TL;DR
This paper introduces a novel "switchable" Model Predictive Control (MPC) strategy for electric vehicle regenerative braking. By dividing the braking system into three distinct operational modes and employing explicit MPC, the system achieves more precise control over brake pressure and piston displacement, leading to improved energy regeneration efficiency and a more consistent, natural pedal feel across various braking conditions.
What Problem Does This Paper Address?
Electric vehicles (EVs) are increasingly adopting electric braking boosters (E-Boosters) to enhance regenerative braking efficiency and provide a better driver experience. However, coordinating the hydraulic braking system (including the E-Booster and electronic stability control) with the electric motor's regenerative braking presents a significant challenge. Existing methods often struggle to simultaneously achieve precise control of brake pressure and piston displacement, maintain a consistent and natural "pedal feel" for the driver, and adapt effectively to different braking scenarios (like normal braking, regenerative braking, and situations with hydraulic "dead zones"). This paper aims to bridge this gap by developing a control strategy that can seamlessly manage these objectives across various operating conditions.
Methods
The core of the proposed solution is a switchable Model Predictive Control (MPC) approach. Here's a breakdown of the key methods:
- Flow Regulation Model: A new flow model is developed as the foundation for controlling the hydraulic system, focusing on the interplay between the E-Booster, master cylinder, and wheel cylinders.
- System Partitioning and State Machine: The complex electro-hydraulic system is divided into three distinct operational modes: Fully Hydraulic Braking (FHB), Distributed Electro-Hydraulic Braking (DEHB), and Coordinated Electro-Hydraulic Braking (CEHB). A state machine is used to manage the transitions between these modes based on the current braking requirements.
- Explicit MPC: For each of the three modes, a linearized model is derived, and an explicit MPC controller is designed. Explicit MPC pre-computes the control law offline, allowing for faster execution during real-time operation compared to traditional online MPC solvers, which is crucial for automotive applications.
- Non-linear Extended Kalman Filter (EKF): To accurately estimate the system's states (like piston displacement and pressure) in the presence of non-linear dynamics and time-varying process noise, an EKF is employed. This helps to improve the robustness and accuracy of the MPC controller.
- Multi-Objective Optimization: The MPC controller is formulated to simultaneously track reference signals for piston displacement (related to pedal feel) and wheel cylinder pressure, while also considering input constraints and optimizing for regenerative braking efficiency.
Datasets
Not discussed in the paper. The paper relies on simulation results based on a developed model of the electro-hydraulic braking system.
Key Insights and Ideas
- Switchable Control is Key: The paper demonstrates that a single, monolithic controller struggles to handle the diverse dynamics of regenerative braking. By partitioning the system into distinct modes (FHB, DEHB, CEHB) and using mode-specific controllers, performance is significantly improved.
- Explicit MPC for Real-Time Performance: The adoption of explicit MPC is crucial for achieving the fast response times required in automotive braking systems, enabling the controller to react quickly to changing conditions without overwhelming the vehicle's processing capabilities.
- Pedal Feel Consistency is Achievable: A major contribution is the ability to maintain a consistent and natural pedal feel even during regenerative braking, which often introduces hydraulic hysteresis and can lead to a disconnected driver experience. This is achieved by decoupling the pedal feel (linked to piston displacement) from the wheel cylinder pressure control.
- EKF for Robust State Estimation: The use of an EKF with time-varying process noise is vital for accurately estimating system states, especially when transitioning between braking modes, preventing cumulative errors that could destabilize the MPC.
Key Results
- The proposed switchable MPC controller significantly decreases latency by 85 milliseconds compared to a single MPC.
- It improves accuracy by 22.6%.
- The controller demonstrates consistent pedal feel across different braking conditions, including regenerative braking.
- In simulations, the switchable MPC method saves 9.8% electric energy compared to braking without regeneration and improves regenerative efficiency by 7.7% in a specific test case.
- The controller shows improved tracking performance for piston displacement (±0.20 mm precision) compared to a single MPC (±0.45 mm).
Limitations and Future Work
- Future Case Studies: The authors suggest future work should investigate the effects of low-pressure accumulator volume and lateral motion on accuracy.
- Flow Control Accuracy: Improving the accuracy of flow control, particularly considering hydraulic hysteresis, is identified as an area for future enhancement to accelerate convergence speed during the return stroke.
- Real-World Validation: While simulations are presented, the paper does not detail experimental validation on a physical vehicle. Future work would likely involve implementing and testing this system on a real electric vehicle.
Summary updated