MULTIPLE will integrate laser-based sensors for the detection of gases reaching sensitivity up to the MWIR spectral range.

Spectrometric technology for chemometric analysis has gained popularity thanks to its non-destructive nature and the capability to provide highly specific and accurate measure MWIR range are still synonyms of either sensitivity issues or expensive equipment. MULTIPLE will develop different types of chemometrics models including neural network models and analytical models for robust estimation of gas concentration ratios of different compounds, including CO/CO2/O2, NOx, and acetone using laser-based IR spectroscopy. Different types and structures of neural networks will be studied and developed in order to reach the optimum structure for the application.

The final choice of chemometrics model will be based accuracy and precision of results, immunity to cross interferences form other gas constituents in the sample, immunity to temperature and pressure variations and complexity for introduction into serial production. A work will be carried out towards implementation of analyzer auto calibration features as well as self-diagnostics functionality which is crucial for long term autonomous operation.