A multi-agent approach using perceptron-based learning for robust operation of distributed chemical reactor networks

Controlling the individual reactors of a chemical reactor network producing different grades of a product requires intelligent reconfiguration strategies. Agent-based approaches are ideal for such distributed manufacturing processes, since they provide flexible, robust, and emergent solutions under dynamically changing process conditions. This paper proposes a multi-layered, multi-agent framework based on a decentralized online learning approach for the supervision of grade transitions in autocatalytic reactor networks. The values for the manipulated variables and the path to the target reactor are determined to give the least disturbance to the system. Case studies illustrate the performance of the approach in managing grade transition and disturbance rejection in a reactor network.

► We control autocatalytic reactors in a reactor network using an agent-based approach. ► Our decentralized multi-agent framework uses local controller agents on each reactor. ► Controller agents are capable of online learning using perceptrons and communication. ► The framework reconfigures the network and manages transitions among various grades. ► It can also track desired product grade and reject disturbances for robust operation.