|
|
Use of Neural Networks for the Stiffness Calculation of a Spur Gear Transmission
Planka, Michal ; Krpalek, David (referee) ; Lošák, Petr (advisor)
The aim of this master's thesis is to build artificial neural network that is able to calculate varying single tooth-pair mesh stiffness of spur gear for given input parameters. The training set for this network was determined by computational modelling by finite element method. Therefore, creating of computational model and mesh stiffness calculating were a partial aim of this thesis. Input parameters for stiffness calculation were number of driving and driven gear teeth and gear loading. Creating of computational model and performing series of simulations was followed by creating artificial neural network. Multilayer neural network with backpropagation training was chosen as a type of the network. Created neural network is sufficiently efficient and can determine varying mesh stiffness in input set range for learned input parameters and for values of parameters that are not included in training set as well. This neural network can be used for varying single tooth-pair mesh stiffness estimation in input set range.
|
|
|
Dynamic Model of Harmonic Gearbox
Garami, Boris ; Lošák, Petr (referee) ; Hadaš, Zdeněk (advisor)
This thesis deals with the design of a dynamic model of a harmonic drive. It includes a theoretical study aimed at the analysis of the harmonic drive gearing principle and its nonlinear properties. The first part of the practical section deals with the analytical calculation of the nondeformed geometry of the Flexspline. Based on these results, several simulations in ANSYS are created to identify torsional characteristics of a harmonic drive. These simulation models are further enhanced by the analysis of clearance, backlash and inaccuracies and their impact on torsional properties. By using MATLAB /Simulink, several dynamic submodels are created representing the individual characteristics of nonlinearities in harmonic drives. Furthermore, a comprehensive dynamic model is created of the mechatronic system which is describing all nonlinearities and kinematic error of the transmission. The dynamic model is also experimentally verified based on its damping properties.
|
|
|
Use of Neural Networks for the Stiffness Calculation of a Spur Gear Transmission
Planka, Michal ; Krpalek, David (referee) ; Lošák, Petr (advisor)
The aim of this master's thesis is to build artificial neural network that is able to calculate varying single tooth-pair mesh stiffness of spur gear for given input parameters. The training set for this network was determined by computational modelling by finite element method. Therefore, creating of computational model and mesh stiffness calculating were a partial aim of this thesis. Input parameters for stiffness calculation were number of driving and driven gear teeth and gear loading. Creating of computational model and performing series of simulations was followed by creating artificial neural network. Multilayer neural network with backpropagation training was chosen as a type of the network. Created neural network is sufficiently efficient and can determine varying mesh stiffness in input set range for learned input parameters and for values of parameters that are not included in training set as well. This neural network can be used for varying single tooth-pair mesh stiffness estimation in input set range.
|
|
|
Dynamic Model of Harmonic Gearbox
Garami, Boris ; Lošák, Petr (referee) ; Hadaš, Zdeněk (advisor)
This thesis deals with the design of a dynamic model of a harmonic drive. It includes a theoretical study aimed at the analysis of the harmonic drive gearing principle and its nonlinear properties. The first part of the practical section deals with the analytical calculation of the nondeformed geometry of the Flexspline. Based on these results, several simulations in ANSYS are created to identify torsional characteristics of a harmonic drive. These simulation models are further enhanced by the analysis of clearance, backlash and inaccuracies and their impact on torsional properties. By using MATLAB /Simulink, several dynamic submodels are created representing the individual characteristics of nonlinearities in harmonic drives. Furthermore, a comprehensive dynamic model is created of the mechatronic system which is describing all nonlinearities and kinematic error of the transmission. The dynamic model is also experimentally verified based on its damping properties.
|
guest ::
