Failure management in spatio-temporal redundant, integrated navigation and flight control reference-systems

Abstract

Failure management techniques for highly reliable, fault-tolerant inertial reference systems are described. Cost, weight, and power considerations imply the use of a minimum number of inertial sensors in a skewed geometry. Fault-tolerant hardware performance is obtained by spatially separated channels with a preceptron-type information flow. Data diversity in temporally separated software channels yields software fault tolerance. Advanced vector space procedures for fault detection, localization, masking, and dynamic system reconfiguration permit safe and quick response, yielding minimal data and recovery latency.<>