INPUT VECTOR IMPACT ON SHORT-TERM HEAT LOAD PREDICTION OF SMALL DISTRICT HEATING SYSTEM

Short-term load prediction is very important for advanced decision making in district heating systems. The idea is to achieve quality prediction for a short period in order to reduce the consumption of heat energy production and increased coefficient of exploitation of equipment. The common thing for each way of prediction is usage of historical data for certain last period which makes possible development of many methodologies for adequate prediction and control. In this paper, application of feedforward artificial neural network for short-term load prediction for period of 1, 3 and 7 days, of one small district heating system, is presented. Three different input vectors are implemented and their impact on quality of prediction discussed. The simulation results are compared and detailed analysis is done where operation in transient regime is of special importance. Satisfied prediction average error is obtained