Mechanical Fault Detection from Undersampled Multi-Delay Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fault diagnosis and predictive maintenance solutions for mechanical components in electric drive systems face challenges with high sampling frequencies, which increase complexity and cost, while low sampling frequencies result in unreliable diagnosis due to aliasing, making it difficult to identify mechanical faults effectively and efficiently.
Innovation Solution
A method using undersampled measurement data collected at different delays and sampling frequencies, analyzed through machine learning-based abnormality identification models, including characteristic extraction and deep learning algorithms, to identify abnormalities in mechanical components with a single or multiple sensors, reducing the need for high-frequency sampling and complex equipment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high sampling frequency signals are used in fault diagnosis, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent changes the parameter of sampling frequency from high to low (undersampled), and compensates for the resulting information loss by introducing multiple delay parameters (Δt1, Δt2, ..., Δtn) and using machine learning models to reconstruct fault features, thereby maintaining measurement precision while reducing device complexity
Solution Approach 2:
The patent introduces machine learning models (including deep learning neural networks) as intermediaries between the undersampled measurement data and fault diagnosis results. These models process the limited data and extract fault features, enabling accurate diagnosis without requiring high sampling frequencies
2Device complexity
If low sampling frequency signals are used in fault diagnosis, then device complexity is reduced, but measurement precision deteriorates due to aliasing
Solution Approach 1:
The patent segments the signal acquisition process by collecting multiple sets of undersampled data with different delay parameters. Instead of relying on a single high-frequency signal, the system divides the measurement into multiple lower-frequency acquisitions, each providing complementary information that collectively enables accurate fault diagnosis
Solution Approach 2:
The patent adds a new dimension to the data by introducing multiple delay parameters (Δt1, Δt2, ..., Δtn) as an additional variable. This transforms the problem from a single-time-point measurement to a multi-dimensional dataset, allowing machine learning models to reconstruct fault features that would be lost in traditional single-channel undersampled signals
3Reliability
If high sampling frequency signals are used, then fault identification reliability is improved, but energy consumption increases
Solution Approach 1:
The patent changes the sampling frequency parameter from high to low, directly reducing the energy consumption of sensors. The reliability is maintained by compensating with multiple delay parameters and using energy-efficient machine learning algorithms that can accurately identify faults from the reduced-rate data
4Measurement precision
If high sampling frequency signals are used, then fault diagnosis accuracy is improved, but cost increases
Solution Approach 1:
The patent replaces expensive high-frequency sampling hardware with cheaper low-frequency sensors. The system uses inexpensive sensors that collect undersampled data, relying on software-based machine learning models (which can be updated or replaced without hardware changes) to achieve accurate fault diagnosis, thereby reducing overall system cost
Data Source
AI summary
A method for identifying an abnormality in a mechanical apparatus or mechanical component includes: i) acquiring at least two classes of undersampled measurement data collected in or on a mechanical apparatus or mechanical component, all of the at least two classes of undersampled measurement data being different from one another in either one of or both of the following aspects: delay relative to occurrence time of a trigger event, and sampling frequency; and ii) based on the at least two classes of undersampled measurement data acquired, using an abnormality identification model to identify an abnormality in the mechanical apparatus or mechanical component, the abnormality identification model being based on machine learning and used for identifying an abnormality in the mechanical apparatus or mechanical component. Also disclosed is a method for training an abnormality identification model based on machine learning, a computer apparatus, a computer program product, and a detection apparatus. The above provides a fault diagnosis or predictive maintenance solution which is cost-effective and gives reliable results.