This paper presents a new method of measuring the variation of the input electrical impedance of a welding process. The input current and voltage signals measured at the operating frequency 50 Hz are processed to obtain the input elect rical impedance Z_(in). The obtained real and imaginary part of Z_(in) represents the variations of the real electrical resistance and inductance along time in a welding process respectively. Experimental studies show that the time records of resistance differs in details from wheat has been reported in literature using "dynamic electrical resistance", and faithfully reflect the phases of a welding process under different welding conditions. The method is further automated via a pattern recognition algorithm based on neural network and is successfully exploited to distinguish in real time welds formed under different welding parameters as well as the resulted weld strength.
展开▼
