The Peak-Shaver provides the tools to implement artificial intelligence (LSTM's and Reinorcement Learning) in order to level the power consumption of a production system. The possibility of using a energy storages was explored, to shift power peaks into periods with lower power requirements. The main focus of the project was the investigation to which extent machine learning can be used for this problem: An LSTM was used to make predictions about future power requirements and based on this (different) reinforcement learning agents decided when to charge and discharge. Finally the resluts were compared with heuristic approaches to check if such a solution is a suitable for this type of problem. For more information about the project and a general guide for the code, check out the documentation.
maik97/peak-shaver
Shaving power peaks with artificial intelligence (lstm predictions and reinforcment learning)
Python