Belt conveyor intelligent state monitoring system based on sensor arrangement

Abstract

The invention relates to the technical field of conveying equipment, in particular to a belt conveyor intelligent state monitoring system based on sensor arrangement, which comprises a multi-source sensing module, a dynamic fusion module, a field reconstruction module, a feature extraction module, a root cause analysis module and a game decision module, and is used for collecting multi-modal original data flow; calculating an information entropy value of the multi-modal original data stream to determine a dynamic contribution weight; according to the dynamic contribution weight, performing weighted fusion on the modal data to obtain an initial fusion data field; performing space-time interpolation reconstruction on the initial fusion data field to obtain holographic operation field data; mining the holographic operation field data, and extracting an intrinsic mode feature set; performing root cause diagnosis on the intrinsic mode feature set to generate a multi-scene deduction path; calculating the cost and income of the pre-maintenance strategy, generating an optimal risk hedging type pre-maintenance decision sequence, and transmitting the optimal risk hedging type pre-maintenance decision sequence to the terminal; according to the invention, the accuracy of intelligent state monitoring of the belt conveyor based on sensor arrangement can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of conveying equipment, in particular to a belt conveyor intelligent state monitoring system based on sensor arrangement. BACKGROUND

[0002] The prior art has significant deficiencies in the data acquisition and fusion link of belt conveyor state monitoring. Heterogeneous sensor arrays are not deployed according to the three-dimensional spatial topological relationship of the equipment, only a small number of sensors or a single type of sensor are randomly arranged, resulting in incomplete multi-modal raw data stream acquisition and inability to completely capture the dynamic state of key parts such as drive drums and roller groups; the information entropy value of multi-modal data is not calculated to determine the dynamic contribution weight, only simple splicing or average fusion of each modal data is performed, the differentiated contribution of different modal data to the monitoring result is ignored, resulting in low accuracy of the initial fusion data field and inability to provide a reliable data basis for subsequent analysis, making it difficult to adapt to the complex operation monitoring requirements of the belt conveyor.

[0003] The prior art does not perform spatiotemporal interpolation reconstruction on the initial fusion data field, only analyzes based on limited acquisition data, resulting in data missing or blank areas and inability to form a holographic operation field data covering all parts and time sequences; the characteristics of nonlinear dynamic systems are not deeply mined and the intrinsic modal feature set is not extracted, only surface data or a single feature is used for state judgment, making it difficult to accurately identify potential equipment failures; root cause diagnosis is not performed in combination with physical constraints and fault propagation maps, and dynamic game calculation is not used to optimize maintenance strategies, only experience is used to determine fault causes and develop fixed maintenance plans, resulting in inaccurate fault location, lack of pertinence and economy of maintenance strategies, and inability to meet the maintenance requirements of efficient and stable operation of the belt conveyor. SUMMARY

[0004] To achieve the above-mentioned purpose, the belt conveyor intelligent state monitoring system based on sensor arrangement provided by the present application is characterized in that the system comprises a multi-source perception module, a dynamic fusion module, a field reconstruction module, a feature extraction module, a root cause analysis module and a game decision module, wherein:

[0005] The multi-source perception module is used to deploy a heterogeneous sensor array according to the three-dimensional spatial topological relationship of the belt conveyor, so as to drive the acquisition of multi-modal raw data streams of the belt conveyor;

[0006] The dynamic fusion module is used to calculate the information entropy value of the multi-modal raw data streams to determine the dynamic contribution weight of the multi-modal raw data streams in the fusion process; and according to the dynamic contribution weight, each modal data is weighted and fused to obtain the initial fusion data field of the belt conveyor;

[0007] The field reconstruction module is configured to perform spatio-temporal interpolation reconstruction on the initial fusion data field based on the three-dimensional spatial topological relationship of the device, so as to obtain the holographic operation field data of the belt conveyor.

[0008] The feature extraction module is configured to perform nonlinear dynamic system feature mining on the holographic operation field data, so as to extract an intrinsic modal feature set of the belt conveyor.

[0009] The root cause analysis module is configured to perform root cause diagnosis on the intrinsic modal feature set according to the physical constraints and fault propagation graph of the belt conveyor, so as to generate a multi-scenario deduction path of the belt conveyor.

[0010] The game decision module is configured to perform dynamic game calculation on the cost and benefit of the pre-maintenance strategy based on the multi-scenario deduction path, so as to generate an optimal risk hedging type pre-maintenance decision sequence of the belt conveyor and deliver the sequence to the terminal of the belt conveyor.

[0011] In a preferred embodiment, the multi-source perception module is configured to deploy a heterogeneous sensor array according to the three-dimensional spatial topological relationship of the device of the belt conveyor, so as to drive the collection of multi-modal raw data streams of the belt conveyor, specifically for:

[0012] According to the spatial geometric connection relationship of the devices in the belt conveyor, a three-dimensional topological structure model of the belt conveyor is constructed;

[0013] Based on the three-dimensional topological structure model, key monitoring nodes representing the dynamic state of the device in the belt conveyor are determined;

[0014] Physical sensors matched with the physical quantities to be measured are arranged at the key monitoring nodes, and the physical sensors are driven to perform synchronous sampling, so as to obtain the belt deviation, material load distribution, drum speed deviation, and roller force balance of the belt conveyor;

[0015] The belt deviation, the material load distribution, the drum speed deviation, and the roller force balance are integrated as the multi-modal raw data streams of the belt conveyor.

[0016] In a preferred embodiment, the dynamic fusion module is configured to calculate the information entropy value of the multi-modal raw data streams, so as to determine the dynamic contribution weight of the multi-modal raw data streams in the fusion process; and according to the dynamic contribution weight, the modal data is weighted and fused, so as to obtain the initial fusion data field of the belt conveyor, specifically for:

[0017] Single-modal data sequences representing different operating states are parsed from the multi-modal raw data streams;

[0018] within an observation window, determine a distribution probability of data values in the single-modal data sequence;

[0019] according to the distribution probability, calculate an information entropy value of the single-modal data sequence;

[0020] based on the information entropy value, determine a dynamic contribution weight of the belt conveyor;

[0021] according to the dynamic contribution weight, perform weighted superposition fusion on the corresponding single-modal data sequence to obtain a comprehensive data sequence of the belt conveyor;

[0022] organize the comprehensive data sequence into an initial fusion data field with the dynamic contribution weight and covering all monitored physical quantities of the belt conveyor.

[0023] In a preferred embodiment, the calculation formula of the information entropy value is:

[0024] ;

[0025] wherein, is the information entropy value, is a distribution probability of a k-th discrete data value in the single-modal data sequence, is a logarithmic function, is a total number of discrete data values.

[0026] In a preferred embodiment, when the field reconstruction module performs spatio-temporal interpolation reconstruction on the initial fusion data field based on the device three-dimensional spatial topological relationship to obtain holographic operation field data of the belt conveyor, it is specifically used for:

[0027] obtaining spatial coordinates and component connection relationships of a driving drum, a redirection drum, a load roller group, a return roller group and a continuous conveyor belt section in the belt conveyor to construct a device three-dimensional spatial topological relationship of the belt conveyor;

[0028] identifying a source sensor associated with a data point in the initial fusion data field and a corresponding first physical component position, and mapping the data point to the corresponding first physical component position in the device three-dimensional spatial topological relationship;

[0029] in the device three-dimensional spatial topological relationship, for a second physical component position which does not have the data point directly mapped, according to a spatial distance and a connection relationship between the second physical component position and the first physical component position, deriving a supplementary data point corresponding to the second physical component position from the data point mapped from the first physical component position; ​

[0030] correlating the data points and the supplementary data points in time sequence to obtain time series data of the belt conveyor for the same physical component position;

[0031] integrating the data points, the supplementary data points and the time series data in the physical component positions in the three-dimensional spatial topological relationship of the device to generate holographic operation field data of the belt conveyor.

[0032] In a preferred embodiment, when the feature extraction module performs nonlinear dynamic system feature mining on the holographic operation field data to extract the intrinsic modal feature set of the belt conveyor, it is specifically used for:

[0033] According to the mechanical structure parameters and load distribution of the belt conveyor, the system damping and stiffness distribution characteristics of the belt conveyor are determined, and the geometric structure of the initial phase space in the belt conveyor is physically constrained and weighted reconstructed by using the stiffness distribution characteristics to obtain a physically constrained phase space of the belt conveyor.

[0034] In the physically constrained phase space, the dynamic attractors representing stable operating states and typical fault modes in the belt conveyor are identified.

[0035] From the holographic operation field data, a dynamic response component dominated by the dynamic attractors and gathered in the time-frequency domain is separated out.

[0036] According to the principle of physical consistency, the aliasing part of the dynamic response component which is inconsistent with the mechanical law is removed to obtain an intrinsic modal component of the dynamic response component.

[0037] The intrinsic modal component is classified and coded, and integrated into an intrinsic modal feature set of the belt conveyor.

[0038] In a preferred embodiment, when the feature extraction module performs nonlinear dynamic system feature mining on the holographic operation field data to extract the intrinsic modal feature set of the belt conveyor, it is specifically used for:

[0039] According to the stiffness distribution characteristics, the differences in support stiffness of the rollers in the carrying section and the return section of the belt conveyor, and the local concentrated stiffness characteristics at the driving drum and the redirection drum are identified.

[0040] Based on the initial phase space, dominant state variables reflecting transverse vibration, longitudinal vibration and driving torque fluctuation in the belt conveyor are extracted.

[0041] According to the difference in the supporting stiffness of the carrier roller and the local concentrated stiffness characteristics, different physical constraint weights are applied to components in the dominant state variables representing the vibration coupling and torque transmission relationship of different sections, respectively;

[0042] The state variable components with the physical constraint weights are used to determine the weighted Mahalanobis distance between state vectors in the initial phase space to generate a physical constraint phase space of the belt conveyor.

[0043] In a preferred embodiment, when the root cause analysis module performs root cause diagnosis on the intrinsic modal feature set according to the physical constraint and fault propagation graph of the belt conveyor to generate multi-scenario deduction paths of the belt conveyor, it is specifically used for:

[0044] Matching the intrinsic modal feature set with the fault propagation graph to identify potential initial fault nodes, and tracing upstream influencing factors based on the connection relationship in the fault propagation graph to generate a candidate root cause set of the belt conveyor;

[0045] Judging the compliance of the root causes in the candidate root cause set with the device operating boundary conditions and component action logic relationship in the physical constraint, and screening out effective root causes of the candidate root cause set;

[0046] Combining the event propagation rules and logical order in the fault propagation graph, deducing continuous fault evolution stages triggered by the effective root causes that comply with the physical constraint to build multi-scenario deduction paths of the belt conveyor.

[0047] In a preferred embodiment, when the game decision module performs dynamic game calculation on the cost and benefit of the pre-maintenance strategy based on the multi-scenario deduction paths to generate an optimal risk hedging type pre-maintenance decision sequence of the belt conveyor, it is specifically used for:

[0048] Matching the fault evolution scenarios in the multi-scenario deduction paths with a preset maintenance strategy knowledge base to obtain a strategy set of the belt conveyor;

[0049] Evaluating the resource cost attribute and operation risk avoidance benefit attribute of the candidate pre-maintenance strategy in the strategy set under the corresponding scenario;

[0050] Taking the fault evolution scenarios in the multi-scenario deduction paths as decision nodes, the candidate pre-maintenance strategies corresponding to the decision nodes as decision branches, and the next state directed by the decision branches as child nodes, a multi-stage game decision tree of the belt conveyor is constructed;

[0051] The optimal risk-hedging pre-maintenance decision sequence for the belt conveyor is generated by backward induction starting from the end node of the multi-stage game decision tree.

[0052] The optimal risk-hedging pre-maintenance decision sequence for the belt conveyor is generated by backward induction starting from the end node of the multi-stage game decision tree.

[0053] The optimal risk-hedging pre-maintenance decision sequence is delivered to the end of the belt conveyor to ensure the safe transport of the belt conveyor.

[0054] In a preferred embodiment, when the game decision-making module evaluates the resource cost attributes and operational risk aversion benefit attributes of candidate pre-maintenance strategies in the strategy set under the corresponding scenario, it is specifically used for:

[0055] Obtain resource consumption data and risk impact data associated with the candidate pre-maintenance strategies;

[0056] Based on the resource consumption data and the risk impact data, the comprehensive evaluation value of the candidate pre-maintenance strategy is assessed, wherein the formula for calculating the comprehensive evaluation value is:

[0057] ;

[0058] In the formula, The comprehensive evaluation value is... The preset resource cost weighting coefficient, The preset risk aversion benefit weighting coefficient, The baseline resource cost of the belt conveyor. For the first Estimated resource costs for each candidate pre-maintenance strategy For the first Quantitative values ​​of the estimated risk aversion benefits of each candidate pre-maintenance strategy. The benchmark risk avoidance benefit for the belt conveyor;

[0059] The candidate pre-maintenance strategies in the strategy set are sorted according to the comprehensive evaluation value to generate a strategy priority sequence for the belt conveyor.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] 1.The present application provides comprehensive and high-quality data support for belt conveyor state monitoring through multi-source data accurate collection and fusion reconstruction. Heterogeneous sensor arrays are deployed according to the three-dimensional spatial topological relationship of the equipment, and multi-modal raw data streams are synchronously collected to ensure the integrity and spatio-temporal consistency of the monitoring data. The dynamic contribution weight is determined by calculating the data information entropy value, the initial fusion data field is generated by weighted fusion of each modal data, and the data of the unmonitored part is supplemented by spatio-temporal interpolation reconstruction, and the holographic running field data covering all parts and full time sequence are integrated to present the equipment running state completely, which meets the monitoring needs of continuous conveying equipment.

[0062] 2.The present application significantly improves the accuracy of state monitoring and the scientificity of maintenance decision by means of deep feature mining and intelligent decision optimization. Nonlinear dynamic system features are mined from the holographic running field data, intrinsic modal feature sets are extracted, fault root causes are located combined with physical constraints and fault propagation atlas, and multi-scenario deduction paths are generated. The multi-stage game decision tree is constructed based on the deduction path, the cost and benefit of the pre-maintenance strategy are dynamically calculated, the optimal risk hedging type pre-maintenance decision sequence is generated and delivered to the terminal, which provides reliable protection for the stable operation of the equipment and meets the maintenance requirements of continuous conveying equipment for efficient and safe operation. BRIEF DESCRIPTION OF DRAWINGS

[0063] Figure 1 The system architecture diagram of the belt conveyor intelligent state monitoring system based on sensor arrangement provided by an embodiment of the present application is shown.

[0064] The implementation, functional characteristics and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0065] In order to make the purpose, technical scheme and advantages of the embodiments of the present application more clear, the technical scheme of the embodiments of the present application will be described clearly and completely below with reference to the drawings of the embodiments of the present application. Obviously, the described embodiments belong to part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0066] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "said" and "this" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates otherwise. "Multiple" generally includes at least two.

[0067] Depending on the context, the word "if" or "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0068] Furthermore, the timing of the steps in the following method embodiments is merely an example and not a strict limitation.

[0069] In practice, the server-side equipment deployed in a sensor-based intelligent condition monitoring system for belt conveyors may consist of one or more devices. This sensor-based intelligent condition monitoring system can be implemented as a business instance, a virtual machine, or hardware devices. For example, it can be implemented as a business instance deployed on one or more devices in a cloud node. Simply put, it can be understood as software deployed on a cloud node, providing sensor-based intelligent condition monitoring for belt conveyors to various user terminals. Alternatively, it can be implemented as a virtual machine deployed on one or more devices in a cloud node, with application software installed to manage various user terminals. Or, it can also be implemented as a server composed of numerous identical or different types of hardware devices, with one or more hardware devices configured to provide sensor-based intelligent condition monitoring for belt conveyors to various user terminals.

[0070] In terms of implementation, the sensor-based intelligent condition monitoring system for belt conveyors and the user terminal are mutually compatible. Specifically, if the sensor-based intelligent condition monitoring system for belt conveyors is implemented as an application installed on a cloud service platform, then the user terminal acts as a client establishing a communication connection with that application; or if the sensor-based intelligent condition monitoring system for belt conveyors is implemented as a website, then the user terminal acts as a webpage; or if the sensor-based intelligent condition monitoring system for belt conveyors is implemented as a cloud service platform, then the user terminal acts as a mini-program within an instant messaging application.

[0071] like Figure 1 The figure shown is a system architecture diagram of an intelligent condition monitoring system for belt conveyors based on sensor arrangement, provided in an embodiment of the present invention.

[0072] The sensor arrangement-based intelligent state monitoring system 100 of the belt conveyor can be arranged in a cloud server, and in terms of implementation, can be used as one or more service devices, or can be installed on a cloud (such as a server of a mobile service operator, a server cluster, etc.), or can be developed as a website. According to the functions implemented, the sensor arrangement-based intelligent state monitoring system 100 of the belt conveyor can include a multi-source perception module 101, a dynamic fusion module 102, a field reconstruction module 103, a feature extraction module 104, a root cause analysis module 105, and a game decision module 106. The modules of the present application can also be referred to as units, which refer to a series of computer program segments that can be executed by an electronic device processor and can complete a fixed function, and are stored in the memory of the electronic device.

[0073] In the embodiment of the present application, each of the above modules in the sensor arrangement-based intelligent state monitoring system of the belt conveyor can be independently implemented and called by other modules. The calling here can be understood as that a module can be connected to multiple modules of another type and provide corresponding services for the connected multiple modules. In the sensor arrangement-based intelligent state monitoring system of the belt conveyor provided in the embodiment of the present application, the applicable range of the architecture of the sensor arrangement-based intelligent state monitoring system can be adjusted by adding modules and directly calling without modifying program codes, so as to achieve cluster-type horizontal expansion, so as to achieve the purpose of quickly and flexibly expanding the sensor arrangement-based intelligent state monitoring system of the belt conveyor. In actual application, the above modules can be arranged in the same device or different devices, or can be arranged in a virtual device, such as a service instance in a cloud server.

[0074] The following will be described in combination with specific embodiments, respectively for each component and specific working process of the sensor arrangement-based intelligent state monitoring system of the belt conveyor:

[0075] The multi-source perception module 101 is configured to deploy a heterogeneous sensor array according to a three-dimensional spatial topological relationship of the belt conveyor, so as to drive the collection of multi-modal raw data streams of the belt conveyor.

[0076] In the embodiment of the present application, the multi-source perception module deploys a heterogeneous sensor array according to a three-dimensional spatial topological relationship of the belt conveyor, so as to drive the collection of the belt deviation amount, the material load distribution, the roller speed deviation, and the force balance degree of the rollers of the belt conveyor as multi-modal raw data streams of the belt conveyor, and is specifically used for:

[0077] According to the spatial geometric connection relationship of the devices in the belt conveyor, a three-dimensional topological structure model of the belt conveyor is constructed.

[0078] Based on the three-dimensional topological structure model, determine the key monitoring nodes in the belt conveyor that represent the dynamic state of the equipment;

[0079] Lay out physical sensors that match the physical quantities to be measured on the key monitoring nodes, and drive the physical sensors to perform synchronous sampling to obtain the belt deviation, material load distribution, drum speed deviation, and force balance of the rollers of the belt conveyor;

[0080] Integrate the belt deviation, material load distribution, drum speed deviation, and force balance of the rollers as the multi-modal raw data stream of the belt conveyor.

[0081] Collect the overall structural information of the belt conveyor, including the size specifications, installation positions, and connection methods between components of all devices such as the conveyor belt, rollers, drums, driving devices, and tensioning devices. According to the actual spatial geometric connection relationship of each device, integrate the field scene information, structural drawings, and other information to accurately model the entire conveyor and each component, restore the relative positions, connection relationships, and assembly logic of each device in three-dimensional space, and form a three-dimensional topological structure model that completely presents the device layout and connection relationship of the belt conveyor.

[0082] Based on the completed three-dimensional topological structure model, comprehensively analyze the dynamic transmission path and force characteristics of each device during the operation of the belt conveyor. Focus on key positions that directly affect the stability, carrying capacity, and operation safety of the conveyor, including the fitting area of the conveyor belt and the drum, the support nodes of the roller group, the power output end of the driving device, and the adjustment connection of the tensioning device. These positions can reflect the changes in the operating state of the equipment and are determined as key monitoring nodes that represent the dynamic state of the equipment.

[0083] According to the monitoring requirements of each key monitoring node, determine the corresponding physical quantities to be measured, including speed, vibration, temperature, pressure, displacement, and other indicators directly related to the operating state of the equipment. Select physical sensors that accurately match each physical quantity to be measured to ensure that the measurement range, response speed, and other performance parameters of the sensors fully meet the monitoring requirements. These different types of physical sensors are laid out on the corresponding key monitoring nodes. Start all sensors through a unified control signal to ensure that each sensor synchronously collects data under the same time dimension. The collected speed data, vibration data, temperature data, and other types of raw data are integrated to form a multi-modal raw data stream of the belt conveyor.

[0084] The beneficial effect is that the three-dimensional topological structure model of the device is constructed according to the spatial geometric connection relationship of the belt conveyor device, the relative positions and assembly logic of the core devices such as the conveyor belt, the carrier roller and the drum are completely restored, the overall structure layout and component association of the device are clearly presented, accurate spatial reference basis is provided for subsequent sensor deployment and state monitoring, and dead angle-free monitoring coverage is ensured.

[0085] The key monitoring nodes representing the dynamic state of the device are determined based on the three-dimensional topological structure model, focusing on the core positions such as the adhesion place of the conveyor belt and the drum, the support point of the carrier roller and the output end of the driving device, which can directly reflect the dynamic characteristics such as force transmission and vibration fluctuation of the device, ensuring that the monitoring is focused and avoiding redundant deployment of invalid monitoring points.

[0086] Physical sensors matched with the to-be-measured physical quantities are arranged at the key monitoring nodes, and sensor types suitable for different monitoring needs such as rotation speed, vibration and temperature are selected to ensure the pertinence and accuracy of data acquisition; the sensors are driven by a unified control signal to sample synchronously, ensuring the consistency of the data of various modes in the time dimension, and the collected multi-type original data is integrated to form a multi-modal original data stream of the belt conveyor, providing comprehensive, synchronous and high-quality basic data support for subsequent data fusion and state analysis, which meets the engineering actual needs of belt conveyor operation state monitoring.

[0087] The dynamic fusion module 102 is configured to calculate an information entropy value of the multi-modal original data stream to determine a dynamic contribution weight of the multi-modal original data stream in a fusion process, and perform weighted fusion on the data of each mode according to the dynamic contribution weight to obtain an initial fusion data field of the belt conveyor.

[0088] In the embodiment of the present application, when the dynamic fusion module calculates the information entropy value of the multi-modal original data stream to determine the dynamic contribution weight of the multi-modal original data stream in the fusion process, and performs weighted fusion on the data of each mode according to the dynamic contribution weight to obtain the initial fusion data field of the belt conveyor, it is specifically used for:

[0089] Parsing a single-mode data sequence representing different operating states from the multi-modal original data stream;

[0090] Determining a distribution probability of data values in the single-mode data sequence in the observation window;

[0091] Calculating an information entropy value of the single-mode data sequence according to the distribution probability;

[0092] Determining a dynamic contribution weight of the belt conveyor based on the information entropy value;

[0093] Based on the dynamic contribution weight, the corresponding single-modal data sequences are weighted, superimposed, and fused to obtain the comprehensive data sequence of the belt conveyor;

[0094] The integrated data sequence is organized into an initial fused data field with the dynamic contribution weights and covering all monitored physical quantities of the belt conveyor.

[0095] The formula for calculating the information entropy value is:

[0096] ;

[0097] In the formula, The information entropy value, The first in the single-modal data sequence The probability distribution of the occurrence of discrete data values It is a logarithmic function. This represents the total number of discrete data values.

[0098] This paper comprehensively analyzes the core components of a belt conveyor, including key components such as the conveyor belt, rollers, idlers, and material-bearing structures. It precisely clarifies the spatial relationships, connection methods, and assembly logic between these components, such as the contact and transmission relationship between rollers and the conveyor belt, and the support distribution relationship between idlers and the conveyor belt. Based on these spatial geometric connections, a 3D modeling method is used to recreate the spatial layout and interaction of each component, constructing a 3D topological structure model that fully reflects the structure and spatial relationships of the belt conveyor equipment.

[0099] Based on the constructed three-dimensional topological model and combined with the dynamic operation principle of the belt conveyor, the force transmission path and motion state correlation logic of the equipment during operation are analyzed, and the key parts that play a decisive role in the stability of the equipment operation are identified. These parts are the key monitoring nodes that characterize the dynamic state of the equipment, such as the belt edge area prone to deviation, the roller shaft transmission area, the stress concentration area of ​​the idler support, and the key cross-section of the material bearing, to ensure that the key monitoring nodes can fully cover the core monitoring needs of the equipment's dynamic state.

[0100] For each key monitoring node, a matching physical sensor is selected based on the characteristics of the physical quantity to be measured. For example, displacement sensors are deployed at key monitoring nodes along the conveyor belt edge to detect belt misalignment; pressure sensor arrays are deployed at key monitoring nodes along the material bearing section to detect material load distribution; speed sensors are deployed at key monitoring nodes along the roller shaft to detect speed deviation; and force sensors are deployed at key monitoring nodes along the idler supports to detect force balance. These sensors are precisely installed at their respective key monitoring nodes, ensuring that the sensor detection direction and installation accuracy meet the measurement requirements. A synchronous sampling mechanism is then activated, allowing all sensors to collect data simultaneously over time, thereby obtaining the conveyor belt misalignment, material load distribution, roller speed deviation, and idler force balance of the belt conveyor.

[0101] The four types of data collected synchronously—conveyor belt misalignment, material load distribution, drum speed deviation, and idler force balance—are standardized in format. This standardizes the data recording format, timestamp format, and storage specifications, eliminating format differences between different data types. The standardized data is then integrated and sorted according to the acquisition time sequence to form a multimodal raw data stream that comprehensively reflects the operating status of the belt conveyor, encompassing multiple monitoring dimensions. This ensures that the data stream fully preserves the original monitoring information and time-related characteristics of each physical quantity.

[0102] The information entropy value is obtained by calculating the probability distribution of discrete data values ​​in a single-modal data sequence, and then using logarithmic operations to quantify the effective information content carried by the single-modal data sequence.

[0103] The first single-modal data sequence The probability distribution of the occurrence of the first discrete data value is determined after processing the single-modal data sequence within the observation window. The probability distribution of the first discrete data value within that window is statistically analyzed. The proportion of a discrete data value in the sequence is determined by the number of times it appears, combined with the total number of data points in the window.

[0104] The logarithmic function originates from mathematical operations as a tool used to quantify the probability association information of data distribution. By performing a logarithmic transformation on the distribution probability, the product relationship of probabilities is transformed into an additive relationship, which is adapted to the cumulative calculation logic of information entropy value.

[0105] The total number of discrete data values ​​comes from the statistical results of different discrete data values ​​in a single-modal data sequence. The single-modal data sequence within the observation window is traversed, and the number of all non-repeating discrete data values ​​is counted.

[0106] The formula means that by quantifying the distribution characteristics of discrete data values ​​in a single-modal data sequence, the effective information content contained in the sequence can be accurately calculated.

[0107] In the calculation process, the distribution probability of each discrete data value is first determined, then a logarithmic operation is performed on each distribution probability, the operation result is multiplied by the corresponding distribution probability to obtain a negative value, and finally the calculation results corresponding to all discrete data values are accumulated to obtain the sum of information entropy values.

[0108] This calculation method can intuitively reflect the uniformity of the distribution of single-mode data. The more dispersed the distribution, the higher the information entropy value, indicating that the data contains more effective information. The more concentrated the distribution, the lower the information entropy value, indicating that the effective information is less.

[0109] The information entropy value calculated by the formula can be directly used as the core basis for determining the dynamic contribution weight, ensuring that the multi-modal data fusion of the belt conveyor can allocate appropriate weights according to the information content of each modal data, improving the accuracy and effectiveness of the fused data, and meeting the needs of belt conveyor operation state monitoring.

[0110] The beneficial effects are that the single-mode data sequence representing different operating states is parsed from the multi-modal raw data stream, the data is split according to the type of monitored physical quantity, so that each sequence only corresponds to one type of physical quantity, accurately reflecting the running state of the belt conveyor in one aspect, laying a foundation for subsequent targeted analysis, and avoiding confusion and affecting judgment of different types of data.

[0111] The distribution probability of the data values in the single-mode data sequence is determined within the observation window, clearly presenting the distribution characteristics of the data within a specific time range, providing accurate basic data for information entropy value calculation, ensuring the accuracy of subsequent quantitative analysis, and meeting the fine-grained needs of belt conveyor data monitoring.

[0112] According to the distribution probability, the information entropy value of the single-mode data sequence is calculated, the effective information content carried by each sequence is quantified, and the information entropy value directly reflects the data's ability to represent the operating state, providing an objective basis for distinguishing the importance of different modal data and avoiding the deviation caused by subjective judgment.

[0113] Based on the information entropy value, the dynamic contribution weight is determined, so that the single-mode data with high information content and great influence on the monitoring result obtains a higher weight, and vice versa, ensuring the dominant position of core effective data in the fusion process and improving the pertinence and effectiveness of the fused data.

[0114] According to the dynamic contribution weight, the single-mode data sequences are weighted and superimposed, the key data is highlighted and the redundant information is weakened through weight allocation, the different modal data is organically integrated into a comprehensive data sequence, and the overall operation state of the belt conveyor is comprehensively and accurately reflected, avoiding the problem of data disorder caused by simple splicing.

[0115] The integrated data sequence is organized into an initial fusion data field with dynamic contribution weights and covering all monitored physical quantities, which not only retains the weight attribute of each modal data, but also ensures that the data is covered without omission, providing high-quality data support with complete structure and accurate information for subsequent field reconstruction, feature extraction and other links, which meets the engineering practice of intelligent state monitoring of the belt conveyor.

[0116] The information entropy value calculation formula quantifies the distribution characteristics of the single-modal data sequence, accurately measures the effective information content carried by the data, and provides an objective quantitative basis for multi-modal data fusion of the belt conveyor, avoiding the weight allocation deviation caused by subjective judgment.

[0117] The formula takes the distribution probability of each discrete data value in the single-modal data sequence as the core calculation element, and the distribution probability directly reflects the appearance law of the data in the observation window. The probability relationship is converted into an accumulative information quantization form through logarithmic operation, which comprehensively captures the uniformity and discrete characteristics of data distribution.

[0118] By accumulating the product of the probability and the logarithmic probability of all discrete data values and then taking the inverse, the calculation result can intuitively reflect the size of the data information. The higher the information entropy value, the more dispersed the distribution and the richer the sequence containing the running state information, and vice versa, which clearly distinguishes the contribution difference of different modal data to the state monitoring of the belt conveyor.

[0119] The information entropy value calculated by the formula can directly support the determination of dynamic contribution weights, so that the modal data with high information content and strong running state representation ability can obtain higher fusion weights, ensuring that the initial fusion data field can highlight the core effective information and weaken the redundant interference, improving the data basis quality of the belt conveyor state monitoring, and meeting the demand for data accuracy of the intelligent monitoring of the belt conveyor.

[0120] The field reconstruction module 103 is configured to perform spatio-temporal interpolation reconstruction on the initial fusion data field based on the three-dimensional spatial topological relationship of the device to obtain holographic running field data of the belt conveyor.

[0121] In the embodiment of the present application, when the field reconstruction module performs spatio-temporal interpolation reconstruction on the initial fusion data field based on the three-dimensional spatial topological relationship of the device to obtain holographic running field data of the belt conveyor, it is specifically used for:

[0122] Obtaining the spatial coordinates and component connection relationship of the driving drum, the redirection drum, the load roller group, the return roller group and the continuous conveyor belt section in the belt conveyor to construct the three-dimensional spatial topological relationship of the device of the belt conveyor.

[0123] identify the source sensor associated with the data point in the initial fusion data field and the corresponding first physical component position, and map the data point to the corresponding first physical component position in the device three-dimensional spatial topological relationship;

[0124] In the device three-dimensional spatial topological relationship, for the second physical component position which does not have the data point directly mapped, according to the spatial distance and connection relationship between the second physical component position and the first physical component position, the supplementary data point corresponding to the second physical component position is derived from the data point mapped from the first physical component position;

[0125] For the same physical component position, the data point and the supplementary data point are arranged in time sequence to obtain the time sequence data of the belt conveyor;

[0126] Integrate the data points, supplementary data points and time sequence data of the physical component positions in the device three-dimensional spatial topological relationship to generate the holographic operation field data of the belt conveyor.

[0127] Comprehensively collect the geometric information of the core components of the belt conveyor, including the actual installation position coordinates of the driving drum, the redirection drum, the load roller group, the return roller group and the continuous conveying belt section, and sort out the assembly connection mode, the power transmission path and the support cooperation relationship between the components. Systematically integrate these spatial coordinate data and component connection logic, restore the relative positions and association relationships of the components in the three-dimensional space according to the actual physical layout, and construct the device three-dimensional spatial topological relationship which completely presents the structure layout and component association of the belt conveyor.

[0128] Each data point in the initial fusion data field is analyzed one by one, the source sensor corresponding to each data point is determined through data tracing, and then the associated first physical component position is determined according to the layout position of the source sensor, so that each data point can be accurately corresponded to a specific component of the belt conveyor. According to the position coordinates of the physical components in the device three-dimensional spatial topological relationship, each data point is accurately mapped to the corresponding position in the three-dimensional spatial topological model according to the associated first physical component position, so as to realize the spatial binding of the data point and the physical component.

[0129] In the three-dimensional spatial topology of the equipment, all physical component locations are comprehensively examined to identify second physical component locations that are not directly mapped to data points. For each second physical component location, the spatial distance between it and the surrounding first physical component locations with mapped data points is analyzed. Simultaneously, the degree of correlation between the two is determined by considering the connection relationships and power transmission logic between components. From the data points mapped to the closely correlated first physical component locations, supplementary data points that conform to the operational characteristics of the second physical component locations are derived based on spatial distance and the strength of the connection relationship, ensuring that the supplementary data points accurately reflect the actual operating status of that location.

[0130] For the same physical component location of a belt conveyor, all directly mapped data points and derived supplementary data points are collected and sorted according to the chronological order of data acquisition. During the sorting process, it is ensured that the timestamps of each data point correspond accurately to avoid chronological disorder. These time-arranged data are then integrated to form continuous time-series data, fully presenting the changes in the operational status of the physical component location at different time points.

[0131] By spatially integrating data points and supplementary data points at all physical component locations within the equipment's three-dimensional spatial topology, ensuring that each component location has corresponding complete data support, and simultaneously incorporating time-series data from each location, spatial and temporal data are organically combined. Through this integration method, a dataset is formed that covers all physical components of the belt conveyor, encompasses the complete time dimension, and reflects the spatial relationships and dynamic operating status of each component, ultimately generating holographic operating field data for the belt conveyor.

[0132] The beneficial effects are that it obtains the spatial coordinates and component connection relationships of the core components of the belt conveyor and constructs the three-dimensional spatial topology of the equipment, fully restores the spatial layout and assembly logic of key components such as drive rollers and redirecting rollers, provides a precise spatial framework for data mapping and interpolation reconstruction, and ensures that subsequent data processing fits the actual structure of the equipment.

[0133] The source sensor and corresponding first physical component location of data points in the initial fused data field are identified to achieve precise binding between data points and physical components. Then, the data points are mapped to the corresponding positions in the three-dimensional spatial topology, so that the data has clear spatial attributes, laying the foundation for subsequent supplementation of data from unmonitored areas.

[0134] For the location of the second physical component that does not have directly mapped data points, supplementary data points are derived based on its spatial distance and connection relationship with the location of the first physical component to fill in the data gaps, ensuring that all component locations of the device have corresponding data support, avoiding monitoring omissions due to data loss, and improving the completeness of data coverage.

[0135] The data points of the same physical component position and the supplementary data points are arranged in time sequence to obtain time sequence data, the running state changes of the component at different time nodes are fully presented, the data has time sequence continuity, and a basis is provided for analyzing dynamic evolution of the equipment running state.

[0136] The data points of the same physical component position, the supplementary data points, and the time sequence data in the three-dimensional spatial topological relationship of the equipment are integrated, spatial data and time data are organically fused, holographic running field data are generated, the whole component is covered, complete time sequence information is contained, the overall running state of the belt conveyor can be fully and accurately presented, high-quality and all-around data support is provided for subsequent feature extraction and fault diagnosis, and the engineering needs of intelligent state monitoring of the belt conveyor are met.

[0137] The feature extraction module 104 is configured to perform nonlinear dynamic system feature mining on the holographic running field data to extract an intrinsic modal feature set of the belt conveyor.

[0138] In the embodiment of the application, when the feature extraction module performs nonlinear dynamic system feature mining on the holographic running field data to extract an intrinsic modal feature set of the belt conveyor, the feature extraction module is specifically configured to:

[0139] According to mechanical structure parameters and load distribution of the belt conveyor, system damping and stiffness distribution characteristics of the belt conveyor are determined, and the stiffness distribution characteristics are used to physically constrain and weight reconstruct the geometric structure of an initial phase space in the belt conveyor to obtain a physically constrained phase space of the belt conveyor.

[0140] In the physically constrained phase space, dynamic attractors representing stable running states and typical fault modes in the belt conveyor are identified.

[0141] From the holographic running field data, dynamic response components dominated by the dynamic attractors and gathered in the time-frequency domain are separated out.

[0142] According to the principle of physical consistency, the aliasing parts of the dynamic response components that are inconsistent with mechanical laws are removed to obtain intrinsic modal components of the dynamic response components.

[0143] The intrinsic modal components are classified and coded, and are integrated into an intrinsic modal feature set of the belt conveyor.

[0144] When the feature extraction module performs weight reconstruction of the geometric structure of the initial phase space in the belt conveyor by using the stiffness distribution characteristics to obtain the physically constrained phase space of the belt conveyor, the feature extraction module is specifically configured to:

[0145] According to the stiffness distribution characteristics, differences in roller support stiffness of a load-bearing section and a return section in the belt conveyor, and local concentrated stiffness characteristics at a driving drum and a redirection drum are identified;

[0146] Based on the initial phase space, dominant state variables reflecting transverse vibration, longitudinal vibration and driving torque fluctuation in the belt conveyor are extracted;

[0147] According to the differences in roller support stiffness and the local concentrated stiffness characteristics, differentiated physical constraint weights are applied to components of the dominant state variables representing vibration coupling and torque transmission relationships of different sections, respectively;

[0148] Weighted Mahalanobis distances between state vectors in the initial phase space are determined using state variable components of the physical constraint weights to generate a physically constrained phase space of the belt conveyor.

[0149] Mechanical structure information of the belt conveyor is comprehensively collected, including mechanical structure parameters such as drum diameter, roller spacing, conveyor belt thickness, and rack material. Load distribution under different conveying capacities is obtained through actual operation monitoring. Combining these parameters and distribution, deformation capacity and energy consumption characteristics of each component of the belt conveyor are analyzed to determine system damping and stiffness distribution characteristics, where the stiffness distribution characteristics directly reflect the differences in the ability of each component to resist deformation. Based on the stiffness distribution characteristics, the geometric structure of the initial phase space of the belt conveyor is adjusted. Physical constraints are applied to the regions where the core components with greater stiffness and affecting the running stability are located, while the weight proportion of the region in the phase space is increased to strengthen the performance of the kinetic characteristics of the core components, and the physically constrained phase space of the belt conveyor is obtained.

[0150] Based on historical kinetic data of the belt conveyor during normal operation, the kinetic characteristic benchmark under stable operation state is determined, and kinetic performance data corresponding to typical fault modes such as belt deviation, drum jamming, and roller damage are collected to establish a feature comparison library. In the physically constrained phase space, the kinetic trajectories corresponding to the holographic operation field data are traversed, and the trajectories are matched with the benchmark features and fault features in the feature comparison library one by one to identify those kinetic attractors that converge and meet the stable operation characteristic benchmark, and kinetic attractors that present specific abnormal patterns and match the typical fault mode features. These attractors correspond to the stable operation state and the typical fault mode of the belt conveyor, respectively.

[0151] Deeply analyze the time-frequency domain characteristics of the holographic operating field data, divide different time intervals and frequency ranges, and track the influence range of each dynamic attractor in the time-frequency domain. According to the characteristic signals of the attractor, the dynamic data that gathers in the corresponding time-frequency domain range and whose motion law is dominated by the attractor are selected from the holographic operating field data. These data together constitute the dynamic response component dominated by the dynamic attractor, ensuring that the component can accurately reflect the operating state or fault mode represented by the corresponding attractor.

[0152] According to the classical mechanics law and the mechanical motion principle of the belt conveyor, a physical consistency judgment standard is established, including the force balance relationship, the energy conservation law, and the rationality of the motion trajectory. Each data point in the dynamic response component is compared with the judgment standard to check whether the data conforms to the mechanical law. The mixed parts that are inconsistent with the mechanical law, such as force imbalance, abnormal energy loss, and motion trajectory contradiction, are removed, and the effective data that conform to the physical logic are retained to obtain the intrinsic modal component of the dynamic response component.

[0153] All intrinsic modal components are classified and arranged according to their corresponding stable operating states or typical fault modes, such as classifying the modal components corresponding to the normal operation of the conveyor belt into one category and classifying the modal components corresponding to the fault of the idler into another category. Each intrinsic modal component is uniquely coded to clearly define the correspondence between the code and the operating state or fault mode. Then, all classified and coded intrinsic modal components are integrated to form the intrinsic modal feature set of the belt conveyor, which covers the core features of various operating states and fault modes of the belt conveyor.

[0154] Deeply analyze the stiffness distribution characteristics of the belt conveyor, focus on the differences in the structure of the idlers in the carrying section and the return section, and compare the deformation degrees of the idlers in the two sections under the same external force to determine the difference in the support stiffness of the idlers. At the same time, observe the installation and fixing methods, material strength, and stress characteristics of the driving drum and the redirection drum. The driving drum is affected by the driving torque of the motor, and the redirection drum mainly changes the running direction of the conveyor belt. Both of them show the characteristics that the local area stiffness is significantly higher than that of the surrounding components. Thus, the local concentrated stiffness characteristics at the driving drum and the redirection drum are identified.

[0155] The initial phase space of the belt conveyor is analyzed, which contains various variable information reflecting the running state of the equipment. The key variables related to the vibration and torque of the equipment are selected from the initial phase space, wherein the transverse vibration variable corresponds to the vibration of the left and right deviation direction of the conveyor belt, the longitudinal vibration variable corresponds to the vibration of the forward and backward direction of the conveyor belt, and the driving torque fluctuation variable reflects the change state of the torque during the driving process of the motor driving roller. These variables can directly reflect the core running state of the equipment and are extracted as the dominant state variables reflecting the transverse vibration, longitudinal vibration and driving torque fluctuation of the belt conveyor.

[0156] In combination with the identified differences in the support stiffness of the idler and the local concentrated stiffness characteristics, a differentiated physical constraint weight distribution rule is formulated. For the components in the dominant state variables representing the vibration coupling of the load carrying section and the return section, according to the strong and weak differences in the support stiffness of the two sections, the vibration coupling component corresponding to the load carrying section with higher stiffness is given a higher physical constraint weight, and the component corresponding to the return section with lower stiffness is given a lower weight. For the components representing the driving torque transmission relationship, according to the local concentrated stiffness characteristics of the driving roller and the redirection roller, the torque transmission components in the regions close to the two rollers are given higher constraint weights, and the components in other regions are given relatively lower weights, so as to realize differentiated constraint on different components of the dominant state variables.

[0157] The components of the state variables after applying the physical constraint weights are used as the basis for calculation, and the correlation degree between any two state vectors in the initial phase space is analyzed. By considering the physical constraint weights of the components, the higher weight components have a greater influence on the distance result, and the lower weight components have a smaller influence on the distance result when calculating the distance between the state vectors. In this way, the weighted Mahalanobis distance between the state vectors in the initial phase space is determined. Based on these weighted Mahalanobis distances, the geometric structure of the phase space is reconstructed, so that the phase space is more consistent with the actual stiffness distribution and mechanical characteristics of the belt conveyor, and finally the physical constraint phase space of the belt conveyor is generated.

[0158] The beneficial effects are that the mechanical structure parameters and load distribution of the belt conveyor are used to determine the system damping and stiffness distribution characteristics, and the initial phase space is physically constrained and weighted based on the stiffness distribution characteristics, so that the physical constraint phase space is consistent with the actual mechanical characteristics of the equipment, and the dynamics characteristics of the core components are highlighted, laying a foundation for subsequent accurate identification of the running state and fault mode.

[0159] In the physical constraint phase space, the dynamics attractors representing the stable running state and the typical fault mode are identified, the core attractors corresponding to different running states are accurately locked by matching the historical dynamics characteristic benchmarks, the feature confusion caused by the unconstrained phase space is avoided, and the pertinence of state and fault identification is improved.

[0160] The dynamic response component dominated by the dynamic attractor is separated from the holographic operating field data. The effective data gathered in the time and frequency domain is focused and irrelevant interference information is eliminated so that the separated component can accurately reflect the operating state or fault characteristics represented by the corresponding attractor, thus ensuring the accuracy of feature extraction.

[0161] Based on the principle of physical consistency, overlapping parts in the dynamic response components that contradict mechanical laws are eliminated. Valid data are screened based on mechanical criteria such as force balance and energy conservation to ensure that the obtained intrinsic mode components conform to the mechanical motion logic of the belt conveyor and to avoid invalid data affecting subsequent analysis.

[0162] The intrinsic modal components are classified, coded, and integrated into an intrinsic modal feature set. The system is classified according to operating status and fault mode, and the feature meaning and corresponding scenario of each component are clarified to form a feature set with a clear structure and comprehensive coverage. This provides accurate and reliable feature support for subsequent root cause diagnosis, which meets the engineering requirements of intelligent condition monitoring of belt conveyors.

[0163] Identifying the differences in idler support stiffness between the carrying section and the return section, as well as the local concentrated stiffness characteristics of the drive drum and the redirecting drum, based on stiffness distribution characteristics, accurately captures the differences in mechanical properties of the core components of the belt conveyor. This provides a basis for subsequent physical constraint weight allocation that is in line with the actual equipment, avoiding a disconnect between constraint logic and the mechanical laws of the equipment structure.

[0164] Based on the initial phase space, the dominant state variables reflecting lateral vibration, longitudinal vibration and driving torque fluctuations are extracted. This focuses on the core characterization dimensions of the belt conveyor's operating state, eliminates irrelevant variables, and ensures that subsequent constraint and reconstruction work focuses on key dynamic features, thereby improving the pertinence of feature extraction.

[0165] Based on the differences in support stiffness and local concentrated stiffness characteristics of the idler rollers, differentiated physical constraint weights are applied to different components in the dominant state variables. This allows components with high stiffness that affect operational stability to obtain higher constraint strength for their corresponding variable components, thereby enhancing the performance of core dynamic characteristics, weakening the interference of secondary factors, and making the phase space structure more consistent with the actual operating mechanical logic of the equipment.

[0166] By using the state variable components with physical constraint weights to determine the weighted Mahalanobis distance of the state vector in the initial phase space, the distance calculation results can highlight the differences of the core variable components. The physical constraint phase space reconstructed based on this distance can accurately distinguish the dynamic trajectories of stable operation and fault states, laying a high-quality foundation for subsequent dynamic attractor identification and intrinsic mode feature extraction, which meets the needs of intelligent condition monitoring of belt conveyors for accurate capture of mechanical features.

[0167] The root cause analysis module 105 is configured to perform root cause diagnosis on the intrinsic modal feature set according to the physical constraints and the fault propagation graph of the belt conveyor to generate multi-scenario deduction paths of the belt conveyor.

[0168] In the embodiment of the present application, when the root cause analysis module performs root cause diagnosis on the intrinsic modal feature set according to the physical constraints and the fault propagation graph of the belt conveyor to generate multi-scenario deduction paths of the belt conveyor, it is specifically configured to:

[0169] The intrinsic modal feature set is matched with the fault propagation graph to identify a potential initial fault node, and the upstream influencing factors are traced based on the connection relationship in the fault propagation graph to generate a candidate root cause set of the belt conveyor.

[0170] The root cause in the candidate root cause set is judged for compliance with the device operation boundary condition and the component action logic relationship in the physical constraints, and an effective root cause of the candidate root cause set is screened out.

[0171] The continuous fault evolution stage triggered by the effective root cause and conforming to the physical constraints is deduced in combination with the event propagation rules and the logical order in the fault propagation graph to construct multi-scenario deduction paths of the belt conveyor.

[0172] The running state and the fault feature corresponding to each type of modal component in the intrinsic modal feature set are comprehensively sorted out, and each type of fault mode and the corresponding feature performance recorded in the fault propagation graph are analyzed. The features of each modal component are compared with the fault modes in the fault propagation graph one by one, and the potential initial fault node with the highest matching degree with the modal features is accurately identified, which is the core position where the fault may originate. According to the connection relationship and the propagation path between the nodes in the fault propagation graph, the initial fault node is traced upstream in reverse, and all related factors that may affect the fault of the node are investigated, including component wear, load abnormality, connection loosening, etc. These factors are systematically integrated to form a candidate root cause set of the belt conveyor.

[0173] The device operation boundary conditions of the belt conveyor in the physical constraint are determined, including the rated conveying capacity, the allowable running speed, the temperature and humidity range, and other limited conditions for normal operation of the device. Meanwhile, the action logic relationship between each component is sorted out, such as the transmission logic of the driving roller and the conveyor belt, the support and cooperation logic of the roller and the conveyor belt, etc. For each root cause in the candidate root cause set, it is checked whether it meets the device operation boundary conditions, and it is judged whether the root cause can occur within the normal working range of the device. At the same time, it is verified whether the root cause matches the component action logic relationship, and it is confirmed whether the root cause will break the normal action cooperation between components. Through this double compliance judgment, invalid root causes that do not meet the constraint conditions are eliminated, and effective root causes of the candidate root cause set are screened out.

[0174] The event propagation rules in the fault propagation graph are analyzed in depth, including the triggering conditions, propagation speed, and influence degree of the fault from one component to another component, and the logical order of the occurrence of each fault event is determined, so as to ensure that the deduction process conforms to the objective logic of fault development. Taking the effective root cause as the starting point, combining the event propagation rules, the subsequent fault events caused by the root cause are deduced in turn, each fault event needs to meet the boundary conditions in the physical constraint and the component action logic, forming a continuous and actual fault evolution stage. For different effective root causes and the possible evolution paths of the same root cause under different operating conditions, corresponding fault scenario deduction chains are respectively constructed, and finally all reasonable deduction chains are integrated to form the multi-scenario deduction path of the belt conveyor.

[0175] The beneficial effects are that the intrinsic modal feature set is matched with the fault propagation graph for fault mode matching, the potential initial fault node is accurately located, the upstream influencing factors are traced back in reverse through the connection relationship of the fault propagation graph, all kinds of inducements that may cause faults are covered comprehensively, a complete candidate root cause set is generated, and the omission of fault root causes is avoided, laying a foundation for subsequent accurate diagnosis.

[0176] The candidate root cause set is judged for compliance with the device operation boundary conditions and the component action logic relationship in the physical constraint, and invalid root causes that exceed the device operation range and are contradictory to the component action logic are eliminated, so as to ensure that the effective root causes selected conform to the actual operation rules and mechanical structure characteristics of the belt conveyor, and the accuracy of root cause diagnosis is improved.

[0177] The event propagation rules and logical order of the fault propagation graph are combined, and the continuous fault evolution stage is deduced from the effective root cause as the starting point, and all evolution processes strictly follow the physical constraint, the fault development paths corresponding to different root causes are completely presented, and the multi-scenario deduction path constructed can comprehensively cover all kinds of fault scenarios, providing accurate and comprehensive scenario support for subsequent preventive maintenance strategy making, which meets the engineering actual needs of fault diagnosis and maintenance of the belt conveyor.

[0178] The game decision module 106 is configured to perform dynamic game calculation on the cost and benefit of the pre-maintenance strategy based on the multi-scenario deduction path, so as to generate an optimal risk hedging type pre-maintenance decision sequence of the belt conveyor.

[0179] In the embodiment of the present application, when the game decision module performs the dynamic game calculation on the cost and benefit of the pre-maintenance strategy based on the multi-scenario deduction path, so as to generate an optimal risk hedging type pre-maintenance decision sequence of the belt conveyor, it is specifically configured to:

[0180] match the failure evolution scenario in the multi-scenario deduction path with a preset maintenance strategy knowledge base to obtain a strategy set of the belt conveyor;

[0181] evaluate the resource cost attribute and the operation risk avoidance benefit attribute of the candidate pre-maintenance strategy in the strategy set under the corresponding scenario;

[0182] use the failure evolution scenario in the multi-scenario deduction path as a decision node, use the candidate pre-maintenance strategy corresponding to the decision node as a decision branch, and use the next state directed by the decision branch as a sub-node, to construct a multi-stage game decision tree of the belt conveyor;

[0183] begin to induce reversely from the end node of the multi-stage game decision tree, so as to generate an optimal risk hedging type pre-maintenance decision sequence of the belt conveyor;

[0184] begin to induce reversely from the end node of the multi-stage game decision tree, so as to generate an optimal risk hedging type pre-maintenance decision sequence of the belt conveyor;

[0185] deliver the optimal risk hedging type pre-maintenance decision sequence to the terminal of the belt conveyor, so as to realize the conveying safety of the belt conveyor.

[0186] When the game decision module performs the evaluation on the resource cost attribute and the operation risk avoidance benefit attribute of the candidate pre-maintenance strategy in the strategy set under the corresponding scenario, it is specifically configured to:

[0187] obtain resource consumption data and risk influence data associated with the candidate pre-maintenance strategy;

[0188] evaluate a comprehensive evaluation value of the candidate pre-maintenance strategy according to the resource consumption data and the risk influence data, wherein the calculation formula of the comprehensive evaluation value is:

[0189] ;

[0190] In the formula, the comprehensive evaluation value is ​a preset resource cost weight coefficient, a preset operation risk avoidance benefit weight coefficient, a reference resource cost of the belt conveyor, a first candidate preventive maintenance strategy, a first candidate preventive maintenance strategy, a first candidate preventive maintenance strategy, a first candidate preventive maintenance strategy, a reference risk avoidance benefit of the belt conveyor;

[0191] According to the comprehensive evaluation value, the candidate preventive maintenance strategies in the strategy set are sorted to generate a strategy priority sequence of the belt conveyor.

[0192] The system sorts all fault evolution scenarios contained in the multi-scenario deduction path, and clearly defines the core characteristics of each scenario, such as fault type, development stage, and influence range. The preset maintenance strategy knowledge base is called, which covers various preventive maintenance schemes for different fault scenarios of the belt conveyor, including component replacement, regular maintenance, and state monitoring enhancement. The core characteristics of each fault evolution scenario are accurately matched with the applicable conditions of the preventive maintenance strategies in the knowledge base, and all preventive maintenance schemes suitable for each scenario are selected and integrated to form the strategy set of the belt conveyor.

[0193] For each candidate preventive maintenance strategy in the strategy set, the resource cost attribute is comprehensively calculated in combination with the corresponding fault evolution scenario. The resource cost includes all direct and indirect expenses required for executing the strategy, such as manpower input, spare parts procurement cost, equipment downtime loss, and tool usage cost, ensuring that the cost calculation covers the entire process of strategy execution. At the same time, the operation risk avoidance benefit attribute of the candidate preventive maintenance strategy is evaluated, and the benefits that can be avoided after the implementation of the strategy, such as fault expansion loss, reduced maintenance cost, and improved equipment operation efficiency, are analyzed to determine the actual value of the strategy in risk prevention and control.

[0194] Each fault evolution scenario in the multi-scenario deduction path is taken as an independent decision node, and each decision node corresponds to a specific fault development state. The candidate preventive maintenance strategy corresponding to each decision node is taken as the decision branch of the node, and each branch represents a specific scheme for dealing with the current fault scenario. According to the fault evolution law and the execution effect of the preventive maintenance strategy, the next operation state of the equipment after the implementation of each decision branch is determined, and this state is taken as the child node of the decision node. The child node may be a stable state after the fault is controlled, or a new scenario node of the fault continuous development, and a multi-stage game decision tree of the belt conveyor is constructed through the association of nodes and branches.

[0195] Starting from the end nodes of the multi-stage game decision tree, the end nodes represent the final state of fault evolution or the stable end of equipment operation. For each end node, the comprehensive benefit of the decision branch is judged by weighing the resource cost attribute and the operation risk avoidance benefit attribute corresponding to the branch in combination with its corresponding pre-branch. According to the order from the end to the starting node, the comprehensive benefit of all decision branches of each decision node is evaluated in turn, and the decision branch with the optimal comprehensive benefit at each node is selected. All optimal decision branches are connected in the order of time and logical relationship of fault evolution to form a continuous and coherent scheme that can balance cost and benefit to the greatest extent and effectively hedge against operational risks, and finally generate the optimal risk hedging type preventive maintenance decision sequence of the belt conveyor.

[0196] Retrieve the maintenance resource files and risk monitoring records of the belt conveyor, and collect various data directly related to the candidate preventive maintenance strategy. Resource consumption data covers all direct and indirect consumption information such as manpower allocation, spare parts specifications and quantity, tool usage wear, production loss corresponding to equipment downtime, etc. required to execute the strategy; risk impact data includes the range of possible expansion of the fault if the strategy is not executed, the degree of equipment damage caused, the additional cost of subsequent maintenance, the impact of the interruption on the overall conveying process, etc. Risk-related information ensures that the two types of data are comprehensive and accurately correspond to the candidate preventive maintenance strategy.

[0197] Clearly define the evaluation dimensions of resource consumption data and risk impact data. Resource consumption data measures cost from the perspectives of expenditure size and resource scarcity, etc. Risk impact data judges risk from the perspectives of loss degree and impact range, etc. Convert resource consumption data into cost evaluation results, the less the consumption, the better the cost evaluation results; convert risk impact data into benefit evaluation results, the greater the risk that the strategy can avoid, the better the benefit evaluation results. Consider the cost evaluation results and benefit evaluation results, combine the operation priority and maintenance target of the belt conveyor, and reasonably weigh the two to finally get the comprehensive evaluation value of the candidate preventive maintenance strategy.

[0198] Take the comprehensive evaluation value as the core sorting basis, and develop clear sorting rules. The higher the comprehensive evaluation value, the better the cost-benefit ratio and risk avoidance effect of the candidate preventive maintenance strategy, and the higher the sorting; the lower the comprehensive evaluation value, the lower the sorting. According to this rule, all candidate preventive maintenance strategies in the strategy set are systematically sorted to ensure that the sorting results can intuitively reflect the comprehensive advantages of each strategy, and finally generate the strategy priority sequence of the belt conveyor, providing clear strategy selection basis for subsequent game decision.

[0199] The comprehensive evaluation value is derived by combining the resource cost weighting coefficient and the operational risk aversion benefit weighting coefficient. The cost saving ratio and risk aversion benefit ratio of the candidate pre-maintenance strategies are quantitatively calculated, and the final value reflects the comprehensive advantages of the strategy.

[0200] The preset resource cost weighting coefficient is based on the maintenance and management objectives of the belt conveyor, the scarcity of resources, and the priority of cost control. It is determined by technical personnel in combination with industry maintenance experience and actual equipment operation needs, and is used to adjust the degree of influence of resource cost factors in the comprehensive evaluation.

[0201] The preset operational risk aversion benefit weighting coefficient is based on the failure impact range of the belt conveyor, operational safety requirements, and production continuity needs, and is determined with reference to the risk control experience of similar equipment and the risk tolerance of enterprises. It is used to balance the proportion of risk aversion benefit factors in the comprehensive evaluation.

[0202] The baseline resource cost of a belt conveyor is derived from the average total resource consumption of the belt conveyor during its normal operating cycle without the adoption of specific pre-maintenance strategies. This includes routine maintenance costs, basic costs for spare parts replacement, fixed labor input costs, etc., and is determined as a baseline value after system accounting.

[0203] No. The estimated resource cost of each candidate pre-maintenance strategy is derived from the specific implementation process of the candidate pre-maintenance strategy. It calculates all resource consumption, such as manpower, spare parts, tools, and downtime losses, required during the execution process, and obtains the value through accurate estimation by combining market prices and equipment operation data.

[0204] No. The estimated risk aversion benefit quantification value of each candidate pre-maintenance strategy comes from analyzing the failure losses that can be avoided after the strategy is implemented, including equipment maintenance costs, production interruption losses, and manual repair costs caused by the failure escalation. The result is obtained by converting these avoided losses into quantitative values.

[0205] The benchmark risk avoidance benefit for belt conveyors is determined by setting the maximum risk avoidance benefit that the belt conveyor can achieve under ideal maintenance conditions. This is based on the benefit data of operating without major failures throughout the equipment's entire life cycle and the risk control effect of the optimal maintenance plan.

[0206] The significance of the formula is to comprehensively evaluate the actual value of each strategy by quantifying the cost-saving effect and risk-avoidance benefit of candidate pre-maintenance strategies, combined with preset weight coefficients, and to provide an objective basis for prioritizing pre-maintenance strategies.

[0207] In the calculation process, the difference between the baseline resource cost and the estimated resource cost of the candidate pre-maintenance strategy is first calculated, and then the difference is compared with the baseline resource cost to obtain the cost saving ratio; at the same time, the ratio of the estimated risk aversion benefit of the candidate pre-maintenance strategy to the baseline risk aversion benefit is calculated to obtain the risk aversion benefit ratio.

[0208] Multiply the cost saving ratio by the resource cost weighting coefficient, multiply the risk aversion benefit ratio by the operational risk aversion benefit weighting coefficient, and then add the two products together to get the total value.

[0209] This calculation method considers both the cost control effect of pre-maintenance strategies and the benefits of risk prevention and control. By balancing the influence of the two through weighting coefficients, it ensures that the comprehensive evaluation value can fully and objectively reflect the comprehensive advantages of the strategy, which meets the needs of pre-maintenance decision-making for belt conveyors for balancing costs and benefits, and provides reliable support for generating strategy priority sequences.

[0210] The beneficial effect is that by matching the fault evolution scenarios in the multi-scenario simulation path with the preset maintenance strategy knowledge base, pre-maintenance solutions suitable for different fault scenarios are accurately selected, forming a strategy set for the belt conveyor, ensuring that each fault scenario has a targeted response strategy, and avoiding the disconnect between maintenance solutions and actual needs.

[0211] The resource cost attributes and operational risk avoidance benefit attributes of candidate pre-maintenance strategies in the evaluation strategy set are assessed. The costs of manpower, spare parts, downtime losses, etc. are fully calculated. At the same time, the failure losses and efficiency improvement benefits avoided after the strategy is implemented are quantified. This provides an objective and comprehensive quantitative basis for subsequent game theory calculations, ensuring that the decision takes into account both economic efficiency and risk control effectiveness.

[0212] A multi-stage game decision tree is constructed using fault evolution scenarios as decision nodes, candidate pre-maintenance strategies as decision branches, and branch-oriented states as sub-nodes. This clearly presents the fault development path and final result corresponding to different maintenance strategies, making the comparison of the advantages and disadvantages of each strategy intuitive and providing structured support for optimal decision selection.

[0213] By reverse induction from the terminal nodes of the multi-stage game decision tree, the comprehensive benefits of each decision branch are evaluated one by one. The decision branches with the optimal balance between cost and benefit and the best risk hedging effect at each node are selected. These optimal branches are then connected in logical order to generate the optimal risk hedging type pre-maintenance decision sequence for the belt conveyor. This ensures that the maintenance decisions not only conform to the evolution of equipment failure, but also minimize operational risks and control maintenance costs, thus meeting the maintenance requirements for the efficient and stable operation of the belt conveyor.

[0214] The resource consumption data and risk impact data associated with the candidate pre-maintenance strategy are acquired, overall covering the cost information such as manpower, spare parts, downtime loss and the like required for strategy execution, and the risk information such as loss of fault expansion when the strategy is not executed, and additional cost of maintenance, to provide comprehensive and practical data for the maintenance of the belt conveyor for the comprehensive evaluation.

[0215] The candidate pre-maintenance strategy is quantitatively evaluated by the comprehensive evaluation value calculation formula, the preset resource cost weight coefficient and the operation risk avoidance benefit weight coefficient are combined, the importance of cost control and risk prevention and control is balanced, the benchmark resource cost and the benchmark risk avoidance benefit are taken as references, the cost saving proportion and the risk avoidance benefit proportion are accurately calculated, and the comprehensive evaluation value can objectively reflect the comprehensive benefit of the strategy.

[0216] The candidate pre-maintenance strategy is sorted according to the comprehensive evaluation value, the higher the comprehensive evaluation value is, the better the cost benefit ratio and the risk avoidance effect of the strategy are, the earlier the sorting is, the advantages and disadvantages of each strategy are clearly presented, the strategy priority sequence of the belt conveyor is generated, the strategy selection basis is provided for subsequent game decision tree construction and optimal pre-maintenance decision sequence generation, the scientificity and pertinence of the maintenance decision are ensured, and the engineering requirement of efficient maintenance of the belt conveyor is met.

[0217] It is obvious for those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and the present application can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application.

[0218] The embodiments of the present application can acquire and process related data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology and application system for simulating, extending and expanding human intelligence by using a digital computer or a machine controlled by a digital computer, perceiving the environment, acquiring knowledge and using the knowledge to obtain the best results.

[0219] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present application.

Claims

1. A sensor-based intelligent condition monitoring system for belt conveyors, characterized in that, The system includes a multi-source sensing module, a dynamic fusion module, a field reconstruction module, a feature extraction module, a root cause analysis module, and a game-theoretic decision-making module, wherein: The multi-source sensing module is used to deploy a heterogeneous sensor array according to the three-dimensional spatial topology of the belt conveyor to drive the collection of the belt belt deviation, material load distribution, drum speed deviation, and idler force balance of the belt conveyor as the multimodal raw data stream of the belt conveyor. The dynamic fusion module is used to calculate the information entropy value of the multimodal raw data stream to determine the dynamic contribution weight of the multimodal raw data stream in the fusion process; and to perform weighted fusion of each modal data according to the dynamic contribution weight to obtain the initial fusion data field of the belt conveyor. The field reconstruction module is used to perform spatiotemporal interpolation reconstruction on the initial fused data field based on the three-dimensional spatial topological relationship of the device to obtain the holographic operation field data of the belt conveyor. The feature extraction module is used to perform nonlinear dynamic system feature mining on the holographic operating field data in order to extract the intrinsic mode feature set of the belt conveyor; The root cause analysis module is used to perform root cause diagnosis on the intrinsic mode feature set based on the physical constraints and fault propagation map of the belt conveyor, so as to generate a multi-scenario inference path for the belt conveyor. The game decision module is used to perform dynamic game calculations on the cost and benefit of the pre-maintenance strategy based on the multi-scenario deduction path, so as to generate the optimal risk-hedging pre-maintenance decision sequence of the belt conveyor and deliver it to the terminal of the belt conveyor.

2. The intelligent status monitoring system for belt conveyors based on sensor arrangement as described in claim 1, characterized in that, The multi-source sensing module deploys a heterogeneous sensor array according to the three-dimensional spatial topology of the belt conveyor to drive the acquisition of the conveyor belt deviation, material load distribution, drum speed deviation, and idler force balance, which serve as the multimodal raw data stream of the belt conveyor. Specifically, it is used for: Based on the spatial geometric connection relationships of the devices in the belt conveyor, a three-dimensional topological model of the belt conveyor is constructed. Based on the three-dimensional topology model, the key monitoring nodes characterizing the dynamic state of the equipment in the belt conveyor are determined; Physical sensors matching the physical quantities to be measured are deployed at the key monitoring nodes, and the physical sensors are driven to perform synchronous sampling to obtain the belt deviation, material load distribution, drum speed deviation, and idler force balance of the belt conveyor. The conveyor belt deviation, material load distribution, drum speed deviation, and idler force balance are integrated as the multimodal raw data stream of the belt conveyor.

3. The intelligent status monitoring system for belt conveyors based on sensor arrangement as described in claim 1, characterized in that, When the dynamic fusion module calculates the information entropy value of the multimodal raw data stream to determine the dynamic contribution weight of the multimodal raw data stream in the fusion process, and performs weighted fusion of each modal data according to the dynamic contribution weight to obtain the initial fused data field of the belt conveyor, it is specifically used for: Single-modal data sequences representing different operating states are parsed from the multimodal raw data stream; Within the observation window, determine the probability distribution of data values ​​in the single-modal data sequence; Calculate the information entropy value of the single-modal data sequence based on the distribution probability; Based on the information entropy value, the dynamic contribution weight of the belt conveyor is determined; Based on the dynamic contribution weight, the corresponding single-modal data sequences are weighted, superimposed, and fused to obtain the comprehensive data sequence of the belt conveyor; The integrated data sequence is organized into an initial fused data field with the dynamic contribution weights and covering all monitored physical quantities of the belt conveyor.

4. The intelligent status monitoring system for belt conveyors based on sensor arrangement as described in claim 3, characterized in that, The formula for calculating the information entropy value is: ； In the formula, The information entropy value, The first in the single-modal data sequence The probability distribution of the occurrence of discrete data values It is a logarithmic function. This represents the total number of discrete data values.

5. The intelligent status monitoring system for belt conveyors based on sensor arrangement as described in claim 1, characterized in that, When the field reconstruction module performs spatiotemporal interpolation reconstruction of the initial fused data field based on the three-dimensional spatial topology of the device to obtain the holographic operating field data of the belt conveyor, it is specifically used for: The spatial coordinates and component connection relationships of the drive roller, redirecting roller, carrying idler group, return idler group and continuous conveyor belt section in the belt conveyor are obtained to construct the three-dimensional spatial topology of the belt conveyor. Identify the source sensor and corresponding first physical component location associated with the data points in the initial fused data field, and map the data points to the corresponding first physical component location in the three-dimensional spatial topology of the device; In the three-dimensional spatial topology of the device, for the location of the second physical component that is not directly mapped to the data point, supplementary data points corresponding to the second physical component location are derived from the data points mapped to the first physical component location based on the spatial distance and connection relationship between the second physical component location and the first physical component location; For the same physical component location, the data points and the supplementary data points are associated and arranged in chronological order to obtain the time series data of the belt conveyor; By integrating the data points, supplementary data points, and time series data at the physical component locations in the three-dimensional spatial topology of the equipment, holographic operating field data of the belt conveyor is generated.

6. The intelligent status monitoring system for belt conveyors based on sensor arrangement as described in claim 1, characterized in that, When the feature extraction module performs nonlinear dynamic system feature mining on the holographic operating field data to extract the intrinsic mode feature set of the belt conveyor, it is specifically used for: Based on the mechanical structure parameters and load distribution of the belt conveyor, the system damping and stiffness distribution characteristics of the belt conveyor are determined, and the geometric structure of the initial phase space in the belt conveyor is physically constrained and weighted using the stiffness distribution characteristics to obtain the physical constraint phase space of the belt conveyor. In the physical constraint phase space, identify the dynamic attractors that characterize the stable operating state and typical failure modes in the belt conveyor; The dynamic response components, which are dominated by the dynamic attractor and clustered in the time-frequency domain, are separated from the holographic operating field data. Based on the principle of physical consistency, the overlapping parts in the dynamic response components that contradict the laws of mechanics are removed to obtain the eigenmode components of the dynamic response components. The intrinsic modal components are classified and encoded, and integrated into the intrinsic modal feature set of the belt conveyor.

7. The intelligent status monitoring system for belt conveyors based on sensor arrangement as described in claim 6, characterized in that, When the feature extraction module performs physical constraint and weighted reconstruction of the geometry of the initial phase space in the belt conveyor using the stiffness distribution characteristics to obtain the physically constrained phase space of the belt conveyor, it is specifically used for: Based on the stiffness distribution characteristics, the difference in idler support stiffness between the carrying section and the return section of the belt conveyor, as well as the local concentrated stiffness characteristics at the drive drum and the redirecting drum, are identified. Based on the initial phase space, the dominant state variables reflecting the lateral vibration, longitudinal vibration and driving torque fluctuations in the belt conveyor are extracted. Based on the difference in support stiffness of the idler rollers and the characteristics of local concentrated stiffness, differentiated physical constraint weights are applied to the components of the dominant state variables that characterize the vibration coupling and torque transmission relationship in different sections. Using the state variable components of the physical constraint weights, the weighted Mahalanobis distance between the state vectors in the initial phase space is determined to generate the physical constraint phase space of the belt conveyor.

8. The intelligent status monitoring system for belt conveyors based on sensor arrangement as described in claim 1, characterized in that, When the root cause analysis module performs root cause diagnosis on the intrinsic mode feature set based on the physical constraints and fault propagation map of the belt conveyor to generate a multi-scenario deduction path for the belt conveyor, it is specifically used for: The intrinsic modal feature set is matched with the fault propagation map to identify potential initial fault nodes, and the upstream influencing factors are traced based on the connection relationship in the fault propagation map to generate the candidate root cause set of the belt conveyor. The root causes in the candidate root cause set are compared with the equipment operation boundary conditions and component action logic relationships in the physical constraints to determine the validity of the candidate root cause set. By combining the event propagation rules and logical sequence in the fault propagation graph, the continuous fault evolution stages triggered by the effective root cause and conforming to the physical constraints are deduced, so as to construct the multi-scenario deduction path of the belt conveyor.

9. The intelligent status monitoring system for belt conveyors based on sensor arrangement as described in claim 1, characterized in that, When the game decision-making module performs dynamic game calculations on the costs and benefits of the pre-maintenance strategy based on the multi-scenario deduction path to generate the optimal risk-hedging pre-maintenance decision sequence for the belt conveyor and delivers it to the terminal of the belt conveyor, it is specifically used for: The fault evolution scenarios in the multi-scenario simulation path are matched with the preset maintenance strategy knowledge base to obtain the strategy set of the belt conveyor. Evaluate the resource cost attributes and operational risk avoidance benefit attributes of the candidate pre-maintenance strategies in the strategy set under the corresponding scenarios; Using the fault evolution scenario in the multi-scenario simulation path as the decision node, the candidate pre-maintenance strategy corresponding to the decision node as the decision branch, and the next state led to after the decision branch is made as the child node, a multi-stage game decision tree of the belt conveyor is constructed. The optimal risk-hedging pre-maintenance decision sequence for the belt conveyor is generated by backward induction starting from the end node of the multi-stage game decision tree. The optimal risk-hedging pre-maintenance decision sequence is delivered to the end of the belt conveyor to ensure the safe transport of the belt conveyor.

10. The intelligent status monitoring system for belt conveyors based on sensor arrangement as described in claim 9, characterized in that, When the game decision-making module evaluates the resource cost attributes and operational risk aversion benefit attributes of candidate pre-maintenance strategies in the strategy set under the corresponding scenario, it is specifically used for: Obtain resource consumption data and risk impact data associated with the candidate pre-maintenance strategies; Based on the resource consumption data and the risk impact data, the comprehensive evaluation value of the candidate pre-maintenance strategy is assessed, wherein the formula for calculating the comprehensive evaluation value is: ； In the formula, The comprehensive evaluation value is... The preset resource cost weighting coefficient, The preset risk aversion benefit weighting coefficient, This is the baseline resource cost of the belt conveyor. For the first Estimated resource costs for each candidate pre-maintenance strategy For the first Quantitative values ​​of the estimated risk aversion benefits of each candidate pre-maintenance strategy. The benchmark risk avoidance benefit for the belt conveyor; The candidate pre-maintenance strategies in the strategy set are sorted according to the comprehensive evaluation value to generate a strategy priority sequence for the belt conveyor.

Images