Numerous industries process particles. In this work, we focused on how to efficiently picture the behaviour of particles by means of numerical simulations, laboratory experiments, and Artificial Neural Networks (ANNs). Particle-particle contact laws and particles size distributions determine the macroscopic results in Discrete Element Method (DEM) simulations. Commonly, contact laws depend on semi-empirical parameters which are difficult to obtain by direct microscopic measurements. To clarify this aspect, we present the related elements of the DEM theory. The ANN theory is also introduced to demonstrate ANN effectiveness towards the solution of inverse problems with non linear regression. Later, we describe the series of small scale DEM simulations with different sets of particle-based simulation parameters and particle distributions, which we performed. The macroscopic results of these simulations were used to train dedicated feed-forward ANNs by the backward propagation reinforcement algorithm. Concurrently, the bulk behaviours of raw particles were characterized by means of macroscopic laboratory experiments. These particles were those commonly used by metallurgical industries. At this point, the relationship between macroscopic results and microscopic DEM simulation parameters could be investigated. We subsequently utilized this artificial neural network to predict the macroscopic ensemble behaviour in relation to additional sets of particle-based simulation parameters and particle distributions. By this method, a comprehensive database was established, relating particle-based simulation parameters to macroscopic ensemble output. If compared to an experiment of a specific granular material, this database identifies valid sets of DEM parameters which lead to the same macroscopic results as observed in the experiments. Finally, we applied the results of this method of DEM parameter identification to two industrial scale processes of steel production.