Modeling, estimation, fault detection and fault diagnosis of spacecraft air contaminants

The objective of this dissertation is to develop a framework for the modeling, estimation, fault detection and diagnosis of air contaminants aboard spacecraft. Safe air is a vital resource aboard spacecraft for crewed missions, and especially so in long range missions, where the luxury of returning to earth for a clean-up does not exist. This research uses modern control theory in conjunction with advanced fluid mechanics to achieve the objective of developing an implementable comprehensive monitoring systems, suitable for use on space missions.; First, a three-dimensional transport model is developed in order to model the dispersion of air contaminants. The flow field, which is an important input to the transport model, is obtained by solving the Navier Stokes equations for the cabin geometry and the appropriate boundary conditions, using a finite element method. Steady flow fields are computed for various conditions for both laminar and turbulent cases. Contamination dispersion studies are undertaken both for routine substances introduced through the inlet ducts and for emissions of toxics inside the cabin volume. The dispersion studies indicate that lumped models and even a two-dimensional model are sometimes inadequate to assure that the Spacecraft Maximum Allowable Concentrations (SMACs) are not exceeded locally.; Since the research was targeted at real-time application aboard Spacecraft, a state estimation routine is implemented using Implicit Kalman Filtering. The routine makes use of the model predictions and measurements from the sensor system in order to arrive at an optimal estimate of the state of the system for each time step. Fault detection is accomplished through the use of analytical redundancy, where error residuals from the Kalman filter are monitored in order to detect any faults in the system, and to distinguish between sensor and process faults.; Finally, a fault diagnosis system is developed, which is a combination of sensitivity analysis and an Extended Kalman Filter, which is used to estimate the location and capacity of an unknown source emission in the system. The sensitivity analysis involves pre-calculating sensitivity coefficients, which measure the response of each sensor to a source emission at each location in the cabin, and in the event of a fault, current measurements are used and inverted to arrive at an initial guess for the unknown source that is causing the fault. An Extended Implicit Kalman filter, developed especially for this application then makes use of the initial guess to arrive at an optimal estimate for the unknown source, by minimizing the squared estimation error. The fault diagnosis procedure is successfully tested for various test cases.