A new generic architecture for the implementation of intelligent and distributed control systems

Nowadays distributed systems have demonstrated to be one of the best options for the industrial control systems implementation due their flexibility and power, which are increased with the application of artificial intelligence techniques. In this line, our group has developed several distributed systems applied to chemical analysis and industrial automation. For these implementations a CAN network was used. This article proposes a new layered architecture for the implementation of these intelligent and distributed systems. In that line, several techniques of discrete control may be used, from Petri nets to an expert system based on rule nets, while continuous control loops may be implemented by means of from traditional PID devices to distributed neural network approaches. Finally, for the communication layer implementation an industrial network such as CAN or TTP among others, will be used. This architecture defines the services offered by each layer, and specifies the interface between any two adjacent layers. It also defines the required characteristics of the protocols at any layers, and the capabilities that should be implemented. Finally, an implementation example of this architecture is presented.