OCR Text |
Show the reference is off or both. To check which one is off, another instrument to validate our system. If two instruments agree, but one is off, then it is highly probable that the one that is off is bad. If we wanted to be very sure, we would compare it once again to another instrument. If three agree and one is off, then we are quite confident that the one that deviates from the other three is in error. A Sensor Validation System uses this same. It compares each sensor to peer information. In fact, it compares not just with one peer, but with several. principle. Figure 2 is a schematic of a typical system. Sensor Validation System Predictiom Raw Values T x Fy ....... P z Data Reco neil iatio Figure 2. Flow dagram fur a Sensor Validation System CASE STUDIES T", • To date, PEMS based on high order, nonlinear regression models have been installed on over 45 processes in the United States. The initial focus has been on developing PEMS for CO and NOx emissions from natural gas, refinery gas and hydrogen fired heaters, furnaces, turbines, engines and boilers. In all instances, the PEMS have been able to pass all required certification tests. In the following paragraphs, comparisons of PEMS versus CEMS for three installations are provided. 9 |