An adaptive neural networks controller for the inline surface oil/water separator.

摘要

The objective of this thesis is to develop an adaptive Artificial Neural Networks (ANN) controller to control water extraction of the Inline Surface Oil/Water Separator (ISOWS). The thesis contains both a theoretical part and a practical part of the implementation of the ANN controller. The definition of the process problem is briefly discussed to explain the practical application of the ANN controller. An overview of the ANN types and the calculations commonly used for training Feed Forward ANNs are given. More emphasis is shown on the RBF Networks with online training. The practical part shows the detailed implementation blocks of this project consisting of the system used to control water extraction from the ISOWS, which includes water-in-oil detection probes, input processing, ANN controller, anti-windup routine, output processing, and the motor-operated-valves (MOVs), which are controlled using the output parameters. The results of the practical application of the ANN controller to the ISOWS are illustrated using screen captures of the graphical interface of the developed software and a short video for the actual operation and control of the ISOWS.