Method and system for using a forward calibration model to generate calibrated values for ambient parameters
Embodiments herein generally relate to methods and systems for using a forward calibration model to generate calibrated values for ambient parameters. In at least one example, the method for calibrating a measured un-calibrated ambient parameter (AP) value, includes generating, using an optimization model, a predicted calibrated AP value corresponding to the un-calibrated ambient parameter (AP); inputting, into a trained forward calibration model: (i) the predicted calibrated AP value; and (ii) one or more accuracy-enhancing parameters; generating, using the trained forward calibration model, a predicted un-calibrated AP value; determining if an error difference between the predicted un-calibrated AP value and measured un-calibrated ambient parameter (AP) value, is below a pre-determined threshold; if not, using the optimization model, to generate an updated predicted calibrated AP value and iterating the method, and otherwise outputting the predicted calibrated AP value, as the calibrated AP value.
1. A method performed by a mobile platform comprising:
measuring, by at least one gaseous component concentration sensor, an uncalibrated sensor value (s[k] un-calibrated (actual) ) that indicates a concentration of gas;
receiving, by an analog to digital converter connected to a processor, an analog signal from the gaseous component concentration sensor that represents an uncalibrated sensor value (s[k] un-calibrated (actual) ), which indicates the measured concentration of gas or suspended particulate matter;
detecting proximity to a stationary reference unit that is external to the mobile platform;
determining one or more accuracy-enhancing parameters (p[k]) based on the proximity to the stationary reference unit;
repeating a set of operations until a difference between the s[k] un-calibrated (actual) and a predicted un-calibrated sensor value (s[k] un-calibrated (predicted) ) is below a predetermined threshold, the set of operations comprising:
i) generating, using an optimization model, a predicted calibrated sensor value (s[k] calibrated (predicted) ) based on the s[k] un-calibrated (actual) and the s[k] un-calibrated (predicted) ;
wherein the optimization model utilizes an optimization algorithm selected from the group of dichotomous search, interval halving, Fibonacci search, and golden section search algorithm;
ii) generating the s[k] un-calibrated (predicted) using a trained forward calibration model based on the s[k] calibrated (predicted) and the p[k]; and
transmitting a message to a cloud server that indicates the s[k] calibrated (predicted) as a calibrated sensor value.
2. The method of claim 1 , wherein detecting proximity to the stationary reference unit comprises determining a geographic location of the mobile platform in relation to a geographic location of the stationary reference unit.
3. The method of claim 2 , wherein determining geographic location of the mobile platform in relation to a geographic location of the stationary reference unit comprises:
receiving, from a GPS module connected to the processor, a signal indicating the geographical location of the mobile platform; and
calculating a distance between the geographical location of the mobile platform and the geographic location of the stationary reference unit.
4. The method of claim 1 , wherein the gaseous component concentration sensor is configured to measure the concentration of gas in air, and the gas is at least one of carbon monoxide (CO), carbon dioxide (CO 2 ), nitrous oxide (NO), nitrogen dioxide (NO 2 ), ozone (O 3 ), methane (CH 4 ), or sulfur dioxide (SO 2 ).
5. The method of claim 1 , further comprising retraining the trained forward calibration model.
6. The method of claim 5 , further comprising determining that a trigger condition has occurred, wherein the retraining of the trained forward calibration model is based on the determination that the trigger condition has occurred.
7. The method of claim 6 , wherein the trigger condition comprises at least one of the following: a change of season, a change in weather temperature, proximity to a specified stationary reference unit, proximity to an unrecognized stationary reference unit, replacement of the gaseous component concentration sensor, or elapsing of a specified time.
8. A mobile platform comprising:
a processor;
an analog to digital converter connected to the processor;
at least one gaseous component concentration sensor connected to the analog to digital converter;
memory connected to the processor, wherein the memory comprises instructions that, when read by the processor, cause the processor to:
measure, by the at least one gaseous component concentration sensor, an uncalibrated sensor value (s[k] un-calibrated (actual) ) that indicates a concentration of gas;
receive, via the analog to digital converter, an analog signal from the gaseous component concentration sensor that represents an uncalibrated sensor value (s[k] un-calibrated (actual) ), which indicates the measured concentration of gas or suspended particulate matter;
detect proximity to a stationary reference unit that is external to the mobile platform;
determining one or more accuracy-enhancing parameters (p[k]) based on the proximity to the stationary reference unit;
repeating a set of operations until a difference between the s[k] un-calibrated (actual) and a predicted un-calibrated sensor value (s[k] un-calibrated (predicted) ) is below a predetermined threshold, the set of operations comprising:
i) generating, using an optimization model, a predicted calibrated sensor value (s[k] calibrated (predicted) ) based on the s[k] un-calibrated (actual) and the s[k] un-calibrated (predicted) ;
wherein the optimization model utilizes an optimization algorithm selected from the group of dichotomous search, interval halving, Fibonacci search, and golden section search algorithm; and
ii) generating the s[k] un-calibrated (predicted) using a trained forward calibration model based on the s[k] calibrated (predicted) and the p[k]; and
transmit a message to a cloud server that indicates the s[k] calibrated (predicted) as a calibrated sensor value.
9. The method of claim 8 , wherein detecting proximity to the stationary reference unit comprises determining a geographic location of the mobile platform in relation to a geographic location of the stationary reference unit.
10. The mobile platform of claim 9 , wherein the mobile platform further comprises a GPS module connected to the processor, and wherein determining geographic location of the mobile platform in relation to a geographic location of the stationary reference unit comprises:
receiving, from the GPS module, a signal indicating the geographical location of the mobile platform; and
calculating a distance between the geographical location of the mobile platform and the geographic location of the stationary reference unit.
11. The mobile platform of claim 8 , wherein the gaseous component concentration sensor is configured to measure the concentration of gas in air, and the gas is at least one of carbon monoxide (CO), carbon dioxide (CO 2 ), nitrous oxide (NO), nitrogen dioxide (NO 2 ), ozone (O 3 ), methane (CH 4 ), or sulfur dioxide (SO 2 ).
12. The mobile platform of claim 8 , wherein the memory further comprises instructions that, when read by the processor, cause the processor to retrain the trained forward calibration model.
13. The mobile platform of claim 12 , wherein the memory further comprises instructions that, when read by the processor, cause the processor to determine that a trigger condition has occurred, wherein the retraining of the trained forward calibration model is based on the determination that the trigger condition has occurred.
14. The mobile platform of claim 13 , wherein the trigger condition comprises at least one of the following: a change of season, a change in weather temperature, proximity to a specified stationary reference unit, proximity to an unrecognized stationary reference unit, replacement of the gaseous component concentration sensor, or elapsing of a specified time.
15. A non-transitory computer readable medium that includes instructions that, when read by a processor of a mobile platform, cause the processor to:
cause measurement, by at least one gaseous component concentration sensor, an uncalibrated sensor value (s[k] un-calibrated (actual) ) that indicates a concentration of gas;
receive, by an analog to digital converter connected to the processor, an analog signal from the gaseous component concentration sensor that represents an uncalibrated sensor value (s[k] un-calibrated (actual) ), which indicates the measured concentration of gas or suspended particulate matter;
detect proximity to a stationary reference unit that is external to the mobile platform;
determine one or more accuracy-enhancing parameters (p[k]) based on the proximity to the stationary reference unit;
repeat a set of operations until a difference between the s[k] un-calibrated (actual) and a predicted un-calibrated sensor value (s[k] un-calibrated (predicted) ) is below a predetermined threshold, the set of operations comprising:
i) generating, using an optimization model, a predicted calibrated sensor value (s[k] calibrated (predicted) ) based on the s[k] un-calibrated (actual) and the s[k] un-calibrated (predicted) ;
wherein the optimization model utilizes an optimization algorithm selected from the group of dichotomous search, interval halving, Fibonacci search, and golden section search algorithm;
ii) generating the s[k] un-calibrated (predicted) using a trained forward calibration model based on the s[k] calibrated (predicted) and the p[k]; and
transmit a message to a cloud server that indicates the s[k] calibrated (predicted) as a calibrated sensor value.
16. The computer readable medium of claim 15 , wherein detecting proximity to the stationary reference unit comprises determining a geographic location of the mobile platform in relation to a geographic location of the stationary reference unit.
17. The computer readable medium of claim 16 , wherein determining geographic location of the mobile platform in relation to a geographic location of the stationary reference unit comprises:
receiving, from a GPS module connected to the processor, a signal indicating the geographical location of the mobile platform; and
calculating a distance between the geographical location of the mobile platform and the geographic location of the stationary reference unit.
18. The computer readable medium of claim 15 , wherein the gaseous component concentration sensor is configured to measure the concentration of gas in air, and the gas is at least one of carbon monoxide (CO), carbon dioxide (CO 2 ), nitrous oxide (NO), nitrogen dioxide (NO 2 ), ozone (O 3 ), methane (CH 4 ), or sulfur dioxide (SO 2 ).
19. The computer readable medium of claim 15 , wherein the instructions further cause the processor to retrain the trained forward calibration model.
20. The computer readable medium of claim 19 , wherein the instructions further cause the processor to determine that a trigger condition has occurred, wherein the retraining of the trained forward calibration model is based on the determination that the trigger condition has occurred.