|
|
 |
In this project Matrix optimized a quality control process. In the original process seven different sensors were used. The goal was to reduce the number of required sensors while maintaining the same accuracy.
|
|
Figure 1: Comparison of the original quality measure derived using the measurements of seven sensors (red curve), replacing one sensor information by the remaining ones (blue curve) and the signal prediction using only four sensors (green curve).
|
By identifying the internal nonlinear relationships between the sensors the redundant sensors could be eliminated. As can be seen from figure 1 it was not only possible to eliminate one sensor from the original setup (red curve) and replace its information by the remaining six other sensors (blue curve). Matrix was able to reduce the number of sensors required to four and still maintain sufficient accuracy (green curve).
The shape of the curve essentially remains the same compared to the original summarized quality measure. The information from the full set of sensors is shown in this figure only as a reference and was not used for the calculation.
Matrix was able to reduce costs and process time for this step by almost 50%.
|
|
