Condition Monitoring Method for Rotary Mechanism of Rotational Inertial Navigation System Based on Dual-Channel CNN IMU Information Fusion

With the demand of long endurance, high precision, and low cost for inertial navigation system (INS), the rotational INS (RINS) has become the focus of attention; compared to fixed components (e.g., accelerometers and gyroscopes) in strapdown INSs (SINSs), the introduction of the rotation mechanism reduces system reliability, and it is necessary to carry out the research of early fault diagnosis/condition monitoring for the RINS. In this article, we proposed a dual-channel convolutional neural network (CNN)-based inertial measurement unit (IMU) information fusion method for RINS condition monitoring and fault diagnosis, which makes full use of the RINS’s IMU signals for information fusion and achieves the RINS condition monitoring accuracy up to 99.64% compared with the traditional accelerometer signals-based mechanical fault diagnosis. It verifies that the method has the advantage of anti-interference ability and cross-domain generalization capability.