Control Secondary Air Injection System for RDE Test Cycles

Overview of Technical Issues:

The secondary air injection system's control unit cannot adequately adjust air injection timing and flow rate to match the unpredictable transient conditions of RDE test cycles, resulting in insufficient catalyst heating during rapid load changes and potential emission compliance failures; the goal is to optimize the control strategy to maintain effective emission reduction across all dynamic driving phases in real-world test conditions.

Solution directions generated for this problem

Problem Direction 1 :

ImproveControl system response speed
VS
ConstraintControl algorithm complexity

Inspiration 1 : Cross-domain reference

Application Principle: #35 Parameter changes
Cross-domain applicability Assess applicability
Welding power supply with digital control of duty cycle
Innovative Solution Refine solution

State-indexed pre-computed air injection control for RDE transients

Pre-compute optimal air injection maps offline then switch via fast state recognition
How to solve :
  • Offline compute optimal air injection trajectories for 8 discrete operating states (cold-start, warm-up idle, light/medium/heavy acceleration, cruise, deceleration, hot-restart) using multi-variable optimization on high-performance workstations — store as 2D lookup tables (engine speed × load) with 15×12 grid resolution
  • Implement lightweight state classifier using 3-input decision tree (exhaust temperature, throttle rate-of-change, engine speed) executing in <5ms to identify current state and select corresponding pre-computed map
  • Execute air injection control via direct table lookup at 50ms cycle — valve timing ±2ms tolerance, flow rate ±8% tolerance verified by mass airflow sensor feedback with ±0.5 g/s accuracy
Expected Effect : Response time 200ms→48ms; ECU load +12% vs +600% for real-time optimization; catalyst light-off time -18%
Risk Control :
  • state transition hysteresis causing control discontinuity
  • lookup table interpolation error during state boundaries
  • pre-computed maps not covering edge-case transients

Problem Direction 2 :

ImproveControl system response speed
VS
ConstraintSystem computational resource demand

Inspiration 1 : Cross-domain reference

Application Principle: #10 Preliminary action
Cross-domain applicability Assess applicability
Providing pre-computed hotword models
Innovative Solution Refine solution

Offline trajectory library for RDE air injection control

Pre-compute optimal air injection trajectories offline for RDE transients
How to solve :
  • Build offline trajectory library using high-performance computers to calculate optimal air injection timing and flow profiles for 50+ common RDE transient events (cold start 0-300°C in 60s, 0-60% throttle in 2s, 80-20% deceleration in 3s, etc.) using multi-variable optimization algorithms
  • store trajectories as indexed lookup tables (5KB per trajectory, total 250KB) in ECU flash memory with key parameters: throttle position, engine speed, catalyst temperature, ambient temperature
  • Implement fast pattern matching algorithm executing in <5ms that compares current driving conditions (throttle rate ±5%/s, speed ±200 RPM, temperature ±20°C) against library index to retrieve closest pre-calculated trajectory
  • Execute retrieved trajectory via simple feedforward controller with 50ms update cycle, requiring only table interpolation (0.2ms CPU time) instead of real-time optimization, achieving target response speed with 95% reduction in computational load versus adaptive algorithms
Expected Effect : Response time <50ms; CPU load <2% vs 15-20% for real-time optimization; emission compliance +12%
Risk Control :
  • trajectory library coverage gaps for rare driving patterns
  • pattern matching accuracy under sensor noise
  • flash memory wear from frequent access

Problem Direction 3 :

ImproveExhaust condition measurement precision
VS
ConstraintControl algorithm complexity

Inspiration 1 : Cross-domain reference

Application Principle: #32 Color changes
Cross-domain applicability Assess applicability
A ratiometric oxygen sensing probe, its preparation method and application
Innovative Solution Refine solution

Dual-wavelength ratiometric optical exhaust sensor for transient measurement

Ratiometric optical sensing eliminates algorithm complexity
How to solve :
  • Install dual-wavelength infrared sensor at catalyst inlet measuring oxygen-sensitive fluorescence at 650nm and reference signal at 550nm simultaneously
  • output oxygen concentration as intensity ratio I₆₅₀/I₅₅₀, inherently canceling temperature drift, light source fluctuations, and optical path contamination without software compensation algorithms
  • Integrate analog ratiometric circuit (operational amplifier divider) directly on sensor PCB performing I₆₅₀/I₅₅₀ calculation in hardware within 5ms, delivering calibrated oxygen concentration to ECU as single analog voltage 0-5V, eliminating multi-channel sampling and digital fusion logic
  • Add fast-response thermochromic coating (vanadium dioxide thin film 0.8μm thickness) on sensor window changing reflectance at 680nm proportional to temperature 200-600°C with <10ms thermal time constant
  • measure reflected 680nm intensity against incident beam for direct temperature readout, avoiding thermocouple lag and filtering algorithms
Expected Effect : Transient resolution 8ms; algorithm lines reduced 75%; sensor output pre-calibrated ±2% accuracy
Risk Control :
  • Optical window fouling by soot deposits
  • ratiometric circuit thermal drift ±0.3%/10°C
  • coating adhesion under thermal cycling

Problem Direction 4 :

ImproveExhaust condition measurement precision
VS
ConstraintSystem computational resource demand

Inspiration 1 : Cross-domain reference

Application Principle: #28 Mechanics substitution
Cross-domain applicability Assess applicability
Detecting digital assistant triggers
Innovative Solution Refine solution

Smart sensor with embedded edge processing for transient exhaust measurement

Offload computation to sensor hardware
How to solve :
  • Replace analog sensors with smart sensors featuring integrated ARM Cortex-M4 microprocessors (80 MHz) that perform local Kalman filtering and transient detection at 10 ms intervals, outputting processed data via CAN 2.0B at 500 kbps to reduce ECU load by 70%
  • Deploy wide-band lambda sensor (Bosch LSU 4.9 or equivalent) with embedded ASIC signal processor executing 5-point moving average and derivative calculation (dO2/dt) in hardware, delivering clean transient features instead of raw 100 Hz data streams
  • Install fast-response thermocouple (Type K, response time <15 ms) paired with analog edge-detection circuit using operational amplifiers (TL084) to trigger interrupt signals when temperature rate exceeds 50°C/s threshold, bypassing continuous polling by ECU
Expected Effect : Transient capture resolution 10 ms; ECU computational load +8% vs baseline; measurement precision ±2% for 10-50 ms fluctuations
Risk Control :
  • sensor calibration drift over 100k km
  • CAN bus bandwidth saturation during multi-sensor operation
  • embedded processor firmware update complexity

Problem Direction 5 :

ImproveControl strategy adaptability
VS
ConstraintControl algorithm complexity

Inspiration 1 : Cross-domain reference

Application Principle: #1 Segmentation
Cross-domain applicability Assess applicability
Integrated circuit for setting subbands in multicarrier communication for radio communication base station apparatus
Innovative Solution Refine solution

Subband-partitioned air injection control with fixed-parameter zones

Partition RDE cycle into fixed control zones
How to solve :
  • Divide the exhaust flow range into 5 discrete subbands (0-20%, 20-40%, 40-60%, 60-80%, 80-100% of max flow) with pre-calibrated air injection parameters for each zone
  • Assign each subband a fixed parameter set (injection timing offset ±15°CA, flow rate multiplier 0.8-1.5×) calibrated offline via dynamometer testing across 200+ RDE-representative transients
  • Implement threshold-based zone switching using exhaust pressure sensor (response time <10ms) to trigger instant parameter set changes when crossing subband boundaries, requiring only lookup table access (execution time <5ms)
Expected Effect : Adaptability across diverse RDE patterns achieved; algorithm complexity remains at simple lookup level; response time <50ms; no pattern recognition required
Risk Control :
  • subband boundary oscillation causing frequent switching
  • calibration coverage gaps for extreme transients
  • sensor drift affecting zone detection accuracy

Problem Direction 6 :

ImproveControl strategy adaptability
VS
ConstraintSystem computational resource demand

Inspiration 1 : Cross-domain reference

Application Principle: #23 Feedback
Cross-domain applicability Assess applicability
Augmented reality system for an amusement ride
Innovative Solution Refine solution

Event-triggered adaptive air injection with outcome-based parameter tuning

Outcome-driven adaptation via emission feedback
How to solve :
  • Monitor catalyst light-off time and post-catalyst NOx/HC levels as performance indicators
  • adjust air injection timing ±5-15ms and flow rate ±10-20% incrementally using simple hill-climbing logic that evaluates whether last adjustment improved emissions within 2-second window
  • Implement event-triggered control updates only when throttle rate exceeds 15%/s or exhaust temperature gradient surpasses 50°C/s, maintaining 200ms baseline cycle but activating 50ms fast response selectively, reducing average computational load by 60-75%
  • Store 8-12 pre-optimized parameter sets for discrete operating regimes (cold start <300°C, warm-up 300-450°C, hot operation >450°C × low/medium/high load) in 2KB lookup table
  • use simple threshold logic to switch sets based on temperature and throttle position, achieving regime-appropriate control without pattern recognition algorithms
Expected Effect : Adaptability +70% across RDE cycles; average CPU load +15% vs baseline; emission compliance rate >95%; response time 50-80ms during transients
Risk Control :
  • hill-climbing convergence in multi-modal scenarios
  • event trigger threshold calibration sensitivity
  • parameter set coverage gaps for edge cases
Patsnap Eureka Solution