Characterizing liquids based on features extracted from time-dependent, differential signal measurements

Patent metadata

Field	Value
Publication number	US-12117802-B2
Application number	US-202117645383-A
Country	US
Kind code	B2
Filing date	Dec 21, 2021
Priority date	Dec 21, 2021
Publication date	Oct 15, 2024
Grant date	Oct 15, 2024

Abstract

One or more computer processors obtain one or more time-dependent signals with one or more sensor pairs in a sensing system, respectively, wherein each of the one or more time-dependent signals are obtained as a differential signal of a respective pair of the one or more sensor pairs by successively sensing a reference liquid and each liquid in a set of liquids to be characterized with the respective pair; extracting one or more sets of features from one or more portions of the one or more time-dependent signals, respectively, each of the one or more portions including a signal portion obtained while sensing each liquid in the set of liquids with said respective pair; and characterize each liquid in the set of liquids based on the one or more extracted sets of features.

First claim

What is claimed is:

1. A computer-implemented method comprising:
obtaining, by one or more computer processors, one or more time-dependent signals with one or more sensor pairs in a sensing system, respectively, wherein each of the one or more time-dependent signals are obtained as a differential signal of a respective pair of the one or more sensor pairs by successively sensing a reference liquid and each liquid in a set of liquids to be characterized with the respective pair, wherein the reference liquid is selected to be a center of a plurality of steady-state values associated with each liquid in the set of liquids;
extracting, by one or more computer processors, one or more sets of features from one or more portions of the one or more time-dependent signals, respectively, each of the one or more portions including a signal portion obtained while sensing each liquid in the set of liquids with said respective pair, wherein the one or more sets of features include an average of signal portions with respect to the reference liquid; and
characterizing, by one or more computer processors, each liquid in the set of liquids based on the one or more extracted sets of features.

2. The computer-implemented method of claim 1 , wherein extracting the one or more sets of features for each liquid in the set of liquids, further comprising:
the signal portion of each of the one or more time-dependent signals includes a transient signal response obtained due to a transition from sensing the reference liquid to sensing each liquid in the set of liquids with the respective pair of the one or more sensor pairs; and each of the one or more sets of features extracted includes at least one transient feature.

3. The computer-implemented method of claim 2 , wherein extracting the one or more sets of features for each liquid in the set of liquids, further comprising:
the signal portion of each of the one or more time-dependent signals further includes a steady-state signal response obtained at an end of the transient signal response; and each of the one or more sets of features extracted further includes at least one steady-state feature.

4. The computer-implemented method of claim 1 , wherein:
each of the one or more time-dependent signals is obtained by further sensing the reference liquid again after having successively sensed the reference liquid and each liquid in the set of liquids with said respective pair;
the signal portion of each of the one or more time-dependent signals includes a further transient signal response obtained due to a transition from sensing each liquid in the set of liquids to sensing said reference liquid again with the respective pair of the one or more sensor pairs; and
each of the one or more extracted sets of features includes at least one further transient feature of the further transient signal response.

5. The computer-implemented method of claim 2 , wherein the sensors are potentiometric sensors and each of the one or more time-dependent signals is obtained as a differential, potentiometric signal.

6. The computer-implemented method of claim 5 , wherein:
the one or more extracted sets of features from each of the one or more portions for each liquid in the set of liquids include two features, the latter consisting of:
a feature obtained from a maximum voltage variation in the transient signal response, with respect to a reference value obtained by sensing the reference liquid with said respective pair; and
a feature obtained from a slope of the transient signal response.

7. The computer-implemented method of claim 6 , wherein:
at extracting the one or more sets of features for each liquid in the set of liquids, the signal portion of each of the one or more time-dependent signals further includes a steady-state signal response obtained at an end of the transient signal response and each extracted set of features further includes, for each liquid in the set of liquids and for each of the one or more sets, three features, the latter respectively obtained from:
a final absolute voltage value of the steady-state signal response;
a final relative voltage value of the steady-state signal response; and
an average of a complete signal response with respect to said reference value, the complete signal response including the transient signal response and the steady-state signal response.

8. The computer-implemented method of claim 5 , wherein the sensors are designed to electrochemically interact with each liquid in the set of liquids.

9. The computer-implemented method of claim 5 , wherein the sensing system includes an array of the sensors, the array designed so as to allow each liquid in the set of liquids to be simultaneously sensed by the one or more sensor pairs and the one or more time-dependent signals are simultaneously obtained for each liquid in the set of liquids, by simultaneously sensing each liquid in the set of liquids with the one or more sensor pairs.

10. The computer-implemented method of claim 1 , further comprising:
selecting, by one or more computer processors, the reference liquid for it to be intermediate between each liquid in the set of liquids, with respect to one or more properties.

11. The computer-implemented method of claim 10 , wherein the one or more properties includes one or more voltage signal response values of the signal responses obtained with one or more of the one or more sensor pairs and the one or more voltage signal response values include one or more of a steady-state voltage signal response value, an average voltage signal response value, and a maximal voltage signal response value.

12. The computer-implemented method of claim 10 , further comprising:
prior to selecting the reference liquid, obtaining, by one or more computer processors, one or more signal responses for each liquid in the set of liquids, including the reference liquid, which is not identified as such yet, such that the reference liquid can be selected based on the one or more signal responses obtained.

13. The computer-implemented method of claim 1 , wherein the one or more pairs of sensors are designed such that each of the resulting one or more time-dependent signals is linearly independent of remaining time-dependent signals.

14. The computer-implemented method of claim 1 , wherein each liquid in the set of liquids is characterized so as to classify each liquid.

15. The computer-implemented method of claim 1 , wherein each liquid in the set of liquids is characterized so as to quantify one or more properties thereof.

16. The computer-implemented method of claim 15 , wherein each liquid in the set of liquids is an aqueous mixture of ions and is characterized so as to quantify concentrations of one or more ions therein.

17. The computer-implemented method of claim 1 , wherein each liquid in the set of liquids is characterized using a cognitive model trained based on labelled examples by feeding the one or more sets of features obtained for each liquid in the set of liquids to the trained model for it to produce an inference.

18. The computer-implemented method of claim 17 , wherein the cognitive model includes one or more regression models.

19. The computer-implemented method of claim 18 , wherein the cognitive model includes both a linear regression model and a nonlinear regression model.

20. A system comprising:
a liquid storage including liquid containers adapted for storing respective liquids including a reference liquid and a set of liquids to be characterized;
a sensing system

Assignees

• IBM

Inventors

• Ruch Patrick
• Gabrieli Gianmarco
• Hu Rui

Classifications

• G16C20/70

  Machine learning, data mining or chemometrics · CPC title

• G05B2219/40585

  Chemical, biological sensors · CPC title

• G16C20/20

  Identification of molecular entities, parts thereof or of chemical compositions · CPC title

• G01N27/27

  Association of two or more measuring systems or cells, each measuring a different parameter, where the measurement results may be either used independently, the systems or cells being physically associated, or combined to produce a value for a further parameter · CPC title

• G05B19/4155 Primary

  characterised by program execution, i.e. part program or machine function execution, e.g. selection of a program · CPC title