Evolutionary Data-Driven Modeling

Springback Reduction in Tailor Welded Blank with High Strength Differential by Using Multi-Objective Evolutionary and Genetic Algorithms

“…Metamodeling technique for the problem under investigation is to use the recently developed Evolutionary Neural Network (EvoNN) approach, discussed extensively by. [13,30,33,43,44] Unlike traditional neural nets where the number of connections should be specified, the architecture and training weights from the EvoNN approach are optimized by working out an optimum trade-off between the complexity and accuracy of the models using the Predator-prey type Genetic Algorithm. One of the main features of neural networks obtained by EvoNN is that they have no tendency of over-or under-fitting.…”

“…This leads to a bi-objective Pareto-optimal problem, as simply elaborated in a recent book chapter [13] and an overview. [14] Elaborate mathematical details of Paretooptimality are available elsewhere.…”

“…[14] Elaborate mathematical details of Paretooptimality are available elsewhere. [13][14][15] Such problems are becoming increasingly tractable due to advent of nature inspired algorithms that are now ubiquitous in materials domain. [16,17] In this study we have resolved the Pareto problem using a data-driven evolutionary approach.…”

Evolutionary Data Driven Modeling and Multi Objective Optimization of Noisy Data Set in Blast Furnace Iron Making Process

“…In recent years, the authors have proposed the concept of Evolutionary Neural Network (EvoNN) [19,20] and Bi-objective Genetic Programming (BioGP) [21,22] that follow an evolutionary approach to model and optimize complex systems utilizing the available data set. To avoid over fitting or under fitting of the data [23] these algorithms construct a tradeoff between the complexity and the accuracy of the models, as detailed elsewhere. [17][18][19][20][21] The strategy is to construct optimum blast furnace models based upon the long term operational history of the furnace and both EvoNN and BioGP are capable of finding out the good solutions from the conventionally nonlinear and noisy blast furnace data.…”

“…EvoNN solves complex problems using evolving neural networks. As demonstrated in this study, it creates optimum models from the raw data and ultimately constructs Pareto frontiers [23] between various objectives. The strategy gives rise to a number of alternate optimum models, out of which a statistically suitable one is usually picked up using the corrected Akaike information criterion (AICc).…”