Optimization of a photovoltaic/wind/battery energy-based microgrid in distribution network using machine learning and...

Area of Science:

• Electrical Engineering
• Renewable Energy Systems
• Optimization Algorithms

Background:

• Hybrid microgrids (HMGs) integrating photovoltaic (PV), wind energy, and battery energy storage (BES) are crucial for modern distribution networks.
• Optimizing HMGs involves balancing energy loss, voltage stability, and purchased power costs, especially when using forecasted data.

Purpose of the Study:

• To develop and apply a fuzzy multi-objective framework for optimizing HMGs (PV/WT/BES) in a 33-bus distribution network.
• To minimize energy losses, voltage oscillations, and purchased power costs using forecasted data.
• To evaluate the impact of battery depth of discharge on HMG performance.

Main Methods:

• A multi-objective improved Kepler optimization algorithm (MOIKOA) was developed, incorporating Kepler's laws, a chaotic map, and FDMT.
• A multilayer perceptron artificial neural network (MLP-ANN) was used for forecasting solar radiation, wind speed, temperature, and load.
• Three optimization scenarios were implemented: real data, forecasted data, and varying battery depth of discharge.

Main Results:

• MOIKOA demonstrated superior performance compared to conventional Kepler, PSO, and GA algorithms.
• Optimizing with forecasted data (MLP-ANN) resulted in a 3.50% increase in energy losses, 2.33% in voltage deviation, and 1.98% in purchased power cost versus real data.
• Increased battery depth of discharge enhanced BES participation, leading to greater reductions in network energy losses and voltage deviation.

Conclusions:

• The MOIKOA framework effectively optimizes HMGs for distribution networks, balancing multiple objectives.
• While forecasted data introduces minor cost increases, its integration with advanced algorithms like MLP-ANN is viable.
• Battery energy storage plays a vital role in mitigating HMG operational impacts and improving network stability.