This Chapter aims to describe the stages of developing an appropriate control schematic for dynamic breakage with respect to particle size distribution. The first phase of the work is to establish a proper theoretical background in order to establish the relationship of grinding of one type of particle to another type within a definite size range. Sufficient experimental results obtained from laboratory data, conducted in the Mineral dressing laboratory using a ball mill at Indian Institute of Engineering Science and Technology, Shibpur, India, which supports the theoretical framework. Experimental analysis of the data reveals some interesting results that help to predict the particle size distribution of materials, obtained through grinding operation. A variant of back propagation learning algorithm is used to train the input-output values of different kinds of particles crushed by the ball mill. Instead of a single value, the multi-valued information of the crushed materials, represented using the p-dimensional vector (p > 1), is fed to each node of the input-output layer of a feedforward neural network. The training algorithm has been studied for a wide range of input-output values and satisfactory results are obtained, especially when the output vector is small enough compared to the input vector. Experimental data are verified with the simulation results to establish the fundamental relationships between grinding of different kinds of particles.