Fuzzy ARTMAP based fault diagnosis approach for GPS/INS navigation system

A method of fault diagnosis and parameter identification for GPS/INS tightly coupled navigation systems is presented in this paper. A two state propagator based on state chi-square test (SCST) and a Fuzzy ARTMAP neural network are used in this approach. First, the two state propagators based on state chi-square test is applied to detect the fault. The sufficient condition for determining the time interval to reset the state propagator is given. Then the fault is classified by using a Fuzzy ARTMAP neural network combined with some specially designed flight maneuvers according to results of the state chi-square test. The fault magnitude and fault occurring time are identified by using another Fuzzy ARTMAP neural network. Finally, a simulation example for GPS/INS tightly coupled navigation system is given for illustration. The results show that the method proposed in this paper can detect the fault effectively and identify its parameters accurately in case of one dimensional failure for gyroscope, accelerator or GPS signal.