HR: 0800h
AN: H31B-05    [Abstracts]
TI: Water Diagnosis in Shrimp Aquaculture based on Neural Network
AU: * Carbajal Hernández, J J
EM: carbajalito@hotmail.com
AF: Centro de Investigación en Computación, IPN, Av. Juan de Dios Bátiz s/n casi esq. Miguel Othón de Mendizábal. Unidad Profesional Adolfo López Mateos Col. Nueva Industrial Vallejo, Gustavo A. Madero, DF 07738, Mexico
AU: Sánchez Fernández, L P
EM: lsanchez@cic.ipn.mx
AB: In many countries, the shrimp aquaculture has not advanced computational systems to supervise the artificial habitat of the farms and laboratories. A computational system of this type helps significantly to improve the environmental conditions and to elevate the production and its quality. The main idea of this study is the creation of a system using an artificial neural network (ANN), which can help to recognize patterns of problems and their evolution in shrimp aquaculture, and thus to respond with greater rapidity against the negative effects. Bad control on the shrimp artificial habitat produces organisms with high stress and as consequence losses in their defenses. It generate low nutrition, low reproduction or worse still, they prearrange to acquire lethal diseases. The proposed system helps to control this problem. Environmental variables as pH, temperature, salinity, dissolved oxygen and turbidity have an important effect in the suitable growth of the shrimps and influence in their health. However, the exact mathematical model of this relationship is unspecified; an ANN is useful for establishing a relationship between these variables and to classify a status that describes a problem into the farm. The data classification is made to recognize and to quantify two states within the pool: a) Normal: Everything is well. b) Risk: One, some or all environmental variables are outside of the allowed interval, which generates problems. The neural network will have to recognize the state and to quantify it, in others words, how normal or risky it is, which allows finding trend of the water quality. A study was developed for designing a software tool that allows recognizing the status of the water quality and control problems for the environment into the pond.
UR: http:www.cic.ipn.mx
DE: 0496 Water quality
DE: 0555 Neural networks, fuzzy logic, machine learning
DE: 1880 Water management (6334)
SC: Hydrology [H]
MN: 2007 Joint Assembly