Optimization and equitable distribution of electricity in Lome by using cake cutting algorithm and artificial intelligence

Abstract

The study addresses the energy challenges faced by Lomé, where electricity demand often exceeds supply, leading to frequent load-shedding and exacerbating inequalities in access among sub-zones. It focuses on optimizing the fair distribution of electricity by combining the cake-cutting algorithm with artificial intelligence (AI) models. An analysis of existing heuristic algorithms (Random-Selector Algorithm RSA, Grouper Algorithm GA, Consumption-Sorter Algorithm CSA1, and Cost-Sorter Algorithm CSA2) revealed their limitations in ensuring equity. To address these shortcomings, a hybrid approach was developed, integrating dynamic parameters and convolutional LSTM networks coupled with wavelet transforms. Tests conducted over periods ranging from 1 to 7 days demonstrated significant performance improv ements. For instance, over 7 days, the hybrid approach (α = 0.8, β = 0.5) reduced the maximum connection gap to 20 hours, compared to 105 hours for CSA1 and 91 hours for CSA2. The proposed solution offers an efficient and equitable method for electricity d istribution, reducing inequalities while supporting the sustainable management of urban energy resources.