Hybrid Taguchi-Artificial Neural Network Method for Design and Optimization of Microfluidic Biosensor

A Taguchi
$L_{32}(2^{8})$
orthogonal design combined with an analysis of variance ANOVA will be used to optimize the microfluidic biosensor with confinement. For this, eight factors such as Reynolds number (Re), Damkohler number (Da), relative adsorption capacity
$(\boldsymbol{\sigma})$
, equilibrium dissociation constant (K
D
), Schmidt number (Sc), confinement coefficient
$(\boldsymbol{\alpha})$
, confinement position (X) and inclination angle
$(\boldsymbol{\theta})$
have been analyzed. In order to accurately predict the results, a predictive model based on the ANN artificial neural network was developed. Taguchi's analysis suggests that the optimal moment is obtained when position X is at level 1 and the other factors are at level 2. This combination gave a response time of 9.6. ANOVA analysis shows that the Damkohlar number × relative adsorption capacity interaction (41.35% contribution), the Reynolds number*confinement coefficient interaction (11.35% contribution), the Damkohler number (11.35% contribution) of 9.55%) and the equilibrium dissociation constant (contribution of 6.81%) are the interactions and the most influencing factors on the response time. Response time was predicted by ANN with 99% accuracy. The ANN prediction was more precise than the regression model and the numerical simulation data.