Faculty Publications
Permanent URI for this communityhttps://idr.nitk.ac.in/handle/123456789/18736
Publications by NITK Faculty
Browse
2 results
Search Results
Item Some aspects of location identification of PD source using AE signals by an iterative method(2012) Punekar, G.S.; Jadhav, P.; Bhavani, S.T.; Nagamani, H.N.An acoustic Partial Discharge (PD) location problem modeled mathematically, gives system of sphere equations, which are non-linear. These equations are formed with known acoustic emission (AE) sensors co-ordinates, with PD locations co-ordinates as unknowns. Newton's method is implemented to locate the PD activity using the AE signals. This is an iterative method and the convergence depends on the initial guess. Different aspects such as initial guess, location of sensors (sensor co-ordinates) and tank orientation in space are studied in this paper by numerical experiments on the algorithm implemented using the experimental data published (available) in a literature. The published data considered for the study here uses 8 number of sensors (4 on the front and 4 on the back wall of the transformer tank; laboratory model). The method of locating acoustic emission partial discharge (AEPD) requires at most 4 sensors (three to identify the coordinates of the location and one for arrival time of AE signal). Hence, results of such 70 combinations (i.e. 8C4) are studied using the algorithm implemented. The numerical test runs indicate that some combinations either do not lead to convergence or yield results with high errors. At least such 10 combinations (out of 70) are identified and analyzed. © 2012 IEEE.Item Genetic algorithm in location identification of AEPD source: Some aspects(IEEE Computer Society, 2013) Punekar, G.S.; Antony, D.; Bhavanishanker, T.; Nagamani, H.N.; Kishore, N.K.Using the experimental data obtained from an Acoustic Emission Partial Discharge (AEPD) system, efforts are made to locate the source of Partial discharge (PD) with a transformer tank. The AEPD data with 8 sensors (available in the literature) is numerically experimented with a Genetic Algorithm, although minimum of 4 sensors only are necessary for identifying the location. With eight sensors, four sensors considered at a time, form 70 ( 8C4) combination of sensors. The effect, implication and usage of superfluous sensor data in identifying the location with GA is analyzed and reported. Results are compared with Newton's method. © 2013 IEEE.