A novel strategy for simultaneous active/reactive power design and management using artificial intelligence techniques

Abstract

Adequate capacity and strategy of an energy management system improve technical and economic characteristics, promote flexibility, and encourage higher penetration of distributed generation. This paper presents a new strategy for simultaneously carry out the optimal equipment design and the management of active and reactive power in smart electrical installations. The proposed optimization algorithm defines the capacity of the photovoltaic generation and the battery banks, the capacitor banks and substation configuration, in addition to selecting the optimal parameters of the battery energy and capacitor bank management strategy making the methodology adaptable to any consumption profile and energy billing format. The purpose of the algorithm is to reduce the global cost of energy over a 20-year time horizon using net present value as the metric, considering energy bill, investment, maintenance, and replacement costs. The energy tariff is composed of time-of-use scheme, demand cost, fines associated with low grid power factor, and feed-in remuneration. The approach also considers transformer losses and 3 typologies defined by the k-means technique used to represent the power consumption profile and meteorological data to reduce the computational effort. The Genetic Algorithm has been applied in a real case study whose project’s net present value was reduced by 28.15%, with an internal rate of return of 21.20% and a computation time of 15 s. The results show high algorithm performance with a reduction of 78.16% in energy and 16.43% in demand costs, virtually eliminating costs caused by low power factor and reducing the energy injected into the grid.