A Markovian Statistical Analysis of Single-Event Upsets on Closed-Loop Flight Control Systems

The operation of aircrafts depend heavily on closed-loop digital flight control systems which can be disrupted or upset by errant neutrons at aircraft flight altitudes. These upsets, called single-event upsets, range from data corruption to short circuits which can potentially cause physical damage to digital devices and unexpected behaviors in flight controls. Experiments conducted at the Los Alamos Neutron Science Center consisted of irradiating an experimental flight control computer (FCC) architecture with a high intensity neutron beam while the FCC controlled a simulation of a Boeing 737. It was postulated that the upset statistics on the FCC could be fitted with a Markov model. The main objective of this research was to analyze data from the single-event upset experiments and to determine if the rate of upsets could be modeled as a homogeneous Markov chain of a specific order. The analysis was based on well known X2 hypothesis test for determining independence, stationarity, and the order of a finite Markov chain of a sequence of data. The data analyzed is a sequence of states that indicates when the FCC was under an upset condition.