|
|
Neural networks for EMC modeling of small airplanes
Koudelka, Vlastimil ; Goňa,, Stanislav (referee) ; Raida, Zbyněk (advisor)
This thesis deals with neural modeling of electromagnetic field inside small aircrafts, witch can contain composite materials in their construction. Introduction to neural networks and its application in EMC of small airplanes is discussed in the first part of the text. In the second part of this thesis we design a simple EM model of small airplane. The airplane is simulated by two parallel dielectric layers (the left-hand side wall and the right hand side wall of the airplane). The layers are put into a rectangular metallic waveguide terminated by the absorber in order to simulate the illumination of the airplane by the external wave (both of the harmonic nature and pulse one). Numerical analyses are performed to search the relations between the distribution of an electromagnetic field inside the aircraft and electric parameters of model walls. The results of numerical analyses are used to train two types of neural network. In this way we can obtain accurate continuous model of electromagnetic field inside the aircraft. For the comparison with neural networks a multi-dimensional cubic spline interpolation is provided also. Neural classifiers are also investigated. We use them for classification of imaginary composite materials in terms of EMC. The nearest neighbour algorithm is applied as a classic approach to problem of classification.
|
|
|
Model-based evolutionary optimization methods
Bajer, Lukáš ; Holeňa, Martin (advisor) ; Brockhoff, Dimo (referee) ; Pošík, Petr (referee)
Model-based black-box optimization is a topic that has been intensively studied both in academia and industry. Especially real-world optimization tasks are often characterized by expensive or time-demanding objective functions for which statistical models can save resources or speed-up the optimization. Each of three parts of the thesis concerns one such model: first, copulas are used instead of a graphical model in estimation of distribution algorithms, second, RBF networks serve as surrogate models in mixed-variable genetic algorithms, and third, Gaussian processes are employed in Bayesian optimization algorithms as a sampling model and in the Covariance matrix adaptation Evolutionary strategy (CMA-ES) as a surrogate model. The last combination, described in the core part of the thesis, resulted in the Doubly trained surrogate CMA-ES (DTS-CMA-ES). This algorithm uses the uncertainty prediction of a Gaussian process for selecting only a part of the CMA-ES population for evaluation with the expensive objective function while the mean prediction is used for the rest. The DTS-CMA-ES improves upon the state-of-the-art surrogate continuous optimizers in several benchmark tests.
|
|
| |
|
|
Model-based evolutionary optimization methods
Bajer, Lukáš ; Holeňa, Martin (advisor) ; Brockhoff, Dimo (referee) ; Pošík, Petr (referee)
Model-based black-box optimization is a topic that has been intensively studied both in academia and industry. Especially real-world optimization tasks are often characterized by expensive or time-demanding objective functions for which statistical models can save resources or speed-up the optimization. Each of three parts of the thesis concerns one such model: first, copulas are used instead of a graphical model in estimation of distribution algorithms, second, RBF networks serve as surrogate models in mixed-variable genetic algorithms, and third, Gaussian processes are employed in Bayesian optimization algorithms as a sampling model and in the Covariance matrix adaptation Evolutionary strategy (CMA-ES) as a surrogate model. The last combination, described in the core part of the thesis, resulted in the Doubly trained surrogate CMA-ES (DTS-CMA-ES). This algorithm uses the uncertainty prediction of a Gaussian process for selecting only a part of the CMA-ES population for evaluation with the expensive objective function while the mean prediction is used for the rest. The DTS-CMA-ES improves upon the state-of-the-art surrogate continuous optimizers in several benchmark tests.
|
|
|
Neural networks for EMC modeling of small airplanes
Koudelka, Vlastimil ; Goňa,, Stanislav (referee) ; Raida, Zbyněk (advisor)
This thesis deals with neural modeling of electromagnetic field inside small aircrafts, witch can contain composite materials in their construction. Introduction to neural networks and its application in EMC of small airplanes is discussed in the first part of the text. In the second part of this thesis we design a simple EM model of small airplane. The airplane is simulated by two parallel dielectric layers (the left-hand side wall and the right hand side wall of the airplane). The layers are put into a rectangular metallic waveguide terminated by the absorber in order to simulate the illumination of the airplane by the external wave (both of the harmonic nature and pulse one). Numerical analyses are performed to search the relations between the distribution of an electromagnetic field inside the aircraft and electric parameters of model walls. The results of numerical analyses are used to train two types of neural network. In this way we can obtain accurate continuous model of electromagnetic field inside the aircraft. For the comparison with neural networks a multi-dimensional cubic spline interpolation is provided also. Neural classifiers are also investigated. We use them for classification of imaginary composite materials in terms of EMC. The nearest neighbour algorithm is applied as a classic approach to problem of classification.
|
guest ::
