The inclinometer system based on fiber Bragg grating assisted by neural network

In this paper, a fiber grating inclinometer system based on neural network is proposed to improve the accuracy and efficiency of slope displacement monitoring. The system combines a fully connected neural network (FCNN) model to process complex spectral data and evaluate the degree of tilt deformation. Firstly, the traditional differential equation method (DEM) strain displacement algorithm struggles to compensate for complex physical nonlinearities, resulting in insufficient measurement accuracy. In order to solve this problem, this paper introduces the FCNN, which can effectively capture the complex relationship between the fiber grating signal and displacement deformation by using its advantages in high-dimensional data processing and automatic feature extraction. On this basis, the FCNN model based on a multi-source sensor data fusion multi-layer perceptron (MLP) architecture is constructed. By analyzing the spectral data detected by the fiber Bragg grating and the field temperature data, feature extraction and model training are carried out to predict the displacement deformation of the inclinometer tube with measurement errors controlled below 2%. Furthermore, through comprehensive error analysis and validation assessment systems, errors in core areas are consistently controlled within 0.01mm. Statistical analyses including cross-validation and multiple independent training runs ultimately confirm the experimental system’s significant advantages and excellent stability. This method ultimately enhances the accuracy of inclination deformation measurements, providing a robust and highly practical solution for slope monitoring and early warning.