DECIPHERING SIZE DISTRIBUTION OF AEROSOL PARTICLES BY APPLYING GENETIC ALGORITHM TO THE DATA OBTAINED FROM ELECTRICAL MOBILITY SPECTROMETER

Diffusion charging of sub-micron particles is a stochastic phenomenon. Since there is a probability that particles of the same diameter may obtain different number of elementary charges, it is difficult to estimate size distribution of aerosol particles from analysing Electrical Mobility Spectrometer (EMS) data. This work presents a new approach for obtaining size distribution of captured particles of known charge distribution. In our approach, data signals obtained from EMS, are assumed to be a superposition of the signals resulted from charges on particles belonging to different diameter classes and having different concentrations. A set of linear equations with unknown parameters (unknown concentrations) and constant values can represent this superposition. The constants are derived from the calibration of the EMS instrument, either analytically or experimentally. The unknown parameters are calculated via Genetic Algorithm (GA) because the number of unknowns exceeds the number of equations and because of unavoidable noises in the signals. For the proposed approach, a benchmark analytic replicate of EMS is used as a test. The estimated size distribution agrees well with delivered size distribution.