Pulsed Mechanical Effect Detection in Plant Components
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting pulse-like mechanical actions on plant parts, such as loose parts carried by gas turbines, face challenges in distinguishing these events from high and temporally varying background noise, leading to false alarms and missed detections due to the high operating noise levels.
Innovation Solution
A method involving continuous recording of operating noise with sensors, processing the measurement signals through mathematical transformations, and using an evaluation function derived from Fourier transforms, combined with adaptive quantile-based averaging to enhance signal detection and reduce noise interference, allowing for the reliable identification of pulse-like mechanical effects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensors continuously record operating noise to detect pulse-like mechanical actions, then detection capability is improved, but false alarms increase due to high background noise levels
Solution Approach 1:
The patent transforms the measurement signal from time domain to frequency domain using Fourier transformation, changing the parameter representation from amplitude over time to frequency spectrum. This allows identification of characteristic frequency patterns that distinguish impulse events from background noise, resolving the contradiction between detection sensitivity and false alarm rate
Solution Approach 2:
The system continuously monitors the frequency spectrum and dynamically compares current spectral characteristics against reference patterns. This feedback mechanism enables real-time distinction between genuine impulse events and background noise variations, maintaining high detection accuracy while minimizing false alarms
2Loss of energy
If band or high-pass filtering is applied to improve signal-to-noise ratio, then background noise is reduced, but impulse-like events may be attenuated or distorted
Solution Approach 1:
Instead of filtering in the time domain, the patent transitions to frequency domain analysis using Fourier transformation. This dimensional change allows selective identification and evaluation of impulse characteristics in the frequency spectrum without applying time-domain filters that would distort the impulse signal itself
3Loss of energy
If simple averaging of measurement signals is used to reduce noise, then noise suppression is achieved, but rapid noise level changes cause false alarms
Solution Approach 1:
The system pre-calculates and stores reference frequency spectra representing normal operating conditions before impulse events occur. During monitoring, current spectra are compared against these pre-established references, enabling rapid distinction between genuine impulses and noise transitions without requiring complex real-time adaptive averaging
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 approach effectively differentiates burst signals from background noise, enabling accurate detection of pulse-like mechanical events even during high noise conditions, such as turbine humming, thereby preventing false alarms and ensuring timely detection of potential damage.
Implementation Method 1
A plurality of measured value pickups or sensors 4, in particular piezoelectric acceleration pickups, are arranged on a plant part 2
Implementation Method 2
the absolute value A(f i ,t j )) of the transform of the measurement signal (M) is determined in time steps of time windows that follow one another in time using specified parameters of a mathematical transformation rule
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
The invention relates to a method and device for detecting a pulsed mechanical effect on a plant component (2), whereby an operating noise in the plant component (2) is continuously recorded by means of a sensor (4) which is arranged on the plant component (2) and converted by the above into a measured signal (M) which is subjected to a mathematical transformation. According to the invention, a sliding mean is calculated by determining the quantiles from the determined transformations and the evaluation function (K(t)), which displays the application of a pulsed mechanical effect on the plant component (2), is derived from said sliding mean.