This article presents two algorithms for online mode thermal stress determination in machinery components based on Green's Function Technique approach and on the artificial neural network, respectively. A modification of the GFT algorithm is proposed involving the use of the component metal temperature rather than the temperature of the flowing fluid. The algorithms developed were used to calculate stresses in rotors and casings of new generation turbines. A possibility was also shown of using the algorithms to solve the inverse problem, i.e., to control the rate of temperature changes while keeping stresses within an allowable range.