*2.1. Data Preparation*

Data preparation was the first step in the development of the ANN. The goal was to simulate a machining process, calculating the machining time of several plastic injection mold plates using CAM programming software (MasterCAM® 2022), and a library of tools created for this purpose. As seen before, the amount of data used for training the ANN influences its accuracy, making it necessary to calculate the machining time of hundreds of different parts to generate enough data. The traditional method for calculating machining times requires a 3D model and the CAM programming of each part, which is time-consuming considering the high number of parts needed; usually this calculated time is then compared to the actual required machining time to produce a said part. Therefore, an empirical-based method was created where the machining time of each element is calculated individually and subsequently assigned to each part at random, allowing the calculation of a multitude of different parts. The workpiece material and the machined geometries are defined based on common requests obtained to produce injection mold parts. These parts are usually standardized. Furthermore, literature research provided insight on the best-suited parameters and strategies to produce said parts. This information is complemented by contacting manufacturers of these standardized components. They usually offer practical and industrially acquired knowledge, which can be very useful when defining these parameters. Then, the machining times of 30 modeled parts are calculated by the traditional (simulation and modeling, offering a comparison of measurements made after component production) and empirical methods to compare the obtained values and validate the devised empirical method. A definition of these methods can be found below:

