Calibration and Validation of Process Sensors
In 1989, Harald Martens and Tormod Næ´s published their now seminal book, Multivariate Calibration (John Wiley & Sons Ltd, 1989). Although the book’s primary focus is on what we now refer to as chemometric modeling — “soft modeling” to be more precise — they comment on an aspect of modeling that is at the present time very relevant to the focus of this chapter. To quote:
The same holds true for process sensors. It will be shown that, through transduction and transmission of signal into impulse, or through calculation into a determined value, the flow of process data into process information, which then may be transformed into process knowledge, can be turned into useful process understanding only if (a) the knowledge behind the sensor systems intended application can be shown to be valid; and (b) the quality of the data is accurately and precisely verifiable when that information is needed and is based on appropriate measuring reference systems. A further condition that is necessary is that the system must be adjustable (preferably in realor near-real time)
DBURNS: “7393_c012” — 2007/7/27 — 17:12 — page 246 — #2
and intensity, as information is fed back to the Supervisory Control and Data Acquisition (SCADA) system to induce a change to the process variable being affected.