Publicação
Application of computational intelligence techniques in optimization and inverse heat transfer problems
| Resumo: | Computational Intelligence can be defined as a collection of methods that solve problems by means of intelligence, reasoning, and tolerance to imprecision. In this thesis, Computational Intelligence techniques such as optimization metaheuristics, neural networks, fuzzy logic, and bayesian inference are applied to solve inverse heat transfer problems. To achieve this, five case studies with that aim are presented The first study deals with the problem of identification of contact failures in the bonding of different materials. This is formulated as an Inverse Heat Conduction Problem and solved with an Artificial Neural Network. The second one employs the Fuzzy Levenberg-Marquardt method with Tikhonov Regularization in order to solve a combined conduction-radiation inverse problem. In the third, the performance of the Topographical Global Optimization algorithm is compared when initialized with two different point generators and finalized with two different local search methods for the solutions of inverse heat transfer problems. The fourth study presents a method to predict the optimal value of the Tikhonov’s regularization parameter by solving simplified versions of the inverse problems considered, where tests are performed on three function estimation inverse heat transfer problems. In the fifth study, the formulation and solution of a bioheat transfer model of the brain containing a Deep Brain Stimulation (DBS) lead is presented and, furthermore, this solution is used for solving an inverse problem of determining the thermal and electrical conductivities of the brain tissue using measurements supposedly obtained with a sensor inside the DBS lead. In conclusion, this thesis shows the capabilities of various Computational Intelligence techniques in solving inverse heat transfer problems. The results presented contribute to two main categories: the field of inverse heat transfer, its applications in science and engineering; and the development of efficient computational procedures based on Computational Intelligence techniques. |
|---|---|
| Autores principais: | Jardim, Lucas Correia da Silva |
| Assunto: | Inverse heat transfer problem Computational intelligence Optimization Bayesian inference Neural networks Fuzzy logic Problema inverso em transferência de calor Inteligência computacional Otimização Redes neurais Inferência bayesiana Lógica difusa Problemas inversos (Equações diferenciais) Calor Transmissão CIENCIAS EXATAS E DA TERRA::MATEMATICA::MATEMATICA APLICADA |
| Ano: | 2023 |
| País: | Brasil |
| Tipo de documento: | tese de doutoramento |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | IBICT |
| Idioma: | português |
| Origem: | Oasisbr |
| Resumo: | Computational Intelligence can be defined as a collection of methods that solve problems by means of intelligence, reasoning, and tolerance to imprecision. In this thesis, Computational Intelligence techniques such as optimization metaheuristics, neural networks, fuzzy logic, and bayesian inference are applied to solve inverse heat transfer problems. To achieve this, five case studies with that aim are presented The first study deals with the problem of identification of contact failures in the bonding of different materials. This is formulated as an Inverse Heat Conduction Problem and solved with an Artificial Neural Network. The second one employs the Fuzzy Levenberg-Marquardt method with Tikhonov Regularization in order to solve a combined conduction-radiation inverse problem. In the third, the performance of the Topographical Global Optimization algorithm is compared when initialized with two different point generators and finalized with two different local search methods for the solutions of inverse heat transfer problems. The fourth study presents a method to predict the optimal value of the Tikhonov’s regularization parameter by solving simplified versions of the inverse problems considered, where tests are performed on three function estimation inverse heat transfer problems. In the fifth study, the formulation and solution of a bioheat transfer model of the brain containing a Deep Brain Stimulation (DBS) lead is presented and, furthermore, this solution is used for solving an inverse problem of determining the thermal and electrical conductivities of the brain tissue using measurements supposedly obtained with a sensor inside the DBS lead. In conclusion, this thesis shows the capabilities of various Computational Intelligence techniques in solving inverse heat transfer problems. The results presented contribute to two main categories: the field of inverse heat transfer, its applications in science and engineering; and the development of efficient computational procedures based on Computational Intelligence techniques. |
|---|
Atividades financiadas
Não foram encontrados projetos financiados.