In previous mechanism optimization studies, the active parameters were selected based
either on local sensitivity coefficients or on products of the local sensitivity coefficient
and the uncertainty of the parameters. In this work, we propose a very efficient novel
method, called PCALIN, which uses not only the local sensitivities, but also considers
the uncertainty and the correlation of parameters, and furthermore the uncertainty
and the weights of the experimental data. The method also identifies the relevant
subset of the experimental data collection, thereby allows significant savings in
computation time at mechanism optimization.
