Belt conveyor state monitoring system based on multi-modal data

Abstract

The invention relates to the technical field of belt conveyor state monitoring, and discloses a belt conveyor state monitoring system based on multi-modal data, which comprises a data acquisition unit, a feature extraction unit, an anomaly analysis unit and a fault display unit, by fusing vibration, temperature and acoustics multi-mode data, a three-dimensional system for belt conveyor state monitoring is constructed, and the comprehensiveness of fault detection is improved; mechanical abnormity of the driving roller bearing is captured in real time, and the sensitivity is superior to that of traditional single threshold detection; local overheating or global temperature rise of the belt is accurately recognized, and environmental interference is effectively eliminated through secondary temperature analysis; the broadband acoustic sensor is combined with short-time Fourier transform and an acoustic wave feature library matching technology, and the tail wheel abnormal sound fault type is accurately positioned according to the frequency band energy proportion difference; vibration abnormity is instantly alarmed, temperature abnormity is verified through local abrupt change and global mean value, acoustic faults are dynamically matched and classified based on similarity of a feature library, and real-time performance and reliability are both considered.

Description

Technical Field

[0001] The present invention relates to the technical field of belt conveyor state monitoring, and in particular to a belt conveyor state monitoring system based on multimodal data. Background Art

[0002] Belt conveyors, as core equipment for continuous material transportation, are widely used in industries such as mining, metallurgy, ports, and power generation. Their operational stability directly impacts production efficiency and equipment safety. However, under long-term high loads and complex operating conditions, belt conveyors are prone to failures such as drive roller bearing wear, belt slippage, localized overheating, and abnormal tail pulley noise. Failure to provide timely warnings can lead to equipment downtime or even accidents.

[0003] Traditional condition monitoring methods often use single-modal sensor data, such as detecting bearing anomalies through vibration sensors or monitoring belt surface temperature using temperature sensors. These methods have obvious limitations: a single sensor can only reflect local conditions and cannot fully capture the correlation anomalies between mechanical vibration, thermal distribution, and acoustic characteristics. For example, early bearing wear may be accompanied by slight vibration changes, but a single vibration threshold makes it difficult to distinguish between normal fluctuations and precursors to failure; local belt slippage can cause a sudden temperature rise, but global temperature average monitoring tends to ignore local hotspots; industrial field environments are complex, vibration signals are easily affected by equipment-coupled vibration interference, temperature data is easily affected by environmental temperature drift, and acoustic signals are mixed with background noise, resulting in a high false alarm rate in traditional threshold judgment methods; existing acoustic diagnosis mostly relies on manual experience or fixed-frequency band analysis, lacks acoustic feature library support for different faults (such as bearing damage and belt tearing), and is difficult to achieve accurate classification.

[0004] In recent years, multimodal data fusion technology has provided new insights for equipment condition monitoring. However, effectively extracting cross-modal key features and establishing hierarchical diagnostic strategies remain technical challenges. For example, issues such as the time-series alignment of vibration and temperature data and the dynamic matching of acoustic signal frequency band energy ratios have yet to be fully addressed.

[0005] Based on this, there is an urgent need for a belt conveyor condition monitoring system that integrates multimodal sensing, has anti-interference feature extraction capabilities, and can realize multi-level abnormality judgment, so as to improve the comprehensiveness, accuracy and real-time performance of fault detection. Summary of the Invention

[0006] The purpose of the present invention is to provide a belt conveyor condition monitoring system based on multimodal data, which solves the technical problems raised in the background technology.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] A belt conveyor condition monitoring system based on multimodal data, comprising:

[0009] Data acquisition unit, used to collect status monitoring data of the belt conveyor;

[0010] A feature extraction unit is used to extract key feature values ​​from the condition monitoring data of the belt conveyor;

[0011] Key characteristic values ​​include: the vibration intensity growth rate of the belt conveyor drive roller bearing seat, the temperature variation amplitude of the belt surface, and the proportion of energy in different frequency bands at the tail pulley to the total energy;

[0012] An abnormality analysis unit, used to analyze the abnormal state of the belt conveyor based on the key feature values ​​extracted by the feature extraction unit;

[0013] The fault display unit is used to display the results of the abnormality analysis unit to relevant management personnel.

[0014] As a further solution of the present invention: the status monitoring data includes:

[0015] The vibration sensor group installed at the belt conveyor drive roller bearing seat collects the amplitude data of the belt conveyor drive roller bearing seat;

[0016] Among them, the vibration sensor group uses a 3-axis accelerometer to measure the vibration amplitude corresponding to the X-axis, Y-axis, and Z-axis, and marks them as Ax, Ay, and Az respectively;

[0017] The surface temperature of the belt conveyor is collected by a temperature sensor group installed under the belt bearing section;

[0018] The temperature sensor group is an infrared thermal array temperature sensor, and the belt surface temperature it collects is marked as T i，j , i = 1, 2, ... n, m represents the number of rows of the infrared thermal array temperature sensor, j = 1, 2, ... m, m represents the number of columns of the infrared thermal array temperature sensor; T i，j is the belt surface temperature collected at row i and column j;

[0019] The acoustic sensor installed at the tail wheel of the belt conveyor collects the acoustic wave signal of the corresponding tail wheel of the belt conveyor;

[0020] Among them, the acoustic sensor uses a broadband microphone and collects sound wave signals through it.

[0021] As a further solution of the present invention: the key feature value extraction method is as follows:

[0022] Step 1: Extraction of vibration intensity growth rate:

[0023] At the current acquisition node, the amplitude data collected by the vibration sensor group is extracted, and the modulus of the three-axis vibration composite vector is calculated based on the amplitude data;

[0024] The calculation method is as follows: at the current acquisition node, the amplitude data collected by the vibration sensor group is extracted, and then the square of the vibration amplitude corresponding to the X-axis, Y-axis, and Z-axis is calculated respectively. Then, the square root of the sum is added and taken to obtain the modulus of the composite vector of the three-axis vibration. At the same time, the modulus of the composite vector of the three-axis vibration at the previous specified acquisition node is calculated.

[0025] The calculation formula for the triaxial vibration composite vector modulus is:

[0026] At the same time, the modulus of the three-axis vibration composite vector on the previous specified acquisition node is calculated;

[0027] Then, the vibration intensity growth rate is calculated based on the modulus of the three-axis vibration synthesis vector at the current acquisition node and the previous designated acquisition node;

[0028] The calculation method is: subtract the modulus of the three-axis vibration composite vector of the current acquisition node from the modulus of the three-axis vibration composite vector of the previous designated acquisition node, and then divide the result by the modulus of the three-axis vibration composite vector of the previous designated acquisition node to obtain the vibration intensity growth rate;

[0029] The calculation formula for the vibration intensity growth rate is:

[0030] Where ZA is the vibration intensity growth rate, A1 is the modulus of the three-axis vibration composite vector of the current acquisition node, and A2 is the modulus of the three-axis vibration composite vector of the previous designated acquisition node;

[0031] Step 2: Extract temperature variation:

[0032] First, extract the belt surface temperature T corresponding to the belt conveyor i，j , and extract the maximum and minimum belt surface temperatures, and subtract the minimum belt surface temperature from the belt surface temperature corresponding to the belt conveyor, and at the same time subtract the minimum belt surface temperature from the maximum belt surface temperature, then divide the previous difference result by the next difference result, and then multiply by 100 to obtain the normalized belt surface temperature;

[0033] The calculation formula is as follows:

[0034] pass: Get the normalized belt surface temperature T1 i，j ;

[0035] Where, T max and T min are the belt surface temperatures with the maximum and minimum values ​​respectively;

[0036] Then select a normalized belt surface temperature T1 i，j , and extract multiple normalized belt surface temperatures T1 adjacent to the belt surface temperature i+1，j 、T1 i-1，j 、T1 i，j+1 、T1 i，j-1 、T1 i+1，j+1 、T1 i-1，j-1 、T1 i+1，j-1 、T1 i-1，j+1 ;

[0037] Calculating the absolute value of the temperature difference between the selected normalized belt surface temperature and multiple adjacent normalized belt surface temperatures, then averaging the multiple absolute values ​​of the temperature differences, and recording the average as the temperature variation amplitude of the belt surface temperature at the i-th row and j-th column;

[0038] The formula is:

[0039] Among them, TC i，j is the temperature variation of the belt surface temperature at row i and column j;

[0040] Step 3: Extract the proportion of energy in different frequency bands to the total energy:

[0041] Extract the acoustic wave signal of the tail wheel of the belt conveyor, then use short-time Fourier transform to analyze the time-frequency characteristics of the acoustic wave signal, and obtain the energy distribution of the sound signal at different times and frequencies, that is, obtain the time-frequency diagram;

[0042] Extract the maximum and minimum values ​​of the frequency from the time-frequency graph, then form the entire frequency range based on the maximum and minimum values ​​of the frequency, and then divide it into several frequency segments. Then select a frequency segment and obtain the energy value of the frequency segment based on the time-frequency graph;

[0043] Extract the total energy value of the entire frequency range from the time-frequency diagram;

[0044] The proportion of energy in different frequency bands to the total energy is obtained by dividing the energy value of each frequency band by the total energy value of the entire frequency range.

[0045] As a further solution of the present invention: the abnormal state analysis method is as follows:

[0046] The first step is to compare the vibration intensity growth rate of the belt conveyor drive roller bearing seat with the preset vibration intensity growth rate threshold:

[0047] When the vibration intensity growth rate is greater than or equal to the vibration intensity growth rate threshold, it is determined that the vibration of the belt conveyor drive roller bearing seat is abnormal;

[0048] When the vibration intensity growth rate is less than the vibration intensity growth rate threshold, it is determined that the vibration of the belt conveyor drive roller bearing seat is normal;

[0049] The second step is to compare the temperature variation at each position on the belt surface with the preset temperature variation threshold:

[0050] When the temperature variation amplitude of one position among all positions on the belt surface is greater than or equal to the temperature variation amplitude threshold, it is determined that the temperature of the belt surface is abnormal;

[0051] The third step is to select a variety of belt conveyors with different fault types and measure their acoustic wave signals through experiments. Then, a feature extraction unit is used to extract the energy ratio of the belt conveyors with different fault types in different frequency bands to the total energy ratio. Then, sound comparison samples are generated for the belt conveyors with different fault types and their energy ratio in different frequency bands to the total energy ratio. An acoustic wave feature library is established based on the multiple sound comparison samples.

[0052] Then, the proportion of the belt conveyor's energy in different frequency bands to the total energy is extracted, and the similarity is determined with the acoustic wave feature library to obtain the corresponding similarity value;

[0053] Then, the similarity between the proportion of the energy of the belt conveyor in different frequency bands and the sound comparison samples in the sound wave feature library is compared with the corresponding preset similarity thresholds, and then the fault type corresponding to the sound comparison sample with the similarity ≥ the similarity threshold and the largest similarity value is selected as the fault type of the belt conveyor.

[0054] As a further solution of the present invention: when the temperature variation amplitude of all positions on the belt surface is less than the temperature variation amplitude threshold, it is determined that the belt surface is subjected to secondary temperature analysis;

[0055] The secondary temperature analysis method is as follows:

[0056] Extract all belt surface temperatures T1 after normalization i，j , then calculate the surface temperature T1 of all belts i，j The average value is then compared to a preset belt temperature threshold:

[0057] If the average value of all belt surface temperatures is ≥ the belt temperature threshold, the belt surface temperature is determined to be abnormal;

[0058] If the average value of all belt surface temperatures is less than the belt temperature threshold, it is determined that the belt surface temperature is normal.

[0059] As a further solution of the present invention: the similarity determination method is:

[0060] Select a sound comparison sample from the sound wave feature library;

[0061] pass: Select the frequency segment with the largest energy proportion to the total energy from the sound comparison samples;

[0062] At the same time through Select the frequency segment with the largest energy proportion to the total energy from the belt conveyor;

[0063] Wherein, Ek represents the proportion of the energy of the belt conveyor in the kth frequency segment to the total energy, k = 1, 2, ... v, v represents the number of different frequency segments in the belt conveyor, EYg represents the proportion of the energy in the gth frequency segment of the selected sound comparison sample to the total energy, g = 1, 2, ... u, u represents the number of different frequency segments in the sound comparison sample;

[0064] The two selected frequency bands are then compared:

[0065] When the two frequency bands are consistent, the following is achieved: Calculate the similarity EC between the ratio of the energy of the belt conveyor in different frequency bands to the total energy and the sound comparison sample;

[0066] When the two frequency segments are inconsistent, the frequency segment with the largest energy ratio in the belt conveyor is selected, and the same frequency segment is selected from the sound comparison sample, and the energy ratio of the frequency segment to the total energy is obtained, and it is marked as M(E k ), followed by: Calculate the similarity EC between the ratio of the energy of the belt conveyor in different frequency bands to the total energy and the sound comparison sample;

[0067] By analogy, the proportion of the belt conveyor's energy in different frequency bands to the total energy and the similarity between the various sound comparison samples in the sound wave feature library are calculated.

[0068] Beneficial effects of the present invention:

[0069] This invention integrates vibration, temperature, and acoustic data to create a comprehensive monitoring system. The vibration intensity growth rate dynamically detects mechanical anomalies in the bearing housing, the infrared thermal array accurately identifies localized belt overheating or global temperature rise, and acoustic frequency band energy analysis effectively locates abnormal tail pulley noise. These three collaborative features enable comprehensive monitoring of the conveyor belt's mechanical dynamics, thermal state, and acoustic characteristics, significantly improving fault detection coverage.

[0070] In the present invention, the vibration intensity growth rate is calculated by the hourly vibration mode length change rate, which can detect gradual faults such as bearing wear and loosening in advance; temperature monitoring adopts normalization processing and adjacent node difference mean analysis, which can not only identify local drastic temperature changes, but also eliminate environmental interference through secondary temperature analysis, reducing the false alarm rate; the acoustic feature library similarity matching technology combined with short-time Fourier transform can distinguish the characteristic frequency bands of different fault types such as bearing damage and belt tearing, thereby improving diagnostic specificity.

[0071] The anomaly analysis unit in this invention employs a multi-level decision logic: vibration anomalies directly trigger an alarm, temperature anomalies are verified through both local mutation thresholds and global mean thresholds, and acoustic faults are dynamically matched with a sound wave signature library to locate the fault type. This strategy balances real-time response with precise judgment, preventing misjudgments from a single sensor and enhancing the credibility of the results through cross-validation of multi-source data.

[0072] This invention uses a three-axis composite vector modulus calculation to extract vibration features, eliminating interference from single-direction vibrations. Temperature data is normalized to eliminate the influence of individual sensor differences. Acoustic signals are analyzed through frequency band energy ratio analysis to effectively suppress background noise. These processing methods ensure that the eigenvalues ​​truly reflect the device status, enhancing the system's robustness in complex industrial environments.

[0073] This invention builds an acoustic signature library based on experimental samples, supporting the continuous addition of acoustic signature data for new fault types, enabling the system to self-learn. The similarity calculation model is compatible with different frequency band division methods, adapting to the diagnosis needs of diverse belt conveyor fault types and providing a dynamic optimization foundation for equipment health management.

[0074] This invention uses multimodal feature fusion analysis to differentiate between mechanical faults (primarily vibration anomalies), thermal faults (primarily temperature anomalies), and complex faults (multiple indicators exhibiting concurrent anomalies), helping managers quickly locate the root cause. The fault display unit integrates vibration alarms, temperature heat maps, and acoustic spectrum comparison visualization, significantly reducing manual inspection workload and improving equipment maintenance efficiency and safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] The present invention will be further described below with reference to the accompanying drawings.

[0076] Figure 1 This is a system block diagram of a belt conveyor condition monitoring system based on multimodal data of the present invention.

[0077] Figure 2 It is a flow chart of a feature extraction unit in a belt conveyor condition monitoring system based on multimodal data of the present invention. DETAILED DESCRIPTION

[0078] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0079] Example 1

[0080] See also Figure 1 and Figure 2 As shown, the present invention is a belt conveyor condition monitoring system based on multimodal data, comprising:

[0081] Data acquisition unit, used to collect status monitoring data of the belt conveyor;

[0082] Condition monitoring data includes:

[0083] The vibration sensor group installed at the belt conveyor drive roller bearing seat collects the amplitude data of the belt conveyor drive roller bearing seat;

[0084] Among them, the vibration sensor group uses a 3-axis accelerometer to measure the vibration amplitude corresponding to the X-axis, Y-axis, and Z-axis, and marks them as Ax, Ay, and Az respectively;

[0085] The feature extraction unit is used to extract key eigenvalues ​​from the condition monitoring data of the belt conveyor. The key eigenvalues ​​include: the vibration intensity growth rate of the belt conveyor drive roller bearing seat; the key eigenvalue extraction method is as follows:

[0086] At the current acquisition node, the amplitude data collected by the vibration sensor group is extracted, and the modulus of the three-axis vibration composite vector is calculated based on the amplitude data;

[0087] The calculation formula for the triaxial vibration composite vector modulus is:

[0088] At the same time, the modulus of the three-axis vibration composite vector on the previous specified acquisition node is calculated;

[0089] Then, the vibration intensity growth rate is calculated based on the modulus of the three-axis vibration synthesis vector at the current acquisition node and the previous designated acquisition node;

[0090] The calculation formula for the vibration intensity growth rate is:

[0091] Where ZA is the vibration intensity growth rate, A1 is the modulus of the three-axis vibration composite vector of the current acquisition node, and A2 is the modulus of the three-axis vibration composite vector of the previous designated acquisition node;

[0092] In this embodiment, the time interval between the current collection node and the previous designated collection node is 1 hour;

[0093] An abnormality analysis unit, used to analyze the abnormal state of the belt conveyor based on the key feature values ​​extracted by the feature extraction unit;

[0094] The abnormal status analysis method is as follows:

[0095] Compare the vibration intensity increase rate of the belt conveyor drive roller bearing seat with the preset vibration intensity increase rate threshold:

[0096] When the vibration intensity growth rate is greater than or equal to the vibration intensity growth rate threshold, it is determined that the vibration of the belt conveyor drive roller bearing seat is abnormal;

[0097] When the vibration intensity growth rate is less than the vibration intensity growth rate threshold, it is determined that the vibration of the belt conveyor drive roller bearing seat is normal;

[0098] The fault display unit is used to display the results of the abnormality analysis unit to relevant management personnel.

[0099] This embodiment uses the three-axis vibration composite vector modulus to calculate the vibration intensity growth rate in real time, accurately capturing abnormal vibration trends in the belt conveyor's drive roller bearing seat. A dynamic comparison mechanism, performed at hourly intervals, effectively identifies sudden changes in vibration intensity, avoiding the lag inherent in traditional single-threshold detection and significantly enhancing early warning capabilities for mechanical faults such as bearing wear and loosening. Furthermore, by directly comparing the vibration intensity growth rate with the threshold, the system simplifies the abnormality determination logic, reduces system computational complexity, facilitates rapid feedback to management personnel, and improves maintenance response efficiency.

[0100] Example 2

[0101] As the second embodiment of the present invention, when this application is specifically implemented, compared with the first embodiment, the technical solution of this embodiment differs from that of the first embodiment only in that:

[0102] Condition monitoring data also includes:

[0103] The surface temperature of the belt conveyor is collected by a temperature sensor group installed under the belt bearing section;

[0104] The temperature sensor group is an infrared thermal array temperature sensor, and the belt surface temperature it collects is marked as T i，j , i = 1, 2, ... n, m represents the number of rows of the infrared thermal array temperature sensor, j = 1, 2, ... m, m represents the number of columns of the infrared thermal array temperature sensor; T i，j is the belt surface temperature collected at row i and column j;

[0105] The key eigenvalues ​​include: the temperature variation of the belt surface; the key eigenvalue extraction method is as follows:

[0106] First, extract the belt surface temperature T corresponding to the belt conveyor i，j The maximum and minimum belt surface temperatures are extracted and marked as T max and T min ;

[0107] pass: Get the normalized belt surface temperature T1 i，j ;

[0108] Then select a normalized belt surface temperature T1 i，j , and extract multiple normalized belt surface temperatures T1 adjacent to the belt surface temperature i+1，j 、T1 i-1，j 、T1 i，j+1 、T1 i，j-1 、T1 i+1，j+1 、T1 i-1，j-1 、T1 i+1，j-1 、T1 i-1，j+1 ;

[0109] pass:

[0110] Calculate the temperature variation of the belt surface temperature at row i and column j;

[0111] The abnormal state analysis method is as follows:

[0112] Compare the temperature variation at each location on the belt surface with the preset temperature variation threshold:

[0113] When the temperature variation amplitude of one position among all positions on the belt surface is greater than or equal to the temperature variation amplitude threshold, it is determined that the temperature of the belt surface is abnormal;

[0114] When the temperature variation amplitude at all positions on the belt surface is less than the temperature variation amplitude threshold, the belt surface is determined to undergo secondary temperature analysis;

[0115] The secondary temperature analysis method is as follows:

[0116] Extract all belt surface temperatures T1 after normalization i，j , then calculate the surface temperature T1 of all belts i，j The average value is then compared to a preset belt temperature threshold:

[0117] If the average value of all belt surface temperatures is ≥ the belt temperature threshold, the belt surface temperature is determined to be abnormal;

[0118] If the average value of all belt surface temperatures is less than the belt temperature threshold, the belt surface temperature is determined to be normal;

[0119] This embodiment introduces an infrared thermistor array temperature sensor group based on vibration monitoring. Through normalization processing and averaging of adjacent temperature differences, it achieves highly sensitive detection of belt surface temperature anomalies. Its secondary temperature analysis mechanism (combining the local temperature change amplitude with the overall temperature mean threshold) effectively distinguishes between local overheating and global temperature rise scenarios, avoiding misjudgments caused by ambient temperature fluctuations. This solution is particularly suitable for detecting temperature anomalies caused by belt slippage, uneven load, or abnormal friction, compensating for the shortcomings of single vibration monitoring and forming a multi-dimensional fault protection system.

[0120] Example 3

[0121] As the third embodiment of the present invention, when the present application is specifically implemented, compared with the first and second embodiments, the technical solution of this embodiment is to combine the solutions of the first and second embodiments. The technical solution of this embodiment differs from those of the first and second embodiments only in that:

[0122] Condition monitoring data also includes:

[0123] The acoustic sensor installed at the tail wheel of the belt conveyor collects the acoustic wave signal of the corresponding tail wheel of the belt conveyor;

[0124] Among them, the acoustic sensor uses a broadband microphone and collects sound wave signals through it;

[0125] The key characteristic values ​​include: the temperature variation of the belt surface, the proportion of energy in different frequency bands at the tail pulley to the total energy;

[0126] The key eigenvalue extraction method is as follows: extract the acoustic wave signal of the tail wheel corresponding to the belt conveyor, then use short-time Fourier transform to analyze the time-frequency characteristics of the acoustic wave signal, and obtain the energy distribution of the sound signal at different times and frequencies, that is, obtain a time-frequency diagram;

[0127] Extract the maximum and minimum values ​​of the frequency from the time-frequency graph, then form the entire frequency range based on the maximum and minimum values ​​of the frequency, and then divide it into several frequency segments. Then select a frequency segment and obtain the energy value of the frequency segment based on the time-frequency graph;

[0128] Extract the total energy value of the entire frequency range from the time-frequency diagram;

[0129] The energy value of each frequency band is divided by the total energy value of the entire frequency range to obtain the proportion of energy in different frequency bands to the total energy;

[0130] The abnormal status analysis method is as follows:

[0131] A variety of belt conveyors with different fault types were selected, and their acoustic wave signals were measured experimentally. Subsequently, a feature extraction unit was used to extract the energy ratio of the belt conveyors with different fault types in different frequency bands to the total energy ratio. Then, sound comparison samples were generated for the belt conveyors with different fault types and their energy ratio in different frequency bands to the total energy ratio. An acoustic wave feature library was established based on the multiple sound comparison samples.

[0132] Then, the proportion of the belt conveyor's energy in different frequency bands to the total energy is extracted, and the similarity is determined by comparing it with the acoustic wave feature library. The similarity determination method is as follows:

[0133] Select a sound comparison sample from the sound wave feature library;

[0134] pass: Select the frequency segment with the largest energy proportion to the total energy from the sound comparison samples;

[0135] At the same time through Select the frequency segment with the largest energy proportion to the total energy from the belt conveyor;

[0136] Wherein, Ek represents the proportion of the energy of the belt conveyor in the kth frequency segment to the total energy, k = 1, 2, ... v, v represents the number of different frequency segments in the belt conveyor, EYg represents the proportion of the energy in the gth frequency segment of the selected sound comparison sample to the total energy, g = 1, 2, ... u, u represents the number of different frequency segments in the sound comparison sample;

[0137] The two selected frequency bands are then compared:

[0138] When the two frequency bands are consistent, the following is achieved: Calculate the similarity EC between the ratio of the energy of the belt conveyor in different frequency bands to the total energy and the sound comparison sample;

[0139] When the two frequency segments are inconsistent, the frequency segment with the largest energy ratio in the belt conveyor is selected, and the same frequency segment is selected from the sound comparison sample, and the energy ratio of the frequency segment to the total energy is obtained, and it is marked as M(E k ), followed by: Calculate the similarity EC between the ratio of the energy of the belt conveyor in different frequency bands to the total energy and the sound comparison sample;

[0140] By analogy, the similarity between the proportion of the belt conveyor's energy in different frequency bands to the total energy and the sound comparison samples in the sound wave feature library is calculated;

[0141] Then, the similarity between the proportion of the energy of the belt conveyor in different frequency bands to the total energy and each sound comparison sample in the sound wave feature library is compared with the corresponding preset similarity threshold, and then the fault type corresponding to the sound comparison sample with the largest similarity value and a similarity greater than or equal to the similarity threshold is selected as the fault type of the belt conveyor;

[0142] This embodiment further integrates acoustic sensors with short-time Fourier transform technology. By analyzing the time-frequency energy distribution characteristics of the tail wheel acoustic wave signal, constructing an acoustic wave feature library, and calculating similarity matching, it achieves accurate diagnosis of various fault types (such as bearing damage and belt tearing). This method utilizes the differences in the energy contribution of different faults in specific frequency bands to significantly improve the accuracy of fault classification. Combining vibration and temperature monitoring forms a multimodal collaborative analysis mechanism that can distinguish the different root causes of mechanical failures and thermal anomalies, enhancing the system's comprehensive judgment capabilities for complex fault scenarios.

[0143] Example 4

[0144] As the fourth embodiment of the present invention, when this application is specifically implemented, compared with the first, second and third embodiments, the technical solution of this embodiment is to combine and implement the solutions of the above-mentioned first, second and third embodiments.

[0145] This embodiment integrates vibration, temperature, and acoustic trimodal data to build a multi-dimensional collaborative monitoring system, leveraging the strengths of each sensor: vibration monitoring captures mechanical dynamic anomalies, temperature monitoring identifies thermal anomalies, and acoustic analysis locates the type of fault. The complementary nature of these three sensors effectively reduces the false alarm rate of a single sensor and covers the full range of characteristics of the conveyor belt's operating status. For example, it can simultaneously detect multiple concurrent faults, such as abnormal bearing vibration, localized belt overheating, and abnormal tail pulley noise. This integrated solution significantly improves the system's robustness and diagnostic comprehensiveness, providing highly reliable decision-making support for equipment health management.

[0146] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field according to actual conditions.

[0147] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A belt conveyor condition monitoring system based on multimodal data, characterized in that: include: A feature extraction unit is used to extract key feature values ​​from the condition monitoring data collected from the belt conveyor; The condition monitoring data contains acoustic wave signals, and the key eigenvalues ​​include the proportion of energy in different frequency bands to the total energy; The abnormality analysis unit is used to analyze the abnormal state of the belt conveyor based on the key feature values ​​extracted by the feature extraction unit, and determine the fault type of the belt conveyor based on the abnormal state analysis results.

2. The belt conveyor condition monitoring system based on multimodal data according to claim 1, characterized in that: The key feature value extraction method is as follows: Extract the acoustic wave signal of the tail wheel of the belt conveyor, then use short-time Fourier transform to analyze the time-frequency characteristics of the acoustic wave signal, and obtain the energy distribution of the sound signal at different times and frequencies, that is, obtain the time-frequency diagram; Extract the maximum and minimum values ​​of the frequency from the time-frequency graph, then form the entire frequency range based on the maximum and minimum values ​​of the frequency, and then divide it into several frequency segments. Then select a frequency segment and obtain the energy value of the frequency segment based on the time-frequency graph; Extract the total energy value of the entire frequency range from the time-frequency graph; The proportion of energy in different frequency bands to the total energy is obtained by dividing the energy value of each frequency band by the total energy value of the entire frequency range.

3. The belt conveyor condition monitoring system based on multimodal data according to claim 2, characterized in that: The abnormal status analysis method is as follows: A variety of belt conveyors with different fault types were selected, and their acoustic wave signals were measured experimentally. Subsequently, a feature extraction unit was used to extract the energy ratio of the belt conveyors with different fault types in different frequency bands to the total energy ratio. Then, sound comparison samples were generated for the belt conveyors with different fault types and their energy ratio in different frequency bands to the total energy ratio. An acoustic wave feature library was established based on the multiple sound comparison samples. Then, the proportion of the belt conveyor's energy in different frequency bands to the total energy is extracted, and the similarity is determined with the acoustic wave feature library to obtain the corresponding similarity value; Then, the similarity between the proportion of the energy of the belt conveyor in different frequency bands and the sound comparison samples in the sound wave feature library is compared with the corresponding preset similarity thresholds, and then the fault type corresponding to the sound comparison sample with the similarity ≥ the similarity threshold and the largest similarity value is selected as the fault type of the belt conveyor.

4. The belt conveyor condition monitoring system based on multimodal data according to claim 3 is characterized in that: The similarity determination method is: Select a sound comparison sample from the sound wave feature library; pass: Select the frequency segment with the largest energy proportion to the total energy from the sound comparison samples; At the same time through Select the frequency segment with the largest energy proportion to the total energy from the belt conveyor; Wherein, Ek represents the proportion of the energy of the belt conveyor in the kth frequency segment to the total energy, k = 1, 2, ... v, v represents the number of different frequency segments in the belt conveyor, EYg represents the proportion of the energy in the gth frequency segment of the selected sound comparison sample to the total energy, g = 1, 2, ... u, u represents the number of different frequency segments in the sound comparison sample; The two selected frequency bands are then compared: When the two frequency bands are consistent, the following is achieved: Calculate the similarity EC between the ratio of the energy of the belt conveyor in different frequency bands to the total energy and the sound comparison sample; By analogy, the proportion of the belt conveyor's energy in different frequency bands to the total energy and the similarity between the various sound comparison samples in the sound wave feature library are calculated.

5. The belt conveyor condition monitoring system based on multimodal data according to claim 4 is characterized in that: When the two frequency segments are inconsistent, the frequency segment with the largest energy ratio in the belt conveyor is selected, and the same frequency segment is selected from the sound comparison sample, and the energy ratio of the frequency segment to the total energy is obtained, and it is marked as M(E k ), followed by: Calculate the similarity EC between the proportion of the belt conveyor's energy in different frequency bands to the total energy and the sound comparison sample.

6. The belt conveyor condition monitoring system based on multimodal data according to claim 1, characterized in that: Condition monitoring data also includes amplitude data and belt surface temperature. Key characteristic values ​​also include vibration intensity growth rate and temperature change amplitude. The key feature value extraction method is as follows: At the current acquisition node, the amplitude data collected by the vibration sensor group is extracted. Then, the squares of the vibration amplitudes corresponding to the X-axis, Y-axis, and Z-axis are calculated respectively. Then, the square root of the sum is taken to obtain the modulus of the three-axis vibration composite vector. At the same time, the modulus of the three-axis vibration composite vector at the previous designated acquisition node is calculated. Then, the modulus of the three-axis vibration composite vector of the current acquisition node is subtracted from the modulus of the three-axis vibration composite vector of the previous designated acquisition node, and the result is divided by the modulus of the three-axis vibration composite vector of the previous designated acquisition node to obtain the vibration intensity growth rate; and extracting the belt surface temperature corresponding to the belt conveyor, and extracting the maximum and minimum belt surface temperatures therefrom; The belt surface temperature after normalization is obtained by subtracting the minimum belt surface temperature from the belt surface temperature corresponding to the belt conveyor and subtracting the minimum belt surface temperature from the maximum belt surface temperature. The previous difference result is divided by the next difference result and then multiplied by 100. Then, a normalized belt surface temperature is selected, and multiple normalized belt surface temperatures adjacent to the belt surface temperature are extracted; Calculate the absolute value of the temperature difference between the selected normalized belt surface temperature and its multiple adjacent normalized belt surface temperatures, then calculate the average of the multiple temperature difference absolute values, and record the average as the temperature change amplitude of the belt surface temperature at the i-th row and j-th column.

7. The belt conveyor condition monitoring system based on multimodal data according to claim 6, characterized in that: The abnormal state analysis method is as follows: Compare the vibration intensity increase rate of the belt conveyor drive roller bearing seat with the preset vibration intensity increase rate threshold: When the vibration intensity growth rate is greater than or equal to the vibration intensity growth rate threshold, it is determined that the vibration of the belt conveyor drive roller bearing seat is abnormal; When the vibration intensity growth rate is less than the vibration intensity growth rate threshold, it is determined that the vibration of the belt conveyor drive roller bearing seat is normal; Compare the temperature variation at each location on the belt surface with the preset temperature variation threshold: When the temperature variation amplitude of one position among various positions on the belt surface is greater than or equal to the temperature variation amplitude threshold, it is determined that the temperature of the belt surface is abnormal.

8. The belt conveyor condition monitoring system based on multimodal data according to claim 7, characterized in that: When the temperature variation amplitude at all positions on the belt surface is less than the temperature variation amplitude threshold, the belt surface is determined to undergo secondary temperature analysis; The secondary temperature analysis method is as follows: Extract all belt surface temperatures T1 after normalization i，j , then calculate the surface temperature T1 of all belts i，j The average value is then compared to a preset belt temperature threshold: If the average value of all belt surface temperatures is ≥ the belt temperature threshold, the belt surface temperature is determined to be abnormal; If the average value of all belt surface temperatures is less than the belt temperature threshold, it is determined that the belt surface temperature is normal.

9. The belt conveyor condition monitoring system based on multimodal data according to claim 6, characterized in that: The amplitude data is collected by a vibration sensor group installed on the bearing seat of the belt conveyor drive roller. The vibration sensor group uses a three-axis accelerometer to measure the vibration amplitude corresponding to the X-axis, Y-axis, and Z-axis. The belt surface temperature is collected by a temperature sensor group installed under the belt load section; wherein the temperature sensor group is an infrared thermal array temperature sensor that collects multiple belt surface temperatures; The acoustic wave signal is collected by an acoustic sensor installed at the tail wheel of the belt conveyor; wherein the acoustic sensor adopts a broadband microphone and collects the acoustic wave signal.

10. The belt conveyor condition monitoring system based on multimodal data according to claim 9, characterized in that: Also includes: Data acquisition unit, used to collect status monitoring data of the belt conveyor; The fault display unit is used to display the results of the abnormality analysis unit to relevant management personnel.

Images