Application of neural networks in predicting airtightness of residential units

• Database containing measured air permeability values of residential units has been created.
• Corrective factors for input parameters for neural network learning have been determined.
• A successful air permeability prediction model has been obtained by using neural networks.

The paper describes the need to investigate airtightness of existing residential units in order to achieve adequate energy efficiency. There is a brief explanation of the blower door method used in field testing of airtightness in residential units. The mentioned method was used to create a database of airtightness test results obtained in situ at 58 residential units in the local area. The analysis of the results of these measurements indicates that there are four input parameters, which had been previously investigated and evaluated, that had a dominant effect on the test results. The examining of the possibility of successfully predicting airtightness was conducted by applying neural networks. The response results indicate that there is potential for successful prediction of airtightness, which was confirmed by independent validation through 20 additional measurements. In order to be able to generalize the results, further research needs to be conducted, using a bigger sample of measured data, and additional configurations of neural networks need to be examined.