Neural model of residential building air infiltration process

• Main Authors: Pawel Malinowski, Iwona Polarczyk, Jerzy Piotrowski
• Format: Article
• Language: English
• Published: Vilnius Gediminas Technical University 2006-03-01
• Series: Journal of Civil Engineering and Management
• Subjects: infiltration process; neural networks; neural models
• Online Access: http://localhost/journals.vgtu.lt/index.php/JCEM/article/view/8787
• ISSN: 1392-3730; 1822-3605

For the needs of this paper, certain area of issues associated with the infiltration process and neural networks were selected. In the scope of infiltration process, the measurements and analysis of air infiltration through the buildings have shown how difficult this process is in terms of strict mathematical models formulation. In the scope of neural networks, attention was given to multiâ€layer perceptrons and systems composed of them. It must be noted that, due to the novelty of this method, a detailed description of the neuron network theory was presented in this paper as well as a thorough discussion of the neuron modelling process. The work was carried out in two stages, with the first stage covering the development of detailed methodology for the construction of neural model and the second stage in which a developed methodology was applied to modelling real objects. Neuroninis infiltracijos proceso gyvenamajame name modelis Santrauka Å iam straipsniui pasirinkti aspektai, susijÄ™ su infiltracijos procesu ir neuroniniais tinklais. Infiltracijos pastatuose matavimai parodÄ—, koks sudÄ—tingas yra griežtas matematinis Å¡io proceso modeliavimas. Analizuojant infiltracijos procesÄ… neuroniniu tinklu požiÅ«riu, dÄ—mesys buvo skiriamas daugiasluoksniams perceptronams ir iÅ¡ jų sudarytoms sistemoms. Atsižvelgiant į Å¡io metodo naujumÄ…, straipsnyje pateikiamas tiek neuronų tinklo teorijos, tiek iÅ¡samus neuroninio modeliavimo proceso apraÅ¡ymas. Darbas buvo atliktas dviem etapais: pirmajame sukurta neuroninio modelio sudarymo metodika, antrajame â€Å¡i metodika buvo pritaikyta realiems objektams modeliuoti. Raktiniai žodžiai: infiltracijos procesas, neuroniniai tinklai, neuroniniai modeliai. First Published Online: 14 Oct 2010