Edge cloud computing cooperative processing method and system for steel wire rope detection data

Abstract

The invention discloses an edge cloud computing cooperative processing method and system for steel wire rope detection data, and relates to the technical field of steel wire rope detection.The method includes the steps that original monitoring data of multiple collection points are preprocessed, a steel wire rope short-time energy data set is constructed, and a noise fusion energy feature Efusion is obtained through calculation; a normal section and an abnormal noise section are accurately judged in combination with an energy threshold Efth generated based on extremely cold environment historical data statistics, and meanwhile noise section subdivision is further achieved through an ice hanging energy threshold EP2 h, a wind shear energy threshold EP2 wind and a tower pole resonance energy threshold EP3th. The method can effectively distinguish the ice hanging impact noise, the wind shear disturbance noise, the pole resonance noise and the normal operation section, and solves the problem that in the prior art, different noise sections are mixed into the same energy interval in the extremely cold environment and cannot be accurately identified, so that the accuracy and robustness of the operation noise judgment of the cableway steel wire rope in the extremely cold area are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of wire rope testing technology, specifically to an edge cloud computing collaborative processing method and system for wire rope testing data. Background Technology

[0002] With the rapid development of intelligent operation and maintenance technology, edge computing technology, and IoT monitoring technology, structural health monitoring based on real-time perception and distributed computing has gradually become an important direction in the field of mechanical equipment operation and maintenance. Among the many structural monitoring objects, the steel wire ropes of mountain cableways are directly related to the stability of the cableway transportation system and the safety of users' lives due to their long-term exposure to complex environments such as extreme cold, high winds, and icicles. Therefore, online real-time monitoring of the operating status of steel wire ropes has become one of the key application scenarios in structural health monitoring technology. In this type of monitoring scenario, collecting multi-source monitoring data such as magnetic flux density, vibration response, tower micro-vibration, and temperature changes through multiple sensors, and performing real-time analysis at edge computing nodes, is an important technical means to improve the operational safety of steel wire ropes.

[0003] Currently, existing online wire rope inspection technologies generally suffer from a lack of diverse data processing methods. They primarily employ fixed sampling frequencies, fixed analysis windows, and fixed feature extraction techniques, making it difficult to effectively differentiate between various types of non-structural disturbances that frequently occur in extremely cold environments, such as noise from icicle shearing, wind shear disturbances, and tower resonance. Furthermore, existing systems often directly upload all collected raw monitoring data to cloud servers for computation, resulting in a large amount of invalid data in high-noise environments. This not only consumes limited communication bandwidth but also significantly increases the cloud computing load, causing truly structural information relevant to wire rope damage to be submerged in the noise data, thus affecting the accurate extraction of wire rope damage features.

[0004] The fundamental reason for the aforementioned technical deficiencies lies in the fact that the operating environment of wire ropes in extremely cold mountainous areas is characterized by high noise intensity, diverse noise types, uneven noise temporal distribution, and significant differences in energy structure among different noise segments. If a fixed sampling and uniform analysis window are still used, it will be impossible to adapt to the characteristics of different noise segments. Inappropriate sampling resolution will cause high-frequency impact events such as icicles to be masked by low resolution, wind shear disturbances to be lost due to short window segmentation, and low-frequency changes such as tower resonance to be diluted or even misjudged at high sampling frequencies, making it difficult to effectively distinguish between noise segments and structural damage segments. This not only results in more than 60% of the uploaded data being noise or redundant information unrelated to the structure, but also increases the processing latency of the cloud server and reduces the accuracy of model recognition. Ultimately, this may lead to missed, false, or delayed detection of wire rope damage signals, thus posing a potential risk to the safe operation of the cableway system. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an edge cloud computing collaborative processing method and system for wire rope detection data, which solves the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution, comprising the following steps: S1. Set up collection points along the cableway's steel wire rope running path in the mountainous area to collect raw data in real time and upload it to the edge computing node. In the edge computing node, preprocess the raw data to obtain the short-time energy dataset of the steel wire rope. S2. In the edge computing node, the short-time energy dataset of the wire rope is fused to obtain the noise fusion energy feature Efusion. The energy threshold Efth is set and the noise fusion energy feature Efusion is compared and evaluated in advance. Based on the preliminary comparison and evaluation results, the noise segment label Ns is output. S3. Based on the noise segment label Ns, trigger the variable resolution processing operation, automatically select the corresponding resolution level according to the noise segment label Ns to construct the multi-scale structural feature vector Fraw, and obtain the structural mutation index Istruct based on the multi-scale structural feature vector Fraw. S4. Preset the structural mutation threshold Is, and use the structural mutation threshold Is and the structural mutation index Istruct to perform structural mutation assessment. Based on the structural mutation assessment results, proceed to the collaborative verification stage.

[0007] Preferably, S1 includes S11; S11. Multiple data collection points are set up along the cableway's steel cable running path, and sensor groups are installed at the data collection points to collect raw data in real time. The collected raw data is then transmitted to the edge computing node via the wireless transmission module of the corresponding sensor at the data collection point using a local area wireless network. The edge computing node is equipped with edge processing capabilities by being set up on the cableway tower platform and is connected to the cloud collaborative node to receive raw data in real time. The collection points include a first collection point P1, a second collection point P2, a third collection point P3, and a fourth collection point P4; The sensor group includes a magnetic flux sensor, a micro-vibration acceleration sensor, a micro-vibration pickup sensor, and a temperature change rate sensor; The raw data includes the magnetic flux density of the wire rope XP1, the surface vibration acceleration of the wire rope XP2, the micro-vibration acceleration of the tower structure XP3, and the temperature change rate of the tower surface XP4.

[0008] Preferably, S1 further includes S12; S12. Preprocess the raw data in the edge computing node to obtain the short-time energy dataset of the wire rope; The preprocessing includes timestamp alignment, short-time energy extraction, and normalization. The short-time energy dataset of the wire rope includes the short-time energy EP1(t) of the first acquisition point P1 at time t, the short-time energy EP2(t) of the second acquisition point P2 at time t, the short-time energy EP3(t) of the third acquisition point P3 at time t, and the short-time energy EP4(t) of the fourth acquisition point P4 at time t. The timestamp alignment is achieved by setting a unified clock in the edge computing nodes. When the raw data is received, all parameters in the raw data are timestamped according to the unified clock. The short-time energy extraction is performed by extracting the short-time energy of the corresponding collection point using the short-time energy calculation formula. The standardization process uses the z-score standardization method to standardize all extracted short-time energies, eliminating the influence of unit dimensions, and then summarizing them to obtain a short-time energy dataset of the wire rope.

[0009] Preferably, S2 includes S21; S21. Based on the short-time energy dataset of the wire rope, construct an energy calculation model, and input the real-time acquired short-time energy dataset of the wire rope into the energy calculation model to perform noise energy fusion and obtain the noise fusion energy feature Efusion. The noise fusion energy characteristic Efusion is obtained through the following energy calculation model; In the formula, Efusion(t) represents the noise fusion energy characteristics at time t, and a1, a2, a3 and a4 represent the preset weight values ​​of the short-time energy EP1 of the first acquisition point P1, the short-time energy EP2 of the second acquisition point P2, the short-time energy EP3 of the third acquisition point P3 and the short-time energy EP4 of the fourth acquisition point P4, respectively, and a1+a2+a3+a4=1, the specific values ​​of which are set by the user.

[0010] Preferably, S2 further includes S22; S22. Construct an energy calculation model based on the short-time energy dataset of steel wire ropes under historical extreme cold conditions to obtain the short-time energy distribution range under extreme cold conditions, and determine the energy threshold Efth that distinguishes between normal segments and abnormal noise segments based on the short-time energy distribution range. A preliminary comparison and evaluation is conducted based on the acquired energy threshold Efth and the real-time acquired noise fusion energy characteristic Efusion to determine the abnormal noise segment of the current time period of the wire rope. The specific evaluation content is as follows: When the noise fusion energy characteristic Efusion(t) at time t is less than the energy threshold Efth, the output noise label Ns(t) at time t is a normal segment, indicating that the current time period of the wire rope is a normal segment; When the noise fusion energy feature Efusion(t) at time t is greater than or equal to the energy threshold Efth, it indicates that the current time period of the wire rope is an abnormal noise segment, and at this time, the noise segment label Ns(t) at time t is generated. The noise segment label Ns(t) at time t is generated using the following method: Based on the energy distribution characteristics of the short-time energy dataset of the wire rope at time t, the noise segment is subdivided, where: When the short-time energy EP2(t) of the second collection point P2 at time t is greater than the ice hanging energy threshold EP2h, the noise segment label Ns(t) at time t is determined to be the ice hanging noise segment; When the short-time energy EP2(t) of the second acquisition point P2 at time t is greater than the wind shear energy threshold EP2wind and the short-time energy EP3(t) of the third acquisition point P3 at time t is less than the tower resonance energy threshold EP3th, the noise segment label Ns(t) at time t is determined to be the wind shear disturbance segment. When the short-time energy EP3(t) of the third acquisition point P3 at time t is greater than the tower resonance energy threshold EP3th, the noise segment label Ns(t) is determined to be the tower resonance segment.

[0011] Preferably, S3 includes S31; S31. After obtaining the noise segment label Ns(t) at time t, the edge computing node automatically performs variable resolution processing based on the noise segment label Ns(t) at time t; the specific operation is as follows: Wherein: when the noise segment label Ns(t) at time t is the ice hanging noise segment, the sampling frequency fs in the range of 9kHz-20kHz is selected as the high-frequency resolution; When the noise segment label Ns(t) at time t is the wind shear disturbance segment, the sampling frequency fs in the range of 3kHz-8kHz is selected as the intermediate frequency resolution; When the noise segment label Ns(t) at time t is the tower resonance segment, the sampling frequency fs in the range of 500Hz-2kHz is selected as the low-frequency resolution. When the noise segment label Ns(t) at time t is a normal segment, select the sampling frequency fs in the range of 100Hz-500Hz, which corresponds to the energy-saving mode with the lowest sampling resolution. The edge computing node performs frequency domain analysis and time domain analysis on the wire rope detection signal according to the sampling frequency fs of the corresponding interval range, obtains the time domain feature set TD and the frequency domain feature set FD corresponding to the current noise segment, and constructs the multi-scale feature vector Fraw(t) corresponding to the current noise segment at time t based on the time domain feature set TD and the frequency domain feature set FD. The multi-scale feature vector Fraw(t) at time t is specifically and formally expressed as: Fraw(t) = TD∪FD.

[0012] Preferably, S3 further includes S32; S32. After obtaining the multi-scale feature vector Fraw(t) at time t, the edge computing node performs difference calculation on the multi-scale feature vector Fraw(t) at time t and the multi-scale feature vector Fraw(t−Δt) of the previous feature update cycle based on the first-order difference structural catastrophe analysis method in the existing technology, and obtains the structural catastrophe index Istruct through the following structural catastrophe formula: Istruct(t) = |Fraw(t)−Fraw(t−Δt)|; where Istruct(t) represents the structural mutation index at time t.

[0013] Preferably, S4 includes S41; S41. Based on the statistical distribution of the difference amplitude of the multi-scale feature vectors of the stable section of the wire rope structure in long-term historical operating data, the edge computing node extracts the reference difference interval of the structural mutation index Istruct during the stable period, and obtains the structural threshold Is for the critical degree of structural mutation. After obtaining the structural mutation index Istruct(t) at time t, the edge computing node performs structural mutation assessment by comparing the structural mutation index Istruct(t) at time t with the structural threshold Is to determine the current structural mutation status of the wire rope. The specific assessment content is as follows: When the structural mutation index Istruct(t) at time t is greater than or equal to the structural threshold Is, it is determined that there is a significant mutation in the current wire rope structural state and a structural mutation assessment result is generated as a mutation has occurred. When the structural mutation index Istruct(t) at time t is less than the structural threshold Is, the current wire rope structure is determined to be stable and the structural mutation assessment result is generated as no mutation.

[0014] Preferably, S4 further includes S42; S42. When the structural mutation assessment result indicates that a mutation has occurred, the edge computing node sends a structural anomaly reporting information to the cloud. The cloud generates a collaborative control command based on the energy change range corresponding to different structural states in long-term historical operating data and sends it to the edge computing node to execute the cloud collaborative control mechanism triggered by the edge computing node and the cloud collaborative node. The cooperative control command increases the determination energy threshold Efth of the noise segment label Ns(t) at time t by 5%-20% compared to the original value; This allows edge computing nodes to increase the sampling frequency fs range by 10%-30% in subsequent monitoring cycles. After receiving the collaborative control command, the edge computing node automatically updates its local parameters and performs monitoring and control according to the updated threshold and sampling resolution in subsequent cycles. When the structural mutation assessment result is no mutation, the edge computing node does not perform parameter adjustment, but only records the structural state of the current period for subsequent periodic calibration, without triggering the cloud collaborative control mechanism.

[0015] The edge cloud computing collaborative processing system for steel wire rope detection data includes an edge acquisition module, a short-time energy analysis module, a resolution processing module, and a cloud collaboration module; The edge acquisition module collects raw data in real time by setting up acquisition points along the operating path of the cableway steel wire rope in the mountainous area, and uploads it to the edge computing node. The raw data is preprocessed in the edge computing node to obtain the short-time energy dataset of the steel wire rope. The short-time energy analysis module performs fusion processing on the short-time energy dataset of the wire rope in the edge computing node to obtain the noise fusion energy feature Efusion, sets the energy threshold Efth and performs a preliminary comparison and evaluation with the noise fusion energy feature Efusion, and outputs the noise segment label Ns based on the preliminary comparison and evaluation results. The resolution processing module triggers a variable resolution processing operation based on the noise segment label Ns, automatically selects the corresponding resolution level according to the noise segment label Ns to construct a multi-scale structural feature vector Fraw, and obtains the structural mutation index Istruct based on the multi-scale structural feature vector Fraw. The cloud-based collaborative module performs structural mutation assessment by setting a structural mutation threshold Is and combining the structural mutation threshold Is with the structural mutation index Istruct. Based on the structural mutation assessment results, it then enters the collaborative verification stage.

[0016] This invention provides an edge computing collaborative processing method and system for steel wire rope detection data. It has the following beneficial effects: (1) This method constructs a short-time energy dataset for steel wire ropes by performing timestamp alignment, short-time energy extraction, and standardization on raw monitoring data from multiple acquisition points in edge computing nodes. It then obtains the noise fusion energy feature Efusion based on a short-time energy fusion model, and combines this with the energy threshold Efth generated from historical data of extremely cold environments to accurately distinguish between normal and abnormal noise segments. Furthermore, it utilizes the icefall energy threshold EP2h, wind shear energy threshold EP2wind, and tower resonance energy threshold EP3th to further subdivide noise segments. This method effectively distinguishes between icefall impact noise, wind shear disturbance noise, tower resonance noise, and normal operating segments, solving the problem in existing technologies where different noise segments are mixed in the same energy range in extremely cold environments, leading to inaccurate identification. This significantly improves the accuracy and robustness of determining the operating noise of cableway steel wire ropes in extremely cold regions.

[0017] (2) After obtaining the noise segment label Ns(t), the edge computing node automatically selects the matching sampling frequency fs interval according to different noise segment types, realizing adaptive switching between high frequency, medium frequency, low frequency and energy-saving mode. At the corresponding sampling resolution, the time domain feature set TD and frequency domain feature set FD are extracted to construct the multi-scale structural feature vector Fraw, and then the structural mutation index Istruct is obtained based on the first-order difference analysis method. This innovation avoids the problem of loss of structural mutation features caused by "fixed sampling frequency and fixed analysis window length" in the existing technology, so that the high frequency pulse of icefall impact, the medium frequency disturbance of wind shear and the low frequency resonance feature of tower can be accurately captured at the optimal resolution, thereby significantly improving the extraction sensitivity and analysis accuracy of the structural features of wire rope operation.

[0018] (3) This method achieves a complete edge-cloud collaborative processing link by cascading the noise segment identification model with variable resolution structural feature extraction, from automatic noise segment identification, adaptive resolution switching, structural mutation index calculation to cloud-based collaborative verification and dynamic threshold adjustment. The cloud dynamically adjusts the threshold Efth and sampling resolution range by 5%–20% and 10%–30% based on historical structural data, enabling the edge detection strategy to converge and optimize in real time as the environment changes. This collaborative mechanism can filter out more than 60% of invalid and structure-irrelevant data, reporting only information related to structural mutations. This not only significantly reduces communication bandwidth usage and cloud computing pressure, but also enhances the effective signal ratio of the wire rope damage feature map, improves the detectability rate of structural anomalies, and solves the problems of extreme cold noise masking damage signals and fixed sampling mode diluting structural information in existing technologies. It achieves a comprehensive technical effect of reducing data volume, improving feature quality, and simultaneously enhancing the ability to detect structural changes. Attached Figure Description

[0019] Figure 1This is a schematic diagram of the edge cloud computing collaborative processing method for steel wire rope detection data according to the present invention; Figure 2 This is a schematic diagram of the edge cloud computing collaborative processing system for steel wire rope detection data according to the present invention; Figure 3 A schematic diagram of the sensor deployment structure along the cableway's wire rope running path in mountainous areas. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example 1

[0021] Please see Figure 1 This invention provides an edge computing collaborative processing method for steel wire rope detection data. To achieve the above objectives, this invention employs the following technical solution, comprising the following steps: S1. Set up collection points along the cableway's steel wire rope running path in the mountainous area to collect raw data in real time and upload it to the edge computing node. In the edge computing node, preprocess the raw data to obtain the short-time energy dataset of the steel wire rope. S2. In the edge computing node, the short-time energy dataset of the wire rope is fused to obtain the noise fusion energy feature Efusion. The energy threshold Efth is set and the noise fusion energy feature Efusion is compared and evaluated in advance. Based on the preliminary comparison and evaluation results, the noise segment label Ns is output. S3. Based on the noise segment label Ns, trigger the variable resolution processing operation, automatically select the corresponding resolution level according to the noise segment label Ns to construct the multi-scale structural feature vector Fraw, and obtain the structural mutation index Istruct based on the multi-scale structural feature vector Fraw. S4. Preset the structural mutation threshold Is, and use the structural mutation threshold Is and the structural mutation index Istruct to perform structural mutation assessment. Based on the structural mutation assessment results, proceed to the collaborative verification stage.

[0022] In this embodiment, the edge computing collaborative processing method for wire rope detection data forms an adaptively evolving detection link through continuous operations S1 to S4. First, in S1, short-time energy is performed on the raw data from multiple acquisition points because noise in extremely cold mountainous areas is characterized by "suddenness, randomness, and intermittency." If the raw waveform is used directly, the impact of icicles will be smoothed out by long windows, and the resonance of towers will be diluted by high resolution, causing the true structural changes to be masked by noise. Short-time energy is equivalent to capturing "instantaneous activity," accurately reflecting the energy characteristics of different disturbances such as icicles, wind shear, and resonance within a millisecond window. It is the most fundamental quantity for distinguishing noise types in extremely cold environments. In S2, weighted fusion of the four types of short-time energy and setting an energy threshold Efth are performed to solve the problem of "false high energy" caused by the simultaneous superposition of multiple noises. For example, wind shear continuously pushes up mid-frequency energy, while icicle shearing instantly raises high-frequency energy; the superposition of the two can cause the system to misjudge structural anomalies. By fusing the model and segmented thresholds, it becomes clear who is raising the energy, fundamentally avoiding false alarms, and promptly outputting noise segment labels Ns, enabling the system to understand the current noise environment. In S3, the sampling resolution is automatically switched based on the noise segment label Ns to avoid the "inadequate sampling" problem caused by a fixed sampling mode. For example, icefall impacts can only be fully captured at 9kHz–20kHz; if a low sampling rate is still used, the pulse energy will be distorted, leading to a distortion of the structural mutation index Istruct. Conversely, if a high sampling rate is used for low-frequency resonance of towers, the low-frequency envelope will become indistinct. The significance of variable resolution is that each noise segment is analyzed at the resolution that "best shows it," thereby realistically constructing a multi-scale structural feature vector Fraw, and accurately reflecting whether a structural state mutation has occurred through first-order difference. In S4, the structural mutation threshold Is is used to determine whether the structure has changed, and further triggers cloud-based collaborative verification, aiming to avoid the "false alarm risk caused by isolated judgments at the edge." For example, a single wind shear event might cause a short-term energy spike, but if it doesn't last, it's not considered a structural anomaly; conversely, a low-frequency resonance that doesn't match historical models might indicate tower loosening. By combining edge detection (Is) with secondary comparison in the cloud, real-time decisions can be integrated with long-term trends, achieving a higher confidence level in identifying structural anomalies. Example 2

[0023] Please see Figure 1 and Figure 3 Specifically: S1 includes S11; S11. Multiple data collection points are set up along the cableway's steel cable running path, and sensor groups are installed at the data collection points to collect raw data in real time. The collected raw data is then transmitted to the edge computing node via the wireless transmission module of the corresponding sensor at the data collection point using a local area wireless network. The edge computing node receives the raw data in real time through a cloud connection with a cloud collaborative node set up on the cableway tower platform that has edge processing capabilities. The data collection points include the first data collection point P1, the second data collection point P2, the third data collection point P3, and the fourth data collection point P4; The sensor group includes a magnetic flux sensor, a micro-vibration acceleration sensor, a micro-vibration pickup sensor, and a temperature change rate sensor; The raw data include the magnetic flux density of the wire rope XP1, the surface vibration acceleration of the wire rope XP2, the micro-vibration acceleration of the tower structure XP3, and the temperature change rate of the tower surface XP4; Among them: the first acquisition point P1 is installed in the middle area of ​​the cableway wire rope bearing section and is equipped with a magnetic flux density sensor to obtain the magnetic flux density XP1 of the wire rope; The second acquisition point P2 is installed on the outer surface of the cable car support or the wire rope and is equipped with a micro-vibration acceleration sensor to acquire the vibration acceleration XP2 of the wire rope surface. The third acquisition point P3 is installed in the low-frequency vibration sensitive area of ​​the wire rope connection base or tower support structure, and is equipped with a micro-vibration pickup sensor to acquire the micro-vibration acceleration XP3 of the tower structure. The fourth sampling point P4 is set up on the surface of the tower and equipped with a low temperature temperature change rate sensor to obtain the temperature change rate XP4 on the surface of the tower.

[0024] S1 also includes S12; S12. Preprocess the raw data in the edge computing node to obtain the short-time energy dataset of the wire rope; Preprocessing includes timestamp alignment, short-time energy extraction, and normalization. The short-time energy dataset of the wire rope includes the short-time energy EP1(t) of the first acquisition point P1 at time t, the short-time energy EP2(t) of the second acquisition point P2 at time t, the short-time energy EP3(t) of the third acquisition point P3 at time t, and the short-time energy EP4(t) of the fourth acquisition point P4 at time t. Timestamp alignment is achieved by setting a unified clock in edge computing nodes. When raw data is received, all parameters in the raw data are timestamped according to the unified clock. Short-time energy extraction is performed by extracting the short-time energy of the corresponding acquisition points using the short-time energy calculation formula. The standardization process uses the z-score standardization method to standardize all extracted short-time energies, eliminating the influence of unit dimensions, and then summarizing them to obtain the wire rope short-time energy dataset. The formula for calculating short-time energy is: Where T represents the length of the acquisition time window, which is set by the user; EPn(t) represents the short-time energy corresponding to the nth acquisition point at time t; and Xn(t) represents the raw data corresponding to the nth acquisition point at time t. The essential meaning of short-time energy is that it measures the degree of instantaneous change of a signal within a very short time interval, reflecting: signal strength, whether an impact event has occurred, whether vibration has increased, and whether there has been a sudden change in magnetic flux; it can be simply understood as: "how strong and active the signal is within this small window"; Short-time energy is a commonly used analytical metric in signal processing, used to measure the instantaneous intensity of a signal within a very short time window. Essentially, it is a statistical measure of the sum of the squares of the signal amplitude within that window. When steel wire ropes operate in extremely cold mountainous areas, they are affected by various noise segments, including impacts from falling ice, wind shear disturbances, low-frequency resonance of the tower, and normal operation. Each noise segment exhibits significant differences in signal energy structure. For example, the ice noise segment is characterized by peak pulse energy, the wind shear disturbance segment by a long-term rise in mid-to-high frequency energy, the tower resonance segment by a continuous increase in low-frequency energy, while the normal segment maintains stable energy. Therefore, short-time energy can serve as an effective indicator for distinguishing different noise segment types.

[0025] In this embodiment, the method deploys a first collection point P1, a second collection point P2, a third collection point P3, and a fourth collection point P4 along the operating path of the cableway wire rope in the mountainous area. These points are equipped with magnetic flux density sensors, micro-vibration acceleration sensors, micro-vibration pickup sensors, and temperature change rate sensors, respectively. This allows for the simultaneous acquisition of multiple raw data sources, including changes in magnetic flux within the wire rope, surface vibration of the wire rope, vibration of the tower structure, and disturbances from extremely cold temperatures. The aim is to avoid misinterpreting wind shear, tower vibration, or icicle detachment as abnormalities in the wire rope using a single-parameter monitoring mode. For example, relying solely on magnetic flux density XP1, slight fluctuations in magnetic flux caused by tower resonance could be falsely reported as wire rope damage. By deploying multiple collection points, "external disturbances" and "changes within the wire rope itself" can be effectively separated physically. Subsequently, timestamp alignment is performed in the edge computing nodes using a unified clock, ensuring that XP1, XP2, XP3, and XP4 from different acquisition points have a consistent time reference at the same moment, avoiding "event misalignment" caused by clock drift at the acquisition points. Without time alignment, the same icefall event might occur tens of milliseconds earlier in P2, leading to complete distortion of subsequent feature fusion. Next, a short-time energy formula is used to extract short-time energy from the raw signals of different acquisition points, quantifying signal strength, impact amplitude, vibration activity, and magnetic flux abrupt changes within a millisecond window. This makes the differences between the peak energy of icefall impact, the sustained mid-frequency rise of wind shear, the low-frequency enhancement of tower resonance, and the stable characteristics of normal operation very prominent. Short-time energy is equivalent to a physical characterization of "instantaneous activity," avoiding the masking of impact events by long-window averaging. Finally, z-score standardization brings EP1(t)–EP4(t) to a unified dimension, preventing the inherent dominance of a certain indicator due to differences in sensor physical units, ensuring fair and effective subsequent fusion processing. Through the above implementation methods, the short-time energy dataset of the wire rope can accurately reflect the real impact of different disturbances on the structural state of the wire rope under extremely cold conditions, providing stable, reliable and physically meaningful basic data for subsequent noise segment identification, variable resolution selection and structural change judgment; in terms of engineering effect, it significantly reduces noise misjudgment, improves the accuracy of wire rope operation status identification under extremely cold conditions, and enhances the stability and sensitivity of subsequent structural anomaly detection. Example 3

[0026] Please see Figure 1 Specifically: S2 includes S21; S21. Based on the short-time energy dataset of the wire rope, construct an energy calculation model, and input the real-time acquired short-time energy dataset of the wire rope into the energy calculation model to perform noise energy fusion and obtain the noise fusion energy feature Efusion. The noise fusion energy characteristic (Efusion) is obtained through the following energy calculation model; In the formula, Efusion(t) represents the noise fusion energy characteristics at time t, and a1, a2, a3 and a4 represent the preset weight values ​​of the short-time energy EP1 of the first acquisition point P1, the short-time energy EP2 of the second acquisition point P2, the short-time energy EP3 of the third acquisition point P3 and the short-time energy EP4 of the fourth acquisition point P4, respectively, and a1+a2+a3+a4=1, the specific values ​​of which are set by the user; The short-time energy calculation formula used in this paper is derived from the classic short-time energy model in the field of signal processing. It is a fundamental formula widely used in mathematical and engineering signal processing, defined as the sum of the squares of the signal amplitude within a short time window centered at t, reflecting the instantaneous intensity of the signal during that time period. Based on this principle, this formula calculates the short-time energy for all acquisition points, obtaining a short-time energy dataset for the wire rope. Furthermore, this application constructs a weighted summation model of noise fusion energy characteristics based on multi-sensor weighted linear fusion theory. This model characterizes the contribution of different acquisition points to the overall energy and can be set or dynamically updated based on historical operating data in extremely cold environments or cloud-trained models. The aforementioned weighted fusion method is a modification of the classic linear fusion model. The engineering application of the fusion model can convert the short-time energy indices of different types of sensors into a unified energy scale for the assessment of overall noise intensity. Although the original physical dimensions of the four types of short-time energy are not consistent (including magnetic flux density, vibration acceleration, micro-vibration and temperature change rate, etc.), short-time energy, as a statistical quantity of summation of squares, is essentially used to represent signal strength and has no direct correspondence with the absolute original dimensions. Through the dimensional compensation effect of the weight parameters, EP1(t)–EP4(t) can be fused under a unified calculation framework, so that Efusion(t) has dimensionless or engineering scale meaning, thereby meeting the requirement of dimensional consistency on both sides of the formula and ensuring that the fusion result has the equivalent expression capability of actual signal energy.

[0027] S2 also includes S22; S22. Construct an energy calculation model based on the short-time energy dataset of steel wire ropes under historical extreme cold conditions to obtain the short-time energy distribution range under extreme cold conditions, and determine the energy threshold Efth that distinguishes between normal segments and abnormal noise segments based on the short-time energy distribution range. A preliminary comparison and evaluation is conducted based on the acquired energy threshold Efth and the real-time acquired noise fusion energy characteristic Efusion to determine the abnormal noise segment of the current time period of the wire rope. The specific evaluation content is as follows: When the noise fusion energy characteristic Efusion(t) at time t is less than the energy threshold Efth, the output noise label Ns(t) at time t is a normal segment, indicating that the current time period of the wire rope is a normal segment; When the noise fusion energy feature Efusion(t) at time t is greater than or equal to the energy threshold Efth, it indicates that the current time period of the wire rope is an abnormal noise segment, and at this time, the noise segment label Ns(t) at time t is generated. The noise segment label Ns(t) at time t is generated using the following method: Based on the energy distribution characteristics of the short-time energy dataset of the wire rope at time t, the noise segment is subdivided, where: When the short-time energy EP2(t) of the second collection point P2 at time t is greater than the ice hanging energy threshold EP2h, the noise segment label Ns(t) at time t is determined to be the ice hanging noise segment; When the short-time energy EP2(t) of the second acquisition point P2 at time t is greater than the wind shear energy threshold EP2wind and the short-time energy EP3(t) of the third acquisition point P3 at time t is less than the tower resonance energy threshold EP3th, the noise segment label Ns(t) at time t is determined to be the wind shear disturbance segment. When the short-time energy EP3(t) of the third acquisition point P3 at time t is greater than the tower resonance energy threshold EP3th, the noise segment label Ns(t) is determined to be the tower resonance segment. The setting methods for the ice shear energy threshold EP2h, wind shear energy threshold EP2wind, and tower resonance energy threshold EP3th are as follows: the edge computing node constructs an energy calculation model based on the short-time energy dataset of steel wire rope under historical extreme cold environmental conditions. By statistically fitting the short-time energy distribution intervals corresponding to different noise events, the energy threshold ranges of the corresponding normal segment and abnormal noise segment are determined respectively. The critical value is taken as the relevant threshold. If it exceeds the threshold, it is an abnormal noise segment, and if it is below the threshold, it is a normal segment. The threshold is periodically sent to the edge computing node for the classification and determination of the noise segment at the current moment.

[0028] In this embodiment, the short-time energy datasets of the wire rope from P1–P4 are input into the energy calculation model for weighted fusion to obtain the noise fusion energy feature Efusion. The core of this approach is to address the problem of "multiple noise sources simultaneously amplifying the energy of a single sensor" in extremely cold mountainous areas. For example, the high-frequency pulses from falling ice can instantly increase EP2, while wind shear can raise the energy of the same channel for a prolonged period. If EP1–EP4 are not fused uniformly, a situation will arise where "a sudden increase in the energy of a certain sensor equals a misjudgment of structural abnormality." By introducing weighted linear fusion of a1–a4, it is possible to physically distinguish between "changes in the wire rope itself" and "external superimposed disturbances," allowing the noise fusion energy feature Efusion to represent the "overall noise intensity" rather than "arbitrary single-point anomaly," thus reducing the probability of misjudgment caused by sudden noise at the source. Subsequently, a preliminary comparison is made using the energy threshold Efth constructed based on historical data from extremely cold environments in S22 to establish the "upper limit of energy in the normal segment." In extremely cold environments, a sudden drop in temperature causes an overall increase in XP4, while slight tower swaying causes a periodic increase in XP3. Using a fixed threshold directly would result in numerous false alarms of "normal but colder" or "normal but windier." Therefore, the energy threshold Efth is set based on the distribution range of the actual environment, making the judgment closer to real-world conditions. Only when the noise fusion energy characteristic Efusion exceeds the energy threshold Efth does the noise segment subdivision judgment begin. The subdivision mechanism of the noise segment label Ns (ice shear segment, wind shear segment, tower resonance segment) uses EP2h, EP2wind, and EP3th because the energy structures of the three types of noise are completely different: ice shear is sudden and predominantly high-frequency, wind shear is continuous and predominantly mid-frequency, and tower resonance is slow and predominantly low-frequency. If only the noise fusion energy characteristic Efusion is relied upon, regardless of the type of noise, it would simply be described as "energy increasing," making further differentiation impossible. However, by comparing the characteristics of EP2(t) and EP3(t) at different energy thresholds, the type of noise can be immediately determined, avoiding the misreporting of "ice falling" as "steel wire rope structure jumping." Through the above implementation methods, the direct purpose of this processing step is to decompose the "energy anomaly caused by the mixing of different noise sources in extremely cold environments" into noise categories with clear physical meanings, allowing the system to complete a "noise purification" process before proceeding to subsequent analysis. The resulting improvements are: not only significantly reducing misjudgments caused by unstructured noise, but also providing accurate noise segment labels for the subsequent variable resolution processing, ensuring that subsequent analysis is always conducted based on the most suitable sampling resolution, thereby greatly improving the stability and accuracy of structural change detection. Example 4

[0029] Please see Figure 1 Specifically: S3 includes S31; S31. After obtaining the noise segment label Ns(t) at time t, the edge computing node automatically performs variable resolution processing based on the noise segment label Ns(t) at time t; the specific operation is as follows: Wherein: when the noise segment label Ns(t) at time t is the ice-hanging noise segment, the sampling frequency fs in the range of 9kHz-20kHz is selected as the high-frequency resolution. When the noise segment label Ns(t) at time t is the wind shear disturbance segment, the sampling frequency fs in the range of 3kHz-8kHz is selected as the intermediate frequency resolution; When the noise segment label Ns(t) at time t is the tower resonance segment, the sampling frequency fs in the range of 500Hz-2kHz is selected as the low-frequency resolution. When the noise segment label Ns(t) at time t is a normal segment, select the sampling frequency fs in the range of 100Hz-500Hz, which corresponds to the energy-saving mode with the lowest sampling resolution. The edge computing node performs frequency domain analysis and time domain analysis on the wire rope detection signal according to the sampling frequency fs of the corresponding interval range, obtains the time domain feature set TD and the frequency domain feature set FD corresponding to the current noise segment, and constructs the multi-scale feature vector Fraw(t) corresponding to the current noise segment at time t based on the time domain feature set TD and the frequency domain feature set FD. The multi-scale feature vector Fraw(t) at time t is formally expressed as: Fraw(t) = TD∪FD; For example: the ice-cave noise segment Fraw(t) Includes: high-frequency pulse peak value, high-frequency energy, and high spectral bandwidth; Wind shear disturbance segment Fraw(t) Includes: mid-frequency energy stability, mid-frequency spectral morphology, and mid-frequency vibration duration; Tower resonance segment Fraw(t) Includes: low-frequency energy, low-frequency resonance period, and low-frequency envelope trend. Normal segment Fraw(t) Includes: operational stability, ultra-low frequency energy, and low variability indicators; It should be noted that the multi-scale feature vector Fraw(t) must be composed of "time domain features + frequency domain features" because the effect of a disturbance on a steel wire rope is not single-dimensional. Rather, the time domain reflects the "shape" of the disturbance (impact, waveform, transition); the frequency domain reflects the "components" of the disturbance (frequency band energy, spectral distribution, center frequency). If only a single domain is used, it is impossible to accurately characterize the physical phenomenon.

[0030] S3 also includes S32; S32. After obtaining the multi-scale feature vector Fraw(t) at time t, the edge computing node performs difference calculation on the multi-scale feature vector Fraw(t) at time t and the multi-scale feature vector Fraw(t−Δt) of the previous feature update cycle based on the first-order difference structural catastrophe analysis method in the existing technology, and obtains the structural catastrophe index Istruct through the following structural catastrophe formula: Istruct(t) = |Fraw(t)−Fraw(t−Δt)|; where Istruct(t) represents the structural mutation index at time t, which is used to characterize the degree of mutation of the wire rope structural features in the current time period.

[0031] In this embodiment, method S3 achieves a structural mutation analysis link of "segmentation first, then clear observation" through the linkage control of noise segment label Ns(t) and sampling frequency fs: After obtaining the noise segment label Ns(t) at time t, the edge computing node does not adopt a fixed sampling strategy indiscriminately, but selects sampling frequencies fs of 9kHz-20kHz, 3kHz-8kHz, 500Hz-2kHz, and 100Hz-500Hz for icefall noise segment, wind shear disturbance segment, tower resonance segment, and normal segment, respectively. The direct purpose of this setting is to avoid the problems of "seeing too coarsely and not being able to see details" or "seeing too closely and breaking the overall trend" in extremely cold mountainous areas: For example, icefall impact is mainly manifested as short-term high-frequency pulses. If low-frequency sampling is still used, the high-frequency details will be sampled as "fuzzy envelopes", which will weaken the real impact event; conversely, tower resonance is a slow low-frequency form. If an excessively high sampling frequency fs is used, the low-frequency trend will be cut into a large number of point data, which is not conducive to identifying the resonance mode. Through variable resolution processing driven by noise segment labels Ns(t), edge computing nodes perform time-domain and frequency-domain analysis on the wire rope detection signal in the most matching frequency band, extracting the time-domain feature set TD and the frequency-domain feature set FD, and constructing a multi-scale feature vector Fraw(t) = TD∪FD. This ensures that each type of noise segment has both "morphological information" (such as impact intensity, waveform envelope, and vibration duration) and "compositional information" (such as frequency band energy, spectral morphology, and resonance period), avoiding the problem of being unable to distinguish different frequency band sources when only looking at the time domain, and losing the timing and duration of events when only looking at the frequency domain. Building upon this foundation, the edge computing node employs the existing first-order differential structural mutation analysis method to differ the current multi-scale feature vector Fraw(t) from the multi-scale feature vector Fraw(t−Δt) of the previous feature update cycle, calculating the structural mutation index Istruct. Its true physical meaning is to quantify "how much the overall structural response of this window has changed compared to the previous window," thereby effectively distinguishing structurally related changes such as a sudden increase in icicles, a significant enhancement of wind disturbance, and amplified tower resonance from short-term random noise or measurement jitter. Through this process, on the one hand, the noise segment label Ns is used to select the correct "observation magnifying glass" at the frequency scale; on the other hand, the multi-scale feature vector Fraw and the structural mutation index Istruct accurately characterize structural state changes at the feature scale. Compared to the traditional fixed sampling + single-domain feature method, this not only significantly reduces the probability of missed and false detections of structural mutations but also improves the sensitivity and reliability of subtle structural anomalies in extremely cold environments, providing a stable and reliable input foundation for subsequent collaborative verification and edge-cloud linkage control based on the structural mutation threshold Is. Example 5

[0032] Please see Figure 1 Specifically: S4 includes S41; S41. Based on the statistical distribution of the difference amplitude of the multi-scale feature vectors of the stable section of the wire rope structure in long-term historical operating data, the edge computing node extracts the reference difference interval of the structural mutation index Istruct during the stable period, and obtains the structural threshold Is for the critical degree of structural mutation. After obtaining the structural mutation index Istruct(t) at time t, the edge computing node performs structural mutation assessment by comparing the structural mutation index Istruct(t) at time t with the structural threshold Is to determine the current structural mutation status of the wire rope. The specific assessment content is as follows: When the structural mutation index Istruct(t) at time t is greater than or equal to the structural threshold Is, it is determined that there is a significant mutation in the current wire rope structural state and a structural mutation assessment result is generated as a mutation has occurred. When the structural mutation index Istruct(t) at time t is less than the structural threshold Is, the current wire rope structure is determined to be stable and the structural mutation assessment result is generated as no mutation.

[0033] S4 also includes S42; S42. When the structural mutation assessment result indicates that a mutation has occurred, the edge computing node sends a structural anomaly reporting information to the cloud. The cloud generates a collaborative control command based on the energy change range corresponding to different structural states in long-term historical operating data and sends it to the edge computing node to execute the cloud collaborative control mechanism triggered by the edge computing node and the cloud collaborative node. The collaborative control command adjusts the noise segment judgment threshold with the goal of reducing the probability of misjudgment, so that the judgment energy threshold Efth of the noise segment label Ns(t) at time t is increased by 5%-20% compared with the original value, thereby improving the judgment stability of structural anomaly segments. With the goal of improving the accuracy of structural response monitoring, the sampling resolution range is adjusted so that the edge computing nodes can increase the sampling frequency fs range by 10%-30% in subsequent monitoring cycles, thereby enhancing the response sensitivity to potential structural anomalies. After receiving the collaborative control command, the edge computing node automatically updates its local parameters and performs monitoring and control according to the updated threshold and sampling resolution in subsequent cycles, so as to realize the dynamic adaptive adjustment of the edge computing node monitoring strategy according to the cloud verification results. When the structural mutation assessment result is no mutation, the edge computing node does not perform parameter adjustment, but only records the structural state of the current period for subsequent periodic calibration, without triggering the cloud collaborative control mechanism.

[0034] In this embodiment, method S4 achieves a closed-loop control link by linking the structural mutation threshold Is with the cloud-based collaborative control command, which "first clarifies whether the structure has changed, and then decides whether the system should tighten or loosen": The edge computing node does not arbitrarily set a fixed threshold, but extracts the reference difference interval of the structural mutation index Istruct under "normal fluctuation" based on the multi-scale feature vector difference amplitude distribution of the stable section of the wire rope structure in long-term historical operating data, and uses its upper boundary as the structural threshold Is for the critical degree of structural mutation. The real physical meaning of this setting is to first draw a "normal sway safety belt" for the system using real stable operating conditions, avoiding misjudging slow temperature changes, slight wind disturbances, or daily load changes as structural anomalies; On this basis, when the structural mutation index Istruct(t) at time t ≥ the structural threshold Is, it indicates that the structural response change within the current window has significantly exceeded the range of historical stable fluctuations. The edge computing node judges the structural mutation assessment result as a mutation and sends structural anomaly reporting information to the cloud. The cloud leverages energy variation ranges corresponding to different structural states over a long historical period to generate collaborative control commands, which are then sent to the edge. On one hand, to reduce misjudgments, the noise threshold Efth is moderately increased by 5%–20%, avoiding a "hypersensitive reaction" caused by overall energy increases due to extreme weather. On the other hand, to improve detection capabilities, the sampling frequency fs range is adjusted upward by 10%–30%, essentially "automatically increasing the microscope magnification" when there is a risk of structural mutation, enhancing the sensitivity to potential structural anomalies in subsequent monitoring cycles. After receiving the collaborative control commands, the edge computing nodes automatically update their local thresholds and resolution parameters, making subsequent noise judgment and feature extraction more aligned with the current environment and structural state, forming a dynamically adaptive monitoring strategy based on cloud verification results. When Istruct(t) < structural threshold Is, and the evaluation result is no mutation, only the current period's structural state is recorded for subsequent calibration, without triggering any parameter adjustments, thus effectively avoiding frequent strategy adjustments under conditions of high noise, strong disturbances, but with a normal structure. Through the above implementation process, the core problems of "whether structural mutations are real changes" and "whether the system needs to be adjusted" in extremely cold and complex environments have been solved incisively. This has enabled the system to reduce false alarms and stabilize alarm boundaries, while automatically improving monitoring accuracy and sensitivity when real structural anomalies occur, thus greatly enhancing the reliability and practicality of wire rope structural anomaly detection. Example 6

[0035] Please see Figure 1 and Figure 2 The edge cloud computing collaborative processing system for steel wire rope detection data includes an edge acquisition module, a short-time energy analysis module, a resolution processing module, and a cloud collaboration module. The edge acquisition module collects raw data in real time by setting up acquisition points along the operating path of the cableway steel wire rope in the mountainous area, and uploads it to the edge computing node. The raw data is preprocessed in the edge computing node to obtain the short-time energy dataset of the steel wire rope. The short-time energy analysis module performs fusion processing on the short-time energy dataset of the wire rope in the edge computing node to obtain the noise fusion energy feature Efusion, sets the energy threshold Efth and performs a preliminary comparison and evaluation with the noise fusion energy feature Efusion, and outputs the noise segment label Ns based on the preliminary comparison and evaluation results. The resolution processing module triggers a variable resolution processing operation based on the noise segment label Ns, automatically selects the corresponding resolution level according to the noise segment label Ns to construct a multi-scale structural feature vector Fraw, and obtains the structural mutation index Istruct based on the multi-scale structural feature vector Fraw. The cloud-based collaborative module performs structural mutation assessment by setting a structural mutation threshold Is and combining the structural mutation threshold Is with the structural mutation index Istruct. Based on the structural mutation assessment results, it then enters the collaborative verification stage.

[0036] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

Claims

1. An edge computing collaborative processing method for steel wire rope detection data, characterized in that: Includes the following steps: S1. Set up collection points along the cableway's steel wire rope running path in the mountainous area to collect raw data in real time and upload it to the edge computing node. In the edge computing node, preprocess the raw data to obtain the short-time energy dataset of the steel wire rope. S2. In the edge computing node, the short-time energy dataset of the wire rope is fused to obtain the noise fusion energy feature Efusion. The energy threshold Efth is set and the noise fusion energy feature Efusion is compared and evaluated in advance. Based on the preliminary comparison and evaluation results, the noise segment label Ns is output. S3. Based on the noise segment label Ns, trigger the variable resolution processing operation, automatically select the corresponding resolution level according to the noise segment label Ns to construct the multi-scale structural feature vector Fraw, and obtain the structural mutation index Istruct based on the multi-scale structural feature vector Fraw. S4. Preset the structural mutation threshold Is, and use the structural mutation threshold Is and the structural mutation index Istruct to perform structural mutation assessment. Based on the structural mutation assessment results, proceed to the collaborative verification stage.

2. The edge computing collaborative processing method for steel wire rope detection data according to claim 1, characterized in that: S1 includes S11; S11. Multiple data collection points are set up along the cableway's steel cable running path, and sensor groups are installed at the data collection points to collect raw data in real time. The collected raw data is then transmitted to the edge computing node via the wireless transmission module of the corresponding sensor at the data collection point using a local area wireless network. The edge computing node is equipped with edge processing capabilities by being set up on the cableway tower platform and is connected to the cloud collaborative node to receive raw data in real time. The collection points include a first collection point P1, a second collection point P2, a third collection point P3, and a fourth collection point P4; The sensor group includes a magnetic flux sensor, a micro-vibration acceleration sensor, a micro-vibration pickup sensor, and a temperature change rate sensor; The raw data includes the magnetic flux density of the wire rope XP1, the surface vibration acceleration of the wire rope XP2, the micro-vibration acceleration of the tower structure XP3, and the temperature change rate of the tower surface XP4.

3. The edge computing collaborative processing method for steel wire rope detection data according to claim 2, characterized in that: S1 further includes S12; S12. Preprocess the raw data in the edge computing node to obtain the short-time energy dataset of the wire rope; The preprocessing includes timestamp alignment, short-time energy extraction, and normalization. The short-time energy dataset of the wire rope includes the short-time energy EP1(t) of the first acquisition point P1 at time t, the short-time energy EP2(t) of the second acquisition point P2 at time t, the short-time energy EP3(t) of the third acquisition point P3 at time t, and the short-time energy EP4(t) of the fourth acquisition point P4 at time t. The timestamp alignment is achieved by setting a unified clock in the edge computing nodes. When the raw data is received, all parameters in the raw data are timestamped according to the unified clock. The short-time energy extraction is performed by extracting the short-time energy of the corresponding collection point using the short-time energy calculation formula. The standardization process uses the z-score standardization method to standardize all extracted short-time energies, eliminating the influence of unit dimensions, and then summarizing them to obtain a short-time energy dataset of the wire rope.

4. The edge cloud computing collaborative processing method for steel wire rope detection data according to claim 3, characterized in that: S2 includes S21; S21. Based on the short-time energy dataset of the wire rope, construct an energy calculation model, and input the real-time acquired short-time energy dataset of the wire rope into the energy calculation model to perform noise energy fusion and obtain the noise fusion energy feature Efusion. The noise fusion energy characteristic Efusion is obtained through the following energy calculation model; In the formula, Efusion(t) represents the noise fusion energy characteristics at time t, and a1, a2, a3 and a4 represent the preset weight values ​​of the short-time energy EP1 of the first acquisition point P1, the short-time energy EP2 of the second acquisition point P2, the short-time energy EP3 of the third acquisition point P3 and the short-time energy EP4 of the fourth acquisition point P4, respectively, and a1+a2+a3+a4=1, the specific values ​​of which are set by the user.

5. The edge cloud computing collaborative processing method for steel wire rope detection data according to claim 4, characterized in that: S2 further includes S22; S22. Construct an energy calculation model based on the short-time energy dataset of steel wire ropes under historical extreme cold conditions to obtain the short-time energy distribution range under extreme cold conditions, and determine the energy threshold Efth that distinguishes between normal segments and abnormal noise segments based on the short-time energy distribution range. A preliminary comparison and evaluation is conducted based on the acquired energy threshold Efth and the real-time acquired noise fusion energy characteristic Efusion to determine the abnormal noise segment of the current time period of the wire rope. The specific evaluation content is as follows: When the noise fusion energy characteristic Efusion(t) at time t is less than the energy threshold Efth, the output noise label Ns(t) at time t is a normal segment, indicating that the current time period of the wire rope is a normal segment; When the noise fusion energy feature Efusion(t) at time t is greater than or equal to the energy threshold Efth, it indicates that the current time period of the wire rope is an abnormal noise segment, and at this time, the noise segment label Ns(t) at time t is generated. The noise segment label Ns(t) at time t is generated using the following method: Based on the energy distribution characteristics of the short-time energy dataset of the wire rope at time t, the noise segment is subdivided, where: When the short-time energy EP2(t) of the second collection point P2 at time t is greater than the ice hanging energy threshold EP2h, the noise segment label Ns(t) at time t is determined to be the ice hanging noise segment; When the short-time energy EP2(t) of the second acquisition point P2 at time t is greater than the wind shear energy threshold EP2wind and the short-time energy EP3(t) of the third acquisition point P3 at time t is less than the tower resonance energy threshold EP3th, the noise segment label Ns(t) at time t is determined to be the wind shear disturbance segment. When the short-time energy EP3(t) of the third acquisition point P3 at time t is greater than the tower resonance energy threshold EP3th, the noise segment label Ns(t) is determined to be the tower resonance segment.

6. The edge cloud computing collaborative processing method for steel wire rope detection data according to claim 5, characterized in that: S3 includes S31; S31. After obtaining the noise segment label Ns(t) at time t, the edge computing node automatically performs variable resolution processing based on the noise segment label Ns(t) at time t; the specific operation is as follows: Wherein: when the noise segment label Ns(t) at time t is the ice hanging noise segment, the sampling frequency fs in the range of 9kHz-20kHz is selected as the high-frequency resolution; When the noise segment label Ns(t) at time t is the wind shear disturbance segment, the sampling frequency fs in the range of 3kHz-8kHz is selected as the intermediate frequency resolution; When the noise segment label Ns(t) at time t is the tower resonance segment, the sampling frequency fs in the range of 500Hz-2kHz is selected as the low-frequency resolution. When the noise segment label Ns(t) at time t is a normal segment, select the sampling frequency fs in the range of 100Hz-500Hz, which corresponds to the energy-saving mode with the lowest sampling resolution. The edge computing node performs frequency domain analysis and time domain analysis on the wire rope detection signal according to the sampling frequency fs of the corresponding interval range, obtains the time domain feature set TD and the frequency domain feature set FD corresponding to the current noise segment, and constructs the multi-scale feature vector Fraw(t) corresponding to the current noise segment at time t based on the time domain feature set TD and the frequency domain feature set FD. The multi-scale feature vector Fraw(t) at time t is specifically and formally expressed as: Fraw(t) = TD∪FD.

7. The edge cloud computing collaborative processing method for steel wire rope detection data according to claim 6, characterized in that: S3 further includes S32; S32. After obtaining the multi-scale feature vector Fraw(t) at time t, the edge computing node performs difference calculation on the multi-scale feature vector Fraw(t) at time t and the multi-scale feature vector Fraw(t−Δt) of the previous feature update cycle based on the first-order difference structural catastrophe analysis method in the existing technology, and obtains the structural catastrophe index Istruct through the following structural catastrophe formula: Istruct(t) = |Fraw(t)−Fraw(t−Δt)|; where Istruct(t) represents the structural mutation index at time t.

8. The edge computing collaborative processing method for steel wire rope detection data according to claim 7, characterized in that: S4 includes S41; S41. Based on the statistical distribution of the difference amplitude of the multi-scale feature vectors of the stable section of the wire rope structure in long-term historical operating data, the edge computing node extracts the reference difference interval of the structural mutation index Istruct during the stable period, and obtains the structural threshold Is for the critical degree of structural mutation. After obtaining the structural mutation index Istruct(t) at time t, the edge computing node performs structural mutation assessment by comparing the structural mutation index Istruct(t) at time t with the structural threshold Is to determine the current structural mutation status of the wire rope. The specific assessment content is as follows: When the structural mutation index Istruct(t) at time t is greater than or equal to the structural threshold Is, it is determined that there is a significant mutation in the current wire rope structural state and a structural mutation assessment result is generated as a mutation has occurred. When the structural mutation index Istruct(t) at time t is less than the structural threshold Is, the current wire rope structure is determined to be stable and the structural mutation assessment result is generated as no mutation.

9. The edge cloud computing collaborative processing method for steel wire rope detection data according to claim 8, characterized in that: S4 further includes S42; S42. When the structural mutation assessment result indicates that a mutation has occurred, the edge computing node sends a structural anomaly reporting information to the cloud. The cloud generates a collaborative control command based on the energy change range corresponding to different structural states in long-term historical operating data and sends it to the edge computing node to execute the cloud collaborative control mechanism triggered by the edge computing node and the cloud collaborative node. The cooperative control command increases the determination energy threshold Efth of the noise segment label Ns(t) at time t by 5%-20% compared to the original value; This allows edge computing nodes to increase the sampling frequency fs range by 10%-30% in subsequent monitoring cycles. After receiving the collaborative control command, the edge computing node automatically updates its local parameters and performs monitoring and control according to the updated threshold and sampling resolution in subsequent cycles. When the structural mutation assessment result is no mutation, the edge computing node does not perform parameter adjustment, but only records the structural state of the current period for subsequent periodic calibration, without triggering the cloud collaborative control mechanism.

10. An edge computing collaborative processing system for wire rope inspection data, applied to the edge computing collaborative processing method for wire rope inspection data as described in any one of claims 1-9, characterized in that: It includes an edge acquisition module, a short-time energy analysis module, a resolution processing module, and a cloud collaboration module; The edge acquisition module collects raw data in real time by setting up acquisition points along the operating path of the cableway steel wire rope in the mountainous area, and uploads it to the edge computing node. The raw data is preprocessed in the edge computing node to obtain the short-time energy dataset of the steel wire rope. The short-time energy analysis module performs fusion processing on the short-time energy dataset of the wire rope in the edge computing node to obtain the noise fusion energy feature Efusion, sets the energy threshold Efth and performs a preliminary comparison and evaluation with the noise fusion energy feature Efusion, and outputs the noise segment label Ns based on the preliminary comparison and evaluation results. The resolution processing module triggers a variable resolution processing operation based on the noise segment label Ns, automatically selects the corresponding resolution level according to the noise segment label Ns to construct a multi-scale structural feature vector Fraw, and obtains the structural mutation index Istruct based on the multi-scale structural feature vector Fraw. The cloud-based collaborative module performs structural mutation assessment by setting a structural mutation threshold Is and combining the structural mutation threshold Is with the structural mutation index Istruct. Based on the structural mutation assessment results, it then enters the collaborative verification stage.

Images