Abstarct: Since much of the cost and energy consumption (about 60%) belong to the mineral processing sector, crushing machines, so measures must be considered in addition to achieving a desired size, the crushing excessive and indiscriminate use of materials power to prevent such an important role in the economic feasibility of the product and its price. In most cases, due to the inherent complexity of operations, the multiplicity of factors involved and the lack of good basic design, performance crushing circuit is less than anticipated. Hence, numerous mathematical models and optimization techniques for determining the optimum operating conditions shredding machines and processes used. Today most engineering problems using evolutionary algorithms for economic reasons, and time management is growing strongly. However, the equations for determining the power consumption due to the many variables that are often not available, can not effectively predict the mill power draw to be applied. So at the end of a valid equations and parameters were examined to calculate the power of the two models can be normalized mill empirical model (Austin) for estimating industrial mills, more adapted to the actual data. Then the equations defining the objective function baxsed on Particle Swarm Optimization was used to minimize the cost function baxsed on the actual amount of power devices and power draw system modeling was considered as the global optimum. The results of the two models showed that the system has excellent performance for optimization of power draw is in mill and power equation in mill normalized error value 0/0061 has the best performance.