ML Based Design Algorithms for Battery Systems as Well as EV Charging Systems

• Published in: 2023 International Conference on Power Energy, Environment & Intelligent Control (PEEIC) pp. 128 - 131
• Main Author: Kalele, Girish
• Format: Conference Proceeding
• Language: English
• Published: IEEE 19.12.2023
• Subjects: Costs; DDPG Method; Energy management; Energy Storage; Genetic algorithms; Job shop scheduling; Lithium batteries; ML Optimization Techniques; Optimization; photovoltaic energy; Power demand
• DOI: 10.1109/PEEIC59336.2023.10451000

Clean energy, characterized by solar energy and wind energy, is utilised to change the energy structure and address concerns with energy and the environment. However, the production of wind or solar power is unreliable due to environmental effects. Eliminating the instability is a key strategy, and lithium batteries are an approach that is becoming more established. A capacity that is too great would result in waste and higher costs, while a capacity that is too little would interfere with scheduling. The amount of energy that can be stored is also influenced by power use, even if extensive industrial power demand is unpredictable and erratic. For this difficult challenge, a design that dispatches as well as has a significant loading capacity is needed. A battery installation model and a lithium battery's cycle life were taken into account as part of the creation of an energy management model using data on the electric power consumption and photovoltaic power generation of an industrial estate. The performance of these lithium batteries and grid demand placed restrictions on the energy management system's ability to achieve optimal energy storage capacity while also lowering operational expenses. With operational cost and an appropriate battery volume because the optimization targets and lithium battery performance and power grid requirements as the constraints, the DDPG method (Deep-Deterministic-Policy-Gradient) was employed. The simulation findings show that, in comparison to the current energy sources, the three approaches of management of energy reduced the cost of volume and operation of the energy storing system by 19.1%.