The paper presents an adapted least squares identification method for reduced-order parametric models. On the example of the open velocity loop, different model approaches were implemented in a motion control system. Furthermore, it is demonstrated how the accuracy of the method can be improved. Finally, experimental results are shown.

In this study, a procedure for optimal selection of measurement points using the D-optimality criterion to find the best calibration curves of measurement sensors is proposed. The coefficients of calibration curve are evaluated by applying the classical Least Squares Method (LSM). As an example, the problem of optimal selection for standard pressure setters when calibrating a differential pressure sensor is solved. The values obtained from the D-optimum measurement points for calibration of the differential pressure sensor are compared with those from actual experiments. Comparison of the calibration errors corresponding to the D-optimal, A-optimal and Equidistant calibration curves is done.

Reliable knowledge of thermo-physical properties of materials is essential for the interpretation of solidification behaviour, forming, heat treatment and joining of metallic systems. It is also a precondition for precise simulation calculations of technological processes. Numerical calculations usually require the knowledge of temperature dependencies of three basic thermo-physical properties: thermal conductivity, heat capacity and density. The objective of this work is to find a correlation that fits the thermal conductivity of selected steel grades as a function of temperature (within the range of 0–800°C) and carbon content (within the range of 0.1–0.6%). The starting point for the analysis are the experimental data on thermal conductivity taken from literature. Using the method of least squares it was possible to fit an equation which allows calculating the thermal conductivity of steel depending on the temperature and carbon content. Two kinds of equations have been analyzed: a linear one (a linear model) and a second degree polynomial (a non-linear model). The thermal conductivity obtained by linear and nonlinear models varies on average from the measured values by 3% and 2.6% respectively.