An ordinal optimization based method for power distribution system control

This work presents an Ordinal Optimization Theory based method for power distribution system control. The solution procedure bases on an evolutionary search scheme. The method aims at minimizing the total power losses on network feeders while keeping the node voltages at near rated values against changing loading conditions. By relaxing the definition of optimality and softening the goal of optimization, ordinal optimization makes the problem easier as well as the solution process faster. A solution can be defined as good enough in performance value if it is one of the best m-percent solutions on the search space with probability P%. A solution can be defined as good enough also if its performance value approximates to the optimal value to a pre-defined extent. This work estimates the performance value, the total power loss, of the optimum state of power distribution network at a given loading condition. Then, with the two relaxed definitions for optimality, it finds the optimal settings of on-load tap changers and capacitors in a 24-h time frame. Results verify our method is proper for on-line and off-line operation.