The purpose of this work is to create a new algorithm for solving an inverse transient heat conduction problem in any extended complex domain. If the problem cannot be solved in the entire region under analysis, the domain is divided into parts and free boundary conditions are introduced between them. Then the solution is carried out in subsequent parts of the field. Depending on the shape and dimensions of the analysed part, an appropriate location of the temperature sensors, the amount and size of the control volumes, the time step and a convenient way to filter the temperature distribution over time and space can be proposed. Such actions make it possible to achieve better stabilization of the inverse method and obtain the solution in the whole area. The proposed procedure in subsequent parts of the domain belongs to the group of space marching methods. The analysis starts on the surface where temperature sensors are located and marches through space sequentially to the surface with an unknown boundary condition. The proposed algorithm can be used to optimize the power unit start-up and shutdown operation. It may also enable a reduction in the heat loss arising during the process and extend the power unit life. The presented procedure can be applied in monitoring systems working both in conventional and nuclear power plants. (C) 2018 Elsevier Ltd. All rights reserved.