Sensitivity analysis of control networks in terms of minimal detectable displacements

Abstract

Sensitivity analysis is one aspect of the calculation process in displacement measurement. For this purpose, sensitivity measures are used in the form of minimal detectable displacements ( ), derived from the definition of minimal detectable bias ( ). Analyses were performed based on disturbing the parameter vector with the  value (calculated from principal component analysis). Analyses were considered for displacements of the levelling network based on the global vector ( ). The calculations were conducted using the least squares adjustment with pseudo-random observations. The mean success rate ( ) was used to perform the detection analyses. The effectiveness of the global test agrees with the assumed power of the test. Local tests for a single point showed 48.6 % displacements of more than one point. It should be considered using another local test for the case of displacements of multiple points. This study concludes that in addition to the accuracy of the designed network points, the network configuration is also important in sensitivity analyses. The research shows that sensitivity analyses should be considered at the design stage of control networks, i.e. to determine at what level a given network is able to detect displacement.