| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 17.06 MB | Adobe PDF | |||
| 1.74 KB | License |
Authors
Advisor(s)
Abstract(s)
Um dos fatores mais importantes na produção de moldes para injeção de plástico é a estimativa do custo dos serviços de maquinagem, que representam uma parte significativa do preço final do molde. O custo destes serviços é habitualmente determinado em função do tempo de maquinagem, cujo cálculo é geralmente longo e dispendioso. Se for considerado que as peças dos moldes de injeção são todas diferentes, compreende-se que a correta e célere estimativa de tempos de maquinagem é de grande importância para o sucesso de uma empresa. Esta dissertação apresenta uma proposta de aplicação de redes neuronais artificiais na estimativa de tempos de maquinagem de peças standard de moldes de injeção de plástico. Para o efeito, foram simuladas peças e calculados os tempos de maquinagem para recolher dados suficientes para o treino das redes neuronais. Foi estudada a influência da arquitetura de rede, da quantidade de dados de entrada e das variáveis utilizadas no treino da rede, de forma a encontrar a rede neuronal com maior precisão. A aplicação de redes neuronais neste trabalho revelou-se uma forma célere e eficaz de calcular os tempos de corte, podendo dar um forte contributo a empresas do setor.
One of the most important factors in the production of plastic injection molds is the cost estimation of machining services which represents a significant part of the final mold price. The cost of these services is commonly determined as a function of the machining time, which is usually long and expensive to calculate. If it is considered that the injection mold parts are all different, it is understood that the correct and quick estimation of machining times is of great importance for the success of a company. This dissertation presents a proposal for the application of artificial neural networks in machining time estimation for standard injection molds parts. For this purpose, parts were simulated and machining times were calculated to collect enough data for training the neural networks. The influence of the network architecture, the amount of input data and the variables used in the training of the network were studied in order to find the neural network with greater precision. The application of neural networks in this work proved to be a quick and efficient way to calculate cutting times, which can give a strong contribution to companies in the sector.
One of the most important factors in the production of plastic injection molds is the cost estimation of machining services which represents a significant part of the final mold price. The cost of these services is commonly determined as a function of the machining time, which is usually long and expensive to calculate. If it is considered that the injection mold parts are all different, it is understood that the correct and quick estimation of machining times is of great importance for the success of a company. This dissertation presents a proposal for the application of artificial neural networks in machining time estimation for standard injection molds parts. For this purpose, parts were simulated and machining times were calculated to collect enough data for training the neural networks. The influence of the network architecture, the amount of input data and the variables used in the training of the network were studied in order to find the neural network with greater precision. The application of neural networks in this work proved to be a quick and efficient way to calculate cutting times, which can give a strong contribution to companies in the sector.
Description
Keywords
Maquinagem Estimativa de tempos Moldes de injeção Redes Neuronais Artificiais Machining Time estimation Injection molds Artificial Neural Networks