The Quality Prediction of Fiber-Optic Gyroscope Based on the Grey Theory and BP Neural Network

Abstract

Fiber-optic gyroscope has become a new generation of leading device of inertial measurement system. Fiber-optic gyroscope is small batch product with high precision and high cost. Its development process has the characteristics of heavy testing work, complex data analysis, poor stability so that its product quality is hard to control. Therefore, we use the grey theory combined with BP neural network to build the model for the quality prediction of fiber-optic gyroscope. As the fiber-optic gyroscope is small batch product and its quality is influenced by multiple factors, we add "The Sample Selector" and optimize the results in the traditional grey prediction method. "The Sample Selector" is used to process the relevant historical data of similar fiber-optic gyroscope. The grey model is used to predict the quality of the fiber-optic gyroscope. Furthermore, we use BP neural network to correct residual error to improve the accuracy of prediction. We verified that the method has a high accuracy rate and meets the production demands. According to the prediction, it can take timely measures to control the product quality in advance.