Abstract:
A gyroscope is an inertial instrument which is used to determine the angle of orientation or
the rate of rotation relative to an inertial frame of reference. This work emphasises on type
of gyroscope called a tuning fork gyroscope (TFG), which operates on the principle of coro lis effect. Due to factors such as gyroscope self-heating and, variations in ambient conditions,
the performance of a high precision strapdown inertial navigation system (SINS) gets affected
and the gyroscope bias varies. The characteristics of bias drift in TFGs and various methods
to compensate for it must therefore be studied.We suggest using a Long Short-Term Memory
(LSTM) neural network model based on Empirical Mode Decomposition (EMD) and Seasonal
Decomposition to forecast bias compensation accurately.
The purpose of this work is to employ artificial intelligence to create and evaluate bias
compensation for tuning fork gyroscopes. To determine the strength and degree of linear re lationships between sets of data, different types of correlations are compared. Denoising is
essential to improve system performance, and this can be achieved through various methods,
such as seasonal decomposition and empirical mode decomposition (EMD). The denoised data,
or trend, is fed individually into Long-Short Term Memory (LSTM) neural networks to fur ther enhance their performance. LSTM neural networks are a suitable option for predicting
time series data and can be used effectively for forecasting bias compensation using historical
data.By using a hybrid EMD-LSTM predictive algorithm, bias prediction and sensor perfor mance improvement can be achieved. The proposed seasonal decomposition-LSTM method
and EMD-LSTM hybrid forecasting model outperforms traditional LSTM neural networks
