On identification of dimensional function of many variables

A method for identification of dimensional functions of many variables is described. It is based on the combined use of dimensional analysis and statistical methods. The identified function is transformed into an equivalent dimensionless function which has a smaller number of variables, and, by using the preliminary measurements, the dimensionless function is decomposed into a product of functions of single variables. These are identified in succession and the final form of the dimension function is then constructed.<
>