Optimization model for owner-based microgrids using LSTM predicted demand for rural development

Over the past several years, microgrids have been setup in remote villages in developing countries such as India, Kenya and China to boost the standards of living of the less privileged citizens, mostly by private companies. However, these systems succumb to increase in demand and maintenance issues over time. A method for scaling the capacity of solar powered microgrids is presented in this paper. The scaling is based on both the needs of the owner and those of the consumers. Data acquired from rural villages characterizes the electrical use with respect to time. Further, it employees a Long-Short Term Memory (LSTM) deep learning model that can help the owner predict future demand trends. This is followed by a model to determine the optimum increase in capacity required to meet the predicted demand. The model is based on empowering the owner to make informed decisions and the equity of energy distribution is the key motivation for this paper. The models are applied to a village in Eastern India to test its applicability. Acknowledging the highly varying nature of demand for electricity and its applications, we propose a rule-based adaptive power management strategy which can be tailored specifically in accordance to the preference of the communities. This will ensure a fair distribution of power for everyone using the system, thereby making it applicable anywhere in the world. We propose to incorporate social and demographic conditions of the user in the optimization to ensure that the profit of the owner does not outweigh the needs of the users.