Prediction of Surface Roughness of a Nimonic C-238 by Using the ANN Technique

Artificial Neural Network (ANN) approach is employed to forecast surface roughness (Ra)in the turning of Nimonic C238 using cutting parameters such as Feed, Speed, and Depth of cut. Experiments using carbide inserts were carried out, and the outcomes of these experiments were measured. The measured responses were fed into an ANN, which was then trained to predict the surface roughness. It has been discovered that ANN models provide better surface roughness (Ra) predictions than mechanical assessment techniques. Furthermore, the outcomes of ANN predictions are compared to the surface roughness (Ra)value obtained through experiments. Finally, it was established that the stylus measuring technique is less accurate than ANN models in determining surface roughness. These forecasts are useful for real-time operation control, which is required to obtain the required surface roughness. The created ANN approach can accurately estimate the Surface roughness (Ra)value of the Surface roughness (Ra)with an error percentage of less than 3.45%.

I. Introduction

When it comes to the process of machining Nimonic C238, a correct prediction of the surface roughness is of the utmost importance. This is done to ensure that the machined components will have the required quality and performance. Nimonic C238, a high-performance nickel-based superalloy, is put to use in a variety of industries due to the remarkable mechanical qualities and heat resistance that it possesses. These industries include aerospace, power generation, and others. The machining of Nimonic C238 is difficult due to the material's high hardness, low thermal conductivity, and strong inclination to work harden. [1] These characteristics make the material difficult to work with. Historically, the prediction of surface roughness in machining operations has relied on empirical models that were developed from a large number of comprehensive experimental trials. However, these methods frequently have shortcomings, particularly when it comes to effectively capturing the intricate interactions that exist between the parameters of the machining process and the surface roughness. The application of Artificial Neural Networks, often known as ANN s, has emerged as a potentially useful method for overcoming these limitations[2].