Self-refreshing SOM for dynamic process state monitoring in a circulating fluidized bed energy plant

Recent studies have shown that intelligent systems approaches have potential for solving complex industrial problems. In addition, process data is valuable for the process state monitoring and knowledge management of the process and it can be used for the optimization of productivity. In this paper novel approach for dynamic process state monitoring is presented and applied in a circulating fluidized bed energy plant. The proposed method is based on self-refreshing modification of the self-organizing map where previously learned data was used recursively to avoid catastrophic forgetting. The results of the simulations showed that method is a useful tool for monitoring process states of dynamic processes and produced reasonable results also in the case of a circulating fluidized bed energy plant