Energy demand estimation for a great scale wood board production plant

Wood and forest products are a sector that has a %3 share in the country's trade volüme and continues to grow. The most important input items of forest industry enterprises operating in the sector can be listed as the cost of wood, glue and electricenergy. Producing a solition to the problem experienced in one of the secost items will enable the sector to gain a competitive advantage in global markets. In this thesis, it is tried to produce a solution to the problem in the estimation of electrical energy consumption of the great scale chip board production facility for the next month. For the prediction of electricity energy consumption artifical neural network (ANNs), which produce successful estimation results with nonlinear model, were subjected to training and testing with seven models with different data distribution and number of neurons.The lowest error value in the test process, which is 2.48% mean absolute percent error (MAPE) value, 85% training-15% of the test data distribution has been twelve hidden neurons model.A multiple linear regression model was established including the data used to test ANN for estimating electricity consumption in 2018. The ANN estimated the MAPE value of 2.67% in the first five months, while the multiple regression estimated the MAPE value of 7.05%. Inspite of the extraordinary circumstances in 2018, ANN made an estimate of 13.2% MAPE for one year. According to Lewis, those with MAPE values below 20% are good estimation models. With reference to this definition, the model is included in the good estimation group. The study shows that ANN can also produce successful results in estimating electricity consumption in enterprises.