The Research Of Fault Detection Methods Of Redundant Inertial Navigation Based On MSPCA

With the continuous advancement of technology, the development of inertial navigation technology more and more rapidly, the study of how to improve the accuracy of inertial navigation system at the same time, inertial navigation system reliability requirements are also increasing. Improve the reliability of the inertial navigation system can be configured to implement redundant gyro, when the gyroscope failure, the need for timely fault detection and isolation out. So for redundant inertial navigation system, the introduction of fault detection techniques to detect faults and locate the fault, and further improve the reliability of redundant inertial navigation system, ensure that the system works effectively for a long time.Subject to marine redundant SINS for the study, the multiscale principal component analysis into marine redundant SINS fault detection, the reference method in chemical process fault detection, and combined marine redundancy Features inertial navigation system,fault detection research. Before using the traditional method of multiscale principal component analysis for fault detection using wavelet denoising gyroscope signal processing,it will experience two wavelet decomposition, reconstruction, for the problem, de-noising and wavelet multiscale PCA wavelet analysis section combine to achieve the purpose of saving time algorithm.The multiscale principal component analysis method is applied to redundant inertial navigation system fault detection for false alarm problems, the study of the noise impact of Multiscale PCA fault detection, fault detection, and is divided into the establishment of the main results of the meta model two implications. By adding different gyro signal to noise ratio of the noise, the noise can prove the main element will affect the complexity of the model and generate false alarms phenomenon makes fault detection process, so the gyro signal denoising is necessary. Based on the characteristics of fiber optic gyro output signal in-depth analysis to real-time gyro signal processing to analyze the premise of designing real-time wavelet threshold denoising method based on fiber-optic gyro random noise removal, combined FOG characteristics of the output signal of the wavelet threshold selection criteria optimization and improvement. Through real-time de-noising results measured signal analysis method is verified in real-time to meet the gyro signal premise is valid for the gyroscope signal.For multiscale principal component analysis of the problem of the complex history of the model library can not complete coverage, while retaining the traditional history of the traditional library model on the basis of merit, adding gyroscopes online data collection, site modeling process, can make up the traditional model of library history All the models can not cover all issues. Based on the algorithm and wavelet denoising MSPCA-depth study on the part of wavelet de-noising wavelet decomposition process and MSPCA algorithm together,solve the problem of time-consuming due to repeated wavelet transform algorithm, can save about 12.57%-19.15% time. The improved algorithm is applied to redundant inertial navigation system fault detection, simulation results show that this method can effectively detect the fault, locate the fault, and can effectively avoid the phenomenon of false alarms.