Bulk material conveying belt wear monitoring system fusing digital twinning

Abstract

The application provides a kind of fusion digital twin's bulk conveying belt wear monitoring system, it is related to data processing technical field, the system includes: sensing data acquisition module, for collecting native data in the running process of conveying belt in multiple monitoring positions along the line;Data preprocessing module is used for formatting, noise reduction and calibration processing to native data;Digital twin construction module is used for dynamically generating the virtual model of conveying belt, and data is mapped to the corresponding area of preset virtual model according to space and time dimension;Abnormal identification module is used for detecting the wear characteristics of each monitoring area of belt, and identifying and positioning the local area with abnormal change trend;Early warning trigger module is used for generating early warning information containing abnormal position and wear degree, and output to host computer or operation and maintenance terminal;The application improves the accuracy of conveying belt wear monitoring system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a bulk material conveying belt wear monitoring system fusing digital twinning. BACKGROUND

[0002] In the prior art, the wear monitoring of the bulk material conveying belt mainly relies on the combination of sensor arrays and regular manual inspection. Usually, strain sensors, temperature sensors and vibration sensors are arranged at key positions of the belt, and the wear condition of the belt surface or the inside is judged by real-time acquisition of physical signals. At the same time, the monitoring system will statistically analyze the sensor data to form a wear trend report, supplemented by manual inspection verification. Some systems also introduce image acquisition and processing modules to capture belt surface images through cameras and perform image recognition, thereby improving the accuracy of detection.

[0003] In actual application, when the belt conveys large-particle and hard bulk materials, due to the limitation of the number and arrangement position of sensors, the existing monitoring system often has difficulty in timely and comprehensively capturing early signals of local severe wear. Taking the conveying of mine raw materials as an example, due to frequent material particle impact, uneven wear is prone to occur at the edge or joint position of the belt. However, due to the existence of monitoring blind area, the existing system is prone to miss abnormal wear, which increases the risk of equipment operation. In this scenario, the system has the technical defect of slow response to local sudden wear, which affects subsequent maintenance and operation safety. SUMMARY

[0004] The purpose of the present application is to provide a bulk material conveying belt wear monitoring system fusing digital twinning, which aims to solve the problems mentioned in the background.

[0005] To solve the above technical problems, the technical scheme of the present application is as follows:

[0006] A bulk material conveying belt wear monitoring system fusing digital twinning, the system comprises:

[0007] A sensor data acquisition module for acquiring raw data in the conveying belt running process at multiple monitoring positions along the conveying belt, the raw data including physical signal data and image data signals;

[0008] A data preprocessing module for formatting, denoising and calibrating the raw data to obtain analysis data;

[0009] A digital twinning construction module for dynamically generating a virtual model of the conveying belt according to the analysis data, mapping the analysis data to the corresponding regions of the preset virtual model in space and time dimensions to obtain a set of belt multi-region state parameters;

[0010] An anomaly identification module is configured to perform partition detection on the wear characteristics of each monitoring area of the belt according to a belt multi-area state parameter set, identify and locate a local area with an abnormal change trend, and obtain an anomaly identification result.

[0011] A warning trigger module is configured to generate warning information containing an abnormal position and a wear degree according to the anomaly identification result, and output the warning information to an upper computer or an operation and maintenance terminal.

[0012] Preferably, the digital twin construction module comprises:

[0013] A virtual model construction submodule is configured to generate a virtual model corresponding to the conveying belt structure according to the monitoring positions in the analysis data and the conveying belt structure information, and initialize the attributes of each spatial node.

[0014] A data mapping submodule is configured to map the analysis data to corresponding spatial nodes and time indexes in the virtual model according to a preset spatial partition and time period, update the original data sequence of each spatial node in different time periods, and the spatial partition is a grouping of the monitoring positions arranged along the belt, the time period is a grouping of the analysis data collection time, and the time index is a numbering of the time period.

[0015] A parameter extraction submodule is configured to calculate the maximum value, the minimum value, the average value and the change rate of each spatial node through its original data sequence, and generate a state parameter set of the spatial node corresponding to the spatial position and the time index of the spatial node.

[0016] A parameter sequence combination submodule is configured to serialize and arrange the state parameter sets of all spatial nodes in each time period according to the order of the spatial partition and the time index, generate a multi-dimensional parameter matrix, and perform multi-dimensional statistics on various state parameter data in the multi-dimensional parameter matrix to generate a belt multi-area state parameter set.

[0017] Preferably, the anomaly identification module comprises:

[0018] A parameter sequence analysis submodule is configured to perform step-by-step analysis on the state parameter sequences of each spatial partition in different time periods according to the belt multi-area state parameter set, calculate the parameter change rate of adjacent time points, compare the parameter change rate with a preset abnormal threshold, and generate an abnormal state parameter candidate set.

[0019] A clustering identification submodule is configured to cluster the abnormal state parameter candidate set according to spatial proximity and change characteristics, identify spatial partition areas with similar abnormal characteristics, and generate a clustered abnormal partition set.

[0020] An anomaly positioning submodule is configured to calibrate the abnormal occurrence position and the time period according to the clustered abnormal partition set in combination with the spatial partition and the time period information, and form an anomaly identification result.

[0021] Preferably, the parameter extraction submodule includes:

[0022] An extreme value calculation unit is used to determine the maximum and minimum values ​​of all sampled values ​​according to the original data sequence of each spatial node in each time period, and output the maximum and minimum values ​​as extreme value parameters corresponding to the spatial position and time index of the spatial node respectively;

[0023] The mean calculation unit is used to accumulate all the sampled values ​​in the original data sequence of each spatial node in each time period and divide them by the number of samples to obtain an average value parameter. The average value is output in a one-to-one correspondence with the spatial position and time index of the spatial node;

[0024] A change rate calculation unit is used to calculate the difference of all adjacent sampling values ​​in sequence according to the original data sequence of each spatial node in each time period to obtain a difference sequence, and divide the sum of the absolute values ​​of the difference sequence by the number of differences to obtain a change rate parameter. The change rate parameter is output in a one-to-one correspondence with the spatial position and time index of the spatial node;

[0025] The parameter aggregation unit is used to uniformly aggregate extreme value parameters, average value parameters and change rate parameters according to the spatial position and time index of each spatial node to form a state parameter set for each spatial node and each time period.

[0026] Preferably, the parameter sequence combination submodule includes:

[0027] The sequence arrangement unit is used to arrange the state parameter sets of all spatial nodes in each time period in sequence according to the spatial partition position order and the increasing order of time index to generate a multi-dimensional parameter matrix;

[0028] The multidimensional statistical unit is used to perform statistics on various state parameter data in the multidimensional parameter matrix according to the following steps:

[0029] For each spatial partition, according to the time index order, for the same state parameters, calculate the mean, range, and maximum continuous increasing interval length of each time period to generate the partition time series statistics;

[0030] For the same state parameters of all spatial partitions under each time index, calculate the spatial maximum, minimum, mean and standard deviation at that time point and generate the corresponding spatial statistical results;

[0031] The parameter integration unit is used to merge the time series statistical results of all spatial partitions and the spatial statistical results of each time period according to the two dimensions of spatial partition and time, and form a multi-region belt state parameter set indexed by spatial partition and time.

[0032] Preferably, the parameter sequence analysis submodule comprises:

[0033] a rate of change calculation unit configured to, for the state parameter sequence of each spatial partition, sequentially calculate the difference value of the state parameter at adjacent time indexes, and divide the sum of the absolute values of all the difference values by the number of the difference values to obtain the rate of change sequence of the spatial partition on the state parameter;

[0034] a threshold comparison unit configured to compare each rate of change value in the rate of change sequence with a preset abnormality determination threshold, and if the rate of change value is greater than the abnormality determination threshold, mark the corresponding spatial partition, time index and state parameter as an abnormality candidate to generate an abnormal state parameter candidate set.

[0035] Preferably, the clustering identification submodule comprises:

[0036] a spatial clustering unit configured to generate an initial spatial clustering cluster set by taking the abnormal state parameters adjacent in spatial position as the same initial clustering cluster according to the abnormal state parameter candidate set;

[0037] a feature similarity discrimination unit configured to, according to the initial spatial clustering cluster set, classify the abnormal points with the same abnormal state parameter type and change direction in each initial spatial clustering cluster according to the state parameter type and change feature to obtain feature consistency subclusters;

[0038] a clustering merging unit configured to judge the spatial distance and parameter change amplitude between the feature consistency subclusters, and if the spatial distance is less than a preset spatial distance threshold and the change amplitude is less than a preset absolute difference threshold, merge the feature consistency subclusters satisfying the conditions into a clustering partition to form a clustering abnormal partition set.

[0039] Preferably, the abnormality positioning submodule comprises:

[0040] an abnormal interval determination unit configured to extract the spatial partition range and corresponding time range of each clustering partition according to the clustering abnormal partition set to generate abnormal interval information;

[0041] an abnormal type determination unit configured to extract the corresponding state parameter according to the abnormal interval information, analyze the change feature thereof over time, determine the state parameter as persistent wear if it continuously remains abnormal in the entire time range, determine the state parameter as sudden wear if it suddenly changes at individual time points, determine the state parameter as periodic abnormality if the change presents regular fluctuations, and generate abnormal type information;

[0042] an abnormality recognition result generation unit configured to generate and output the abnormality recognition result according to the spatial partition range, time range and abnormal type information of each abnormal interval.

[0043] Preferably, the feature similarity discrimination unit comprises:

[0044] a parameter type grouping unit, configured to classify the abnormal state parameters in each cluster according to parameter types to form parameter type groups according to the initial spatial clustering clusters;

[0045] a change trend analysis unit, configured to determine whether the change directions of the abnormal state parameters in each parameter type group are consistent according to the parameter value changes of the abnormal state parameters in adjacent time periods, and if the change directions are consistent, the abnormal state parameters are classified into the same trend group, and if the change directions are inconsistent, the abnormal state parameters with different change directions are respectively classified into new trend groups;

[0046] a magnitude consistency screening unit, configured to calculate the change magnitudes of the abnormal state parameters in each trend group, and compare the change magnitudes with the average change magnitude in the group, if the difference between the change magnitude of a parameter and the average change magnitude in the group is less than a preset magnitude difference threshold, the parameter is retained, otherwise, the parameter is removed, and finally, a feature consistency sub-cluster is generated.

[0047] Preferably, the cluster merging unit comprises:

[0048] a spatial distance determination unit, configured to calculate the distances between the center spatial partitions of the feature consistency sub-clusters, and if the distance is less than a spatial distance threshold, the corresponding sub-cluster is classified as a mergable object;

[0049] a change magnitude determination unit, configured to calculate the average parameter change magnitudes of the sub-clusters in each group of mergable objects, and then calculate the absolute difference values of the average change magnitudes between the groups, and if the absolute difference value is less than an absolute difference threshold, it is determined that the corresponding sub-clusters of the group meet the merging condition;

[0050] a merging generation unit, configured to merge the feature consistency sub-clusters that meet the spatial distance threshold and the absolute difference threshold into one cluster partition, and a set of all the cluster partitions constitutes a cluster anomaly partition set

[0051] The above scheme of the present application at least has the following beneficial effects:

[0052] The application greatly expands the monitoring coverage by simultaneously collecting physical signal data and image data signals at multiple monitoring positions along the conveying belt, compared with the limited number of sensors and the monitoring blind area existing in manual inspection in the prior art, and can more carefully capture the wear state of each area of the belt, especially the damaged positions such as the edges and joints. The data preprocessing module formats, denoises and calibrates the original data, effectively eliminates the interference caused by the external environment or abnormal data collection, and improves the accuracy and reliability of wear identification. Through the digital twin construction module, the system can dynamically generate a virtual model based on the analysis data, realize the spatial and temporal multi-dimensional mapping of the actual belt running state, and then obtain a set of belt multi-region state parameters, thereby significantly improving the early perception ability of the belt wear change. The abnormality identification module can not only automatically partition and detect each monitoring area of the whole line, but also can accurately identify and locate the abnormal change trend of the local area, so as to timely find the local sudden wear caused by material impact, uneven particle distribution and the like. The early warning triggering module further outputs the abnormal position and wear degree information to the upper computer or operation and maintenance terminal in real time, so that the operation and maintenance personnel can respond and maintain at the first time, and effectively reduce the equipment failure risk caused by monitoring blind area and slow response. In typical application scenarios such as mine raw ore and heavy load bulk material, the application can significantly improve the identification efficiency and monitoring accuracy of early local wear, and provides a strong technical guarantee for the long-period safe operation of the bulk material conveying belt. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 is an architecture diagram of a bulk material conveying belt wear monitoring system fused with digital twin provided by an embodiment of the application. DETAILED DESCRIPTION

[0054] Exemplary embodiments of the present disclosure will be described in greater detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be accurately conveyed to those skilled in the art.

[0055] As Figure 1 shown, an embodiment of the application proposes a bulk material conveying belt wear monitoring system fused with digital twin, which comprises:

[0056] A sensor data acquisition module is configured to acquire original data in the conveying belt running process at multiple monitoring positions along the conveying belt, wherein the original data comprises physical signal data and image data signals.

[0057] A data preprocessing module is configured to format, denoise and calibrate the original data to obtain analysis data.

[0058] a digital twin construction module configured to dynamically generate a virtual model of the conveying belt according to the analysis data, map the analysis data to a preset corresponding region of the virtual model in terms of spatial and temporal dimensions, and obtain a plurality of regional state parameter sets of the belt;

[0059] an abnormality identification module configured to perform partition detection on the wear characteristics of each monitoring region of the belt according to the plurality of regional state parameter sets of the belt, identify and locate a local region having an abnormal change trend, and obtain an abnormality identification result;

[0060] a warning triggering module configured to generate warning information containing an abnormal position and a wear degree according to the abnormality identification result, and output the warning information to an upper computer or an operation and maintenance terminal.

[0061] In the embodiment of the present application, the system can collect physical signal data and image data signals in real time at a plurality of monitoring positions along the conveying belt through the sensing data acquisition module, covering the multi-dimensional state of the belt during operation. After the format, noise reduction and calibration processing of the data preprocessing module, the noise and abnormal points in the original data are effectively eliminated, and the accuracy and robustness of subsequent data analysis are improved. The digital twin construction module dynamically generates a virtual model of the conveying belt according to the analysis data, realizes real-time mapping and simulation of the actual conveying belt running state, and can fully reflect the physical state of each monitoring region. Through the mapping of the analysis data in terms of spatial and temporal dimensions, the system can generate a fine-grained plurality of regional state parameter sets of the belt, ensuring that the running characteristics of each spatial partition and each time point can be recorded and tracked. The abnormality identification module can realize accurate partition detection of the wear characteristics of each monitoring region of the belt based on the above state parameter sets, and can timely identify and locate the abnormal change trend of the local region, so as to discover the running risk of the belt in advance. Finally, the warning triggering module generates warning information containing the abnormal position and the wear degree according to the abnormality identification result, and can flexibly output the warning information to the upper computer or the operation and maintenance terminal, so that the operation and maintenance personnel can obtain the abnormal situation in the first time and take measures. For example, in an actual application scenario, if the state parameter of a monitoring position suddenly changes in a short time, the system can quickly determine that the region is a potential wear risk point and timely issue a warning to ensure the safe and stable operation of the conveying equipment.

[0062] In the present application, the digital twin not only includes virtual modeling of the conveying belt structure, but also dynamically synchronizes the analysis data to the virtual model through continuous collection and mapping of actual monitoring data, realizes real-time linkage between the physical object and the virtual space. The virtual model as the core component of the digital twin continuously updates the attributes with the change of data, realizes accurate dynamic mapping of the running state of the conveying belt in the whole life cycle, and thus constructs the digital twin system of the conveying belt.

[0063] The sensing data collection module is configured to collect raw data of the conveying belt during operation at a plurality of monitoring positions along the conveying belt, and specifically includes:

[0064] Strain sensors, temperature sensors, vibration sensors, and industrial cameras are installed at different spatial positions of the conveying belt (such as the edge, middle, and joint of the belt) to collect the mechanical state, temperature change, vibration condition, and surface image of the belt in real time or periodically. Each sensor can be connected to the main controller through a data acquisition card or a data bus. Physical signal data such as strain value, temperature value, and vibration amplitude are collected at a set sampling frequency and assigned with a timestamp and a spatial number to realize the correspondence between the collected data and the specific monitoring position and collection time. Image data signals are collected by high-resolution cameras to obtain panoramic or local detail pictures of the belt surface, and the image files are labeled with spatial position and collection time for subsequent analysis and tracing. For example, in the scenario of mine bulk material conveying, the system can continuously collect and store the physical quantities of key positions and the corresponding images at the time to ensure comprehensive coverage of the entire belt operation state.

[0065] The data preprocessing module is configured to preprocess the raw data to obtain analysis data, and specifically includes:

[0066] After receiving the raw data output by the sensing data collection module, the physical signal data is first standardized in terms of data type, the sensor data from different sources is unified into a specified format, and the data is sorted and archived according to the monitoring position and collection time. For physical signal data, sliding average or median filtering methods are applied to remove fluctuation noise caused by environmental interference or transient anomalies, ensuring data smoothness and reliability. For possible abnormal values or missing data, outlier detection algorithms are used to automatically remove outliers, and interpolation methods are used to fill in missing data segments. For image data signals, format conversion and size normalization are performed to unify the resolution, and image denoising algorithms (such as bilateral filtering) are used to remove noise during shooting. All preprocessed data is associated with the original collection time and spatial position to form a structured analysis data set. For example, during continuous collection, if an abnormal high value appears in a strain sensor, the system automatically identifies and removes the value by comparing adjacent time data, and fills in the value with the mean value of adjacent valid data, finally ensuring the effectiveness of downstream analysis.

[0067] The warning triggering module is configured to generate warning information containing the abnormal position and the degree of wear according to the abnormal identification result, and output the warning information to the upper computer or the operation and maintenance terminal, and specifically includes:

[0068] After receiving the abnormality identification result output by the abnormality identification module, the system first extracts the spatial partition number, time range and wear characteristic parameter of each abnormal region. According to the position of the abnormality, the type of the abnormality and the degree of wear, the abnormal event is divided into general warning, serious warning and other levels according to the set warning grading standard, and the corresponding warning information is generated. The warning information content includes the spatial position of the abnormality, the start and end time, the abnormality type description, the wear parameter (such as the maximum value, the change rate, etc.) and the risk suggestion, etc. The system can automatically judge whether it needs to alarm immediately (such as reaching the set threshold or the abnormal duration exceeding the limit), and push the warning information to the upper computer, the control center or the mobile terminal of the operation and maintenance personnel through wired or wireless network, so as to realize the real-time notification and response of the abnormal situation. For example, if the wear change rate of a certain section exceeds the warning line significantly in continuous multiple time periods, the system will automatically generate detailed warning information including the abnormal partition, the time range and the wear parameter, and prompt the relevant operation and maintenance personnel to check and maintain the area in time.

[0069] In a preferred embodiment of the present application, the digital twin construction module comprises:

[0070] A virtual model construction submodule is configured to generate a virtual model corresponding to the conveying belt structure according to the monitoring positions in the analysis data and the conveying belt structure information, and initialize attributes of each spatial node.

[0071] A data mapping submodule is configured to map the analysis data to corresponding spatial nodes and time indexes in the virtual model according to preset spatial partitions and time periods, update original data sequences of each spatial node in different time periods, wherein the spatial partitions are groupings of the monitoring positions arranged along the belt, the time periods are groupings of analysis data collection times, and the time indexes are numbers of the time periods.

[0072] A parameter extraction submodule is configured to calculate maximum values, minimum values, average values and change rates from the original data sequences of each spatial node, and generate a state parameter set of the spatial node corresponding to the spatial position and the time index of the spatial node.

[0073] A parameter sequence combination submodule is configured to serialize and arrange the state parameter sets of all spatial nodes in each time period according to the order of the spatial partitions and the time indexes, generate a multi-dimensional parameter matrix, and perform multi-dimensional statistics on various state parameter data in the multi-dimensional parameter matrix to generate a belt multi-region state parameter set.

[0074] In the embodiment of the present application, the digital twin construction module can construct a virtual model corresponding to the actual conveying belt structure based on the monitoring positions in the analysis data and the conveying belt structure information, and make the properties of each spatial node consistent with the actual monitoring points through initialization operation. Through the data mapping submodule, the analysis data is mapped to the corresponding spatial nodes and time indexes in the virtual model according to the preset spatial partition and time period, so that each spatial node can obtain a complete original data sequence in different time periods, realizing full coverage and dynamic update in space-time dimension. The parameter extraction submodule can perform multi-dimensional statistical analysis such as extreme value, mean value and change rate on the original data sequence of each spatial node, thereby obtaining a state parameter set that can accurately reflect the running state of the conveying belt. The parameter sequence combination submodule further sequentially arranges the state parameter set of all spatial nodes in each time period and generates a multi-dimensional parameter matrix, and extracts the running and wear characteristic parameters of the belt in multiple regions through multi-dimensional statistics. In this way, the system can not only realize traceability analysis of the historical and current state of each monitoring point, but also support subsequent anomaly identification, prediction and decision-making. For example, in the actual running process, if the average value of a certain spatial partition decreases gradually and the change rate increases in multiple consecutive time periods, the system can capture this abnormal trend in time through parameter matrix analysis, thereby providing a solid data foundation for subsequent anomaly identification and risk warning.

[0075] The virtual model construction submodule is configured to generate a virtual model corresponding to the conveying belt structure according to the monitoring positions in the analysis data and the conveying belt structure information, and initialize the properties of each spatial node, and specifically includes:

[0076] Firstly, according to the monitoring position data on site, combined with the basic design parameters of the conveying belt such as structure size, trend, segmentation, connection mode, etc., the conveying belt skeleton model in virtual space is constructed. For example, a three-dimensional modeling tool or a self-developed digital modeling program can be used to generate model nodes and connections in virtual space one by one for each key part of the physical belt (such as belt section, turning point, connection joint, etc.). For each virtual model node, the system is positioned according to the spatial coordinates of the actual monitoring position, and the node is given the same attribute parameters as the actual physical position, including spatial coordinates, number, node type (such as edge, center, joint), initialization time, etc. For the entire belt structure, all nodes can be connected in turn according to the actual arrangement order, and the distance, angle and other attributes between nodes can be set according to the design parameters. If the belt is divided into multiple spatial partitions or has different structural characteristics, the system can assign exclusive attribute sets to different partitions. After modeling, each node in the virtual model has a unique number and spatial coordinates, supporting subsequent dynamic data mapping and real-time attribute updating. For example, in a 500-meter-long belt along the line, every 5 meters is equipped with a monitoring position, and the virtual model includes 100 spatial nodes, each of which has attributes corresponding to the actual monitoring point on site.

[0077] The data mapping submodule is configured to map the analysis data to the corresponding spatial nodes and time indexes in the virtual model according to the preset spatial partitions and time periods, and update the original data sequence of each spatial node in different time periods. Specifically, the data mapping submodule comprises the following steps:

[0078] The system first groups all the actual monitoring positions according to the preset spatial partition rules, for example, a number of consecutive monitoring points along the belt direction are divided into the same spatial partition; the nodes in each partition are numbered in physical position order. For analysis data, the system divides all data into multiple time periods according to the set grouping interval of collection time (such as every 1 minute, every 10 minutes, or custom according to process requirements), and assigns a unique time index number to each time period. Then, the system iterates through each spatial partition and each time period, and assigns or links all the original physical signal data and image data signals collected by the monitoring nodes in the spatial partition within the corresponding time period to the corresponding spatial nodes and time indexes in the virtual model, forming the original data sequence. For nodes and time periods with missing or abnormal data, interpolation or historical average value can be used to fill in the data to ensure that each node has complete original data sequence records for each time period. For example, in a day of belt operation, a time period is divided every 10 minutes, then each spatial node will correspond to 144 time indexes, and the system will automatically fill the analysis data collected in the 144 time periods into the virtual model nodes in turn, laying a foundation for subsequent parameter extraction and anomaly analysis.

[0079] In a preferred embodiment of the present application, the anomaly identification module comprises:

[0080] a parameter sequence analysis submodule, configured to perform step-by-step analysis on the state parameter sequences of each spatial partition in different time periods according to the belt multi-region state parameter set, calculate the parameter change rate at adjacent time points, and compare the parameter change rate with a preset anomaly threshold to generate a candidate set of abnormal state parameters;

[0081] a clustering identification submodule, configured to cluster the candidate set of abnormal state parameters according to spatial proximity and change characteristics, identify spatial partition regions with similar abnormal characteristics, and generate a clustered abnormal partition set;

[0082] an anomaly positioning submodule, configured to determine the location and time period of the anomaly occurrence according to the clustered abnormal partition set in combination with the spatial partition and time period information, and form an anomaly identification result.

[0083] In the embodiments of the present application, the anomaly identification module performs step-by-step analysis on the state parameter sequences of each spatial partition in different time periods in the belt multi-region state parameter set through the parameter sequence analysis submodule, and sequentially calculates the parameter change rate at adjacent time points, thereby effectively identifying parameter fluctuation anomalies and trend changes. Further, the clustering identification submodule can cluster the candidate set of abnormal state parameters in the spatial and characteristic dimensions, and group spatial partition regions with similar abnormal characteristics into the same group, thereby significantly improving the accuracy and pertinence of anomaly identification. After obtaining the clustered abnormal partition set, the anomaly positioning submodule can accurately determine the location and time period of the anomaly occurrence in combination with the spatial partition and time period information, thereby ensuring that the anomaly information is positioned to a specific monitoring region and time period. For example, in an actual application scenario, if a parameter change rate anomaly continuously occurs in a region for a period of time, the module can automatically identify the region as an abnormal partition and determine the specific location, and finally form an anomaly identification result, thereby providing a scientific basis for subsequent early warning and maintenance, and effectively reducing the risk of failure and maintenance blind spots caused by wear.

[0084] In a preferred embodiment of the present application, the parameter extraction submodule comprises:

[0085] an extreme value calculation unit, configured to determine the maximum value and the minimum value in all sampling values according to the original data sequence of each spatial node in each time period, and output the maximum value and the minimum value as extreme value parameters corresponding to the spatial position and the time index of the spatial node, respectively;

[0086] a mean value calculation unit, configured to accumulate all sampling values in the original data sequence of each spatial node in each time period, and divide the sum by the number of samples to obtain an average value parameter, wherein the average value is output corresponding to the spatial position and the time index of the spatial node;

[0087] a change rate calculation unit configured to sequentially calculate differences between adjacent sampling values according to the original data sequence of each spatial node in each time period, obtain a difference sequence, and divide a sum of absolute values of the difference sequence by a number of differences to obtain a change rate parameter, which is one-to-one corresponding to a spatial position and a time index of the spatial node and is output;

[0088] a parameter collection unit configured to uniformly collect the extreme value parameter, the average value parameter and the change rate parameter according to the spatial position and the time index of each spatial node to form a state parameter set of each spatial node in each time period.

[0089] In the embodiment of the present application, the parameter extraction sub-module can perform multi-dimensional feature extraction on the original data sequence of each spatial node in each time period through the extreme value calculation unit, the average value calculation unit and the change rate calculation unit. The system can not only obtain the maximum value and the minimum value to reflect the running limit, but also obtain the average value to reflect the overall running level, and identify the dynamic trend of data change through the change rate index. The parameter collection unit uniformly collects the above parameter results according to the spatial position and the time index of each spatial node, effectively ensuring that all feature data of each spatial node in each time period can be completely retained and subsequently traced. This technical solution can digitize and standardize the running state of each monitoring point in any time period, greatly improving the efficiency and accuracy of subsequent statistics, comparison and analysis. For example, when the maximum value of a node along a certain conveying belt continuously decreases and the change rate suddenly increases, the system can directly locate the node and the specific time period, and the operation and maintenance personnel can focus on and take maintenance measures in advance to reduce the risk of equipment failure.

[0090] In a preferred embodiment of the present application, the parameter sequence combination sub-module comprises:

[0091] a sequence arrangement unit configured to arrange the state parameter set of all spatial nodes in each time period in order according to the spatial partition position sequence and the time index increasing sequence to generate a multi-dimensional parameter matrix;

[0092] a multi-dimensional statistical unit configured to statistically analyze each type of state parameter data in the multi-dimensional parameter matrix according to the following steps:

[0093] for each spatial partition, calculating the mean value, the range, and the maximum continuous increasing interval length of each time period for the same type of state parameter according to the time index sequence to generate a partition time sequence statistical result;

[0094] for the same type of state parameter of all spatial partitions at each time index, calculating the spatial maximum value, the minimum value, the mean value and the standard deviation at the time point to generate a corresponding spatial statistical result;

[0095] A parameter integration unit is configured to merge the time sequence statistical results of all spatial partitions and the spatial statistical results in each time period according to the two dimensions of spatial partition and time, to form a belt multi-region state parameter set indexed by spatial partition and time.

[0096] In the embodiment of the application, the parameter sequence combination sub-module first arranges the state parameter sets of all spatial nodes in each time period in order of spatial partition position and increasing order of time index by the sequence arrangement unit, to generate a multi-dimensional parameter matrix, so as to realize systematic expression of the space and time of the entire conveying belt running state. The multi-dimensional statistical unit can perform mean, range, maximum continuous increasing interval length and other statistical analysis on the state parameters of each spatial partition in different time periods, to further mine the change characteristics of the partition in the running process; meanwhile, the maximum value, minimum value, mean value and standard deviation of the spatial distribution of the state parameters of all spatial partitions in each time period can be calculated, to realize horizontal comparison and global fluctuation evaluation. The parameter integration unit merges the above statistical results according to the two dimensions of spatial partition and time, to finally form a belt multi-region state parameter set indexed by spatial partition and time, to provide structured and quantifiable multi-dimensional basic data for subsequent anomaly recognition and trend judgment. For example, the operation and maintenance personnel can discover that the range anomaly of a region is greatly enlarged or the standard deviation is significantly increased in a period of time through the parameter set, to discover the potential abnormal evolution trend of the region in advance.

[0097] The sequence arrangement unit is configured to arrange the state parameter sets of all spatial nodes in each time period in order of spatial partition position and increasing order of time index, to generate a multi-dimensional parameter matrix, and specifically includes:

[0098] Firstly, the system determines the physical location order of all space nodes according to the virtual model and the setting of spatial partitioning, such as the actual layout sequence number from the belt starting point to the end point. Then, the state parameters (such as maximum value, minimum value, average value, change rate, etc.) extracted by each node in different time periods are summarized. For each time period, the system arranges the state parameters of all nodes in the same time period in order according to the location order of the space nodes, forming a row of data. Multiple rows of data correspond to multiple consecutive time periods, so these data rows are spliced in order of increasing time index, and finally a parameter matrix is generated in the space and time dimensions. Each column of the matrix corresponds to a space node, each row corresponds to a time period, and each matrix element is a certain state parameter of a specific space node in a specific time period. In this way, the running state of the entire conveying belt in a period of time is structuredly mapped to a multi-dimensional data table, which is convenient for subsequent batch processing and analysis. For example, if a belt has 20 space nodes and the analysis period is 100 time periods, the system can generate a 20-column x 100-row parameter matrix, providing an efficient data organization form for big data analysis and trend identification.

[0099] Among them, the multi-dimensional statistical unit is used to statistically analyze various state parameter data in the multi-dimensional parameter matrix according to the following steps: for each spatial partition, according to the time index order, the mean, range, maximum continuous increasing interval length of the same type of state parameter in each time period are calculated, and the partition time series statistical result is generated; for the same type of state parameter of all spatial partitions at each time index, the spatial maximum value, minimum value, mean value and standard deviation at this time point are calculated, and the corresponding spatial statistical result is generated, which specifically includes:

[0100] For the partition time series statistics, the system traverses the same type of state parameters (such as the average value sequence of all nodes) in all time periods for each spatial partition. For each type of parameter, the system first calculates the value of each time period in the partition to obtain the average value of all time periods. Then, the system compares the maximum value and the minimum value of the parameters in all time periods, and the difference between the two is the range. For the maximum continuous increasing interval length, the system traverses the state parameter time series of each partition from the beginning, records the length of each continuous increasing interval, and selects the longest interval length as the maximum increasing interval of the parameter of the partition. In terms of spatial statistics, for all spatial partitions, the system compares the maximum value and the minimum value of the same type of state parameters in each time period, calculates the average value of the parameter of all partitions, and finally obtains the standard deviation by calculating the square of the difference between each partition parameter and the average value and taking the average. This statistical method can reveal the concentration trend, extreme fluctuation and overall dispersion degree of the state parameters in different spaces and different times. For example, in a certain statistics, if the average value of a partition is continuously increasing in multiple time periods and the range is large, the system can determine that the wear of the partition is increasing; if the standard deviation of the parameters of multiple spatial partitions at the same time is large, it indicates that the overall running state is discrete and needs to be paid attention to by the operation and maintenance.

[0101] The parameter integration unit is configured to merge the time series statistics results of all spatial partitions and the spatial statistics results in each time period according to the space partition and time, to form a belt multi-region state parameter set indexed by space partition and time, and specifically includes:

[0102] The system reorganizes the time series statistics results of each partition output by the multi-dimensional statistics unit (such as the average value, range, maximum increasing interval length of each partition in all time periods) and the spatial statistics results in each time period (such as the maximum value, minimum value, average value and standard deviation of each partition at the moment) according to the partition number and time index. For each spatial partition, the system arranges all time series statistics parameters related to the partition with the time index as the horizontal coordinate to form a multi-dimensional time series feature of the partition. For each time period, the system arranges all spatial statistics parameters with the spatial partition as the horizontal coordinate to form the spatial distribution feature at the moment. Finally, all these structured parameter sets are merged into a belt multi-region state parameter set with “spatial partition-time period” as the two-dimensional index. This data set not only provides a data basis for subsequent anomaly detection, trend analysis and other processes, but also facilitates export for further analysis by external systems or experts. For example, operation and maintenance engineers can quickly trace the historical state of any partition at any moment according to the parameter set, or compare the running differences of multiple partitions in the same time period, to improve the scientificity and accuracy of wear monitoring.

[0103] In a preferred embodiment of the present application, the parameter sequence analysis submodule comprises:

[0104] a change rate calculation unit configured to calculate, for each state parameter sequence of each spatial partition, a difference value of the state parameter at adjacent time indexes, and divide a sum of absolute values of all the difference values by a number of the difference values to obtain a change rate sequence of the spatial partition on the state parameter;

[0105] a threshold comparison unit configured to compare each change rate value in the change rate sequence with a preset abnormality determination threshold, and if the change rate value is greater than the abnormality determination threshold, mark a corresponding spatial partition, time index and state parameter as an abnormality candidate to generate an abnormal state parameter candidate set.

[0106] In the embodiments of the present application, the parameter sequence analysis submodule analyzes the state parameter sequence of each spatial partition point by point through the change rate calculation unit, calculates the parameter difference value of adjacent time points, and obtains the change rate sequence, thereby realizing continuous monitoring and sensitive capture of parameter changes in the time dimension. The threshold comparison unit strictly compares each change rate with the preset abnormality determination threshold, and once the change rate is found to be greater than the threshold, it is automatically marked as an abnormality candidate, effectively reducing the subjectivity and omission risk of manual discrimination. This automatic analysis and screening method can quickly and accurately screen out abnormal trends or mutations hidden in massive monitoring data, greatly improving the intelligence and real-time performance of large-scale conveying system wear monitoring. For example, in the operation of a conveying belt, if the change rate of a certain spatial partition exceeds the set threshold continuously for multiple times, the system can immediately mark the parameters of this partition in the relevant time period as an abnormality candidate, and the operation and maintenance personnel can intervene and investigate in a timely manner, thereby effectively avoiding significant production stoppage caused by wear deterioration.

[0107] The preset abnormality determination threshold is a reference standard for determining whether the state parameter change rate constitutes an abnormality, and specifically includes:

[0108] The system can set one or more numerical thresholds for abnormality discrimination through historical operation data statistics, expert experience or on-site working conditions before actual application. Usually, engineers will analyze a large amount of normal operation and known abnormal case data, and statistically analyze the maximum fluctuation amplitude, mean value and standard deviation of the change rate of each state parameter under normal conditions. According to the statistical results, a reasonable judgment threshold is set, for example, the mean value of the normal change rate plus twice the standard deviation can be selected as the criterion, or the threshold can be set by the equipment manufacturer or operation and maintenance experts according to safety requirements. When the change rate calculated by the system in real time is higher than the threshold, it is determined as a potential abnormal state. The threshold can be set for different parameter types (such as maximum value, mean value, change rate, etc.), and it can also be adjusted periodically according to different operating environments. For example, during high load operation, the change rate threshold of wear parameters can be appropriately relaxed to prevent false positives; while during low load or critical equipment inspection, the threshold can be appropriately tightened to improve the sensitivity of abnormality detection. The system supports manual or automatic periodic correction of the threshold to ensure that the abnormality judgment is both accurate and meets the actual needs of the field.

[0109] In a preferred embodiment of the present application, the cluster identification submodule comprises:

[0110] A spatial clustering unit is configured to generate an initial spatial cluster set according to the candidate set of abnormal state parameters, by taking abnormal state parameters with adjacent spatial positions as the same initial cluster.

[0111] A feature similarity discrimination unit is configured to, according to the initial spatial cluster set, classify abnormal points with the same abnormal state parameter type and change direction in each initial spatial cluster according to the state parameter type and change feature, to obtain feature consistency sub-clusters.

[0112] A cluster merging unit is configured to judge the spatial distance and parameter change amplitude between the feature consistency sub-clusters, and if the spatial distance is less than a preset spatial distance threshold and the change amplitude is less than a preset absolute difference threshold, the feature consistency sub-clusters that meet the conditions are merged into a cluster partition to form a cluster abnormal partition set.

[0113] In the embodiment of the present application, the clustering identification submodule can automatically divide the abnormal state parameters with adjacent spatial positions into initial spatial clustering clusters according to the candidate set of abnormal state parameters through the spatial clustering unit, so as to automatically discover spatial correlation anomalies in large-scale monitoring data. The feature similarity discrimination unit further subdivides the abnormal state parameters in the clustering cluster, classifies them according to the parameter types and change characteristics, and ensures that only the anomalies with consistent characteristics are further aggregated, thereby effectively improving the homogeneity and interpretability of the abnormal partition. The clustering merging unit merges the sub-clusters with a spatial distance less than a preset threshold and a similar parameter change amplitude into a larger range of clustering partition through judging the spatial distance and parameter change amplitude of the feature consistency sub-cluster, and finally forms a clustering abnormal partition set. This series of processing not only enables the system to quickly identify single-point anomalies, but also accurately locates and induces multiple spatially adjacent and feature-similar abnormal regions, which helps to comprehensively grasp the overall running state of the bulk material conveying belt. For example, when the parameters of multiple monitoring points in a certain section of the belt all show the same type of anomaly and are close in spatial distance, the system can automatically cluster them into a large-area wear abnormal partition, thereby providing a scientific basis for equipment maintenance and risk classification.

[0114] The spatial clustering unit is configured to generate an initial spatial clustering cluster set by taking the abnormal state parameters with adjacent spatial positions as the same initial clustering cluster according to the candidate set of abnormal state parameters, and specifically includes the following steps:

[0115] The system first arranges all the spatial nodes or partitions marked as abnormal according to their spatial positions. For spatial nodes that detect anomalies at adjacent positions, the system automatically classifies them into the same initial clustering cluster. For example, if the abnormal state parameters of the 8th, 9th and 10th spatial nodes of the belt are all abnormal and the three nodes are physically connected, the three nodes are merged into an initial spatial clustering cluster. For abnormal nodes with a gap in physical position, the system determines whether the distance between them exceeds the spatial distance threshold (see below), and if it does, the nodes are classified into different initial clustering clusters. This method refers to the existing geographic information clustering and industrial monitoring field segmentation clustering technology, and realizes automatic grouping of spatial anomalies by traversing the abnormal point group in spatial order. The initial spatial clustering cluster provides a basis for subsequent feature similarity discrimination and further merging processing, and significantly improves the flexibility and adaptability of spatial anomaly identification.

[0116] The preset spatial distance threshold is a spatial distance threshold for determining whether two feature consistency sub-clusters or abnormal spatial nodes can be classified into the same clustering partition, and specifically includes the following steps:

[0117] The system sets an upper limit of spatial distance before deployment according to the actual length of the conveying belt, the arrangement density of the monitoring nodes, and the engineering requirements on site. Generally, the threshold value can be an integer multiple of the distance between adjacent monitoring nodes, or the maximum physical distance in which anomalies of the same type are usually distributed according to historical data statistics. For example, if the center distance between two abnormal nodes is less than the set threshold value of 10 meters, it is considered to be physically related and can be aggregated into the same cluster partition; if the distance is greater than the threshold value, it is considered to belong to different spatial anomalies and is processed separately. The spatial distance can be obtained according to the physical distance between nodes, the actual arrangement of the conveying path, or latitude / longitude / three-dimensional coordinate calculation. The threshold value should support customization by on-site engineers, and can also be dynamically adjusted according to the operation of the equipment, the monitoring density, and the spatial distribution. Reasonable setting of the spatial distance threshold value can prevent misclustering of unrelated anomalies in space, and also ensure accurate identification of spatial continuous anomalies.

[0118] The preset absolute difference threshold value refers to a parameter that is used to determine whether the variation amplitude of each feature consistency sub-cluster in the clustering process can be considered as a consistent amplitude threshold, and specifically includes:

[0119] After preliminary spatial clustering, the system calculates the average parameter variation amplitude of each group of spatially adjacent feature consistency sub-clusters. For example, the average value of all abnormal state parameters in each sub-cluster is calculated, and the difference between the average values of different sub-clusters is compared. The preset absolute difference threshold value is a standard for determining whether the parameter variation amplitudes are close enough. If the difference between the average variation amplitudes of two sub-clusters is lower than the threshold value, it is considered that they have high consistency in parameter variation, and can be further merged into a larger cluster partition. If the threshold value is exceeded, it indicates that the abnormal features may come from different sources, and it is more reasonable to cluster them separately. The threshold value can be determined in combination with historical wear data analysis, equipment safety specifications, and operation experience, and can be dynamically adjusted according to the actual monitoring object. By reasonably setting the absolute difference threshold value, only abnormal points with similar variation amplitudes can be clustered together, improving the homogeneity and reliability of the clustering results.

[0120] In a preferred embodiment of the present application, the abnormal positioning sub-module comprises:

[0121] An abnormal interval determination unit is configured to extract the spatial partition range and the corresponding time range of each cluster partition according to the cluster abnormal partition set, and generate abnormal interval information;

[0122] An abnormal type determination unit is configured to extract the corresponding state parameters according to the abnormal interval information, analyze the variation characteristics of the state parameters over time, determine the type of the abnormal parameters, and generate abnormal type information.

[0123] The abnormality identification result generation unit is configured to generate and output an abnormality identification result according to the spatial partition range, the time range and the abnormality type information of each abnormality interval.

[0124] In the embodiment of the present application, the abnormality positioning sub-module first extracts the spatial partition range and the corresponding time range of each clustering partition according to the clustering abnormal partition set through the abnormality interval determination unit, and accurately locks the specific interval of abnormality occurrence. The abnormality type determination unit analyzes the variation law of the state parameter with time for each abnormality interval, and can intelligently distinguish different types such as continuous wear, sudden wear or periodic abnormality, thereby realizing automatic identification of the nature of the abnormality. The abnormality identification result generation unit automatically generates a complete abnormality identification result according to the identified spatial partition range, time range and abnormality type information, and outputs the result to the subsequent early warning link. This process can ensure that the system automatically and finely classifies and locates the abnormality with complex spatial and time variation, thereby significantly improving the operation and maintenance efficiency and the emergency response speed. For example, in actual application, the system can not only determine that a certain spatial partition is continuously worn in a specific time period, but also mark the specific type, continuous interval and evolution trend, thereby providing decision support for on-site personnel decision-making and remote monitoring.

[0125] The abnormality interval determination unit is configured to extract the spatial partition range and the corresponding time range of each clustering partition according to the clustering abnormal partition set, and generate abnormality interval information, which specifically includes:

[0126] After the system obtains the clustering abnormal partition set, it first retrieves all the spatial partition numbers and the associated time period numbers contained in each clustering abnormal partition. For the spatial partition range, the system finds the spatial nodes with the smallest and largest partition numbers in the order of the spatial partition numbers, to determine the start and end positions of the abnormal area on the belt. For the time range, the system finds all the time period numbers involved in the clustering partition to determine the earliest and latest time points, and obtains the occurrence interval of the abnormality. The system can automatically organize and archive the spatial start and end partitions and the time start and end segments of each abnormal area by traversing all the clustering partitions, to form structured abnormality interval information. The abnormality interval information not only includes the number range, start and end time of the abnormal partition, but also can further record the abnormality type, parameter characteristics and other information in the interval, thereby facilitating subsequent type determination and abnormality notification. For example, in a conveying operation, if the clustering abnormal partition set contains spatial partitions 5 to 8 and time periods 10 to 14, the system combines and determines this segment and period as the same abnormality interval, and associates the abnormal parameters, to ensure accurate positioning and efficient processing.

[0127] In a preferred embodiment of the present application, the feature similarity discrimination unit includes:

[0128] a parameter type grouping unit, configured to classify the abnormal state parameters in each cluster according to parameter types based on the initial spatial clustering cluster set, to form parameter type groups;

[0129] a change trend analysis unit, configured to determine whether the change directions of the abnormal state parameters in each parameter type group are consistent according to the parameter value changes of the abnormal state parameters in adjacent time periods, and if consistent, classify the abnormal state parameters into the same trend group, and if not consistent, divide the abnormal state parameters with different change directions into new trend groups respectively;

[0130] a magnitude consistency screening unit, configured to calculate the change magnitudes of the abnormal state parameters in each trend group, and compare the change magnitudes with the average change magnitude in the group, and if the difference between the change magnitude of a parameter and the average change magnitude in the group is less than a preset magnitude difference threshold, the parameter is retained, otherwise, the parameter is removed, to finally generate a feature consistency sub-cluster.

[0131] In the embodiment of the application, the feature similarity discrimination unit can perform multi-level and multi-angle detailed screening on the abnormal state parameters in the initial spatial clustering cluster set through the parameter type grouping unit, the change trend analysis unit and the magnitude consistency screening unit. The parameter type grouping unit first classifies abnormal state parameters of different types to ensure that subsequent analysis is only performed between parameters of the same type, avoiding feature confusion. The change trend analysis unit can identify and divide abnormal parameters with consistent or different change directions, and through dynamic trend discrimination, parameters with the same change trend are classified into the same trend group. The magnitude consistency screening unit compares and screens the parameter change magnitudes in the trend group with the average value in the group, and only retains abnormal parameters with consistent magnitudes, to finally form a feature consistency sub-cluster with high purity. This technical solution effectively filters "pseudo-abnormalities" in spatial distribution and feature performance, greatly improving the homogeneity and discrimination accuracy of clustering partition. For example, in actual monitoring, if there are a few fluctuating abnormal interference points in a cluster, the system can automatically remove the interference points through the above screening process, and only output the most representative abnormal feature group, significantly reducing the false positive and false negative rates.

[0132] The preset magnitude difference threshold is a limit standard for judging the difference between the change magnitudes of the abnormal state parameters in the same trend group and the average value in the group in the feature similarity discrimination process, and specifically includes:

[0133] In the amplitude consistency screening of each trend group, the system first calculates the amplitude of change of all abnormal state parameters in the group, and then calculates the average amplitude of change of the group. Next, the actual amplitude of change of each abnormal state parameter is compared with the average amplitude of change of the group, and the difference between the two is calculated. If the difference is less than the preset amplitude difference threshold, it is considered that the parameter is highly consistent with the change trend in the group and can be retained in the group; otherwise, it is considered that the change characteristics do not match, and it is excluded from the trend group. The setting of the amplitude difference threshold usually depends on the analysis of historical monitoring data and expert experience. For example, within the normal wear fluctuation range, the amplitude difference threshold is set as a certain percentage of the historical average amplitude of change, so as to avoid misjudgment caused by individual abnormal fluctuations. By reasonably setting and dynamically adjusting the amplitude difference threshold, the system can effectively screen out abnormal state parameters that are consistent in the same trend group, and improve the homogeneity and reliability of the clustering results.

[0134] In a preferred embodiment of the present application, the cluster merging unit comprises:

[0135] a spatial distance determination unit for calculating the distance between the center spatial partitions of each sub-cluster according to the feature consistency sub-clusters, and if the distance is less than the spatial distance threshold, the corresponding sub-cluster is classified as a mergable object;

[0136] a change amplitude determination unit for calculating the average parameter change amplitude of each sub-cluster in each group of mergable objects, and then calculating the absolute difference of the average change amplitudes between groups, and if the absolute difference is less than the absolute difference threshold, it is determined that the corresponding sub-clusters of the group meet the merging condition;

[0137] a merging generation unit for merging the feature consistency sub-clusters that simultaneously meet the spatial distance threshold and the absolute difference threshold into one cluster partition, and the set of all cluster partitions constitutes the cluster abnormal partition set

[0138] In the embodiment of the present application, the cluster merging unit realizes intelligent aggregation of feature consistency sub-clusters through the spatial distance determination unit, the change amplitude determination unit and the merging generation unit. The spatial distance determination unit can calculate the distance between the center spatial partition positions of all feature consistency sub-clusters, and automatically identify the sub-clusters with a spatial distance less than the set threshold as mergable objects, thereby ensuring that aggregation only occurs between abnormal groups that are physically close enough. The change amplitude determination unit further analyzes the average parameter change amplitudes of these mergable objects and calculates the absolute difference between groups. Only when the difference is lower than the preset absolute difference threshold, it is considered that the parameter features are highly consistent, ensuring the consistency and reliability of the cluster partition in the expression of abnormal features. The merging generation unit merges the feature consistency sub-clusters that meet the conditions of spatial distance and parameter amplitude into one cluster partition, and the set of all cluster partitions finally forms the cluster abnormal partition set and serves as the basic data for subsequent abnormal positioning.

[0139] Through the above process, the system not only realizes the cooperative aggregation of the spatial and feature dimensions of the abnormal data, but also avoids the misclustering of abnormal data with dispersed physical positions or excessively large parameter feature differences, effectively improving the accuracy and engineering usability of the clustering abnormal partition. In actual application, if multiple monitoring nodes in a certain area of the belt are adjacent in space and the parameter variation amplitude is also extremely close, the system can automatically merge the abnormalities of these nodes into the same clustering partition and include them in the clustering abnormal partition set, greatly facilitating the subsequent analysis, tracing, and early warning response of the abnormal partition, thereby assisting in precise maintenance and risk management.

[0140] The above is the preferred embodiment of the present application. It should be noted that for those skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, which should also be considered within the scope of protection of the present application.

Claims

1. A bulk material conveyor belt wear monitoring system integrating digital twin, characterized by: The system comprises: A sensor data acquisition module is used to collect raw data during the operation of the conveyor belt at multiple monitoring locations along the conveyor belt, wherein the raw data includes physical signal data and image data signals; The data preprocessing module is used to format, reduce noise and calibrate the raw data to obtain analytical data; The digital twin construction module is used to dynamically generate a virtual model of the conveyor belt based on the analysis data, map the analysis data to the corresponding areas of the preset virtual model according to the spatial and temporal dimensions, and obtain the multi-region state parameter set of the belt; The anomaly recognition module is used to perform partition detection on the wear characteristics of each monitoring area of ​​the belt based on the multi-area state parameter set of the belt, identify and locate local areas with abnormal change trends, and obtain anomaly recognition results; The early warning trigger module is used to generate early warning information including the abnormal location and wear degree based on the abnormality identification results, and output it to the host computer or operation and maintenance terminal; The digital twin building blocks include: The virtual model construction submodule is used to generate a virtual model corresponding to the conveyor belt structure according to the monitoring position and conveyor belt structure information in the analysis data, and initialize the attributes of each spatial node; The data mapping submodule is used to map the analysis data to the corresponding spatial nodes and time indexes in the virtual model according to preset spatial partitions and time periods, and to update the original data sequence of each spatial node in different time periods. The spatial partitions are groupings of monitoring locations along the belt, the time periods are groupings of analysis data collection times, and the time indexes are numberings of time periods. The parameter extraction submodule is used to calculate the maximum value, minimum value, average value and rate of change of each spatial node through its original data sequence, and to correspond them to the spatial position and time index of the spatial node one by one to generate the state parameter set of the spatial node; The parameter sequence combination submodule is used to serialize the state parameter sets of all spatial nodes in each time period in the order of spatial partition and time index to generate a multi-dimensional parameter matrix, and perform multi-dimensional statistics on the various state parameter data to generate a multi-region state parameter set for the belt; The abnormality identification module includes: The parameter sequence analysis submodule is used to analyze the state parameter sequences of each spatial partition in different time periods based on the multi-region state parameter set of the belt, calculate the parameter change rate at adjacent time points, and compare it with the preset abnormal threshold to generate a candidate set of abnormal state parameters; The cluster identification submodule is used to cluster the candidate set of abnormal state parameters according to spatial proximity and change characteristics, identify spatial partition areas with similar abnormal characteristics, and generate clustered abnormal partition sets; The anomaly location submodule is used to calibrate the anomaly location and time period based on the clustered anomaly partition set, combined with the spatial partition and time period information, to form an anomaly identification result; The parameter extraction submodule includes: An extreme value calculation unit is used to determine the maximum and minimum values ​​of all sampled values ​​according to the original data sequence of each spatial node in each time period, and output the maximum and minimum values ​​as extreme value parameters corresponding to the spatial position and time index of the spatial node respectively; The mean calculation unit is used to accumulate all the sampled values ​​in the original data sequence of each spatial node in each time period and divide them by the number of samples to obtain an average value parameter. The average value is output in a one-to-one correspondence with the spatial position and time index of the spatial node; A change rate calculation unit is used to calculate the difference of all adjacent sampling values ​​in sequence according to the original data sequence of each spatial node in each time period to obtain a difference sequence, and divide the sum of the absolute values ​​of the difference sequence by the number of differences to obtain a change rate parameter. The change rate parameter is output in a one-to-one correspondence with the spatial position and time index of the spatial node; The parameter aggregation unit is used to uniformly aggregate extreme value parameters, average value parameters and change rate parameters according to the spatial position and time index of each spatial node to form a state parameter set for each spatial node and each time period.

2. The bulk material conveyor belt wear monitoring system integrating digital twin according to claim 1 is characterized in that: The parameter sequence combination submodule includes: The sequence arrangement unit is used to arrange the state parameter sets of all spatial nodes in each time period in sequence according to the spatial partition position order and the increasing order of time index to generate a multi-dimensional parameter matrix; The multidimensional statistical unit is used to perform statistics on various state parameter data in the multidimensional parameter matrix according to the following steps: For each spatial partition, according to the time index order, for the same state parameters, calculate the mean, range, and maximum continuous increasing interval length of each time period to generate the partition time series statistics; For the same state parameters of all spatial partitions under each time index, calculate the spatial maximum, minimum, mean and standard deviation at that time point and generate the corresponding spatial statistical results; The parameter integration unit is used to merge the time series statistical results of all spatial partitions and the spatial statistical results of each time period according to the two dimensions of spatial partition and time, and form a multi-region belt state parameter set indexed by spatial partition and time.

3. The bulk material conveyor belt wear monitoring system integrating digital twin according to claim 1 is characterized in that: The parameter sequence analysis submodule includes: A change rate calculation unit is used to calculate the difference of the state parameters at adjacent time indexes for each spatial partition state parameter sequence, and divide the sum of the absolute values ​​of all differences by the number of differences to obtain the change rate sequence of the spatial partition on the state parameter; The threshold comparison unit is used to compare each change rate value in the change rate sequence with the preset abnormality judgment threshold. If the change rate value is greater than the abnormality judgment threshold, the corresponding spatial partition, time index and state parameter are marked as abnormal candidates to generate an abnormal state parameter candidate set.

4. The bulk material conveyor belt wear monitoring system integrating digital twin according to claim 1 is characterized in that: The cluster identification submodule includes: The spatial clustering unit is used to take the abnormal state parameters with adjacent spatial positions as the same initial cluster according to the abnormal state parameter candidate set to generate an initial spatial cluster set; The feature similarity discrimination unit is used to classify the abnormal points with the same abnormal state parameter type and change direction according to the state parameter type and change characteristics in each initial spatial cluster according to the initial spatial cluster cluster set, and obtain feature consistency subclusters; The cluster merging unit is used to judge the spatial distance and parameter change amplitude between feature consistency subclusters. If the spatial distance is less than the preset spatial distance threshold and the change amplitude is less than the preset absolute difference threshold, the feature consistency subclusters that meet the conditions will be merged into cluster partitions to form a cluster anomaly partition set.

5. The bulk material conveyor belt wear monitoring system integrating digital twin according to claim 1 is characterized in that: The abnormality locating submodule includes: An abnormal interval determination unit is used to extract the spatial partition range and corresponding time range of each cluster partition according to the cluster abnormal partition set, and generate abnormal interval information; The abnormality type determination unit is used to extract the corresponding state parameters based on the abnormal interval information and analyze its change characteristics over time. If the abnormality persists throughout the entire time range, it is determined to be continuous wear. If a sudden change occurs at a certain moment, it is determined to be sudden wear. If the change shows regular fluctuations, it is determined to be a periodic abnormality and generates abnormality type information; The abnormality identification result generating unit is used to generate and output the abnormality identification result according to the spatial partition range, time range and abnormality type information of each abnormal interval.

6. The bulk material conveyor belt wear monitoring system integrating digital twin according to claim 4 is characterized in that: The feature similarity determination unit includes: The parameter type grouping unit is used to classify the abnormal state parameters in each cluster according to the parameter type according to the initial spatial clustering cluster set to form parameter type groups; The change trend analysis unit is used to determine whether the change directions of the abnormal state parameters in each parameter type group are consistent based on the changes in their parameter values ​​in adjacent time periods. If they are consistent, they are classified into the same trend group. If they are inconsistent, the abnormal state parameters with different change directions are divided into new trend groups and classified separately; The amplitude consistency screening unit is used to calculate the change amplitude of each parameter in each trend group for the abnormal state parameters, and compare it with the average change amplitude in the group. If the difference between the change amplitude of the parameter and the average value in the group is less than the preset amplitude difference threshold, the parameter is retained; otherwise, it is eliminated, and finally a feature consistency subcluster is generated.

7. The bulk material conveyor belt wear monitoring system integrating digital twin according to claim 4 is characterized in that: The cluster merging unit includes: The spatial distance determination unit is used to calculate the distance between the central spatial partitions of each sub-cluster based on the feature consistency sub-clusters. If the distance is less than the spatial distance threshold, the corresponding sub-cluster is classified as a mergeable object; A change amplitude determination unit is used to calculate the average parameter change amplitude of each sub-cluster in each group of mergeable objects, and then calculate the absolute difference of the average change amplitudes between the groups. If the absolute difference is less than the absolute difference threshold, it is determined that the sub-cluster corresponding to the group meets the merging condition; The merging generation unit is used to merge the feature consistency subclusters that meet both the spatial distance threshold and the absolute difference threshold into a cluster partition, and the set of all cluster partitions constitutes a cluster anomaly partition set.

Images