Expert air monitoring system (EXAMS): An expert system to use Fourier transform infrared spectrometry for air monitoring in industrial environments.

Abstract

The goal of this research was to develop a prototype expert system (EXAMS) for monitoring workplace air by Fourier transform infrared (FT-IR) spectrometry. FT-IR can provide real-time analysis of complex gas-phase mixtures without the need for time-averaging sample pre-concentration or pre-separation of individual components. The expert system enables a non-spectroscopist, or a spectroscopist with limited occupational health training, to utilize this powerful analytical technique for assessment of health hazards. The expert system operates in MS-Windows$\sp\circler$ 3.1 and includes an occupational hygiene and toxicological evaluation of the air monitoring results. EXAMS currently operates in occupational environments of known atmospheric contaminants, but with the ability to identify the presence of unanticipated compounds. EXAMS is a hybrid blackboard system utilizing three different knowledge-based systems that communicate with each other to accomplish the system goal. The mid-infrared spectrum of pure compounds is analyzed for analytical regions using a hybrid algorithm/heuristic expert system written in Borland C. The selection of optimal analytical regions for sensitivity and selectivity within a sample mixture is based on the compounds present in the environment. The hypertext user interface and database routines were written with Borland Object Vision$\sp\circler$ 2.0. Additional heuristic or rule-based programming for quality assurance and decision support is provided with Information Builder's Level5 Object$\sp\circler$ inference engine. The expert system was evaluated for individual compounds against previously published information using a metric for measuring the health protection value of proposed analytical regions. EXAMS performed as well as or better than the expert spectroscopists in determining analytical regions for these samples. Quantitative evaluation of EXAMS using five known composition mixtures of varied complexity against eight expert spectroscopists produced results in which EXAMS was statistically superior (P $<$ 0.05) to the experts. Implementation of the fully developed EXAMS expert system can dramatically improve the accessibility of FT-IR to non-spectroscopists and increase its range of applicability for spectroscopists by expanding its use to health hazard assessments.