Artificial neural-network-assisted stochastic process optimization strategies

This article presents two hybrid robust process optimization approaches int egrating artificial neural networks (ANN) and stochastic optimization forma lisms-genetic algorithms (GAs) and simultaneous perturbation stochastic app roximation (SPSA). An ANN-based process model was developed solely from pro cess input - output data and then its input space comprising design and ope rating variables was optimized by employing either the GA or the SPSA metho dology. These methods possess certain advantages over widely used determini stic gradient-based techniques. The efficacy of ANN-GA and ANN-SPSA formali sms in the presence of noise-free as well as noisy process data was demonst rated for a representative system involving a nonisothermal CSTR. The case study considered a nontrivial optimization objective, which, in addition to the conventional parameter design also addresses the issue of optimal tole rance design. Comparison of the results with those from a robust determinis tic modeling/optimization strategy suggests that the hybrid methodologies c an be gainfully employed for process optimization.