Skip to main content

Process optimization in sheet metal presslines by using artificial neural network technique

Objective



-When deviation outside tolerance limits occurs after the last operation in a line of presses it is today impossible to decide which of the presses that need adjustments and which parameters that must be changed. This makes it very difficult to force the process in a desired direction The operator have to do all adjustments by instinct. The result is insufficient quality, time consuming and expensive operations for correction, low productivity and unnecessary waste of natural resources because of rejected details. To find a mathematical model to describe how different parameters influence the press result is very difficult or impossible. -A method to create transfer functions without having to specify a mathematical model is to use artificial neural networks. This network can be trained on measured parameters to find a relation between parameters such as blank holding forces, greasing, tool temperature and sheet metal quality as input data and forming error or cracking as output data. -In the project we will use a neural network design tool to create a transfer function that can predict the result by training it on measured parameters from a press line.

Funding Scheme

EAW - Exploratory awards

Coordinator

AP&T Lagan AB
Address
6,Industrigatan
340 14 Lagan
Sweden