Neural Network Angle Sensor Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing angle sensors in the automotive industry face challenges with misalignment and external magnetic fields, leading to reduced accuracy and increased production costs due to the need for complex calibration and shielding.
Innovation Solution
A device and method that train a neural network using system data and error data to determine rotation angles, accounting for geometric deviations and external magnetic fields, allowing for accurate angle determination with low-cost, simple angle sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex production processes and calibration are used to minimize misalignment influence, then measurement precision is improved, but device complexity and production costs increase
Solution Approach 1:
The neural network is trained in advance with simulated sensor data that includes various misalignment scenarios and external magnetic field conditions. This preliminary training enables the network to compensate for these errors during actual operation without requiring complex physical calibration processes or specialized production procedures.
Solution Approach 2:
The patent replaces complex mechanical calibration procedures and physical shielding mechanisms with a software-based neural network approach. The neural network processes sensor data and compensates for misalignment and magnetic field interference through computational methods, eliminating the need for elaborate mechanical adjustment and calibration systems.
2Measurement precision
If magnets with larger diameter are used to produce stronger magnetic field, then measurement precision is improved, but device dimensions and cost increase
Solution Approach 1:
Instead of changing the physical parameter of magnet diameter, the patent changes the processing parameters of the sensor data through neural network algorithms. The network adjusts how the magnetic field data is interpreted and compensates for weaknesses in the magnetic field strength through computational correction, allowing accurate measurement without requiring larger magnets.
Solution Approach 2:
The patent creates a virtual model of the sensor system including the magnet and sensors, and trains the neural network using simulated measurement data from this model. This virtual training allows the system to learn optimal compensation strategies without physically modifying the magnet size or conducting extensive physical calibration experiments.
3Measurement precision
If shielding is used to reduce external magnetic field influence, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent replaces physical magnetic shielding structures with a software-based compensation approach. The neural network analyzes the sensor data and mathematically compensates for the effects of external magnetic fields, eliminating the need for complex magnetic shielding materials and structures that would add device complexity and cost.
Solution Approach 2:
Instead of trying to physically block external magnetic fields, the patent converts the harmful effect into useful information. The neural network learns to recognize patterns caused by external fields and uses this knowledge to compensate for their effects, transforming the problem of magnetic interference into a solvable computational task.
4Measurement precision
If differential measurement with multiple magnetic sensors is used to correct external magnetic fields, then measurement precision is improved, but susceptibility to misalignment increases
Solution Approach 1:
The patent introduces the neural network as an intermediary processing layer between the simple magnetic sensors and the final angle determination. The network receives data from the sensors and performs the complex compensation calculations, allowing the use of simple, inexpensive sensors without requiring them to be precisely aligned, as the neural network compensates for alignment errors.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The solution enables precise rotation angle determination with reduced errors using low-cost, uncomplicated angle sensors, avoiding the need for complex production processes or expensive calibration and shielding.
Implementation Method 1
receive system data about a sensor system for measuring a magnetic field in order to determine the rotation angle
Data Source
AI summary
An exemplary embodiment relates to a device for training a neural network for determining a rotation angle of an object. The device is configured to receive system data via a sensor system for measuring a magnetic field in order to determine the rotation angle. The device is also configured to generate error data which includes at least one deviation of the system data from a target state of the sensor system or the strength of the components of a superimposed external magnetic field. Furthermore, the device is configured to create training data using the system data and the error data and to train the neural network using the training data.