Predição da resistencia a compressão de um cimento industrial utilizando tecnicas de redes neurais artificiais
AUTOR(ES)
Fabiola Andrea Chandia Moreno
DATA DE PUBLICAÇÃO
2001
RESUMO
The industry of cement production has been growing each year and with this the trend in if applying techniques that also optimize its process. The process of cement production does not imply bigger difficulties, but as in any industry, it has its problems due to the high consumption of energy and represents the quality control of the final product, the cement, for the generation of the concrete. Mainly in respect to the Compression Strength, which is the main variable that indicates the quality ofthe cement The techniques of artificial neural networks can be applied the industrial majority of the systems or processes and if have become practical one, nowadays, carried through with very satisfactory results. By its easiness in simulating, among others characteristic, it can also be used to predict and shape referring industrial data to the process of cement production. In this work a model based on neural computation with the objective of prediction of the Compression Strength is developed in 3 days (R3), with enough approach in which had great advantage of being able in accordance with to anticipate with confidence the time for dispatching being the cement the norms, or then to be overcome the due precautions and corrections, in case that it is it are. The optimized model is represented by 15 neurons in the layer of input, 17 neurons in the intermediate layer and 1 neuron in the layer of output, that for each one of these neurons it was used the function of sigmoidal transference and algorithm of back propagation for the correction of the weights with aid of the structure Delta bar Delta for the update of the weights. In this mode it resulted satisfactory with the use of the data supplied for the company and with the use of the statistical techniques of bracket was arrived and the techniques of artificial neural networks the modelling of the system
ASSUNTO(S)
controle de qualidade cimento portland redes neurais (computação) simulação (computadores)
ACESSO AO ARTIGO
http://libdigi.unicamp.br/document/?code=vtls000219988Documentos Relacionados
- Predição da intensidade do campo elétrico da onda de superfície utilizando redes neurais artificiais
- Ensaios de ondas ultrassônicas e redes neurais artificiais na avaliação da resistência à compressão do concreto
- Modelagem térmica de um forno panela utilizando redes neurais artificiais
- Um modelo estratégico para a análise de crédito utilizando redes neurais artificiais
- Caracterização de imagens utilizando redes neurais artificiais