Presentation is loading. Please wait.

Presentation is loading. Please wait.

“Intelligent” nanostructured sensors for a monitoring grid of urban environmental pollutants Authors and affiliations: M. Chiesa 1, M. Paderno 1, G. Gagliotti.

Similar presentations


Presentation on theme: "“Intelligent” nanostructured sensors for a monitoring grid of urban environmental pollutants Authors and affiliations: M. Chiesa 1, M. Paderno 1, G. Gagliotti."— Presentation transcript:

1 “Intelligent” nanostructured sensors for a monitoring grid of urban environmental pollutants Authors and affiliations: M. Chiesa 1, M. Paderno 1, G. Gagliotti 1, L. Schiavina 2,3, P. Borghetti, M. Bertoni, L. Sangaletti *2, A. Goldoni 4, A. Ballarin Denti, ARPA et al. 1 Università Cattolica del Sacro Cuore, CRASL, Italy; 2 Università Cattolica del Sacro Cuore, Dept. of Mathematics and Physics, Italy; 3 EDOR M.Q. S.r.l., Italy; 4 Sincrotrone Trieste S.C.p.A, Italy * sangalet@dmf.bs.unicatt.itsangalet@dmf.bs.unicatt.it

2 Introduction - 1 Aim of the project: a nanostructured sensor to determine (with a sensitivity of a few ppb) the environmental concentration of ammonia (C NH3 ) in Milano Municipality (Italy). The sensor output signal is related to a gas mixture, containing ammonia and some interfering pollutants, nitrogen dioxide (NO 2 ) above all. The sensor will integrate a diffuse grid where NO 2 can be independently monitored. A twofold approach: the sensitivity of the nanostructured sensor with respect to NH 3 measurement was increased by CNTs functionalisation or by addition of clusters of suitable metallic catalysts. Then historical continuous concentration data (over the year 2008) of urban pollutants had been processed by an Expert System, based on neurofuzzy networks and genetic algorithms. The calculated C NH3 values had been extrapolated from the output signal of the gas sensor with a maximum error of 5% with respect to real values (on a ppb scale). 2

3 Materials and methods - 1 3 Carbon NanoTubes (CNT) deposed on plastic supports with two different densities….

4 Materials and methods - 2 4 ….

5 Materials and methods - 3 Main screen of FuzzyWorld, a software which is a framework for a neuro- fuzzy expert system training for environmental data elaboration. The system is a direct tree-structured application composed by antecedents (in red, which are the premises) and a consequent (in blue, i.e. the output). The direct application gives the value of an electrical signal (impedance or tension) that corresponds to the environmental concentration of a gas mixture (NH 3 + O 3 + NO 2 + NO x ). 5

6 6 Materials and methods - 4 Box diagram representing the different components of the direct (on the left) and the inverse function (on the right) developed with FuzzyWorld. The fundamental phase of the process is the selection of the fuzzy rules associated to certain antecedents and consequents.

7 Materials and methods - 5 7 On the right: the window of the inverse function, which has two inputs: the signal of the gases mixture and NO 2 concentration. The expert system selects the fuzzy rules related to those values and gives the ammonia concentration which corresponds to the two given inputs. The calculated output has to be compared to the real value of NH 3 concentration.

8 Results - 1 Calibration curve for ammonia concentrations measured in the range [0 ppm ; 5,2 ppm]. The first imagine is about data collected with Figaro commercial sensor; the second one describes the calibration done for CNT sensor. Calibration test focused on NH 3 environmental concentration range [0-1 ppm] are ongoing. 8

9 Results - 2 Linear correlation between NH 3 real concentration and NH 3 concentration calculated by the neuro-fuzzy expert system realized with FuzzyWorld 9

10 10 Results - 3 All the values of NH 3 concentration elaborated by FuzzyWorld through the inverse function have been reported, counted and related to the corresponding real mean NH 3 concentration.

11 Weekly training of Neurofuzzy Expert Systems 11 Weekly and monthly training compared for direct application (March 2008) Monthly and weekly training results of inverse application (March 2008) to be compared to the first results (last row)

12 Conclusions NH 3 environmental concentration data treatment for detecting ammonia in a gas mixture with fuzzy logic applied to expert systems algorithms had been carried out. The neurofuzzy Expert System developed for the project is composed by a direct application, to obtain the electrical signal corresponding to the gas mixture, knowing the single gas concentrations. Then an inverse application has been built to determine the concentration of NH 3 within the signal, with a further data, i.e. the concentration of NO 2 associated to the inputs explained above. Meteorogical data had been analyzed and coupled to concentration trends to find eventual correlations, which haven’t been found. Acknowledgements: This project is co-funded by the Municipality of Milan in the framework of the PROLIFE project “Nanostructured sensors for diffuse air quality monitoring in Milan”. 12


Download ppt "“Intelligent” nanostructured sensors for a monitoring grid of urban environmental pollutants Authors and affiliations: M. Chiesa 1, M. Paderno 1, G. Gagliotti."

Similar presentations


Ads by Google