Chemo-Resistive Gas Sensing With Adaptive Drift Correction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Chemo-resistive gas sensors experience drift over time, leading to inaccurate gas concentration estimation, which existing methods struggle to address effectively.

Innovation Solution

A gas sensing device employs a drift correction processor with a mapping function and a second trained model based algorithm processor to adaptively correct for sensor drift, utilizing reinforcement learning to select the best update for the mapping function and improve long-term stability.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If chemo-resistive gas sensors are used for gas concentration sensing, then gas concentration can be determined by measuring electrical properties, but the sensor outputs drift over time making estimation harder and eventually failing completely

Engineering Contradiction:
Improvegas concentration estimation accuracyVSAvoidlong-term operational stability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The system continuously monitors sensor outputs and uses machine learning models to detect drift patterns, applying corrective transformations in real-time. The feedback loop compares expected vs. actual sensor responses and adjusts the mapping function accordingly, maintaining measurement accuracy despite drift over time.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent transforms sensor outputs using learned mapping functions that dynamically adjust parameters based on detected drift. By changing the transformation parameters rather than the physical sensor, the system compensates for drift while maintaining the original sensing capability.

Inventive Principle:
Principle #35Parameter changes

2Reliability

If drift correction methods are applied to maintain measurement accuracy, then gas concentration estimation remains reliable, but the device complexity increases with additional processors and algorithms

Engineering Contradiction:
Improvedrift-corrected measurement reliabilityVSAvoidprocessing system complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system uses the sensor's own drifted outputs to train and update the correction models, eliminating the need for external calibration equipment or reference sensors. The sensor essentially calibrates itself by learning its own drift patterns from historical data.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces complex mechanical or chemical drift compensation mechanisms with software-based machine learning models. Instead of physically adjusting or replacing sensors, the system uses algorithmic transformations to correct drift, reducing mechanical complexity.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Adaptability or versatility

If machine learning models are trained on drifted data to adapt to long-term changes, then the system can maintain accuracy, but training on drifted data is challenging and requires sophisticated algorithms

Engineering Contradiction:
Improveadaptation to drifted conditionsVSAvoidtraining on drifted data difficulty
Core Design Contradiction:
Adaptability or versatilityVSDifficulty of detecting and measuring

Solution Approach 1:

The system performs preliminary training on clean, non-drifted data first to establish baseline performance. This preliminary action creates a reference model that can later be fine-tuned or adapted as drift occurs, making the overall adaptation process more manageable and effective.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent implements dynamic model updating where the machine learning models continuously adapt to drifting conditions rather than being static. The system transitions from static training to dynamic online learning, allowing the models to evolve with the sensor's drift characteristics.

Inventive Principle:
Principle #15Dynamics

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

The device provides improved long-term stability and accuracy in gas concentration estimation by effectively addressing sensor drift, even in real-world conditions where training on drifted data is challenging.

Implementation Method 1

Chemo-resistive gas sensors change their electrical properties in the presence of gases such as nitrogen dioxide (NO2), ozone (O3), ammonia (NH3) or carbon monoxide (CO) so that a concentration of such gases in a mixture of gases may be determined by measuring the electrical properties of one or more chemo-resistive gas sensors

Methodology Applied
Scientific EffectChemo-resistive effect: Electrical Resistance

Implementation Method 2

the interaction between graphene sheets and absorbed gas analytes influences the electronic structure of the material depending on the mixture of gases, resulting in altered charge carrier concentration and changed electrical conductance

Methodology Applied
Scientific EffectAdsorption: Adsorption

Implementation Method 3

the interaction between graphene sheets and absorbed gas analytes influences the electronic structure of the material depending on the mixture of gases, resulting in altered charge carrier concentration and changed electrical conductance

Methodology Applied
Scientific EffectElectrical conductance change: Conduction (electrical)

Data Source

PatentEP4414699B1Gas sensing device and method for operating a gas sensing device
Publication Date: 2025.09.24 INFINEON TECHNOLOGIES AG
  • EP4414699B1 patent drawingFigure 1
  • EP4414699B1 patent drawingFigure 2
  • EP4414699B1 patent drawingFigure 3

AI summary

A gas sensing device for sensing one or more gases in a mixture of gases is provided. The gas sensing device (1) comprises: one or more chemo-resistive gas sensors (2); a preprocessing processor (3) configured for generating a preprocessed signal sample (PSS) for each of the signal samples (SIG); a drift correction processor (4) configured for applying a mapping function (MPF) to each of the preprocessed signal samples (PSS) in order to create a drift-corrected signal (DCS) sample for each of the preprocessed signal samples (PSS); a feature extraction processor (5) configured for extracting a set of feature values (FV) from each of the drift-corrected signal samples (DCS); a sensing result processor (6) configured for creating for each of the gases a sensing result (SR) for each of the sets of feature values (FV); a state processor (7) configured for determining a state (STA) of the gas sensing device (1); and an update processor (8) configured for providing updates (UPD) to the drift correction processor (4), wherein a quality value (QV) for each possible update action of a plurality of update actions is provided, wherein an update action processor (10) is configured for calculating the update (UPD) for the subsequent update step by applying the possible update action, which has a highest quality value of the quality values (QV) of the current update step, to the mapping function (MPF) of the current update step.