Prediction of Hole Quality in deep hole drilling of AISI 316 Steel: A Hybrid AI Approach

• Experimental measurement of hole quality characteristics after deep hole drilling of AISI 316 using CNC peck drilling approach. • Literature review of Deep hole drilling applications and its optimization. • Range of drill tool materials used are HSS, M35, M42, TiAlN, Solid carbide. • Assessment of drilled hole quality metrics through descriptive statistics. • Development of GAN model for multi-response prediction of deep hole drilling process. • Hybrid GANAPSO methodology is used for model optimization and prediction performance of regression, GAN and optimized GAN are comapred. The growing demand for deep hole drilling of high-strength AISI 316 steels, commonly used in critical fields like aerospace, biomedical, and marine industries, calls for advanced modelling and optimization techniques to achieve the best machining results. This study explores how different tool types, including HSS, M35, M42, TiAlN-coated, and solid carbide drills, affect the hole quality of AISI 316 steel under various drilling parameters such as spindle speed, feed rate, depth of cut, and drill diameter. Conducted in two stages (trials 1-9 and trials 10-18), the experiments measured the hole quality metrics like surface roughness, machining time, hole diameter, and circularity error using standard methods. A new hybrid approach involving Generative Adversarial Network and Adaptive Particle Swarm Optimization (GAN+APSO) has been developed to model and predict hole quality. This model captures the nonlinear relation between drilling parameters and resulting hole quality metrics. Verified against experimental data, it showed high reliability and accuracy in estimating machining outcomes. The results indicate that the GAN+APSO model outperforms traditional predictive methods, with minimum prediction error. This paper presents a robust optimization framework for deep hole drilling of AISI 316 steel in industrial applications, highlighting the value of combining advanced AI models with experimental validation.