IMPROVED COST-FUNCTIONS FOR ECONOMIC-DISPATCH COMPUTATIONS

Representation of generator fuel cost curves by polynomils in real-time economic dispatch is standard industry practice. Two polynomial curve fitting methods, Gram-Schmidt orthonormalization and least-squares, are evaluated here. It is shown that the Gram-Schmidt method, which does not require matrix inversion, gives more accurate fuel cost curves. Also, a constraint equation can be included in both methods to insure that incremental cost curves are monotonically increasing. Numerical examples are included to demonstrate the results.