Ant colony system-based applications to electrical distribution system optimization

Electrical distribution networks are structurally weakly meshed, but are typically operated in radial configurations to simplify the network protection schemes. This implies the need to carry out suitable selection of the redundant network branches to open in normal conditions, as well as to define the control variables to set up in order to guarantee effective system operation and voltage quality. Furthermore, the structure of the distribution networks has to be upgraded to meet future needs, according to economic objective functions defined for distribution system planning. Finally, distribution systems need to face possible interruptions through the formulation of appropriate service restoration strategies, based on switching on and off a number of branches, reducing the effects of faults occurring in the networks by defining a time-dependent path to restore electricity supply to all consumers. All these aspects can be addressed by formulating and solving suitable optimization problems for distribution network operation and planning. Typically these problems are defined within discrete domains for the decision variables. However, for large real distribution systems the highly-dimensional and/or combinatorial nature of the optimization problems make it impracticable to resort to exhaustive search, since the number of combinations would be excessively high to be processed in reasonable computation times. Global optimizations can then be solved in terms of adopting suitable meta-heuristics. In this context, ant colony optimization (ACO) provides viable solutions to the various distribution system optimization problems. Key features of ACO, such as parallel search, shortest path finding, adaptability to changes in the search space, long-term memory and information sharing, have been fully exploited in their classical formulations, in advanced versions such as the hyper-cube ACO framework, as well as in hybrid formulations combining the ACO properties with those of other meta-heuristics. This chapter summarizes the formulation and use of ACO algorithms and variants to solve different types of electrical distribution system optimization problems, namely: • Distribution system optimal reconfiguration and voltage support. Various objective functions to minimize can be set up in the presence of a set of structural and operational constraints. The objectives include total distribution system losses, voltage deviations with respect to their reference values, maximum load on system components, load balancing, and capacitor placement. Performing network configuration changes and setting up the level of insertion of compensating devices or distributed generation and