A hybrid control strategy for oil separators based on electrical capacitance tomography images

A knowledge-based hybrid control strategy based on principal component analysis and cluster analysis has been developed for oil separators, which use electrical capacitance tomography (ECT) images. ECT is considered as soft-field tomography and it requires complicated image reconstruction due to the non-linear relationship between the capacitance and the permittivity distribution. PCA is used for generating precise prior information using a model-based approach and then cluster analysis is used to find the closest stored condition to the current. When an ECT image presenting the current process situation of oil separator is generated, again percentages of water, oil and air are calculated using the method of image interpretation. The hybrid control approach provides an intelligent control system with capability for making decisions within some pre-defined range.