PERFORMANCE OPTIMIZATION OF BS970 MILD STEEL TURNING PARAMETERS UNDER SiC-MoS2 HYBRID NANOFLUID USING HYBRID DEEP BELIEF NETWORK BASED COOT OPTIMIZATION

Abstract

In the modern world of competitive manufacturing, turning is a basic and important process that needs to be optimized for the fast-evolving industrial nature. It has its application in various fields involving numerous materials. This research study deals with modeling with optimization on the turning process parameters of turning E2-BS970 mild steel with cryogenically treated tungsten carbide tool under NFMQL having [Formula: see text] nanofluid. The parameters considered as the input factors are spindle speed and depth of cut along with feed rate and lubrication condition. The design of the experiment is done based on Box–Behnken design for 27 trials in the response surface method. The Deep Neural Network (DBN) and the Coot optimization algorithm are employed using MATLAB software for the prediction modeling. The prediction model developed by the DBN sigmoid function gave minimal error with prediction regression values of 0.99988 for MRR, 0.99516 for Cutting temperature, and 0.99545 for Tool life. The optimized result by CO shows a value of 188.89 rpm for spindle speed, 0.2318[Formula: see text]mm/rev for feed rate, Doc of 1.45[Formula: see text]mm, and 0.5015[Formula: see text]vol.% nanofluid MQL condition. The optimized values of MRR are 3.35[Formula: see text]mm3/min, [Formula: see text]C cutting temperature, and 1543.12[Formula: see text]s of tool life. The DBN model combined with the Coot algorithm shows minimal deviation compared to the experimental results validation and confirmatory analysis and is deemed suitable for efficient prediction.