Improvement in signal evaluation methods for semiconductor gas sensors

Pattern Recognition and Multicomponent Analysis

“…In this case the "Chemnet" was designed for Taguchi sensors. * Further applications of neural nets are given in [70] (general treatment, different sensor types), [71] (semiconductor gas sensors), [72] (interdigital capacitors with polymeric coatings), [73] (thick film pellistor array), and [74] (voltammetric measurements in solutions containing nitrophenols). A more complex model consisting of a combination of a self-organizing map (Kohonen type) or a pattern recognition (PR) perception (classification) with an additional perceptron (BPN) to perform the quantification was proposed by Krauss et al [75].…”

“…Further examples for extracting simple parameters in describing transient sensor signal are given in [71], [80] and [81].…”

“…It was shown that the prediction of mixtures could fail if the system was calibrated with pure gases only, since cross interferences were not included in the calibration. Endres et al [71] applied a three-layer artificial neural network (ANN) with an optimized number of hidden neurons to the evaluation of pre-processed data. The pre-processing consisted of a fit of the response behavior and the short-term drift by an exponential decay.…”

Interdigital sensors and transducers

“…On the other hand, a common disadvantage of ANNs is the immense amount of training data needed for the calibration. A new method, the dynamic test point distribution (DTPD), was invented to choose a higher density of test points in the region of interest and a modest density elsewhere [122]. This method achieves a significant reduction in the calibration time, together with a high accuracy in calculating the gas concentration.…”

Complementary Metal Oxide Semiconductors Microelectromechanical Systems Integration

“…Integrated with circuitry for amplification multiplexing, spike detection and the wireless transmission of power and bi-directional data, they are sparking a revolution in environmental monitoring and are facilitating on-chip future sensor devices to combat several environmental disorders like underground coalmining explosions 34 . Due to the lack of selectivity and separability of most common chemical sensors, the use of sensor arrays in combination with artificial neural network (ANN) signal processing is widely used [35][36] . Graf 37 , et al published some reports where MEMS-based gas sensor and necessary signal processing circuits were on the same chip.…”