Development of an integrated sensor to measure odors

• Published in: Environmental monitoring and assessment Vol. 144; no. 1-3; pp. 277 - 283
• Main Authors: Qu, Guoliang, Omotoso, Moremi M, Gamal El-Din, Mohamed, Feddes, John J. R
• Format: Journal Article
• Language: English
• Published: Dordrecht Dordrecht : Springer Netherlands 01.09.2008; Springer Netherlands; Springer; Springer Nature B.V
• Subjects: Accuracy; Air Pollutants - analysis; Ammonia; Ammonia - analysis; Analysis methods; Animals; Applied sciences; AromaScan; Atmospheric pollution; Atmospheric Protection/Air Quality Control/Air Pollution; Concentration; Earth and Environmental Science; Ecology; Ecotoxicology; electronic nose; Environment; Environmental Management; Environmental monitoring; Exact sciences and technology; Expected values; Gas detectors; Humans; Hydrogen; hydrogen sulfide; Hydrogen Sulfide - analysis; Hypotheses; Laboratories; Manures; measurement; Monitoring/Environmental Analysis; Neural networks; Neural Networks (Computer); Nose - physiology; Odorants - analysis; Odors; Pollution; Principal Component Analysis; Principal components analysis; Regression Analysis; Sensors; Smell; Software; Studies; Measurement; Hydrogen sulfide; Ammonia; Electronic nose; AromaScan; Concentration; Odor; Sulfur compounds; Olfactometry; Measurement sensor; Nitrogen compounds; Hydrogen sulfides; Odor, Electronic nose; Measuring instrument; Concentration measurement; Accuracy; Air pollution; Integrated system; Performance
• ISSN: 0167-6369; 1573-2959
• DOI: 10.1007/s10661-007-9991-9

Odorous air samples collected from several sources were presented to an olfactometer, an electronic nose, a hydrogen sulfide (H₂S) detector and an ammonia (NH₃) detector. The olfactometry measurements were used as the expected values while measurements from the other instrumentation values became input variables. Five hypotheses were established to relate the input variables and the expected values. Both linear regression and artificial neural network analyses were used to test the hypotheses. Principal component analysis was utilized to reduce the dimensionality of the electronic nose measurements from 33 to 3 without significant loss of information. The electronic nose or the H₂S detector can individually predict odor concentration measurements with similar accuracy (R ² = 0.46 and 0.50, respectively). Although the NH₃ detector alone has a very poor relationship with odor concentration measurements, combining the H₂S and NH₃ detectors can predict odor concentrations more accurately (R ² = 0.58) than either individual instrument. Data from the integration of the electronic nose, H₂S, and NH₃ detectors produce the best prediction of odor concentrations (R ² = 0.75). With this accuracy, odor concentration measurements can be confidently represented by integrating an electronic nose, and H₂S and NH₃ detectors.