Experimental data for frequency response obtained from a linear dynamic system is processed to obtain the transfer function as a ratio of two frequency-dependent polynomials. The difference between the absolute magnitudes of the actual function and the polynomial ratio is the error considered. The polynomial coefficients are evaluated as the result of minimizing the sum of the squares of the above errors at the experimental points. The magnitude and phase angle of the transfer function are evaluated at various frequencies by naeans of the computed polynomial ratio and are compared with the observed data. The numerical solution of this problem was obtained by using an IBM 704 FORTRAN program. The inethod presented here gives an analytic description of the comiplex transfer function superior to that given by minimization of the "weighted" sum of the squares of the errors in magnitude. This method is applicable to both minimum and nonminimumi phase systems. ''Setting q-1, does not restrict the problem in any way.

No abstract

A method for obtaining accurate redq allh + a n + a z uced order models for hydro-turbines with long pen-12 13(1) stocks is presented. The inadequacy of the customary mt = a21h +.a22n + a23z first order model is demonstrated. It is also shown that an appropriately chosen second order model pro, where a.. represent the appropriate partial derivatives vides the necessary accuracy while preserving model associatad with the discharge and the torque "surfaces". simplicity which is essential for ease in controllerThe partials with respect to speed, a12 and a22, are design, in assessing stability margins, and in power usually negligible. system simulation.If it is assumed that the hydraulic conduit can be

No abstract

No abstract