Gas Sensor Stabilization via Machine Learning Drift Compensation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Ultrasonic mass-based gas sensors are vulnerable to noise, particularly drift, which can lead to inaccurate readings, as they are sensitive to environmental changes and rapid gas concentration fluctuations, and current methods to compensate for noise sources are inadequate for all types of noise.
Innovation Solution
A system and method that utilize a processor, memory, and machine learning engine to stabilize sensor readings by filtering out noise through environmental parameter measurements and modeling drift as a polynomial expression, reducing the impact of residual errors on output values, and compensating for drift by identifying and subtracting residual offsets using linear models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ultrasonic mass-based gas sensors are used to measure gas concentration, then sensitivity to gas changes is improved, but vulnerability to noise and drift increases
Solution Approach 1:
The patent introduces control measurements taken at time points before and after the main measurement as intermediary data. These control measurements serve as mediators to characterize and quantify the drift and noise conditions, which are then used by the machine learning engine to compensate for the harmful effects on the main measurement.
Solution Approach 2:
The system implements a feedback mechanism where control measurements are continuously taken and used to update the machine learning model's understanding of drift conditions. The model uses this feedback to dynamically adjust and compensate for noise and drift in real-time, improving measurement accuracy while maintaining sensitivity.
2Measurement precision
If machine learning with control parameters is used to compensate for noise, then measurement accuracy is improved, but processing power requirements increase
Solution Approach 1:
Instead of using complex comprehensive models, the patent applies partial action by using a machine learning engine that processes only the essential control parameters (measurements at before and after time points) to compensate for drift. This selective approach achieves sufficient accuracy improvement without requiring excessive processing power, making it suitable for portable devices.
3Stability of the object's composition
If drift compensation methods are applied to sensor readings, then reading stability is improved, but model complexity increases
Solution Approach 1:
The patent stabilizes sensor readings by changing the parameter representation of drift - instead of using complex time-varying models, it transforms drift into a polynomial expression based on control measurements. This parameter transformation simplifies the model while maintaining reading stability, as the polynomial form is computationally efficient and easier to implement.
Data Source
AI summary
A system for stabilizing sensor readings. The system includes a processor; a memory communicatively coupled to the processor; a receiver to receive from a sensing device, a measurement and an environmental parameter. The system also includes a machine learning engine executed on the processor wherein the machine learning engine receives, as inputs: the environmental parameters; the measurement; and control parameters calculated based on control measurements made at time points before and after the measurement.