Electrical Systems, Energy and bioenergy, Metaheuristics, Home Energy Management System, GridLAB-D, Energy Conservation.

摘要

The performance of centrifugal pumps depends mainly onrnthe intake pressure and the rotational speed. In the case ofrnelectric submersible pumps working with both gas and liquid,rnthe performance also depends on the gas amount thatrncan cause operation instabilities and a pronounced drop inrnthe pump pressure increment for lower flows than the surgingrnpoint (point of maximum pressure in pressure incrementrnversus liquid flow rate curve). So, it is important tornfind the condition that surging occurs and keep the pumprnoperating in a considered safe region. In this paper, a controlrntechnique to keep the pump operating in the desiredrnpoint is proposed considering different intake conditions.rnFor this, two artificial neural networks are used: the first tornmodel the ESP behavior and another to control the pumprnoperation. To control, the Direct Inverse ControlMethod isrnused considering the rotational speed as the control input.rnData used to train and verify the artificial neural networksrnbehavior were obtained from experimental tests of a pumprnoperating with water-air flow.