IMPROVEMENT OF OPTIMIZATION ALGORITHM IN ACTIVE GAS ODOR SENSING SYSTEM/

In the field of gas or odor sensing, it is difficult to quantify the c omposition of a mixture of aromas accurately. We proposed earlier the method of active sensing for this purpose and quantification was succe ssfully performed in principle. Here, an optimization algorithm such a s the gradient descent method for quantifying the mixture composition is improved by the introduction of least-squares and singular-value de composition methods so that the stability of the system can be improve d. Better convergence is achieved after the improvement.