Fault detection and isolation using parity space approach in leveling networks

It is well known that faults have a significant effect on parameter estimation in geodesy as well as many other engineering disciplines. In geodesy Tests for outlier and robust methods are used to detect outliers. The parity space approach is a model-based fault detection and isolation (FDI) technique. In this study; the applicability of the parity space approach to the linear regression and leveling network is tackled. For this aim; the linear regression model and leveling network have been simulated. The mean success rates (MSR) of approaches are obtained by using Monte Carlo simulation technique. Singular value and QR decomposition of the coefficient matrix are used among the matrix decomposition methods as an alternative way to the Potter algorithm used in standard parity space and optimal parity vector approaches.