Drilling is one of the oldest and the most widely used of all machining processes, comprising about one third of all metal- machining operations. The present work consist of drilling an aluminum silicon nitride composite forged plate using high-speed steel drill bit and was carried by varying the cutting speed and feed. Theoretical analysis in the present work involved monitoring of drilled hole status of composite based on surface roughness and cylindricity using the independent variables machining time, tool tip temperature, vibration, flank wear (both average and maximum), cylindricity and cutting conditions by sophisticated methods of signal analysis like Multiple Regression Analysis (MRA), Group method Data Handling Technique (GMDH) and Pattern Recognition Technique (PRT) like Back Propagation Neural Network (BPNN) was used. Comparisons of the three theoretical methods for estimation of surface roughness and cylindricity with measured one were carried out. The influence of network architecture is used to know the drilled hole status based on surface roughness and cylindricity was studied.