St. Petersburg Polytechnic University Journal: Physics and Mathematics Научно-технические ведомости СПбГПУ. Физико-математические науки 2304-9782, 2618-8686, 2405-7223 24 Parametrical neural network models of classical and nonclassical problems for heat conduction equation Параметрические нейросетевые модели классических и неклассических задач для уравнения теплопроводности Vasiliev Alexander a.n.vasilycv@gmail.com Tarkhov Dmitry dtarkhov@gmail.com 10 09 2012 3 153 136 144

Problems of mathematical modeling for complex systems are considered in the paper in terms of neural network technique. System parameters are given in some variation intervals. Neural network model of nonstationary temperature field inthe case of both classical and non-classical problem statement is cited as an example. Results of neurocomputing for accurate and noise data are given.

 boundary value problem (bvp) interval parameter modeling artificial neural network training error functional global optimization