Virtual Sensor for Vehicle Variables via Neural Network Derivatives
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern vehicles face challenges in estimating unmeasurable quantities in real-time due to the absence of physical sensors, which limits their operational capabilities and health monitoring.
Innovation Solution
A processor-implemented method using a trained feedforward neural network, combined with a model-free derivatives estimator, estimates unmeasurable variables by computing time derivatives of measurable signals, allowing for real-time estimation and operation without requiring a physical sensor.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a physical sensor is installed to measure an unmeasurable quantity, then measurement capability is improved, but device complexity and cost increase
Solution Approach 1:
The patent creates a virtual sensor that copies the measurement capability of a physical sensor by using a neural network model. The virtual sensor processes data from existing sensors to estimate unmeasurable quantities, providing the same measurement capability without installing additional physical hardware, thereby resolving the contradiction between measurement capability and device complexity
Solution Approach 2:
The patent replaces the mechanical/physical sensor system with a computational/software-based virtual sensor. Instead of using physical components to measure quantities, the system uses neural network algorithms running on existing processors to estimate unmeasurable variables, substituting mechanical measurement with computational estimation
2Device complexity
If computational methods are used to estimate unmeasurable variables, then device complexity is reduced, but measurement precision may deteriorate
Solution Approach 1:
The patent applies preliminary action by training the neural network offline using labeled data before deployment. The virtual sensor is pre-trained with comprehensive datasets that include both measurable and unmeasurable variables, allowing it to learn accurate relationships in advance. This offline training ensures high estimation accuracy when the virtual sensor operates with limited real-time computational resources
Solution Approach 2:
The patent replaces physical sensing mechanisms with computational estimation using neural networks. The virtual sensor uses algorithms to infer unmeasurable variables from measurable ones, achieving acceptable measurement precision through software-based estimation rather than hardware-based measurement
3Productivity
If real-time estimation is implemented without physical sensors, then productivity and response time are improved, but measurement precision may be compromised
Solution Approach 1:
The neural network is trained offline in advance with comprehensive datasets, preparing the virtual sensor for real-time operation. This preliminary training allows the system to perform accurate estimations during real-time vehicle operation without requiring additional computation time, achieving both high productivity and measurement precision
Solution Approach 2:
The virtual sensor uses existing sensor data and computational resources already available in the vehicle system. It serves itself by processing data from existing sensors through the neural network model, eliminating the need for additional physical sensors or external measurement systems while maintaining real-time capability
Data Source
AI summary
Provided is a processor-implemented method and a processor in a vehicle for estimating the value of a quantity for which a physical sensor is not available for measurement. The method includes: receiving a plurality of measured signals representing values of measurable variables; computing, in real-time, time derivatives of the measured signals; and applying a trained feedforward neural network, in real-time, to estimate values for a plurality of unmeasurable variables, the unmeasurable variables being variables that are unmeasurable in real-time, the feedforward neural network having been trained using test data containing time derivatives of values for the measurable variables and values for the unmeasurable variables; wherein the vehicle uses the estimated values for the unmeasurable variables for vehicle operation.