Information Technology Reference
In-Depth Information
experts, implements Learning and Decision-support System for fishing
ground analysis and prediction.
5.11.1
Problem Analysis and Case Representation
In addition to make assessment on fishery resources (medium and long-term
predictions), the important function of the system is to make the short-term
prediction (the fishing ground centre) . The fishing ground refers to the ocean
areas with intensive fish or other economic marine animals in certain season
that are worth developing and utilizing. The central fishing ground is the
fishing ground with higher average net product and gross total product.
Accurate predictions can directly improve the product and efficiency of the
fishery production and brings great economic benefits.
The migration of fish and formation of central fishing ground have been
influenced by the following factors: temperature of sea water (including
marine surface temperature, the temperature of marine ground floor), data
about platform, such as the salt degree of sea water, salt gradient ion, the flow
rate of Changjiang River, wind directions and velocities, and marine
chlorophyll density. However, the migration of the fish is influenced by a lot
of factors and is too complicated to be described by traditional mathematics
method and model. Furthermore, the knowledge of experts about the fishing
ground is inaccuracy, and incomplete in some sense. Fortunately, we have
already collected the data about fishery conditions in China East Sea for 20
years. These materials are very valuable in that a lot of useful information and
knowledge can be elicited out from them, which can be used to predict the
trend of the fishing ground. The system has been adopted CBR in that CBR is
very suitable in situations where there is little domain knowledge and many
exceptions for the general rules, but only a large number of historical data are
available, domains are described in terms of cases, and the problems are not
totally understood.
According to the fact that information about sea states is usually
collected weekly and for the sake of easy computation, we simplify the
demand according to the actual conditions and make a week as the predication
cycle. In this way, the problem can be formulated as how to predicate the
central fishing ground in next week according to the information of aquatic
products in this week (the position, the output and size). Even so, the problem
