The paper presents a neural model, based on a GMDH neural network, used for modelling the power generated by a wind turbine. The research results of this model are presented compared to the research results of two stochastic models (WPPT and GWPPT). The presented statistical evaluation of three models shows that the most preferred model for the power generated by a wind turbine is a GMDH neural model.

Construction of functions is considered as an important part of measurement, which is essential for its quality. Since the traditional least-squares method is only applicable within the strict regression model of data, other methods are required in practice. In the paper, some proper extensions of regression models are considered, which enable consistent estimates for functions. The confluent estimates based on these models are studied; the accuracy characteristics are primarily considered.

Many measurements cannot be performed without computer-aided data processing. Such indirect measurements require dealing with uncertainty propagation through metrological software. The combination of automatic differentiation and uncertainty characteristics representation as fuzzy intervals can be the basis of fully automatic metrological estimation of final results of data processing. This paper is devoted to clarifications of this approach.

The Bayesian analysis of a series of correlated indications of an unknown quantity is here presented when they are modelled by a joint Gaussian distribution and their covariance is assumed to be the (known) squared uncertainty associated with a systematic effect common to all the indications. An interesting application of the obtained results to the conformity assessment of a series production is also presented. A criterion is derived so that at least a portion p₁ of the series production shows to have a characteristic value below a prescribed limit, with a probability not less than p₂.

The paper is a digest of some problems to be considered should the proposed use of fundamental constants in the definition of measurement units of the SI be implemented: (a) more base units being multi-dimensional, instead of fixing the present problems in this respect; (b) the multidimensionality in the definitions; (c) how the magnitude of the base units is established; (d) the use of CODATA adjusted values of the constants for this specific purpose; (e) formal issues in stipulating algebraic expressions of the definitions, and in respect to the rounding or truncation of the numerical values; (f) formal issues with the use of the integer number N_A; (g) limitations in new determinations that might arise from the stipulation of the values of several constants, for the CODATA to continue performing in future meaningful least squares adjustments of the fundamental constants taking into account future data; (h) implementation at the NMIs and Society level.