Abstract
Distributed driving electric vehicles (EVs) utilize independent in-wheel motors to enhance maneuverability but face significant stability risks on slippery surfaces due to uneven wheel slip. This paper presents an Acceleration Slip Regulation (ASR) strategy that integrates real-time road surface identification with a fuzzy logic controller. By dynamically estimating the road's friction coefficient and adjusting individual motor torques, the system maintains wheel slip within optimal ranges, significantly improving vehicle stability and acceleration performance across diverse road conditions.
Introduction
Distributed drive electric vehicles represent a significant advancement in electric mobility, featuring four independent in-wheel motors that allow for precise torque distribution. However, this architecture introduces a major challenge: if driving forces become unbalanced—particularly during acceleration on low-friction surfaces like ice or wet asphalt—the vehicle can easily lose stability or spin out.
Traditional Acceleration Slip Regulation (ASR) systems often rely on a fixed "optimum slip rate." Because different road surfaces (e.g., dry asphalt vs. ice) have vastly different optimal slip points, rigid controllers often fail to adapt to changing environments. This research addresses the challenge of dynamically identifying road surfaces in real-time to adjust the torque of each individual motor, preventing excessive wheel spin and ensuring safety.
Modeling and Road Identification
The proposed system utilizes a closed-loop control architecture combining vehicle physics modeling with fuzzy logic. The foundation of the system is the Burckhardt tire model, which maps the non-linear relationship between wheel slip rate and longitudinal tire force.
To ensure the controller adapts to the environment, a real-time road identification module uses the Levenberg-Marquardt algorithm to continuously estimate the road's peak attachment coefficient (friction). Based on this calculation, a 1-D lookup table determines the precise optimum slip rate for the current surface, allowing the vehicle to transition seamlessly between different road types.
Fuzzy Logic Control Strategy
A fuzzy logic controller is deployed for each wheel to manage torque adjustments smoothly. The controller processes two primary inputs: the slip error (the difference between actual and optimal slip rates) and the rate of change of that error.
The controller utilizes specific membership functions to ensure smooth transitions, outputting precise torque adjustments. This approach prevents the harsh, jerky torque changes often associated with traditional PID controllers, leading to smoother acceleration and improved passenger comfort. Because distributed EVs can control each motor independently, ASR can be applied asymmetrically—reducing torque only on the wheels losing traction while maintaining power on wheels with better grip.
Key Results
The effectiveness of the strategy was evaluated through co-simulation using CarSim and Simulink:
- Low Friction Performance: On highly slippery roads, the road identification module identified the optimal slip rate for the rear wheels in just 0.31 seconds. The controller stabilized the slip rate at approximately 0.008, enabling smooth acceleration to 11.24 m/s without loss of control.
- Variable Friction (Docking Road): When simulating a sudden drop in friction (from 0.6 to 0.2) at 50 km/h, the system adapted in 0.162 seconds. It maintained a tightly controlled slip rate (0.011 to 0.012), allowing the vehicle to safely accelerate to 22.5 m/s.
Conclusion
By combining real-time road identification with independent fuzzy logic torque control, this ASR strategy significantly enhances the stability of distributed drive EVs. While the system shows some slip rate fluctuation during low-speed start-ups on slippery surfaces, its rapid adaptation to changing road conditions demonstrates its potential for improving the safety and performance of next-generation electric vehicles. Future work will focus on refining the algorithm to improve accuracy during the initial start-up phase.
Summary updated