Artificial neural networks applications in the field of separation science optimisation

Optimisation procedures in chromatography usually exploit "hard" model approaches or methods based on the coupling of experimental design techniques and surface response methods. A powerful alternative has been recently provided by Artificial Neural Networks (ANNs), which allow to obtain "soft" models, not based on the a-priori knowledge of the mechanisms involved in the separation, and permit to model non-linear relationships. Most of ANNs applications in chromatography regard multivariate calibration and prediction or studies on structure-activity relationships. They have also been recently applied to the optimisation of process and mobile phase composition parameters: in these applications they are usually coupled to response surface methods and/or experimental design techniques. This review reports the main applications of ANNs to the optimisation of different separation techniques: high-performance liquid-chromatography, ion and gas chromatography, electro-separation methods. A section describing the main experimental designs and the theory of ANNs is also present.