Inertial Sensor Noise Estimation Using Allan Variance Tangents
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Inertial navigation systems based on gyroscopes and accelerometers suffer from systematic errors leading to drift in position indication, making them unreliable without periodic calibration, especially with low-cost sensors, and existing noise analysis methods like Allan Variance do not effectively improve measurement accuracy.
Innovation Solution
A measuring device using at least two sensors with a noise extraction unit that generates a weighted difference signal, independent of the vectorial physical quantity, and a low-frequency noise estimator to estimate and subtract correlated noise, improving measurement accuracy by reducing noise contributions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If inertial sensors are used for navigation, then position and velocity can be estimated through numerical integration, but systematic errors cause drift in position indication making the system unreliable without periodic calibration
Solution Approach 1:
The patent extracts and separates the noise component from the sensor signal by using multiple sensors and statistical analysis. The noise extraction unit identifies and isolates the correlated noise portion from the total signal, allowing it to be removed independently from the useful measurement data.
Solution Approach 2:
The patent implements a feedback mechanism where the measured noise characteristics are continuously used to adjust and compensate the sensor readings. The low-frequency noise estimator provides feedback about the noise level, which is then used to correct the position and velocity calculations in real-time.
2Ease of manufacture
If low-cost sensors are used, then device cost is reduced, but measurement accuracy deteriorates rapidly within minutes without bias compensation
Solution Approach 1:
The patent enables the navigation system to self-correct its measurements by automatically detecting and compensating for its own noise characteristics. The system uses its multiple sensors to monitor its performance and adjust its calculations without external intervention, maintaining accuracy throughout the navigation process.
Solution Approach 2:
The patent changes the parameters used in navigation calculations dynamically based on the measured noise characteristics. By adjusting the integration weights and compensation factors according to the actual sensor performance, the system optimizes accuracy for the specific sensor quality being used.
3Measurement precision
If Allan Variance analysis is used to determine sensor stability, then noise characteristics can be analyzed, but the method requires long integration times reaching the Allan minimum time which limits productivity
Solution Approach 1:
The patent performs preliminary noise characterization during the manufacturing or initialization phase by analyzing multiple sensors simultaneously. This pre-characterization allows the system to establish noise models in advance, eliminating the need for lengthy field calibration procedures and enabling faster deployment.
Solution Approach 2:
The patent combines the outputs of multiple sensors to improve the signal-to-noise ratio during calibration. By processing several sensors together, the system achieves more accurate noise characterization in shorter time periods compared to analyzing single sensors independently.
Data Source
AI summary
A measuring device includes a pair of sensors, an actuator, a noise extraction unit and a low frequency noise (LFN) estimator. The sensors each generate with sample time Ts a sense signal indicating a value of a component of a vectorial physical quantity. The sensors have an Allan variance curve with a minimum value for a first integration time T1. The curve has a first and second tangent line being tangent at integration time 0, and integration time T1 respectively. The tangent lines intersect each other at an intersection point for a second integration time T2. The estimator having an effective integration time Teff determined by, Ts, T1 and T2, generates an estimated noise signal indicative for the estimated value of the noise component from the difference signal and from information about the relative rotation between the sensors.