Extremal Convolutional-Matrix Method for Analysis of Measuring Results

H. H. Asadov, A. Dj. Aliyeva · Вісник Національного технічного університету України Київський політехнічний інститут. Серія: Радіотехніка. Радіоапаратобудування · 2021

Matrix methods of measurements intruded into theory and practice of control and diagnostics of different multi-measure apparatus and installations and methodic of their functioning. This an intrusion is firstly linked with multi-measure structure of researched object itself. The new matrix method for processing of measuring results is suggested which make it possible to determine value of measuring signals in dynamics. In order to determine the value of measuring signal the principle of occurrence of extremum in convolutions value is utilized. Extremal property of this convolution is based on principle of multi-criteria optimization. The suggested matrix method envisages serial realization of these operations as (a) forming of extremal scalar convolution composed of two anti-phase partial criteria multiplied by normed weight coefficients. The anti-phase partial criteria being the functions of measuring optical parameter of media should has the have anti-phase dynamics of change, i.e. increase of one should accompanied by decrease of another one; (b) development of analytical formula to calculate the value of measuring signal depending of values of other parameters at moment of occurrence of extreme value of convolution; (c) carrying out analytical measuring operations on detection of extreme value of calculated convolution of partial criteria upon given weight coefficients; (d) regulation of weight coefficients to provide the maximum authentic registration of extreme value of convolution; (d) to compose the matrix of values of measured researched parameter depending of weight coefficients of partial criteria. Results of model researches on approbation of suggested method for processing measuring results are given which confirm operability of suggested method.

Read the paper · More papers on PaperTik