An apparatus for monitoring quality of electrical welding processes on bodies (10) for producing cans is described. Welding-parameter values (Fi) which have been measured on a number of bodies (10) found to be acceptable are statistically analysed in an analysing device (22). For this purpose, an average welding-parameter profile is determined. Then a reject limit-value band (Fg+/-) enveloping the average welding-parameter profile is determined from the average values and the product of the standard deviation of the welding- parameter values (Fi), assumed to be normally distributed, and at least one sensitivity factor. A permissible number of mistaken rejects per one million welding spots are selected as a criterion for the sensitivity factor so that, for example in the case of 100 welding spots on a body (10) of a medium-sized can, every ten-thousandth body on average is inadvertently rejected. For bodies (10) produced subsequent to the "learning phase", a comparator (26) establishes whether a welding-parameter profile measured for these bodies (10) lies within the reject limit-value band (Fg+/-). This quality-monitoring apparatus, working according to statistical methods, is sensitive and accurate, since a fixed value is not simply selected for the width of the reject limit-value band. IMAGE
展开▼
