A Data Analysis Method and System for Coal Mining Equipment Based on a Big Data Platform

Abstract

This invention discloses a data analysis method and system for coal mining equipment based on a big data platform, comprising: collecting multi-source data from coal mining equipment and preprocessing the data; extracting features from the preprocessed data, fusing features from different data sources to form a unified set of equipment operating status features; constructing and training a deep learning model, continuously optimizing model parameters and structure through learning to identify normal and abnormal equipment states; and performing maintenance actions based on abnormal states. This invention effectively integrates multi-source data, providing a comprehensive assessment of equipment operating status; utilizes a deep learning model to improve the accuracy of anomaly detection; dynamically adjusts model parameters to adapt to changes in equipment state; and promptly identifies potential faults and optimizes maintenance strategies, thereby reducing downtime and improving the safety and economy of coal mine production.

Description

Technical Field

[0001] This invention relates to the field of industrial equipment monitoring and safety protection technology, specifically to a data analysis method and system for coal mining equipment based on a big data platform. Background Technology

[0002] With the development of industrial automation and informatization, the monitoring and maintenance of coal mining equipment has become a crucial link in ensuring the safety and efficiency of coal mine production. Traditional equipment monitoring mainly relies on periodic manual inspections and monitoring data from single sensors. However, with the improvement of sensor technology and data acquisition capabilities, coal mining equipment can generate a large amount of multi-source data, including various parameters such as vibration, temperature, and current. This data can provide rich information, helping to understand the operating status of the equipment more comprehensively. However, because traditional monitoring methods lack the ability to effectively integrate and analyze multi-source data, they often struggle to identify abnormal equipment conditions in real time and accurately. This situation makes it difficult to detect and prevent equipment failures in a timely manner, increasing equipment downtime and maintenance costs.

[0003] Existing data analysis technologies for coal mining equipment have several shortcomings. First, most methods are limited to processing data from a single sensor, failing to fully utilize the complementary information from multiple data sources, thus unable to provide a comprehensive assessment of equipment operating status. Second, existing technologies typically rely on simple statistical analysis or rule-based methods in data processing and analysis. These methods have limited ability to identify complex anomaly patterns and are ill-equipped to handle abnormal situations under varying operating conditions. Furthermore, traditional data analysis methods lack adaptability and cannot dynamically adjust model parameters according to changes in equipment operating status, resulting in low accuracy in anomaly detection. In addition, many existing methods lack the application of deep learning models for equipment operating status, making them inadequate when faced with large amounts of unstructured data and unable to uncover deeper features within the data. Therefore, the shortcomings of existing technologies in equipment health status assessment and anomaly detection limit their effectiveness and widespread application in practice.

[0004] Existing coal mining equipment data analysis technologies mainly face problems such as insufficient data integration, low anomaly detection accuracy, and poor model adaptability. Our invented coal mining equipment data analysis method based on a big data platform collects and preprocesses multi-source data, extracts and fuses different features to form a unified equipment operating status feature set, and utilizes a deep learning model for continuous optimization and anomaly detection. This effectively solves the aforementioned problems and belongs to the field of industrial equipment monitoring and safety protection technology. This method not only improves the accuracy of equipment status assessment but also promptly detects potential equipment failures and optimizes maintenance strategies, thereby significantly improving the safety and economy of coal mine production. Summary of the Invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, the technical problem solved by this invention is: how to make full use of multi-source data from coal mining equipment, conduct comprehensive evaluation of equipment operating status through effective data preprocessing and feature extraction, and combine deep learning models, timely detect and identify abnormal states, optimize equipment maintenance strategies, thereby improving the accuracy of equipment fault detection and maintenance efficiency, reducing downtime, and ensuring the safety and economy of coal mine production.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a data analysis method for coal mining equipment based on a big data platform, comprising: collecting multi-source data from coal mining equipment and preprocessing the data;

[0008] Features are extracted from the preprocessed data, and features from different data sources are merged to form a unified set of equipment operating status features.

[0009] Build and train deep learning models, and continuously learn to optimize model parameters and structure to identify normal and abnormal states of devices;

[0010] Perform maintenance actions based on the abnormal status.

[0011] As a preferred embodiment of the coal mining equipment data analysis method based on a big data platform described in this invention, the multi-source data collected from the coal mining equipment includes vibration data, temperature data, acoustic data, current data, and historical operating data.

[0012] The preprocessing includes data cleaning, noise reduction, standardization, and time synchronization.

[0013] As a preferred embodiment of the coal mining equipment data analysis method based on a big data platform described in this invention, the step of extracting features from the preprocessed data includes obtaining the frequency domain features X(f) of the coal mining equipment vibration signal through frequency domain analysis, the expression of which is:

[0014]

[0015] Where x[n] represents the time domain signal, N represents the signal length, f represents the frequency, and j represents the imaginary unit. The weighted X(f) is used as part of the feature set, and the weight matrix is ​​W1.

[0016] The trend characteristic T of temperature data is obtained through time series analysis, and its expression is:

[0017]

[0018] Where t0 and t NThese represent the start and end points of the time interval, respectively, where t represents the time variable, and φ... i The coefficients of the autoregressive model are represented by p, which represents the lag order of the autoregressive model, and x represents the coefficients of the autoregressive model. t-i Represents temperature data at a lag time point, ∈ t Let T represent white noise, and α represent the attenuation coefficient of the time-series feature; add T to the feature set;

[0019] The Mel frequency cepstral coefficient feature MFCC(n) obtained from acoustic data through voiceprint recognition is expressed as follows:

[0020]

[0021] Where K represents the total number of Mel filters, E k Let W represent the energy of the k-th filter; MFCC(n) is weighted and added to the feature set, with the weight matrix being W2.

[0022] The final feature set F is represented as:

[0023] F = {X(f), T, MFCC(n)}

[0024] Where β represents the characteristic attenuation coefficient.

[0025] As a preferred embodiment of the coal mining equipment data analysis method based on a big data platform described in this invention, the construction and training of the deep learning model includes performing forward propagation calculations, using the feature set F as input, the formula of which is:

[0026] z (l) =g(W (l) F+b (l) )

[0027] Among them, z (l) W represents the activation value of the l-th layer, where l is any layer number from 1 to L. (l) Let b represent the weight matrix of the l-th layer. (l) Let g(·) represent the bias vector of the l-th layer, and g(·) represent the activation function.

[0028] The loss function is calculated using the following formula:

[0029]

[0030] in, Let m represent the loss value, m represent the number of samples, and y represent the loss value. i Indicates the actual value. This represents the final predicted value;

[0031] The formula for parameter optimization and feature fusion is as follows:

[0032]

[0033] Where Z represents the optimized comprehensive feature, and λ and μ represent the regularization coefficients; This represents the regularization term, which controls the change of the weight matrix over time.

[0034] As a preferred embodiment of the coal mining equipment data analysis method based on a big data platform described in this invention, the identification of the normal and abnormal states of the equipment includes, during equipment operation, if the optimized comprehensive feature Z is within the normal range, i.e., Z∈[0, Z...], ... high The system adopts a normal operation mode based on historical data to continuously monitor the equipment;

[0035] Normal state threshold Z high The system determines the status of key components of the equipment through feature clustering and density estimation methods, performs real-time telemetry on all collected sensor data, and monitors and adjusts the status of key components of the equipment through remote control functions to keep the equipment in optimal operating condition.

[0036] If the comprehensive feature Z is greater than the abnormal state threshold Z threshold Z threshold <Z high The system switches to early warning mode, increasing the monitoring frequency of key features;

[0037] When the comprehensive feature Z exceeds the normal range, i.e. Z>Z high The system enters anomaly handling mode, initiates anomaly identification and isolation mechanisms, and uses multi-level anomaly detection algorithms to analyze and classify anomalies, determining the anomaly type and severity.

[0038] As a preferred embodiment of the coal mining equipment data analysis method based on a big data platform described in this invention, the step of analyzing and classifying anomalies using a multi-level anomaly detection algorithm includes preliminary anomaly detection on the comprehensive feature representation Z, and screening out anomalies that deviate from the normal range. The specific formula is as follows:

[0039]

[0040] Among them, Z i Let μ represent the i-th eigenvalue. Z and σ Z Let Z represent the mean and standard deviation of the feature set Z, respectively.

[0041] If Anomaly i If the value exceeds the preset threshold θ, it is determined to be a preliminary anomaly.

[0042] For the anomalies initially identified, a multi-level clustering algorithm is used for forest segmentation and classification.

[0043] The DBSCAN algorithm is applied to perform density clustering on outliers to identify high-density outlier groups and noise points. The formula is as follows:

[0044]

[0045] Where ∈ represents the cluster radius parameter, Cluster i Representing point x i The clustering results centered on x are used; points that cannot be classified into any density cluster are marked as noise points; x represents any data point in the data space. This represents a d-dimensional real space, where d represents the dimension of the data.

[0046] Based on the clustering results, the severity of outliers is probabilistically assessed using a Geometric Matrix (GMM). The GMM is fitted to each clustering result, calculating the probability that each outlier belongs to a different category, and the outlier type is determined based on maximum likelihood estimation. The specific formula is as follows:

[0047]

[0048] Wherein, P(x i ) represents point x i The probability density, π k Denotes the mixing coefficients of the k-th Gaussian distribution. The mean is μ k The covariance is ∑ k Gaussian distribution; determine the anomaly type based on the category with the highest probability density; when P(x i If the value exceeds the set threshold Pθ, the current fault state of the device is considered to be severely abnormal.

[0049] As a preferred embodiment of the coal mining equipment data analysis method based on a big data platform described in this invention, the maintenance action includes: when the early warning mode is triggered, the system automatically increases the remote control of the equipment, and adjusts the equipment operating parameters to prevent the occurrence of abnormal states. At the same time, the early warning information is fed back to the management personnel in real time through the remote communication module.

[0050] When the equipment experiences a severe malfunction, the equipment protection program is automatically triggered to perform remote shutdown or isolation operations. During this process, remote fault indicators and multimodal sensor fusion technology are used to accurately locate the fault location and send an anomaly report to the maintenance team. The system records and analyzes the entire process of the anomaly, and through deep learning of the data and continuous optimization of the model, it continuously improves the system's fault detection and handling capabilities. When the abnormal state is resolved, the system automatically performs a self-healing process to restore the equipment to normal operation.

[0051] A coal mining equipment data analysis system based on a big data platform, characterized in that it includes:

[0052] Data acquisition module: Acquires multi-source data from coal mining equipment and preprocesses the data;

[0053] Data feature fusion module: Extracts features from preprocessed data, merges features from different data sources, and forms a unified set of equipment operating status features;

[0054] Model building module: Builds and trains deep learning models, continuously learns and optimizes model parameters and structure to identify normal and abnormal states of devices;

[0055] Anomaly alarm module: Based on abnormal conditions, it will trigger alarms and perform maintenance actions.

[0056] The beneficial effects of this invention are: it effectively integrates multi-source data to provide a comprehensive assessment of equipment operating status; it utilizes deep learning models to improve the accuracy of anomaly detection; it dynamically adjusts model parameters to adapt to changes in equipment status; and it promptly identifies potential faults and optimizes maintenance strategies, thereby reducing downtime and improving the safety and economy of coal mine production. Attached Figure Description

[0057] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0058] Figure 1 The flowchart illustrates a data analysis method for coal mining equipment based on a big data platform, as provided in the first embodiment of the present invention. Detailed Implementation

[0059] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0060] Example 1, referring to Figure 1 As an embodiment of the present invention, a data analysis method for coal mining equipment based on a big data platform is provided, including:

[0061] S1: Collect multi-source data from coal mining equipment and preprocess the data.

[0062] The multi-source data collected from coal mining equipment includes vibration data, temperature data, acoustic data, current data, and historical operating data.

[0063] Preprocessing includes data cleaning, noise reduction, standardization, and time synchronization.

[0064] It should be noted that collecting multi-source data from coal mining equipment is fundamental to achieving accurate monitoring and analysis. These data sources include vibration data, temperature data, acoustic data, current data, and historical operating data. Each data type provides different dimensions of information about the equipment's operating status. Vibration data reflects the dynamic characteristics of mechanical components, temperature data indicates thermal energy status, acoustic data provides the sound characteristics of mechanical operation, current data reveals the operating status of the electrical system, and historical operating data provides background information for analyzing trends in equipment status changes. By integrating this multi-source data, a comprehensive and three-dimensional picture of the equipment's status can be constructed, facilitating a deeper understanding of the equipment's operating condition and potential problems.

[0065] Because the sampling frequencies and time points of multi-source data may differ, direct data fusion can lead to inconsistencies in time-series characteristics, thus affecting the accuracy of the analysis. Time synchronization processing, through precise time alignment techniques, maps data collected at different times onto the same time axis. This not only ensures the accuracy of data fusion but also provides a foundation for dynamic feature analysis. For example, the relationship between mechanical vibration and current fluctuations can only be accurately reflected when time is synchronized.

[0066] S2: Extract features from the preprocessed data, integrate features from different data sources, and form a unified set of equipment operating status features.

[0067] The frequency domain characteristics X(f) of the vibration signal of the coal mining equipment are obtained through frequency domain analysis, and its expression is:

[0068]

[0069] Where x[n] represents the time domain signal, N represents the signal length, f represents the frequency, and j represents the imaginary unit. The weighted X(f) is used as part of the feature set, and the weight matrix is ​​W1.

[0070] The trend characteristic T of temperature data is obtained through time series analysis, and its expression is:

[0071]

[0072] Where t0 and t N These represent the start and end points of the time interval, respectively, where t represents the time variable, and φ... i The coefficients of the autoregressive model are represented by p, which represents the lag order of the autoregressive model, and x represents the coefficients of the autoregressive model. t-iRepresents temperature data at a lag time point, ∈ t Let T represent white noise, and α represent the attenuation coefficient of the time series feature; add T to the feature set.

[0073] The Mel frequency cepstral coefficient feature MFCC(n) obtained from acoustic data through voiceprint recognition is expressed as follows:

[0074]

[0075] Where K represents the total number of Mel filters, E k Let W represent the energy of the k-th filter; MFCC(n) is weighted and added to the feature set, with the weight matrix being W2.

[0076] The final feature set F is represented as:

[0077] F = {X(f), T, MFCC(n)}

[0078] Where β represents the characteristic attenuation coefficient.

[0079] It should be noted that the extraction and fusion of features from multiple data sources significantly improves the accuracy and comprehensiveness of equipment operating status assessment. While frequency domain analysis, time series analysis, and voiceprint recognition technologies are all effective when applied independently in equipment fault diagnosis, their fusion allows for a more comprehensive capture of equipment operating characteristics, enhancing the sensitivity and accuracy of fault identification. Furthermore, this multi-dimensional feature fusion method addresses the issue of insufficient information from a single data source, enabling more effective support for preventative maintenance and fault early warning, thereby improving equipment operational reliability and maintenance efficiency.

[0080] S3: Build and train deep learning models, continuously learn and optimize model parameters and structure to identify normal and abnormal states of devices.

[0081] Forward propagation computation is performed, using the feature set F as input, and the formula is as follows:

[0082] z (l) =g(W (l) F+b (l) )

[0083] Among them, z (l) W represents the activation value of the l-th layer, where l is any layer number from 1 to L. (l) Let b represent the weight matrix of the l-th layer. (l) Let g(·) represent the bias vector of the l-th layer, and g(·) represent the activation function.

[0084] It should be noted that deep features of the equipment's operating status are captured through multi-layer neural network calculations. Unlike traditional shallow feature extraction methods, deep learning models can automatically extract complex feature patterns from data, exhibiting significant advantages, particularly when processing nonlinear and high-dimensional data. This characteristic enables the present invention to more accurately identify subtle abnormal states of equipment, improving the sensitivity of fault detection.

[0085] The loss function is calculated using the following formula:

[0086]

[0087] in, Let m represent the loss value, m represent the number of samples, and y represent the loss value. i Indicates the actual value. This represents the final predicted value.

[0088] The formula for parameter optimization and feature fusion is as follows:

[0089]

[0090] Where Z represents the optimized comprehensive feature, and λ and μ represent the regularization coefficients; This represents the regularization term, controlling the change of the weight matrix over time. Regularization effectively avoids overfitting and ensures the model's generalization ability. In particular, the introduced time-varying control term dynamically adjusts the model parameters, allowing them to better adapt to changes in equipment status. This dynamic adaptability is especially important in scenarios where equipment operating states change frequently, providing more accurate and timely anomaly detection.

[0091] During equipment operation, if the optimized comprehensive feature Z is within the normal range, i.e. Z∈[0, Z], then... high The system adopts a normal operation mode based on historical data to continuously monitor the equipment.

[0092] Normal state threshold Z high By using feature clustering and density estimation methods, the system performs real-time telemetry on all collected sensor data and monitors and adjusts the status of key components of the equipment through remote control functions to keep the equipment in optimal operating condition.

[0093] If the comprehensive feature Z is greater than the abnormal state threshold Z threshold Z threshold <Z high The system switches to early warning mode, increasing the monitoring frequency of key features.

[0094] When the comprehensive feature Z exceeds the normal range, i.e. Z>Z highThe system enters anomaly handling mode, initiates anomaly identification and isolation mechanisms, and uses multi-level anomaly detection algorithms to analyze and classify anomalies, determining the anomaly type and severity.

[0095] The specific formula for performing preliminary anomaly detection on the comprehensive feature representation Z and filtering out outliers that deviate from the normal range is as follows:

[0096]

[0097] Among them, Z i Let μ represent the i-th eigenvalue. Z and σ Z Let Z represent the mean and standard deviation of the feature set Z, respectively.

[0098] If Anomaly i If the value exceeds the preset threshold θ, it is determined to be a preliminary anomaly.

[0099] For the outliers initially identified, a multi-level clustering algorithm is used for foresting and classification.

[0100] The DBSCAN algorithm is applied to perform density clustering on outliers to identify high-density outlier groups and noise points. The formula is as follows:

[0101]

[0102] Where ∈ represents the cluster radius parameter, Cluster i Representing point x i The clustering results centered on x are used; points that cannot be classified into any density cluster are marked as noise points; x represents any data point in the data space. Let represent a d-dimensional real space, where d represents the dimension of the data.

[0103] Based on the clustering results, the severity of outliers is probabilistically assessed using a Geometric Matrix (GMM). The GMM is fitted to each clustering result, calculating the probability that each outlier belongs to a different category, and the outlier type is determined based on maximum likelihood estimation. The specific formula is as follows:

[0104]

[0105] Wherein, P(x i ) represents point x i The probability density, π k Denotes the mixing coefficients of the k-th Gaussian distribution. The mean is μ k The covariance is ∑ k Gaussian distribution; determine the anomaly type based on the category with the highest probability density; when P(x iIf the value exceeds the set threshold Pθ, the current fault state of the device is considered to be severely abnormal.

[0106] It should be noted that the combined application of preliminary anomaly detection, density clustering, and GMM effectively solves the problems of unclear anomaly type identification and difficulty in quantifying severity in traditional methods. First, standardized deviation calculation effectively quantifies the anomaly degree of feature values, filtering out noteworthy preliminary anomalies. Second, the DBSCAN algorithm accurately identifies the group characteristics of anomalies by clustering the density distribution of the data. Unlike traditional methods such as K-means, this algorithm does not require a preset number of clusters and can naturally handle noisy points. Finally, GMM further refines the classification and severity assessment of anomalies, providing an intuitive quantitative evaluation standard through a probability density function. This series of steps not only enhances the early warning capability for equipment failures but also provides clear classification criteria and quantitative assessment for subsequent fault handling, offering crucial support for equipment maintenance and management.

[0107] S4: Perform maintenance actions based on abnormal conditions.

[0108] When the warning mode is triggered, the system automatically increases the remote control of the equipment and adjusts the equipment operating parameters to prevent abnormal conditions from occurring. At the same time, the warning information is fed back to the management personnel in real time through the remote communication module.

[0109] When the equipment experiences a severe malfunction, the equipment protection program is automatically triggered to perform remote shutdown or isolation operations. During this process, remote fault indicators and multimodal sensor fusion technology are used to accurately locate the fault location and send an anomaly report to the maintenance team. The system records and analyzes the entire process of the anomaly, and through deep learning of the data and continuous optimization of the model, it continuously improves the system's fault detection and handling capabilities. When the abnormal state is resolved, the system automatically performs a self-healing process to restore the equipment to normal operation.

[0110] It should be noted that the system performs corresponding maintenance actions based on the abnormal state of the equipment to ensure safe operation and efficient maintenance. When the system triggers the early warning mode, the early warning mechanism is activated, and the system automatically increases remote control over the equipment. This includes adjusting key operating parameters (such as power output, voltage, and temperature) to prevent the further development of potential abnormalities. At this time, the early warning information is fed back to the management personnel in real time through the remote communication module, allowing operators to understand the equipment status in a timely manner and perform necessary manual intervention. This remote control and real-time feedback mechanism not only helps prevent the escalation of faults but also identifies and resolves problems in their early stages, reducing equipment downtime and improving system reliability and stability.

[0111] When a serious anomaly is detected in the equipment, the system automatically triggers the equipment protection program, performing remote shutdown or isolation operations. This operation is achieved through a remote fault indicator and multimodal sensor fusion technology, which can precisely locate the specific location of the equipment fault. This localization technology combines data from multiple sensors, including vibration, temperature, and acoustic data, to provide a multi-dimensional view of the fault, ensuring the accuracy and reliability of the location. In the event of a serious anomaly, immediately shutting down or isolating the faulty equipment can effectively prevent the expansion and spread of the fault, protecting the safety of other equipment and systems. Simultaneously, the system generates a detailed anomaly report and sends it to the maintenance team via remote communication, enabling maintenance personnel to quickly understand the fault details and develop appropriate repair plans.

[0112] Throughout the recording and analysis of anomalies, the system employs deep learning and continuous model optimization techniques to continuously improve its fault detection and handling capabilities. This adaptive learning capability enables the system to adjust its detection model based on emerging fault modes, enhancing the accuracy of fault prediction and response speed. During fault handling, the system continuously accumulates and learns from new data, constantly optimizing and updating model parameters. This not only improves its ability to identify unknown faults but also allows it to better adapt to changes in the equipment's operating environment.

[0113] Once the abnormal condition is resolved, the system automatically initiates a self-healing process to restore the equipment to normal operation. This process includes readjusting the equipment's operating parameters, recalibrating sensors, and restoring normal monitoring and control modes. This automated self-healing process enables the equipment to quickly return to normal operation, reducing downtime, lowering maintenance costs, and ensuring continuous and stable operation.

[0114] The above embodiments also include a coal mining equipment data analysis system based on a big data platform, specifically:

[0115] Data acquisition module: Acquires multi-source data from coal mining equipment and preprocesses the data;

[0116] Data feature fusion module: Extracts features from preprocessed data, merges features from different data sources, and forms a unified set of equipment operating status features;

[0117] Model building module: Builds and trains deep learning models, continuously learns and optimizes model parameters and structure to identify normal and abnormal states of devices;

[0118] Anomaly alarm module: Based on abnormal conditions, it will trigger alarms and perform maintenance actions.

[0119] The computer device can be a server. This computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores data cluster data from the power monitoring system. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a data analysis method for coal mining equipment based on a big data platform.

[0120] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0121] Example 2 is an embodiment of the present invention, which provides a data analysis method for coal mining equipment based on a big data platform. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation / comparative experiments.

[0122] This experiment aims to verify the effectiveness of a data analysis method for coal mining equipment based on a big data platform in equipment fault detection and maintenance. The experimental subjects included six coal mining machines, numbered M1 to M6. Each machine was equipped with multiple sensors to collect multi-source data, covering vibration, temperature, acoustics, current, and historical operating information. Specifically, vibration data was recorded by an accelerometer, temperature data was acquired by thermocouples, acoustic data was collected by an acoustic sensor, and current data was monitored by a current transformer. In addition, historical operating data of the equipment was collected, including cumulative operating hours and past fault records.

[0123] The data preprocessing step first cleans the collected data to remove outliers and noise. Then, all data undergoes standardization to eliminate differences caused by variations in units and magnitudes. Time synchronization ensures time consistency across different data sources, unifying the timestamps of each data stream.

[0124] In the feature extraction and fusion stage, the main frequency components were extracted from the vibration signal through frequency domain analysis, the time-series characteristics of the temperature data were analyzed using an autoregressive model, and the Mel-frequency cepstral coefficients (MFCCs) of the acoustic data were extracted through voiceprint recognition. These features were fused into a comprehensive feature set to represent the overall operating status of the equipment.

[0125] Next, based on the extracted feature set, a deep learning model was constructed and trained. The model employs a multi-layer neural network structure and is optimized using the cross-entropy loss function. During training, the model parameters are continuously adjusted using forward and backpropagation algorithms to minimize prediction error. To prevent overfitting, early stopping and cross-validation techniques were used in the experiments.

[0126] In practical applications, the system executes different maintenance measures based on detected abnormal states. For equipment entering early warning mode, the system automatically adjusts its operating parameters and sends early warning information to management personnel in real time via remote communication. If a serious abnormality is detected, the system immediately implements protection mechanisms, including remote shutdown and equipment isolation. Simultaneously, multimodal sensor technology is used to accurately locate the fault point and generate a detailed anomaly report. Specific experimental data are shown in Table 1.

[0127] Table 1 Implementation Data Table

[0128]

[0129] The experimental data revealed significant differences in multi-source data across the various devices, particularly in vibration, temperature, and acoustic data. This data provides crucial information for device condition assessment. For instance, devices M5 and M6 exhibited high vibration and temperature data, reaching 0.70g and 0.68g respectively, and 85℃ and 83℃ respectively. Their acoustic data also reached 95dB and 93dB, exceeding the normal operating levels of other devices. The increased current data further indicates that these devices may be operating under high load. Historical operating data also shows that M5 and M6 had the longest cumulative operating time, at 6500 hours and 6400 hours respectively, and the most failures, at 5 and 4 respectively.

[0130] Through the training and application of deep learning models, the system can accurately identify the status of each device. M5 and M6 were detected as being in an abnormal state, consistent with their historical fault records and current monitoring data. In contrast, M1 and M3 were judged to be in a normal state, and actual data also showed that their parameters were within the normal range. M2 and M4 were in a warning state, which is consistent with their slightly higher vibration and temperature data, as well as a relatively high number of historical faults.

[0131] Compared to traditional single-parameter monitoring, this invention provides a more comprehensive equipment condition assessment capability through deep fusion and analysis of multi-source data. In particular, the extraction of data features and the application of deep learning models significantly improve the accuracy and sensitivity of fault detection. The combined use of the DBSCAN algorithm and GMM not only effectively identifies abnormal groups but also quantifies the severity of anomalies, providing a quantitative basis for maintenance decisions.

[0132] Furthermore, experimental results show that the system can effectively reduce the impact of equipment failures through early warning modes and automatic maintenance measures under abnormal conditions. For example, in early warning mode, timely parameter adjustments and early warning information feedback reduce the risk of potential equipment damage. In abnormal conditions, automatic shutdown and fault location ensure rapid isolation and handling of faults. This fully automated monitoring and maintenance mechanism not only improves equipment operational safety but also reduces maintenance costs and downtime, providing strong support for industrial production.

[0133] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A data analysis method for coal mining equipment based on a big data platform, characterized in that, include: Collect multi-source data from coal mining equipment and preprocess the data; Features are extracted from the preprocessed data, and features from different data sources are merged to form a unified set of equipment operating status features. Build and train deep learning models, and continuously learn to optimize model parameters and structure to identify normal and abnormal states of devices; Perform maintenance actions based on the abnormal status; The multi-source data collected from the coal mining equipment includes vibration data, temperature data, acoustic data, current data, and historical operating data. The preprocessing includes data cleaning, noise reduction, standardization, and time synchronization. The feature extraction from the preprocessed data includes obtaining the frequency domain features X(f) of the coal mining equipment vibration signal through frequency domain analysis, the expression of which is: Where x[n] represents the time domain signal, N represents the signal length, f represents the frequency, and j represents the imaginary unit. The weighted X(f) is used as part of the feature set, and the weight matrix is ​​W1. The trend characteristic T of temperature data is obtained through time series analysis, and its expression is: Where t0 and t N These represent the start and end points of the time interval, respectively, where t represents the time variable, and φ... i The coefficients of the autoregressive model are represented by p, which represents the lag order of the autoregressive model, and x represents the coefficients of the autoregressive model. t-i Represents temperature data at a lag time point, ∈ t Let T represent white noise, and α represent the attenuation coefficient of the time-series feature; add T to the feature set; The Mel frequency cepstral coefficient feature MFCC(n) obtained from acoustic data through voiceprint recognition is expressed as follows: Where K represents the total number of Mel filters, E k Let W represent the energy of the k-th filter; MFCC(n) is weighted and added to the feature set, with the weight matrix being W2. The final feature set F is represented as: F = {X(f), T, MFCC(n)} Where β represents the characteristic attenuation coefficient.

2. The data analysis method for coal mining equipment based on a big data platform as described in claim 1, characterized in that: The construction and training of the deep learning model includes performing forward propagation calculations, using the feature set F as input, with the following formula: z (l) =g(W (l) F+b (l) ) Among them, z (l) W represents the activation value of the l-th layer, where l is any layer number from 1 to L. (l) Let b represent the weight matrix of the l-th layer. (l) Let g(·) represent the bias vector of the l-th layer, and g(·) represent the activation function. The loss function is calculated using the following formula: in, Let m represent the loss value, m represent the number of samples, and y represent the loss value. i Indicates the actual value. This represents the final predicted value; The formula for parameter optimization and feature fusion is as follows: Where Z represents the optimized comprehensive feature, and λ and μ represent the regularization coefficients; This represents the regularization term, which controls the change of the weight matrix over time.

3. The data analysis method for coal mining equipment based on a big data platform as described in claim 2, characterized in that: The normal and abnormal states of the identification device include, during device operation, if the optimized comprehensive feature Z is within the normal range, i.e., Z∈[0,Z...] high The system adopts a normal operation mode based on historical data to continuously monitor the equipment; Normal state threshold Z high The system determines the status of key components of the equipment through feature clustering and density estimation methods, performs real-time telemetry on all collected sensor data, and monitors and adjusts the status of key components of the equipment through remote control functions to keep the equipment in optimal operating condition. If the comprehensive feature Z is greater than the abnormal state threshold Z threshold Z threshold <Z high The system switches to early warning mode, increasing the monitoring frequency of key features; When the comprehensive feature Z exceeds the normal range, i.e. Z>Z high The system enters anomaly handling mode, initiates anomaly identification and isolation mechanisms, and uses multi-level anomaly detection algorithms to analyze and classify anomalies, determining the anomaly type and severity.

4. The data analysis method for coal mining equipment based on a big data platform as described in claim 3, characterized in that: The analysis and classification of outliers using a multi-level anomaly detection algorithm includes preliminary anomaly detection on the comprehensive feature representation Z, and screening out outliers that deviate from the normal range. The specific formula is as follows: Among them, Z i Let μ represent the i-th eigenvalue. Z and σ Z Let Z represent the mean and standard deviation of the feature set Z, respectively. If Anomaly i If the value exceeds the preset threshold θ, it is determined to be a preliminary anomaly. For the anomalies initially identified, a multi-level clustering algorithm is used for forest segmentation and classification. The DBSCAN algorithm is applied to perform density clustering on outliers to identify high-density outlier groups and noise points. The formula is as follows: Where ∈ represents the cluster radius parameter, Cluster i Representing point x i The clustering results centered on x are used; points that cannot be classified into any density cluster are marked as noise points; x represents any data point in the data space. This represents a d-dimensional real space, where d represents the dimension of the data. Based on the clustering results, the severity of outliers is probabilistically assessed using a Geometric Matrix (GMM). The GMM is fitted to each clustering result, calculating the probability that each outlier belongs to a different category, and the outlier type is determined based on maximum likelihood estimation. The specific formula is as follows: Wherein, P(x i ) represents point x i The probability density, π k Denotes the mixing coefficients of the k-th Gaussian distribution. The mean is μ k The covariance is ∑ k Gaussian distribution; determine the anomaly type based on the category with the highest probability density; when P(x i If the value exceeds the set threshold Pθ, the current fault state of the device is considered to be severely abnormal.

5. The data analysis method for coal mining equipment based on a big data platform as described in claim 4, characterized in that: The maintenance actions include, when the early warning mode is triggered, the system automatically increases the remote control of the equipment, adjusts the equipment operating parameters to prevent abnormal conditions from occurring, and at the same time, the early warning information is fed back to the management personnel in real time through the remote communication module. When the equipment experiences a severe malfunction, the equipment protection program is automatically triggered to perform remote shutdown or isolation operations. During this process, remote fault indicators and multimodal sensor fusion technology are used to accurately locate the fault location and send an anomaly report to the maintenance team. The system records and analyzes the entire process of the anomaly, and through deep learning of the data and continuous optimization of the model, it continuously improves the system's fault detection and handling capabilities. When the abnormal state is resolved, the system automatically performs a self-healing process to restore the equipment to normal operation.

6. A coal mining equipment data analysis system based on a big data platform, employing the method described in any one of claims 1-5, characterized in that: Data acquisition module: Acquires multi-source data from coal mining equipment and preprocesses the data; Data feature fusion module: Extracts features from preprocessed data, merges features from different data sources, and forms a unified set of equipment operating status features; Model building module: Builds and trains deep learning models, continuously learns and optimizes model parameters and structure to identify normal and abnormal states of devices; Anomaly alarm module: Based on abnormal conditions, it will trigger alarms and perform maintenance actions.

7. A computer device 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 according to any one of claims 1 to 5.

8. 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 according to any one of claims 1 to 5.

Images