An application of mean field theory to optimal power flow

This paper aims at proposing a new method based on mean field theory to cope with the mixed nonlinear integer programming, especially with optimal power flow problems. That is, the first step is to establish the energy function as well as the related partition function, and then to take advantage of the characteristics of the original problem to integrate out the discrete variables. The second step is to derive mean field equations by carrying out saddle approximation. Numerical simulations have verified effectiveness of this approach for a small power system.