Modelling of demand response for utility's loac forecasting

In the new era of electrical power industry with more emphasis on green energy resources and active customer participation, the distribution utilities (DISCOMs) are being challenged. Being an important link between wholesale and retail electricity markets, these DISCOMs are exposed to risks on both sides. Under such circumstances, they are looking for new analytics to optimize operations and maximize profits. Load forecasting is one such predictive analytics used by DISCOMs to minimize risks. With increasing level of penetration of intermittent wind and solar energy in the generation at bulk as well as distributed level, both supply and demand have become uncertain. With certain benefits and incentives offered to the customers, demand can be made flexible and controllable in nature. Such flexible demand can help in minimizing the demand supply gap. Also the response of load is faster than the conventional generation resources which have machine inertia. Flexible demand is thus one of vital feature of future grid. Demand Response (DR) can be considered as a way to utilize this flexibility of load by adjusting the consumption profile thus assisting in dealing with the increased uncertainty and improving the power system operational efficiency. With the DR programs implemented, the DISCOMs will now have to forecast not only the demand of electricity but the net demand adjusted after accommodating demand response. In this paper, we analyse the impact of DR on the net demand profile of DISCOM considering the end consumers with varying demand profiles and varying preferences. We propose a new approach to forecast demand with modelling of the DR. The proposed approach gives a Probability Density Forecast (PDF) of demand using a non-parametric approach based on Kernel Density Estimation (KDE). The proposed model will help DISCOMs for developing demand bidding strategies in a market where DR programs are being implemented.