AI-Based LVDT Position Sensing Beyond the Linear Range
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional Linear Variable Differential Transformers (LVDTs) are limited by their Stroke to Length (STL) ratio, which restricts their use in applications with limited space, as their bodies need to be twice as long as the nominal operating region, making them unsuitable for certain size and stroke requirements.
Innovation Solution
An improved LVDT arrangement with a digital signal processing component using artificial intelligence or machine learning methods to extend the linear response range beyond the nominal operating region, allowing for a higher STL ratio and enabling operation in previously unsuitable spaces by producing output signals for core positions within and outside the linear response range.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Length of moving object
If conventional LVDT design is used, then the sensor provides reliable measurement in linear response range, but the body length must be twice as long as the nominal operating region (STL ratio ≈ 50%)
Solution Approach 1:
The patent changes the operational parameters by applying artificial intelligence/machine learning algorithms to process the electrical output signals. This transforms the measurement approach from direct linear correlation to a model-based interpretation that can extract position information beyond the traditional linear response range, effectively increasing the measurement range without extending the physical body length.
Solution Approach 2:
The patent replaces the traditional mechanical/electrical direct conversion system with a digital signal processing system using AI/ML models. Instead of relying solely on the physical linear relationship between core position and voltage output, the system uses trained models to interpret electrical features and determine position, enabling operation beyond the nominal linear range.
2Length of moving object
If HSTL LVDT design is used, then the STL ratio is improved to 25-30%, but not all LVDTs with specific size and stroke requirements can be designed following that approach
Solution Approach 1:
The patent creates a universal solution that can be applied to LVDTs with various size and stroke requirements. The AI/ML-based approach is not limited to specific HSTL designs but can be adapted to different LVDT configurations, providing a multi-functional methodology that enhances design flexibility across various application scenarios.
Solution Approach 2:
By changing from a fixed hardware-based linear response limitation to a flexible software-based AI/ML interpretation system, the patent enables adaptability across different LVDT designs. The trained models can be configured for specific size and stroke requirements, providing universal applicability rather than being constrained to particular HSTL configurations.
3Measurement precision
If the displacement range is extended beyond the linear response range, then the STL ratio is increased, but the linearity of output signals is compromised
Solution Approach 1:
The patent introduces an intermediary AI/ML processing layer between the raw electrical output signals and the final position determination. This intermediary model learns the complex non-linear relationships during training and provides accurate position estimates even when the direct linear relationship between voltage and position breaks down in extended ranges.
Solution Approach 2:
The patent substitutes the direct linear electrical-mechanical relationship with a data-driven AI/ML model that captures the true non-linear behavior. This replacement allows the system to maintain measurement precision across extended ranges by using learned patterns rather than assuming linearity, effectively decoupling measurement accuracy from signal linearity.
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
This solution increases the STL ratio, making LVDTs more versatile and suitable for applications with limited space, while also providing a scalable and adaptable solution that can be replicated across various LVDTs with different characteristics, and includes a protection circuit to prevent damage from excessive currents.
Implementation Method 1
LVDTs generally comprise a high design robustness... a primary coil and secondary coils... when an excitation current is supplied to the primary coil
Implementation Method 2
a value y of the electrical feature, parameter or characteristic may generally be described by a linear relationship or equation such as y = ax + b
Data Source
Figure 1~2
Figure 3A~3B
Figure 4A~4B
AI summary
An arrangement (100) comprising a linear variable differential transformer (110) including a core (131) displaceable in a bore (105) surrounded by a coil arrangement including a primary coil (111) and secondary coils (112, 113) in a displacement range including a linear response range of the linear variable differential transformer (110), wherein the bore (105) at least extends between a first position at a first end of the coil arrangement and a second end at a second end of the coil arrangement, and wherein the arrangement (100) comprises a control unit (120) configured to produce output signals corresponding to a plurality of positions of the core (131) in the bore (105) when an excitation current is supplied to the primary coil (111). The displacement range is larger than the linear response range, the plurality of positions include positions outside the linear response range, and the control unit (120) is configured to provide the output signals as output signals indicating the plurality of positions of the core (131) in the bore (105) within and outside of the linear response range, and particularly to produce the output signals by evaluating an input dataset including at least one of a voltage of the primary coil (111), of the secondary coils (112, 113), and a temperature of the variable differential transformer (110) using a trained artificial intelligence module (122). Corresponding methods are also provided.