Page 166 - Data Science Algorithms in a Week

150 Djordje Cica and Davorin Kramar

based on the on the geometry and physical characteristics of the machining process.
However, due to the large number of interrelated machining parameters that have a great
influence on cutting forces it is difficult to develop an accurate theoretical cutting forces
analytical model. Therefore, over the last few decades, different modeling methods based
on artificial intelligence (AI) have become the preferred trend and are applied by most
researchers for estimation of different parameters of machining process, including cutting
forces, tool wear, surface roughness, etc. Artificial neural networks (ANN) are by now
the most popular AI method for modeling of various machining process parameters.
There are numerous applications of ANN based modeling of cutting forces in turning
reported in the literature. Szecsi (1999) presented a three-layer feed-forward ANN trained
by the error back-propagation algorithm for modeling of cutting forces. Physical and
chemical characteristics of the machined part, cutting speed, feed, average flank wear and
cutting tool angles were used as input parameters for training ANN. The developed
model is verified and can be used to define threshold force values in cutting tool
condition monitoring systems. Lin, Lee, and Wu (2001) developed a prediction model for
cutting force and surface roughness using abductive ANN during turning of high carbon
steel with carbide inserts. The ANN were trained with depth of cut, feed and cutting
speed as input parameters. Predicted results of cutting force and surface roughness are
found to be more accurate compared to regression analysis. Sharma, Dhiman, Sehgal, and
Sharma (2008) developed ANN model for estimation of cutting forces and surface
roughness for hard turning. Cutting parameters such as approaching angle, speed, feed,
and depth of cut were used as input parameters for training ANN. The ANN model gave
overall 76.4% accuracy. Alajmi¹ and Alfares¹ (2007) modeled cutting forces using back
propagation ANN with an enhancement by differential evolution algorithm. Experimental
machining data such are speed, feed, depth of cut, nose wear, flank wear and, notch wear,
were used in this study to train and evaluate the model. The results have shown an
improvement in the reliability of predicting the cutting forces over the previous work.
Zuperl and Cus (2004) were developed supervised ANN approach for estimation of the
cutting forces generated during end milling process. The predictive capability of using
analytical and ANN models were compared using statistics, which showed that ANN
predictions for three cutting force components were closer to the experimental data
compared to analytical method. Aykut, Gölcü, Semiz, and Ergür (2007) used ANN for
modeling cutting forces with three axes, where cutting speed, feed and depth of cut were
used as input dataset. ANN training has been performed using scaled conjugate gradient
feed-forward back-propagation algorithm. Results show that the ANN model can be used
for accurate prediction of the cutting forces. Cica, Sredanovic, Lakic-Globocki, and
Kramar (2013) investigate prediction of cutting forces using ANN and adaptive networks
based fuzzy inference systems (ANFIS) as a potential modeling techniques. During the
experimental research focus is placed on modeling cutting forces in different cooling and
lubricating conditions (conventional, high pressure jet assisted machining, and minimal