Publicação: Uso de redes neurais artificiais para avaliação comparativa do desenvolvimento de mudas de Eucalyptus spp. produzidas em tubetes de polipropileno e em berço germinador compostável a campo
dc.contributor.advisor | Tomaz, Rafael Simões [UNESP] | |
dc.contributor.author | Mingoranci, Vitória Costa | |
dc.contributor.institution | Universidade Estadual Paulista (Unesp) | |
dc.date.accessioned | 2023-07-04T20:47:01Z | |
dc.date.available | 2023-07-04T20:47:01Z | |
dc.date.issued | 2023-06-07 | |
dc.description.abstract | Atualmente, a sustentabilidade tem se tornado uma questão frequente e importante. Medidas que visam reduzir os impasses para uma produção sustentável estão sendo cada vez mais adotadas, especialmente na produção das florestas plantadas. A cultura do eucalipto é amplamente utilizada na produção de papel e celulose, devido ao seu grande potencial de produtividade tornando o Brasil o segundo maior produtor do mundo. No entanto, a qualidade das mudas desempenha um papel fundamental no crescimento e sobrevivência das plantas pós-plantio. O uso de tubetes é predominante na produção de mudas de eucalipto no setor florestal. Nesse contexto, há uma necessidade de utilizar recipientes renováveis para substituir os tubetes convencionais. Além disso, a inteligência artificial tem desempenhado um papel cada vez mais importante no setor florestal, sendo aplicada em diversas áreas, como medição de volume, diâmetro e altura das árvores, identificação de incêndios florestais, doenças e análises de diversidade genética. As Redes Neurais Artificiais, em particular, são altamente adaptáveis e flexíveis, permitindo a compreensão e simulação das relações presentes nos sistemas complexos, como os ecossistemas florestais. Este trabalho teve por objetivo comparar o desempenho de árvores de Eucalyptus spp produzidas em tubete de polipropileno e em Berço Germinador Compostável em campo, por meio do método baseado em Redes Neurais Artificiais. As plantas foram provenientes de um sistema de produção em estufa agrícola e transplantadas a campo. Foram realizadas avaliações de altura e diâmetro. Na análise estatística, considerou-se o ajuste de modelos de crescimento, lineares e não lineares, incluindo os modelos Logístico, Gompertz e Von Bertalanffy. Além disso, foram utilizadas diferentes arquiteturas e predição de fenótipos pelas Redes Neurais Artificiais, considerando um conjunto do tipo MLP (Perceptron Multicamada) com o treinamento supervisionado com algoritmo backpropagation. Foram consideradas duas camadas ocultas e diferentes quantidades de neurônios, bem como o uso de métricas para avaliar a eficiência. Também foi avaliada a biodegradabilidade dos tubetes SIS BGC®. Os resultados indicam que o uso do tubete SIS BGC® não afetou o desenvolvimento e pode ser considerado uma alternativa ao tubete convencional. No entanto, a taxa de biocompostabilidade não atingiu a porcentagem determinada. Em resumo, os modelos de crescimento não lineares apresentaram um melhor ajuste dos dados e as Redes Neurais Artificiais foram consideradas modelos adequados para predição de altura e diâmetro de árvores da espécie eucalipto. Além disso, a quantidade de neurônios nas arquiteturas testadas das redes não influenciou o desempenho dos modelos. | pt |
dc.description.abstract | Currently, sustainability has become a frequent and important issue. Measures that aim to reduce the impasses for a sustainable production are being increasingly adopted, especially in the production of planted forest. The culture of eucalyptus is widely used in the production of paper and pulp, due to its great potential for productivity, making Brazil the second largest producer in the world. However, the quality of the seedlings plays a key role in the growth and survival of the plants post-planting. The use of tubes is predominant in the production of eucalyptus seedlings in the forestry sector. In this context, there is a need to use renewable containers to replace conventional tubes. In addition, artificial intelligence has been playing an increasingly important role in the forest sector, being applied in several areas, such as volume, diameter, and height measurement of trees, identification of forest fires, diseases, and analysis of genetic diversity. Artificial Neural Networks, in particular, are highly adaptable and flexible, allowing the understanding and simulation of the relationships present in complex systems, such as forest ecosystems. This work aimed to compare the performance of Eucalyptus spp trees grown in polypropylene tubes and in Compostable Germinating Cradle in the field, using the method based on Artificial Neural Networks. The plants came from a production system in an agricultural greenhouse and were transplanted to the field. Evaluations of height and diameter were performed. In the statistical analysis, we considered the adjustment of growth models, linear and non-linear, including the Logistic, Gompertz, and Von Bertalanffy models. In addition, different architectures and phenotype prediction by Artificial Neural Networks were used, considering a MLP (Multilayer Perceptron) type ensemble with supervised training with backpropagation algorithm. Two hidden layers and different numbers of neurons were considered, as well as the use of metrics to evaluate efficiency. The biodegradability of SIS BGC® tubes was also evaluated. The results indicate that the use of SIS BGC® cores did not affect development and can be considered an alternative to conventional cores. However, the biodegradability rate did not reach the determined percentage. In summary, the non-linear growth models presented a better fit to the data and the Artificial Neural Networks were considered suitable models for predicting the height and diameter of trees of the eucalypt species. Furthermore, the number of neurons in the tested network architectures did not influence the performance of the models. | en |
dc.description.sponsorship | Pró-Reitoria de Pesquisa (PROPe UNESP) | |
dc.description.sponsorshipId | PROPE: 1712 | |
dc.identifier.uri | http://hdl.handle.net/11449/244378 | |
dc.language.iso | por | |
dc.publisher | Universidade Estadual Paulista (Unesp) | |
dc.rights.accessRights | Acesso aberto | |
dc.subject | Inteligência artificial | pt |
dc.subject | Mensuração florestal | pt |
dc.subject | Sustentabilidade | pt |
dc.subject | Agrotecnologia | pt |
dc.subject | Inovação | pt |
dc.subject | Artificial intelligence | en |
dc.subject | Forest measurement | en |
dc.subject | Sustainability | en |
dc.subject | Agrotechnology | en |
dc.subject | Innovation | en |
dc.title | Uso de redes neurais artificiais para avaliação comparativa do desenvolvimento de mudas de Eucalyptus spp. produzidas em tubetes de polipropileno e em berço germinador compostável a campo | pt |
dc.title.alternative | Use of artificial neural networks for comparative evaluation of the development of Eucalyptus spp. seedlings produced in polypropylene tubes and in a compostable germinator cradle in the field | en |
dc.type | Trabalho de conclusão de curso | pt |
dspace.entity.type | Publication | |
unesp.campus | Universidade Estadual Paulista (UNESP), Faculdade de Ciências Agrárias e Tecnológicas, Dracena | pt |
unesp.undergraduate | Engenharia Agronômica - FCAT | pt |
Arquivos
Pacote Original
1 - 1 de 1
Carregando...
- Nome:
- mingoranci_vc_tcc_dra.pdf
- Tamanho:
- 2.88 MB
- Formato:
- Adobe Portable Document Format
Licença do Pacote
1 - 2 de 2
Carregando...
- Nome:
- license.txt
- Tamanho:
- 2.44 KB
- Formato:
- Item-specific license agreed upon to submission
- Descrição:
Carregando...
- Nome:
- mingoranci_vc_autorizacao_dra.pdf
- Tamanho:
- 129.44 KB
- Formato:
- Adobe Portable Document Format
- Descrição: