finally the training performances of various algorithms are compared based on a simulation experiment on a benchmark problem of neural network learning, furthermore, a viewpoint that genetic algorithm is subject to "curse of dimension" is proposed.
The method is discussed of analyzing the inflow performance of oil wells by artificial neural network technique. In this method, an oil well is considered as a black box-like nonlinear dynamic system, so there is no need to build up a complicated mathematical model describing the dynamic of the oil well. The corresponding neural network prediction model can be built up by means of network learning of input/output of the dynamic system.
In order to take the advantage of network learning system, this thesis begins with a discussion on the modern learning theory of constructivism, analyses the functions and features of learning system and then discusses how to stimulate learners' initiative by making use of the features of network.
Based on the new generation Web technique environment, the main aims in this thesis is to use the Association Mining method to design the network learning system satisfying the characteristic requests .