A belt abnormal sound early warning method, device and medium based on pattern recognition

Abstract

This invention relates to the field of acoustic and vibration monitoring technology, and discloses a method, device, and medium for early warning of abnormal belt noise based on pattern recognition. The method includes: deploying acoustic and vibration sensing optical cables along the fixed components of the belt conveyor; collecting acoustic and vibration data and spatially calibrating the idlers and rollers to generate a normalized acoustic and vibration energy map and a high-frequency envelope map; calculating the interpretability of the fixed components and the interpretability of belt movement based on the normalized acoustic and vibration energy map to screen candidate objects of anchoring anomalies in the fixed components; calculating the rotation phase reproduction coefficient by combining the rotation phase of the fixed components and the high-frequency envelope map to screen objects of mechanically reproduced abnormal noise; performing nonlinear gated coupling on the objects of mechanically reproduced abnormal noise to obtain the confidence level of the mechanical abnormal noise and updating the health accumulation value, generating a belt noise early warning record, thus improving the relevance and continuity of the early warning results. This invention improves the relevance and continuity of belt noise early warning results.

Description

Technical Field

[0001] This invention relates to the field of acoustic and vibration monitoring technology, and in particular to a method, device, and medium for early warning of abnormal noise from belts based on pattern recognition. Background Technology

[0002] Belt conveyors are widely used in continuous conveying scenarios such as ports, mines, power plants, and metallurgy. During long-term operation, idlers, drums, frame structures, and belts generate complex acoustic and vibration responses related to friction, impact, howling, eccentric vibration, and local structural disturbances. Existing belt conveyor noise monitoring typically uses distributed fiber optic acoustic sensors, acoustic sensors, vibration sensors, or inspection devices to collect operating signals along the belt frame, idler seats, and drum support parts. These signals are then combined with frequency band energy, amplitude thresholds, time-frequency characteristics, acoustic feature recognition, or spatial positioning algorithms to generate abnormal locations, acoustic spectrograms, trend curves, and alarm records. These data are used to assist in judging the operating status of idler bearing friction, local drum anomalies, belt friction, and structural impact.

[0003] In conventional methods, short-term acoustic vibration segments are easily affected by materials carried by belts, local impacts, and environmental noise. It is difficult to distinguish between abnormal noises from fixed components and moving sound sources based on a single time window. In addition, conventional identification results mostly stay at the level of abnormal segment judgment, lacking a joint expression of the fixed component's position anchoring, rotation phase reproduction, and health accumulation process, resulting in an insufficient correlation between the early warning results and the evolution of mechanical deterioration. Summary of the Invention

[0004] In view of the above-mentioned existing problems, the present invention provides a belt noise early warning method based on pattern recognition, in order to solve the problems of difficulty in distinguishing between abnormal noises from fixed components and moving sound sources along the belt, and insufficient correlation between abnormal segment judgment and mechanical deterioration evolution in the existing technology.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a belt conveyor noise early warning method based on pattern recognition, comprising: laying acoustic vibration sensing optical cables along the frame structure corresponding to the fixed components of the belt conveyor; collecting belt conveyor acoustic vibration data; spatially calibrating idlers and rollers as fixed components; recording the center position of the fixed components, the monitoring space range of the fixed components, the diameter of the fixed components, the correspondence between the belt running direction and the optical cable distance direction, and the belt running speed; generating a normalized acoustic vibration energy map and a high-frequency envelope map; calculating the fixed component interpretation degree for each fixed component based on the normalized acoustic vibration energy map; and determining the interpretation degree based on the belt running speed and the belt running direction and the optical cable distance direction. Based on the correspondence between directions, the normalized acoustic vibration energy map is backtracked to determine the position of the moving component, and the explanatory power of the moving component is calculated. Combining the explanatory power of the fixed component and the explanatory power of the moving component, the anchoring coefficient of the fixed component is calculated, and candidates for anchoring anomalies of the fixed component are screened. For only the candidates for anchoring anomalies of the fixed component, the rotation phase of the fixed component is determined based on the diameter of the fixed component and the running speed of the belt, and the rotation phase reproduction coefficient is calculated based on the high-frequency envelope diagram to screen the mechanically reproduced abnormal noise objects. Nonlinear gated coupling is performed on the mechanically reproduced abnormal noise objects to obtain the mechanical abnormal noise confidence level. The health accumulation value is updated based on the mechanical abnormal noise confidence level to generate belt abnormal noise early warning records.

[0006] As a preferred embodiment of the belt vibration early warning method based on pattern recognition described in this invention, the generation of the normalized acoustic vibration energy map and high-frequency envelope map includes: establishing optical cable distance coordinates with the starting end of the acoustic vibration sensing optical cable as the distance zero point; numbering the fixed components of the idlers and rollers to form a basic data table of fixed components; calibrating the corresponding position of each fixed component in the optical cable distance coordinates to obtain the center position of the fixed component; determining the monitoring space range of the fixed component based on the center position of the fixed component and the installation spacing between adjacent fixed components; recording the correspondence between the diameter of the fixed component, the belt running direction and the optical cable distance direction, and the belt running speed; dividing the belt conveyor acoustic vibration data into analysis periods, extracting the energy of the vibration-sensitive frequency band and establishing a normal acoustic vibration energy benchmark, combining the vibration-sensitive frequency band energy and the normal acoustic vibration energy benchmark to generate a normalized acoustic vibration energy map, and generating a high-frequency envelope map through high-frequency filtering and envelope extraction.

[0007] As a preferred embodiment of the belt noise early warning method based on pattern recognition described in this invention, the step of generating a normalized acoustic vibration energy map by combining the noise-sensitive frequency band energy with the normal acoustic vibration energy benchmark includes: comparing the noise-sensitive frequency band energy at the same optical cable distance position within the analysis period with the median value of the normal energy in the normal acoustic vibration energy benchmark; when the noise-sensitive frequency band energy is not higher than the median value of the normal energy, setting the normalized acoustic vibration energy value of the corresponding optical cable distance position within the corresponding analysis period to zero; when the noise-sensitive frequency band energy is higher than the median value of the normal energy, performing normalization processing based on the discrete value of the normal energy of the normal acoustic vibration energy benchmark to obtain the normalized acoustic vibration energy value of the corresponding optical cable distance position within the corresponding analysis period.

[0008] As a preferred embodiment of the belt noise early warning method based on pattern recognition described in this invention, the step of calculating the interpretability of each fixed component includes: reading the monitoring space range of each fixed component from the fixed component basic data table; removing invalid optical cable distance positions within the monitoring space range of the fixed component, and arranging the valid optical cable distance positions according to the optical cable distance coordinates; setting the number of continuous analysis cycles participating in competitive identification; and averaging the normalized acoustic and vibration energy values ​​at the valid optical cable distance positions within the corresponding fixed component monitoring space range in the current analysis cycle and the continuous analysis cycles going back in the historical direction to obtain the interpretability of the fixed component.

[0009] As a preferred embodiment of the belt noise early warning method based on pattern recognition described in this invention, the step of backtracking the belt movement position of the normalized acoustic energy map and calculating the belt movement interpretability includes: accumulating the belt movement distance between the historical analysis period and the current analysis period based on the average belt running speed and the analysis period update interval within each analysis period; and backtracking the effective optical cable distance position within the monitoring space of the fixed component based on the belt movement distance and the correspondence between the belt running direction and the optical cable distance direction to obtain the belt movement... Backtracking position; compare the backtracking position with the effective acquisition range of the acoustic vibration sensing optical cable. The backtracking position that falls within the effective acquisition range of the acoustic vibration sensing optical cable is included in the calculation of the backtracking position ...

[0010] As a preferred embodiment of the belt noise early warning method based on pattern recognition described in this invention, the step of calculating the anchoring coefficient of the fixed component and screening candidate objects for fixed component anchoring anomalies by combining the interpretability of the fixed component and the interpretability of the belt movement includes: when the sum of the interpretability of the fixed component and the interpretability of the belt movement is greater than zero, calculating the anchoring coefficient of the fixed component based on the ratio between the difference between the interpretability of the fixed component and the interpretability of the belt movement and the sum of the interpretability of the fixed component and the interpretability of the belt movement; when the sum of the interpretability of the fixed component and the interpretability of the belt movement is not greater than zero, recording the anchoring coefficient of the fixed component as zero; when the fixed component simultaneously satisfies the conditions that the interpretability of the fixed component is not less than the interpretability threshold of the fixed component and the anchoring coefficient of the fixed component is not less than the anchoring threshold of the fixed component within the corresponding analysis period, marking the corresponding fixed component as a candidate object for fixed component anchoring anomalies.

[0011] As a preferred embodiment of the belt noise early warning method based on pattern recognition described in this invention, the step of determining the rotation phase of the fixed component based on the diameter of the fixed component and the belt running speed, and calculating the rotation phase reproduction coefficient in combination with the high-frequency envelope map to screen mechanical noise reproduction objects includes: performing rotation phase reproduction identification only on candidates of fixed component anchoring abnormalities; calculating the rotation phase of the fixed component by sampling based on the diameter of the fixed component and the belt running speed; extracting the high-frequency envelope value within the monitoring space range of the fixed component from the high-frequency envelope map, and averaging the high-frequency envelope value at the effective optical cable distance position to obtain the fixed component envelope value; dividing a complete rotation cycle into phase intervals, and distributing the fixed component envelope value to the corresponding phase interval according to the fixed component rotation phase to obtain the phase envelope distribution value within each effective complete rotation cycle; calculating the rotation phase reproduction coefficient based on the phase envelope distribution similarity between several effective complete rotation cycles and summarizing the phase envelope distribution similarity; when the rotation phase reproduction coefficient is not less than the rotation phase reproduction threshold, the corresponding fixed component is marked as a mechanical noise reproduction object.

[0012] As a preferred embodiment of the belt noise early warning method based on pattern recognition described in this invention, the step of performing nonlinear gating coupling on the mechanically reproduced noise object to obtain the mechanical noise confidence level, updating the health accumulation value based on the mechanical noise confidence level, and generating a belt noise early warning record includes: converting the fixed component anchoring coefficient into a fixed component anchoring gating factor, converting the rotation phase reproduction coefficient into a phase reproduction gating factor, converting the fixed component explanatory power into a fixed component explanatory strength factor, calculating the fixed component dominance factor based on the fixed component explanatory power and the belt movement explanatory power, determining the fixed component explanatory power growth based on the difference between the current time period's average fixed component explanatory power and the historical adjacent time period's average fixed component explanatory power, and calculating the increase based on the fixed component explanatory power growth. The system employs a long-gated control factor. It nonlinearly couples the anchoring gating factor and the phase reproduction gating factor of the fixed component to obtain the anchoring phase coupling factor. It then multiplies and couples the anchoring phase coupling factor, the fixed component interpretation strength factor, and the fixed component dominance factor, adjusting the product coupling result through an increasing gating factor to obtain the mechanical noise confidence level. Based on the mechanical noise confidence level and the health accumulation value of the previous analysis period, it updates the health accumulation value of the current analysis period. According to the health accumulation value, the fixed component is divided into healthy state, early concern stage, slight deterioration stage, continuous deterioration stage, and fault alarm stage. When the fixed component's health stage within the analysis period is the early concern stage, slight deterioration stage, continuous deterioration stage, or fault alarm stage, a belt noise early warning record is generated.

[0013] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements any step of the belt noise warning method based on pattern recognition as described in the first aspect of the present invention.

[0014] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements any step of the belt noise warning method based on pattern recognition as described in the first aspect of the present invention.

[0015] The beneficial effects of this invention are as follows: by calculating the interpretation degree of the fixed component and the interpretation degree of the accompanying movement and forming the anchoring coefficient of the fixed component, the abnormal sound vibration can be distinguished between the coordinates of the fixed component and the trajectory of the accompanying movement, so that the candidate objects of the fixed component anchoring anomaly have a clear spatial belonging; by combining the rotation phase reproduction coefficient and performing nonlinear gated coupling on the mechanical reproduction abnormal noise object, the confidence degree of the mechanical abnormal noise is obtained and the health accumulation value is updated, so that the belt abnormal noise early warning record simultaneously reflects the spatial anchoring, mechanical reproduction and deterioration accumulation process, thereby improving the pertinence and continuity of the belt abnormal noise early warning results. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of a belt conveyor noise early warning method based on pattern recognition.

[0018] Figure 2 The flowchart for generating the normalized acoustic vibration energy map and high-frequency envelope map.

[0019] Figure 3 A flowchart for screening candidate objects for anchorage anomalies of fixed components.

[0020] Figure 4 A flowchart for generating belt noise early warning records.

[0021] Figure 5 An anchoring identification response curve for a fixed component abnormality source scenario.

[0022] Figure 6 The response curve is used to explain the trajectory of a moving sound source. Detailed Implementation

[0023] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0024] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0025] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0026] Reference Figures 1-6 As one embodiment of the present invention, this embodiment provides a belt noise early warning method based on pattern recognition, including the following steps: S1. Lay out acoustic and vibration sensing optical cables along the frame structure corresponding to the fixed components of the belt conveyor, collect acoustic and vibration data of the belt conveyor, use idlers and drums as fixed components for spatial calibration, record the center position of the fixed components, the monitoring space range of the fixed components, the diameter of the fixed components, the correspondence between the belt running direction and the optical cable distance direction, and the belt running speed, and generate a normalized acoustic and vibration energy map and a high-frequency envelope map.

[0027] It should be noted that in this embodiment, belt conveyor refers to a belt conveyor, and belt refers to the conveyor belt in a belt conveyor used to carry and transport materials.

[0028] Specifically, step S1 includes the following: S1.1 Lay out the acoustic and vibration sensing optical cable and establish the optical cable distance coordinates, and number the fixed components of the idler roller and drum.

[0029] Furthermore, after the belt conveyor is installed, an acoustic vibration sensing optical cable is laid along the frame structure corresponding to the fixed components of the belt conveyor; the acoustic vibration sensing optical cable is connected to a distributed acoustic vibration acquisition device, which uses the starting end of the acoustic vibration sensing optical cable as the zero point of distance and establishes the optical cable distance coordinate along the length of the acoustic vibration sensing optical cable; the distributed acoustic vibration acquisition device continuously collects the acoustic vibration data of the belt conveyor during its operation.

[0030] Among them, the optical cable distance coordinate is used to represent the distance position of different acoustic vibration sampling positions on the acoustic vibration sensing optical cable; the optical cable distance direction refers to the positive direction of the optical cable distance coordinate, which is the direction of increasing along the length of the acoustic vibration sensing optical cable from the starting end of the optical cable.

[0031] It should be noted that the fixed component refers to the idler roller or drum whose installation position is fixed relative to the frame structure of the belt conveyor, and the fixed component itself can rotate with the belt.

[0032] It should be noted that the acoustic vibration sensing optical cable is preferably laid on the channel steel on both sides of the belt conveyor, the idler bracket, the drum support seat, or the frame structure rigidly connected to the idler bracket.

[0033] It should be noted that the acoustic vibration sensing optical cable is fixed to the frame structure by clamps, so that the acoustic vibration disturbances generated by the friction of the rollers, drums, and belts and local impacts can be transmitted to the acoustic vibration sensing optical cable through the frame structure.

[0034] It should be noted that a distributed acoustic vibration acquisition device refers to an acquisition device that sends a probe light signal to an acoustic vibration sensing optical cable, receives the light response signal returned along the optical cable, and demodulates the light response signal into acoustic vibration data arranged according to the distance and sampling time of the optical cable.

[0035] Furthermore, the idlers and rollers on the belt conveyor are uniformly numbered, with each idler or roller serving as a fixed component. Each fixed component corresponds to a fixed component number, and the type of each fixed component is recorded. The fixed component number, fixed component type, and belt segment information are then written into the fixed component basic data table.

[0036] The fixed component number is used for subsequent association with acoustic and vibration data, early warning records, and maintenance records.

[0037] It should be noted that for idlers, record the belt segment where the idler is located, the location of the idler group, and the idler number; for rollers, record the belt segment where the roller is located, the roller type, and the roller number.

[0038] S1.2. Calibrate the center position of the fixed component, determine the monitoring space range of the fixed component, record the diameter of the fixed component, record the correspondence between the belt running direction and the optical cable distance direction, synchronously collect the belt running speed, and form a basic data table of the fixed component.

[0039] Furthermore, after the acoustic vibration sensing optical cable is laid, the corresponding position of each fixed component in the optical cable distance coordinate is calibrated. For the current fixed component, the center distance position of the current fixed component in the optical cable distance coordinate is recorded as the center position of the fixed component.

[0040] It should be noted that the center position of the fixed component is obtained through construction surveying, optical cable laying path recording, on-site point calibration, or acoustic and vibration excitation calibration. For example, during the commissioning phase, an identifiable slight structural excitation is applied near the current fixed component, and the optical cable distance position corresponding to the current fixed component is determined based on the peak position returned by the distributed acoustic and vibration acquisition equipment, which is then used as the center position of the fixed component.

[0041] Furthermore, based on the center position of the fixed component and the installation spacing between adjacent fixed components, the monitoring space range corresponding to each fixed component is calculated, and the calculated monitoring space range of the fixed component is written into the basic data table of the fixed component.

[0042] It should be noted that when the first When the fixed component is an idler with adjacent idlers on both sides, the first... The spatial half-width of a fixed component is represented as: ; in, Indicates the first Half the width of the space of a fixed component Indicates the first The center position of each fixed component Indicates the first The center position of the adjacent idler roller before the fixed component Indicates the first The center position of the next adjacent idler roller after the fixed component.

[0043] It should be noted that when the first When the fixed component is a roller located at the end, the first... Half the distance between the center of the fixed component and the center of the only adjacent idler roller is used as the first... The space half-width of a fixed component.

[0044] It should be noted that when the first When the first fixed component is a roller, half of the actual coverage area of ​​the roller support, roller bearing housing, or roller installation area in the optical cable distance direction shall be taken as the first fixed component. The space half-width of a fixed component.

[0045] It should be noted that the monitoring space range for each fixed component is calculated and expressed as follows: ; in, Indicates the first The monitoring space range of each fixed component.

[0046] Furthermore, record the diameter of the fixing component. Specifically, when the fixing component is a roller, record the roller diameter; when the fixing component is a drum, record the drum diameter.

[0047] It should be noted that the diameter of the fixed component is used to calculate the rotation phase of the fixed component based on the belt running speed.

[0048] It should be noted that when the current fixed component does not participate in the rotation phase reproduction identification, the current fixed component is marked as not participating in the rotation phase reproduction calculation in the fixed component basic data table.

[0049] Furthermore, for the current location of the fixed component, record the correspondence between the belt running direction and the optical cable distance direction. When the belt running direction is consistent with the positive direction of the optical cable distance coordinate, record the correspondence between the belt running direction and the optical cable distance direction at the current fixed component as 1; when the belt running direction is opposite to the positive direction of the optical cable distance coordinate, record the correspondence between the belt running direction and the optical cable distance direction at the current fixed component as -1.

[0050] The correspondence between the belt running direction at the fixed component and the optical cable distance direction is used to subsequently trace the position of the belt movement in the normalized acoustic and vibration energy map.

[0051] It should be noted that when the abnormal sound source moves with the belt, the position of the corresponding abnormal sound source in the historical analysis period is traced back upstream according to the correspondence between the belt running speed and the belt running direction and the optical cable distance direction.

[0052] Furthermore, the belt speed is simultaneously acquired during the operation of the belt conveyor.

[0053] It should be noted that the belt running speed is output through any of the following methods: belt speed measuring device, roller speed conversion result, or belt conveyor control system.

[0054] Among them, the belt speed measuring device refers to the speed measuring device installed on the belt conveyor and directly outputs the belt linear speed; the drum speed conversion result refers to the belt linear speed calculated based on the drum angular velocity and drum diameter; the belt conveyor control system refers to the field control system used to control the start and stop of the belt conveyor, speed setting, and operation status acquisition.

[0055] It should be noted that when the belt running speed is output at fixed time intervals, the speed values ​​falling within the same analysis period are averaged, and the averaged speed value is used as the average belt running speed for the corresponding analysis period.

[0056] Furthermore, the fixed component number, fixed component type, fixed component center position, fixed component monitoring space range, fixed component diameter, the correspondence between the belt running direction and the optical cable distance direction, the belt running speed data source, and the belt segment to which the fixed component belongs are written into the fixed component basic data table.

[0057] S1.3 Divide the belt conveyor sound and vibration data into analysis periods, extract the energy of the abnormal noise sensitive frequency band and establish a normal sound and vibration energy benchmark, combine the abnormal noise sensitive frequency band energy and the normal sound and vibration energy benchmark to generate a normalized sound and vibration energy map, and generate a high-frequency envelope map through high-frequency filtering and envelope extraction.

[0058] Furthermore, after the distributed acoustic and vibration acquisition equipment continuously acquires the acoustic and vibration data of the belt conveyor, it divides the data into several analysis cycles according to a fixed time window.

[0059] It should be noted that adjacent analysis cycles slide according to the analysis cycle update interval. The analysis cycle update interval is determined based on the data refresh cycle of the distributed acoustic and vibration acquisition equipment, the belt running speed, the center distance between adjacent fixed components, and the requirement for continuous tracking of acoustic and vibration anomalies. Furthermore, the analysis cycle update interval is not greater than the length of the time window, so that adjacent analysis cycles are continuously connected in the sampling time and form overlapping analysis.

[0060] It should be noted that when the belt speed increases or the center-to-center distance between adjacent fixed components decreases, the analysis cycle update interval is set according to the principle that the belt movement distance between adjacent analysis cycles does not exceed half of the center-to-center distance between adjacent fixed components; when the data refresh cycle of the distributed acoustic and vibration acquisition device is greater than the analysis cycle update interval determined according to the belt movement distance, the data refresh cycle of the distributed acoustic and vibration acquisition device is used as the analysis cycle update interval.

[0061] It should be noted that the analysis cycle update interval is usually between 0.1 seconds and 5 seconds. The example uses 1 second. The reason for using 1 second is that 1 second can generate continuous acoustic and vibration anomaly tracking results under normal belt running speed, and avoids the increase in computation caused by excessively frequent analysis cycle updates.

[0062] It should be noted that for each analysis period, the energy of the corresponding abnormal noise sensitive frequency band is calculated and the high-frequency envelope value is extracted.

[0063] Furthermore, spectral analysis is performed on the vibration data of the belt conveyor during the analysis period to obtain the energy distribution at different frequencies at the optical cable distance position during the analysis period, and the noise-sensitive frequency band is selected; the energy of the optical cable distance position located in the noise-sensitive frequency band during the analysis period is integrated to obtain the noise-sensitive frequency band energy.

[0064] It should be noted that the abnormal noise sensitive frequency band is determined based on idler roller squealing, idler roller bearing friction, local impact, roller abnormality, and on-site calibration results. Specifically, during the commissioning phase, normal operation samples and abnormal noise samples confirmed manually are collected respectively. The frequency axis is divided into several continuous frequency bands. The energy increment of the abnormal noise sample in each continuous frequency band relative to the normal operation sample is calculated. The frequency range where the energy increment is higher than the frequency band increment threshold and is continuous with the adjacent high increment frequency band is determined as the abnormal noise sensitive frequency band. When there are no abnormal noise samples confirmed manually, 1000 Hz to 5000 Hz is used as the initial abnormal noise sensitive frequency band.

[0065] It should be noted that the frequency band increment threshold is determined offline through normal operation samples and abnormal noise samples confirmed by manual adjustment during the debugging phase; the value range of the frequency band increment threshold is usually 3 dB to 10 dB.

[0066] Furthermore, during the normal and stable operation phase of the belt conveyor, acoustic and vibration data of the belt conveyor are collected to establish a normal acoustic and vibration energy benchmark. For each optical cable distance position, the median value of the energy in the abnormal noise sensitive frequency band in the normal and stable operation sample is calculated, and the energy discrete value of the current optical cable distance position in the normal and stable operation sample is calculated to obtain the normal energy median value and the normal energy discrete value of the optical cable distance position.

[0067] Among them, the median value of normal energy and the discrete value of normal energy are used to describe the normal acoustic and vibration energy reference of the current optical cable distance position under normal operating conditions.

[0068] It should be noted that the normal and stable operation phase refers to the operation phase in which the belt conveyor has no abnormal maintenance records for idlers, rollers, misalignment and friction alarms, and maintenance alarms, and the belt speed fluctuation rate does not exceed the speed fluctuation threshold within the analysis period.

[0069] It should be noted that the speed fluctuation threshold is determined based on the speed stability accuracy of the belt conveyor control system and the speed operation statistics during normal operation, and the value range is usually 1% to 5%.

[0070] It should be noted that the speed fluctuation rate is the proportion of the difference between the maximum and minimum belt speeds within the same analysis period to the average belt speed within the analysis period.

[0071] Furthermore, during real-time operation, the energy of the abnormal noise sensitive frequency band within the analysis period is compared with the normal acoustic and vibration energy benchmark at the same optical cable distance position, and a positive deviation normalization algorithm is used to obtain a normalized acoustic and vibration energy map.

[0072] It should be noted that the positive deviation normalization algorithm is expressed as: ; in, Indicates the distance and location of the optical cable. Indicates the analysis period number. Indicates the distance and location of the optical cable at the th position. Normalized acoustic vibration energy values ​​within each analysis period Indicates the distance and location of the optical cable at the th position. Energy in the sensitive frequency band of abnormal noise within each analysis period This represents the median normal energy value at the location of the optical cable. This represents the normal energy dispersion value at the distance of the optical cable. This represents the smallest discrete value.

[0073] It should be noted that the minimum discrete value is determined based on the offline calibration of the energy noise level of the distributed acoustic and vibration acquisition equipment under the condition of no external structural excitation. The range of the minimum discrete value is usually 1 to 3 times the energy discrete value of the sensitive frequency band of abnormal noise under the condition of no external structural excitation, and the minimum discrete value is greater than zero.

[0074] It should be noted that, through the positive deviation normalization algorithm, energy deviations that are not higher than the normal acoustic vibration energy benchmark are set to zero; for energy deviations that are higher than the normal acoustic vibration energy benchmark, the normalized acoustic vibration energy value is calculated.

[0075] It should be noted that the normalized acoustic energy values ​​of each optical cable distance position in each analysis period are arranged according to the optical cable distance position and the analysis period to form a normalized acoustic energy map.

[0076] Furthermore, high-frequency filtering and envelope extraction are performed on the belt conveyor acoustic vibration data to generate a high-frequency envelope map. Specifically, the belt conveyor acoustic vibration data that has not undergone high-frequency filtering and envelope extraction is bandpass filtered according to the high-frequency analysis frequency band to obtain high-frequency acoustic vibration data; envelope extraction is performed on the high-frequency acoustic vibration data to obtain the high-frequency envelope value of each optical cable distance position at each sampling time; the high-frequency envelope values ​​are arranged according to the optical cable distance position and sampling time to form a high-frequency envelope map.

[0077] It should be noted that envelope extraction is performed using any one of the following methods: Hilbert envelope, square detection low-pass envelope, and absolute value smoothing envelope.

[0078] It should be noted that the high-frequency analysis band is determined through offline calibration using manually confirmed idler bearing impact samples, drum local friction samples, idler eccentricity samples, or idler looseness samples. During offline calibration, the manually confirmed idler bearing impact samples, drum local friction samples, idler eccentricity samples, or idler looseness samples are used as manually confirmed mechanical noise samples. Within the effective frequency response range of the distributed acoustic and vibration acquisition equipment, the frequency axis is divided into several continuous frequency bands. The envelope energy statistics of the normal operation samples in each continuous frequency band are statistically analyzed, and the envelope energy statistics of the manually confirmed mechanical noise samples in the same continuous frequency band are also statistically analyzed. The envelope energy statistics of the normal operation samples are subtracted from the envelope energy statistics of the manually confirmed mechanical noise samples to obtain the envelope energy increment of the corresponding continuous frequency band.

[0079] It should be noted that when the envelope energy increment of a continuous frequency band is not less than the high-frequency envelope increment threshold, and it is connected to an adjacent continuous frequency band whose envelope energy increment is not less than the high-frequency envelope increment threshold, the connected continuous frequency bands are merged to form a high-frequency candidate frequency interval, and the high-frequency candidate frequency interval is determined as the high-frequency analysis frequency band.

[0080] It should be noted that the high-frequency envelope increment threshold is determined offline based on the envelope energy increment distribution between normal operation samples and manually confirmed mechanical noise samples. When using mean and standard deviation for tuning, the high-frequency envelope increment threshold is the mean of the envelope energy increment distribution plus 2 to 4 times the standard deviation. When using quantile value for tuning, the high-frequency envelope increment threshold is the 95th to 99th percentile of the envelope energy increment distribution. For example, the 97.5th percentile is used. The reason for using the 97.5th percentile is that it can exclude the high-frequency envelope fluctuations located at the tail end in the normal operation samples and retain the response capability to the high-frequency envelope enhancement frequency band in the mechanical noise samples of fixed components.

[0081] It should be noted that the candidate frequency range for the high-frequency analysis band is usually from 1000 Hz to 8000 Hz, and the upper limit of the high-frequency analysis band does not exceed half of the sampling frequency of the distributed acoustic and vibration acquisition equipment; when field calibration results are lacking, the high-frequency analysis band adopts the same band as the abnormal noise sensitive band.

[0082] The next step is S2.

[0083] S2. Based on the normalized acoustic and vibration energy map, calculate the interpretation degree of each fixed component. According to the correspondence between the belt running speed and the belt running direction and the optical cable distance direction, backtrack the normalized acoustic and vibration energy map to calculate the interpretation degree of belt movement. Combine the interpretation degree of fixed components and the interpretation degree of belt movement to calculate the anchoring coefficient of fixed components and screen the candidates for fixed component anchoring anomalies.

[0084] Specifically, step S2 includes the following: S2.1. Based on the normalized acoustic vibration energy map, determine the number of continuous analysis cycles participating in the competitive identification and the number of effective optical cable distance positions within the monitoring space of the fixed component, and calculate the fixed component interpretation degree for each fixed component.

[0085] Furthermore, the number of consecutive analysis cycles participating in competition identification is set, for the first... The first fixed component is read from the fixed component basic data table. The monitoring space range of the fixed component corresponding to the fixed component is determined. The number of effective optical cable distance locations within the monitoring space of each fixed component is calculated, and these locations are arranged in ascending order of optical cable distance coordinates. Based on the normalized acoustic and vibration energy map, a fixed component dwell interpretation algorithm is used to calculate the number of effective optical cable distance locations within the monitoring space of each fixed component. The normalized acoustic and vibration energy values ​​of each fixed component are averaged over the current analysis period and consecutive analysis periods going back to the historical direction to obtain the interpretation degree of the fixed component.

[0086] It should be noted that competitive identification refers to interpreting the same abnormal sound vibration segment using the normalized sound vibration energy distribution under the coordinates of a fixed component and the historical normalized sound vibration energy distribution under the trajectory of the moving component, respectively, within the same continuous analysis period. The difference between the interpretation degree of the fixed component and the interpretation degree of the moving component determines whether the abnormal sound vibration is more consistent with the sound source characteristics of the fixed component or the sound source characteristics of the moving component.

[0087] It should be noted that the number of consecutive analysis cycles participating in competitive identification is determined based on the belt running speed, the installation spacing of adjacent fixed components, and the analysis cycle update interval, ensuring that the number of adjacent fixed component spacings covered by the moving sound source within the consecutive analysis cycle is not less than 1. The value range of the number of consecutive analysis cycles participating in competitive identification is usually 3 to 20.

[0088] It should be noted that when there are invalid optical cable distance locations within the monitoring space of the fixed component, the invalid optical cable distance locations will be removed from the number of valid optical cable distance locations. Invalid optical cable distance locations include locations where the distributed acoustic and vibration acquisition equipment has not returned acoustic and vibration data, locations where the acoustic and vibration sensing optical cable is broken, and locations marked as unusable after debugging.

[0089] It should be noted that the algorithm for interpreting the dwell time of fixed components is expressed as follows: ; in, Indicates the first The fixing component is in the first The degree of explanation of fixed components within a single analysis cycle This indicates the number of consecutive analysis cycles involved in the competition identification process. Indicates the first The number of optical cable distance locations within the monitoring space of each fixed component that participate in the normalized acoustic and vibration energy summation. The sequence number indicating the effective optical cable distance location. Indicates the first Within the monitoring space range of the fixed component corresponding to the fixed component, the first One effective optical cable distance location, This indicates the analysis period number used to trace back from the current analysis period to historical periods. Indicates the first The first fixed component corresponds to the first The effective optical cable distance position is at the [number]th [location]. Normalized acoustic vibration energy values ​​within each analysis period.

[0090] It should be noted that the fixed component interpretation degree is used to characterize abnormal acoustic vibration in the first... The degree to which each fixed component appears continuously within the monitoring space.

[0091] S2.2. Based on the belt running speed, analysis cycle update interval, the correspondence between the belt running direction and the optical cable distance direction, and the monitoring space range of the fixed components, the normalized acoustic and vibration energy map is backtracked to the location of belt movement, the historical normalized acoustic and vibration energy value corresponding to the backtracked location of belt movement is read, and the explanatory power of belt movement is calculated.

[0092] Furthermore, the average belt speed within each analysis cycle is obtained, and the belt displacement accumulation algorithm is used to calculate the belt movement distance between the historical analysis cycle and the current analysis cycle, based on the analysis cycle update interval.

[0093] It should be noted that the algorithm for cumulative displacement is expressed as follows: ; in, Indicates from the first The analysis cycle to the 1st The travel distance between analysis cycles, This indicates the analysis period number involved in the velocity accumulation. Indicates the first The average belt speed within each analysis period This indicates the analysis cycle update interval between adjacent analysis cycles.

[0094] It should be noted that when When the distance moved is zero, it means that no historical position backtracking is performed within the current analysis period.

[0095] Furthermore, regarding the first The fixed component will be the first For each fixed component, the distance position of each optical cable within the monitoring space is determined. The accompanying position backtracking algorithm is used to backtrack the historical position according to the correspondence between the belt running direction and the optical cable distance direction and the accompanying movement distance, so as to obtain the accompanying movement backtracking position.

[0096] It should be noted that the accompanying position backtracking algorithm is expressed as: ; in, Indicates the first The fixing component is in the first Within the analysis period, for the first The effective optical cable distance position to the first The position of the backtracking along with the movement obtained from each analysis cycle. Indicates the first The correspondence between the belt running direction and the optical cable distance direction at each fixed component.

[0097] It should be noted that when the belt's running direction is consistent with the positive direction of the optical cable distance coordinate, The value is 1; when the belt running direction is opposite to the positive direction of the optical cable distance coordinate, The value is -1.

[0098] It should be noted that the follow-up position is used to represent the historical position of the follow-up sound source within the kr-th analysis period; when the belt running direction is consistent with the positive direction of the optical cable distance coordinate, the follow-up position is located on the side where the optical cable distance decreases along the optical cable distance coordinate; when the belt running direction is opposite to the positive direction of the optical cable distance coordinate, the follow-up position is located on the side where the optical cable distance increases along the optical cable distance coordinate.

[0099] It should be noted that when the follow-up position does not fall on the actual optical cable distance sampling point in the normalized acoustic energy map, the linear interpolation result of two adjacent optical cable distance sampling points is used as the normalized acoustic energy value corresponding to the follow-up position.

[0100] It should be noted that when the location of the follow-up movement exceeds the effective acquisition range of the acoustic vibration sensing optical cable, the corresponding follow-up item will be marked as an invalid follow-up item, and invalid follow-up items will not participate in the calculation of the follow-up movement interpretation degree.

[0101] Furthermore, based on the follow-up movement backtracking position, the normalized acoustic and vibration energy values ​​corresponding to the historical analysis period are read from the normalized acoustic and vibration energy map, and invalid backtracking items are excluded by valid backtracking indicator values. The follow-up trajectory interpretation algorithm is used to calculate the follow-up movement interpretation degree.

[0102] It should be noted that the effective backtracking position refers to the backtracking position that does not exceed the effective acquisition range of the acoustic vibration sensing optical cable. When the effective backtracking position does not fall on the actual optical cable distance sampling point, the linear interpolation result of two adjacent optical cable distance sampling points is used to participate in the calculation of the accompanying trajectory interpretation algorithm.

[0103] Furthermore, invalid backtracking entries are not included in the calculation of the explanatory power of the accompanying movement; for each backtracking position of the accompanying movement, a valid backtracking indicator value is set, and the number of valid backtracking entries is calculated based on the valid backtracking indicator value.

[0104] It should be noted that the effective backtracking indication value is set as follows: the position of the follow-up movement backtracking is compared with the effective acquisition range of the acoustic vibration sensing optical cable; when the follow-up movement backtracking position falls within the effective acquisition range of the acoustic vibration sensing optical cable, the effective backtracking indication value of the corresponding follow-up movement backtracking position is recorded as 1; when the follow-up movement backtracking position falls outside the effective acquisition range of the acoustic vibration sensing optical cable, the effective backtracking indication value of the corresponding follow-up movement backtracking position is recorded as 0.

[0105] It should be noted that the number of valid backtracking terms is calculated as follows: ; in, Indicates the first The fixing component is in the first The number of valid backtracking items actually used in the calculation of the explanatory power of the moving component within each analysis period. Indicates the first The fixing component is in the first Within the analysis period, for the first The effective optical cable distance position to the first Valid backtracking indicator value during each analysis cycle backtracking.

[0106] It should be noted that when the number of valid backtracking items is greater than zero, the accompanying trajectory interpretation algorithm is expressed as follows: ; in, Indicates the first The fixing component is in the first The explanatory power of the carryover movement within each analysis period, Indicates the position of the backtracking movement. In the The normalized acoustic vibration energy value corresponding to each analysis period.

[0107] It should be noted that the belt-movement interpretability is used to indicate whether the same anomalous acoustic vibration segment can be interpreted as an acoustic vibration response generated by an anomalous sound source moving with the belt passing near the current fixed component.

[0108] It should be noted that when the explanatory power of the accompanying movement is higher than that of the fixed component, it means that the abnormal sound vibration is more consistent with the historical movement trajectory of the accompanying sound source; when the explanatory power of the accompanying movement is lower than that of the fixed component, it means that the abnormal sound vibration is more consistent with the characteristics of a sound source fixed near the current fixed component.

[0109] It should be noted that when the first The fixing component is in the first If no valid backtracking item exists within the first analysis period, the explanatory power of the accompanying movement is not calculated, and the first analysis period is not included. One fixed component is selected as a candidate for fixed component anchorage anomaly.

[0110] S2.3. Combining the interpretation degree of the fixed component and the interpretation degree of the accompanying movement, calculate the degree of dominance of the coordinate interpretation of the fixed component relative to the coordinate interpretation of the accompanying movement, and calculate the anchorage coefficient of the fixed component based on the degree of dominance.

[0111] Furthermore, when the sum of the fixed component's explanatory power and the accompanying movement's explanatory power is greater than zero, the fixed component's anchorage discrimination algorithm is used to calculate the fixed component's anchorage coefficient.

[0112] It should be noted that the algorithm for determining the anchorage of fixed components is expressed as follows: ; in, Indicates the first The fixing component is in the first Anchorage coefficient of fixed components within each analysis cycle.

[0113] It should be noted that when the sum of the fixed component's explanation degree and the accompanying movement's explanation degree is not greater than zero, the anchorage coefficient of the fixed component is recorded as zero, and the first... One fixed component is selected as a candidate for fixed component anchorage anomaly.

[0114] It should be noted that the fixed component anchorage factor is used to indicate the degree of dominance of the interpretation of the same anomalous acoustic vibration segment by the fixed component coordinates relative to the interpretation by the accompanying moving coordinates; when the fixed component interpretation is greater than the accompanying moving interpretation, the fixed component anchorage factor is positive; when the fixed component interpretation is not greater than the accompanying moving interpretation, the fixed component anchorage factor is not greater than zero.

[0115] S2.4 Determine the fixed component interpretation threshold and fixed component anchoring threshold. Combine the fixed component interpretation degree, fixed component anchoring coefficient, fixed component interpretation threshold and fixed component anchoring threshold to screen for fixed component anchoring anomaly candidates.

[0116] Furthermore, during the normal and stable operation phase of the belt conveyor, based on the explanation degree and anchoring coefficient of each fixed component in the normal and stable operation sample, the explanation threshold and anchoring threshold of the fixed components are determined respectively. The fixing component is in the first When the candidate conditions for anchorage anomalies of fixed components are met within the analysis cycle, the first analysis cycle will be... Each fixed component is marked as a candidate for fixed component anchoring anomaly.

[0117] It should be noted that the candidate conditions for fixed component anchorage anomaly include a first candidate condition for fixed component anchorage anomaly and a second candidate condition for fixed component anchorage anomaly. When both the first and second candidate conditions for fixed component anchorage anomaly are met, it is determined that the candidate conditions for fixed component anchorage anomaly are met.

[0118] It should be noted that the first candidate condition for anchorage anomaly of fixed components is expressed as: ; in, Indicates the first The fixed component interpretation threshold corresponding to each fixed component.

[0119] It should be noted that the fixed component interpretation threshold is determined based on the upper quantile of the fixed component interpretation degree in the normal stable operation sample. The upper quantile of the fixed component interpretation threshold is usually between the 95th and 99.5th percentiles. An example of the upper quantile of the fixed component interpretation threshold is the 97.5th percentile. The reason for choosing the 97.5th percentile is that the 97.5th percentile can exclude the accidental energy fluctuations located at the tail in the normal stable operation sample, while retaining the response sensitivity to early fixed component anomalies.

[0120] It should be noted that the second candidate condition for anchorage anomaly of fixed components is expressed as follows: ; in, Indicates the first The anchoring threshold of a fixed component is corresponding to a fixed component.

[0121] It should be noted that the fixed component anchoring threshold is determined based on the upper quantile of the fixed component anchoring coefficient in the normal stable operation sample. The upper quantile of the fixed component anchoring threshold is usually between the 95th and 99.5th percentiles. For example, the upper quantile of the fixed component anchoring threshold is the 95th percentile. The reason for choosing the 95th percentile is that the fixed component anchoring coefficient is used to determine the dominance of the fixed component coordinate interpretation relative to the accompanying moving coordinate interpretation. Choosing the 95th percentile can reduce the random dominant segments entering the candidate object during normal operation, and at the same time avoid the early fixed component anchoring anomalies being excluded due to the threshold being too high.

[0122] It should be noted that fixed components that do not meet the first or second conditions for candidates of anchorage anomalies will not be included in the rotation phase reproduction identification.

[0123] In this embodiment, to verify the ability of the present invention to locate persistent abnormal acoustic vibrations near a fixed component, the trends of the fixed component's interpretability, accompanying movement interpretability, and anchoring coefficient with the analysis period number are recorded in the scenario of the fixed component's abnormal source. Figure 5As can be seen in the overview diagram, the interpretability of the fixed component gradually increases and remains at a high level as the analysis cycle progresses, indicating that the abnormal acoustic vibrations are continuously distributed within the monitoring space of the fixed component corresponding to the target fixed component. Although the interpretability of the accompanying movement changes over time, it is generally lower than that of the fixed component, indicating that the abnormal acoustic vibrations are more consistent with the dwell response under the coordinates of the fixed component. The anchoring coefficient of the fixed component remains positively stable, indicating that the interpretation of the fixed component coordinates is dominant over the interpretation of the accompanying movement trajectory. Through the local magnified diagram, the characteristic peaks of the interpretability of the fixed component and the characteristic peaks of the interpretability of the accompanying movement are separated, and the point with the greatest difference between the two curves corresponds to the analysis cycle in which the spatial attribution of the abnormal acoustic vibrations is clearest. This shows that the present invention can transform the abnormal acoustic vibrations that continuously occur near the fixed component into screening results with the attribution relationship of the fixed component.

[0124] In this embodiment, to verify the ability of the present invention to eliminate acoustic vibration sources from non-fixed components moving with the belt, the trends of the fixed component interpretation degree, the moving component interpretation degree, and the fixed component anchoring coefficient with the analysis period number are recorded in a scenario of moving sound sources. Figure 6 As can be seen in the overview diagram, when the moving sound source passes near the target fixed component, both the fixed component explanatory power and the moving sound source explanatory power exhibit short-term peaks, indicating that the moving sound source can form a local acoustic vibration response when passing through the monitoring space of the fixed component. However, the fixed component anchoring coefficient does not continuously meet the fixed component anchoring threshold within the continuous analysis period, and it does not form a candidate condition for fixed component anchoring anomalies together with the fixed component explanatory power, indicating that the abnormal acoustic vibration has not formed an anchoring feature that is long-term dominated by the fixed component coordinates. Through the local magnified diagram, the peak position of the moving sound source explanatory power characteristic is more prominent than that of the fixed component explanatory power characteristic peak position, and the point of maximum difference between the two curves can reflect the degree of adaptation of the moving sound source trajectory to the abnormal acoustic vibration. This shows that the present invention can interpret the moving sound source by tracing back the moving sound source position, thereby reducing the probability of the moving sound source being misattributed to the fixed component.

[0125] The next step is step S3.

[0126] S3. For candidates with abnormal anchoring of fixed components only, determine the rotation phase of the fixed component based on the diameter of the fixed component and the belt running speed, and calculate the rotation phase reproduction coefficient in combination with the high-frequency envelope diagram to screen the mechanically reproduced abnormal noise objects.

[0127] Specifically, step S3 includes the following: S3.1 Only perform rotation phase reproduction identification on candidates of anchorage anomalies of fixed components. Determine the rotation phase of the fixed component based on the diameter of the fixed component and the belt running speed, and extract the envelope value of the fixed component from the high-frequency envelope map.

[0128] Furthermore, the fixed component number corresponding to the fixed component anchoring anomaly candidate object is read. For fixed components that are not marked as fixed component anchoring anomaly candidate objects, rotation phase reproduction identification is not performed.

[0129] It should be noted that when the current fixed component is marked as not participating in the rotation phase reproduction calculation in the fixed component basic data table, the rotation phase reproduction identification is not performed on the current fixed component, and the current fixed component is not selected as a mechanical reproduction abnormal noise object.

[0130] Furthermore, for the i-th fixed component anchoring anomaly candidate, the diameter of the i-th fixed component is read from the fixed component basic data table, and the belt running speed corresponding to the current analysis cycle is read; using the current analysis cycle as the basis for phase analysis, the minimum number of complete rotation cycles required for rotation phase reproduction identification is set. When the number of complete rotation cycles formed in the current analysis cycle is less than the minimum number of complete rotation cycles, adjacent analysis cycles are merged in the historical direction according to the analysis cycle number to form a phase analysis time range, until the number of complete rotation cycles formed within the phase analysis time range is not less than the minimum number of complete rotation cycles.

[0131] It should be noted that the minimum number of complete rotation cycles is adjusted based on the requirements for comparing the periodic similarity of the rotation phase recurrence coefficient, the on-site noise level, and the stability of the rotation speed of the fixed component. The minimum number of complete rotation cycles is usually in the range of 2 to 5. In this example, we take 2. The reason for taking 2 is that two complete rotation cycles can form at least one similarity comparison between the phase envelope distributions, so that the rotation phase recurrence coefficient can be judged as reproducible.

[0132] It should be noted that when the number of merged analysis cycles reaches the maximum number of merged cycles, and the number of complete rotation cycles formed is still less than the minimum number of complete rotation cycles, the rotation phase reproduction coefficient of the current fixed component in the current analysis cycle will not be calculated, and the current fixed component will not be screened as a mechanically reproduced abnormal noise object.

[0133] It should be noted that the maximum number of merges is determined based on the time required for the fixed component to complete two revolutions under low-speed operation of the belt conveyor and the analysis cycle update interval. The maximum number of merges is usually in the range of 3 to 30.

[0134] Furthermore, within the phase analysis time range, the initial sampling time of the phase analysis time range is taken as the initial time, and the rotation phase of the fixed component at the initial time is recorded as zero. The rotation phase of the fixed component is calculated by sampling according to the diameter of the fixed component and the running speed of the belt.

[0135] It should be noted that the rotation phase of the fixed component is calculated as follows: ; in, Indicates the first A fixed component during sampling time The rotation phase of the fixed component below, Indicates the sampling time. Indicates the time interval between adjacent sampling times. Indicates sampling time The belt running speed below, Indicates the first The diameter of the fixing component of each fixing component. This indicates the modulo operation.

[0136] It should be noted that the rotation phase of the fixed component is used to characterize the rotational angular position of the fixed component within the phase analysis time range; the rotation phase of the fixed component reaches [a certain value] every [period]. Then start accumulating from scratch again.

[0137] Furthermore, extract the first from the high-frequency envelope graph. The high-frequency envelope value of the fixed component within the monitoring space range corresponding to the fixed component is measured, and the value of the fixed component is measured for the first fixed component. The high-frequency envelope value of the fixed component is obtained by averaging the effective optical cable distance position corresponding to each fixed component.

[0138] It should be noted that, regarding the first The high-frequency envelope values ​​of the effective optical cable distance positions corresponding to each fixed component are averaged and expressed as follows: ; in, Indicates the first A fixed component during sampling time The envelope value of the fixed component below, Indicates the first The first fixed component corresponds to the first The effective optical cable distance location at the sampling time The high-frequency envelope value below.

[0139] It should be noted that the envelope value of the fixed component is used to characterize the first... The intensity of high-frequency acoustic vibration components within the monitoring space of each fixed component as a function of sampling time.

[0140] S3.2 Divide a complete rotation cycle into several phase intervals, and distribute the envelope value of the fixed component to the corresponding phase interval according to the rotation phase of the fixed component, forming the phase envelope distribution value within each effective complete rotation cycle.

[0141] Furthermore, the number of rotation phase intervals is set, and a complete rotation cycle is divided into equal-width phase intervals.

[0142] It should be noted that dividing a complete rotation cycle into equal-width phase intervals is represented as follows: ; in, Indicates the first One phase interval, Indicates the phase interval number, Indicates the number of rotation phase intervals.

[0143] It should be noted that the number of rotation phase intervals is determined based on the resolution requirements of the high-frequency envelope change within one revolution of the fixed component. The value range of the number of rotation phase intervals is usually from 12 to 72, and the example number of rotation phase intervals is 24.

[0144] Furthermore, the sampling time within the phase analysis time range is divided into several complete rotation cycles according to the rotation phase of the fixed component, and the complete rotation cycles with a total envelope energy of zero are removed to obtain the effective complete rotation cycles for participating in the rotation phase reproduction calculation.

[0145] It should be noted that when the number of effective complete rotation cycles is less than the number of minimum complete rotation cycles, the rotation phase reproduction coefficient of the current fixed component in the current analysis cycle is not calculated, and the current fixed component is not selected as a mechanically reproduced abnormal noise object.

[0146] Furthermore, for each effective complete rotation cycle, the envelope value of the fixed component is accumulated according to the rotation phase interval, and the ratio of the accumulated envelope value in each rotation phase interval to the sum of the envelope values ​​in the corresponding effective complete rotation cycle is calculated to obtain the phase envelope distribution value.

[0147] It should be noted that the phase envelope distribution algorithm is expressed as: ; in, Indicates the first The fixing component is in the first Within the first analysis period In the first effective complete rotation cycle Phase envelope distribution values ​​for each phase interval. Indicates the number of a valid complete rotation cycle. Indicates the first The fixing component is in the first Within the first analysis period The set of sampling times corresponding to each valid complete rotation cycle.

[0148] It should be noted that the phase envelope distribution value is used to represent the proportion of high-frequency envelope energy carried by different rotation phase intervals within a single complete rotation cycle of the fixed component.

[0149] S3.3 Calculate the rotation phase reproduction coefficient based on the phase envelope distribution similarity between several effective complete rotation cycles, and screen out the objects of mechanical reproduction abnormal noise.

[0150] Furthermore, when the number of effective complete rotation cycles is not less than the number of minimum complete rotation cycles, the rotation phase reproduction algorithm is used to calculate the rotation phase reproduction coefficient.

[0151] It should be noted that the rotation phase reproduction algorithm is expressed as follows: ; ; in, Indicates the first The fixing component is in the first Normalized coefficients for the recurrence of rotating phases within each analysis period Indicates the first The fixing component is in the first The number of valid complete rotation cycles involved in the rotation phase reproduction calculation within each analysis cycle. Indicates the first The fixing component is in the first Rotational phase reproduction coefficient within each analysis period Indicates the number of the first valid complete rotation cycle to be compared. Indicates the number of the second valid complete rotation cycle to be compared.

[0152] It should be noted that the rotation phase reproduction coefficient is used to characterize the consistency of the phase intervals in which the high-frequency envelope values ​​of a fixed component appear in several effective complete rotation cycles; the closer the phase envelope distributions of several effective complete rotation cycles are, the higher the rotation phase reproduction coefficient is.

[0153] Furthermore, during the normal and stable operation phase of the belt conveyor, the rotational phase reproduction threshold is determined based on the rotational phase reproduction coefficient of each fixed component in the normal and stable operation sample. The fixing component is in the first When the mechanical reproduction abnormal noise screening conditions are met within the first analysis cycle, the first... A fixed component is marked as an object for mechanically reproducing abnormal noise.

[0154] It should be noted that the screening criteria for mechanically reproducing abnormal noises are expressed as follows: ; in, Indicates the first The threshold for reproducing the rotational phase corresponding to a fixed component.

[0155] It should be noted that the rotation phase reproduction threshold is determined based on the upper quantile of the rotation phase reproduction coefficient in the normal stable operation sample. The upper quantile of the rotation phase reproduction threshold is usually between the 95th and 99.5th percentiles. For example, the upper quantile of the rotation phase reproduction threshold is taken as the 97.5th percentile. The reason for taking the 97.5th percentile is that the 97.5th percentile can exclude the accidental phase concentration phenomenon in the normal stable operation sample and retain the response capability to early mechanical reproduction abnormal noise of fixed components.

[0156] It should be noted that anchorage anomalies of fixed components that do not meet the screening criteria for mechanical noise reproduction will not be included in the mechanical noise confidence calculation.

[0157] Finally, there is step S4.

[0158] S4. Perform nonlinear gating coupling on the mechanically reproduced abnormal noise object to obtain the mechanical abnormal noise confidence level, update the health accumulation value according to the mechanical abnormal noise confidence level, and generate belt abnormal noise early warning record.

[0159] Specifically, step S4 includes the following: S4.1 Based on the anchoring coefficient of the fixed component, the rotation phase reproduction coefficient, the explanation degree of the fixed component, the explanation degree of the accompanying movement, and the growth of the explanation degree of the fixed component, construct the anchoring gate factor, the phase reproduction gate factor, the explanation strength factor of the fixed component, the dominance factor of the fixed component, and the growth gate factor.

[0160] Furthermore, the fixed component number corresponding to the mechanically reproduced abnormal noise object is read. For fixed components not marked as mechanically reproduced abnormal noise objects, no mechanical abnormal noise confidence is calculated, and only the health accumulation value of the corresponding fixed component is updated according to the health accumulation decay algorithm. For mechanically reproduced abnormal noise objects, the fixed component anchoring coefficient, rotation phase reproduction coefficient, fixed component explanatory power and accompanying movement explanatory power of the fixed component in the current analysis period are read, and the fixed component explanation threshold, fixed component anchoring threshold and rotation phase reproduction threshold corresponding to the fixed component are read.

[0161] Furthermore, a fixed component anchoring gating algorithm is adopted to convert the fixed component anchoring coefficient into a fixed component anchoring gating factor.

[0162] It should be noted that the fixed component anchoring gating algorithm is expressed as follows: ; in, Indicates the first The fixing component is in the first Fixed component anchoring gating factor within each analysis cycle.

[0163] It should be noted that the fixed component anchoring gating factor is used to convert the degree to which the fixed component anchoring coefficient exceeds the fixed component anchoring threshold into a gating value between 0 and 1.

[0164] It should be noted that, under the condition that the anchoring threshold of the fixed component is equal to 1, if the anchoring coefficient of the fixed component is not less than the anchoring threshold of the fixed component, the anchoring gate factor of the fixed component is recorded as 1; if the anchoring coefficient of the fixed component is less than the anchoring threshold of the fixed component, the anchoring gate factor of the fixed component is recorded as 0.

[0165] Furthermore, a phase reproduction gating algorithm is adopted to convert the rotation phase reproduction coefficients into phase reproduction gating factors.

[0166] It should be noted that the phase reconstruction gating algorithm is expressed as: ; in, Indicates the first The fixing component is in the first Phase reproduction gating factor within each analysis cycle.

[0167] It should be noted that the phase reproduction gating factor is used to convert the degree to which the rotation phase reproduction coefficient exceeds the rotation phase reproduction threshold into a gating value between 0 and 1.

[0168] It should be noted that, under the condition that the rotation phase reproduction threshold is equal to 1, if the rotation phase reproduction coefficient is not less than the rotation phase reproduction threshold, the phase reproduction gating factor is recorded as 1; if the rotation phase reproduction coefficient is less than the rotation phase reproduction threshold, the phase reproduction gating factor is recorded as 0.

[0169] Furthermore, a fixed member interpretation strength algorithm is adopted to convert the fixed member interpretation degree into a fixed member interpretation strength factor.

[0170] It should be noted that the strength calculation algorithm for fixed components is expressed as follows: ; in, Indicates the first The fixing component is in the first Explanation of strength factors for fixed components within each analysis cycle.

[0171] It should be noted that the fixed component interpretation strength factor is used to characterize the degree of enhancement after the fixed component interpretation degree exceeds the fixed component interpretation threshold; when the fixed component interpretation degree does not exceed the fixed component interpretation threshold, the fixed component interpretation strength factor is 0.

[0172] Furthermore, when the sum of the fixed component interpretation degree and the accompanying movement interpretation degree is greater than zero, the fixed component dominance algorithm is used to calculate the fixed component dominance factor.

[0173] It should be noted that the fixed component dominance algorithm is expressed as follows: ; in, Indicates the first The fixing component is in the first Dominance factor of fixed components within each analysis period.

[0174] It should be noted that when the sum of the explanation degree of the fixed component and the explanation degree of the accompanying movement is not greater than zero, the dominance factor of the fixed component is recorded as 0.

[0175] Furthermore, a fixed component explanation growth algorithm is adopted to calculate the increase in the explanation degree of fixed components, and the increase in the explanation degree of fixed components is converted into a growth gating factor.

[0176] It should be noted that the increase in the explanatory power of fixed components is determined by the difference between the average explanatory power of fixed components in the current time period and the average explanatory power of fixed components in adjacent historical time periods.

[0177] It should be noted that the fixed component interpretation growth algorithm is expressed as follows: ; in, Indicates the first The fixing component is in the first Increase in the explanatory power of fixed components within an analysis period. This indicates the number of analysis periods used for comparing the current time period with adjacent historical time periods.

[0178] It should be noted that when the number of historical analysis periods is less than At that time, the increase in the explanatory power of fixed components is not calculated, and the confidence level of mechanical noise of fixed components is not calculated in the current analysis period. Instead, the health cumulative decay algorithm is used to update the health cumulative value of the corresponding fixed components.

[0179] It should be noted that the number of comparative analysis cycles is determined based on the stability of the belt running speed and the length of observation time for the deterioration process of the fixed components. The value range of the number of comparative analysis cycles is usually 3 to 20.

[0180] It should be noted that the increase in the explanatory power of fixed components is converted into a growth gating factor, expressed as: ; in, Indicates the first The fixing component is in the first Growth gating factor within an analysis period.

[0181] It should be noted that the growth gating factor is used to characterize the degree to which the explanatory power of a fixed component increases from adjacent historical time periods to the current time period; when the explanatory power of a fixed component does not show an increase, the growth gating factor is 0.

[0182] S4.2. Nonlinearly couple the anchoring gating factor and the phase reproduction gating factor of the fixed component to obtain the anchoring phase coupling factor.

[0183] Furthermore, when the sum of the anchoring gating factor and the phase reproduction gating factor of the fixed component is greater than zero, the anchoring phase coupling algorithm is used to calculate the anchoring phase coupling factor.

[0184] It should be noted that the anchored phase coupling algorithm is expressed as: ; in, Indicates the first The fixing component is in the first Anchoring phase coupling factor within each analysis cycle.

[0185] It should be noted that when the sum of the anchoring gating factor and the phase reproduction gating factor of the fixed component is not greater than zero, the anchoring phase coupling factor is recorded as 0.

[0186] It should be noted that the anchoring phase coupling factor is formed through harmonic coupling. When the anchoring gate factor or the phase reproduction gate factor of the fixed component is 0, the anchoring phase coupling factor is 0. When both the anchoring gate factor of the fixed component and the phase reproduction gate factor are greater than 0, the anchoring phase coupling factor is constrained by the value of the smaller of the two factors and decreases as the value of the smaller factor decreases.

[0187] S4.3. Based on the anchoring phase coupling factor, the fixed component interpretation strength factor, the fixed component dominance factor, and the growth gating factor, a nonlinear gating coupling algorithm is used to obtain the confidence level of mechanical abnormal noise.

[0188] Furthermore, based on the anchoring phase coupling factor, the fixed component interpretation strength factor, the fixed component dominance factor, and the growth gating factor, a nonlinear gating coupling algorithm is used to calculate the confidence level of the mechanical noise of the fixed component within the analysis period for the mechanically reproduced abnormal noise object.

[0189] It should be noted that the nonlinear gated coupling algorithm is expressed as: ; in, Indicates the first The fixing component is in the first Confidence level of mechanical abnormal noise within each analysis period.

[0190] It should be noted that the nonlinear gated coupling algorithm performs product coupling on the anchoring phase coupling factor, the fixed component explained strength factor, the fixed component dominance factor, and the growth gate factor; when any one of the anchoring phase coupling factor, the fixed component explained strength factor, and the fixed component dominance factor is 0, the confidence level of the mechanical abnormal noise is 0; the growth gate factor is obtained through... Adjusting the confidence level of mechanical noise, when the explanatory power of the fixed component does not show an increase, the confidence level of mechanical noise retains half of the product of the anchoring phase coupling factor, the explanatory strength factor of the fixed component, and the dominance factor of the fixed component. When the explanatory power of the fixed component shows an increase, the confidence level of mechanical noise is improved relative to the state without increase.

[0191] S4.4 Update the cumulative health value based on the confidence level of the mechanical noise, divide the health stage according to the cumulative health value, and generate belt noise early warning records according to the health stage.

[0192] Furthermore, regarding the first For each mechanically reproduced abnormal noise object, the health cumulative value for the current analysis period is updated using a health cumulative update algorithm based on the mechanical abnormal noise confidence level and the health cumulative value of the previous analysis period.

[0193] It should be noted that the health cumulative update algorithm is expressed as: ; in, Indicates the first The fixing component is in the first Cumulative health values ​​within each analysis period Indicates the first The fixing component is in the first Cumulative health values ​​within each analysis period This represents the historical retention coefficient.

[0194] It should be noted that the health cumulative update algorithm is used to jointly update the health cumulative value of the previous analysis period and the mechanical noise confidence level of the current analysis period; the larger the historical retention coefficient, the stronger the retention effect of the health cumulative value of the previous analysis period on the current health cumulative value; the larger the mechanical noise confidence level, the greater the extent to which the current health cumulative value is updated in the direction of high risk.

[0195] It should be noted that the historical retention coefficient is adjusted based on the belt conveyor maintenance cycle, the expected advance of early warning, and the permissible false alarm rate on site. The value range of the historical retention coefficient is usually 0.7 to 0.98.

[0196] It should be noted that when the first The fixing component is in the first If an object is not marked as a mechanically reproduced abnormal noise object within a certain analysis period, the health accumulation value is updated using the health accumulation decay algorithm.

[0197] It should be noted that the cumulative health decay algorithm is expressed as: ; It should be noted that when the first If a fixed component does not have a health cumulative value from the previous analysis period, the health cumulative value from the previous analysis period will be recorded as 0.

[0198] Furthermore, based on the samples of normal and stable operation and the samples of manual inspection and confirmation, the health stage threshold is determined, and the health stage is divided according to the health stage threshold. Specifically, according to the health stage threshold, the fixed components are divided into the healthy state, the early attention stage, the slight deterioration stage, the continuous deterioration stage, and the fault alarm stage.

[0199] It should be noted that the health stage thresholds include the early attention stage threshold, the minor degradation stage threshold, the continuous degradation stage threshold, and the fault alarm stage threshold, and satisfy the following conditions: the early attention stage threshold is less than the minor degradation stage threshold, less than the continuous degradation stage threshold, and less than the fault alarm stage threshold; when the cumulative health value is less than the early attention stage threshold, the fixed component is classified as healthy; when the cumulative health value is not less than the early attention stage threshold and less than the minor degradation stage threshold, the fixed component is classified as early attention stage; when the cumulative health value is not less than the minor degradation stage threshold and less than the continuous degradation stage threshold, the fixed component is classified as minor degradation stage; when the cumulative health value is not less than the continuous degradation stage threshold and less than the fault alarm stage threshold, the fixed component is classified as continuous degradation stage; when the cumulative health value is not less than the fault alarm stage threshold, the fixed component is classified as fault alarm stage.

[0200] It should be noted that the health stage threshold is jointly tuned based on the distribution of cumulative health values ​​in the normal and stable operation samples and the distribution of cumulative health values ​​in the manually inspected and confirmed samples. When manually inspected and confirmed samples are available, the threshold for the early concern stage is the 95th percentile of the cumulative health values ​​in the normal and stable operation samples, the threshold for the slight degradation stage is the lower quantile of the cumulative health values ​​in the manually inspected and confirmed slightly degradation samples, the threshold for the continuous degradation stage is the lower quantile of the cumulative health values ​​in the manually inspected and confirmed continuous degradation samples, and the threshold for the fault alarm stage is the lower quantile of the cumulative health values ​​in the manually inspected and confirmed fault samples. The lower quantile value is usually taken from the 5th percentile to the 25th percentile of the cumulative health values ​​in the corresponding manually inspected and confirmed samples. For example, the 10th percentile is taken. The reason for taking the 10th percentile is that the 10th percentile can retain the low-end boundary characteristics of the cumulative health values ​​in the corresponding manually inspected and confirmed samples and reduce the impact of a single low-value sample on the health stage threshold.

[0201] It should be noted that when the number of any of the following samples—slightly deteriorated, continuously deteriorated, or faulty—is lower than the minimum number of samples to be manually inspected, the lower quantile of the corresponding category of manually inspected samples is not used to determine the health stage threshold. Instead, the health cumulative value quantile initialization rule for when there are no manually inspected samples is used to determine the corresponding health stage threshold. After the number of samples in the corresponding category reaches the minimum number of manually inspected samples, the corresponding health stage threshold is readjusted based on the distribution of the health cumulative value in the manually inspected samples.

[0202] It should be noted that the minimum number of manual inspection confirmation samples is determined based on the dispersion of the cumulative health value distribution and the stability of the quantile values ​​in the manual inspection confirmation samples. The minimum number of manual inspection confirmation samples is usually between 20 and 50, and 30 is used in the example. The reason for using 30 is that 30 samples can form a preliminary statistical result of the cumulative health value distribution under the corresponding manual inspection confirmation category, and reduce the impact of a single inspection sample on the lower quantile value.

[0203] It should be noted that when manual inspection and confirmation samples are lacking, the health stage threshold is initialized based on the percentile of the cumulative health value in the normal and stable operation samples; the threshold range for the early concern stage is usually from the 95th percentile to the 97.5th percentile, the threshold range for the slight degradation stage is usually from the 97.5th percentile to the 99th percentile, the threshold range for the continuous degradation stage is usually from the 99th percentile to the 99.5th percentile, and the threshold range for the fault alarm stage is usually from the 99.5th percentile to the 99.9th percentile.

[0204] Furthermore, when the fixed component is in the early concern stage, slight deterioration stage, continuous deterioration stage, or fault alarm stage during the analysis period, a belt noise early warning record is generated.

[0205] It should be noted that the belt noise warning record includes the fixed component number, the center position of the fixed component, the monitoring space range of the fixed component, the interpretation degree of the fixed component, the interpretation degree of the belt movement, the anchoring coefficient of the fixed component, the rotation phase reproduction coefficient, the confidence degree of the mechanical noise, the cumulative health value, the health stage, and the corresponding analysis cycle number.

[0206] It should be noted that when the fixed component is in a healthy state during the analysis period, no belt noise warning record is generated; only the cumulative health value and health stage update result of the corresponding fixed component are retained.

[0207] This embodiment also provides a computer device applicable to the belt noise early warning method based on pattern recognition, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the belt noise early warning method based on pattern recognition as proposed in the above embodiment.

[0208] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0209] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the belt noise warning method based on pattern recognition as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0210] In summary, this invention distinguishes between the coordinates of the fixed component and the trajectory of the accompanying movement by calculating the interpretation degree of the fixed component and forming the anchoring coefficient of the fixed component, thus giving the candidate objects of the fixed component anchoring anomaly a clear spatial assignment; by combining the rotation phase reproduction coefficient and performing nonlinear gating coupling on the mechanical reproduction abnormal noise object, the confidence degree of the mechanical abnormal noise is obtained and the health accumulation value is updated, so that the belt abnormal noise early warning record simultaneously reflects the spatial anchoring, mechanical reproduction and deterioration accumulation process, thereby improving the pertinence and continuity of the belt abnormal noise early warning results.

[0211] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A belt conveyor noise early warning method based on pattern recognition, characterized in that, include: A sound and vibration sensing optical cable is laid along the frame structure corresponding to the fixed components of the belt conveyor to collect sound and vibration data of the belt conveyor. The idler rollers and drums are used as fixed components for spatial calibration. The center position of the fixed component, the monitoring space range of the fixed component, the diameter of the fixed component, the correspondence between the belt running direction and the optical cable distance direction, and the belt running speed are recorded to generate a normalized sound and vibration energy map and a high frequency envelope map. Based on the normalized acoustic and vibration energy map, the interpretation degree of the fixed component is calculated for each fixed component. According to the correspondence between the belt running speed and the belt running direction and the optical cable distance direction, the normalized acoustic and vibration energy map is backtracked to the position of belt movement, and the interpretation degree of belt movement is calculated. Combining the interpretation degree of the fixed component and the interpretation degree of belt movement, the anchoring coefficient of the fixed component is calculated and the candidate objects of fixed component anchoring anomalies are screened. For candidates with abnormal anchoring of fixed components, the rotation phase of the fixed component is determined based on the diameter of the fixed component and the belt running speed. The rotation phase reproduction coefficient is calculated by combining the high-frequency envelope diagram to screen out mechanically reproduced abnormal noise objects. Nonlinear gating coupling is applied to the mechanically reproduced abnormal noise object to obtain the mechanical abnormal noise confidence level. The health accumulation value is updated based on the mechanical abnormal noise confidence level to generate belt abnormal noise early warning records.

2. The belt noise early warning method based on pattern recognition as described in claim 1, characterized in that, The generation of the normalized acoustic vibration energy map and the high-frequency envelope map includes: The optical cable distance coordinates are established by taking the starting end of the acoustic vibration sensing optical cable as the zero point. Number the fixed components of the idlers and rollers to form a basic data table of fixed components; The corresponding position of each fixed component in the optical cable distance coordinate is calibrated to obtain the center position of the fixed component; The monitoring space range of the fixed component is determined based on the center position of the fixed component and the installation spacing between adjacent fixed components. Record the correspondence between the diameter of the fixed component, the direction of belt running and the direction of optical cable distance, and the belt running speed; The belt conveyor acoustic and vibration data are divided into analysis periods. The energy of the abnormal noise sensitive frequency band is extracted and a normal acoustic and vibration energy benchmark is established. The normalized acoustic and vibration energy map is generated by combining the abnormal noise sensitive frequency band energy and the normal acoustic and vibration energy benchmark. A high-frequency envelope map is generated by high-frequency filtering and envelope extraction.

3. The belt noise early warning method based on pattern recognition as described in claim 2, characterized in that, The process of generating a normalized acoustic vibration energy map by combining the energy of the abnormal noise-sensitive frequency band with the normal acoustic vibration energy benchmark includes: The energy of the sensitive frequency band of abnormal noise at the same optical cable distance within the analysis period is compared with the median value of normal energy in the normal acoustic and vibration energy benchmark. When the energy of the sensitive frequency band of abnormal noise is not higher than the median value of normal energy, the normalized acoustic vibration energy value of the corresponding optical cable distance position in the corresponding analysis period is set to zero; When the energy of the sensitive frequency band of abnormal noise is higher than the median value of normal energy, it is normalized according to the discrete value of normal acoustic and vibration energy benchmark to obtain the normalized acoustic and vibration energy value of the corresponding optical cable distance position in the corresponding analysis period.

4. The belt noise early warning method based on pattern recognition as described in claim 1, characterized in that, The calculation of the explanation degree of each fixed component includes: Read the monitoring space range of each fixed component from the fixed component basic data table; After removing invalid optical cable distance positions within the monitoring space of the fixed components, the valid optical cable distance positions are arranged according to the optical cable distance coordinates; Set the number of consecutive analysis cycles for competition identification; The normalized acoustic and vibration energy values ​​at the effective optical cable distance positions within the monitoring space range of the corresponding fixed component are averaged within the current analysis period and the continuous analysis periods going back to the historical direction to obtain the fixed component interpretation degree.

5. The belt noise early warning method based on pattern recognition as described in claim 4, characterized in that, The step of performing position backtracking on the normalized acoustic vibration energy map with accompanying shift and calculating the explanatory power of accompanying shift includes: Based on the average belt speed and the analysis cycle update interval within each analysis cycle, the belt travel distance between the historical analysis cycle and the current analysis cycle is accumulated. Based on the correspondence between the belt movement distance and the belt running direction and the optical cable distance direction, the effective optical cable distance position within the monitoring space of the fixed component is traced back to its historical position to obtain the belt movement traceback position. The position of the follow-up movement is compared with the effective acquisition range of the acoustic vibration sensing optical cable. The follow-up movement position that falls within the effective acquisition range of the acoustic vibration sensing optical cable is included in the calculation of the follow-up movement interpretation degree, while the follow-up movement position that falls outside the effective acquisition range of the acoustic vibration sensing optical cable is not included in the calculation of the follow-up movement interpretation degree. The normalized acoustic and vibration energy values ​​of the retrospective positions involved in the calculation of the accompanying motion interpretation degree are read from the normalized acoustic and vibration energy map within the corresponding historical analysis period. The normalized acoustic and vibration energy values ​​read are averaged to obtain the accompanying motion interpretation degree.

6. The belt noise early warning method based on pattern recognition as described in claim 5, characterized in that, The process of combining the interpretation degree of the fixed component with the interpretation degree of the accompanying movement to calculate the anchorage coefficient of the fixed component and to screen candidate objects of fixed component anchorage anomalies includes: When the sum of the fixed component's explanation degree and the accompanying movement's explanation degree is greater than zero, the anchorage coefficient of the fixed component is calculated based on the ratio between the difference between the fixed component's explanation degree and the accompanying movement's explanation degree and the sum of the fixed component's explanation degree and the accompanying movement's explanation degree. When the sum of the fixed component's explanation degree and the accompanying movement's explanation degree is not greater than zero, the anchorage coefficient of the fixed component is recorded as zero; When a fixed component simultaneously satisfies both the fixed component interpretability not less than the fixed component interpretability threshold and the fixed component anchoring coefficient not less than the fixed component anchoring threshold within the corresponding analysis period, the corresponding fixed component will be marked as a candidate object for fixed component anchoring anomaly.

7. The belt noise early warning method based on pattern recognition as described in claim 1 or 6, characterized in that, The process of determining the rotation phase of the fixed component based on its diameter and belt speed, and calculating the rotation phase reproduction coefficient using a high-frequency envelope diagram, to screen for mechanically reproduced abnormal noises includes: Rotation phase reproduction identification is performed only on candidates with anchorage anomalies in fixed components; The rotation phase of the fixed component is calculated by sampling based on the diameter of the fixed component and the speed of the belt; The high-frequency envelope values ​​of the fixed component within the monitoring space are extracted from the high-frequency envelope map, and the high-frequency envelope values ​​at the effective optical cable distance locations are averaged to obtain the fixed component envelope value. Divide a complete rotation cycle into phase intervals, and distribute the envelope value of the fixed component to the corresponding phase interval according to the rotation phase of the fixed component, so as to obtain the phase envelope distribution value in each effective complete rotation cycle. The rotation phase recurrence coefficient is calculated based on the phase envelope distribution similarity between several valid complete rotation cycles and by summarizing the phase envelope distribution similarity. When the rotation phase reproduction coefficient is not less than the rotation phase reproduction threshold, the corresponding fixed component is marked as a mechanically reproduced abnormal noise object.

8. The belt noise early warning method based on pattern recognition as described in claim 1, characterized in that, The process of performing nonlinear gated coupling on the mechanically reproduced abnormal noise object, obtaining the mechanical abnormal noise confidence level, updating the health accumulation value based on the mechanical abnormal noise confidence level, and generating a belt abnormal noise early warning record includes: The anchoring coefficient of the fixed component is converted into the anchoring gating factor of the fixed component, the rotation phase reproduction coefficient is converted into the phase reproduction gating factor, the explanatory power of the fixed component is converted into the explanatory strength factor of the fixed component, the dominant factor of the fixed component is calculated based on the explanatory power of the fixed component and the explanatory power of the accompanying movement, the growth amount of the explanatory power of the fixed component is determined based on the difference between the average explanatory power of the fixed component in the current time period and the average explanatory power of the fixed component in adjacent historical time periods, and the growth gating factor is calculated based on the growth amount of the explanatory power of the fixed component. The anchoring gating factor and the phase reproduction gating factor of the fixed component are nonlinearly coupled to obtain the anchoring phase coupling factor; The anchoring phase coupling factor, the fixed component interpretation strength factor, and the fixed component dominance factor are product-coupled, and the product-coupled result is adjusted by the growth gating factor to obtain the confidence level of mechanical abnormal noise. Update the health cumulative value for the current analysis period based on the confidence level of mechanical abnormal noise and the health cumulative value of the previous analysis period; Based on the accumulated health value, fixed components are divided into the healthy state, early attention stage, slight deterioration stage, continuous deterioration stage, and fault alarm stage. When the fixed component is in the early concern stage, slight deterioration stage, continuous deterioration stage, or fault alarm stage during the analysis period, a belt noise early warning record is generated.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the belt noise early warning method based on pattern recognition as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the belt noise early warning method based on pattern recognition as described in any one of claims 1 to 8.

Images