This paper presents a comparison of swarm intelligence and evolutionary techniques based approaches for minimization of system losses and improvement of voltage profiles in a power network. Efficient distribution of reactive power in an electric network can be achieved by adjusting the excitation on generators, the on-load tap changer positions of transformers, and proper switching of discrete portions of inductors or capacitors. This is a mixed integer non-linear optimization problem where metaheuristics techniques have proven suitable for providing optimal solutions. Four algorithms explored in this paper include differential evolution (DE), particle swarm optimization (PSO), a hybrid combination of DE and PSO, and a mutated PSO (MPSO) algorithm. The effectiveness of these algorithms is evaluated based on their solution quality and convergence characteristic. Simulation studies on the Nigerian power system show that a PSO based solution is more effective than a DE approach in reducing real power losses while keeping the voltage profiles within acceptable limits. The results also show that MPSO allows for further reduction of the real power losses while maintaining a satisfactory voltage profile.

Applying expert systems (ESs) to power system operation means embedding them in an existing environment which consists of computational facilities (SCADA, EMS) on one side, and the operating staff on the other side. An interface between process data and ES data is required to achieve this goal. For an unambiguous and clear information exchange with and good acceptance by the operators, an approximate natural language dialogue of the ES is desirable. Several approaches to these problems are described.>