Employing acoustic emission sensors for detection of partial discharge, PD, introduces many advantages. Besides easy installation and replacement, they are non-invasive and immune to electromagnetic noise and interference and their sensitivity does not vary with object capacitance. For PD allocating utilizing AE sensors, distance calculations are based on the arrival time of acoustic waves to the sensors. Considering structure-borne waves of higher speed, the peaks of some of indirect path AE signals with significant contribution are mistakenly considered as peaks of direct path AE signals. Furthermore, the acoustic signals are propagating through certain parts of the transformer, such as the windings, and this complicates the partial discharge detection and allocation. These would imply an incorrect distance between the source and sensor. A method based on a heuristic algorithm has been proposed which considering all possible indirect paths with the relevant propagation times and all the barriers on the travel path of acoustic signal, calculates the more precise arrival times to sensors. A test chamber has been utilized and artificial PD signals are produced at various points. Output results of algorithm have been compared with results of classic method. It has been shown that proposed method significantly reduces the positioning errors.