An AI Approach to Noise Prediction

The goal of the work described here is to produce a computationally efficient technique, with output of data that can be easily visualised, for problems involving noise prediction.

An AI (Artificial Intelligence) approach is adopted. The particular techniques being applied as a part of the approach here involves, firstly, defining the geometry of the volume that we are interested in (the Geographical Space) and then later elements within that space such as objects and constraints. This is the aspect of the work which is particularly interesting since we use a technique which effectively linearises three dimensional space. The result is a significant reduction in computational requirements and a consequent increase in speed and efficiency of analysis. The technique used here for spatial representation is termed Tesseral Addressing.

This technique can be combined with that for performing analysis (more correctly reasoning) within the Geographical Space. The application of the combined techniques is illustrated in the form of an analysis of a potential noise pollution problem at a residential building.