sensors. The theory is based on the neural network that creates an inverse function to the uncalibrated sensor's transfer function. Learning process of the neural network uses a gradient methodology applying total differential on the scalar error equation. The analyzed theoretical principles are supplemented by simulations and experimental measurements. The performed simulations and experiments confirmed that the algorithm successfully converges, which enables a precise estimation of the calibration constants. Other advantage of this methodology lies in the attitude independent sensor discrete random rotation in the 3D space during the calibration procedure without the need of any precision positioning calibration platforms.