| The selection of sensor films and sensor signal pattern recognition are two of the decisive factors in gas identification with chemical sensors. This dissertation is particularly focused on the following research topics: optimum sensor materials selection/optimization and sensor array signal recognition.;The influence of nanoparticales interparticle spatial properties and interdigited microelectrodes design parameter on sensor response characteristics were investigated and demonstrated. Different sensor ranking and optimal sensor array searching algorithms were developed targeted to increase the selectivity and diversity of the array.;Sensor signal analysis in this research includes siangl preprocessing and pattern recognition. The signal was normalized and then the signal drift was removed with polynomial fit algorithm. Different pattern recognition techniques were explored. The combination of principle component analysis (PCA) technique, cluster analysis (CA), and Neural Networks (ANN) were demonstrated to provide a rapid and accurate recognition of the tested vapors. The sensor response characteristics for both pure VOCs and binary vapor mixtures were also investigated and modeled.;Finally, a comprehensive software platform which support real time sensing experiment control, signal acquisition, and analysis was developed.;The combined results of our investigations have important implications for the design of intelligent sensing system for portable or remote chemical detection with high sensitivity, high selectivity, quick response time and low detection limit. |
