Inferential Sensor Framework for Automotive Emissions Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern internal combustion engines face complexity due to stringent emission regulations, requiring additional subsystems and sensors, which increase cost, complexity, and maintenance, making direct measurement of certain quantities like NOx and ammonia emissions impractical.
Innovation Solution
A framework for designing inferential sensors that replace physical sensors with models, using mathematical techniques to estimate unmeasurable quantities in automotive subsystems, such as ammonia storage and NOx/NO2 ratios, by preparing models, populating real-time templates with data, and running inferential sensors in real-time to obtain estimated variables.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physical sensors are used to directly measure quantities like NOx and ammonia emissions, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent creates virtual copies of physical sensors through mathematical models that replicate sensor functionality without physical hardware. These virtual sensors compute estimated values by processing data from existing sensors and applying domain-specific algorithms, thereby achieving measurement capabilities without adding physical sensor complexity
Solution Approach 2:
The patent replaces physical sensing mechanisms with computational models. Instead of using additional physical sensors to detect quantities like NOx and ammonia, the system substitutes these with software-based inferential sensors that calculate estimated values through mathematical relationships and data processing
2Measurement precision
If additional physical sensors are added to meet emission regulations, then measurement precision is improved, but manufacturing cost increases
Solution Approach 1:
The patent creates virtual copies of physical sensors through mathematical models that replicate sensor functionality without physical hardware. These virtual sensors compute estimated values by processing data from existing sensors and applying domain-specific algorithms, thereby achieving measurement capabilities without adding physical sensor complexity
Solution Approach 2:
The patent uses computationally inexpensive virtual sensor models that can be rapidly deployed and modified through software updates rather than hardware changes. These virtual sensors provide measurement capabilities at minimal marginal cost compared to physical sensor installation
3Measurement precision
If more physical sensors are installed to control emissions, then measurement precision is improved, but maintenance complexity increases
Solution Approach 1:
The patent creates virtual copies of physical sensors through mathematical models that replicate sensor functionality without physical hardware. These virtual sensors compute estimated values by processing data from existing sensors and applying domain-specific algorithms, thereby achieving measurement capabilities without adding physical sensor complexity
Solution Approach 2:
The virtual sensor system automatically adapts and calibrates itself using existing sensor data and mathematical models, eliminating the need for manual maintenance, calibration, or repair that would be required for additional physical sensors
4Device complexity
If a framework approach is used to replace physical sensors with models, then device complexity is reduced, but measurement precision may worsen
Solution Approach 1:
The patent implements feedback mechanisms where virtual sensor estimates are continuously refined using actual sensor measurements and system state information. This closed-loop approach allows the mathematical models to adapt and improve their accuracy over time while maintaining the simplicity of the virtual sensor architecture
Solution Approach 2:
The patent dynamically adjusts model parameters and computational approaches based on operating conditions to optimize measurement accuracy. By changing parameters such as model complexity, data weighting, and computational methods in response to system state, the virtual sensors maintain high precision across varying operational scenarios
Data Source
Figure 1
Figure 2
Figure 3~6
AI summary
A system that provides an estimation of quantities, which are not necessarily directly measurable in a subsystem. The estimates of these quantities may be inferred from other variables. This approach may be referred to as inferential sensing. Physical sensors may be replaced with models or virtual sensors, also known as inferential sensors. The present approach may be a framework for designing inferential sensors in automotive subsystems. The framework may incorporate preparing a model for an observed subsystem, populating a real-time template with data, and running an inferential sensor periodically together with a model in real-time to obtain estimated variables. Once implemented, the framework may be reused for virtually any automotive subsystem without needing significant software code changes.