Method for detecting and warning damage of belt conveyor based on multi-spectrum vibration sensing

Abstract

This application provides a method for damage detection and early warning of conveyor belts based on multispectral vibration sensing, applied in the fields of mine safety monitoring and industrial equipment fault diagnosis. The method includes: collecting the multispectral radiation response and structural vibration response of the conveyor belt during operation; constructing a joint distribution map of energy absorption and phase hysteresis to identify spectral anomaly regions; analyzing the consistency of vibration propagation to locate regions of enhanced vibration reflection or abnormal dissipation; performing spatiotemporal consistency verification on spectral anomalies and vibration distortion characteristics to confirm structural damage and its spatial location; tracking the rate of change of vibration reflection and the spectral spread rate to construct physical criteria for damage evolution and output risk level warnings. This invention, through multispectral and vibration dual-mode fusion detection and spatiotemporal verification, achieves early identification and precise location of internal damage to the conveyor belt, effectively preventing serious accidents such as belt breakage and improving the predictive maintenance level of equipment operation.

Description

Technical Field

[0001] This application relates to the field of mine safety monitoring and industrial equipment fault diagnosis technology, and in particular to a method for damage detection and early warning of belt conveyors based on multispectral vibration sensing. Background Technology

[0002] Belt conveyors are core transportation equipment in mining and other fields, and their operational reliability directly affects production efficiency and safety. During long-term heavy-load operation, conveyor belts are prone to internal hidden damage such as wire rope breakage and capping layer delamination. Once these damages develop to a critical state, they can lead to serious accidents such as belt breakage, causing significant losses.

[0003] Existing detection technologies mainly include X-ray, electromagnetic detection, and infrared thermal imaging, but all are single-physical-field detection and each has blind spots. X-rays pose radiation risks and are difficult to perform online; electromagnetic detection is insensitive to non-wire rope damage; and infrared thermal imaging is slow to detect internal changes. These technologies struggle to comprehensively identify different types of early damage, lack spatial correlation in detection results, and have high false alarm and false negative rates.

[0004] Multispectral imaging and vibration analysis in visual intelligence have been applied to equipment monitoring in recent years, but they are usually used independently, failing to spatially correlate and fuse spectral and vibration information, resulting in insufficient detection reliability and positioning accuracy. Existing methods are mostly static threshold alarms, lacking the ability to dynamically track damage evolution trends and predict failure risks, making it difficult to achieve predictive maintenance. Summary of the Invention

[0005] The embodiments of this application provide a method for damage detection and early warning of conveyor belts based on multispectral vibration sensing. This method enables early identification and precise location of internal damage to the conveyor belt, and dynamically tracks the damage evolution trend for risk warning, thereby effectively preventing serious accidents such as belt breakage. To achieve the above objectives, this application adopts the following technical solution: A method for damage detection and early warning of belt conveyors based on multispectral vibration sensing, the method comprising: During the stable operation of the conveyor belt, multispectral radiation response and structural vibration response are collected to form multispectral response set and structural vibration response set, respectively. Based on the aforementioned multispectral response set, a joint distribution map of energy absorption and phase hysteresis is constructed to identify spectral anomaly regions where absorption intensity and phase hysteresis are simultaneously unbalanced due to internal damage to the tape. Using the set of structural vibration responses as input, and combining it with the spatial location corresponding to the spectral anomaly region, the consistency of vibration energy propagation is analyzed, the region of enhanced vibration reflection or abnormal dissipation caused by decreased structural integrity is located, and vibration propagation distortion characteristics spatially associated with the spectral anomaly region are formed. The spatiotemporal consistency of the spectral anomaly region and the vibration propagation distortion characteristics is checked. Only when the two form a stable correspondence on the tape running path is it determined that there is structural damage, a damage confirmation result is generated, and the specific spatial location of the damage in the length and width directions of the tape is determined. Based on the specific spatial location corresponding to the damage confirmation results, the changing trend of vibration energy reflection ratio over time and the spatial expansion rate of spectral anomaly regions are analyzed. Physical criteria for the evolution of damage from local defects to operational failure are constructed, and an early warning of the operational risk level of the belt conveyor is output based on the physical criteria.

[0006] In some possible implementations, the acquisition of multispectral radiation response and structural vibration response, forming a multispectral response set and a structural vibration response set respectively, includes: In the detection section of the conveyor belt's running path, multiple sets of multispectral imaging sensors and vibration sensors are deployed along the belt's running direction and width direction; The multispectral imaging sensor is controlled to emit detection radiation including at least three preset characteristic bands to the surface and near-surface of the tape, and the reflection intensity and scattering intensity data of the tape to the detection radiation are received and recorded to form a multispectral response set. The vibration acceleration time-domain signal output by the vibration sensor is collected to form a set of structural vibration responses.

[0007] In some possible implementations, constructing a joint distribution map of energy absorption and phase hysteresis based on the multispectral response set includes: The multispectral response set is processed to obtain the absorption attenuation coefficient sequence and phase hysteresis sequence of the tape in each characteristic band. The absorption attenuation coefficient and phase hysteresis in the same characteristic band are paired and correlated to form paired correlation data. Multiple sets of paired and correlated data are mapped and fitted in a two-dimensional coordinate system consisting of the absorption attenuation coefficient axis and the phase hysteresis axis to generate a joint distribution map for identifying internal damage.

[0008] In some possible implementations, identifying spectral anomalies caused by an imbalance between absorption intensity and phase hysteresis due to internal damage to the tape includes: In the joint distribution map, a baseline distribution region characterizing the normal material state is defined based on the historical multispectral response set of the tape in its intact state. For the newly acquired multispectral response set, the statistical distance between the corresponding data point in the joint distribution map and the baseline distribution area is calculated as a deviation index. Determine whether the deviation index exceeds a preset damage assessment threshold; When the deviation index exceeds the damage determination threshold, the spatial location of the tape that generated the data point is determined as a suspected damage point. Based on the spatial clustering of multiple consecutive suspected damage points, the spectral anomaly region is determined.

[0009] In some possible implementations, the spatiotemporal consistency check between the spectral anomaly region and the vibration propagation distortion characteristics is performed, and structural damage is determined to exist only when the two form a stable correspondence along the tape's running path. This includes: The center coordinates of the spectral anomaly region are obtained as the first position, and the center coordinates of the vibration propagation distortion feature are obtained as the second position. Calculate the Euclidean distance between the first position and the second position in the tape coordinate system; Within multiple consecutive detection cycles, it is determined whether the Euclidean distance is less than a preset spatial tolerance threshold, and whether the spectral anomaly region and the vibration propagation distortion feature are detected synchronously within the corresponding cycle; when the above conditions are met simultaneously, it is determined that the spectral anomaly region and the vibration propagation distortion feature form a stable correspondence in space and time, confirming the existence of structural damage.

[0010] In some possible implementations, the output of the operational risk level warning for the belt conveyor based on the physical criteria includes: Multiple incremental risk levels are predefined, and corresponding vibration reflection intensity thresholds and spectral anomaly spread rate thresholds are set for each level. The quantized value of the change trend of the detected vibration reflection intensity is compared with the vibration reflection intensity threshold, and the expansion rate value of the detected spectral anomaly region is compared with the spectral anomaly expansion rate threshold corresponding to the current risk level. The risk level is raised to the next level only when the vibration reflection intensity exceeds the corresponding threshold and the expansion rate value exceeds the corresponding threshold, and a warning signal including the current risk level is generated and output.

[0011] In some possible implementations, the method further includes: After obtaining the damage confirmation result, the characteristic data of the spectral anomaly region, the characteristic data of the vibration propagation distortion characteristics, and the spatiotemporal correlation between the two are bound and integrated with the determined specific spatial location to form a historical damage record and stored in the damage pattern database. In subsequent detection, when new suspected damage is identified, the corresponding spectral and vibration feature data are extracted, and the composite similarity with each historical damage record in the damage pattern database is calculated. If the composite similarity is greater than the preset matching threshold, the current suspected damage and the corresponding historical record are determined to belong to the same damage pattern, and the evolutionary information included in the historical record is output as an auxiliary diagnostic conclusion for the current damage.

[0012] In some possible implementations, the location of regions of enhanced vibration reflection or anomalous dissipation due to decreased structural integrity forms vibration propagation distortion features spatially associated with the spectral anomaly regions, including: A vibration signal analysis window is defined in the tape coordinate system, centered on the spectral anomaly region. From the set of structural vibration responses, extract the vibration acceleration time-domain signal corresponding to the spatiotemporal range of the vibration signal analysis window; The intercepted vibration acceleration time-domain signal is transformed and analyzed in the frequency domain to calculate the ratio of reflected wave energy related to the excitation source to the total signal energy within the preset characteristic frequency band. The reflected energy ratio is compared with the normal threshold range calibrated based on healthy adhesive tape samples: If the reflected energy ratio is higher than the upper limit of the normal threshold range, the area corresponding to the window is determined to be a vibration reflection enhancement area caused by structural weakening. If the reflected energy ratio is lower than the lower limit of the normal threshold range, the area corresponding to the window is determined to be the abnormal dissipation area of ​​vibration energy caused by structural damage. Within multiple consecutive detection cycles, spatiotemporal clustering is performed on the identified regions to form stable vibration propagation distortion characteristics.

[0013] In some possible implementations, the analysis of the consistency of vibration energy propagation includes: Based on the aforementioned structural vibration response set, the longitudinal propagation velocity of the vibration wave along the belt running direction and the transverse vibration energy amplitude along the belt width direction are calculated respectively. Based on the spatial location of the spectral anomaly region, the target analysis segment covered along the running direction is determined, the vibration signals corresponding to both ends of the target analysis segment are extracted, and the longitudinal propagation velocity change rate and the transverse vibration energy amplitude attenuation rate before and after the vibration wave is transmitted are calculated. The calculated rate of change and the attenuation rate are compared with preset health baseline thresholds to quantitatively evaluate the consistency of vibration energy propagation of the tape in the target analysis section.

[0014] In some possible implementations, the method further includes: A predefined mapping strategy between risk levels, damage modes, and belt conveyor control commands; Obtain the operational risk level output by the physical criteria, and the auxiliary diagnostic conclusions obtained by matching historical damage records; When the operational risk level reaches a preset high-risk threshold, or when the auxiliary diagnostic conclusion indicates a specific type of known damage mode, a corresponding control command is generated according to the mapping strategy. Executing the control command triggers at least one control action, including slow-down operation, emergency shutdown, or planned maintenance scheduling.

[0015] As can be seen from the above technical solution, this application has the following beneficial effects: 1. This invention achieves dual-modal sensing of internal structural anomalies and mechanical performance degradation in conveyor belts by collecting multispectral radiation responses and structural vibration responses, respectively constructing a joint absorption phase distribution map and vibration propagation distortion characteristics. Furthermore, through spatiotemporal consistency verification, it precisely correlates spectral anomaly regions with vibration distortion characteristics in space, confirming damage only when the two stably correspond, effectively eliminating false alarms from single detection methods and significantly improving the accuracy and reliability of damage identification. This method can detect early internal damage in real time during conveyor belt operation and accurately locate the specific position of the damage in the length and width directions, providing accurate operational guidance for maintenance personnel.

[0016] 2. This invention constructs physical criteria for the evolution of damage from local defects to operational failure by continuously tracking the rate of change of vibration reflection energy ratio and the expansion rate of spectral anomaly regions, achieving quantitative assessment of damage development trends and dynamic early warning of risk levels. Combined with similarity matching from a historical damage database, it can provide auxiliary diagnostic conclusions for newly discovered damage and predict its evolution path based on historical experience. Finally, it automatically generates control commands such as speed reduction, shutdown, or maintenance scheduling based on risk level and damage pattern, forming a closed-loop management system. This upgrades the maintenance mode of belt conveyors from reactive repair to predictive maintenance, significantly reducing the risk of serious accidents such as belt breakage, ensuring continuous production, and yielding significant economic and safety benefits. Attached Figure Description

[0017] The invention will now be further described with reference to the accompanying drawings.

[0018] Figure 1 This is an overall flowchart of the method provided in the embodiments of this application; Figure 2 This is a flowchart of multispectral data acquisition and processing provided in an embodiment of this application; Figure 3 This is a flowchart of vibration signal acquisition and processing provided in an embodiment of this application; Figure 4 A flowchart for identifying spectral anomaly regions provided in this application embodiment; Figure 5 A flowchart for locating vibration propagation distortion features provided in this application embodiment; Figure 6A flowchart for spatiotemporal consistency verification and damage confirmation provided in the embodiments of this application; Figure 7 A flowchart illustrating the damage evolution tracking and risk warning process provided in this application embodiment; Figure 8 A flowchart illustrating historical database matching and assisted diagnosis provided for embodiments of this application. Detailed Implementation

[0019] The terms "first," "second," and "third," etc., used in this application specification, claims, and drawings are for distinguishing different objects, not for specifying a particular order.

[0020] In the embodiments of this application, the words "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the words "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0021] Research has revealed that existing tape damage detection technologies rely on a single physical field, and each method has a detection blind spot, making it difficult to comprehensively identify different types of early internal damage. Furthermore, most of them use static thresholds and isolated alarm methods, which cannot spatially correlate and verify spectral and vibration information, resulting in high false alarm and false negative rates. They also lack the ability to dynamically track damage evolution trends and predict failure risks.

[0022] To address the aforementioned issues, this application provides a method for damage detection and early warning of belt conveyors based on multispectral vibration sensing: Example 1

[0023] like Figures 1-8 As shown, this embodiment demonstrates deployment and application verification on a belt conveyor in the main inclined shaft of a large coal mine. This conveyor is responsible for hoisting raw coal for the entire mine, with a total length of approximately 3,000 meters, a belt width of 1.4 meters, and a belt speed of 3.5 meters per second. The conveyor belt type is a steel wire rope core flame-retardant conveyor belt. Due to long-term heavy-load operation, the belt faces the risk of internal damage such as steel wire rope breakage, cover layer wear, and joint fatigue. These damages are highly concealed in their early stages, making them difficult to accurately identify using traditional single detection methods. Once they develop into failure, they will lead to serious accidents such as belt breakage and tearing, causing huge economic losses and safety hazards.

[0024] To address the aforementioned issues, this method establishes three fixed detection zones at key locations on the conveyor: 50 meters behind the head drive station, 30 meters before and after the middle loading point, and 50 meters in front of the tail redirecting roller. Each detection zone is equipped with multiple sets of sensors along the belt's running direction, including a multispectral imaging sensor array and a vibration acceleration sensor array, enabling online, non-contact comprehensive detection of the conveyor belt during operation. The detection data is transmitted in real-time to a surface monitoring server via the underground industrial ring network. The server runs the various algorithm models involved in this method, ultimately outputting damage confirmation results and risk warnings, and linking with the conveyor control system to achieve automatic protection.

[0025] 1. Collect multispectral radiation response and structural vibration response.

[0026] During stable operation of the conveyor belt, the system activates all sensors within the detection section for synchronous data acquisition. The multispectral imaging sensor employs multi-band active illumination, emitting detection radiation in at least three preset characteristic bands towards the surface and near-surface of the conveyor belt. The selection of these bands is based on the absorption characteristics of the conveyor belt material; for example, the near-infrared band (center wavelength 950 nm) is sensitive to debonding at the interface between the steel wire rope and rubber, the mid-infrared band (center wavelength 3400 nm) is sensitive to rubber aging, and the ultraviolet band (center wavelength 365 nm) is sensitive to surface microcracks. The sensor receives data on the reflection and scattering intensity of the detection radiation from the conveyor belt and correlates this data with the corresponding spatial location (along the length and width of the conveyor belt) and a timestamp to form a multispectral response set. This set is stored in the form of a three-dimensional data block, with dimensions including: spatial location (x-direction: running length, y-direction: bandwidth), band channel, and radiation intensity value.

[0027] Meanwhile, vibration acceleration sensors deployed beneath the conveyor belt collect time-domain vibration acceleration signals during belt operation at a preset sampling frequency (e.g., 2,000 points per second). The vibration sensors are charge-amplified output accelerometers, mounted on support rollers tightly attached to the cover adhesive beneath the belt, to effectively capture vibration characteristics caused by internal damage to the belt. Multiple vibration sensors are arranged at regular intervals along the belt's running direction, forming a sensor array. The collected vibration signals, after anti-aliasing filtering and analog-to-digital conversion, are correlated with timestamps and sensor positions to form a structural vibration response set.

[0028] 2. Construct a joint distribution map of energy absorption and phase hysteresis to identify spectral anomaly regions.

[0029] The system first processes the multispectral response set. For each characteristic band, based on the reflection and scattering intensity data, combined with a preset reference reflectivity (based on intact tape calibration), the absorption attenuation coefficient of the tape in that band is calculated. The absorption attenuation coefficient is defined as: absorption attenuation coefficient = -ln(actual reflection intensity / reference reflection intensity), characterizing the degree of absorption of the probe radiation by the tape material. Its physical meaning lies in reflecting the attenuation effect of material density, chemical composition, and internal defects on radiation energy. Simultaneously, the phase lag is obtained by analyzing the phase delay of the reflected signal relative to the transmitted signal. The phase lag reflects the phase change generated during multiple scattering and absorption of the probe radiation within the tape. Its physical meaning lies in indicating the homogeneity of the internal structure of the material—uniform materials have stable phase lag, while non-uniform or damaged areas exhibit abnormal phase lag. The phase lag is extracted using an orthogonal demodulation method: assuming the transmitted signal is a modulated sine wave (… The received signal is Let (r(t)) be compared with the in-phase reference signal. and quadrature reference signals Multiply, and then low-pass filtered to obtain the in-phase component. and orthogonal components ( Then the phase lag ( The average value is taken over multiple periods to suppress noise.

[0030] The absorption attenuation coefficient and phase hysteresis under the same characteristic band are paired and correlated to form a set of paired correlation data. For the three characteristic bands, each spatial location corresponds to three sets of data. Then, all paired correlation data are mapped in a two-dimensional coordinate system composed of the absorption attenuation coefficient axis and the phase hysteresis axis. Each data point represents the absorption-phase characteristics of a certain spatial location of the tape under a certain band. By performing density clustering or fitting on these data points, a joint distribution map of energy absorption and phase hysteresis is generated. This distribution map can intuitively show the characteristic distribution pattern of the tape material under different states. Its purpose is to transform abstract spectral data into a visualized feature space in order to separate outliers.

[0031] On the joint distribution map, based on the historical multispectral response set of the tape in good condition (i.e., tape test data after new installation or recent maintenance), a baseline distribution area characterizing the normal material state is defined. The method for defining the baseline distribution area is as follows: collect multiple sets of good data from the initial stage of tape operation (e.g., the first month) or after each major overhaul; for each band, calculate the mean absorption attenuation coefficient and the mean phase hysteresis of all data points to obtain the mean vector (…). ), and calculate the covariance matrix ( The sample size should be no less than one hundred groups to ensure statistical stability. In two-dimensional space, with ( Centered on the Mahalanobis distance, an elliptical region is defined as the baseline distribution region. The boundary of the ellipse corresponds to a certain quantile of the Mahalanobis distance (such as the distance corresponding to the 95% confidence level). This quantile is the reference benchmark for the damage determination threshold.

[0032] For a newly acquired multispectral response set, the system calculates the value of each data point ( Mahalanobis distance to the baseline distribution region: The Mahalanobis distance integrates the correlation between absorption and phase, making it more scientific than a single threshold. This distance is then compared with a preset damage assessment threshold (…). Compare. The determination method can be based on statistical distribution, taking the 95th percentile of the Mahalanobis distance of the benchmark sample, or optimized through historical accident data. When ( When a data point is identified as having a synchronous imbalance between absorption intensity and phase lag at its corresponding spatial location on the tape, indicating an abnormal spectral characteristic, it is marked as a suspected damage point. Multiple consecutive suspected damage points are spatially clustered (e.g., five consecutive points along the running direction and adjacent in the bandwidth direction) to form a spectral anomaly region. The extent of this region is defined by the minimum bounding rectangle of the cluster points. The clustering algorithm can be DBSCAN, using distance along the running direction and bandwidth direction as features, setting a neighborhood radius (e.g., 20 cm in the running direction and 5 cm in the bandwidth direction) and a minimum number of points (e.g., five points). After clustering, the center coordinates of each cluster are... The boundaries are recorded.

[0033] 3. Analyze the characteristics of vibration propagation distortion to ensure consistent vibration propagation.

[0034] Centered on the spatial location of the spectral anomaly region, the system defines a vibration signal analysis window in the conveyor belt coordinate system. The analysis window extends a certain length along the running direction (e.g., one meter forward and backward), covering the entire spectral anomaly region and its surrounding area. The vibration acceleration time-domain signal corresponding to the spatiotemporal range of this window is extracted from the structural vibration response set. Since the vibration sensors are discretely deployed, the window may include multiple sensors. The system selects the signal from the sensor closest to the center of the window, or performs a weighted average of the signals from multiple sensors within the window. A fast Fourier transform is performed on the extracted vibration acceleration time-domain signal to obtain the spectrum. The preset characteristic frequency band is determined based on the characteristics of the excitation source: based on the roller passing frequency (…). (in( ) is the belt speed, ( ( ) is the roller spacing) and its multiples ( As the center of the characteristic frequency band, the bandwidth of each frequency band is set to ( It can dynamically track and adjust for changes in band speed. In the spectrum, the energy within the characteristic frequency band is calculated ( and total signal energy ( (Limited to the effective frequency band). Define the reflected energy ratio ( This ratio reflects the reflection or dissipation effect caused by structural abrupt changes (such as internal damage) encountered when the vibration wave propagates in the tape.

[0035] Will( Compared with the normal threshold range calibrated based on healthy tape samples ( Comparison: like( If so, the area corresponding to the window is determined to be the area of ​​enhanced vibration reflection caused by structural weakening.

[0036] like( If the region corresponding to the window is determined to be the area of ​​abnormal dissipation of vibration energy caused by structural damage, then the region is identified.

[0037] If it is within the range, it is considered normal.

[0038] In multiple consecutive detection cycles, spatiotemporal clustering (similar to clustering of spectral anomaly regions) is performed on the identified enhancement or dissipation regions to form stable vibration propagation distortion characteristics, and their center coordinates are recorded. The abnormality type (enhancement / dissipation) and abnormal intensity (i.e., the degree of deviation of the reflected energy ratio from the threshold) are considered. The physical significance of vibration propagation distortion characteristics lies in the fact that a decrease in structural integrity will lead to changes in the propagation path of vibration waves, manifested as localized enhanced reflection or intensified energy absorption. The spatial correlation between these regions and spectral anomaly regions is key to determining the presence of damage.

[0039] 4. Spatiotemporal consistency check to determine structural damage.

[0040] The system extracts the center coordinates of the spectral anomaly regions respectively. As the first position, and the center coordinates of the vibration propagation distortion characteristics ( As the second position. Calculate the Euclidean distance between these two center coordinates in the tape coordinate system: Set spatial tolerance threshold ( (For example, 10 centimeters), this threshold takes into account the positioning errors of the two detection methods. Within multiple consecutive detection cycles (for example, five cycles), the following two conditions are checked: Condition 1: The calculated (d) for each period is less than ( .

[0041] Condition 2: In each cycle, both spectral anomaly regions and vibration propagation distortion characteristics are detected (i.e., they coexist).

[0042] If both conditions are met, it is determined that a stable correspondence has been formed between the two in space and time, confirming the presence of structural damage at that location. This multimodal information fusion verification method effectively eliminates false alarms that may occur with a single detection method, improving the accuracy of damage confirmation.

[0043] 5. Determine the specific spatial location of the damage.

[0044] Based on the damage confirmation results, the system further refines the spatial location of the damage along the length and width of the conveyor belt. The length position is obtained by adding the coordinates of the starting point of the detection section to the projected distance of the center of the damaged area in the direction of travel, which can be converted into a specific number of meters on the conveyor belt (e.g., how many meters from the machine head). The width position is determined by the ratio of the distance from the center of the damaged area to the left edge of the conveyor belt to the total width of the conveyor belt, for example, "0.3 meters from the left edge." These two coordinates together constitute the precise location of the damage on the conveyor belt plane.

[0045] 6. Construct physical criteria for damage evolution and output risk level warnings.

[0046] Based on the specific spatial location corresponding to the damage confirmation results, the system begins to continuously track the evolution trend of the damage. The tracked physical quantities include two key indicators: the changing trend of the vibration-reflection energy ratio and the spatial expansion rate of the spectral anomaly region.

[0047] The rate of change (R) of vibration-reflected energy is defined as: in( The ratio of reflected energy at the time of initial damage confirmation, ( This represents the reflected energy ratio during the current detection cycle. This indicator quantifies the evolution of vibration reflection characteristics at the damage site.

[0048] Spectral anomaly region expansion rate ( The calculation method for each detection cycle is as follows: Record the coordinates of the front and rear boundaries of the spectral anomaly region along the direction of travel. and( Then the length of the region ( Expansion rate ( ,in( This refers to the detection cycle duration. A smoothed value from multiple cycles can be used as the current expansion rate, expressed in millimeters per day or millimeters per 10,000 revolutions.

[0049] The system constructs physical criteria for the evolution of damage from local defects to operational failure based on these two indicators. Specifically, the physical criteria are as follows: when the vibration-reflected energy ratio continuously increases and exceeds a preset critical threshold (…),… Meanwhile, the rate at which the spectral anomaly region expands along the operating direction exceeds a preset rate threshold. At that time, it was considered that the damage had entered a rapid development phase and was about to lead to operational failure. Here, continuous increase is defined as within multiple consecutive cycles ( .

[0050] Based on the degree to which physical criteria are met, the system outputs a corresponding operational risk level warning. The risk level classification is related to subsequent grading thresholds; here, only the failure stage is determined.

[0051] Glossary of terms involved: Multispectral response set: refers to a dataset acquired by a multispectral imaging sensor, which includes data on the reflection and scattering intensity of the tape at different spectral bands and is associated with its spatial location.

[0052] Structural vibration response set: refers to a dataset collected by a vibration sensor array, including the time-domain signal of vibration acceleration during the operation of the conveyor belt, and associated with the sensor positions.

[0053] Absorption attenuation coefficient: characterizes the degree of absorption of radiation in a specific wavelength band by the tape material, and is calculated using the following formula: The physical meaning reflects the attenuation effect of material density, chemical composition, and internal defects on radiation energy.

[0054] Phase hysteresis: The phase delay caused by multiple scattering and absorption of the probe radiation as it propagates inside the tape reflects the uniformity of the material's internal structure and is extracted using the orthogonal demodulation method.

[0055] Joint distribution map of energy absorption and phase hysteresis: A two-dimensional scatter plot constructed with the absorption attenuation coefficient as the horizontal axis and the phase hysteresis as the vertical axis, used to visualize the state distribution of tape materials and identify anomalies.

[0056] Baseline distribution area: Based on historical data of intact tape, the area on the joint distribution map that represents the normal material state is an elliptical region defined by the mean vector and covariance matrix.

[0057] Mahalanobis distance: A distance metric that takes into account both absorption and phase correlation, used to quantify the degree of deviation of data points from the baseline distribution area.

[0058] Spectral anomaly region: The region formed by the clustering of the tape spatial locations corresponding to data points that deviate from the baseline distribution region on the joint distribution map, indicating that there is an anomalous absorption and phase synchronization in this region.

[0059] Vibration propagation distortion characteristics: regions of enhanced vibration reflection or abnormal energy dissipation caused by decreased structural integrity are identified by vibration signal analysis and confirmed by spatiotemporal clustering, including center coordinates, anomaly type and anomaly intensity.

[0060] Reflection energy ratio: The ratio of vibration energy to total energy within a characteristic frequency band, quantifying the degree of reflection or dissipation of vibration waves at the damaged interface.

[0061] Spatiotemporal consistency verification: The spectral anomaly region is compared and verified with the vibration propagation distortion characteristics in the temporal and spatial dimensions. The existence of damage is confirmed only when the two correspond stably.

[0062] Physical criteria for damage evolution: A quantitative criterion based on the rate of change of vibration reflection energy ratio and the rate of spectral anomaly expansion is used to determine whether damage is progressing towards failure.

[0063] Existing belt conveyor damage detection technologies suffer from the following main drawbacks: First, they rely on a single detection method, such as using only X-rays to detect broken steel wire rope cores or only infrared thermal imaging to detect wear on the cover layer. Each method has blind spots, making it difficult to comprehensively identify early internal damage. Second, the detection results lack spatial correlation; data obtained from different detection methods are isolated and cannot be mutually verified, leading to frequent false alarms and missed alarms. Third, they rely on static detection modes, such as manual inspection during shutdown, which cannot capture the dynamic evolution characteristics during operation and make it difficult to predict failure trends. Fourth, the early warning information is simple, only indicating the existence of damage without specifying the location, type, and severity of the damage, making it difficult for on-site maintenance personnel to handle the situation quickly and accurately.

[0064] To address the aforementioned issues, this application utilizes multispectral radiation response to detect the absorption and phase characteristics of the adhesive tape material, making it sensitive to internal structural changes such as wire rope breakage and interface debonding; and it detects changes in the mechanical integrity of the tape through structural vibration response, making it sensitive to fluctuation reflections caused by decreased structural stiffness and localized damage. These two physical mechanisms are different but complementary; spatially correlated analysis allows for mutual verification and the elimination of interference. Furthermore, by constructing a joint absorption-phase distribution map, abstract spectral data is transformed into a visualized feature space, and statistical distance is used to identify anomalies, solving the problem of identifying damage from complex spectral data; vibration propagation consistency analysis locates regions of decreased structural integrity, solving the problem of locating damage from vibration signals; spatiotemporal consistency verification confirms damage only when the two characteristics stably correspond, solving the problem of avoiding false alarms from a single method; and by tracking damage evolution trends to construct physical criteria, the application upgrades from static detection to dynamic prediction, solving the problem of predicting failure risks.

[0065] The system can simultaneously utilize the physical properties of light and vibration to identify early internal damage in real time and accurately during conveyor belt operation. It can also precisely indicate the location, type, and progression of the damage, enabling maintenance personnel to take timely and targeted measures, such as planned repairs or replacements. This effectively prevents serious accidents such as belt breakage and tearing, transforming the conveyor maintenance mode from reactive repair to predictive maintenance, and significantly reducing downtime losses and safety risks.

[0066] Multispectral radiation response and structural vibration response were collected.

[0067] 1. Sensor deployment.

[0068] In each detection section along the conveyor belt's running path, multiple sets of multispectral imaging sensors and vibration sensors are deployed along the belt's running direction and width. The multispectral imaging sensors employ a linear array scanning method, with each sensor covering a specific bandwidth; multiple sensors arranged side-by-side achieve coverage of the entire bandwidth. The sensors are mounted on a crossbeam approximately 0.5 meters above the conveyor belt, angled downwards towards the belt surface to avoid contact. Multiple sets of sensors are deployed along the running direction in each detection section, for example, one set every two meters, to ensure continuous coverage of the passing conveyor belt. The vibration sensors are intrinsically safe mining accelerometers, mounted on supports beneath the conveyor belt and mounted one every meter along the running direction in each detection section, forming a dense array.

[0069] 2. Multispectral response acquisition.

[0070] A multispectral imaging sensor is controlled to emit detection radiation in at least three preset characteristic bands towards the surface and near-surface of the tape. These bands are: the first band (near-infrared, center wavelength 950 nm) is sensitive to the interface between the steel wire rope and the rubber; the second band (mid-infrared, center wavelength 3400 nm) is sensitive to rubber aging; and the third band (ultraviolet, center wavelength 365 nm) is sensitive to surface microcracks. The sensor receives the reflection and scattering intensity data of the tape to the detection radiation, and the intensity value of each band is recorded as a 16-bit grayscale value. Simultaneously, the encoder acquires the tape's running position pulses, associating each scan line with the absolute position (in meters) on the tape. The data of each band are organized according to spatial position to form a multispectral response set, which is a three-dimensional array: the first dimension is the running direction position, the second dimension is the bandwidth direction position, and the third dimension is the band channel.

[0071] 3. Acquisition of structural vibration response.

[0072] The vibration acceleration time-domain signal output from the vibration sensors is acquired. Each sensor continuously acquires data at a sampling rate of 2,000 points per second, and the data is stored as a 16-bit integer after anti-aliasing filtering. Each sampling point is accompanied by a precise timestamp and is associated with the sensor location (installation coordinates). Data from all sensors is acquired synchronously to ensure that vibration signals from different locations at the same time have a unified time reference. This data constitutes a set of structural vibration responses, stored in time series format, with each channel corresponding to a sensor location.

[0073] In the scenario of a conveyor belt in a main inclined shaft of a coal mine, this technical solution achieves the following specific effects: By deploying multiple sets of sensors along the running direction and width direction, it achieves full coverage of the entire width and length of the conveyor belt without omissions, solving the problem of potential missed detections in point-based detection; by emitting three characteristic wavebands through a multispectral imaging sensor, targeting different damage types (interface, aging, cracks), the system can distinguish different types of early defects, improving the targeting of detection; through a dense array of vibration sensors, it can capture subtle changes in the propagation of vibration waves along the conveyor belt, with positioning accuracy reaching the centimeter level, providing high-resolution location information for subsequent spatial correlation; through synchronous acquisition and location correlation, it lays a solid foundation for the spatiotemporal fusion of multimodal data, ensuring the reliability of subsequent verification.

[0074] Construct a joint distribution map of energy absorption and phase hysteresis.

[0075] 1. Obtain the absorption attenuation coefficient sequence and the phase hysteresis sequence.

[0076] For each characteristic band in the multispectral response set, the system calculates the absorption attenuation coefficient point by point. Let the reference reflection intensity at a certain point in a certain band be ( (The test is performed using intact tape at the beginning of the test section; periodic calibration is required.) The actual reflection intensity is (I), then the absorption attenuation coefficient is ( The scattering attenuation coefficient can also be calculated for scattering intensity, but this embodiment mainly uses reflection intensity. Simultaneously, the phase lag is extracted using the aforementioned orthogonal demodulation method. .

[0077] 2. Pairing association and two-dimensional mapping.

[0078] By pairing the absorption attenuation coefficient and phase hysteresis under the same characteristic band, data pairs are obtained. Each spatial location corresponds to a data pair in each band. All data pairs are plotted as a scatter plot in a two-dimensional coordinate system consisting of the absorption attenuation coefficient axis and the phase hysteresis axis. To enhance visualization, the scatter points can be density-colored to form a joint distribution map. This map visually demonstrates the characteristic distribution of the tape material in different states: intact material typically clusters in a central region, while damaged material deviates from this region.

[0079] By pairing and correlating the absorption attenuation coefficient with the phase hysteresis, two independent physical quantities in the multispectral signal—amplitude information and phase information—are fully utilized. Compared with methods that only use intensity, this provides richer material state characteristics and can identify internal damage (such as interface debonding) that has no obvious absorption changes but abnormal phase. By constructing a two-dimensional joint distribution map, the high-dimensional spectral data is visualized by reducing its dimensionality, allowing operators to intuitively understand the distribution of the tape's health status and facilitating the algorithm's automatic identification of abnormal clusters. Pairing within the same band ensures the homology of features and avoids ambiguity caused by cross-band feature aliasing.

[0080] Identify regions of spectral anomalies.

[0081] 1. Define the baseline distribution area.

[0082] Immediately after tape replacement or overhaul, the system collects a set of multispectral response data as a sample of good condition. The absorption-phase data pairs of these samples in each band are statistically analyzed to calculate the mean vector. Sum of covariance matrix ( In two-dimensional space, an elliptical region is defined as the reference distribution region, and the center of the ellipse is ( The boundary is determined by Mahalanobis distance (e.g., the distance corresponding to a 95% confidence level). This ellipse represents the spectral characteristic distribution range of normal adhesive tape materials. The sample size should be large enough (e.g., at least one hundred groups) to ensure statistical stability.

[0083] 2. Calculate the deviation index.

[0084] For each newly collected data point ( Calculate its Mahalanobis distance to the baseline distribution region: This distance measures how far the current point deviates from the normal distribution.

[0085] 3. Threshold determination and suspected damage point marking.

[0086] The calculated Mahalanobis distance is compared with the preset damage assessment threshold. Compare. The determination method can be: taking the 95th quantile of the Mahalanobis distance of the benchmark sample, or optimizing it through historical accident data, for example, to balance the false alarm rate and the false negative rate. If ( If the point is identified as a suspected damage point, its spatial coordinates are recorded.

[0087] 4. Spatial clustering forms spectral anomaly regions.

[0088] Spatial clustering was performed on suspected damage points marked across multiple consecutive detection cycles. A density-based clustering algorithm was employed, using distance along the running direction and bandwidth direction as clustering features. Neighborhood radii were set, such as 20 cm along the running direction and 5 cm along the bandwidth direction, with a minimum number of points, such as 5. After clustering, the minimum bounding rectangle of each cluster was defined as the spectral anomaly region, and the center coordinates and boundary of this region were calculated. The clustering process eliminated isolated noise points, giving the detected damage regions practical physical meaning.

[0089] By defining the baseline distribution area based on intact data, the damage assessment becomes adaptive, eliminating the influence of factors such as differences in tape materials from different batches and changes in ambient temperature, thus improving the robustness of the criteria. The deviation is calculated by Mahalanobis distance, which comprehensively considers the correlation between absorption and phase, making it more scientific than a single threshold. Spatial clustering merges isolated suspected points into continuous regions, eliminating single-point noise interference and giving the detected damage areas actual physical significance, such as how a broken steel wire rope often leads to an anomaly in an entire area.

[0090] Spatiotemporal consistency check.

[0091] 1. Extract the center coordinates.

[0092] Extracting the center coordinates of spectral anomaly regions ( ,in( The position along the direction of travel (with the starting point of the detection section as the origin), The location is along the bandwidth (originating from the left edge). The center coordinates are extracted from the vibration propagation distortion features. Obtained using the same method.

[0093] 2. Calculate the Euclidean distance.

[0094] Calculate the Euclidean distance between the two center coordinates in the tape coordinate system: 3. Multi-period stability correspondence determination.

[0095] Set spatial tolerance threshold ( (For example, 10 centimeters). Within multiple consecutive testing cycles (for example, five cycles), check the following two conditions: Condition 1: The calculated (d) for each period is less than ( .

[0096] Condition 2: In each cycle, both spectral anomaly regions and vibration propagation distortion characteristics are detected (i.e., they coexist).

[0097] If both conditions are met, it is determined that the two have formed a stable correspondence in space and time, confirming that there is structural damage at that location.

[0098] By requiring multiple cycles to continuously meet spatial proximity conditions, single-time mismatches caused by accidental factors such as tape vibration and sensor noise are effectively eliminated, improving the reliability of damage confirmation. By requiring both to occur synchronously, misjudgments caused by time asynchrony are avoided, such as spectral anomalies disappearing but vibration anomalies remaining. The 10-centimeter spatial tolerance takes into account the positioning errors of the two detection methods, ensuring correlation accuracy while allowing for a certain error tolerance. The finally confirmed damage location integrates information from both modes, which is more accurate than a single mode, providing a reliable foundation for subsequent precise positioning and evolution tracking.

[0099] Risk level warnings are issued based on physical criteria.

[0100] 1. Predefine risk levels and thresholds.

[0101] The system predefines three progressively higher risk levels: Level 1 (Attention Level), Level 2 (Warning Level), and Level 3 (Danger Level). Each level corresponds to a set of thresholds. Level 1 threshold: Threshold for the rate of change of vibration-reflected energy ( Spectral anomaly spread rate threshold ( .

[0102] Secondary threshold: , .

[0103] Level 3 threshold: , .

[0104] These thresholds are determined based on the characteristics of the tape material and historical failure case data, for example, by analyzing historical tape breakage data to determine the critical values ​​for each level.

[0105] 2. Quantitative evolution indicators.

[0106] From subsequent tests, the rate of change of vibration reflection energy ratio (R) and the rate of spectral anomaly spread (E) were calculated. The calculation method has been detailed in the independent patent.

[0107] 3. Level determination and early warning output.

[0108] Compare (R) and (E) with the threshold corresponding to the current risk level. The system starts from level one; if ( and( If the condition is met, a Level 1 warning will be issued; in subsequent detections, if ( and( If the threshold is exceeded, the alert level is raised to Level 2; and so on. The alert level is only upgraded when both indicators simultaneously exceed their corresponding thresholds. This reflects the rigor of the two-factor criterion and ensures the reliability of the alert.

[0109] By combining two independent indicators, the rate of change of vibration reflection energy ratio and the spectral expansion rate, damage development is comprehensively assessed from both mechanical and material dimensions, avoiding misjudgments that may arise from a single indicator (for example, increased vibration reflection may be caused by roller failure, but if the spectrum does not expand, it may be a misjudgment). The graded early warning mechanism provides maintenance personnel with clear priorities for handling: a level 1 warning allows for planned inspections, a level 2 warning requires shortening the inspection cycle, and a level 3 warning requires immediate shutdown, achieving refined risk management. The system only escalates when both indicators exceed limits simultaneously, reflecting a conservative safety principle and reducing unnecessary downtime losses.

[0110] Construction and application of historical damage database.

[0111] 1. Construct a historical damage record.

[0112] After each damage confirmation, the system integrates the relevant information into a historical damage record. The record includes: Specific spatial location of the damage (length in meters, width in width).

[0113] Characteristic data of spectral anomaly regions: average absorption attenuation coefficient, average phase lag, and region area for each band.

[0114] Characteristic data of vibration propagation distortion: reflected energy ratio, anomaly type (enhancement / dissipation), center frequency, etc.

[0115] Spatiotemporal correlation: distance to center coordinates, number of stable corresponding periods.

[0116] Time of first discovery of damage, time of last update.

[0117] Subsequent evolution tracking data: rate of change of vibration-reflection energy ratio, expansion rate, etc.

[0118] These records are stored in a damage pattern database, which uses a relational or time-series database structure.

[0119] 2. Similarity calculation.

[0120] In subsequent detection, when the system identifies new suspected damage, it extracts the corresponding spectral feature data (multi-band absorption-phase mean) and vibration feature data (reflectance energy ratio, anomaly type) to construct a feature vector. For each historical record in the database, extract its stored feature data ( Calculate composite similarity: in( Weights can be preset through expert experience, such as... Alternatively, learn from historical data.

[0121] Spectral similarity ( Cosine similarity is used: in( It is a vector composed of multi-band absorption coefficients and phase lag.

[0122] Vibration similarity ( Using normalized Euclidean distance: in( This includes the reflection energy ratio and anomaly type encoding, such as enhancement = 1, dissipation = 0, ( ) is the preset maximum distance normalization factor, such as taking the maximum distance among all historical samples.

[0123] 3. Auxiliary diagnosis.

[0124] If the composite similarity is greater than the preset matching threshold (e.g., 0.85), the current suspected damage is determined to belong to the same damage pattern as the corresponding historical record. The system outputs the evolution information stored in the historical record as an auxiliary diagnostic conclusion, such as "This damage pattern is similar to the damage at joint No. 3 last year. The subsequent expansion rate is about two millimeters per day. It is recommended to arrange an examination within two weeks."

[0125] By constructing a historical damage database, the experience gained from each inspection is accumulated, enabling the system to learn. New damage can be compared with historical cases to obtain its possible development trends and treatment suggestions, providing a reference for maintenance decisions. Composite similarity, combined with spectral and vibration characteristics, makes the matching more accurate and avoids mismatches caused by matching based on a single feature. The auxiliary diagnostic conclusions can prompt maintenance personnel to refer to historical experience, especially for repetitive damage (such as specific joint locations), which can predict its typical evolution path in advance, enabling more accurate predictive maintenance.

[0126] Location vibration propagation distortion characteristics.

[0127] 1. Define the vibration signal analysis window.

[0128] Centered on the spectral anomaly region, a rectangular window is formed by extending a certain length (e.g., one meter) forward and backward along the direction of tape travel, and a certain width (e.g., twenty centimeters) to the left and right along the tape's width direction. The coordinate range of the window is determined based on the tape coordinate system.

[0129] 2. Extract vibration signals.

[0130] From the set of structural vibration responses, the time-domain signals of all vibration sensors within the spatiotemporal range of the window are extracted. Since the vibration sensors are discretely deployed, the window may include multiple sensors. The system selects the sensor signal closest to the center of the window, or performs a weighted average of the signals from multiple sensors within the window (the weight is inversely proportional to the distance).

[0131] 3. Frequency domain analysis and calculation of reflected energy ratio.

[0132] The extracted vibration acceleration time-domain signal is subjected to a Fast Fourier Transform (FFT) to obtain the spectrum. The preset characteristic frequency band is determined based on the characteristics of the excitation source: based on the roller passing frequency (…). in v) is the belt speed, ( ( ) is the roller spacing) and its multiples ( As the center of the characteristic frequency band, the bandwidth of each frequency band is set to ( It can dynamically track and adjust for changes in band speed. In the spectrum, the energy within the characteristic frequency band is calculated ( and total signal energy ( (Limited to the effective frequency band). Reflected energy ratio ( .

[0133] 4. Region determination.

[0134] Will( Compared with the normal threshold range calibrated based on healthy tape samples ( Comparison: like( If so, it is determined to be a region of enhanced vibration reflection caused by structural weakening.

[0135] like( If so, it is determined to be a region of abnormal dissipation of vibration energy caused by structural damage.

[0136] If it is within the range, it is considered normal.

[0137] In multiple consecutive detection cycles, spatiotemporal clustering is performed on the identified enhancement or dissipation regions, similar to the clustering of spectral anomaly regions, to form stable vibration propagation distortion characteristics, and the center coordinates and anomaly types are recorded.

[0138] By defining a window centered on the spectral anomaly region, preliminary spatial correlation between the two modes was achieved, allowing vibration analysis to focus on potential damage areas and improving computational efficiency. The reflection energy ratio calculation simplifies complex vibration signals into a physically meaningful index, intuitively reflecting changes in structural integrity. Distinguishing between enhancement and dissipation anomalies allows for preliminary assessment of damage characteristics; for example, enhanced reflection may correspond to cracks, while dissipation may correspond to loosening, providing more information for subsequent diagnosis. Multi-period spatiotemporal clustering ensures feature stability and eliminates transient interference.

[0139] Analyze the consistency of vibration energy propagation.

[0140] 1. Calculate the longitudinal propagation velocity and the transverse vibration energy amplitude.

[0141] Based on the structural vibration response set, the longitudinal propagation velocity of the vibration wave along the conveyor belt running direction is calculated using cross-correlation analysis of signals from two adjacent vibration sensors. The specific method is as follows: Select two sensors, the distance between which is known ( ), calculate the cross-correlation function of its vibration signal. The time delay corresponding to the peak ( That is, the wave propagation time, then ( Simultaneously, the vibration energy amplitude of each sensor in the bandwidth direction (lateral direction) is extracted. The root mean square value can be obtained by bandpass filtering (to remove low-frequency interference) and then calculating the root mean square value.

[0142] 2. Determine the target analysis segment.

[0143] Based on the spatial location of the spectral anomaly region, determine the target analysis segment covered along the operating direction. For example, the anomaly region starts from ( arrive( The analysis segment is ( ,in( To extend the length (e.g., 0.5 meters) to ensure the entire affected area is included.

[0144] 3. Calculate the propagation consistency index.

[0145] Extract the vibration sensor signals from both ends of the target analysis segment, i.e., the foremost and last points. Calculate the longitudinal propagation velocity of the vibration wave before it enters the segment (before the start of the analysis segment) and after it exits (before the end of the analysis segment), respectively, to obtain ( and( Calculate the rate of change of longitudinal propagation velocity: Simultaneously, the lateral vibration energy amplitude of each sensor within the analysis section is extracted, and the average value is calculated. A section at least two meters long, confirmed by historical data to be undamaged, is selected in front of the damaged area (away from the damaged side) as a healthy reference section. The average value of the lateral energy amplitude of all vibration sensors within this section is taken as the baseline of the healthy section. (Updated every monitoring cycle to track changes in operating conditions). Calculate the lateral vibration energy amplitude attenuation rate: 4. Quantitative assessment.

[0146] Will( and( Each is compared with a preset health baseline threshold, such as , By comparing ( ), the consistency of vibration energy propagation of the tape in the target analysis section is quantitatively evaluated. If ( or( Exceeding a threshold indicates a decrease in propagation consistency, suggesting structural damage. These indicators can be combined with reflection energy ratio analysis to provide a more comprehensive profile of vibration damage.

[0147] The longitudinal propagation velocity change rate reflects the change in the overall stiffness of the conveyor belt, while the transverse energy attenuation rate reflects the change in local damping characteristics. These two indicators characterize the impact of damage on vibration propagation from different perspectives and complement each other. By comparing the signals at both ends of the analysis section, the influence of excitation source differences is effectively eliminated, making the indicators more sensitive to damage. The dynamic updating of the baseline of the healthy section ensures that the indicators adapt to changes in working conditions. This analysis provides more evidence for spatiotemporal verification and enhances the reliability of damage confirmation.

[0148] The generation and execution of control commands.

[0149] 1. Predefined mapping strategy.

[0150] The system predefines a mapping strategy table between risk levels, damage modes, and belt conveyor control commands. This mapping strategy is based on safety procedures and expert experience, for example: Level 1 warning (attention level) and damage mode is "minor wear of the overlay": generate "recommend planned inspection" instruction.

[0151] Level 2 warning (alert level) and damage mode is "wire rope core breakage": generate the instruction "reduce speed to 50% and strengthen monitoring".

[0152] Level 3 warning (danger level) and damage mode is "severe fatigue of joint": generate "emergency stop" command.

[0153] If a historical record shows that the damage pattern has caused a belt break, an "emergency stop" command is generated directly, regardless of the current level.

[0154] 2. Obtain the current risk level and auxiliary diagnostic conclusions.

[0155] Auxiliary diagnostic findings may indicate a specific known damage pattern (such as fatigue of joint No. 3).

[0156] 3. Generate control commands.

[0157] Based on the mapping strategy, and combined with the current risk level and auxiliary diagnostic conclusions, corresponding control instructions are generated. For example, if the current level is a Level 2 warning and the auxiliary diagnostic conclusions are similar to the pattern before the belt break last year, the mapping strategy may trigger an emergency shutdown instruction.

[0158] 4. Perform control actions.

[0159] The generated control commands are sent to the PLC control system of the belt conveyor via the industrial ring network. The PLC executes corresponding actions, such as reducing speed via the frequency converter, triggering the emergency stop circuit to stop the machine, or sending a work order to the maintenance dispatch system. Simultaneously, the system records the command execution results for subsequent evaluation.

[0160] By mapping risk levels and damage modes to specific control commands, a closed loop from detection results to automated handling is achieved, avoiding delays caused by manual decision-making. Utilizing historical damage modes to assist decision-making enables stricter controls, such as direct shutdown, to be implemented in advance for known high-risk modes, demonstrating experience-driven intelligent protection. Multiple control actions (speed reduction, shutdown, and scheduling) provide differentiated responses for different risk levels, ensuring safety while avoiding unnecessary downtime losses. This mechanism endows the conveyor with self-protection capabilities, enabling it to automatically take measures before internal damage develops to a dangerous stage, significantly reducing the probability of serious accidents such as belt breakage.

[0161] The foregoing has shown and described the basic principles, main features, and advantages of this application. Those skilled in the art should understand that this application is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this application. Various changes and modifications can be made to this application without departing from the spirit and scope thereof, and all such changes and modifications fall within the scope of this application as claimed. The scope of protection of this application is defined by the appended claims and their equivalents.

Claims

1. A method for damage detection and early warning of belt conveyors based on multispectral vibration sensing, characterized in that, The method includes: During the stable operation of the conveyor belt, multispectral radiation response and structural vibration response are collected to form multispectral response set and structural vibration response set, respectively. Based on the aforementioned multispectral response set, a joint distribution map of energy absorption and phase hysteresis is constructed to identify spectral anomaly regions where absorption intensity and phase hysteresis are simultaneously unbalanced due to internal damage to the tape. Using the set of structural vibration responses as input, and combining it with the spatial location corresponding to the spectral anomaly region, the consistency of vibration energy propagation is analyzed, the region of enhanced vibration reflection or abnormal dissipation caused by decreased structural integrity is located, and vibration propagation distortion characteristics spatially associated with the spectral anomaly region are formed. The spatiotemporal consistency of the spectral anomaly region and the vibration propagation distortion characteristics is checked. Only when the two form a stable correspondence on the tape running path is it determined that there is structural damage, a damage confirmation result is generated, and the specific spatial location of the damage in the length and width directions of the tape is determined. Based on the specific spatial location corresponding to the damage confirmation results, the changing trend of vibration energy reflection ratio over time and the spatial expansion rate of spectral anomaly regions are analyzed. Physical criteria for the evolution of damage from local defects to operational failure are constructed, and an early warning of the operational risk level of the belt conveyor is output based on the physical criteria.

2. The method according to claim 1, characterized in that, The acquisition of multispectral radiation response and structural vibration response forms a multispectral response set and a structural vibration response set, respectively, including: In the detection section of the conveyor belt's running path, multiple sets of multispectral imaging sensors and vibration sensors are deployed along the belt's running direction and width direction; The multispectral imaging sensor is controlled to emit detection radiation including at least three preset characteristic bands to the surface and near-surface of the tape, and the reflection intensity and scattering intensity data of the tape to the detection radiation are received and recorded to form a multispectral response set. The vibration acceleration time-domain signal output by the vibration sensor is collected to form a set of structural vibration responses.

3. The method according to claim 1, characterized in that, The construction of a joint distribution map of energy absorption and phase hysteresis based on the multispectral response set includes: The multispectral response set is processed to obtain the absorption attenuation coefficient sequence and phase hysteresis sequence of the tape in each characteristic band. The absorption attenuation coefficient and phase hysteresis in the same characteristic band are paired and correlated to form paired correlation data. Multiple sets of paired and correlated data are mapped and fitted in a two-dimensional coordinate system consisting of the absorption attenuation coefficient axis and the phase hysteresis axis to generate a joint distribution map for identifying internal damage.

4. The method according to claim 3, characterized in that, The identification of spectral anomalies caused by internal damage to the tape, resulting in a synchronous imbalance between absorption intensity and phase hysteresis, includes: In the joint distribution map, a baseline distribution region characterizing the normal material state is defined based on the historical multispectral response set of the tape in its intact state. For the newly acquired multispectral response set, the statistical distance between the corresponding data point in the joint distribution map and the baseline distribution area is calculated as a deviation index. Determine whether the deviation index exceeds a preset damage assessment threshold; When the deviation index exceeds the damage determination threshold, the spatial location of the tape that generated the data point is determined as a suspected damage point. Based on the spatial clustering of multiple consecutive suspected damage points, the spectral anomaly region is determined.

5. The method according to claim 1, characterized in that, The step of performing a spatiotemporal consistency check between the spectral anomaly region and the vibration propagation distortion characteristics, and determining the existence of structural damage only when the two form a stable correspondence on the conveyor belt's running path, includes: The center coordinates of the spectral anomaly region are obtained as the first position, and the center coordinates of the vibration propagation distortion feature are obtained as the second position. Calculate the Euclidean distance between the first position and the second position in the tape coordinate system; Within multiple consecutive detection cycles, it is determined whether the Euclidean distance is less than a preset spatial tolerance threshold, and whether the spectral anomaly region and the vibration propagation distortion feature are detected synchronously within the corresponding cycle; when the above conditions are met simultaneously, it is determined that the spectral anomaly region and the vibration propagation distortion feature form a stable correspondence in space and time, confirming the existence of structural damage.

6. The method according to claim 1, characterized in that, The method of outputting an operational risk level warning for the belt conveyor based on the physical criteria includes: Multiple incremental risk levels are predefined, and corresponding vibration reflection intensity thresholds and spectral anomaly spread rate thresholds are set for each level. The quantized value of the change trend of the detected vibration reflection intensity is compared with the vibration reflection intensity threshold, and the expansion rate value of the detected spectral anomaly region is compared with the spectral anomaly expansion rate threshold corresponding to the current risk level. The risk level is raised to the next level only when the vibration reflection intensity exceeds the corresponding threshold and the expansion rate value exceeds the corresponding threshold, and a warning signal including the current risk level is generated and output.

7. The method according to claim 1, characterized in that, The method further includes: After obtaining the damage confirmation result, the characteristic data of the spectral anomaly region, the characteristic data of the vibration propagation distortion characteristics, and the spatiotemporal correlation between the two are bound and integrated with the determined specific spatial location to form a historical damage record and stored in the damage pattern database. In subsequent detection, when new suspected damage is identified, the corresponding spectral and vibration feature data are extracted, and the composite similarity with each historical damage record in the damage pattern database is calculated. If the composite similarity is greater than the preset matching threshold, the current suspected damage and the corresponding historical record are determined to belong to the same damage pattern, and the evolutionary information included in the historical record is output as an auxiliary diagnostic conclusion for the current damage.

8. The method according to claim 1, characterized in that, The location of the region with enhanced vibration reflection or abnormal dissipation due to decreased structural integrity forms a vibration propagation distortion feature spatially associated with the spectral anomaly region, including: A vibration signal analysis window is defined in the tape coordinate system, centered on the spectral anomaly region. From the set of structural vibration responses, extract the vibration acceleration time-domain signal corresponding to the spatiotemporal range of the vibration signal analysis window; The intercepted vibration acceleration time-domain signal is transformed and analyzed in the frequency domain to calculate the ratio of reflected wave energy related to the excitation source to the total signal energy within the preset characteristic frequency band. The reflected energy ratio is compared with the normal threshold range calibrated based on healthy adhesive tape samples: If the reflected energy ratio is higher than the upper limit of the normal threshold range, the area corresponding to the window is determined to be a vibration reflection enhancement area caused by structural weakening. If the reflected energy ratio is lower than the lower limit of the normal threshold range, the area corresponding to the window is determined to be the abnormal dissipation area of ​​vibration energy caused by structural damage. Within multiple consecutive detection cycles, spatiotemporal clustering is performed on the identified regions to form stable vibration propagation distortion characteristics.

9. The method according to claim 1, characterized in that, The analysis of the consistency of vibration energy propagation includes: Based on the aforementioned structural vibration response set, the longitudinal propagation velocity of the vibration wave along the belt running direction and the transverse vibration energy amplitude along the belt width direction are calculated respectively. Based on the spatial location of the spectral anomaly region, the target analysis segment covered along the running direction is determined, the vibration signals corresponding to both ends of the target analysis segment are extracted, and the longitudinal propagation velocity change rate and the transverse vibration energy amplitude attenuation rate before and after the vibration wave is transmitted are calculated. The calculated rate of change and the attenuation rate are compared with preset health baseline thresholds to quantitatively evaluate the consistency of vibration energy propagation of the tape in the target analysis section.

10. The method according to claim 1, characterized in that, The method further includes: A predefined mapping strategy between risk levels, damage modes, and belt conveyor control commands; Obtain the operational risk level output by the physical criteria, and the auxiliary diagnostic conclusions obtained by matching historical damage records; When the operational risk level reaches a preset high-risk threshold, or when the auxiliary diagnostic conclusion indicates a specific type of known damage mode, a corresponding control command is generated according to the mapping strategy. Executing the control command triggers at least one control action, including slow-down operation, emergency shutdown, or planned maintenance scheduling.

Images