Vehicle Mass Estimation with Two-Stage Kalman Filter Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle mass estimation methods, such as extended Kalman filters, are locally convergent, leading to inaccurate or slow convergence of mass estimation when initial assumptions significantly deviate from the true mass, particularly in vehicles with varying payloads like trucks.
Innovation Solution
A computer-implemented method using a non-linear model with two Kalman filter tunings: an aggressive initial tuning to quickly approach the true mass value and a soft tuning for accurate, rapid convergence, adjusting filter parameters like state uncertainties and measurement noise variances to improve mass estimation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single extended Kalman filter with fixed tuning is used for mass estimation, then the filter parameters are simple to manage, but the estimation accuracy is poor when initial assumptions deviate from true mass
Solution Approach 1:
The patent applies dynamics by transitioning from a static single-filter approach to a dynamic two-filter system. The first filter uses aggressive tuning (high process noise variance) for rapid initial convergence, then the system switches to the second filter with conservative tuning (low process noise variance) for refined accuracy. This dynamic adaptation of filter characteristics resolves the contradiction between quick convergence and final accuracy.
Solution Approach 2:
The patent changes filter parameters by implementing two distinct tuning configurations: the first filter uses high process noise variance (Q1) for fast adaptation, while the second filter uses low process noise variance (Q2) for precision. This parameter transformation allows the system to overcome the limitation of fixed-parameter filters and achieve both rapid convergence and high accuracy.
2Speed
If an aggressive filter tuning is used, then the mass estimation converges quickly to the true value, but the estimation fluctuates significantly during convergence
Solution Approach 1:
The patent segments the mass estimation process into two distinct phases: a first phase using an aggressively tuned filter for rapid convergence, and a second phase using a conservatively tuned filter for stable refinement. This segmentation allows the system to achieve both fast convergence and stability by applying different filter characteristics at different stages.
Solution Approach 2:
The patent performs preliminary action by using the first aggressively tuned filter to quickly bring the mass estimation close to the true value before switching to the second filter. This preliminary convergence phase prepares the system for the subsequent stable refinement phase, ensuring both speed and stability are achieved.
3Stability of the object's composition
If a conservative filter tuning is used, then the mass estimation is stable, but the convergence to the true mass value takes a very long time
Solution Approach 1:
The patent applies dynamics by switching from a static conservative filter to a dynamic two-filter system. The system initially uses an aggressively tuned filter to achieve rapid convergence, then transitions to a conservatively tuned filter for stable refinement. This dynamic approach eliminates the time loss associated with purely conservative filtering while maintaining its stability benefits.
Solution Approach 2:
The patent performs preliminary action by using the aggressive filter to quickly reduce the estimation error before switching to the conservative filter. This preliminary convergence phase significantly reduces the time required for the conservative filter to achieve accurate results, resolving the time loss problem.
4Measurement precision
If the filter tuning is switched from aggressive to conservative, then the mass estimation accuracy is improved, but the system complexity increases
Solution Approach 1:
The patent implements feedback by monitoring the convergence state of the mass estimation and using this information to determine when to switch between filters. The switching decision is based on feedback from the estimation process itself, allowing the system to achieve high accuracy through a manageable switching mechanism triggered by convergence criteria.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
The invention relates to a computer implemented method for controlling a vehicle, said control being performed by use of a non-linear model representing dynamics of the vehicle, the non-linear model giving a relationship (1) between a plurality of vehicle state parameters, the vehicle state parameters including the mass (m) of the vehicle and one or more further state parameters (v, cr, cair,α), the method comprising: - obtaining a value (m1) of the mass of the vehicle, - receiving a plurality of time sequential measured first values (v1-v14, α1-α14) of one or more of the further state parameters, - calculating a first plurality of time sequential values (m2-m14) of the vehicle mass, including a first calculated mass value (m14), using the plurality of measured first values (v1-v14, α1-α14) of the one or more further state parameters, the non-linear model, and an extended Kalman filter with a first filter tuning, with the obtained mass value (m1) as a start value, - receiving a plurality of time sequential measured second values (v15-v45, α15-α45) of the one or more of the further state parameters, and - calculating a second plurality of time sequential values (m15-m45) of the vehicle mass, including a second calculated mass value (m45), using the plurality of measured second values (v15-v45, α15-α45) of the one or more further state parameters, the non-linear model, and an extended Kalman filter with a second filter tuning, with the first calculated mass value (m14) as a start value, - wherein the second filter tuning is made less aggressive than the first filter tuning, - the method further comprising using the second calculated mass value (m45) in the control of the vehicle.