| Because of more and more serious air pollution and more harmful gases resulting in health damage for people in recent years, it is in great need for rapid and convenient gas detection. A single gas sensor has such shortcomings as poor selectivity and crosssensitivity. Arrays containing selective different sensors could improve poor selectivity, and resistance type gas sensor arrays are chosen. By contrasting the characteristics of the experimental data and the difference between algorithms, the artificial neural network algorithm was adopted, because artificial neural network method has the following advantages: simulation of human thinking, nonlinear dynamical system, distributed storage and parallel collaborative information processing, and it is suitable for quantitative test of sample data. So back-end information processing algorithms are studied, the sensor converting circuit is designed, BP algorithm and its Java implement are studied. The main work includes the following:1. Basing on the characteristics of the sensor sensitive membrane and the principle of converting circuit, a converting circuit for interdigital electrode gas sensor arrays was designed. And the results showed that the Circuit can convert the resistance of the gas sensor to voltage, which meet the requirement of measuring accuracy for the sample data.2. After the principle of quantitative detection algorithm is analyzed, algorithm model used for gas sensor array is established. The standard BP algorithm was improved. The results showed that the elastic BP algorithm is more superior to the standard BP algorithm, and it is suitable for processing the sample data of the gas sensor array.3. A program for the standard BP algorithm was designed with Java language, which lays a good foundation for embedded sensor devices. |
PDF Full Text Request