Adaptively Weighted Kalman Filter for Speed Estimation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing vibration signal processing systems fail to accurately estimate rotation speed, especially when the signal-to-noise ratio drops due to non-stationary noises and transient changes in operational conditions, leading to erroneous speed estimations and spurious results.

Innovation Solution

An adaptively weighted Kalman filter is used to minimize incorrect speed estimations by modifying the adaptive weight based on a test statistic, such as Mahalanobis distance, to enforce continuity in time domain and reject spurious speed estimations, thereby improving the accuracy of rotation speed estimation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Object-affected harmful factors

If conventional filters are used to attenuate noises, then noise reduction is achieved, but non-stationary noises cannot be effectively attenuated and speed estimation accuracy deteriorates under transient changes

Engineering Contradiction:
Improvenoise attenuationVSAvoidspeed estimation accuracy
Core Design Contradiction:
Object-affected harmful factorsVSMeasurement precision

Solution Approach 1:

The patent applies dynamics by making the filter adaptive rather than static. The Kalman filter continuously adjusts its parameters (process noise covariance Q, measurement noise covariance R) based on current system conditions and estimation errors. This dynamic adaptation allows the filter to effectively handle non-stationary noises and transient changes while maintaining speed estimation accuracy, resolving the contradiction between noise attenuation and measurement precision.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent implements feedback through the Kalman filter's recursive structure that uses estimation residuals (difference between measured and predicted speeds) to continuously update state estimates and adjust filter parameters. This feedback mechanism enables the system to learn from past errors and adapt to changing noise characteristics, achieving both effective noise attenuation and maintained estimation accuracy under varying operational conditions.

Inventive Principle:
Principle #23Feedback

2Productivity

If standard Kalman filter is used for speed estimation, then continuous estimation is provided, but spurious speed estimations occur during transient operational changes

Engineering Contradiction:
Improvecontinuous speed estimationVSAvoidestimation correctness
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent uses feedback through residual analysis and adaptive parameter adjustment. When transient changes occur, the filter detects increased estimation residuals and responds by adjusting the process noise covariance Q or measurement noise covariance R to reduce the influence of spurious measurements. This feedback-based adaptation maintains continuous estimation while filtering out incorrect speed estimates during transient operational changes.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent applies parameter changes by dynamically modifying the Kalman filter's noise covariance parameters (Q and R) based on operational conditions and estimation quality indicators. During transient changes, the filter increases process noise covariance to allow faster adaptation or increases measurement noise covariance to discount potentially spurious measurements. This dynamic parameter adjustment enables the system to maintain reliable continuous estimation across varying operational conditions.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3196657B1Speed estimation systems
Publication Date: 2019.12.11 SIMMONDS PRECISION PRODUCTS INC
  • EP3196657B1 patent drawingFigure 1~2A
  • EP3196657B1 patent drawingFigure 2B~2C
  • EP3196657B1 patent drawing

AI summary

A method for estimating rotational speed of a system includes receiving vibrational data from a sensor (101), estimating a speed from the vibrational data to create estimated speed data, and filtering the estimated speed data through an adaptively weighted filter (105) to minimize incorrect speed estimation. The filter can be a Kalman filter, having a Kalman gain which can be modified taking into consideration a deviation between the predicted and estimated speed. In this way, the negative effect of outliers on the speed estimation can be reduced.