A mixed integer quadratic programming for dynamic economic dispatch with valve point effect

In this paper a mixed integer quadratic programming (MIQP) is proposed to solve the dynamic economic dispatch (DED) with valve-point effect (VPE) where the non-linear and non-smooth cost caused by VPE is piecewise linearized. However if the DED with VPE is directly solved by the MIQP in a single step, the optimization suffers convergence stagnancy and will run out of memory. In this paper the multi-step method, the warm start technique and the range restriction scheme are combined with the MIQP. The optimization process can then break the convergence stagnancy and the computation efficiency can be greatly improved. When the system loss is considered, the loss formula is piecewise linearized. A post-processing procedure is proposed to eliminate the approximation error caused by linearization of the loss formula. The effectiveness of the proposed method is demonstrated by seven cases and the results are compared with those obtained by the previous published methods.