Forecasting of Heat Production in Combined Heat and Power Plants Using Generalized Additive Models

The paper presents a developed methodology of short-term forecasting for heat production in combined and power (CHP) plants using big data-driven model. An accurate prediction an hourly load the day-ahead horizon allows better planning optimization energy by cogeneration units. method training testing predictive model with use generalized additive (GAM) was presented. weather data as input variables were discussed to show impact conditions on quality predicted load. new approach focuses application moving window learning set from last several days order adaptively train influence size accuracy forecasts evaluated. Different versions model, depending GAM parameters compared. results presented obtained coming real district heating system European city. methods during different periods season performed comparing demand actual values, measuring located case study CHP plant. As result, averaged percentage error analyzed period (November–March) less than 7% obtained.