1990 Volume 8 Issue 1 Pages 84-90
A method is shown of knowledge acquisition of arc-welding by the use of fuzzy expression of experimental data, and its application is discussed to a decision support system of tuning of weld conditions (i.e. welding speed and arc-voltage) for gas metal arc-welding.
Although an experimental data itself is quantitative and non-fuzzy information, it is very common and useful for man to extend the applicability of the data by qualitative inference. In such a process, knoweldge is often explained as fuzzy information. From the point of view of knowledge engineering, it is important to accomplish such a process on computer level.
Fuzzy rules are constructed of the form of "if...then..." for tuning weld conditions from an experimental data which explains relationships between bead shape and weld conditions of CO2 arc-welding as a map. The map of the experimental data is divided into a few rectangular fuzzy regions where the information of the data is represented as a set of fuzzy rules to adjust weld conditions. The rules consist of a few input parameters in "if" part and an output one in "then" part. The input parameters of the rule are some of welding current, welding speed, arc voltage, bead shape, bead width, etc. The output parameters of the rule is adjustment value of arc voltage or weld speed. Fuzzy labels of the rule parameters are defined by simple triangular or trapezoidal membership functions.
The fuzzy reasoning system is implemented on a personal computer and simulation results are shown.
