Computer aided detection of sensitive parameters in additive manufacturing
In additive manufacturing a variety of errors can occur that lead to defective products. This is because there are a multitude of settings that can be changed to affect the printed product. With a lot of experience and trial and error these settings can be adjusted almost optimally. Consequently, it is expensive and cost a lot of time to adjust these settings.
In this research, these settings are handled as parameters thus a sensitivity analysis can be carried out. Sensitivity analyses are commonly used to determine whether input parameters are sensitive parameters and have an influence on a target value or not. In additive manufacturing, it is necessary to determine the influences of a multitude of input parameters. This research work deals with the spectral sensitivity analyses Fourier amplitude sensitivity test (FAST) and different variants.
The outcome for the spectral sensitivity analyses is calculated using the Fourier transformation. For example, in FAST values of input parameters are oscillating in different frequencies over the experiments. Thus frequencies in the output of the experiments can be traced back to main effects or interactions between input parameters. The variants of the FAST method are used to split the parameters in groups and analyse them separately but over the same experiments. This way less experiments are needed and all parameters of the whole system can be analysed at the same time.