Soft-sensors and numerical filters in mineral processing

- Organization:
- Canadian Institute of Mining, Metallurgy and Petroleum
- Pages:
- 2
- File Size:
- 1602 KB
- Publication Date:
- Jan 1, 1992
Abstract
"Soft-sensors are a computer software alternative to inadequate hardware instrumentation. Filters are averaging methods that reduce the noise in any sensor's signal. This paper introduces these topics and places them in a mineral processing context using then-product formula.Soft SensorsAs their name implies, ""soft-sensors"" are computer programs that estimate values for process variables (Tham, 1991). This approach is useful when the appropriate hardware sensor does not exist, is deficient, supplies only infrequent values, etc. For example, in mineral concentrators the traditional magnetic flowmeter/ nuclear density gauge instrumentation for measuring slurry flows costs $5,000 or more per flow and requires frequent maintenance. However, metal assays are commonly available from online X-ray fluorescence analyzers and application of the n-product formula (Taggart, 1945) to these assays conveniently provides software estimates of the flowrates (albeit only when the circuit is in steady-state).Soft-sensors are used by inferential controllers that base their control actions on the process values ""inferred"" by the soft-sensor. Such was the case at Brenda Mines where the rod mill feed tonnage controller used an inferred value of the cyclone overflow particle size (Bradburn, 1976). It was also the case at Brunswick Mining where reagent controllers adjusted reagent addition rates throughout the plant according to the various ore flowrates inferred by an expert system (Spring, 1991). A unique feature of the expert system soft-sensor was its dynamic flow models which compensated for residence times, material transport lags, etc. Thus the soft-sensor values were valid under both steady-state and nonsteady- state conditions."
Citation
APA:
(1992) Soft-sensors and numerical filters in mineral processingMLA: Soft-sensors and numerical filters in mineral processing. Canadian Institute of Mining, Metallurgy and Petroleum, 1992.