Artificial Neural Networks for Nonlinear Control of Industrial Processes

摘要

In this paper, the state-of-the-art methodology for model identification and control of fluid injection into petroleum reservoir is presented. The methodology uses several techniques: Dynamic Neural Network Control (DNNC) network structure, neuro-statistical (neural network, conventional statistical, and non-parametric statistical such as ACE; Alternative Conditional Expectation) techniques, Geometric Model-Based Control Strategy, and stability analysis techniques such as Liapunov theory. In this study, the DNNC model is used because it is much easier to update and adapt the network on-line. The neuro-statistical technique is used because the model can deal with uncertainty in data. The neuro-statistical model uses neural network techniques, since the functional structure of the data is unknown and uses statistical technique because the data and our requirements are imperfect. In addition, this technique in conjunction with Levenberge-Marquardt algorithm can be used as a more robust technique for network training and optimization purposes. The ACE technique is used for scaling the network's input-output data and can be used to find the input structure of the network. The result from Liapunov theory is used to find optimal neural network structure. In addition, a special neural network structure is used to insure the stability of the network for ling-term prediction. In this model, the current information from the input layer is presented into a pseudo hidden layer. This model minimizes not only the conventional error in the output layer but also minimizes the filtered value of the output. This technique is a tradeoff between the accuracy of the actual and filtered prediction which will result in the stability of the long-term prediction of the network model. In this paper, an optimal neural network controller will be designed for a well-head controller for steam injection into low permeability petroleum reservoirs for oil displacement.