Real-Time and Time-of-Use Demand Response Model : A Reinforcement Learning Approach

This work puts forward a demand response model based on both prices and incentives, which strengthens the accuracy of the demand response approach influencing the reduction or increase of the customer demand. This possibility arises from offering customers real-time and time-of-use pricing if they actively increase or decrease their demand. In this way, the positive influence of including characteristic parameters such as the internal or external coincidence factor within the classification of customers based on the k-means method is shown in this work. In addition, the reinforcement learning method obtains prices and incentives that maximize benefits for both customers and energy distribution companies. Finally, a sensitivity analysis on the elasticity of customers shows how clustering and reinforcement learning algorithms are dynamically coupled to the customer behavior