Object-Centric Thermal Mapping: A wireless sensor network perspective

This work mainly focuses on developing a concept towards building an object-centric thermal mapping (OCT Map) environment based on the use of wireless sensor networks. The sensor network is represented through infrastructural sensors of any functional space like the cool-room or a food-store. These sensors facilitate initial values for the calculation of the temperature at a given location within the space. In this work, artificial neural networks (ANN) are used for predicting the thermal map of a given space. It is found that the estimated temperature distributions at any arbitrary points have considerably low root mean square error (RMSE). However the accuracy of the prediction and number of infrastructural nodes are concerned, at 8 infrastructural nodes and the room volume of 4500 m³, the temperature can be predicted with at least RMS error of 0.02°C. The results divulged that the design of using neural networks for temperature mapping is feasible.