Pre-tensioned Component Quality Control via Frequency Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing quality control methods for sensors and actuators that utilize piezoelectric or ferroelectric converters struggle to accurately assess the compressive load, leading to reduced effectiveness or failure due to excessive or insufficient pressure, and require direct access during manufacturing, which is not feasible in finished components.
Innovation Solution
A method involving frequency dependency analysis of electrical parameters influenced by AC voltage, using a combination of parameterized physical models and machine learning modules to evaluate the compressive force and quality of the component, allowing for non-destructive, post-manufacturing quality control without direct access to the component's interior.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If direct access to component interior is used during manufacturing for quality control, then measurement precision of compressive force is improved, but device complexity and manufacturing process requirements increase
Solution Approach 1:
The patent replaces direct mechanical access and physical measurement methods with an electrical field-based measurement system. By applying AC voltage and analyzing the frequency dependence of electrical parameters (current, impedance), the compressive force is measured indirectly through electrical characteristics that are influenced by the mechanical stress state of the piezoelectric/ferroelectric transducer element, eliminating the need for direct interior access.
Solution Approach 2:
The patent changes the measurement parameter from direct mechanical observation to electrical parameter frequency dependence. By measuring how electrical parameters (current, impedance) vary with frequency under different compressive loads, the system extracts quality information without requiring physical access to the component interior during manufacturing.
2Reliability
If frequency dependence analysis with AC voltage is used for quality control, then non-destructive testing capability is improved, but measurement precision of compressive force may deteriorate
Solution Approach 1:
The patent employs feedback by using the measured frequency dependence of electrical parameters to iteratively determine the compressive force. The system measures the electrical response at multiple frequencies, analyzes the frequency dependence pattern, and uses this feedback information to accurately infer the compressive force state, maintaining measurement precision while enabling non-destructive testing.
Solution Approach 2:
The patent applies mechanical vibration principles by using AC voltage to induce dynamic electrical responses in the transducer element. By analyzing the frequency dependence of these responses (including resonance behavior), the system extracts information about the mechanical stress state without applying destructive mechanical loads, thus achieving non-destructive testing with adequate precision.
3Measurement precision
If hysteresis and mechanical changes are accounted for in the evaluation model, then quality assessment accuracy is improved, but device complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-training machine learning modules and parameterized physical models with data that includes hysteresis effects and mechanical changes. These models are prepared in advance to account for non-linear behaviors, so that during actual quality control measurements, the complex hysteresis compensation is already built into the evaluation algorithm, reducing the apparent complexity during operation.
Solution Approach 2:
The patent uses a composite evaluation approach that combines parameterized physical models with machine learning modules. This hybrid model integrates the strengths of both approaches: the physical model provides theoretical framework for hysteresis and mechanical changes, while the machine learning component learns from data to accurately predict quality parameters, creating a composite system that handles complexity effectively.
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
Enables reliable determination of compressive force and quality assessment independently of manufacturing processes, accounting for hysteresis and mechanical changes, ensuring consistent quality classification and efficient series production with 100% non-destructive testing capability.
Implementation Method 1
piezoelectric or ferroelectric transducers, made of electrically polarized materials, are used in sensors and actuators that measure or exert forces or pressures
Implementation Method 2
piezoelectric or ferroelectric transducers, made of electrically polarized materials
Implementation Method 3
Conversely, such a transducer element can be deformed by applying an electrical voltage and thus exert a force
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method (100) for the quality control of a component (1), wherein this component (1) comprises at least one converter element (2) which converts between a force (3) exerted on the converter element (2) or by the converter element (2) and a change in at least one electrical variable (4) of the converter element (2), wherein the converter element (2) is pressure-loaded by a preloading element (9), the method comprising the steps of: - obtaining (110) a frequency dependence (1d, 1d*) of an electrical characteristic variable (1b, 1c) of the component (1), which variable is influenced by applying, to the component (1), a voltage (1a) having an alternative voltage component; - evaluating (120b), from the frequency dependence (1d, 1d*), an evaluation variable (5a); - evaluating (130), from the evaluation variable (5a), at least one desired quality variable (6), which is a measure of the quality of the component (1). The invention also relates to a method (200) for training a machine-learning module (5) for use in the method (100).