A data-driven technique for investment strategy of distribution systems upgrade

This paper presents an investment funds allocation model for distribution system upgrade based on data-driven technique. First, a bilevel programming evaluation system is established, which takes upgrade measures as element layer indices and comprehensive evaluation parameters as performance layer indices. In order to mine the correlation between these indices, enormous indices data generated by a virtual distribution system program is analyzed through artificial neutral network (ANN). Second, the trained ANN is embedded in the distribution network upgrade optimization model, which is solved by genetic algorithm (GA). Finally, effectiveness and feasibility of model are verified by a city distribution system.