Robust least square approach for optimal development of quadratic fuel quantity function for steam power stations

Abstract

Ordinary Least Square (OLS) and Robust Least Square (RLS) consisting of Least Absolute Residual and Bi-square approaches were deployed to obtain the fuel consumption characteristic curve and the coefficients of the quadratic fuel consumption function for thermal stations in Nigeria. Results were compared based on convergence property, Root Mean Square Error, R-Square value, Adjusted R-Square value, and Width Interval of coefficients. Valve Point loading effects of Egbin and Sapele power stations were used to develop the quadratic fuel consumption characteristic curve and function. The average difference in width interval for the coefficients a, b, c, d, and e of the two stations, after comparing Bi-square and OLS technique, were 0.02084, 8.5055, 1856.565, 520.8855, and 0.0082, respectively. The R-square values obtained from the Bi-Square technique were superior to the OLS technique with arithmetic differences of 0.6196 and 0.5254 for Egbin and Sapele generating stations, respectively. Bi-Square technique also offered better adjusted R-Square value for Egbin and Sapele stations with arithmetic differences of 0.622 and 0.5287, respectively. Bi-Square technique offered smaller Root Mean Square Error than the Ordinary Least Square technique for both stations. The coefficients obtained from Bi-Square technique were used to develop the fuel quantity function for the studied stations.