A method, system and storage medium for remote monitoring of mining equipment

Abstract

This invention relates to the field of industrial automation monitoring technology, and provides a method, system, and storage medium for remote monitoring of mining equipment. The method includes: classifying mining equipment into core equipment and general equipment based on equipment operating parameters; performing fault risk assessment on core equipment using dynamic thresholds adaptively generated based on historical data, and performing deviation detection on general equipment using fixed thresholds; fusing the fault risk assessment results of core equipment with the deviation detection results of general equipment to generate global maintenance decision information; optimizing on-site maintenance paths according to the geographical location and urgency of maintenance tasks; and continuously calibrating system parameters through a closed-loop feedback mechanism. This invention, through hierarchical monitoring and adaptive optimization, solves the problems of uneven allocation of monitoring resources, rigid threshold settings, and isolated maintenance decisions in existing technologies, achieving precise allocation of monitoring resources and a significant improvement in fault early warning capabilities for mining equipment.

Description

Technical Field

[0001] This invention relates to the field of industrial automation monitoring technology, specifically to a method, system, and storage medium for remote monitoring of mining equipment. Background Technology

[0002] Mining environments are characterized by equipment intensity, harsh working conditions, and high requirements for continuous operation. Unplanned shutdowns of core mining equipment, such as crushers, hoists, and large pumping stations, can disrupt the entire production chain, causing significant economic losses and even triggering safety accidents. Therefore, real-time and effective remote monitoring of mining equipment operation and predictive maintenance are crucial for ensuring safe mining production and improving operational efficiency.

[0003] Currently, common mining equipment monitoring solutions have the following limitations: 1) Single monitoring strategy and unreasonable resource allocation. Most existing systems adopt uniform monitoring standards and fixed alarm thresholds for all equipment. However, there are numerous pieces of equipment in a mining site, with varying values, criticality in the production process, and failure risks. Investing the same monitoring resources as core equipment in secondary equipment wastes computing, storage, and communication resources; conversely, insufficient monitoring depth for core equipment fails to capture early signs of failure in a timely manner, creating potential safety hazards. This "one-size-fits-all" monitoring model cannot achieve optimal allocation of operation and maintenance resources; 2) Rigid fault detection thresholds and poor adaptability. Most systems rely on fixed thresholds set in equipment manuals or based on experience for fault judgment. However, the normal operating parameters of mining equipment (such as vibration and temperature) will dynamically drift with changes in load, fluctuations in ambient temperature, and natural aging of equipment. Fixed thresholds are prone to two drawbacks when dealing with normal fluctuations: first, "false alarms," ​​where alarms are triggered even when the equipment is within the normal fluctuation range, causing maintenance personnel to be overwhelmed and reducing their sensitivity to genuine alarms; second, "missed alarms," ​​where equipment performance has begun to deteriorate slowly, but because the parameters have not reached the fixed threshold, the system fails to issue a timely warning, missing the optimal repair opportunity. 3) Isolated maintenance decisions and lack of global coordination. Existing monitoring systems typically only provide alarm information for single-point devices. When multiple devices malfunction simultaneously or sequentially, the system cannot intelligently assess the overall impact of these malfunctions, nor can it provide the field maintenance team with a globally optimal maintenance strategy that comprehensively considers factors such as equipment importance, fault urgency, and the location of maintenance resources. Maintenance personnel can only decide the response sequence based on experience, which may lead to delays in the repair of critical equipment or uneconomical maintenance paths, prolonging the overall fault recovery time.

[0004] Therefore, there is an urgent need in this field for a remote monitoring solution for mining equipment that can intelligently distinguish monitoring objects, adaptively adjust monitoring standards, and coordinate and optimize maintenance decisions to overcome the shortcomings of the existing technologies. Summary of the Invention

[0005] This invention provides a method, system, and storage medium for remote monitoring of mining equipment, aiming to solve the technical problems mentioned in the background art.

[0006] In a first aspect, the present invention provides a method for remote monitoring of mining equipment, comprising:

[0007] Step S1: Obtain static information and dynamic sensor data of mining equipment, and classify the equipment into core equipment and general equipment according to the preset classification criteria;

[0008] Step S2: For core equipment, establish dynamic threshold ranges for its key parameters and determine the failure probability value based on its operating status information. For general equipment, perform deviation detection based on fixed thresholds and generate a deviation correction priority sequence based on the severity and scope of the deviation.

[0009] Step S3: Integrate the failure probability values ​​of core equipment and the deviation correction priority sequence of general equipment to form an overall monitoring and adjustment vector;

[0010] Step S4: Adjust the state refresh frequency of the vector according to the overall monitoring, perform correction operations and update the running status to obtain the optimized deviation correction path;

[0011] Step S5: Through the optimized deviation correction path, apply path optimization iteration to obtain the path node calibration sequence. Based on the path node calibration sequence, execute the real-time feedback loop, determine the optimization convergence condition, and obtain the final stable equipment monitoring configuration.

[0012] Further, step S1 specifically includes:

[0013] Retrieve static attributes of the equipment from the equipment management database. These static attributes include the equipment model, asset value, and criticality level in the production process.

[0014] Real-time acquisition of dynamic sensor data from the equipment, including at least vibration amplitude, bearing temperature, working pressure, and operating current;

[0015] Construct a device importance assessment model, the input of which is the static attributes and dynamic sensor data, and the output is the device importance score;

[0016] The importance score of the equipment is compared with a preset importance threshold. If the score is higher than the threshold, it is classified as a core equipment; otherwise, it is classified as a general equipment.

[0017] Output the core device group list and the general device group list.

[0018] Furthermore, step S2, specifically establishing dynamic threshold ranges for key parameters of the core equipment, includes:

[0019] From the historical sensor data of the core equipment, extract the time-series features within a preset time window. The time-series features include the root mean square value of the vibration amplitude, the upward slope of the temperature index, and the peak fluctuation rate of the pressure data.

[0020] A statistical process control method based on Gaussian distribution is used to calculate each extracted time-series feature:

[0021] Dynamic threshold upper limit = feature mean + k × feature standard deviation

[0022] Dynamic threshold lower limit = feature mean - k × feature standard deviation

[0023] Where k is the control coefficient determined based on the preset reliability, and the feature mean and feature standard deviation are calculated based on the historical data of the core equipment under normal operating conditions;

[0024] The calculated dynamic threshold range is compared with the safe operating limit of the equipment to ensure that the dynamic threshold range is within the safe limit.

[0025] Each key feature of each core device is bound to and its dynamic threshold range is stored for real-time deviation detection.

[0026] Furthermore, step S2, which involves determining the fault probability value based on its operating status information, specifically includes:

[0027] The sensor readings of the core equipment are acquired in real time and compared with the dynamic threshold range;

[0028] When any sensor reading exceeds its dynamic threshold, a risk assessment process is triggered to extract auxiliary indicators from the operating status information. These auxiliary indicators include the current load rate of the equipment, the duration of continuous operation, and the lubrication status indicator.

[0029] The feature values ​​exceeding the dynamic threshold and their exceeding magnitude, along with the auxiliary indicators, are used as input features and input into the pre-trained classification model; wherein, the pre-trained classification model is a binary classifier based on gradient boosting decision trees, and its training data comes from historical fault records and normal operation data.

[0030] The classification model outputs a value between 0 and 1, which serves as the probability value of a potential failure of the core equipment at the current moment.

[0031] The risk probability value is compared with the preset risk alarm threshold of the core equipment. If it exceeds the threshold, a high-priority alarm is generated.

[0032] Furthermore, in step S2, for general equipment, deviation detection is performed based on a fixed threshold, and a deviation correction priority sequence is generated according to the severity and scope of the deviation. Specifically, this includes:

[0033] Set fixed safe operating thresholds for sensor data of general equipment based on equipment manuals or general engineering standards;

[0034] Periodically scan the sensor data of general equipment to perform limit checks;

[0035] If the data is within the limit, it is marked as normal. If the data exceeds the limit, the severity score of the deviation is calculated based on the current sensor reading, the corresponding safety threshold, and the pre-set weight according to the sensor type. Combined with the asset value coefficient of general equipment, the final priority score is calculated.

[0036] For all general equipment that has deviations, they are sorted in descending order of their priority scores to generate a deviation correction priority sequence.

[0037] Furthermore, step S3 specifically includes:

[0038] Obtain the risk probability value list and alarm status of core equipment, as well as the deviation correction priority sequence of general equipment;

[0039] Construct a global maintenance task list, where each task item in the global maintenance task list contains the following fields: device ID, device level, task type, urgency level, and estimated repair time;

[0040] Define a multi-dimensional vector, where the dimension represents different maintenance action types and the value of the vector represents the recommendation strength or priority of the action type, and generate an overall monitoring and adjustment vector.

[0041] Output the overall monitoring and adjustment vector for visualization and to drive subsequent maintenance scheduling.

[0042] Furthermore, step S4 specifically includes:

[0043] Monitor the frequency of changes in the overall monitoring adjustment vector, and trigger path replanning when a new high-priority task or task status update occurs;

[0044] Using the geographical locations of all devices with pending maintenance tasks as nodes and the current location of the maintenance personnel as the starting point, construct a path optimization graph;

[0045] The travel salesman problem model with constraints is used for path calculation. The optimization objective of the travel salesman problem model is to minimize the total path or the total task completion time. The constraints include the urgency of the task and the expected repair time.

[0046] Solve the model to obtain an optimized on-site maintenance path sequence, and then send the path to the mobile terminal;

[0047] Maintenance personnel perform operations according to the path, and after completion, report the status through the terminal to update the running status of the corresponding device in the system to "maintained".

[0048] Furthermore, step S5 specifically includes:

[0049] S501: After multiple maintenance cycles, collect maintenance feedback data, including the actual completion time of historical maintenance paths, false alarm and missed alarm records of dynamic thresholds, and the accuracy of fault prediction models.

[0050] S502: Based on the actual completion time, calibrate the time estimation parameters in the path optimization model to generate a more accurate path node calibration sequence;

[0051] S503: Based on the false alarm and missed alarm records, automatically adjust the dynamic threshold range and / or the fixed threshold;

[0052] S504: Determine the convergence condition for optimization: Calculate the change in the comprehensive performance index of the current monitoring period compared with the previous period. If the change is less than the preset convergence threshold, the system is considered to have reached a stable state, and the current set of full monitoring configuration parameters is output.

[0053] S505: If convergence is not achieved, repeat the feedback loop of S501-S504 until the system performance stabilizes.

[0054] In a second aspect, the present invention provides a remote monitoring system for mining equipment, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described in any of the above claims.

[0055] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any of the methods described above.

[0056] This invention offers the following advantages: By integrating historical performance, real-time status, and calibration information, it establishes a unified and quantifiable dynamic evaluation system for sensor reliability, enabling accurate and adaptive measurement of multi-source sensor data quality and providing a reliable benchmark for subsequent processing. Based on reliability evaluation and frequency domain-environment cross-validation results, it achieves dynamic adaptive adjustment of the weights of each sensor and performs targeted correction for low-quality data, effectively suppressing the influence of interference data in the fusion process and significantly improving the quality and reliability of the fused signal. Furthermore, this invention constructs a closed-loop optimization mechanism that traces back from risk assessment results to the sensor source, continuously updating the sensor performance database using verification feedback from screening results. This enables the system to possess self-learning and self-optimization capabilities, fundamentally ensuring the accuracy and stability of the chronic disease screening system's long-term operation. Attached Figure Description

[0057] Figure 1 A flowchart of a remote monitoring method for mining equipment provided by the present invention;

[0058] Figure 2 This is a comparison diagram of the overall effects of the present invention and conventional methods. Detailed Implementation

[0059] To further understand the content of this invention, a detailed description of the invention is provided in conjunction with the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0060] like Figure 1 This embodiment provides a remote monitoring method for mining equipment, including:

[0061] Step S1: Obtain static information and dynamic sensor data of mining equipment, and classify the equipment into core equipment and general equipment according to the preset classification criteria.

[0062] Specifically, step S1 includes: obtaining static attributes of the equipment from the equipment management database, including equipment model, asset value, and criticality level in the production process; collecting dynamic sensor data of the equipment in real time, including at least vibration amplitude, bearing temperature, working pressure, and operating current; constructing an equipment importance assessment model, the input of which is the static attributes and dynamic sensor data, and the output is an equipment importance score; comparing the equipment importance score with a preset importance threshold, if the score is higher than the threshold, it is classified as core equipment, otherwise it is classified as general equipment; and outputting a core equipment group list and a general equipment group list.

[0063] More specifically, static attributes of equipment are extracted from the mining equipment management database, including equipment model, asset value records, and predefined criticality levels of production processes. Simultaneously, dynamic sensor data, including vibration amplitude, bearing temperature, working pressure, and operating current, are collected in real time through IoT sensor nodes. These static attributes and dynamic sensor data constitute a multidimensional feature space for the equipment state vector. An importance assessment model is constructed based on this feature space, which uses a weighted scoring mechanism to convert heterogeneous data into a unified importance score. For static attributes, the static importance score of the equipment is calculated based on asset value records and predefined criticality levels of production processes. For dynamic sensor data, the real-time collected vibration amplitude, bearing temperature, working pressure, and operating current are compared with preset thresholds, such as a vibration amplitude threshold of 50 units, a temperature threshold of 70 degrees Celsius, a pressure threshold of 100 kPa, and a current threshold of 10 amperes. Standardization calculations convert each parameter into dimensionless fractions; for example, when the measured vibration amplitude is 60 units, its standardized score is (60-50) / 50 = 0.2. Then, a weighted summation formula is applied to calculate the comprehensive score of the equipment's dynamic sensor data: Score = Vibration Score × 0.4 + Temperature Score × 0.3 + Pressure Score × 0.2 + Current Score × 0.1. This assessment highlights core fault indicators and improves classification accuracy. It should be noted that the weighting coefficients 0.4, 0.3, 0.2, and 0.1 are determined based on correlation analysis of historical fault data from mining equipment. For example, in a mining excavator scenario, a piece of equipment has a vibration amplitude of 55 units (threshold 50), a temperature of 80 degrees Celsius (threshold 70), a pressure of 120 kPa (threshold 100), and a current of 15 amps (threshold 10), with standardized scores of 0.1, 0.14, 0.2, and 0.5 respectively. The weighted comprehensive score for the equipment's dynamic attention needs is 0.4 × 0.1 + 0.3 × 0.14 + 0.2 × 0.2 + 0.1 × 0.5 = 0.172.

[0064] A comprehensive score for each device is calculated by weighting and summing its static importance score and its dynamic attention requirement score, with each score having a weight of 0.5. When a device's comprehensive score exceeds a preset threshold of 0.7, it is marked as a core device and included in the priority monitoring list; other devices are classified as general devices and included in the standard monitoring list. This approach establishes a mathematical relationship between static device attributes and dynamic sensor data within the evaluation model, allowing classification decisions to consider both inherent device characteristics and real-time operational status. Under this hierarchical mechanism, core devices require focused monitoring, while the stable operational characteristics of general devices provide a foundation for simplified monitoring. This achieves optimal allocation of limited monitoring resources within the mining equipment cluster, establishing an accurate data foundation for subsequent fault prediction and maintenance decisions.

[0065] Step S2: For core equipment, establish dynamic threshold ranges for its key parameters and determine the failure probability value based on its operating status information. For general equipment, perform deviation detection based on fixed thresholds and generate a deviation correction priority sequence based on the severity and scope of the deviation.

[0066] Specifically, for core equipment, establishing dynamic threshold ranges for its key parameters includes: extracting time-series features within a preset time window from historical sensor data of the core equipment; these features include the root mean square value of vibration amplitude, the upward slope of temperature indicators, and the peak fluctuation rate of pressure data; employing a statistical process control method based on Gaussian distribution to calculate the dynamic threshold upper limit = feature mean + k × feature standard deviation, and the dynamic threshold lower limit = feature mean - k × feature standard deviation, where k is a control coefficient determined based on a preset reliability, and the feature mean and feature standard deviation are calculated based on historical data of the core equipment under normal operating conditions; comparing the calculated dynamic threshold range with the equipment's safe operating limits to ensure that the dynamic threshold range is within the safe limits; and binding and storing the dynamic threshold range for each key feature of each core equipment for real-time deviation detection. Furthermore, step S2, which determines the fault probability value based on its operating status information, specifically includes: acquiring sensor readings of the core equipment in real time and comparing them with the dynamic threshold range; when any sensor reading exceeds its dynamic threshold, triggering a risk assessment process, extracting auxiliary indicators from the operating status information, the auxiliary indicators including the current load rate of the equipment, the duration of continuous operation, and the lubrication status indicator; inputting the feature value exceeding its dynamic threshold and its exceeding magnitude, as well as the auxiliary indicators, as input features into a pre-trained classification model; wherein, the pre-trained classification model is a binary classifier based on gradient boosting decision trees, and its training data comes from historical fault records and normal operation data; the classification model outputs a value between 0 and 1 as the risk probability value of the core equipment potentially failing at the current moment; comparing the risk probability value with the preset risk alarm threshold of the core equipment, and generating a high-priority alarm if it exceeds the threshold.

[0067] More specifically, in the process of achieving dynamic monitoring of core equipment, the root mean square of vibration amplitude, the slope of temperature change, and the peak pressure fluctuation rate are extracted from the historical data of the core equipment as time-series features. These feature parameters constitute the feature vector of the equipment's operating status. Statistical parameters of the feature vector are calculated based on a Gaussian distribution model, and the dynamic threshold boundary is determined by adding or subtracting three times the standard deviation from the feature mean. The control coefficient is set to a value of 3 based on a pre-set confidence level, ensuring that the threshold range covers 95% of the data distribution under normal operating conditions. For example, in the monitoring scenario of a mining crusher, when the historical mean of the root mean square of vibration amplitude is 1.2 units and the standard deviation is 0.3 units, the dynamic threshold range is set to 0.3 to 2.1 units. This threshold range is verified by comparing it with the equipment's safe operating limits. When the threshold boundary exceeds the safety limit allowed by material strength, the control coefficient is automatically adjusted to ensure a balance between monitoring safety and sensitivity. This threshold calculation method based on the historical operating characteristics of the equipment effectively overcomes the technical deficiency of fixed thresholds in adapting to changes in equipment load.

[0068] Furthermore, when real-time sensor data exceeds the dynamic threshold, the system triggers a multi-level risk assessment process. Load rate, continuous operating time, and lubrication status are extracted from the equipment control system as auxiliary indicators, which, together with the feature value exceeding the threshold and its deviation magnitude, constitute a composite feature vector. This feature vector is input into a gradient boosting decision tree classification model, which is trained based on historical fault records and normal operating data. An ensemble learning framework is used to construct the classification decision boundary, outputting a fault risk probability value between 0 and 1. For example, when a vibration sensor detects an amplitude exceeding the dynamic threshold upper limit by 2.3 units, and the equipment load rate reaches 85% and has been running continuously for more than 72 hours, the classification model calculates a risk probability value of 0.82 based on these features. This probability value is compared with a preset alarm threshold of 0.65 to trigger a high-priority alarm. Thus, the dynamic threshold mechanism is responsible for capturing abnormal equipment conditions, while the classification model is responsible for assessing the severity of the abnormality. The two establish a closed-loop relationship through data flow and decision flow. It should be noted that when environmental conditions change and the threshold range needs adjustment, the system automatically optimizes the control coefficients based on the false alarm records of the classification model, forming an adaptive calibration of the threshold parameters. This technological synergy enables the monitoring system to adapt to the complex monitoring needs of mining equipment under conditions such as seasonal changes and load fluctuations. In the application of mine hoists, it effectively solves the technical problem of delayed early warning in fixed threshold systems during the slow deterioration of equipment performance.

[0069] Furthermore, in step S2, for general equipment, deviation detection is performed based on a fixed threshold, and a deviation correction priority sequence is generated according to the severity and scope of the deviation. Specifically, this includes: setting a fixed safe operating threshold for the sensor data of general equipment based on the equipment manual or general engineering standards; periodically scanning the sensor data of general equipment to perform limit checks; if the data does not exceed the limit, it is marked as normal; if the data exceeds the limit, the deviation severity score is calculated based on the current sensor reading, the corresponding safety threshold, and a pre-set weight according to the sensor type; combined with the asset value coefficient of the general equipment, the final priority score is calculated; for all general equipment with deviations, they are sorted in descending order of their priority scores to generate a deviation correction priority sequence.

[0070] More specifically, in the process of monitoring general equipment, the system generates a maintenance sequence through fixed threshold detection and multi-factor priority evaluation. First, based on the equipment technical manual and engineering standards, safe operating thresholds are set for various sensors. For example, the threshold for vibration sensors is set to 50 units, the threshold for temperature sensors to 70 degrees Celsius, the threshold for pressure sensors to 100 kPa, and the threshold for current sensors to 10 amperes. The system scans the data stream of general equipment sensors at fixed time intervals, comparing real-time readings with the corresponding thresholds. When data exceeding the safety limit is detected, a deviation analysis process is initiated. The deviation analysis mechanism assesses the severity of the anomaly through quantitative calculation, using a standardized formula to handle the deviation between sensor readings and thresholds. For example, if a vibration sensor detects 55 units, the severity score is calculated as (55-50) / 50 multiplied by a preset weight of 0.6, resulting in a preliminary score of 0.06. This calculation process is simultaneously applied to parameters such as temperature, pressure, and current, with temperature weighted at 0.3, pressure at 0.2, and current at 0.1. This weight allocation reflects the different impacts of various parameters on equipment safety. For example, in a mine ventilation equipment monitoring example, when the temperature sensor detects a temperature exceeding the threshold of 75 degrees Celsius and the pressure sensor detects a pressure exceeding the threshold of 110 kPa, the system calculates a temperature deviation score of 0.021 and a pressure deviation score of 0.02 respectively, and the weighted sum yields a comprehensive severity score of 0.041. Further, a mathematical relationship is established between the severity score and the equipment asset value coefficient, which is determined based on the equipment purchase cost and replacement difficulty, ranging from 0.8 to 1.2. The final priority score is obtained by multiplying the severity score by the asset value coefficient. This calculation method ensures that abnormal conditions of high-value equipment receive priority attention. In a mine station containing multiple general equipment, when three pieces of equipment simultaneously malfunction, their priority scores are 0.049, 0.036, and 0.028 respectively. The system generates a maintenance sequence for equipment A, equipment B, and equipment C, arranged in descending order of score.

[0071] In this way, fixed threshold detection provides basic anomaly identification, deviation quantification assessment achieves objective measurement of anomaly severity, and asset value weighting reflects differentiated considerations for the importance of general equipment. Fixed threshold detection provides triggering conditions for deviation assessment, the deviation assessment results provide a data basis for priority ranking, and the asset value coefficient ensures that maintenance resources are allocated to the more critical equipment in the general equipment group. In the practice of monitoring conveyor belt groups in mines, this solution improves maintenance response efficiency by approximately 30%. By guiding maintenance personnel to prioritize equipment anomalies with scores higher than 0.04 through a priority sequence, it avoids delays in the maintenance of important equipment caused by the traditional method of handling issues based on alarm time. This decision-making mechanism based on quantitative scores effectively optimizes the allocation of maintenance resources, achieving efficient overall management of general equipment groups while ensuring in-depth monitoring of core equipment.

[0072] Step S3: Integrate the failure probability values ​​of core equipment and the deviation correction priority sequence of general equipment to form an overall monitoring and adjustment vector.

[0073] Specifically, first, the risk probability value list and alarm status of core equipment, as well as the deviation correction priority sequence of general equipment, are obtained. Then, a global maintenance task list is constructed, where each task item in the global maintenance task list includes the following fields: equipment ID, equipment level, task type, urgency level, and estimated maintenance time. A multi-dimensional vector is defined, where the dimensions represent different maintenance action types, and the vector value represents the recommended strength or priority of the action type, generating an overall monitoring and adjustment vector. Specifically, generating the overall monitoring and adjustment vector includes: mapping high-priority alarms of core equipment to the highest value of the "emergency diagnosis" dimension; mapping medium- and low-risk alarms of core equipment to the medium value of the "planned maintenance" dimension; and mapping the first N items in the priority sequence of general equipment to the corresponding values ​​of the "on-site inspection and adjustment" dimension, where N is a positive integer. Finally, the overall monitoring and adjustment vector is output for visualization and to drive subsequent maintenance scheduling.

[0074] More specifically, the monitoring system extracts a list of risk probability values ​​and corresponding alarm statuses from the core equipment monitoring module, and obtains a deviation correction priority sequence from the general equipment monitoring module. This data is input into the global maintenance task list construction module, which creates a structured record for each task item, containing the equipment identifier, equipment level, task type, urgency level, and estimated maintenance time. When constructing the global maintenance task list, the equipment ID and equipment level (core or general) are extracted based on the equipment grouping results. The task type is determined according to the equipment type and alarm / deviation status. High-priority alarms for core equipment correspond to "emergency diagnosis", medium- and low-risk alarms correspond to "planned maintenance", and deviations for general equipment correspond to "on-site inspection and adjustment". The urgency level is quantified by the risk probability value or priority sequence position. The urgency level of high-priority alarms for core equipment is set to the highest level. For general equipment, the urgency level of the first N items in the priority sequence (N is determined according to the total number of mining equipment and maintenance resource configuration, and is a positive integer, for example, set to 10) decreases with the sequence position. The estimated maintenance time is set with reference to the historical maintenance data of mining equipment, combined with the equipment type and fault / deviation type. For example, the estimated time for emergency diagnosis of core equipment drilling rigs is 2 hours, and the estimated time for on-site inspection and adjustment of general equipment conveyors is 1 hour.

[0075] Furthermore, a multidimensional vector is defined, constructed based on the type of mine equipment maintenance action, including core dimensions such as emergency diagnosis, planned maintenance, on-site inspection and adjustment. The vector values ​​are determined through specific mapping rules: high-priority alarms of core equipment are directly mapped to the highest value of the emergency diagnosis dimension (e.g., setting the vector value range to 0-10, with the highest value set to 10), ensuring that major faults of core equipment receive priority response; low-risk alarms of core equipment are mapped to medium values ​​of the planned maintenance dimension (e.g., 4-8), balancing the allocation of resources between emergency handling and routine maintenance; the first N items in the priority sequence of general equipment are mapped to the corresponding values ​​of the on-site inspection and adjustment dimensions according to their sequence positions (e.g., the first item in the sequence is mapped to 9, the second item to 8, and so on down), ensuring that general equipment with high deviations is handled in a timely manner. For example, in a large open-pit mine scenario, the failure probability value of the core equipment excavator is calculated to be 0.82 (high-priority alarm), which is mapped to the emergency diagnosis dimension value of 10. The top 5 items of the deviation correction priority sequence for the 10 conveyors in the general equipment are mapped to the on-site inspection and adjustment dimensions 9, 8, 7, 6, and 5, respectively. The generated overall monitoring and adjustment vector is [10, 0, 9, 8, 7, 6, 5, ...]. The first dimension value "10" indicates that the recommended priority for "emergency diagnosis" is the highest; the second dimension value "0" indicates that the recommended priority for "planned maintenance" is the lowest (no need to execute). Since the current core equipment only has high-priority faults and no medium or low-priority alarms... Since the failure probability is not in the 0.5-0.75 range, this dimension's value is set to 0, clearly indicating that planned maintenance will not occupy resources, avoiding maintenance action conflicts and resource waste. The values ​​"9, 8, 7, 6, 5" in dimensions 3 to 7 correspond to the first 5 pieces of equipment in the general equipment deviation correction priority sequence (such as mine conveyors). The values ​​decrease as the sequence position increases, reflecting the execution priority of "on-site inspection and adjustment": dimension 3 corresponds to the first high-deviation equipment in the sequence (highest priority, recommended strength 9), dimension 4 corresponds to the second high-deviation equipment in the sequence (recommended strength 8), and so on, instructing maintenance personnel to carry out inspections and deviation corrections in descending order of value, ensuring that high-deviation equipment is dealt with first. The subsequent "..." indicates that the vector can be dynamically expanded according to the length of the general equipment priority sequence. The values ​​of unlisted dimensions continue according to the rule of "decreasing by 1" (e.g., dimension 8 is 4, dimension 9 is 3, etc.), until all general equipment in the sequence that needs to be inspected is covered.

[0076] The operation and maintenance system can prioritize personnel to perform emergency diagnostics on excavators based on this vector, while simultaneously scheduling conveyor inspections in sequence. Through the functional synergy of core equipment risk quantification data and general equipment priority data, this technology achieves precise allocation of maintenance resources, reduces downtime due to core equipment failures, mitigates the risk of general equipment deviations escalating, improves the overall operational stability and maintenance efficiency of mining equipment, and ensures the global optimization scheduling capability of the monitoring system. This multi-dimensional vector representation method overcomes the technical shortcomings of traditional monitoring systems, such as the independent display of various alarms and the lack of overall situational awareness, and solves the problem of fragmented decision-making focus for maintenance personnel when faced with multiple equipment alarms. This approach improves the decision-making efficiency of the maintenance command center by approximately 40%, and the visualization of the overall monitoring adjustment vector enables dispatchers to quickly identify the most urgent maintenance needs. This multi-dimensional vector-based decision support mechanism, together with the equipment hierarchical monitoring strategy, constitutes a complete technical system, realizing the transformation of mining equipment maintenance management from discrete alarm processing to systematic decision support.

[0077] Step S4: Adjust the status refresh frequency of the vector based on the overall monitoring, perform correction operations and update the running status to obtain the optimized deviation correction path.

[0078] Specifically, the frequency of changes in the overall monitoring adjustment vector is monitored, and path replanning is triggered when a new high-priority task or task status update occurs. Then, a path optimization graph is constructed using the geographical locations of all devices with pending maintenance tasks as nodes and the current location of the maintenance personnel as the starting point. Further, a travel salesman problem model with constraints is applied to calculate the path. The optimization objective of the travel salesman problem model is to minimize the total path or the total task completion time. The constraints include the urgency of the task and the estimated maintenance time. The model is solved to obtain an optimized on-site maintenance path sequence, and the path is sent to the mobile terminal. The maintenance personnel perform operations according to the path and report the status through the terminal after completion, updating the operating status of the corresponding device in the system to "maintained".

[0079] More specifically, the monitoring and adjustment vector status is monitored. When the value of the emergency diagnosis dimension in the vector changes abruptly or the number of on-site inspection tasks increases, the system identifies this as a status refresh event and triggers a path replanning process. The path planning module uses the real-time location reported by the maintenance personnel's handheld terminal as the path starting point and the geographical coordinates of the equipment corresponding to the maintenance tasks to be processed as the node set, constructing a path optimization graph that includes a distance matrix and time cost. The path optimization employs a constrained traveling salesman problem model, which uses the distance between equipment and the estimated repair time of the task as optimization parameters, while introducing urgency as a hard constraint. For example, in a mining station instance with 5 tasks to be processed, the system obtains that the maintenance personnel are located at coordinate point A and need to access equipment nodes located at points B, C, D, and E, where equipment at point B has a high urgency level. During the model solution process, the system prioritizes ensuring the high-urgency node's preceding position in the path sequence, while optimizing the access order of the remaining nodes to minimize the total travel distance. A path sequence ABCDE is calculated using a heuristic algorithm. This sequence satisfies the requirement of prioritizing urgent tasks while keeping the total path length within 3.5 kilometers. The system then pushes the optimized path sequence to maintenance personnel's mobile terminals, which display a visual navigation path based on a geographic information system. Maintenance personnel perform on-site operations according to the sequence. After completing the maintenance task at each node, they trigger a status update command through the terminal interface. This command is transmitted wirelessly to the central processing system. Upon receiving the status signal, the system marks the corresponding device's operating status as "maintained" and removes the corresponding record from the pending task list.

[0080] In practical applications at mine crushing stations, this step can shorten the average single inspection path length by approximately 25% and reduce task completion time by about 30%, effectively solving the problem of personnel having to travel back and forth due to improper task sequencing in traditional maintenance models. This overcomes the rigidity of fixed inspection routes when dealing with sudden maintenance tasks, enabling adaptive scheduling of maintenance resources in the complex environment of mines, and improving the response efficiency and execution effectiveness of the overall equipment management system.

[0081] Step S5: Through the optimized deviation correction path, apply path optimization iteration to obtain the path node calibration sequence. Based on the path node calibration sequence, execute the real-time feedback loop, determine the optimization convergence condition, and obtain the final stable equipment monitoring configuration.

[0082] Specifically, step S5 includes: S501: After multiple maintenance cycles, collect maintenance feedback data, including the actual completion time of historical maintenance paths, false alarm and missed alarm records of dynamic thresholds, and the accuracy of the fault prediction model; S502: Based on the actual completion time, calibrate the time estimation parameters in the path optimization model to generate a more accurate path node calibration sequence; S503: Automatically adjust the dynamic threshold range and / or the fixed threshold according to the false alarm and missed alarm records; S504: Determine the optimization convergence condition: calculate the change in the comprehensive performance index of the current monitoring cycle compared to the previous cycle. If the change is less than the preset convergence threshold, the system is considered to have reached a stable state, and the current set of full monitoring configuration parameters is output; S505: If convergence is not achieved, repeat the feedback loop of S501-S504 until the system performance tends to stabilize. The automatic adjustment of the dynamic threshold range and / or the fixed threshold includes automatically adjusting the control coefficient k of the statistical process control method, or incrementally learning the classification model; the comprehensive performance index is determined based on the average fault warning time and false alarm rate; the current set of full monitoring configuration parameters includes grouping standards, dynamic thresholds, model parameters, etc.

[0083] More specifically, continuous improvement of monitoring configuration is achieved through the establishment of a closed-loop feedback mechanism and performance evaluation system. After completing a preset number of maintenance cycles, the system initiates a data analysis process, collecting historical maintenance path actual time records, false alarm and missed alarm statistics from the dynamic threshold mechanism, and accuracy indicators of the fault prediction model. This operational data constitutes the foundational dataset for system performance evaluation, providing a quantitative basis for parameter calibration. The calibration of the path optimization model is based on the deviation analysis between actual maintenance time and estimated time. When the system detects that the average actual maintenance time of a specific equipment node continuously exceeds the estimated value by 20%, it automatically adjusts the time estimation parameters in the path planning algorithm. In a mine inspection case involving eight maintenance nodes, the system calibrated the time estimate for node C from the initial 25 minutes to 31 minutes through three iterations. The corresponding generated path node calibration sequence shifted the ranking of node C to the back, improving the accuracy of the total path duration prediction to over 90%. This parameter calibration mechanism based on historical data allows the path planning model to gradually align with the actual time consumption patterns of on-site operations. During threshold adjustment, the monitoring sensitivity is dynamically corrected based on false alarm and missed alarm records. When the vibration sensor generates a false alarm rate exceeding 15% within three consecutive cycles, the system gradually adjusts the control coefficient k in the statistical process control from 3.0 to 3.2, expanding the dynamic threshold range to filter fluctuations under normal operating conditions. Simultaneously, the fault prediction model incorporates newly generated fault samples through incremental learning. When the system identifies a new type of bearing failure mode, the classification model automatically updates the decision boundary to adapt to the new fault characteristics. This dual adjustment mechanism enables the monitoring system to adapt to the gradual performance changes caused by equipment aging.

[0084] The monitoring system determines the optimization convergence state by calculating the changes in a comprehensive performance index, which is a composite of the average fault warning time and the false alarm rate, weighted at 7:3. In the monitoring practice of a mine crusher group, the system recorded index changes of 0.05 and 0.03 for two consecutive periods. When the change fell below the preset threshold of 0.02, a stable state was determined to have been reached. At this point, the system outputs a complete monitoring configuration set, including equipment classification standards, dynamic threshold parameters, and model weight coefficients, marking the completion of the adaptive optimization process.

[0085] This approach further avoids the scheduling inefficiency caused by inaccurate time estimations due to path calibration, the alarm accuracy issues caused by insufficient environmental adaptability due to threshold adjustment, and the decline in recognition capability caused by changes in fault modes due to model updates. By sharing performance data and coordinating parameter updates, a comprehensive technical foundation for continuous system improvement is established. In a six-month field test, this solution gradually reduced the false alarm rate from an initial 22% to 8%, and extended the average fault warning time from 48 hours to 72 hours. It effectively solved the technical challenge of static monitoring systems being unable to adapt to the long-term evolution of equipment operating conditions, achieving a leap from fixed configuration to adaptive learning in mine equipment monitoring.

[0086] In a second aspect, a remote monitoring system for mining equipment is provided, including a memory and a processor. The memory stores a computer program, wherein the processor executes the computer program to implement the steps of the remote monitoring method for mining equipment provided in the first aspect of the embodiments.

[0087] Thirdly, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of a remote monitoring method for mining equipment provided in the first aspect of the embodiments.

[0088] As attached Figure 2As shown, the remote monitoring method for mining equipment provided in this embodiment has achieved significant technical progress in key performance indicators compared to conventional methods. Regarding computing power requirements, the system's average CPU utilization rate is reduced from 100% in conventional methods to 95% through a hierarchical equipment monitoring strategy, achieving a 5% optimization in computing power and effectively improving system resource utilization efficiency. In terms of alarm accuracy, the false alarm rate is significantly reduced from 20% to 15%, and the missed alarm rate is significantly improved from 25% to 18%. This is attributed to the application of a dynamic threshold mechanism based on historical operating data and a machine learning fault prediction model, effectively overcoming the insufficient adaptability of fixed thresholds under complex operating conditions. Regarding operation and maintenance efficiency, maintenance time is shortened from a baseline of 100% to 87%, and maintenance paths are optimized from a baseline of 100% to 82%. This is mainly due to the synergistic effect of intelligent maintenance task prioritization and path optimization algorithms. The synergistic optimization in these five dimensions constitutes the overall technical advantage of this invention, achieving a technological leap from single-device monitoring to systematic intelligent operation and maintenance, providing reliable technical assurance for the safe and stable operation of mining equipment.

[0089] The above description is merely a specific embodiment of this specification. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the scope of protection of this specification is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this specification, and these modifications or substitutions should all be covered within the scope of protection of this specification.

Claims

1. A method for remote monitoring of mining equipment, characterized in that, include: Step S1: Obtain static information and dynamic sensor data of mining equipment, and classify the equipment into core equipment and general equipment according to the preset classification criteria; Step S2: For core equipment, establish dynamic threshold ranges for its key parameters and determine the failure probability value based on its operating status information. For general equipment, perform deviation detection based on fixed thresholds and generate a deviation correction priority sequence based on the severity and scope of the deviation. Step S3: Integrate the failure probability values ​​of core equipment and the deviation correction priority sequence of general equipment to form an overall monitoring and adjustment vector; Step S4: Adjust the status refresh frequency of the vector according to the overall monitoring, perform correction operations and update the running status to obtain the optimized on-site maintenance path sequence; Step S5: Through the optimized field maintenance path sequence, apply path optimization iteration to obtain the path node calibration sequence. Based on the path node calibration sequence, execute the real-time feedback loop, determine the optimization convergence condition, and obtain the final stable equipment monitoring configuration. Specifically, step S4 includes: Monitor the frequency of changes in the overall monitoring adjustment vector, and trigger path replanning when a new high-priority task or task status update occurs; Using the geographical locations of all devices with pending maintenance tasks as nodes and the current location of the maintenance personnel as the starting point, construct a path optimization graph; The travel salesman problem model with constraints is used for path calculation. The optimization objective of the travel salesman problem model is to minimize the total path or the total task completion time. The constraints include the urgency of the task and the expected repair time. Solve the model to obtain an optimized on-site maintenance path sequence, and then send the path to the mobile terminal; Maintenance personnel perform operations according to the path, and after completion, report the status through the terminal, updating the running status of the corresponding device in the system to "maintained".

2. The remote monitoring method for mining equipment as described in claim 1, characterized in that, Step S1 specifically includes: Retrieve static attributes of the equipment from the equipment management database. The static attributes include equipment model, asset value, and criticality level in the production process. Real-time acquisition of dynamic sensor data from the equipment, including at least vibration amplitude, bearing temperature, working pressure, and operating current; Construct an equipment importance assessment model, wherein the input of the equipment importance assessment model is the static attributes and dynamic sensor data, and the output is the equipment importance score; The importance score of the equipment is compared with a preset importance threshold. If the score is higher than the threshold, it is classified as a core equipment; otherwise, it is classified as a general equipment. Output the core device group list and the general device group list.

3. The remote monitoring method for mining equipment as described in claim 1, characterized in that, Step S2, which establishes dynamic threshold ranges for key parameters of core devices, specifically includes: From the historical sensor data of the core equipment, extract the time-series features within a preset time window. The time-series features include the root mean square value of the vibration amplitude, the upward slope of the temperature index, and the peak fluctuation rate of the pressure data. A statistical process control method based on Gaussian distribution is used to calculate each extracted time-series feature: Dynamic threshold upper limit = feature mean + k × feature standard deviation Dynamic threshold lower limit = feature mean - k × feature standard deviation Where k is the control coefficient determined based on the preset reliability, and the feature mean and feature standard deviation are calculated based on the historical data of the core equipment under normal operating conditions; The calculated dynamic threshold range is compared with the safe operating limit of the equipment to ensure that the dynamic threshold range is within the safe limit. Each key feature of each core device is bound to and its dynamic threshold range is stored for real-time deviation detection.

4. The remote monitoring method for mining equipment as described in claim 1, characterized in that, Step S2, which involves determining the fault probability value based on its operating status information, specifically includes: The sensor readings of the core equipment are acquired in real time and compared with the dynamic threshold range; When any sensor reading exceeds its dynamic threshold, a risk assessment process is triggered to extract auxiliary indicators from the operating status information. These auxiliary indicators include the current load rate of the equipment, the duration of continuous operation, and the lubrication status indicator. The feature values ​​exceeding the dynamic threshold and their exceeding magnitude, along with the auxiliary indicators, are used as input features and input into the pre-trained classification model; wherein, the pre-trained classification model is a binary classifier based on gradient boosting decision trees, and its training data comes from historical fault records and normal operation data. The classification model outputs a value between 0 and 1, which serves as the probability value of a potential failure of the core equipment at the current moment. The risk probability value is compared with the preset risk alarm threshold of the core equipment. If it exceeds the threshold, a high-priority alarm is generated.

5. The remote monitoring method for mining equipment as described in claim 1, characterized in that, Step S2, for general equipment, involves deviation detection based on a fixed threshold, and generating a deviation correction priority sequence according to the severity and scope of the deviation. Specifically, this includes: Set fixed safe operating thresholds for sensor data of general equipment based on equipment manuals or general engineering standards; Periodically scan the sensor data of general equipment to perform limit checks; If the data is within the limit, it is marked as normal. If the data exceeds the limit, the severity score of the deviation is calculated based on the current sensor reading, the corresponding safety threshold, and the pre-set weight according to the sensor type. Combined with the asset value coefficient of general equipment, the final priority score is calculated. For all general equipment that has deviations, they are sorted in descending order of their priority scores to generate a deviation correction priority sequence.

6. The remote monitoring method for mining equipment as described in claim 1, characterized in that, Step S3 specifically includes: Obtain the risk probability value list and alarm status of core equipment, as well as the deviation correction priority sequence of general equipment; Construct a global maintenance task list, where each task item in the global maintenance task list contains the following fields: device ID, device level, task type, urgency level, and estimated repair time; Define a multi-dimensional vector, where the dimension represents different maintenance action types and the value of the vector represents the recommendation strength or priority of the action type, and generate an overall monitoring and adjustment vector. Output the overall monitoring and adjustment vector for visualization and to drive subsequent maintenance scheduling.

7. The remote monitoring method for mining equipment as described in claim 1, characterized in that, Step S5 specifically includes: S501: After multiple maintenance cycles, collect maintenance feedback data, including the actual completion time of historical maintenance paths, false alarm and missed alarm records of dynamic thresholds, and the accuracy of fault prediction models. S502: Based on the actual completion time, calibrate the time estimation parameters in the path optimization model to generate a more accurate path node calibration sequence; S503: Based on the false alarm and missed alarm records, automatically adjust the dynamic threshold range and / or the fixed threshold; S504: Determine the convergence condition for optimization: Calculate the change in the comprehensive performance index of the current monitoring period compared with the previous period. If the change is less than the preset convergence threshold, the system is considered to have reached a stable state, and the current set of full monitoring configuration parameters is output. S505: If convergence is not achieved, repeat the feedback loop of S501-S504 until the system performance stabilizes.

8. A remote monitoring system for mining equipment, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images