Scraper chain state remote monitoring system based on AI learning

Abstract

The invention relates to the technical field of scraper chain monitoring, and discloses a scraper chain state remote monitoring system based on AI learning. A stress distribution monitoring module of the system calls a chain link stress peak value, the deformation difference quantity of adjacent chain links and transmission path rate data, analyzes the matching degree of stress distribution and structural stability, and generates a stress distribution control data set; an anomaly sensing module extracts a key node displacement value and a gradient offset based on the data set, optimizes node displacement distribution balance, and generates a node displacement optimization parameter set; the dynamic compensation module analyzes the permeation and diffusion of the lubricating medium, adjusts the injection rate and the path distribution proportion, and generates a dynamic compensation parameter set; the learning intervention module extracts a real-time vibration fluctuation rate and a tension offset, adjusts a lubricating medium distribution path and a vibration-tension ratio, and generates an intervention instruction data set; the quality optimization module analyzes chain link wear characteristics and degradation time, adjusts monitoring path parameters, and generates a chain link health state evaluation table.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of scraper chain monitoring, in particular to a scraper chain state remote monitoring system based on AI learning. BACKGROUND

[0002] In the scene of thermal power plants, scraper chains are often used for coal conveying and ash transfer, and need to withstand high temperature, variable load, dust erosion and corrosive gas for a long time, and the working conditions are more complex and special. Scraper chains are prone to wear, deformation, breakage and other failures. Once a failure occurs, it will not only cause production interruption, but also may cause safety accidents and serious economic losses.

[0003] The monitoring of the state of the scraper chain is mostly dependent on manual inspection or traditional sensor monitoring methods. Manual inspection is limited by the experience and energy of the inspectors, and it is difficult to achieve real-time and comprehensive perception of the running state of the scraper chain, and there is a risk of missed detection and misjudgment. Especially in the scene of high-speed operation of equipment or harsh environment, the limitations of manual inspection are more prominent. The traditional sensor monitoring method can obtain some running parameters, but it often only monitors a single indicator, such as simply monitoring the stress value or vibration frequency of the chain link, and lacks comprehensive analysis and correlation judgment of multiple parameters.

[0004] In the aspect of stress distribution monitoring, the existing technology cannot accurately capture the stress peak value changes of the chain link in different running stages, nor can it effectively analyze the influence of the deformation difference between adjacent chain links on the overall structural stability, resulting in a lack of scientific basis for stress regulation and easy occurrence of local stress concentration phenomenon, which accelerates the fatigue damage of the chain link. For node displacement monitoring, the traditional method mostly uses a fixed threshold alarm mechanism, which cannot dynamically adjust the displacement balance according to the real-time running state of the scraper chain. When the node displacement gradient deviates, it is difficult to intervene in time, thereby affecting the running stability of the scraper chain.

[0005] The reasonable supply of lubricating medium is the key to ensure the long-term operation of the scraper chain, but the existing lubrication control mostly uses a fixed rate injection method, which cannot adjust the injection rate and path distribution of the lubricating medium according to the actual penetration and diffusion of the chain link meshing surface, resulting in insufficient lubrication in some areas and aggravating wear, or excessive lubrication in some areas causing resource waste. At the same time, in the aspects of vibration and tension monitoring, the existing system cannot real-time correlate the influence of vibration fluctuation rate and tension deviation on the chain link wear rate, and cannot dynamically adjust the running parameters to reduce the wear risk.

[0006] The existing monitoring technology lacks effective learning and optimization mechanism, and each monitoring module is independent, so the data cannot be effectively fused and fed back, resulting in that the evaluation of the health state of the scraper chain lags behind the actual running state, and it is difficult to realize forward-looking maintenance, and the overall monitoring efficiency and reliability need to be improved. SUMMARY

[0007] The purpose of this invention is to provide a remote monitoring system for scraper chain status based on AI learning, so as to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides a remote monitoring system for the status of a scraper chain based on AI learning, the system comprising: The stress distribution monitoring module, based on the scraper chain's operating status information, calls up the peak stress parameters of the chain links, the deformation difference between adjacent chain links, and the transmission path speed data. It analyzes the degree of matching between stress distribution and structural stability, allocates stress control values ​​and transmission balance parameters for chain links, and generates a stress distribution control dataset. Based on the stress distribution control dataset, the anomaly perception module extracts the displacement values ​​and gradient offsets of key nodes in the scraper chain, analyzes the impact of node displacement fluctuations on operational stability, adjusts the uniformity of node displacement distribution, and generates a set of node displacement optimization parameters. Based on the node displacement optimization parameter set, the dynamic compensation module analyzes the penetration and diffusion of lubricating medium on the meshing surface of the chain links, adjusts the lubricating medium injection rate and flow path distribution ratio, redistributes the distribution trend and dynamic parameter values ​​of lubricating medium concentration, and generates a dynamic compensation parameter set. Based on the dynamic compensation parameter set, the learning intervention module extracts the real-time vibration fluctuation rate and tension offset during the operation of the scraper chain, analyzes the influence of the fluctuation range on the wear rate of the chain links, dynamically adjusts the distribution path of the lubricating medium and the vibration-tension ratio within the target range, and generates an intervention instruction dataset. Based on the intervention instruction dataset, the quality optimization module analyzes the distribution ratio and deterioration time of the wear characteristics of the target links, adjusts the parameters of the target monitoring path, and generates a link health status assessment table.

[0009] Preferably, the step of obtaining the degree of matching between the analytical stress distribution and structural stability specifically includes: Based on the scraper chain's operating status information, the peak stress parameters, deformation differences, and transmission path speed data of the chain links are extracted. A time window is set, and time points are selected to match the data. By comparing the data correlation and filtering the data, the peak stress parameters and deformation differences are obtained. Based on the stress peak parameter and deformation difference, the transmission path is matched and verified, the deviation between the stress peak and deformation difference is calculated, the stress distribution and deformation distribution are corrected by combining the rate change, and the path parameters are adjusted by the influence of the rate on the data to obtain the matching situation between the stress peak and deformation. Based on the stress peak and deformation matching, structural stability analysis is performed, structural stability analysis standards are set, and stress distribution under differentiated rate conditions is evaluated in combination with dynamic changes during operation. Stability indices are compared and rate conditions are optimized to obtain the degree of matching between stress distribution and structural stability.

[0010] Preferably, the step of obtaining the stress distribution control dataset specifically includes: Based on the degree of matching between the stress distribution and structural stability, the stress transmission and stability changes of the scraper chain under different operating conditions are analyzed, and the stress distribution of the chain links is weighted and calculated to obtain the preliminary stress control requirements of the chain links. Based on the initial stress control requirements of the links, the stress balance between links is analyzed, the relationship between stress transfer efficiency and load distribution between links is identified, the stress control parameters of the links are corrected, the stress compensation coefficient after the link adjustment is calculated, and the stress control dataset between links is obtained. By combining the stress control dataset between chain links with the structural stability matching results, the stress between chain links is allocated, optimized and matched to the required equilibrium and stability requirements, and a stress distribution control dataset is generated.

[0011] Preferably, the steps for obtaining the displacement values ​​and gradient offsets of the key nodes of the scraper chain are as follows: Based on the stress distribution control dataset, deformation data at key nodes of the scraper chain are extracted, deformation points in each time period are screened, and deformation fluctuations are analyzed by combining the deformation change trends at different locations of nodes to obtain deformation data at key nodes. Based on the deformation data at the key nodes, the deformation point and its corresponding displacement value are calculated. By analyzing the relationship between deformation and displacement, the displacement change of each measurement point is identified. Combined with the chain structure parameters, the displacement changes at different locations are compared to obtain displacement distribution and gradient distribution data. Based on the displacement distribution and gradient distribution data, the overall displacement distribution of key nodes of the scraper chain is analyzed, the displacement gradient is optimized by combining deformation data, the impact of displacement changes on operating performance is analyzed, the stable displacement configuration under differentiated operating conditions is determined, and the displacement values ​​and gradient offsets of key nodes are obtained.

[0012] Preferably, the steps for obtaining the nodal displacement optimization parameter set are as follows: Based on the displacement values ​​and gradient offsets of the key nodes, the time series of displacement changes is determined, the current displacement value is compared with the initial displacement data, the displacement gradient at each moment is analyzed, and corresponding thresholds are defined according to the running state partition to generate a preliminary displacement change parameter set. The preliminary displacement change parameter set is analyzed to analyze the impact of key node displacements on operational stability, identify the correlation between displacement and operational status, and calculate the section operational stability influence coefficient. By analyzing the operational stability impact coefficient of the section and combining it with the node displacement change parameters, the displacement distribution balance is adjusted, the displacement control data is optimized, and a set of node displacement optimization parameters is generated.

[0013] Preferably, the step of obtaining the dynamic compensation parameter set specifically includes: Based on the node displacement optimization parameter set, lubricant penetration data of the chain link meshing surface is extracted, the penetration rate of lubricant on the surface of differentiated materials is monitored, and the diffusion characteristics of lubricant are inferred by combining external environmental factors such as time and load. The penetration and diffusion rate coefficients are defined, and a dynamic parameter set of penetration and diffusion is generated. The influence of the aforementioned dynamic parameters of permeation and diffusion on the flow rate and distribution of the medium is analyzed. Based on the requirements of the lubricating medium concentration distribution at the meshing surface, the ratio between the flow path and the medium injection rate is optimized, the adjustment coefficient of the lubricating medium concentration distribution trend is calculated, and the medium concentration control results are generated. Analyze the results of the medium concentration control, adjust the ratio between the lubricating medium injection rate and the flow path, allocate the distribution trend of the lubricating medium concentration, and combine the permeation diffusion parameters and adjustment coefficients to obtain a dynamic compensation parameter set.

[0014] Preferably, the step of obtaining the intervention instruction dataset specifically includes: Based on the dynamic compensation parameter set, the vibration fluctuation rate and tension offset during operation are monitored in real time, the fluctuation range is identified, abnormal values ​​of the equipment are eliminated, the average fluctuation rate of the data is analyzed, and vibration and tension fluctuation data are obtained. The influence of the vibration and tension fluctuation range on the wear rate of the chain links is analyzed. Using the known wear rate of the chain links, the relationship between tension and vibration is analyzed, and the wear rate under the different fluctuation range is calculated to obtain the wear rate influence data. Based on the wear rate impact data, the distribution path of the lubricating medium and the vibration ratio within the target range are dynamically adjusted. Adjustments are made based on the relationship between the wear rate impact data and the vibration and tension fluctuation range. The medium flow rate and vibration control range are allocated, and an intervention instruction dataset is generated.

[0015] Preferably, the step of obtaining the health status assessment table for generating the chain segment is as follows: Based on the intervention instruction dataset, the wear characteristic distribution and deterioration time of the target link are analyzed. Wear characteristic data at different monitoring time points are collected, the characteristic time distribution is sorted out, the data change trend is analyzed and the data is classified to obtain the link wear distribution data. Based on the chain link wear distribution data, the target monitoring path parameters are adjusted, the optimal monitoring time and characteristic distribution of chain link wear are analyzed, and the operating conditions of monitoring vibration threshold, time window and medium concentration are adjusted by comparing the characteristic changes under different monitoring conditions to obtain the target monitoring path parameters. Based on the target monitoring path parameters, the monitoring conditions are adjusted according to the current operating parameters, and the relationship between the variables of monitoring time, vibration threshold, and medium concentration is controlled. Real-time evaluation is performed based on the adjusted parameters to generate a chain link health status assessment table.

[0016] Preferably, the system further includes: Based on the chain link health status assessment table, the topology association analysis module extracts structural feature data of adjacent chain links, analyzes the structural similarity between associated chain links according to the displacement vector of the connection point between chain links and the structural deformation gradient, and generates structural association parameters in combination with the running load distribution parameters. Based on the structural association parameters, the state mapping module compares the structural feature differences between healthy and abnormal links, and determines the identity of the associated link structures through displacement vector consistency verification and deformation gradient offset analysis, and outputs structural identity determination parameters. Based on the structural identity determination parameters, the map generation module integrates the target monitoring path parameters and chain link health status data to construct a full-link topology map of the scraper chain.

[0017] Preferably, the map generation module specifically performs the following: Obtain the displacement vector matching result and deformation gradient convergence value of the associated links in the structural identity determination parameters; The positional mapping relationship of the connection points between chain segments is corrected based on the displacement vector matching result, and the chain segment deformation compensation coefficient is adjusted based on the deformation gradient convergence value. By integrating the corrected position mapping relationship, deformation compensation coefficient, and real-time monitoring data from the chain link health status assessment table, a three-dimensional dynamic topology model of the scraper chain is reconstructed, generating a full-link status monitoring map.

[0018] Compared with existing technologies, the advantages of this invention are: the AI-based remote monitoring system for scraper chain status achieves comprehensive and intelligent monitoring of the scraper chain's operating status through the collaborative operation of multiple modules. The stress distribution monitoring module, by comprehensively accessing and analyzing peak stress parameters of chain links, deformation differences between adjacent chain links, and transmission path speed data, can accurately grasp the matching relationship between stress distribution and structural stability. By rationally allocating chain link stress control values ​​and transmission balance parameters, the stress distribution in various parts of the scraper chain becomes more reasonable, reducing structural damage caused by localized stress concentration.

[0019] Based on the stress distribution control dataset, the anomaly detection module focuses on the extraction and analysis of key node displacement values ​​and gradient offsets, and delves into the impact of node displacement fluctuations on operational stability. By dynamically adjusting the uniformity of node displacement distribution, it generates an optimized set of node displacement parameters, which can promptly correct node displacement deviations, avoid operational oscillations caused by displacement imbalance, and enhance the operational stability of the scraper chain under complex working conditions.

[0020] The dynamic compensation module, taking into account the characteristics of the lubricating medium's effect on the meshing surface of the chain links, combines a set of optimized node displacement parameters to meticulously analyze the penetration and diffusion patterns of the lubricating medium. By adjusting the lubricating medium injection rate and flow path distribution ratio, the module redistributes the distribution trend and dynamic parameter values ​​of the lubricating medium concentration, ensuring that the lubricating medium can accurately act on the critical parts requiring lubrication. This reduces chain link wear caused by insufficient or excessive lubrication, extending the effective service life of the chain links.

[0021] The learning intervention module relies on a dynamic compensation parameter set to extract the vibration fluctuation rate and tension offset during the scraper chain operation in real time, and deeply analyzes the impact of the fluctuation range of these two parameters on the wear rate of the chain links. By dynamically adjusting the distribution path of the lubricating medium and the vibration-tension ratio within the target range, targeted intervention commands are generated to ensure that the scraper chain maintains a reasonable vibration and tension state at different operating stages, reducing the cumulative damage to the chain links caused by abnormal vibration and tension fluctuations.

[0022] The quality optimization module integrates the intervention command dataset, comprehensively analyzes the distribution ratio and deterioration time of wear characteristics of target links, and generates a detailed link health status assessment table by adjusting the parameters of the target monitoring path. This assessment table clearly reflects the health status of each link, providing an intuitive reference for maintenance decisions, making maintenance work more targeted, and reducing resource consumption caused by blind maintenance.

[0023] The various modules utilize an AI learning mechanism to achieve efficient data flow and correlation analysis, forming a complete closed loop from status monitoring and parameter optimization to intervention and control and health assessment. This collaborative operation mode breaks the limitations of the isolation between different links in traditional monitoring, enabling the monitoring system to continuously optimize its performance based on the real-time operating status of the scraper chain, improve its adaptability to complex working conditions, and ensure the long-term stable operation of the scraper chain. Attached Figure Description

[0024] Figure 1 This is a schematic diagram illustrating the working principle of the AI-based remote monitoring system for scraper chain status as described in this invention.

[0025] Figure 2 This is a flowchart for analyzing the degree of matching between stress distribution and structural stability.

[0026] Figure 3 This is a flowchart for extracting the displacement values ​​and gradient offsets of key nodes in the scraper chain.

[0027] Figure 4 A flowchart for generating the parameter set for nodal displacement optimization. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] Please see Figure 1 This invention provides a remote monitoring system for the status of a scraper chain based on AI learning, the system comprising: The system collects scraper chain operation status information through a stress distribution monitoring module, including peak stress parameters of chain links, deformation differences between adjacent chain links, and transmission path rate data. It analyzes the matching degree between stress distribution and structural stability, allocates chain link stress control values ​​and transmission balance parameters, and generates a stress distribution control dataset. An anomaly detection module extracts key node displacement values ​​and gradient offsets based on this dataset, analyzes the impact of node displacement fluctuations on operational stability, and generates a node displacement optimization parameter set. A dynamic compensation module adjusts the penetration and diffusion parameters of the lubricating medium on the chain link meshing surface according to the node displacement optimization parameter set, generating a dynamic compensation parameter set. A learning intervention module extracts real-time vibration fluctuation rate and tension offset based on the dynamic compensation parameter set, analyzes the impact of fluctuation range on chain link wear rate, and generates an intervention command dataset. A quality optimization module analyzes the distribution ratio and deterioration time of target chain link wear characteristics based on the intervention command dataset, generating a chain link health status assessment table.

[0030] Example 1: See Figure 2 The analysis process involves determining the degree of matching between stress distribution and structural stability. This process is achieved by dynamically acquiring information on the scraper chain's operating status, including peak stress parameters of chain links, deformation differences between adjacent chain links, and transmission path rate data. The system sets a fixed time window, the range of which is dynamically adjusted according to the transmission device's rotational speed; a shorter time window is used when the speed is higher, and a longer time window is used when the speed is lower. Within the selected time window, the system extracts three types of data matching the timestamps: peak stress parameters record the maximum stress value borne by each chain link, deformation differences reflect the difference in deformation between adjacent chain links, and transmission path rate describes the power transmission speed.

[0031] Data correlation comparison employs multi-dimensional matrix analysis. A three-dimensional data matrix is ​​established, with each dimension corresponding to the time series, chain node position index, and parameter type, respectively. The data matrix is ​​processed using a singular value decomposition algorithm to remove noisy data with singular values ​​below a threshold. The retained data is standardized to eliminate the influence of dimensional differences. The Pearson correlation coefficient between peak stress parameters and deformation differences is calculated; data with an absolute correlation coefficient greater than 0.7 are considered validly correlated. The final output consists of a filtered set of peak stress parameters and a set of deformation differences.

[0032] The transmission path matching verification employs a path topology mapping method. A topology graph of the scraper chain transmission path is constructed, where nodes represent chain links and edges represent transmission relationships. Path weights are calculated based on the transmission path speed data, with weight values ​​inversely proportional to the speed. Stress-deformation deviation analysis is performed on each chain link: the peak stress parameter of the link is extracted, the deformation difference between adjacent links is retrieved, and the Euclidean distance between the peak stress and the deformation difference is calculated as the initial deviation value. Combining the transmission path weights, the weighted least squares method is used to correct the initial deviation value, generating a corrected deviation matrix. The path parameters are adjusted based on the correction results; when the deviation value exceeds a set threshold, the transmission weight coefficient of the path is automatically reduced to optimize stress transmission efficiency. The final output is a data structure containing the peak stress and deformation matching coefficients of each chain link.

[0033] Structural stability analysis employs a dynamic benchmark assessment mechanism. A stability assessment index system is defined, including three core indicators: stress distribution dispersion, deformation gradient change rate, and transmission balance coefficient. An operational state-stability mapping model is established, trained using historical data. The input parameter is the rate change along the transmission path, and the output is the expected value of the stability index. Real-time dynamic operational data is collected, including load change curves, transmission acceleration, and ambient temperature parameters. This real-time data is input into the mapping model to generate the stability benchmark value under the current operating conditions.

[0034] Stress distribution assessment under differentiated speed conditions employs multi-scenario simulation. A speed classification model is established within the control system, dividing the transmission path speed into three intervals: low-speed, medium-speed, and high-speed. Independent analysis is performed for each speed interval: stress distribution data is extracted, and stress concentration factor and distribution uniformity are calculated; the coupling relationship between stress distribution and structural deformation is derived by combining current operating load parameters; and stress transmission paths at different speeds are simulated using finite element simulation to identify high stress concentration areas. Stability indices across the three speed intervals, including structural resonance frequency offset and deformation recovery period, are compared. Based on the comparison results, the speed control strategy is optimized. When a deterioration in stability indices is detected in a certain speed interval, suggested speed adjustment parameters are automatically generated. The final output includes an evaluation result containing a matching coefficient, calculated from the deviation of the actual stability index from the benchmark value, with a value ranging from 0 to 1; a higher value indicates a better matching degree.

[0035] The entire analysis process employs an iterative optimization mechanism. The results generated from each analysis are fed back to the data acquisition module, dynamically adjusting the time window length and sampling frequency. When the matching coefficient falls below 0.6 three times consecutively, a data verification process is automatically triggered: recalibrating sensor accuracy, reviewing the transmission path topology, and verifying the validity of the mapping model parameters. The system establishes an analysis log database to record key parameters and adjustment records during each analysis, providing data support for algorithm optimization. The analysis cycle is dynamically adjusted based on the equipment's operating status. Under normal operating conditions, a complete analysis is performed every 5 minutes; when abnormal vibration or sudden load changes are detected, the analysis frequency is increased to once per minute.

[0036] Example 2: See Figure 3 This system covers the generation process of stress distribution control datasets and the extraction methods for key node displacement values ​​and gradient offsets. In the stress distribution control dataset generation stage, the system analyzes the degree of matching between stress distribution and structural stability. Using this matching degree as an input parameter, the system first analyzes the stress transmission characteristics of the scraper chain under various operating conditions. Operating conditions are divided into three modes based on transmission path speed, load change rate, and ambient temperature parameters: normal operating conditions, variable load conditions, and extreme operating conditions. For each operating condition mode, the system establishes an independent stress transmission analysis model: analyzing the periodic characteristics of stress fluctuations under normal operating conditions; tracking the stress transmission hysteresis effect during sudden load changes under variable load conditions; and monitoring the stress distribution distortion at the material's yield critical point under extreme operating conditions. Through parallel analysis of multiple operating conditions, the system calculates the stress distribution weight coefficient for each chain link under the corresponding operating condition. This coefficient is determined by the ratio of peak stress to deformation recovery capacity, and finally outputs a preliminary stress control requirement dataset containing operating condition identifiers.

[0037] The stress equalization analysis between chain links employs a dynamic load tracking mechanism. The system constructs a chain link load distribution matrix, where row vectors correspond to time series and column vectors identify chain link position indices. A sliding time window algorithm is used to calculate the load transfer efficiency between adjacent chain links in real time. Load transfer efficiency is defined as the ratio of the actual load of the output chain link to the theoretical load of the input chain link. When the transfer efficiency of a specific chain link is detected to be consistently below a threshold, the system initiates stress transfer path diagnosis: reconstructing the transmission topology of the upstream and downstream links of that chain link, detecting the wear clearance of the chain link meshing surfaces, and analyzing the distribution of the lubricating medium. Based on the diagnostic results, the stress control parameters are corrected, introducing a stress compensation coefficient δ, which is determined by the product of the transfer efficiency deviation and the structural deformation recovery rate. The corrected parameters form an inter-link stress control dataset, containing the position identifier, compensation coefficient value, and correction timestamp for each chain link.

[0038] The final dataset generation process employs a multi-objective optimization algorithm. The system establishes a dual objective function of stability and equilibrium. The stability objective function integrates parameters such as structural resonant frequency offset and deformation gradient change rate; the equilibrium objective function includes indicators such as load distribution dispersion and stress concentration coefficient. A Pareto optimal solution search algorithm is used to traverse possible stress distribution schemes based on the inter-link stress control dataset. Each iteration adjusts three key variables: link preload adjustment, transmission path speed compensation value, and load distribution weight. After each adjustment, the dual objective function values ​​are recalculated. When the change in the objective function over five consecutive iterations is less than the convergence threshold, the optimal solution is output as the stress distribution control dataset. This dataset uses a hierarchical storage structure: the top layer is a working condition classification index, the middle layer records link positions and parameter adjustments, and the bottom layer stores historical version data of the optimization process.

[0039] The key node displacement value extraction process starts from the stress distribution control dataset. The system first identifies the key topological nodes of the scraper chain, including three categories: transmission gear meshing points, chain link hinge points, and load concentration points. An independent deformation monitoring channel is established for each type of node to collect high-frequency sampled deformation data. Deformation data preprocessing employs a three-stage filtering mechanism: primary filtering removes equipment vibration noise, secondary filtering eliminates the influence of temperature drift, and tertiary filtering separates the effective deformation caused by the workload. The filtered deformation data is segmented by time window, and within each window, a clustering algorithm is used to identify deformation feature points, labeling them as abrupt deformation, gradual recovery, and periodic fluctuation modes. By analyzing the evolution trend of feature points in adjacent windows, a node deformation fluctuation spectrum is constructed. The horizontal axis of this spectrum represents the time series, the vertical axis represents the deformation amplitude, and the spectral line color depth represents the fluctuation frequency.

[0040] Displacement calculation incorporates spatial coordinate system transformation. A local coordinate system is established for each key node, with the origin at the node's geometric center and the Z-axis perpendicular to the link's motion plane. Deformation monitoring data is converted into three-dimensional deformation data in this coordinate system. A deformation-displacement transformation matrix is ​​used, containing parameters in three dimensions: the link material's elastic modulus, structural geometric parameters, and the real-time load vector. The calculation process employs an iterative approximation method: the initial displacement value is determined by the product of the deformation and the elastic modulus; the displacement direction vector is corrected based on the structural geometric parameters; and the displacement amplitude is adjusted in conjunction with the real-time load vector. After each iteration, the physical rationality of the displacement value is verified, and a recalculation is initiated when the angle between the displacement vector and the motion trajectory exceeds a safety threshold. The final output is a data structure containing a timestamp, node type, and three-dimensional displacement coordinates.

[0041] Gradient offset analysis employs a spatiotemporal joint analysis method. The system constructs a displacement distribution field model, unfolding the scraper chain into a two-dimensional planar coordinate grid, with grid intersections corresponding to key node positions. A displacement rate field is established in the time dimension, and the instantaneous displacement gradient is obtained by solving the time partial derivative of the displacement field. Spatial dimension analysis introduces the Laplace operator to calculate the spatial curvature distribution of the displacement field. The gradient offset is defined as the deviation between the actual displacement gradient and the theoretical motion model, obtained through four steps: first, calculating the displacement vector difference between adjacent nodes; second, solving for the angle between the vector difference and the theoretical motion direction; third, weighted averaging of the angle deviation; and fourth, introducing a transmission path weight coefficient to correct the final gradient offset. The analysis process uses a parallel computing architecture, with each key node allocated an independent computing unit, and a central scheduler synchronizing the calculation results of each node in real time. The final generated gradient offset dataset includes the offset magnitude, offset direction, and confidence index for each node, with the confidence index determined by the data sampling rate and the number of calculation iterations. All data is stored in a time series, forming a complete historical record of node displacement evolution.

[0042] Example 3: See Figure 4 This involves the generation process of the nodal displacement optimization parameter set and the dynamic compensation parameter set. The generation of the nodal displacement optimization parameter set begins with the displacement values ​​and gradient offset data of key nodes. The system first establishes a time-series model of displacement changes. This model uses a sliding window mechanism to process the time-series data. The window length is dynamically adjusted according to the real-time rotational speed of the transmission device; a shorter window is used when the rotational speed is high to capture rapid changes, and a longer window is used when the rotational speed is low to improve data stability. Within each time window, the system performs multi-scale analysis of the displacement data: macro-scale analysis of the overall displacement trend, meso-scale identification of periodic fluctuation characteristics, and micro-scale detection of transient abnormal displacements. Through this three-level scale analysis, a complete feature spectrum of displacement changes is constructed.

[0043] The initial displacement data comparison employs a baseline calibration method. The system maintains a displacement baseline value for each key node, which is derived from historical data collected during equipment operation under no-load conditions. A dynamic tolerance mechanism is introduced during the comparison of the current displacement value with the baseline value; the tolerance range is adjusted proportionally based on the real-time load. The displacement gradient is calculated using the central difference method, considering displacement data from three consecutive time points to smooth instantaneous fluctuations. The operational state partitioning is defined using a fuzzy logic algorithm, taking displacement change, gradient value, and load parameters as input variables, and outputting operational state membership degrees. The membership degree intervals are divided into three categories: stable zone, transition zone, and danger zone.

[0044] The initial displacement change parameter set is generated using feature encoding technology. The system converts the displacement features within each time window into feature vectors, with vector dimensions including parameters such as displacement change amplitude, gradient change rate, and fluctuation frequency. The feature vectors are compressed and encoded using an autoencoder neural network to generate a compact parameter representation. The encoding process preserves the topological structure of the original data, ensuring that similar displacement patterns are close together in the encoding space. The parameter set is organized chronologically, with each record containing a timestamp, encoded features, and a running status partition identifier.

[0045] The operational stability impact analysis employs association rule mining. The system constructs a displacement-operational state association database, recording the correspondence between historical displacement changes and subsequent equipment state changes. Frequently occurring association rules are mined using the Apriori algorithm, such as the pattern of increased vibration after a specific node's displacement gradient exceeds a threshold. When analyzing the impact of critical node displacements on operational stability, the system matches the current displacement pattern with historical association rules in real time and calculates the impact coefficient.

[0046] in: Indicates the impact coefficient on the operational stability of the section. Let be the weight of the i-th matching rule. For rule support, Here, n represents the rule confidence level, and n is the number of matching rules. The weight values ​​are dynamically adjusted based on the recentity of the rules, with recently validated rules having higher weights.

[0047] The displacement distribution balance adjustment employs a distributed optimization strategy. The system divides the scraper chain into multiple logical segments, each containing several adjacent key nodes. Displacement control requirements are calculated independently for each segment, considering factors such as load distribution, structural stiffness, and wear condition. The control process follows a gradual principle, with each adjustment not exceeding 20% ​​of the safety limit. Adjusted displacement data is synchronized between segments via a consistency protocol to ensure a smooth transition in the overall displacement distribution. The final generated set of node displacement optimization parameters uses a hierarchical storage structure: the top layer represents the logical segment divisions, the middle layer records the control parameters for each node, and the bottom layer stores historical versions of the adjustment process.

[0048] The generation of the dynamic compensation parameter set begins with the optimization parameter set for nodal displacements. The system first extracts lubricating medium permeation data from the meshing surfaces of the chain links. This data comes from multi-source sensor fusion: a capacitive sensor measures the medium thickness, an infrared sensor detects temperature distribution, and an acoustic sensor monitors flow noise. The permeation rate is calculated using a multiphysics coupling method, simultaneously considering fluid dynamics characteristics, surface roughness, and contact pressure distribution. The modeling of time-load environmental factors employs a state-space method, abstracting the external environment into system state variables, and linking the permeation data with these state variables through observation equations.

[0049] The generation of the dynamic parameter set for permeation and diffusion employs data assimilation techniques. The system constructs a physical model of the lubricating medium's motion, continuously incorporating sensor observation data into the model correction process. The assimilation algorithm uses ensemble Kalman filtering, maintaining multiple parallel-running model instances and continuously adjusting instance weights based on observation data. The dynamic parameter set includes three core parameters: the permeation rate field, the diffusion coefficient matrix, and the interfacial energy gradient, each accompanied by an uncertainty estimate.

[0050] The optimization of medium flow rate and distribution employs an adaptive control method. The system establishes a coupled control model of injection rate and flow path. The model input is the deviation between the current medium concentration distribution and the target distribution, and the output is the rate adjustment at each injection point. The optimization of the path distribution ratio is based on graph theory algorithms, modeling the lubrication channel as a weighted network and determining the optimal allocation scheme through the maximum flow algorithm. The calculation of the adjustment coefficient incorporates a fuzzy PID controller, dynamically adjusting the proportional, integral, and derivative parameters based on historical adjustment effects.

[0051] The dynamic compensation for lubricant concentration distribution employs a prediction-correction dual-cycle mechanism. The prediction cycle extrapolates the distribution state at future time steps based on a physical model, while the correction cycle adjusts prediction biases based on real-time observation data. The calculation of compensation parameters considers effects at three time scales: instantaneous effects reflect the direct impact of medium injection, medium-term effects include the cumulative effects of the diffusion process, and long-term effects consider surface morphology changes caused by wear. The final generated dynamic compensation parameter set is recorded using a spatiotemporal four-dimensional structure, with three spatial dimensions describing the geometric position of the scraper chain and a time dimension recording the parameter evolution process. Each spatiotemporal unit stores the compensation parameter value and its confidence interval, forming a complete dynamic description system of the lubrication state.

[0052] Example 4: Focusing on the generation process of the intervention instruction dataset and the chain link health status assessment table. During the intervention instruction dataset generation stage, the system initiates a real-time monitoring process based on a dynamic compensation parameter set. Taking a mining scraper conveyor as an example, the system deploys triaxial vibration sensors and tension fiber optic sensors in the drive section, the middle trough, and the tail section, respectively. When the equipment processes raw coal with a gangue content of 35%, the monitoring system captures operational data at a sampling frequency of 200Hz. Vibration fluctuation rate analysis uses wavelet packet decomposition technology to decompose the original signal into 32 sub-frequency bands, extracting the energy proportion of the 1-3kHz high-frequency band as a feature value. Tension offset monitoring uses Bragg grating wavelength demodulation, calculating the strain gradient through the difference in center wavelength offset between adjacent gratings.

[0053] Outlier removal employs a multi-level verification mechanism. First, the system checks the sensor's own status. If a sensor's temperature reading exceeds its 0-70℃ operating range or its power supply voltage fluctuates by more than ±10%, the sensor's data is marked as invalid. Second, data logic verification is performed. For example, if the vibration value at the tail of the machine suddenly reaches three times that of the transmission part, while the tension value decreases by 40% year-on-year, the system automatically triggers a data review process: comparing data from five adjacent nodes, querying historical data patterns, and ultimately confirming that the anomaly is caused by a loose sensor connection, it is removed. The average volatility is calculated using a robust statistical method. For data within each 10-second time window, the maximum and minimum values ​​are first removed, and then the moving average of the remaining data is calculated.

[0054] Wear rate influence analysis established a material wear characteristic database. Taking a certain type of 40CrNiMoA alloy steel chain link as an example, the database records the wear coefficient of this material under different working conditions: when the vibration acceleration is 0.5g and the tension fluctuation range is ±5%, the wear coefficient is... When the vibration increases to 0.8g, the coefficient increases to The system matches the current vibration-tension combination with database records in real time and uses bilinear interpolation to calculate the wear coefficient for intermediate states. For new materials, the system initiates an online learning mode: for the first 72 hours, a conservative estimate is used while recording the actual wear amount, gradually refining the prediction model.

[0055] The dynamic adjustment of intervention parameters adopts a graded response strategy. The system divides the scraper chain into 10-meter-long control sections, with each section making independent decisions. When the vibration-tension ratio (the ratio of vibration RMS value to tension fluctuation amplitude) of a certain section is detected to exceed the threshold, a three-level response is executed: the first-level response adjusts the opening of the lubrication nozzle, the second-level response changes the oil injection frequency, and the third-level response adjusts the transmission speed in a coordinated manner. Taking the fifth section of the central tank as an example, if the vibration-tension ratio is monitored to remain at 0.85 for 3 consecutive minutes (threshold 0.7), the system first increases the lubricating medium flow rate from 20ml / min to 28ml / min; after 5 minutes, the ratio drops to 0.75, but is still higher than the threshold, so the oil injection frequency is adjusted from an interval of 10 seconds to an interval of 7 seconds; finally, by adjusting the output of the transmission frequency converter, the operating speed of this section is reduced from 1.2m / s to 1.0m / s, stabilizing the vibration-tension ratio at 0.68.

[0056] The generation of the chain link health status assessment form begins with wear characteristic analysis. The system employs multispectral imaging technology to scan the chain link surface during maintenance windows. Taking the pin portion of a chain link as an example, images in three bands—visible light, near-infrared, and thermal infrared—are acquired. Wear types are identified through feature fusion: normal wear exhibits a uniform texture, fatigue wear displays a network of cracks, and abrasive wear is accompanied by localized high-temperature points. Temporal distribution analysis establishes a wear evolution model. For instance, if the wear depth of a chain link increases by 0.02 mm in the first month of operation, and the rate of increase accelerates to 0.035 mm / month in the second month, the system automatically marks it as an accelerated deterioration state.

[0057] Orthogonal experimental design was used to optimize the monitoring path parameters. The system was configured with three control factors: vibration threshold, time window, and medium concentration, each with three levels (as shown in Table 1). Nine sets of experimental combinations were used to analyze the impact of different parameter combinations on the sensitivity of wear feature identification. Experimental results show that when using a combination of seven parameters, microcrack propagation signs can be detected earliest.

[0058] Table 1: Experimental plan for optimizing monitoring parameters.

[0059]

[0060] The real-time assessment system constructs a multivariate control model. Taking the advancement of a certain working face as an example, the system monitored the scraper chain load increasing from 50 tons to 80 tons, and simultaneously adjusted three key parameters: the vibration alarm threshold was widened from 0.6g to 0.75g to avoid frequent false alarms; the monitoring time window was shortened from 120 seconds to 90 seconds to improve response speed; and the lubricating medium concentration was increased from 18% to 22% to enhance lubrication effect. The assessment table is generated using dynamic template technology, automatically matching assessment standards according to the chain link position type: the tooth wear rate is mainly assessed for drive chain links, deflection changes are monitored for middle chain links, and the wear angle is checked for transition chain links. The final output health status assessment table includes four status indicators: green normal, yellow attention, orange warning, and red alarm, and also includes a trend prediction curve showing the health status evolution trend over the next 72 hours.

[0061] Example 5: Construction process of the topology map covering the entire scraper chain. The topology association analysis module initiates the data processing flow based on the chain link health status assessment table. This module first parses the structural feature data in the assessment table and extracts three sets of key parameters for adjacent chain links: the connection point displacement vector records the three-dimensional displacement change at the chain link hinge, the structural deformation gradient describes the spatial distribution of the material deformation rate, and the load distribution parameter reflects the pressure distribution on the force transmission path. The system uses feature fusion technology to project the three types of parameters onto a unified metric space and quantifies the structural similarity between chain links by calculating the cosine similarity of the feature vectors. The similarity calculation introduces a time decay factor, with recent data having higher weight than historical data. The analysis process pays special attention to the structural correlation on the load transmission path. When an abnormal load distribution is detected in a chain link, the structural parameters of the three upstream and downstream chain links are automatically traced for correlation analysis. The final generated structural correlation parameter set includes chain link pair identifiers, similarity coefficients, and correlation strength indices, forming the topology basis of the chain link network.

[0062] The state mapping module performs health status comparison analysis. This module establishes a health link feature library, storing benchmark parameters from the equipment commissioning and acceptance phase. The comparison process employs a dual verification mechanism: displacement vector consistency verification is achieved by calculating the angular deviation between the current displacement direction and the benchmark direction, with the system setting a displacement tolerance threshold of 0.15 mm; deformation gradient offset analysis uses differential geometry to calculate the Gaussian curvature difference between the actual deformed surface and the ideal surface. The judgment process introduces a fuzzy decision algorithm, using displacement deviation and deformation difference as input variables, and outputting a structural identity probability value. When the probability value is below 85%, an anomaly marker is triggered, and the system automatically records the anomaly's start time, evolution trend, and spatial distribution characteristics. For five consecutive links where the probability value decreases, a transmission path integrity check is initiated to analyze whether the force transmission path has been distorted. The output structural identity judgment parameters use a hierarchical encoding format, including link identifier, anomaly type code, probability value, and verification timestamp.

[0063] The map generation module performs full-link state integration. This module acquires displacement matching results from the structural identity determination parameters and uses vector field reconstruction technology to correct the connection point position mapping relationship. The correction process considers the thermal expansion effect of chain links and adjusts the coordinate transformation matrix in real time based on temperature sensor data. The deformation gradient convergence value processing uses an iterative optimization algorithm, setting a convergence threshold of 0.001 mm / m, and determining the optimal deformation compensation coefficient through five iterations. Three-dimensional dynamic topology modeling uses real-time rendering technology, discretizing the scraper chain into 100,000 computational units. The model construction process integrates three types of data sources: the corrected position mapping relationship determines the node spatial coordinates, the deformation compensation coefficient adjusts the unit deformation parameters, and real-time monitoring data from the health status assessment table provides the basis for color rendering. The model updates at a frequency of 20 frames per second, with each frame containing complete topological relationship data.

[0064] The generation of the end-to-end status monitoring map employs multi-resolution presentation technology. The system supports three observation modes: macro mode displays the overall topology structure, using gradient color bands to indicate the distribution of health status; meso mode focuses on a 10-meter segment, displaying the force transmission path and displacement vector field between links; and micro mode can locate individual links, presenting a 3D reconstructed image of the surface wear morphology. The map integrates an early warning function, automatically generating a flashing marker at the corresponding location when a topology discontinuity or abrupt deformation gradient change is detected. The historical data tracing function supports playback of the topology status at any point in time, with a time resolution of up to 100 milliseconds. The map data uses distributed storage, and metadata records the topology version number, generation time, and data checksum, ensuring the integrity and traceability of status monitoring. The user interface provides a topology relationship navigation tree, which can organize the link network structure according to multiple dimensions such as transmission path, load partition, or health level.

[0065] Example 6: The load prediction module connects to the data interface of the full-link status monitoring map to extract the historical operating load data and corresponding link health status parameters of the scraper chain stored in the map. The historical operating load data covers the instantaneous load values, load change curves, and load durations at different operating stages. The link health status parameters include the wear characteristic distribution, deterioration time records, and structural deformation data of each link. The extracted data is filtered for validity, removing abnormal data caused by sensor failures, data transmission interruptions, etc., and retaining data samples with normal operating status and complete parameter acquisition. The Min-Max normalization method is used to process the filtered sample data, mapping the load data and health status parameters to the [0,1] interval to eliminate the dimensional differences between different parameters. The weight ratio of load influencing factors is determined by the analytic hierarchy process (AHP). Load influencing factors include link wear degree, operating environment temperature, transmission path speed, and historical load peak value. The corresponding weights are assigned according to the correlation between each factor and load changes.

[0066] Normalized sample data is divided into training and test sets in an 8:2 ratio and input into a pre-defined AI prediction model. The pre-defined AI prediction model uses a Long Short-Term Memory (LSTM) neural network. The input layer dimension matches the feature dimension of the normalized samples, with three hidden layers. The number of neurons in each layer is determined using a grid search method. The output layer represents the predicted load value for a pre-defined future time period. The model is iteratively trained using the training set. After each iteration, the prediction error on the test set is calculated, and gradient descent is used to optimize the model's weights and bias parameters. A pre-defined threshold for the prediction error is set. When the prediction error is below this threshold for five consecutive iterations, model training is stopped, and the trained model parameters are saved.

[0067] The system collects real-time operating environment parameters and current load data of the scraper conveyor chain. Operating environment parameters include real-time temperature, humidity, dust concentration, and current load value. Current load data is collected via a tension sensor at a preset frequency. This real-time data undergoes the same normalization process as the sample data and is then input into the trained AI prediction model. Based on the input data, and combining historical load variation patterns with the evolution characteristics of chain link health status, the model calculates predicted load values ​​for different time points within a preset future time period. The preset time period can be set according to actual monitoring needs, with time point intervals divided in minutes. The predicted load values ​​for each time point are integrated as a time series to form a load prediction dataset containing time markers, predicted load values, and prediction error ranges.

[0068] The maintenance reminder module simultaneously receives the load prediction dataset and the chain link health status assessment table, performs data correlation processing on the two, and establishes a correspondence between the predicted load value and the chain link health parameters. Based on the chain link material performance parameters, design service life, and historical operating data, it analyzes the critical threshold of the chain link health status under the predicted peak load. The critical threshold covers the maximum allowable wear, ultimate deformation degree, and maximum tensile strength of the chain link. It compares the current health parameters of each chain link with the corresponding critical threshold, and sets the maintenance priority level according to the magnitude of the difference; the smaller the difference, the higher the priority. The priority levels are divided into three levels: emergency maintenance, priority maintenance, and routine maintenance. For chains of different priority levels, the maintenance time is determined based on the load prediction dataset to avoid maintenance operations during peak load periods; the maintenance location is identified based on the chain link health status assessment table, and the specific chain links that need to be replaced or repaired are marked; maintenance operation suggestions are formulated based on the distribution of lubricating medium and the wear characteristics of the chain links, including lubricating medium replenishment plans, chain link replacement procedures, and structural adjustment methods. The maintenance time, maintenance location, and maintenance operation suggestions are integrated into a maintenance reminder command and pushed to the operation and maintenance terminal through the remote monitoring platform.

[0069] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0070] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A remote monitoring system for the status of a scraper chain based on AI learning, characterized in that, The system includes: The stress distribution monitoring module, based on the scraper chain's operating status information, calls up the peak stress parameters of the chain links, the deformation difference between adjacent chain links, and the transmission path speed data. It analyzes the degree of matching between stress distribution and structural stability, allocates stress control values ​​and transmission balance parameters for chain links, and generates a stress distribution control dataset. Based on the stress distribution control dataset, the anomaly perception module extracts the displacement values ​​and gradient offsets of key nodes in the scraper chain, analyzes the impact of node displacement fluctuations on operational stability, adjusts the uniformity of node displacement distribution, and generates a set of node displacement optimization parameters. Based on the node displacement optimization parameter set, the dynamic compensation module analyzes the penetration and diffusion of lubricating medium on the meshing surface of the chain links, adjusts the lubricating medium injection rate and flow path distribution ratio, redistributes the distribution trend and dynamic parameter values ​​of lubricating medium concentration, and generates a dynamic compensation parameter set. Based on the dynamic compensation parameter set, the learning intervention module extracts the real-time vibration fluctuation rate and tension offset during the operation of the scraper chain, analyzes the influence of the fluctuation range on the wear rate of the chain links, dynamically adjusts the distribution path of the lubricating medium and the vibration-tension ratio within the target range, and generates an intervention instruction dataset. Based on the intervention instruction dataset, the quality optimization module analyzes the distribution ratio and deterioration time of the wear characteristics of the target links, adjusts the parameters of the target monitoring path, and generates a link health status assessment table.

2. The AI-based remote monitoring system for scraper chain status according to claim 1, characterized in that, The specific steps for obtaining the degree of matching between the analytical stress distribution and structural stability are as follows: Based on the scraper chain's operating status information, the peak stress parameters, deformation differences, and transmission path speed data of the chain links are extracted. A time window is set, and time points are selected to match the data. By comparing the data correlation and filtering the data, the peak stress parameters and deformation differences are obtained. Based on the stress peak parameter and deformation difference, the transmission path is matched and verified, the deviation between the stress peak and deformation difference is calculated, the stress distribution and deformation distribution are corrected by combining the rate change, and the path parameters are adjusted by the influence of the rate on the data to obtain the matching situation between the stress peak and deformation. Based on the stress peak and deformation matching, structural stability analysis is performed, structural stability analysis standards are set, and stress distribution under differentiated rate conditions is evaluated in combination with dynamic changes during operation. Stability indices are compared and rate conditions are optimized to obtain the degree of matching between stress distribution and structural stability.

3. The AI-based remote monitoring system for scraper chain status according to claim 2, characterized in that, The specific steps for obtaining the stress distribution control dataset are as follows: Based on the degree of matching between the stress distribution and structural stability, the stress transmission and stability changes of the scraper chain under different operating conditions are analyzed, and the stress distribution of the chain links is weighted and calculated to obtain the preliminary stress control requirements of the chain links. Based on the initial stress control requirements of the links, the stress balance between links is analyzed, the relationship between stress transfer efficiency and load distribution between links is identified, the stress control parameters of the links are corrected, the stress compensation coefficient after the link adjustment is calculated, and the stress control dataset between links is obtained. By combining the stress control dataset between chain links with the structural stability matching results, the stress between chain links is allocated, optimized and matched to the required equilibrium and stability requirements, and a stress distribution control dataset is generated.

4. The AI-based remote monitoring system for scraper chain status according to claim 3, characterized in that, The specific steps for obtaining the displacement values ​​and gradient offsets of key nodes in the scraper chain are as follows: Based on the stress distribution control dataset, deformation data at key nodes of the scraper chain are extracted, deformation points in each time period are screened, and deformation fluctuations are analyzed by combining the variation trend of deformation at different locations at the nodes to obtain deformation data at key nodes. Based on the deformation data at the key nodes, the deformation point and its corresponding displacement value are calculated. By analyzing the relationship between deformation and displacement, the displacement change of each measurement point is identified. Combined with the chain structure parameters, the displacement changes at different locations are compared to obtain displacement distribution and gradient distribution data. Based on the displacement distribution and gradient distribution data, the overall displacement distribution of key nodes of the scraper chain is analyzed, the displacement gradient is optimized by combining deformation data, the impact of displacement changes on operating performance is analyzed, the stable displacement configuration under differentiated operating conditions is determined, and the displacement values ​​and gradient offsets of key nodes are obtained.

5. The AI-based remote monitoring system for scraper chain status according to claim 4, characterized in that, The specific steps for obtaining the nodal displacement optimization parameter set are as follows: Based on the displacement values ​​and gradient offsets of the key nodes, the time series of displacement changes is determined, the current displacement value is compared with the initial displacement data, the displacement gradient at each moment is analyzed, and corresponding thresholds are defined according to the running state partition to generate a preliminary displacement change parameter set. The preliminary displacement change parameter set is analyzed to analyze the impact of key node displacements on operational stability, identify the correlation between displacement and operational status, and calculate the section operational stability influence coefficient. By analyzing the operational stability impact coefficient of the section and combining it with the node displacement change parameters, the displacement distribution balance is adjusted, the displacement control data is optimized, and a set of node displacement optimization parameters is generated.

6. The AI-based remote monitoring system for scraper chain status according to claim 5, characterized in that, The specific steps for obtaining the dynamic compensation parameter set are as follows: Based on the node displacement optimization parameter set, lubricant penetration data of the chain link meshing surface is extracted, the penetration rate of lubricant on the surface of differentiated materials is monitored, and the diffusion characteristics of lubricant are inferred by combining external environmental factors such as time and load. The penetration and diffusion rate coefficients are defined, and a dynamic parameter set of penetration and diffusion is generated. The influence of the aforementioned dynamic parameters of permeation and diffusion on the flow rate and distribution of the medium is analyzed. Based on the requirements of the lubricating medium concentration distribution at the meshing surface, the ratio between the flow path and the medium injection rate is optimized, the adjustment coefficient of the lubricating medium concentration distribution trend is calculated, and the medium concentration control results are generated. Analyze the results of the medium concentration control, adjust the ratio between the lubricating medium injection rate and the flow path, allocate the distribution trend of the lubricating medium concentration, and combine the permeation diffusion parameters and adjustment coefficients to obtain a dynamic compensation parameter set.

7. The AI-based remote monitoring system for scraper chain status according to claim 6, characterized in that, The specific steps for obtaining the generated intervention instruction dataset are as follows: Based on the dynamic compensation parameter set, the vibration fluctuation rate and tension offset during operation are monitored in real time, the fluctuation range is identified, abnormal values ​​of the equipment are eliminated, the average fluctuation rate of the data is analyzed, and vibration and tension fluctuation data are obtained. The influence of the vibration and tension fluctuation range on the wear rate of the chain links is analyzed. Using the known wear rate of the chain links, the relationship between tension and vibration is analyzed, and the wear rate under the different fluctuation range is calculated to obtain the wear rate influence data. Based on the wear rate impact data, the distribution path of the lubricating medium and the vibration ratio within the target range are dynamically adjusted. Adjustments are made based on the relationship between the wear rate impact data and the vibration and tension fluctuation range. The medium flow rate and vibration control range are allocated, and an intervention instruction dataset is generated.

8. The AI-based remote monitoring system for scraper chain status according to claim 7, characterized in that, The specific steps for obtaining the health status assessment table for the generated chain segment are as follows: Based on the intervention instruction dataset, the wear characteristic distribution and deterioration time of the target link are analyzed. Wear characteristic data at different monitoring time points are collected, the characteristic time distribution is sorted out, the data change trend is analyzed and the data is classified to obtain the link wear distribution data. Based on the chain link wear distribution data, the target monitoring path parameters are adjusted, the optimal monitoring time and characteristic distribution of chain link wear are analyzed, and the operating conditions of monitoring vibration threshold, time window and medium concentration are adjusted by comparing the characteristic changes under different monitoring conditions to obtain the target monitoring path parameters. Based on the target monitoring path parameters, the monitoring conditions are adjusted according to the current operating parameters, and the relationship between the variables of monitoring time, vibration threshold, and medium concentration is controlled. Real-time evaluation is performed based on the adjusted parameters to generate a chain link health status assessment table.

9. The AI-based remote monitoring system for scraper chain status according to claim 8, characterized in that, The system also includes: Based on the chain link health status assessment table, the topology association analysis module extracts structural feature data of adjacent chain links, analyzes the structural similarity between associated chain links according to the displacement vector of the connection point between chain links and the structural deformation gradient, and generates structural association parameters in combination with the running load distribution parameters. Based on the structural association parameters, the state mapping module compares the structural feature differences between healthy and abnormal links, and determines the identity of the associated link structures through displacement vector consistency verification and deformation gradient offset analysis, and outputs structural identity determination parameters. Based on the structural identity determination parameters, the map generation module integrates the target monitoring path parameters and chain link health status data to construct a full-link topology map of the scraper chain.

10. The AI-based remote monitoring system for scraper chain status according to claim 9, characterized in that, The map generation module specifically performs the following: Obtain the displacement vector matching result and deformation gradient convergence value of the associated links in the structural identity determination parameters; The positional mapping relationship of the connection points between chain segments is corrected based on the displacement vector matching result, and the chain segment deformation compensation coefficient is adjusted based on the deformation gradient convergence value. By integrating the corrected position mapping relationship, deformation compensation coefficient, and real-time monitoring data from the chain link health status assessment table, a three-dimensional dynamic topology model of the scraper chain is reconstructed, generating a full-link status monitoring map.

11. The AI-based remote monitoring system for scraper chain status according to claim 10, characterized in that, The system also includes: Based on the full-link status monitoring map, the load prediction module extracts historical operating load data of the scraper chain and corresponding chain link health status parameters. It establishes a correlation model between load changes and chain link deterioration rate through AI algorithms, and combines real-time operating environment parameters to predict the peak load of the scraper chain within a preset time period in the future, generating a load prediction dataset.

12. The AI-based remote monitoring system for scraper chain status according to claim 11, characterized in that, The specific steps for obtaining the load prediction dataset are as follows: Based on the health status data of the links and historical load records in the full-link status monitoring map, valid data samples are screened and normalized to determine the weight ratio of load influence factors. The normalized sample data is input into the preset AI prediction model, and the model parameters are optimized through iterative training so that the model prediction error is controlled within the preset threshold. Input real-time environmental parameters and current load data into the trained model, calculate the load prediction values ​​at different time points within a preset future time period, and integrate them to form a load prediction dataset.

13. The AI-based remote monitoring system for scraper chain status according to claim 12, characterized in that, The system also includes: The maintenance reminder module analyzes the critical threshold of the link health status under the predicted load peak based on the load prediction dataset and the link health status assessment table, compares the difference between the current link health parameters and the critical threshold, sets the maintenance priority level, and generates a maintenance reminder instruction that includes maintenance time, maintenance location, and maintenance operation suggestions.

14. The AI-based remote monitoring system for scraper chain status according to claim 13, characterized in that, The maintenance priority levels are set as follows: emergency maintenance, priority maintenance, and routine maintenance.

Images