Prediction for Combustion States of Waste-to-Energy Plants with Fuzzy Relational Maps of Sensors

In waste-to-energy plants, keeping a stable combustion state is required to decrease hazardous gases and keep the steam volume appropriate for generating electricity. Since waste has various properties and plants have many sensors intricately related to each other, it is very difficult to find a very complex relation of many sensors for keeping a stable combustion state. In this paper, we propose a fuzzy relational map of sensors, FuRMS, as a method to represent relations of many sensors. For data of unknown state, we calculate evaluation values with the map. We apply it to a waste-to-energy plant. We construct a map using 14110 data of the stable state and evaluate other 390 data of unknown state. The result shows that the map can predict combustion states, that is, the evaluation value begins to decrease 10 to 30 minutes earlier than beginnings of unstable states. We propose some methods to extend this method to be more useful in the future.