Catalog Home Page

Automatic inferential bias adjustment: Optimisation of an industrial application

Maloney, Craig (2017) Automatic inferential bias adjustment: Optimisation of an industrial application. Honours thesis, Murdoch University.

PDF - Whole Thesis
Embargoed until December 2018.


Not all process variables can be directly measured, some require alternative measured variables together with their known relationship to enable a calculation of an inferred value of the required variable. As these inferred values rely on a fixed calculation a common practice in maintaining their accuracy is by continued automatic bias adjustment as a result of errors determined through laboratory analysis. The focus of the project was to determine the optimal method of bias adjustment of an industrial application of this process within an alumina refinery. Specifically, the critical digestion control variable the Blow Off Ratio (BOR).

The current method of using a cumulative summation (CUSUM) of the errors resulting from the laboratory analysis with a fixed trigger limit to initiate a bias correction was used as a reference enabling average error reductions to be determined by the simulated alternative methods. The standard CUSUM using a multiple of the process standard deviation to establish a dynamic trigger resulted in average error improvements across all five digestion units of the initial testing. Optimising both CUSUM trigger parameters (fixed and dynamic) resulted in minimal average error solution operating outside the definition of a CUSUM.

Simulation of a standard error filter

Publication Type: Thesis (Honours)
Murdoch Affiliation: School of Engineering and Information Technology
Supervisor: Bahri, Parisa
Item Control Page Item Control Page