This work seeks to establish the impact of the errors in measurements of structural displacement in determination of the loads acting at the height of a cylinder pillar. By determining the impact of the error, more reliable determinations of the intensity of an external load can be made, all on the basis of data on displacements, geometry and construction material that act at some point on a pillar. The thesis is examining the case of an axisymmetric shell that forms a wind turbine pillar. For this purpose, the entire work is based on the Mathcad program package. In this work, a substantially different approach to the problem is going to be taken compared to the usual paradigm in engineering; where the loads that act in a system are known, and where output data are displacements and deformations of the said system. Here, based on the measured displacements in selected points of the structure, loads will be determined by multiplying the pseudoinverse of the measurement vector with the vector array of measured displacements of the structure. The results of the load determination will thus be in certain deviations from the actual loads; with which a shell analysis is initially carried out in order to get the data on the model displacements of the shell. Model displacements are essential here because, on their basis, the measurement vector is being formed. While simulated measured displacements with random error are made by adding a noise in model displacement values. The size of the random error in displacement will thus provide different load values that will be in different deviations from the reference, i.e. actual loads. The load in question is the effect of wind activity, which is by its very nature an asymmetric load type, that requires the shell analysis to be expanded by the use of orthogonal functions and by decoupling them into Fourier series terms. The load solutions that have small random displacement error will show the best correlations with the actual loads, as well as those that have larger amount of measurement data points, that are also more resistant to greater random error deviations.