INVESTIGATION OF SURFACE DEFECTS OF EDMED MONEL 400 AND PROCESS OPTIMIZATION USING K-FOLD NN-INTEGRATED MOBO

The paper investigates the effect of different tool materials on the defects developed on the surface of an electro-discharge-machined Monel 400 and proposes an efficient approach by combining a multi-objective bonobo optimizer (MOBO) with the K-fold neural network (K-fold NN) for enhancing its machining performance. The novelty of this work is implementing a K-fold NN for finding prediction models of an electro-discharge machining (EDM) problem with a small experimental dataset and utilizing a MOBO to optimize the process parameters. During the study, process performance is evaluated by two machining outputs, viz. material removal rate (MRR) and surface roughness (SR). A study on machined surface defects is then accomplished by analyzing their scanning electron microscopy (SEM) pictures. The results indicate that using copper electrodes reduces the chances of fatigue failure of the Monel 400-made machined component by optimizing the surface crack size and density. The obtained dataset is then used in a K-fold NN to build prediction models between factors and responses. After that, the MOBO uses the prediction models as fitness functions to find non-dominated optimal solutions. An improvement in MRR and SR up to 21.237% and 27.166%, respectively, is achieved using the proposed methodology, compared to the experimental results. Confirmatory tests verify the effectiveness of the developed NN models by indicating an average error percentage of 3.353 and 2.662 in the prediction of MRR and SR, respectively. The performance comparison reveals the superiority of MOBO over two popular algorithms, namely multi-objective genetic algorithm (MOGA) and multi-objective particle swarm optimization algorithm (MOPSO), as well as multi-objective dragonfly algorithm (MODA), an algorithm recently used to solve the EDM problem.