Akaninyene J Ntuk’s scientific contributions

Almost every operational physical system in engineering is designed with a concept of mathematical principle which has theories backing them, thus making these systems theory-driven. On the contrary, these systems operate based on parameters which are seen as data. With advanced technique of data acquisition and data processing, these systems can be modeled using the acquired data and parameters thus converting the theory-driven models to data-driven models. In this paper, an automated electrically operated dehydration bed is modelled and analysed using the automatic generation of expression approach. Analysis confirmed a high level of excitation in the input data sets. The excitation orders for all the inputs are [50 50 50] meaning that model with orders higher than 50 will be problematic to estimate.

This paper deals extensively on Fault Detection and Isolation (FDI) techniques, using a three tank system with two inputs, three measured outputs and three States. And describes how the Fault detection and Isolation was carried out on each of the datasets provided. This report also describes how the simulation was developed to confirm the fault earlier detected and isolated. Finally the effects of complete failure of the sensor L3 and the ways of mitigating the adverse effects were carried out. KEYWORDS –Fault detection,Fault isolation, Datasets, Components failure

The operation and management of dehydration systems for the determination of liquid and gas dew points is becoming increasingly complex. Several methods for prediction of the liquid and gas dew points have been developed by researchers. These methods differ in the available inputs, their classification and the horizon of the prediction. In this paper, a nonlinear auto-Regressive with eXogenous (NARX) input model has been proposed. The data driven modelling approach was adopted in the proposed model with System Identification toolbox in MATLAB software. Implementation and simulation of the model which was achieved by feeding the measured input data into the developed model shows that the model is able to reproduce the dew-point (measured output) of the dehydrator bed, and hence gives a good prediction of it.