Influence of Random Error of Temperature Sensors on the Quality of Temperature Compensation of Fog Bias by the Neural Network

Abstract

The formulas are obtained for estimating the «random walk» type noise of algorithmic compensation for the gyro bias. An example of estimating the statistical significance of the factors influencing the bias when calibrating a fiber-optic gyroscope in the operating temperature range and at different rates of their change is given. It is shown that the random error of temperature sensors can play a major role in the “random walk” noise of the algorithmic compensation for the gyro bias and exceed the gyro self noise. An example of obtaining a regression dependence of algorithmic compensation for gyro bias using a neural network with a multilayer perceptron is given. The factors influencing the choice of the time constant of the differentiating low-frequency temperature filter are considered. Experimental dependences of the random error of the bias algorithmic compensation on the value of the random error of temperature sensors are presented and the necessity of using temperature sensors with a minimum random error is shown.