A new hybrid CBSA-GA optimization method and MRMI-LSTM forecasting algorithm for PV-ESS planning in distribution networks

Abstract

Renewable energy sources have gained significant attention as a priority option for cost reduction and emissions reduction from fossil fuels. However, their utilization introduces uncertainty and decreases reliability in distribution networks. To effectively harness these resources and mitigate their uncertainties, this paper proposes a model for determining the optimal size and location of photovoltaic (PV) systems and energy storage systems (ESS) in electrical energy distribution networks. The objective is to minimize network power losses, voltage deviations, customer interruption cost, PV system installation cost, and ESS installation cost. To achieve this, the study employs a hybrid forecasting model based on MRMI feature selection and LSTM deep neural network to forecast daily solar power and electrical load for 12 different days across various months of the year, considering the desired electrical energy distribution system. Subsequently, these predicted values serve as 12 scenarios to determine the optimal location and size of PV systems and ESSs. The optimization problem is tackled using a novel hybrid optimization method known as the Coot Bird Search Algorithm-Genetic Algorithm (CBSA-GA). The efficiency of the proposed algorithm is evaluated by comparing its results with GA, PSO, and CBSA algorithms. Among the optimization methods discussed in this paper, CBSA-GA excels in reducing the objective function by accurately determining the capacities of PV resources and ESS. Compared to scenarios where distributed generation (DG) is not integrated into the distribution network, the implementation of CBSA-GA algorithm results in significant improvements in the objective function for Cases 1 to 3. The reductions achieved in the objective function are 9.10 %, 11.88 %, and 24.39 % for Cases 1, 2, and 3, respectively. These findings highlight the effectiveness of CBSA-GA in optimizing the integration of DG resources, specifically in terms of reducing the objective function and improving system performance.