Micro-electromechanical systems gyro performance improvement through bias correction over temperature using an adaptive neural network-trained fuzzy inference system

This article presents a new method to obtain a miniaturized intelligent gyro sensor. The proposed method uses a fuzzy logic controller to realize an online correction of the error due to the gyro sensor’s bias variation with temperature. In the first phase, gyro experimental testing is performed for various speeds ranging from −150 to 150°/s and temperatures between −10 °C and 70 °C, and the results obtained with classical compensation methods are evaluated. The fuzzy logic controller is developed via a fuzzy inference system (FIS), generated from the gyro testing experimental results. Further, to optimize the membership function parameters of its input–output variables, the FIS is trained with a neuro-fuzzy network. Validation by the assessment of errors shows that the proposed method gives better results than the classical algorithms based on the least squares method.