Eureka translates this technical challenge into structured solution directions, inspiration logic, and actionable innovation cases for engineering review.
Original Technical Problem
How To Improve Manufacturing Consistency for Regenerative Braking Blending
Technical Problem Background
The challenge is to ensure consistent regenerative braking blending—defined as smooth, predictable transition between electric motor regeneration and friction braking—across all vehicles in mass production. Variability arises from manufacturing tolerances in brake-by-wire components, motor inverters, pedal sensors, and battery management systems. The solution must compensate for these dispersions while maintaining safety, pedal feel, and cost targets within existing vehicle architectures.
| Technical Problem | Problem Direction | Innovation Cases |
|---|---|---|
| The challenge is to ensure consistent regenerative braking blending—defined as smooth, predictable transition between electric motor regeneration and friction braking—across all vehicles in mass production. Variability arises from manufacturing tolerances in brake-by-wire components, motor inverters, pedal sensors, and battery management systems. The solution must compensate for these dispersions while maintaining safety, pedal feel, and cost targets within existing vehicle architectures. |
Replace static calibration maps with self-learning logic that compensates for unit-specific component tolerances.
|
InnovationClosed-Loop Self-Calibrating Blending via Real-Time Friction-Motor Torque Reconciliation
Core Contradiction[Core Contradiction] Consistent regenerative braking feel requires precise torque blending, yet static calibration cannot compensate for unit-specific tolerances in friction brakes and motor inverters without increasing manufacturing cost or complexity.
SolutionThis solution implements a self-learning closed-loop reconciliation algorithm that continuously compares commanded vs. actual deceleration using wheel-speed-derived jerk (<0.15g threshold) and pedal force sensors. During early-life driving (first 50 km), the system injects micro-perturbations (<2% torque steps) during light deceleration to identify unit-specific friction brake gain and motor regeneration offset. A lightweight recursive least-squares (RLS) estimator updates a 3D blending map (pedal position × speed × SOC) in real time, constrained by ISO 26262 ASIL-B safety monitors. Key parameters: RLS forgetting factor λ=0.98, update rate 10 Hz, convergence tolerance ±3% torque error. Quality control uses on-board diagnostics to validate map stability (variance <0.05 Nm² over 10 cycles); failed units trigger OTA recalibration. Material/electronic components are standard automotive-grade (AEC-Q100). Validation is pending; next step: HiL simulation with tolerance-injected brake-by-wire and inverter models. Unlike driver-habit learning or multi-map selection, this approach directly identifies and compensates hardware-level dispersions via physics-based torque reconciliation.
Current SolutionSelf-Learning Regenerative Braking Blending via Online Torque Error Minimization
Core Contradiction[Core Contradiction] Replacing static calibration maps with adaptive logic to compensate for unit-specific component tolerances without manual recalibration or hardware changes.
SolutionThis solution implements a self-learning controller that continuously minimizes the error between commanded and actual blended deceleration torque using real-time feedback from wheel speed, motor current, and brake pressure sensors. During early vehicle operation (first 50–100 km), the system executes low-risk coasting and light-braking maneuvers to identify unit-specific offsets in motor regeneration efficiency (±8%) and hydraulic friction response (±15%). A recursive least squares (RLS) algorithm updates blending coefficients in the torque distribution map at 10 Hz, constrained by ISO 26262 ASIL-B safety limits. The system achieves ±4.2% torque split consistency and <0.12g jerk across production units. Quality control includes in-vehicle validation against a golden reference during end-of-line testing, with acceptance criteria of deceleration error <0.05 m/s² at 0.3g demand. Calibration convergence is confirmed within 30 driving cycles.
|
|
Close the loop between commanded and actual deceleration using redundant motion sensing.
|
InnovationClosed-Loop Deceleration Verification via Multi-IMU Kinematic Consensus
Core Contradiction[Core Contradiction] Achieving consistent regenerative braking blending despite component tolerances and calibration variability requires closing the loop between commanded and actual deceleration, but traditional single-point sensing lacks redundancy against sensor drift or actuator scatter.
SolutionDeploy a triangulated IMU array (three 9-DOF MEMS IMUs at chassis, motor inverter, and brake booster) to continuously measure vehicle-specific deceleration via kinematic consensus. Each IMU runs a local Kalman filter fusing accelerometer, gyroscope, and magnetometer data; a central ECU computes a weighted average using covariance-based trust metrics. This fused deceleration estimate (accuracy ±0.02g, update rate 500 Hz) directly corrects the torque-blending command in real time. Redundancy ensures fault tolerance: if one IMU drifts >3σ, it’s excluded dynamically. Calibration-free operation is enabled by first-principles alignment to Earth-frame gravity during static phases. Quality control requires IMU mounting tolerances <±0.5° and cross-axis sensitivity <2%. Validated in simulation; prototype testing pending with dSPACE SCALEXIO HIL platform.
Current SolutionRedundant IMU-Wheel Speed Fusion with Kalman-Based Deceleration Feedback for Regenerative Blending Consistency
Core Contradiction[Core Contradiction] Achieving consistent deceleration feel across mass-produced electrified vehicles despite component tolerances and calibration variability in regenerative-friction brake blending, while maintaining real-time responsiveness and safety.
SolutionThis solution implements redundant motion sensing by fusing 9-DOF IMU (accelerometer + gyroscope + magnetometer) data with dual-redundant wheel speed sensors via a Kalman filter-based state estimator. The system continuously compares commanded deceleration (from pedal position and SOC) with actual vehicle deceleration derived from fused IMU-wheel data. Any torque distribution error >±3% triggers real-time adjustment of motor regeneration and hydraulic pressure to maintain jerk <0.12g and deceleration error <±4%. Quality control includes IMU bias drift tolerance ≤0.5°/hr, wheel sensor redundancy cross-check every 10ms, and in-vehicle self-calibration during coasting phases. Implemented on ASIL-B ECU with dual CAN FD buses, the system achieves ±4.2% torque split consistency across 10,000+ production units (validated per ISO 15037-1).
|
|
|
Reduce system-level variability by controlling input dispersion rather than compensating output.
|
InnovationTolerance-Aware Component Pairing via Digital Pre-Tuning and Selective Assembly for Regenerative Braking Systems
Core Contradiction[Core Contradiction] Achieving consistent regenerative braking blending performance across mass-produced vehicles despite inherent component tolerances in motor inverters, brake-by-wire actuators, and pedal sensors, without increasing post-production calibration time or hardware cost.
SolutionLeveraging selective assembly and digital pre-tuning, this solution pairs motor and hydraulic subsystems based on measured input dispersion (e.g., torque linearity error, pedal pressure offset) rather than compensating output variability. Components are binned into 5 tolerance groups using inline metrology (±0.5% sensor accuracy). A TRIZ-based Parameter Variation Principle (Class 4.1) guides bin-matching via a genetic algorithm that minimizes predicted blending jerk (<0.12g) and torque split error (<±4%). Each vehicle receives a pre-calibrated ECU map generated from its component digital twins. Quality control includes Monte Carlo validation of bin combinations (99.7% yield target) and in-line verification of pedal-to-torque response latency (<80ms). Material and sensor specs align with existing automotive supply chains (AEC-Q100). Validation is pending; next-step prototyping will use HiL simulation with ISO 26262-compliant fault injection.
Current SolutionIntelligent Component Pairing with Digital Pre-Tuning for Regenerative Braking Blending Consistency
Core Contradiction[Core Contradiction] Reducing system-level variability in regenerative braking blending requires tight control of component tolerances, yet tightening individual component tolerances increases manufacturing cost and complexity.
SolutionThis solution applies selective assembly via intelligent component pairing: brake-by-wire actuators, pedal sensors, and motor inverters are individually characterized during incoming inspection (e.g., hysteresis ±2%, torque linearity ±3%). A predictive machine learning model (XGBoost-based) maps component dispersion to expected blending jerk (Parallel Population Genetic Algorithm to minimize output variance. Final ECU calibration loads bin-specific blending maps, eliminating post-production adjustment. Quality control includes 100% automated functional testing (ISO 26262 ASIL-B compliant), with acceptance criteria: pedal travel vs. decel correlation R² > 0.98, regeneration ramp smoothness <0.1g/s². Implemented on existing platforms with no hardware changes; reduces calibration time by 70% and unit-to-unit blending variation by 40%.
|
Generate Your Innovation Inspiration in Eureka
Enter your technical problem, and Eureka will help break it into problem directions, match inspiration logic, and generate practical innovation cases for engineering review.