Resumo:
This dissertation of Master's degree has as main objective to develop an
intelligent method of load distribution among machines (generating groups) of
hydroelectric power plants in way to maximize the global efficiency of the
transformation of the potential energy stored in the reservoirs of water in electric
power. The optimization of the hydroelectric plants operation does with that an
economy of fuel exists, in this case the stored water, for the rational use of the
available energy, with great interest nowadays. The efficiency optimization can be
accomplished through direct measures of efficiency, flow in the turbines entrance or
penstock absolute pressure. In this work, the developed optimization algorithm uses
direct measures of efficiency and power generated by each machine. The search is
guided by the “Steepest Ascent Hill Climbing” heuristic and it presents the
advantage of being executed in real time, where there is not the need of the curves of
machines efficiency, could also be used in conventional optimizations (linear or
nonlinear), which are accomplished through models of the system. The optimization
is accomplished in a hypothetical hydroelectric plant, with ten machines, for a great
variety of demanded power, through simulations accomplished by the developed
algorithm. The efficiency values informed by the search are compared with the
values found by LINGO software. The algorithm was developed in MATLAB, being
presented in the end of the work for using, changes and/or improvements.