Realtime Hydrocarbon Measurement using Software Sensor

The aim of the project is to carry out an in-depth investigation of the current technology with a view to developing and integrating a robust and cost-efficient clustering and classification algorithm into the Interpret software package with the goal of enhancing the predictive execution of software package.

The project will study the different types of clustering algorithms in addition to sample regression/prediction and generate models using MATLAB.

With the clustering and classification algorithms, the project will present a thorough investigation of the structure of the fundamental data and how the model training sets can be progressively picked up while obtaining the best data segregation.

The use of Partial Least Squares and Chemometric Mixing Rules will be reviewed and assessed in conjunction with the chosen clustering algorithm.

The project will involve a detailed assessment previous work done on clustering algorithms with a view to cricising, identifying and developing an algorithm well suited for spectra data.

In conclusion, the aim of the project is to improve on the realtime prediction software package currently in use by the industry sponsor to provide crude blending and data analytics services to their customers.