The manufacturing environment today faces many challenges. Significant changes to cope with the new technologies are noticed in order to achieve customers¿ requirements imposing a complex manufacturing control which is needed to be robustness, reconfigurable to deal with this complex system. In order to achieve the above characteristics and requirements, many theories for new manufacturing generation were proposed, such as Holonic Manufacturing System (HMS). The major focus has been given to the design of the internal architecture of the Proposed Holon (P-Holon) which added a new approach such as feature selection using Bees Algorithm (BA) embedded inside their internal database, and using the Back Propagation Artificial Neural Network for decision making and classifying inside their control unit. This system is applied on two case studies from the real world industrial database from the University of California Irvine (UCI), one is the concrete compressive strength dataset and the other is grape juice data-set. The statistical result proves that the system has the ability to choose informative features, and get optimal classification result with high accuracy.