Mining power supply operation and maintenance system based on fault prediction and health management

Abstract

The invention discloses a mining power supply operation and maintenance system based on fault prediction and health management. The system comprises a data acquisition and preprocessing module, a distributed fault diagnosis module, a cloud collaborative evaluation module and an operation and maintenance execution and closed loop module. And the data acquisition and preprocessing module is used for acquiring multi-dimensional operation data of the lithium battery mining power supply cluster and performing distributed preprocessing through the intelligent edge computing node. The invention relates to the technical field of mining power supply operation and maintenance. According to the mining power supply operation and maintenance system based on fault prediction and health management, through a distributed architecture and a lightweight diagnosis model, real-time fault detection and positioning are carried out on preprocessed data, the downtime of equipment is reduced, a three-level operation and maintenance strategy is adopted, operation and maintenance priorities are divided in combination with health indexes and fault levels, and the operation and maintenance efficiency is improved. Equipment operation and maintenance responses of different priorities are realized, key equipment is ensured to be maintained preferentially, and the operation and maintenance efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of mine power supply operation and maintenance, and more specifically, to a mine power supply operation and maintenance system based on fault prediction and health management. Background Technology

[0002] With the increasing demands for power supply in the mining industry, mining power equipment, especially lithium battery mining power supplies, has gradually become an important tool for ensuring power supply. Lithium battery power supplies are widely used in various equipment in mines due to their high energy efficiency and long lifespan. However, due to the complex environment and harsh working conditions in mines, lithium battery power supplies have a higher risk of failure, leading to frequent equipment malfunctions and consequently affecting production efficiency and safety.

[0003] Current operation and maintenance management of mining power supplies typically relies on manual inspections and periodic maintenance. This approach is not only time-consuming and labor-intensive, but also fails to provide real-time updates on equipment health, leading to delayed fault detection and even situations where faults cannot be addressed promptly. Furthermore, fault prediction and health management technologies for mining power equipment have not been effectively promoted and applied, resulting in a lack of systematic and intelligent operation and maintenance management methods. Traditional operation and maintenance management largely depends on static diagnostic methods, failing to provide dynamic assessments based on real-time equipment data, and lacking sufficient flexibility and accuracy.

[0004] Therefore, there is an urgent need for a mining power supply operation and maintenance system based on advanced technology, which can improve equipment reliability, reduce the frequency of failures, and improve the safety and production efficiency of mining operations through fault prediction and health management. Summary of the Invention

[0005] The purpose of this invention is to provide a mining power supply operation and maintenance system based on fault prediction and health management. This system solves the problem that the existing operation and maintenance management of mining power supplies usually relies on manual inspection and regular maintenance. This method is not only time-consuming and labor-intensive, but also cannot keep track of the health status of the equipment in real time, resulting in delayed fault detection and even situations where faults cannot be dealt with in a timely manner.

[0006] This invention achieves the above objectives through the following technical solution: a mine power supply operation and maintenance system based on fault prediction and health management, the system comprising: Data acquisition and preprocessing module, distributed fault diagnosis module, cloud-based collaborative assessment module, and operation and maintenance execution and closed-loop module; The data acquisition and preprocessing module is used to acquire multi-dimensional operational data of the lithium battery mining power cluster and perform distributed preprocessing through intelligent edge computing nodes. The distributed fault diagnosis module, based on a distributed architecture, uses a lightweight diagnostic model deployed on edge nodes to perform real-time fault detection and location on preprocessed data. The cloud-based collaborative assessment module generates a health assessment report by integrating local diagnostic results with global health status data through collaborative interaction between edge nodes and the cloud-based operation and maintenance platform. The operation and maintenance execution and closed-loop module triggers targeted operation and maintenance execution processes based on health assessment reports and preset operation and maintenance strategies to achieve predictive maintenance.

[0007] Furthermore, the data acquisition and preprocessing module includes: Multiple sensing and detection units are used to collect raw operating data covering multiple dimensions, including voltage, current, temperature, SOC, SOH, and insulation resistance. The sensing and detection unit includes at least one of the following: a single cell voltage sensor, a total voltage sensor, a charge / discharge current sensor, a cell temperature sensor, a SOC detection module, a SOH assessment module, and an insulation monitoring module.

[0008] Furthermore, the intelligent edge computing nodes are deployed according to the mining area, and each edge node is responsible for processing a preset number of lithium battery mining power data within its jurisdiction; The preset quantity is determined based on the CPU clock speed and memory computing power parameters of the edge nodes; Edge nodes preprocess the raw operational data through standardization, outlier filtering, and data gap filling.

[0009] Furthermore, the distributed fault diagnosis architecture includes: The lithium battery device-side data acquisition layer, edge node processing layer, and cloud collaboration layer; The edge node processing layer and the device-side data acquisition layer communicate in real time via industrial Ethernet or wireless sensor network. The transmission delay of key parameters is controlled within a preset threshold. These key parameters include cell voltage, charging and discharging current, and cell temperature.

[0010] Furthermore, the lightweight diagnostic model is a fusion model of convolutional neural network and gradient boosting decision tree, used to extract local temporal feature vectors and global statistical feature vectors of running data, and fuse them to obtain a comprehensive feature vector. Based on the lithium battery-specific fault feature library, fault type identification is achieved through feature similarity matching.

[0011] Furthermore, the cloud-based collaborative assessment module receives local fault diagnosis results, operating status parameters, and processing logs uploaded by edge nodes; The upload frequency is dynamically adjusted based on the lithium battery fault level. The cloud-based operation and maintenance platform builds a global health status database, performs correlation analysis on cross-regional lithium battery cluster data, and identifies systemic failure risks.

[0012] Furthermore, the cloud-based collaborative assessment module uses a lithium battery-specific health index assessment model to calculate the health status; The health index assessment model is based on parameters such as SOH, cumulative fault status, charge and discharge efficiency, and individual cell voltage balance. It is calculated by weighting parameters with preset weighting coefficients, which are determined by the analytic hierarchy process.

[0013] Furthermore, the preset operation and maintenance strategy library is a three-level operation and maintenance strategy, which divides the operation and maintenance priorities according to the health index and fault level; The health index threshold is determined by the K-means clustering algorithm, and different priorities correspond to different operation and maintenance response times, maintenance content, and inspection frequencies.

[0014] Furthermore, the operation and maintenance execution and closed-loop module can automatically generate exclusive operation and maintenance work orders. The work orders include the equipment number, fault-related information, processing procedures and required tool and spare parts information, and push them to the corresponding regional operation and maintenance terminal. After the operation and maintenance is completed, the edge nodes collect the operation data after the repair and continuously monitor it for a preset period of time to verify the effect of fault handling and update the health record and fault feature library.

[0015] Furthermore, it also includes a lifespan and failure prediction unit: Based on historical operation and maintenance data, cycle life data, and health trends of lithium batteries, an improved long short-term memory network is used to predict the remaining service life and the probability of the next failure, providing data support for equipment replacement plans and operation and maintenance resource allocation.

[0016] The beneficial effects of this invention are as follows: 1. Through the data acquisition and preprocessing module, the system can collect multi-dimensional operating data of lithium battery mining power supply in real time, and perform data preprocessing through intelligent edge computing nodes. The data processing process adopts standardization, outlier filtering, data supplementation and other methods to ensure data quality and accuracy, and provide a reliable basis for subsequent fault diagnosis.

[0017] 2. Through a distributed architecture and lightweight diagnostic model, real-time fault detection and location are performed on the preprocessed data. The distributed fault diagnosis architecture can effectively improve the accuracy and response speed of fault detection and reduce equipment downtime.

[0018] 3. Through the cloud-based collaborative assessment module, the system can conduct a global analysis of the health status of equipment in various regions, identify potential systemic failure risks, and provide a basis for operation and maintenance decisions, ensuring that timely countermeasures can be taken when equipment malfunctions.

[0019] 4. Based on health assessment reports and preset operation and maintenance strategies, the system triggers targeted operation and maintenance execution processes to achieve predictive maintenance. After maintenance, the system continuously monitors to verify the effectiveness of fault handling and updates health records and fault feature databases in real time, forming a closed-loop management system.

[0020] 5. Through the lifespan and failure prediction unit, the system predicts the remaining lifespan of lithium batteries and the probability of the next failure based on historical data and health trends, providing data support for equipment replacement plans and operation and maintenance resource allocation, effectively extending equipment lifespan and reducing the probability of sudden failures.

[0021] 6. Adopt a three-level operation and maintenance strategy, and combine health index and fault level to divide operation and maintenance priorities, so as to realize the operation and maintenance response of equipment with different priorities, ensure that critical equipment is given priority maintenance, and improve operation and maintenance efficiency. Attached Figure Description

[0022] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a system block diagram of the present invention; Figure 2 This is a flowchart of the data acquisition and preprocessing process of the present invention; Figure 3 This is a flowchart of the fault diagnosis and maintenance closed-loop process of the present invention. Detailed Implementation

[0023] The present application will now be described in further detail with reference to the accompanying drawings. It should be noted that the following specific embodiments are only used to further illustrate the present application and should not be construed as limiting the scope of protection of the present application. Those skilled in the art can make some non-essential improvements and adjustments to the present application based on the above application content.

[0024] Example 1: Please see Figure 1-3 This invention provides a technical solution: a lithium battery mining power supply operation and maintenance system based on fault prediction and health management, the system comprising: Data acquisition and preprocessing module, distributed fault diagnosis module, cloud-based collaborative assessment module, and operation and maintenance execution and closed-loop module; The data acquisition and preprocessing module is used to acquire multi-dimensional lithium battery operation data of the lithium battery mining power cluster and perform distributed preprocessing of the operation data through intelligent edge computing nodes. Among them, lithium battery mining power supply clusters refer to a collection of multiple lithium battery power supplies used in mining environments. These power supplies work together to provide power support for mining equipment. Due to the complexity of the mining environment, the performance and reliability requirements of the power supplies are high. Multi-dimensional lithium battery operation data covers multiple aspects of lithium battery operation, such as voltage, current, temperature, and internal resistance. This data can comprehensively reflect the working status and performance changes of lithium batteries. Intelligent edge computing nodes are network edge devices deployed close to the data source (i.e., lithium battery mining power supplies). They have certain computing capabilities and can perform preliminary processing and analysis on the collected data, reducing the time and bandwidth requirements for data transmission to the cloud and improving the real-time performance and response speed of the system. Distributed preprocessing utilizes intelligent edge computing nodes to distribute data preprocessing tasks to various edge nodes in parallel, rather than centralizing them in a central node. This approach can improve data processing efficiency, reduce the computing pressure on the central node, and enhance the reliability and scalability of the system. The distributed fault diagnosis module is used to perform real-time fault detection and location on pre-processed lithium battery operating data based on a distributed fault diagnosis architecture and a lightweight diagnostic model deployed on edge nodes. Among them, the distributed fault diagnosis architecture is a system architecture that distributes fault diagnosis tasks across multiple nodes for collaborative completion. Unlike the centralized architecture, the distributed architecture can fully utilize the computing resources of each node, improve the real-time performance and accuracy of fault diagnosis, and enhance the system's fault tolerance. Edge nodes are computing nodes deployed at the network edge, responsible for collecting and processing local data and performing certain computing tasks. In the lithium battery mining power supply operation and maintenance system, edge nodes are deployed close to the power supply, enabling them to quickly acquire the power supply's operating data and perform real-time fault diagnosis. Lightweight diagnostic models, compared to complex deep learning models or other large diagnostic models, have the characteristics of simple model structure, small computational load, and low resource consumption. They can run efficiently on devices with limited computing resources, such as edge nodes, to achieve real-time fault detection and location of lithium battery operating data. Real-time fault detection and location refer to the real-time monitoring of lithium battery operating data to promptly detect the presence of faults. Fault location, after detecting a fault, determines the specific location or component where the fault occurred, so that targeted repair and handling can be carried out subsequently. The cloud-based collaborative assessment module is used to generate a health assessment report for lithium battery mining power supplies by integrating local diagnostic results with global health status data of lithium batteries through collaborative interaction between edge nodes and the cloud-based operation and maintenance platform. The collaborative interaction between edge nodes and the cloud-based operation and maintenance platform involves edge nodes uploading locally processed data and preliminary diagnostic results to the cloud-based platform, while the cloud-based platform can send control commands and global information to the edge nodes. This two-way information exchange enables collaborative work between the edge nodes and the cloud-based platform, fully leveraging the real-time processing capabilities of the edge nodes and the big data processing and comprehensive analysis capabilities of the cloud platform. Local diagnostic results are obtained from fault diagnosis of lithium battery operating data by lightweight diagnostic models on the edge nodes. These results primarily reflect the local operating status and fault conditions of the lithium battery. Global health status data of the lithium battery is obtained on the cloud-based platform by integrating and analyzing data uploaded from multiple edge nodes, providing overall health status information for the entire lithium battery mining power supply cluster, including health trends of each power supply and overall performance evaluation. The health assessment report, generated on the cloud-based platform based on local diagnostic results and global health status data, is a detailed report on the health status of the lithium battery mining power supply. This report typically includes information such as the power supply's health level, potential fault risks, and remaining service life, providing a basis for operation and maintenance decisions. The operation and maintenance execution and closed-loop module is used to trigger targeted operation and maintenance execution processes based on health assessment reports and preset operation and maintenance strategies, so as to realize predictive maintenance of lithium battery mining power supplies. Among them, the health assessment report, along with the health assessment report generated in the aforementioned cloud-based collaborative assessment module, serves as an important basis for operation and maintenance execution. The preset operation and maintenance strategy is a series of pre-defined operation and maintenance rules and processes based on the characteristics and operation and maintenance experience of lithium battery mining power supplies. For example, it includes maintenance measures and maintenance cycles corresponding to different health levels. These strategies provide specific guidance for operation and maintenance execution. The targeted operation and maintenance execution process determines specific operation and maintenance tasks and operational steps based on the health assessment report and the preset operation and maintenance strategy. For example, it involves repairing faulty power supplies, replacing components, and adjusting operating parameters to achieve effective maintenance of lithium battery mining power supplies. Predictive maintenance... Unlike traditional post-failure maintenance, predictive maintenance is based on the assessment and prediction of equipment health status. It identifies potential faults in advance and takes corresponding maintenance measures before a fault occurs, thereby avoiding the occurrence of faults or reducing their impact on the system, improving equipment reliability and availability, and reducing operation and maintenance costs. Closed-loop operation involves feeding the operation and maintenance results back into the system during the operation and maintenance process. This is compared and analyzed with health assessment reports and preset operation and maintenance strategies. Based on the feedback results, the operation and maintenance strategies and execution processes are adjusted to form a closed-loop operation and maintenance management system. Through this closed-loop mechanism, operation and maintenance plans are continuously optimized, improving operation and maintenance effectiveness and the overall performance of the system.

[0025] It should be noted that during use, the data acquisition and preprocessing module utilizes intelligent edge computing nodes to distribute and process multi-dimensional operational data, enabling rapid acquisition and preliminary screening of effective information, reducing data transmission pressure, and improving real-time performance. The distributed fault diagnosis module, with the help of a lightweight model and architecture, achieves real-time fault detection and location at edge nodes, promptly identifying problems. The cloud-based collaborative assessment module, through edge-cloud interaction, integrates local and global data to generate a health assessment report, comprehensively and accurately grasping the power supply status. The operation and maintenance execution and closed-loop module triggers targeted processes based on reports and preset strategies, achieving predictive maintenance, eliminating potential hazards in advance, preventing fault escalation, reducing downtime and maintenance costs, improving the reliability and stability of lithium battery mining power supply operation, and ensuring the continuity and safety of mining operations.

[0026] In one embodiment, multi-dimensional lithium battery operation data of a lithium battery mining power cluster is acquired, and the operation data is preprocessed in a distributed manner through intelligent edge computing nodes, including: Multi-dimensional raw operating data of lithium batteries are collected through individual cell voltage sensors, total voltage sensors, charge / discharge current sensors, cell temperature sensors, SOC detection modules, SOH assessment modules, and insulation monitoring modules deployed at the lithium battery mining power equipment end.

[0027] in, This indicates the number of data dimensions, which must include at least seven dimensions: individual cell voltage, total voltage, charging / discharging current, cell temperature, state of charge (SOC), state of equilibrium (SOH), and insulation resistance. Indicates the data collection time step; The raw operating data of the lithium batteries is divided according to the mining area and distributed to the corresponding intelligent edge computing nodes. Each edge node is responsible for its assigned area. Data processing of lithium battery mining power supplies in Taiwan meets the requirements. ; in, This represents the maximum number of processing devices per edge node, a value determined through edge node computing power testing. When the edge node CPU frequency is ≥2.0GHz and the memory is ≥8GB, ; When the CPU clock speed is 1.5-2.0GHz and the memory is 4-8GB, ; When CPU frequency is <1.5GHz and memory is <4GB, ; The edge node performs standardization processing on the received raw lithium battery operating data. The standardization expression is as follows:

[0028] in, Indicates the first Data on lithium batteries in the first The collected values ​​at each time step Indicates the first The average value of lithium battery data. Indicates the first Standard deviation of lithium battery data; Outlier filtering was performed on the standardized lithium battery dataset, and the 2σ criterion was used to remove outliers from the individual cell voltage data. Outlier data points were removed from the range, and the remaining dimensions of data were removed using the 3σ criterion. Abnormal data points within the range; the selection thresholds for the 2σ and 3σ criteria are determined by the lithium battery manufacturer's technical manual. The manual specifies that the 2σ criterion is used when the allowable fluctuation range of a certain parameter is ≤±5%, and the 3σ criterion is used when the allowable fluctuation range is >±5%. Data gaps are filled using an interpolation method based on the charge and discharge characteristics of lithium batteries. Exponential interpolation is used during the charging phase, with interpolation coefficients... Determined based on the lithium battery charging rate: when the charging rate is ≥1C. When the charging rate is <1C Linear interpolation is used during the discharge phase.

[0029] This design collects multi-dimensional raw data from lithium batteries using various sensors, and distributes it to intelligent edge computing nodes for processing according to region. The multi-dimensional data comprehensively reflects the state of the lithium batteries, providing rich information for subsequent analysis. Distributing data to edge nodes by region reduces data transmission pressure and improves processing efficiency. Furthermore, determining the number of processing devices based on the computing power of the edge nodes allows for the rational use of resources. Standardized processing unifies the data scale, outlier filtering removes interfering data, and interpolation fills in gaps, ensuring data integrity and accuracy. This provides a reliable data foundation for subsequent fault diagnosis and health assessment, improving the overall performance and reliability of the system.

[0030] In one embodiment, based on a distributed fault diagnosis architecture, a lightweight diagnostic model deployed on edge nodes is used to perform real-time fault detection and location on preprocessed lithium battery operating data, including: A distributed fault diagnosis architecture based on intelligent edge computing is constructed. This architecture includes a lithium battery device-side data acquisition layer, an edge node processing layer, and a cloud collaboration layer. The edge node processing layer and the device-side data acquisition layer communicate in real time via industrial Ethernet or wireless sensor network. The transmission delay of key lithium battery parameters is controlled within 30ms. This threshold is determined by the mining power safety standard GB 3836.1-2021 to ensure that the fault response meets the safety requirements of the explosion-proof environment. Key parameters include single cell voltage, charging and discharging current, and cell temperature. Each edge node deploys a lightweight convolutional neural network and gradient boosting decision tree fusion lithium battery fault diagnosis model. The model parameters are pre-trained in the cloud based on the lithium battery fault sample library and then sent to the edge node, supporting online incremental updates based on lithium battery cycle life data. The preprocessed lithium battery operating data is input into a lightweight diagnostic model, and local temporal feature vectors of the lithium battery operating data are extracted through convolutional layers. The study focuses on capturing fault characteristics such as individual cell voltage fluctuations and temperature abrupt changes, and extracts global statistical feature vectors of lithium batteries through gradient boosting decision trees. This includes SOC decay rate, SOH change trend, charge / discharge efficiency, etc., which are fused to obtain a comprehensive feature vector of the lithium battery:

[0031] Based on a dedicated fault feature library for lithium batteries, covering at least 12 typical lithium battery faults such as unbalanced cell voltage, overcharging and over-discharging, thermal runaway of cells, abnormal increase in internal resistance, and insulation damage, fault type identification is achieved through cosine similarity matching. The similarity calculation formula is as follows:

[0032] in The first in the lithium battery fault feature library The standard feature vector of a fault class, when The fault was identified as a corresponding type of lithium battery. The similarity threshold of 0.88 is determined by cross-validation of 1,000 lithium battery fault samples and 500 normal samples. When the threshold is 0.88, the fault identification accuracy is ≥95% and the false positive rate is ≤3%, which meets the accuracy requirements for fault diagnosis of mining equipment. By using an attention mechanism, the lithium battery components and sensors associated with the fault are located. The lithium battery components include specific individual cells, BMS management modules, charging and discharging interfaces, etc. The local fault diagnosis results are output, including the lithium battery fault type, fault level, occurrence time, associated components, and fault impact range assessment.

[0033] This design constructs a distributed architecture, with fusion models deployed on edge nodes. Input preprocessed data is used to extract features and identify faults. The distributed architecture utilizes edge nodes for real-time processing, reducing transmission latency and meeting the safety requirements of explosion-proof environments. The fusion model combines the advantages of convolutional neural networks and gradient boosting decision trees, accurately extracting local and global features. Fault identification is based on a dedicated fault feature library and cosine similarity matching, achieving high accuracy and low false positive rates. An attention mechanism locates fault-related components and outputs detailed diagnostic results, facilitating rapid fault location and resolution, reducing downtime, and ensuring stable operation of the mine power supply.

[0034] In one embodiment, through the collaborative interaction between edge nodes and the cloud-based operation and maintenance platform, local diagnostic results are integrated with global health status data of the lithium battery to generate a health assessment report for the lithium battery mining power supply, including: The edge node uploads the local fault diagnosis results of the lithium battery, the lithium battery operating status parameters and processing logs to the cloud operation and maintenance platform through an encrypted transmission protocol. The operating status parameters include SOC, SOH, cycle count, and cumulative charge and discharge capacity. Upload frequency is dynamically adjusted based on the lithium battery fault level: Level 1 faults are uploaded in real time, Level 2 faults are uploaded every 20 seconds, Level 3 faults are uploaded every minute, and fault-free conditions are uploaded every 3 minutes. Level 1 faults include thermal runaway warnings and severe overcharging and over-discharging; Level 2 faults include slight imbalances in individual cell voltages and slightly excessive internal resistance; and Level 3 faults include a slight decrease in insulation resistance. The upload frequency threshold is determined through data transmission bandwidth testing. When the mining farm network bandwidth is ≥100Mbps, the upload interval under fault-free conditions can be shortened to 2 minutes. When bandwidth is less than 50Mbps, the upload interval under fault-free conditions is extended to 5 minutes to ensure that critical business bandwidth is not occupied. The cloud-based operation and maintenance platform receives uploaded data from all edge nodes, builds a global lithium battery mining power supply health status database, performs correlation analysis on cross-regional lithium battery cluster data, and identifies potential systemic failure risks such as lithium battery consistency degradation and regional power grid fluctuations. Based on the full life cycle operation data of lithium batteries, the health status is calculated using a lithium battery-specific health index assessment model. The health index calculation formula is as follows:

[0035] in: Let be the weighting coefficient, satisfying The determination rules are based on the analytic hierarchy process: a three-layer structure is constructed, consisting of target layer health assessment, criterion layer SOH, fault condition, charge and discharge efficiency, voltage balance, and scheme layer for each lithium battery model. A judgment matrix is ​​constructed using the 1-9 scaling method, and after consistency test CR < 0.1, the values ​​are determined to be 0.45, 0.25, 0.15, and 0.15, respectively. SOH is the state of health value of lithium battery, with a value range of 0-1. It is calculated by combining the capacity decay rate and the internal resistance growth rate of lithium battery. The permissible failure threshold for lithium batteries is determined based on the lithium battery model and application scenario: the permissible failure threshold for lithium iron phosphate mining power supplies is 5 times, with a cycle life of ≥3000 times; the permissible failure threshold for ternary lithium mining power supplies is 3 times, with a cycle life of ≥2000 times. The standard equalization threshold is set to ≤5%, which is based on the lithium battery industry standard GB / T 31484-2015, to ensure that individual cell voltage differences do not affect the overall performance of the power supply. By combining the results of local fault diagnosis of lithium batteries with the global health index, a health assessment report is generated, which includes basic information of lithium battery mining power supply, details of lithium battery operating status, fault diagnosis conclusions, health trend predictions and targeted maintenance suggestions. The basic information includes model, installation time and cycle life. The details of operating status include SOC, SOH, temperature distribution and voltage balance. The health trend predictions include the remaining number of cycles and the remaining service life.

[0036] This design allows edge nodes to upload data to the cloud, where a database is built and analyzed. A dedicated model is used to calculate health indices and generate reports. The upload frequency is dynamically adjusted to adapt to different network bandwidths, ensuring stable data transmission without consuming critical business bandwidth. The cloud also performs cross-regional data correlation analysis to identify systemic failure risks. The dedicated health index assessment model comprehensively considers multiple factors, with scientifically and reasonably determined weighting coefficients, accurately assessing the health status of lithium batteries and generating detailed health assessment reports containing comprehensive information. This provides a complete basis for operation and maintenance decisions, helps to identify potential problems in advance, formulate reasonable maintenance plans, and extend the lifespan of lithium batteries.

[0037] In one embodiment, based on a health assessment report and a preset operation and maintenance strategy, a targeted operation and maintenance execution process is triggered to achieve predictive maintenance of lithium battery mining power supplies, including: A pre-defined three-level lithium battery operation and maintenance strategy library is used, based on health indices. Maintenance priorities are assigned based on lithium battery fault levels. The health index thresholds are determined using data from the entire lifecycle of 500 mining lithium batteries. K-means clustering is employed to divide the health status into three categories, with cluster centers of 0.7 and 0.85 respectively. Therefore, the thresholds are set at 0.7 and 0.85. Level 1 Operations and Maintenance: In the event of a Level 1 fault, an audible and visual alarm will be triggered immediately, the charging and discharging circuit of the faulty lithium battery will be cut off, and maintenance personnel will be dispatched to the site within 1 hour to handle the situation. Handling operations include replacing the faulty individual battery cell and repairing the BMS module. The 1-hour response threshold is determined according to the mine safety management regulations to ensure that major faults will not cause safety accidents. Level 2 Operations and Maintenance: For secondary or level 2 faults, they are included in the 48-hour maintenance plan, generating a task list for lithium battery equalization charging, individual cell testing, and spare parts preparation; the 48-hour maintenance cycle is determined based on the complexity of lithium battery fault repair and the spare parts procurement cycle. Level 3 Operations and Maintenance: And without faults, routine inspections of lithium batteries are carried out every half month. The inspection content includes SOC calibration, voltage balance test, heat dissipation system inspection, and insulation resistance test. The inspection frequency is determined through fault statistical analysis. If lithium batteries in a certain area have been operating without faults for more than 6 months, the inspection frequency can be adjusted to once a month. Based on the maintenance recommendations in the health assessment report, the cloud-based operation and maintenance platform automatically generates a dedicated operation and maintenance work order for lithium batteries. The work order includes the lithium battery device number, fault location, fault type, handling procedure, required tools and spare parts information, and pushes it to the operation and maintenance terminal in the corresponding region. The handling procedure includes equalization charging parameters, cell replacement steps, etc., and the required spare parts include matching model individual cells, BMS testing equipment, etc. After the operation and maintenance is completed, the operational data of the lithium battery repair is collected through the edge node and continuously monitored for 24 hours. The monitoring duration is determined by the fault recurrence rate test. Continuous monitoring for 24 hours can ensure that the fault repair effect is stable for more than 98% of the faults, verify the fault handling effect, including whether the single cell voltage balance has been restored to the standard range, whether the SOH is stable, update the lithium battery health record and fault feature library, and realize closed-loop operation and maintenance management. Based on historical operation and maintenance data, cycle life data, and health trends of lithium batteries, an improved long short-term memory network is used to predict the remaining service life and the probability of the next failure of lithium batteries, providing data support for lithium battery replacement plans and optimized allocation of operation and maintenance resources.

[0038] This design features a pre-defined three-tiered maintenance strategy library, cloud-based work orders, post-maintenance monitoring and file updates, and prediction of remaining lifespan. The three-tiered maintenance strategy prioritizes tasks based on health indices and fault levels, rationally allocating maintenance resources to ensure timely handling of major faults, orderly maintenance of general faults, and regular inspections, thereby improving maintenance efficiency. It automatically generates dedicated work orders, clearly defining processing procedures and required spare parts for easy operation by maintenance personnel. Continuous post-maintenance monitoring verifies effectiveness, and file updates achieve closed-loop management, ensuring maintenance quality. Predicting remaining lifespan and fault probability provides data support for replacement plans and resource optimization, reducing maintenance costs and improving the reliability and economy of mining power supply operation.

[0039] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0040] The above embodiments provide a detailed description of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A mine power supply operation and maintenance system based on fault prediction and health management, characterized in that, The system includes: Data acquisition and preprocessing module, distributed fault diagnosis module, cloud-based collaborative assessment module, and operation and maintenance execution and closed-loop module; The data acquisition and preprocessing module is used to acquire multi-dimensional operational data of the lithium battery mining power cluster and perform distributed preprocessing through intelligent edge computing nodes. The distributed fault diagnosis module, based on a distributed architecture, uses a lightweight diagnostic model deployed on edge nodes to perform real-time fault detection and location on preprocessed data. The cloud-based collaborative assessment module generates a health assessment report by integrating local diagnostic results with global health status data through collaborative interaction between edge nodes and the cloud-based operation and maintenance platform. The operation and maintenance execution and closed-loop module triggers targeted operation and maintenance execution processes based on health assessment reports and preset operation and maintenance strategies to achieve predictive maintenance.

2. The mine power supply operation and maintenance system based on fault prediction and health management according to claim 1, characterized in that, The data acquisition and preprocessing module includes: Multiple sensing and detection units are used to collect raw operating data covering multiple dimensions, including voltage, current, temperature, SOC, SOH, and insulation resistance. The sensing and detection unit includes at least one of the following: a single cell voltage sensor, a total voltage sensor, a charge / discharge current sensor, a cell temperature sensor, a SOC detection module, a SOH assessment module, and an insulation monitoring module.

3. The mine power supply operation and maintenance system based on fault prediction and health management according to claim 1, characterized in that: The intelligent edge computing nodes are deployed according to the mining area, and each edge node is responsible for processing a preset number of lithium battery mining power data within its jurisdiction. The preset quantity is determined based on the CPU clock speed and memory computing power parameters of the edge nodes; Edge nodes preprocess the raw operational data through standardization, outlier filtering, and data gap filling.

4. The mine power supply operation and maintenance system based on fault prediction and health management according to claim 1, characterized in that, The distributed fault diagnosis architecture includes: The lithium battery device-side data acquisition layer, edge node processing layer, and cloud collaboration layer; The edge node processing layer and the device-side data acquisition layer communicate in real time via industrial Ethernet or wireless sensor network. The transmission delay of key parameters is controlled within a preset threshold. These key parameters include cell voltage, charging and discharging current, and cell temperature.

5. The mine power supply operation and maintenance system based on fault prediction and health management according to claim 1, characterized in that: The lightweight diagnostic model is a fusion model of convolutional neural network and gradient boosting decision tree. It is used to extract local temporal feature vectors and global statistical feature vectors from the running data and fuse them to obtain a comprehensive feature vector. Based on the lithium battery-specific fault feature library, fault type identification is achieved through feature similarity matching.

6. The mine power supply operation and maintenance system based on fault prediction and health management according to claim 1, characterized in that: The cloud-based collaborative assessment module receives local fault diagnosis results, operating status parameters, and processing logs uploaded by edge nodes; The upload frequency is dynamically adjusted based on the lithium battery fault level. The cloud-based operation and maintenance platform builds a global health status database, performs correlation analysis on cross-regional lithium battery cluster data, and identifies systemic failure risks.

7. The mine power supply operation and maintenance system based on fault prediction and health management according to claim 1, characterized in that: The cloud-based collaborative assessment module uses a lithium battery-specific health index assessment model to calculate the health status. The health index assessment model is based on parameters such as SOH, cumulative fault status, charge and discharge efficiency, and individual cell voltage balance. It is calculated by weighting parameters with preset weighting coefficients, which are determined by the analytic hierarchy process.

8. The mine power supply operation and maintenance system based on fault prediction and health management according to claim 1, characterized in that: The preset operation and maintenance strategy library is a three-level operation and maintenance strategy, which divides the operation and maintenance priority according to the health index and fault level. The health index threshold is determined by the K-means clustering algorithm, and different priorities correspond to different operation and maintenance response times, maintenance content, and inspection frequencies.

9. The mine power supply operation and maintenance system based on fault prediction and health management according to claim 1, characterized in that: The operation and maintenance execution and closed-loop module can automatically generate exclusive operation and maintenance work orders. The work orders include the equipment number, fault-related information, processing procedures and required tool and spare parts information, and push them to the corresponding regional operation and maintenance terminal. After the operation and maintenance is completed, the edge nodes collect the operation data after the repair and continuously monitor it for a preset period of time to verify the effect of fault handling and update the health record and fault feature library.

10. The mine power supply operation and maintenance system based on fault prediction and health management according to claim 1, characterized in that, It also includes a lifespan and failure prediction unit: Based on historical operation and maintenance data, cycle life data, and health trends of lithium batteries, an improved long short-term memory network is used to predict the remaining service life and the probability of the next failure, providing data support for equipment replacement plans and operation and maintenance resource allocation.

Images