Nupjautojo Lp pasiskirstymo pradiniai ir centriniai momentai

Abstract

By changing the value p in the distribution Lp a new distribution family appears. In the single case, when the parameter value p = 2, we will have the normal distribution law. When analysing the measurement results there occurs the possibility to determine the parameter value p and appropriate distribution law Lp. Seeking to avoid large measurement errors, the obtained results are controlled by the application of tolerances. This procedure creates the situation that the distribution law for measurements and their errors is converted to the cut-off distribution. In this case, for the cut-off distribution new values appear for the initial and central moments and also for variance and correlative moment values. In this article appropriate formulas for calculating the abovementioned moments are derived. A method is proposed for parameter estimations for the distribution Lp using function of a maximum confidence. In cases, when p > 2, the attainable equation systems are nonlinear. Therefore the iterative method by the use of diagonal weight matrix is recommended for their solution.