Signal preprocessing method for diagnosing wire rope defects

Abstract

Various problems in a wire rope, such as internal defects, fatigue damage, wear, and corrosion, are nondestructively analyzed by momentarily magnetizing a wire rope with a permanent magnet for detection and detecting, via sensors, magnetic flux leakage signals that are generated in the magnetized state, but issues arise: defect signals are not clearly delineated due to environmental noise and trends contained in raw magnetic flux leakage signals; and signal distortion caused by the dynamic state of a rope must be removed to detect defects more accurately. To address these issues, data preprocessing comprises: a channel trend removal step for removing low-frequency signals below a threshold after acquiring signal data from a plurality of Hall sensors; a channel noise removal step for identifying high-frequency signals above the threshold as noise and removing same from the signal data that has undergone the trend removal step; a change point detection step for removing the effect of the wire rope moving upward, downward, or being stationary, to distinguish states and calculating information for detecting defects; and a Hilbert transform step for making the signs of the signals to be the same on the basis of the threshold so that the defects of the wire rope can be detected.

Description

Signal Preprocessing Method for Wire Rope Defect Diagnosis

[0001] The present invention relates to a signal preprocessing method before analyzing data measuring the leakage flux of a wire rope in order to diagnose wire ropes used in cranes, elevators, sluice gates, etc., in real time.

[0002] Nondestructive Testing (NDT) refers to a method of inspecting objects or systems without damaging or destroying them to detect physical or structural defects. NDT is widely utilized as a core technology for evaluating structural integrity, ensuring safety, and extending the service life of objects. The purposes of NDT are defect detection, structural integrity assessment, preventive maintenance, and quality control. Various techniques are used for non-destructive diagnostic technology depending on the purpose of inspection, the material of the object, and the characteristics of the defect. These include leakage flux detection technology, which utilizes the principle that leakage flux is generated at the defect site in a magnetized object; ultrasonic inspection technology, which transmits ultrasonic signals into the object and analyzes the reflected signals; radiographic testing (RT) technology, which uses X-rays or gamma rays to photograph the inside of the object; eddy current testing (ECT) technology, which induces eddy currents on the surface of a metal object to measure electromagnetic changes caused by defects; infrared thermography (IRT) technology, which analyzes the thermal characteristics of the object to detect temperature changes occurring at the defect site; and acoustic emission testing (AET) technology, which detects acoustic signals generated when internal defects occur in the object.

[0003] Wire ropes are essential components used in various industries, such as construction, mining, and ports, for supporting or moving heavy loads. However, defects occurring during use, such as wear, cracks, and corrosion, can lead to performance degradation and accident risks. To prevent these issues, non-destructive diagnostic technologies have been introduced, and in particular, Magnetic Flux Leakage (MFL) technology has established itself as a key method for efficiently detecting defects in wire ropes. Conventional MFL technology magnetizes the rope and analyzes defects by collecting leakage flux signals through sensors. However, raw signals are susceptible to noise and environmental factors, which can lower the accuracy of defect detection; therefore, signal preprocessing technology is essential to compensate for this.

[0004] Various problems such as internal defects, fatigue damage, wear, and corrosion in wire ropes are detected non-destructively. Permanent magnets are used to instantaneously magnetize the wire rope, and leakage flux signals generated in the magnetized state are detected and analyzed through sensors; however, leakage flux signals may appear when the magnetized material is not homogeneous. Therefore, preprocessing is required to remove signal distortion caused by large fluctuations or distortion caused by high-frequency signals from the input signal.

[0005] Problems arise where fault signals are not clearly revealed due to environmental noise and trends in the raw leakage flux signal, and where fault detection accuracy is improved by eliminating signal distortion caused by the rope's dynamic state (ascending, descending, stopping, etc.). Additionally, since it is difficult to obtain consistent data by normalizing signal imbalances caused by differences in distance and speed between the sensor and the rope, a method is required to overcome these issues.

[0006] The problems solved by the present invention are not limited to those mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the description below.

[0007] As a means for solving the aforementioned technical problem, according to an embodiment of the present invention, the method comprises: a step of removing channel-specific trends by removing low-frequency signals below a threshold value after obtaining signal data from a plurality of Hall sensors; a step of removing channel-specific noise by determining high-frequency signals above a threshold value as noise and removing them from the signal that has undergone the trend removal step; a step of detecting change points by eliminating mutual influence according to the upward, downward, and stopping movements of the wire rope to distinguish situations and calculate information for detecting defects; and a step of performing a Hilbert transform to make the sign of the signal identical based on a threshold to detect defects in the wire rope.

[0008] The step of removing the above-mentioned channel-specific trends includes a high-pass filter that blocks low-frequency components of the signal and passes high-frequency components after acquiring data from a plurality of Hall sensors.

[0009] The step of removing the above-mentioned channel-specific trend includes a process of stabilizing data-centered fluctuations by modeling the trend through the calculation of the average value of a continuous interval after acquiring data from multiple Hall sensors, and correcting the baseline and average value.

[0010] The step of removing noise per channel described above includes a band filtering process that removes specific bands of the signal to allow only the bands containing defective signals to pass through, or a process of removing noise bands by analyzing the frequency spectrum of each channel through the Fast Fourier Transform. Here, it includes a process of generating a complex signal by removing negative frequency components after transforming the frequency domain through the Fast Fourier Transform, and then determining the transformation in the time domain through the Inverse Fourier Transform.

[0011] The above change point detection step includes a process of detecting a point in time when the values ​​of the distribution, mean, or variance of a signal or data that changes over time suddenly change.

[0012] By applying the technology according to the present invention, environmental noise and trends are removed from the leakage flux signal, thereby distinguishing the dynamic state of the rope and ensuring the reliability of defect detection under the same conditions, which improves the accuracy of defect detection. Defects can be visualized via a radar chart to intuitively identify their location and intensity. Furthermore, data-based defect detection automation reduces user dependency, and signal normalization provides consistent results across various sensor placements and environments, supporting efficient maintenance.

[0013] Fig. 1 is a conventional wire rope diagnostic device

[0014] FIG. 2 is a conceptual diagram graphically representing the leakage flux signal in a wire rope.

[0015] Figure 3 is a signal preprocessing method according to the present invention

[0016] FIG. 4 is a combined radar chart visualizing the magnitude of the combined signal according to the present invention.

[0017] FIG. 5 is a signal graph of a diagnostic device according to the present invention.

[0018] Further objects, features, and advantages of the present invention can be more clearly understood from the following detailed description and the accompanying drawings.

[0019] Before providing a detailed description of the present invention, it should be understood that the present invention is capable of various modifications and may have various embodiments, and that the examples described below and illustrated in the drawings are not intended to limit the present invention to specific embodiments, but rather include all modifications, equivalents, and substitutions that fall within the spirit and scope of the present invention.

[0020] A signal preprocessing method according to the present invention is described with reference to FIG. 2. After obtaining data from 22 Hall sensors, a channel-specific trend removal step is performed to remove low-frequency signals and eliminate trends where the signal fluctuates significantly. Subsequently, a channel-specific noise removal step is performed to remove high-frequency signals and eliminate noise that interferes with coupling judgment. Afterward, a change point detection step is performed to eliminate mutual influences in detecting defects by distinguishing identical situations according to the movement of the rope, followed by a Hilbert transform step to make the signal codes identical in order to perform defect detection based on Threshold.

[0021] Trend removal is a technique that eliminates long-term fluctuation trends (low-frequency components) from time-series or signal data to more clearly reveal local and valid signals (defects, anomalies, etc.). Trends appear as data averages, baseline changes, or low-frequency elements; removing them is essential during signal analysis because they can obscure or distort important defect signals.

[0022] Specifically, after acquiring data from 22 Hall sensors, a high-pass filter may be used to block low-frequency components (trends) of the signal and allow high-frequency components (defect signals) to pass through. Alternatively, after acquiring data from 22 Hall sensors, the signal's baseline or average value can be corrected to stabilize fluctuations in the data center by modeling the trend through a process of calculating the average value of consecutive data segments. Another method may be to use polynomial fitting, which models the data trend as a polynomial and then removes it from the original data. A wavelet transform method may also be applied to precisely separate the trend and the valid signal by adjusting the time-frequency resolution. Whether using hardware or software methods, the appropriate method for the field can be selected and applied.

[0023] Channel-specific noise removal technology is a process that analyzes signal data collected from multiple sensors (or channels) individually to minimize or eliminate noise generated in each channel. This is a data processing step designed to resolve the problem of critical information (such as defective or abnormal signals) being distorted by noise in signal processing.

[0024] Specifically, bandpass filtering, which removes specific bands of a signal and allows only the bands containing defective signals to pass through, and adaptive filtering, which removes noise by adjusting filter coefficients in real time according to sensor noise characteristics, can be used. Alternatively, frequency domain analysis techniques can be applied to remove noise bands by analyzing the frequency spectrum of each channel through the Fast Fourier Transform (FFT). Recently, machine learning-based noise removal technologies have been proposed, and such technologies may also be applied to the present invention.

[0025] The data, having undergone trend removal and noise removal for each of the aforementioned channels, is processed in a step that handles data change values ​​by reflecting the moment a significant change occurs, as a process designed to eliminate the influence of rope movement. This is the change point detection step, which identifies important change points in the data.

[0026] Specifically, this is used to detect the point in time when statistical characteristics, such as the distribution, mean, and variance of signals or data that change over time, suddenly change. For this purpose, statistical methods may be used, or it can be implemented using a machine learning-based method. In the present invention, changes in data according to the state of the rope, such as ascent, descent, and stop, are distinguished based on machine learning.

[0027] Defects are determined by using data containing information on detected change points and noise removal as input. To achieve this, the detected signal data is subjected to a Hilbert transform. The Hilbert transform is a widely used mathematical transformation in signal processing that generates a symmetric complex signal from a real signal to analyze the signal's envelope and phase. This can be implemented by shifting the phase of the signal's frequency components by -90°. More specifically, using a frequency domain-based method, the input signal can be transformed into the frequency domain via the Fourier Transform (FFT), negative frequency components are removed to generate a complex signal, and then transformed into the time domain via the Inverse Fourier Transform (IFFT).

[0028] Here, the amplitude and phase of the signal are calculated using the following formula. Here, A(t) is the signal amplitude value, and φ(t) is the calculated phase value.

[0029]

[0030]

[0031] Hereinafter, specific technical details to be implemented in the present invention will be described in detail with reference to the attached drawings.

[0032] FIG. 1 is a conventional wire rope diagnostic device. As a diagnostic device according to the cited invention (Korean Registered Patent No. 10-2219641), the sensor part of the wire rope diagnostic device may include a magnet part that forms a main magnetic flux path to include a set section on the axial direction of the wire rope, a Hall sensor part that detects leakage magnetic flux generated from a damaged part of the wire rope when the wire rope is magnetized by the magnet part, one or more yokes that fix the magnet part and the Hall sensor part, and a flexible printed circuit board (FPCB) between the Hall sensor part and the one or more yokes.

[0033] Figure 2 is a conceptual diagram graphically representing the leakage flux signal in a wire rope. The magnets above and below instantaneously magnetize the wire rope, and the safety status of the wire rope is verified by analyzing the leakage magnetic signal. Here, if there is damage to the wire, the leakage flux is indicated by a dotted line. Here, the magnitude of data detection varies depending on the distance and speed between the Hall sensor and the rope. The amount of signal change is expressed as a channel-by-channel radar chart, and Figure 4 is an example diagram showing the radar chart representation.

[0034] Figure 5 is an example graph showing the leakage flux signal of a wire rope. Data normalization was performed based on 1, ensuring that the value does not exceed 1 in normal cases, while in the case of defective data, the value exceeds 1 at the defective area.

Claims

1. In a signal preprocessing method for wire rope defect diagnosis, A step of removing channel-specific trends by removing low-frequency signals below a threshold value after obtaining signal data from multiple Hall sensors; A step for removing channel-specific noise in which the signal that has undergone the above trend removal step determines high-frequency signals above a threshold value as noise and removes them; A step of detecting change points to eliminate mutual influence according to the upward, downward, and stopping movements of the wire rope, thereby distinguishing situations and generating information for detecting defects; A signal pretreatment method for wire rope defect diagnosis comprising: a step of performing a Hilbert transform to make the sign of the signal identical based on a threshold to detect defects in the wire rope.

2. In Paragraph 1, The step of removing the above channel-specific trends is A signal pretreatment method for wire rope coupling diagnosis comprising: a high-pass filter that blocks low-frequency components and passes high-frequency components of a signal after acquiring data from a plurality of Hall sensors.

3. In Paragraph 1, A signal pretreatment method for wire rope coupling diagnosis, wherein the step of removing the above-mentioned channel-specific trend includes a process of stabilizing data-centered fluctuations by modeling the trend after acquiring data from multiple Hall sensors by calculating the average value of a continuous interval and correcting the baseline and the average value.

4. In Paragraph 1, The above-mentioned step of removing channel-specific noise is a signal pretreatment method for wire rope coupling diagnosis that includes a band filtering process to remove specific bands of the signal and allow only the bands containing the defective signal to pass through.

5. In Paragraph 1, The step of removing channel-specific noise described above is a signal pretreatment method for wire rope coupling diagnosis that includes the process of removing noise bands by analyzing the frequency spectrum of each channel through the Fast Fourier Transform.

6. In Paragraph 5, A signal pretreatment method for wire rope coupling diagnosis comprising the process of generating a complex signal by removing negative frequency components after transforming the frequency domain through the above Fast Fourier Transform, and determining the transformation of the time domain through the Inverse Fourier Transform.

7. In Paragraph 1, The above-mentioned change point detection step is a signal pretreatment method for wire rope coupling diagnosis that includes a process of detecting a point in time when the values ​​of the distribution, mean, or variance of a signal or data changing over time suddenly change.

Images