|
|
Combination of numerical mathematics and neural network for the model of breakout prediction
Srba, Jan ; Štětina, Josef (referee) ; Mauder, Tomáš (advisor)
This thesis deals with creation of breakout prediction model implemented via artificial neural network in MATLAB, hence it contains theoretical description of artificial neural networks and also presents how the networks were used for the breakout prediction problem. In addition, models of heat conduction through the mold wall were created and used for generating neccesary data for proper training of artificial neural network. The thesis contains proper description of used numerical methods for solving the heat conduction problems. In the thesis are presented partial results and also results of testing created artificial neural network with usage of real data.
|
|
|
Model predictive Control for continuous casting of steel
Zemanová, Hana ; Mauder, Tomáš (referee) ; Štětina, Josef (advisor)
In this thesis an equation of heat conduction including phase and structural changes is derived, involving various boundary conditions. It seems to be the most suitable to calculate the equation by enthalpy method. In this equation not only enthalpy occurs, but also the temperature, and in consequence the relationship between these variables is quite complicated. In this paper I use the values measured or calculated using solidifcation models. The calculation is implemented in Matlab Simulink, which is a very popular blocks scheme in common practice of regulation. The calculation is based on steady state set up with help of experts and as a result, the program could be put into practice. The program calculates the intensity of cooling according to the initial casting speed, casting inlet temperature and the desired temperature curves. The rate of inuence of cooling can be changed according to the given criteria. The thesis compares the surface temperatures and cooling in the case of a predictive controller is or is not applied in the program.
|
|
|
Combination of numerical mathematics and neural network for the model of breakout prediction
Srba, Jan ; Štětina, Josef (referee) ; Mauder, Tomáš (advisor)
This thesis deals with creation of breakout prediction model implemented via artificial neural network in MATLAB, hence it contains theoretical description of artificial neural networks and also presents how the networks were used for the breakout prediction problem. In addition, models of heat conduction through the mold wall were created and used for generating neccesary data for proper training of artificial neural network. The thesis contains proper description of used numerical methods for solving the heat conduction problems. In the thesis are presented partial results and also results of testing created artificial neural network with usage of real data.
|
|
|
Model predictive Control for continuous casting of steel
Zemanová, Hana ; Mauder, Tomáš (referee) ; Štětina, Josef (advisor)
In this thesis an equation of heat conduction including phase and structural changes is derived, involving various boundary conditions. It seems to be the most suitable to calculate the equation by enthalpy method. In this equation not only enthalpy occurs, but also the temperature, and in consequence the relationship between these variables is quite complicated. In this paper I use the values measured or calculated using solidifcation models. The calculation is implemented in Matlab Simulink, which is a very popular blocks scheme in common practice of regulation. The calculation is based on steady state set up with help of experts and as a result, the program could be put into practice. The program calculates the intensity of cooling according to the initial casting speed, casting inlet temperature and the desired temperature curves. The rate of inuence of cooling can be changed according to the given criteria. The thesis compares the surface temperatures and cooling in the case of a predictive controller is or is not applied in the program.
|
guest ::
