Correlative Study On Fault Diagnosis Of GPS/INS Tightly Integrated System

Accompanied by the growth of science, technology and control theory, especially their development in aerospace, marine and other utilities area, integrated navigation system has been a mainstream trend. And navigation system is becoming more complex and larger.So as to highly improve precision and reliability of the navigation system, fault diagnosis technology has become a necessary means. Among methods of fault diagnosis, the introduction and application of artificial neural network research becomes a central issue. It indicates the new direction to fault diagnosis of integrated navigation system (INS), meets the need of application requirements in various fields and provides effective protection support to the modern military.This article takes GPS satellite navigation system as an example. According to different combinations of integrated navigation system, three combination types can be summarized.They are loose, tight and deep portfolio combination. Since the research on loose combination has been relatively mature, this article takes tight combination in integrated navigation system as the research background and performs research on fault detection and diagnosis techniques.At present,there are a lot of fault diagnosis methods. Among them,artificial intelligence method has become a growing trend. Neural network technology with intelligent and biomimetic control functions has been researched for fifty years since the beginning of its creation, and its path of development is extremely difficult. The method is characterized by adaptability, self-learning ability, linearization capability and fault tolerance, which has been widely used in multiple areas such as fault detection and information fusion.Firstly,background and meaning of the article and fault diagnosis and fault-tolerant technology are introduced,including its development status,main types,task and main methods. We also introduce the method to integrated navigation system. Secondly, this article presents the growth and principle of integrated navigation system. Meanwhile, it studies reliability of this method, and analyzes the structure of the error model for each navigation sensor and fault structure in integrated navigation system. We design a method to detect and diagnose fault based on each common sensor fault model in integrated navigation system.This article is divided into five fault models. Complementary effect can be achieved and reliability and precision can be highly improved by combining neural network method and traditional method of fault detection. We apply fault detection algorithms to analyze and explore error in integrated navigation system. Then neural network method is applied to the system. During the research on designing neural network, we find a new method to combine the integration algorithm and neural network and form a new integrated neural network approach which is verified in simulation. Finally, the new method is introduced to fault detection in integrated navigation system. Due to this new method, we design principle and structure of fault detection which is proven feasible and effective in simulation.