摘要 |
A method for monitoring the quality of laser-machining processes (500), in particular laser-cutting or laser-welding processes, including the operations of: acquiring via sensor means (540), in particular optical sensors, a machining-process signal (Xp) and calculating from said signal representing the process (Xp) parameters (E1, E2, DC) that represent the machining quality during the laser-machining process; and making available (130) corresponding reference parameters (E1r, E2r, DCr) representing a given machining quality, which are calculated from a process reference signal (Xr) acquired via said sensor means (540). According to the invention said method further includes the operations of: comparing (230) said parameters (E1, E2, DC) representing the machining quality with said reference parameters (E1r, E2r, DCr) representing a given machining quality to obtain a set of regulated process parameters (N1, N2, NDC); making available (140) a plurality of statistical models (thetak) that identify respective states of the machining process (500), said states including states of defect (D); and executing in real time an operation of recognition (300) that a set of process observations (x; N1, N2, NDC, DIR) acquired in real time which includes said set of regulated process parameters (N1, N2, NDC) belongs to one or more of said plurality of statistical models (thetak) that identify states of the machining process (500), said recognition operation (300) including identification (330) of one or more causes of defect (D) as a function of the states recognized in said recognition operation.
|