Overview
The research introduces a novel adaptive longitudinal platooning method designed for the correct online estimation of powertrain time constants in vehicles. This method addresses challenges associated with existing estimation techniques that either require persistent excitation or a priori knowledge of the input matrix in platooning dynamics. The proposed approach modifies the composite adaptive control framework to specifically handle time constant uncertainty, ensuring the convergence of estimated values to their true counterparts.
Research Context
In longitudinal platooning, powertrain time constants represent a significant source of uncertainty. These time constants influence the input matrix of the platooning dynamics. Current estimation methods for such parameters present inherent difficulties. Methods that rely on persistence of excitation can be impractical in real-world platooning scenarios, where excitation conditions might not always be met. Alternatively, methods requiring the input matrix to be known beforehand are not viable when the time constants, which are part of this matrix, are unknown and need to be estimated.
Approach
The proposed method employs a modification of the composite adaptive control framework. This modification specifically targets the incorporation of time constant uncertainty into the design of the adaptive law. The objective of this adaptation is to facilitate the correct estimation of powertrain time constants during online operation. A key aspect of this approach is its ability to guarantee the convergence of the estimated time constants to their true values without the necessity of persistence of excitation. Instead, the method requires only that the derivative of the acceleration is non-zero over a potentially short transient period. This condition is characterized as an 'extremely relaxed excitation condition'.
The design of the platooning protocol resulting from this approach offers advantages over state-of-the-art platooning solutions. It eliminates the need for measurements of the acceleration derivative and does not require the collection of past data. The robustness and practicality of the proposed design were verified through CarSim-based platooning experiments.
Findings
- The proposed adaptive longitudinal platooning method correctly estimates powertrain time constants.
- The method modifies the composite adaptive control framework to handle time constant uncertainty in the adaptive law design.
- Convergence of estimated time constants to their true values is guaranteed.
- Persistence of excitation is not required for estimation; a non-zero derivative of acceleration over a short transient is sufficient.
- The method does not require measurements of acceleration derivative.
- The method does not require the collection of past data.
- The robustness and practicality of the design were verified using CarSim-based platooning experiments.
Why This Matters
Correct estimation of powertrain time constants is crucial for effective longitudinal platooning. The method's ability to achieve this without demanding persistence of excitation or prior knowledge of the input matrix could enhance the practical deployment of adaptive platooning systems. The reduced requirements for acceleration derivative measurements and past data simplify implementation, addressing challenges encountered by existing platooning solutions.