Planejamento AC da expansão da transmissão: uma abordagem simplificada usando um algoritmo genético e um fluxo de carga convencional

Abstract

Conventionally, the Transmission Expansion Planning problem has been addressed using classical mathematical optimization models, with the AC model being the most complete and complex for this purpose. Due to its complexity, such models typically require specialized optimization software, such as AMPL, GAMS, and Pyomo, for their resolution. However, a limitation of these classical methods is that they yield a single optimal solution (either global or local), which can be a significant disadvantage when seeking a set of viable alternatives for the problem. The superiority of the proposed methodology lies in how the problem is modeled. Unlike previous studies that employ optimal power flow and impose a greater number of constraints4thus limiting the search space4the approach adopted in this study allows for a broader exploration of economically viable solutions. This flexibility enables the identification of multiple planning alternatives, making the methodology a valuable decision-making tool. However, despite its demonstrated efficiency, using conventional power flow simplifies the model, making it more basic compared to literature approaches that apply OPF. Thus, the presented methodology stands out as a robust and flexible alternative for Transmission Expansion Planning, offering more cost-effective solutions and a greater diversity of options. Nevertheless, its potential could be enhanced by incorporating OPF, as seen in more sophisticated models in the literature, allowing for further refinement in the search for optimized solutions. The choice of conventional power flow was motivated by the need to explore a broader set of feasible alternatives without the constraints imposed by an excessive number of restrictions, making the model more accessible and flexible for analysis. In this context, this study proposes an innovative approach that combines Genetic Algorithms with conventional power flow to solve the AC Transmission Expansion Planning problem. This methodology enables the identification of multiple solutions, expanding the range of choices and adaptability to energy planning needs. To validate the proposed approach, the Garver AC system4widely used in Transmission Expansion Planning studies4was employed. Final tests confirmed the efficiency of the GA-based model. Parameter calibration revealed that a 95% recombination rate balances diversity and the preservation of good solutions, while a 5% mutation rate prevents premature convergence. Furthermore, a population of 100 individuals yielded the best performance while maintaining an acceptable execution time. With these optimized parameters, the results obtained surpassed those found in the literature, with 70% of the executions producing solutions superior to the adopted reference. The best solution identified had a cost of 130 monetary units, significantly lower than the reference value of 160.