New software aims to better predict street noise levels
By Darren Quick
June 14, 2012
House hunters could soon have a useful tool to turn to when seeking out a potential new pad. Researchers from the University of Granada have developed software that they claim can accurately predict future noise levels in a street. The system not only predicts the frequency of noise, but the type of noise that potential residents would have to put up with.
While paying a visit to a property will provide potential purchasers with much of the information required to make a decision on whether to buy or not, reliably forecasting the amount of noise will depend largely on the time of the visit. And with noise classified as a pollutant by the World Health Organization (WHO), it is one of the major factors affecting one’s enjoyment and ability to relax at home.
The software designed by researchers from the University of Granada Departments of Computer Sciences and Artificial Intelligence, Civil Engineering and Applied Physics applies a neural network model in a system that they claim can predict urban noise levels with a reliability of 95 percent.
The system takes into account a number of variables, including street type, road conditions, average speed of passing vehicles, and road works, to produce a forecast of future noise levels. The researchers are currently working to reduce the number of variables required to form an accurate noise forecast.
Unlike mathematical models that rely on a specific set of data, which are usually employed to predict noise levels, the team claims theirs is the first urban noise assessment system to use “soft computing”, a field of computer science that tolerates a bit of imprecision and approximation within its calculations.
The research team relied on a set of noise data collected in Granada in 2007 to develop their system, but they are now collecting further data in other cities against which to validate their model.
Source: University of Granada
Just enter your friends and your email address into the form below
For multiple addresses, separate each with a comma