Coal mine environment abnormity monitoring system and method based on machine learning

Abstract

The invention provides a coal mine environment abnormity monitoring system and method based on machine learning, and the system comprises a data access module which is used for accessing multi-dimensional monitoring data of a coal mine environment in a monitoring center; the configuration module is used for establishing and configuring an abnormal rule in the monitoring center; and the exception monitoring module is used for performing exception detection on the multi-dimensional monitoring data according to the configuration result and based on machine learning, and outputting an exception data point and an exception analysis report. Multi-dimensional monitoring data of a coal mine environment is accessed into a monitoring center, and abnormal rules are effectively established and configured in the monitoring center, so that abnormal detection can be performed on the multi-dimensional monitoring data according to a configuration result and based on machine learning, and abnormal data points and abnormal analysis reports are output; according to the method, single-variable abnormity can be detected, abnormity analysis can be carried out by integrating data of multiple dimensions, abnormity reason analysis is provided, and managers are helped to rapidly identify potential risks and take measures.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a coal mine environment anomaly monitoring system and method based on machine learning. Background Art

[0002] At present, in the current coal mine environmental monitoring systems, sensors are usually used to collect various environmental data, including gas concentration, temperature, humidity, water level, flow, equipment status, etc.; these data are transmitted to the monitoring center through the network, and abnormal data are monitored by manual or system preset rules; when a certain indicator data exceeds the set threshold, the system will issue an alarm; the core of the system is to judge the abnormality of a single indicator data; However, the shortcomings of existing technologies include: single rules and lack of flexibility: fixed threshold settings do not take into account dynamic changes in the environment. Data such as oxygen, carbon monoxide, dust, and gas in the coal mine environment may fluctuate greatly in a short period of time, but such fluctuations do not necessarily indicate safety hazards. Relying solely on fixed thresholds will generate a large number of false alarms, interfering with normal monitoring; unable to process multi-dimensional data: most existing technologies can only monitor single sensor data, while safety issues in coal mines are usually the result of the combined effect of multi-dimensional factors. For example, abnormal gas concentration is usually accompanied by changes in wind speed and ventilation equipment status, and existing technologies cannot integrate multiple dimensions for analysis and detection; lack of abnormal data analysis and cause diagnosis: existing systems can detect anomalies, but cannot provide analysis and diagnosis reports of the causes of abnormalities, resulting in managers being unable to take timely and accurate countermeasures, affecting the safety management efficiency of coal mines; insufficient visualization: existing systems are mostly simple digital displays and alarm prompts, lacking intuitive visualization methods to help users understand the distribution and trends of abnormal data; Therefore, in order to overcome the above technical problems, the present invention provides a coal mine environmental abnormality monitoring system and method based on machine learning. Summary of the invention

[0003] The present invention provides a coal mine environment anomaly monitoring system and method based on machine learning, which is used to access multi-dimensional monitoring data of the coal mine environment in a monitoring center, and effectively establish and configure abnormal rules in the monitoring center, so as to facilitate abnormal detection of multi-dimensional monitoring data based on configuration results and based on machine learning, output abnormal data points and abnormal analysis reports, and can not only detect single variable anomalies, but also integrate data of multiple dimensions (such as gas concentration, temperature, ventilation, water level, flow rate and other equipment status and data) for abnormal analysis, and provide abnormal cause analysis to help managers quickly identify potential dangers and take measures.

[0004] The present invention provides a coal mine environment abnormality monitoring system based on machine learning, comprising: Data access module, used to access multi-dimensional monitoring data of coal mine environment in the monitoring center; Configuration module, used to establish and configure exception rules in the monitoring center; The anomaly monitoring module is used to detect anomalies in multi-dimensional monitoring data based on configuration results and machine learning, and output abnormal data points and anomaly analysis reports.

[0005] Preferably, a coal mine environment anomaly monitoring system based on machine learning, a data access module, comprises: A data storage node establishment unit is used to obtain sensors corresponding to each monitoring dimension and obtain the sensor model, and at the same time, establish a data storage node in the monitoring center according to the sensor model; The format conversion unit is used to obtain the data access format of the monitoring center and read the monitoring data of each sensor in real time. At the same time, the monitoring data is converted into a format according to the data access format to obtain standard monitoring data. An information queue generating unit, used for sorting the standard monitoring data corresponding to each sensor and generating an information queue corresponding to each sensor; An access queue generating unit, used to generate a verification code according to the type of the sensor, and encapsulate the verification code with the information queue to generate an access queue; The access and verification unit is used to upload the access queue to the monitoring center, verify the access queue according to the monitoring center, and transmit it to the corresponding data storage node after the verification is passed.

[0006] Preferably, a coal mine environment anomaly monitoring system based on machine learning, an access and verification unit, comprises: The access queue reading subunit is used to read the access queue based on the monitoring center after uploading the access queue to the monitoring center to determine the verification code of the access queue; The matching verification subunit is used to retrieve the sensor management library based on the monitoring center, and input the verification code into the sensor management library for matching, to determine whether there is a sensor model consistent with the verification code; A storage subunit is used to determine a corresponding data storage node according to the sensor model if there is a sensor model that is consistent with the verification code, and transmit the information queue in the access queue to the data storage node; The deletion subunit is used to delete the access queue based on the monitoring center if there is no sensor model consistent with the verification code.

[0007] Preferably, a coal mine environment anomaly monitoring system based on machine learning, a data access module, comprises: A data acquisition unit, used to acquire the obtained multi-dimensional monitoring data and determine the data source of each dimension of monitoring data respectively; Data pre-processing unit, used for: Based on the data source, the basic value interval of the corresponding dimension monitoring data is obtained, and the data value of each dimension monitoring data is matched with the basic value interval for attribution, and the abnormal data in each dimension monitoring data is obtained based on the attribution matching result; Extracting target values ​​of neighboring data of abnormal data, and correcting the data values ​​of abnormal data based on the target values ​​of neighboring data; Based on the correction result, the standardized interval of the monitoring data of each dimension is determined, and the monitoring data of each dimension is standardized based on the standardized interval.

[0008] Preferably, a coal mine environment anomaly monitoring system based on machine learning, the configuration module includes: A baseline safety indicator reading unit is used to read multiple baseline safety indicators of the coal mine environment, and match the multiple baseline safety indicators with multiple abnormal monitoring dimensions to obtain a target baseline safety indicator corresponding to each abnormal monitoring dimension; An abnormal rule construction unit is used to read the target baseline safety index and construct an abnormal rule corresponding to the abnormal monitoring dimension according to the reading result; The configuration unit is used to set the anomaly rules in the monitoring center and complete the configuration of the anomaly rules in the monitoring center according to the setting results.

[0009] Preferably, a coal mine environment anomaly monitoring system based on machine learning, an anomaly monitoring module, comprises: An abnormal data point identification unit is used to identify the abnormal data point and determine the data characteristics of the abnormal data point after outputting the abnormal data point and the abnormal analysis report; A test report generating unit, used to retrieve historical data corresponding to the abnormal data point from a preset database, and generate a test report based on the historical data and data features of the abnormal data point; The storage unit is used to read the abnormality analysis report and store the abnormality analysis report and the detection report in a preset database.

[0010] Preferably, a coal mine environment anomaly monitoring system based on machine learning, an anomaly monitoring module, comprises: A display graph model retrieving unit, used to retrieve the display graph model and read the display rules of the display graph model after the abnormal data point is output, wherein there are multiple display graph models, and the display graph models correspond to the display rules one by one; A mapping unit, used to map the abnormal data points in the display graph model according to the display rule of the display graph model to obtain a target display graph model; The visual unit is used to create a visualization interface, import the target display graph model into the visualization interface, and display the target display graph model in the visualization interface based on the import result.

[0011] Preferably, a coal mine environment anomaly monitoring system based on machine learning, an anomaly monitoring module, comprises: Machine Learning Unit for: Retrieve multi-dimensional historical data samples from the preset database, and analyze the multi-dimensional historical data samples in turn using multiple categories of algorithms based on the configured monitoring center, and obtain the anomaly detection status of the multi-dimensional historical data samples under different categories of algorithms; Based on the anomaly detection status, determine the abnormal data in the multi-dimensional historical sample data, analyze the abnormal data, and obtain the data change trends of the abnormal data under different categories of algorithms; Determine the state representation of abnormal data under different categories of algorithms based on data change trends, and perform machine learning on the monitoring center based on the state representation; Anomaly detection unit, used to: Based on the results of machine learning, anomaly detection is performed on multi-dimensional monitoring data to obtain the abnormal data corresponding to each dimensional monitoring data. At the same time, the topological structure between the multi-dimensional monitoring data and the data features corresponding to the multi-dimensional monitoring data are determined based on the preset knowledge system, and the interaction parameters of the abnormal data in different dimensions are determined based on the topological structure and data features. Correct the abnormal data based on the interaction parameters and recheck the corrected abnormal data; Data isolation unit for: Determine the final abnormal data points based on the re-inspection results, and determine the distribution status of the abnormal data points in the corresponding dimension monitoring data; Construct a target hyperplane, and divide the abnormal data points and normal data points in each dimension of the monitoring data according to the distribution state based on the target hyperplane, and output the abnormal data points based on the division result.

[0012] Preferably, a coal mine environment anomaly monitoring system based on machine learning, an anomaly monitoring module, comprises: A real-time data connection unit is used to construct a real-time data flow channel between the monitoring center and the data source of the multi-dimensional monitoring data based on the configuration results, and input the multi-dimensional monitoring data acquired in real time into the monitoring center for anomaly detection based on the real-time data flow channel; The abnormal data analysis unit analyzes the abnormal data points obtained based on the abnormal detection results to obtain the abnormal causes corresponding to the abnormal data points. At the same time, the target weight of each dimension of monitoring data is determined based on the monitoring requirements, and the obtained abnormal data points are prioritized based on the target weight; The report generation unit is used to create a report template, and based on the priority sorting result, the dimensions corresponding to the abnormal data points and the abnormal causes are recorded in the report template in turn to obtain an abnormal analysis report.

[0013] The present invention provides a method for monitoring coal mine environmental anomalies based on machine learning, comprising: Step 1: Access the multi-dimensional monitoring data of the coal mine environment in the monitoring center; Step 2: Establish and configure exception rules in the monitoring center; Step 3: Perform anomaly detection on multi-dimensional monitoring data based on the configuration results and machine learning, and output abnormal data points and anomaly analysis reports.

[0014] Compared with the prior art, the present invention has the following beneficial effects: By accessing the multi-dimensional monitoring data of the coal mine environment in the monitoring center and effectively establishing and configuring the abnormal rules in the monitoring center, it is beneficial to detect anomalies in the multi-dimensional monitoring data based on the configuration results and machine learning, and output abnormal data points and abnormal analysis reports. It can not only detect single variable anomalies, but also integrate data from multiple dimensions (such as gas concentration, temperature, ventilation, water level, flow rate and other equipment status and data) for abnormal analysis, and provide abnormal cause analysis to help managers quickly identify potential dangers and take measures.

[0015] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures specifically pointed out in this application document.

[0016] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 This is a structural diagram of a coal mine environment anomaly monitoring system based on machine learning in an embodiment of the present invention; Figure 2 This is a structural diagram of a data access module in a coal mine environment abnormality monitoring system based on machine learning in Example 2 of the present invention; Figure 3 The present invention is a flowchart of a method for monitoring coal mine environmental anomalies based on machine learning in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0019] Example 1 This embodiment provides a coal mine environment abnormality monitoring system based on machine learning, such as Figure 1 As shown, including: Data access module, used to access multi-dimensional monitoring data of coal mine environment in the monitoring center; Configuration module, used to establish and configure exception rules in the monitoring center; The anomaly monitoring module is used to detect anomalies in multi-dimensional monitoring data based on configuration results and machine learning, and output abnormal data points and anomaly analysis reports.

[0020] In this embodiment, the user connects the original data to the monitoring center through custom configuration or system built-in rules, and the data can be imported through file import, database connection, etc. The monitoring center can access various sensor data in the coal mine, including basic information and real-time information of various equipment, such as gas concentration, temperature, humidity, water level, flow, real-time status of equipment, etc., and perform preprocessing, including denoising, standardization and missing data filling.

[0021] In this embodiment, the establishment and configuration of abnormal rules in the monitoring center can be: the monitoring center generates identification rules corresponding to abnormal data, as well as data ranges, thresholds, trends and other conditions based on data characteristics and coal mine scenarios where indicators are applied, and formulates abnormal rules according to specific application scenarios and data characteristics, thereby configuring the abnormal rules based on the above information, wherein the abnormal rules include: definition rules for abnormal data identification and abnormal data range, etc.

[0022] In this embodiment, the machine learning algorithm includes: Support Vector Machine (SVM): used to process multi-dimensional data and perform anomaly detection. By constructing a hyperplane, normal data and abnormal data are divided. SVM is good at processing high-dimensional data and effectively distinguishing normal and abnormal points with similar characteristics; Isolation Forest: This algorithm is particularly suitable for high-dimensional and diverse data scenarios. It identifies isolated points, i.e., abnormal points, by randomly dividing the data layer by layer. Isolation Forest has a strong advantage in processing multi-dimensional data; LSTM (Long Short-Term Memory Network): LSTM is particularly effective when processing time series data. The algorithm captures the trend and changes of the data by learning the time dependency of the data, and can identify abnormal points in periodic changes.

[0023] In this embodiment, multi-dimensional anomaly detection: This method supports multi-dimensional abnormal data detection. Users can consider multiple features or attributes at the same time to more accurately identify abnormal data; support simultaneous analysis of multi-dimensional data, taking into account multiple features or attributes, and provide more accurate anomaly detection capabilities. For example: by simultaneously analyzing multi-dimensional features such as sensor temperature, pressure, and humidity, the system can detect abnormal situations for each feature alone or in combination of multiple features. Users can choose a single algorithm (such as Grubbs' Test) or a combination of multiple algorithms (such as SVM + LSTM) for joint analysis according to their needs to improve detection accuracy.

[0024] In this embodiment, the abnormal data point may be abnormal data existing in the multi-dimensional monitoring data.

[0025] In this embodiment, the abnormality analysis report may be a file that records abnormal data and the cause of the abnormality.

[0026] The working principle and beneficial effects of the above technical solution are: by accessing the multi-dimensional monitoring data of the coal mine environment in the monitoring center and effectively establishing and configuring the abnormal rules in the monitoring center, it is beneficial to detect abnormalities in the multi-dimensional monitoring data based on the configuration results and based on machine learning, and output abnormal data points and abnormal analysis reports. It can not only detect single variable anomalies, but also integrate data from multiple dimensions (such as gas concentration, temperature, ventilation, water level, flow rate and other equipment status and data) for abnormal analysis, and provide abnormal cause analysis to help managers quickly identify potential dangers and take measures.

[0027] Example 2 Based on Example 1, this embodiment provides a coal mine environment abnormality monitoring system based on machine learning, such as Figure 2 As shown, the data access module includes: A data storage node establishment unit is used to obtain sensors corresponding to each monitoring dimension and obtain the sensor model, and at the same time, establish a data storage node in the monitoring center according to the sensor model; The format conversion unit is used to obtain the data access format of the monitoring center and read the monitoring data of each sensor in real time. At the same time, the format of the monitoring data is converted according to the data access format to obtain standard monitoring data; An information queue generating unit, used for sorting the standard monitoring data corresponding to each sensor and generating an information queue corresponding to each sensor; An access queue generating unit, used to generate a verification code according to the type of the sensor, and encapsulate the verification code with the information queue to generate an access queue; The access and verification unit is used to upload the access queue to the monitoring center, verify the access queue according to the monitoring center, and transmit it to the corresponding data storage node after the verification is passed.

[0028] In this embodiment, the data storage node may be a cache space for storing data monitored by each sensor.

[0029] In this embodiment, the standard monitoring data may be data that can be effectively stored in a data storage node after format conversion of data monitored by the sensor, meeting the data storage node's requirements for the data format.

[0030] In this embodiment, the access queue may be a result obtained by encapsulating a verification code generated according to the type of the sensor and the information queue.

[0031] The working principle and beneficial effects of the above technical solution are: by determining the sensors corresponding to each monitoring dimension and the corresponding models of the sensors, corresponding data storage nodes are created in the monitoring center according to the models of each sensor, so as to facilitate the storage of data monitored by different sensors; secondly, according to the data access format of the monitoring center, the monitoring data monitored by each sensor is format converted and sorted to obtain the corresponding information queue; finally, a verification code is generated according to the type of each sensor, and the verification code is bound to the obtained information queue, and finally the monitored data is transmitted to the data storage node for effective storage according to the binding result, which provides reliable data support for abnormal coal mine environment monitoring.

[0032] Example 3 Based on Example 2, this embodiment provides a coal mine environment abnormality monitoring system based on machine learning, and an access and verification unit includes: The access queue reading subunit is used to read the access queue based on the monitoring center after uploading the access queue to the monitoring center to determine the verification code of the access queue; The matching verification subunit is used to retrieve the sensor management library based on the monitoring center, and input the verification code into the sensor management library for matching, to determine whether there is a sensor model consistent with the verification code; A storage subunit is used to determine a corresponding data storage node according to the sensor model if there is a sensor model that is consistent with the verification code, and transmit the information queue in the access queue to the data storage node; The deletion subunit is used to delete the access queue based on the monitoring center if there is no sensor model consistent with the verification code.

[0033] In this embodiment, the sensor management library is pre-set and is used to store verification code information that can verify the identity of each sensor.

[0034] The working principle and beneficial effects of the above technical solution are: by verifying the identity of the sensor when receiving the access queue, and after the sensor's identity authentication is passed, the access queue is transmitted to the corresponding data storage node for storage, otherwise, the received access queue is deleted, thereby ensuring the accuracy and reliability of the obtained monitoring data.

[0035] Example 4 On the basis of Example 1, this Example 1 provides a coal mine environment abnormality monitoring system based on machine learning, and a data access module, including: A data acquisition unit, used to acquire the obtained multi-dimensional monitoring data and determine the data source of each dimension of monitoring data respectively; Data pre-processing unit, used for: Based on the data source, the basic value interval of the corresponding dimension monitoring data is obtained, and the data value of each dimension monitoring data is matched with the basic value interval for attribution, and the abnormal data in each dimension monitoring data is obtained based on the attribution matching result; Extracting target values ​​of neighboring data of abnormal data, and correcting the data values ​​of abnormal data based on the target values ​​of neighboring data; Based on the correction result, the standardized interval of the monitoring data of each dimension is determined, and the monitoring data of each dimension is standardized based on the standardized interval.

[0036] In this embodiment, the data source may be a source corresponding to the monitoring data of each dimension, that is, a device or sensor that generates the monitoring data of the corresponding dimension.

[0037] In this embodiment, the basic value interval may be a theoretical value range corresponding to monitoring data of different dimensions, that is, an allowable value range for different types of data.

[0038] In this embodiment, the attribution matching may be to match the data value of each dimensional monitoring data with the corresponding basic value interval, in order to determine whether the data value of each dimensional monitoring data is within the basic value interval.

[0039] In this embodiment, abnormal data may be data whose data value of each dimension of monitoring data is not in the corresponding basic value range.

[0040] In this embodiment, the adjacent data may be two data adjacent to the abnormal data, wherein the target value is the value of the adjacent data.

[0041] In this embodiment, the standardization interval may be a value interval corresponding to the standardization processing of each dimension monitoring data, for example, it may be [0, 1].

[0042] The working principle and beneficial effects of the above technical solution are: by determining the data source of each dimensional monitoring data, the basic value interval corresponding to each dimensional monitoring data is determined according to the data source, so that the abnormal data in each dimensional monitoring data can be locked according to the basic value interval, and secondly, the value of the abnormal data is corrected according to the adjacent data of the abnormal data, and finally, the monitoring data of each dimension is standardized according to the correction result, so as to facilitate the effective processing of the monitoring data of each dimension.

[0043] Example 5 Based on Example 1, this embodiment provides a coal mine environment abnormality monitoring system based on machine learning, and the configuration module includes: A baseline safety indicator reading unit is used to read multiple baseline safety indicators of the coal mine environment, and match the multiple baseline safety indicators with multiple abnormal monitoring dimensions to obtain a target baseline safety indicator corresponding to each abnormal monitoring dimension; An abnormal rule construction unit is used to read the target baseline safety index and construct an abnormal rule corresponding to the abnormal monitoring dimension according to the reading result; The configuration unit is used to set the anomaly rules in the monitoring center and complete the configuration of the anomaly rules in the monitoring center according to the setting results.

[0044] In this embodiment, the benchmark safety index may be a safety requirement index used to characterize the coal mine environment, such as a standard value range corresponding to the concentration of different gases.

[0045] In this embodiment, the anomaly monitoring dimension may be the type of data in which the anomaly occurs.

[0046] In this embodiment, the target baseline safety indicator may be a baseline safety indicator corresponding to each monitoring dimension obtained by matching different abnormal monitoring dimensions with the baseline safety indicator.

[0047] In this embodiment, the anomaly rule may be a strategy or method for analyzing different anomaly monitoring dimensions.

[0048] The working principle and beneficial effects of the above technical solution are: by determining multiple baseline safety indicators of the coal mine environment and matching the baseline safety indicators with the abnormal monitoring dimensions, the target baseline safety indicators for each abnormal monitoring dimension are determined; secondly, the abnormal rules for each abnormal monitoring dimension are constructed according to the determined target baseline safety indicators, and the constructed abnormal rules are set in the monitoring center, so as to facilitate accurate and reliable abnormal monitoring of the coal mine environment according to the setting results.

[0049] Example 6 Based on Example 1, this embodiment provides a coal mine environment abnormality monitoring system based on machine learning, and the abnormality monitoring module includes: An abnormal data point identification unit is used to identify the abnormal data point and determine the data characteristics of the abnormal data point after outputting the abnormal data point and the abnormal analysis report; A test report generating unit, used to retrieve historical data corresponding to the abnormal data point from a preset database, and generate a test report based on the historical data and data features of the abnormal data point; The storage unit is used to read the abnormality analysis report and store the abnormality analysis report and the detection report in a preset database.

[0050] In this embodiment, the monitoring center can store the detected abnormal data, including the specific value of the abnormal data, the time of occurrence, the sensor location and the abnormal type, and generate an abnormality report in combination with historical data, analyze the causes and provide corresponding suggestions, and briefly summarize the causes.

[0051] In this embodiment, the data features of the abnormal data point may include: specific value of the abnormal data, occurrence time, sensor location and abnormal type.

[0052] In this embodiment, the preset database is set in advance and is used to store different types of historical data.

[0053] The working principle and beneficial effects of the above technical solution are: by identifying abnormal data points, the data characteristics of the abnormal data points can be accurately and effectively determined. At the same time, historical data corresponding to the abnormal data points are retrieved from the preset database to generate a detection report based on the historical data and the data characteristics of the abnormal data points. Finally, the abnormal analysis report and the detection report are stored in the preset database at the same time, thereby facilitating accurate and effective understanding of the environmental detection results.

[0054] Example 7 Based on Example 1, this embodiment provides a coal mine environment abnormality monitoring system based on machine learning, characterized in that the abnormality monitoring module includes: A display graph model retrieving unit, used to retrieve the display graph model and read the display rules of the display graph model after the abnormal data point is output, wherein there are multiple display graph models, and the display graph models correspond to the display rules one by one; A mapping unit, used to map the abnormal data points in the display graph model according to the display rule of the display graph model to obtain a target display graph model; The visual unit is used to create a visualization interface, import the target display graph model into the visualization interface, and display the target display graph model in the visualization interface based on the import result.

[0055] In this embodiment, the display graph model includes: a scatter plot, a box plot, and a heat map, wherein the heat map shows the distribution of data points, for example, it can be set as green points for normal data and red points for abnormal data; the upper and lower limit dotted lines intuitively indicate the abnormal thresholds set by the system; the user can view the detailed information of the abnormal point by clicking on the abnormal point; the box plot shows the distribution, extreme values ​​and outliers of the data, which is convenient for quickly identifying the abnormal range of the data; when analyzing multidimensional data, the heat map can show the dense areas of abnormal data in multiple dimensions, providing the user with a global overview.

[0056] In this embodiment, the display rule may be a display restriction condition for data in each display graph model. For example, in a scatter plot, the display rule is that green points represent normal data and red points represent abnormal data.

[0057] In this embodiment, the target display graph model may be a result obtained by visually displaying abnormal data points in the display graph model according to the display rules.

[0058] The working principle and beneficial effects of the above technical solution are: by calling the display graph model and the display rules of the display model, the abnormal data points are mapped in the display graph model according to the display rules; finally, a visualization interface is constructed to display the target display graph model in the visualization interface, so that the management personnel can effectively understand the specific situation of the abnormal data.

[0059] Example 8 Based on Example 1, this embodiment provides a coal mine environment abnormality monitoring system based on machine learning, and the abnormality monitoring module includes: Machine Learning Unit for: Retrieve multi-dimensional historical data samples from the preset database, and analyze the multi-dimensional historical data samples in turn using multiple categories of algorithms based on the configured monitoring center, and obtain the anomaly detection status of the multi-dimensional historical data samples under different categories of algorithms; Based on the anomaly detection status, determine the abnormal data in the multi-dimensional historical sample data, analyze the abnormal data, and obtain the data change trends of the abnormal data under different categories of algorithms; Determine the state representation of abnormal data under different categories of algorithms based on data change trends, and perform machine learning on the monitoring center based on the state representation; Anomaly detection unit, used to: Based on the results of machine learning, anomaly detection is performed on multi-dimensional monitoring data to obtain the abnormal data corresponding to each dimensional monitoring data. At the same time, the topological structure between the multi-dimensional monitoring data and the data features corresponding to the multi-dimensional monitoring data are determined based on the preset knowledge system, and the interaction parameters of the abnormal data in different dimensions are determined based on the topological structure and data features. Correct the abnormal data based on the interaction parameters and recheck the corrected abnormal data; Data isolation unit for: Determine the final abnormal data points based on the re-inspection results, and determine the distribution status of the abnormal data points in the corresponding dimension monitoring data; Construct a target hyperplane, and divide the abnormal data points and normal data points in each dimension of the monitoring data according to the distribution state based on the target hyperplane, and output the abnormal data points based on the division result.

[0060] In this embodiment, the multi-dimensional historical data samples may be data samples of different types retrieved from a preset database.

[0061] In this embodiment, the analysis of multiple categories of algorithms can be performed by analyzing historical data samples of different dimensions through multiple different types of algorithms, thereby facilitating the training of algorithms of different categories and facilitating the analysis of real-time data through trained algorithms.

[0062] In this embodiment, the anomaly detection status may be the corresponding result obtained after different algorithms analyze the multi-dimensional historical data, that is, the result that can be used to train different algorithms.

[0063] In this embodiment, the abnormal data may be erroneous data existing in the multi-dimensional historical sample data.

[0064] In this embodiment, the data change trend may be the change of different abnormal data under different categories of algorithms.

[0065] In this embodiment, the state representation may be a specific manifestation of abnormal data detected under different types of algorithms, such as abnormal data values ​​and abnormal data components.

[0066] In this embodiment, the abnormal data may be the result obtained by performing anomaly detection on multi-dimensional monitoring data through a multi-category algorithm after machine learning.

[0067] In this embodiment, the preset knowledge system may be all data related to coal mine environmental monitoring.

[0068] In this embodiment, the topological structure may represent the association relationship between different multi-dimensional data.

[0069] In this embodiment, the data feature may be information such as data values ​​corresponding to the multi-dimensional monitoring data and corresponding data components.

[0070] In this embodiment, the interaction parameter may be a mutual limitation between different abnormal data.

[0071] In this embodiment, the abnormal data point may be the final abnormal data obtained by retesting the abnormal data initially determined according to the interaction parameter.

[0072] In this embodiment, the distribution status may be the distribution position and value distribution of abnormal data points in each dimension of monitoring data.

[0073] In this embodiment, the target hyperplane may be a parameter for dividing abnormal data points and normal data points in the monitoring data of each dimension.

[0074] The working principle and beneficial effects of the above technical solution are: by retrieving multi-dimensional historical data samples from a preset database and parsing the multi-dimensional historical data samples through a multi-category algorithm, the anomaly detection state under the non-category algorithm can be accurately and effectively determined; secondly, the abnormal data existing in the multi-dimensional historical sample data is determined through the anomaly detection state, and the abnormal data is analyzed to realize machine learning of the monitoring center according to the state representation of the abnormal data, thereby facilitating the effective detection of multi-dimensional detection data through the monitoring center; finally, the multi-dimensional monitoring data is monitored for anomalies through the trained monitoring center, and the abnormal data monitored for anomalies is parsed to realize the determination of the data characteristics of the abnormal data, and the abnormal data is corrected and re-inspected according to the data characteristics, and finally the division of abnormal data points and normal data points is realized, and the abnormal data points are accurately and effectively determined.

[0075] Example 9 Based on Example 1, this embodiment provides a coal mine environment abnormality monitoring system based on machine learning, and the abnormality monitoring module includes: A real-time data connection unit is used to construct a real-time data flow channel between the monitoring center and the data source of the multi-dimensional monitoring data based on the configuration results, and input the multi-dimensional monitoring data acquired in real time into the monitoring center for anomaly detection based on the real-time data flow channel; The abnormal data analysis unit analyzes the abnormal data points obtained based on the abnormal detection results to obtain the abnormal causes corresponding to the abnormal data points. At the same time, the target weight of each dimension of monitoring data is determined based on the monitoring requirements, and the obtained abnormal data points are prioritized based on the target weight; The report generation unit is used to create a report template, and based on the priority sorting result, the dimensions corresponding to the abnormal data points and the abnormal causes are recorded in the report template in turn to obtain an abnormal analysis report.

[0076] In this embodiment, the real-time data stream channel may be a channel for transmitting data between the monitoring center and the data source of the multi-dimensional monitoring data.

[0077] In this embodiment, the monitoring requirements are known in advance and are used to characterize the important situations of the monitoring data in each dimension.

[0078] The working principle and beneficial effects of the above technical solution are: by constructing a real-time data stream channel, the multi-dimensional monitoring data acquired in real time is transmitted to the monitoring center for anomaly detection; secondly, the abnormal data points obtained from the anomaly detection results are analyzed to effectively determine the abnormal causes corresponding to the abnormal data points; finally, the target weight of each dimension of monitoring data is determined according to the monitoring requirements, the abnormal data points are sorted according to the target weight, and the sorting results and abnormal causes are recorded in the report template to effectively determine the abnormal analysis report.

[0079] Embodiment 10: A coal mine environmental anomaly monitoring method based on machine learning, such as Figure 3 As shown, including: Step 1: Access the multi-dimensional monitoring data of the coal mine environment in the monitoring center; Step 2: Establish and configure exception rules in the monitoring center; Step 3: Perform anomaly detection on multi-dimensional monitoring data based on the configuration results and machine learning, and output abnormal data points and anomaly analysis reports.

[0080] The working principle and beneficial effects of the above technical solution are: by accessing the multi-dimensional monitoring data of the coal mine environment in the monitoring center and effectively establishing and configuring the abnormal rules in the monitoring center, it is beneficial to detect abnormalities in the multi-dimensional monitoring data based on the configuration results and based on machine learning, and output abnormal data points and abnormal analysis reports. It can not only detect single variable anomalies, but also integrate data from multiple dimensions (such as gas concentration, temperature, ventilation, water level, flow rate and other equipment status and data) for abnormal analysis, and provide abnormal cause analysis to help managers quickly identify potential dangers and take measures.

[0081] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A coal mine environment anomaly monitoring system based on machine learning, characterized in that: include: Data access module, used to access multi-dimensional monitoring data of coal mine environment in the monitoring center; Configuration module, used to establish and configure exception rules in the monitoring center; The anomaly monitoring module is used to detect anomalies in multi-dimensional monitoring data based on configuration results and machine learning, and output abnormal data points and anomaly analysis reports; Data access module, including: A data storage node establishment unit is used to obtain sensors corresponding to each monitoring dimension and obtain the sensor model, and at the same time, establish a data storage node in the monitoring center according to the sensor model; The format conversion unit is used to obtain the data access format of the monitoring center and read the monitoring data of each sensor in real time. At the same time, the format of the monitoring data is converted according to the data access format to obtain standard monitoring data; An information queue generating unit, used for sorting the standard monitoring data corresponding to each sensor and generating an information queue corresponding to each sensor; An access queue generating unit, used to generate a verification code according to the type of the sensor, and encapsulate the verification code with the information queue to generate an access queue; The access and verification unit is used to upload the access queue to the monitoring center, verify the access queue according to the monitoring center, and transmit it to the corresponding data storage node after the verification is passed.

2. The coal mine environment abnormality monitoring system based on machine learning according to claim 1 is characterized in that: Access and authentication unit, including: The access queue reading subunit is used to read the access queue based on the monitoring center after uploading the access queue to the monitoring center to determine the verification code of the access queue; The matching verification subunit is used to retrieve the sensor management library based on the monitoring center, and input the verification code into the sensor management library for matching, to determine whether there is a sensor model consistent with the verification code; A storage subunit is used to determine a corresponding data storage node according to the sensor model if there is a sensor model that is consistent with the verification code, and transmit the information queue in the access queue to the data storage node; The deletion subunit is used to delete the access queue based on the monitoring center if there is no sensor model consistent with the verification code.

3. The coal mine environment abnormality monitoring system based on machine learning according to claim 1 is characterized in that: Data access module, including: A data acquisition unit, used to acquire the obtained multi-dimensional monitoring data and determine the data source of each dimension of monitoring data respectively; Data pre-processing unit, used for: Based on the data source, the basic value interval of the corresponding dimension monitoring data is obtained, and the data value of each dimension monitoring data is matched with the basic value interval for attribution, and the abnormal data in each dimension monitoring data is obtained based on the attribution matching result; Extracting target values ​​of neighboring data of abnormal data, and correcting the data values ​​of abnormal data based on the target values ​​of neighboring data; Based on the correction result, the standardized interval of the monitoring data of each dimension is determined, and the monitoring data of each dimension is standardized based on the standardized interval.

4. The coal mine environment abnormality monitoring system based on machine learning according to claim 1 is characterized in that: Configuration modules, including: A baseline safety indicator reading unit is used to read multiple baseline safety indicators of the coal mine environment, and match the multiple baseline safety indicators with multiple abnormal monitoring dimensions to obtain a target baseline safety indicator corresponding to each abnormal monitoring dimension; An abnormal rule construction unit is used to read the target baseline safety index and construct an abnormal rule corresponding to the abnormal monitoring dimension according to the reading result; The configuration unit is used to set the anomaly rules in the monitoring center and complete the configuration of the anomaly rules in the monitoring center according to the setting results.

5. The coal mine environment abnormality monitoring system based on machine learning according to claim 1 is characterized in that: Abnormal monitoring module, including: An abnormal data point identification unit is used to identify the abnormal data point and determine the data characteristics of the abnormal data point after outputting the abnormal data point and the abnormal analysis report; A test report generating unit, used to retrieve historical data corresponding to the abnormal data point from a preset database, and generate a test report based on the historical data and data features of the abnormal data point; The storage unit is used to read the abnormality analysis report and store the abnormality analysis report and the detection report in a preset database.

6. The coal mine environment abnormality monitoring system based on machine learning according to claim 1 is characterized in that: Abnormal monitoring module, including: A display graph model retrieving unit, used to retrieve the display graph model and read the display rules of the display graph model after the abnormal data point is output, wherein there are multiple display graph models, and the display graph models correspond to the display rules one by one; A mapping unit, used to map the abnormal data points in the display graph model according to the display rule of the display graph model to obtain a target display graph model; The visual unit is used to create a visualization interface, import the target display graph model into the visualization interface, and display the target display graph model in the visualization interface based on the import result.

7. The coal mine environment abnormality monitoring system based on machine learning according to claim 1 is characterized in that: Abnormal monitoring module, including: Machine Learning Unit for: Retrieve multi-dimensional historical data samples from the preset database, and analyze the multi-dimensional historical data samples in turn using multiple categories of algorithms based on the configured monitoring center, and obtain the anomaly detection status of the multi-dimensional historical data samples under different categories of algorithms; Based on the anomaly detection status, determine the abnormal data in the multi-dimensional historical sample data, analyze the abnormal data, and obtain the data change trends of the abnormal data under different categories of algorithms; Determine the state representation of abnormal data under different categories of algorithms based on data change trends, and perform machine learning on the monitoring center based on the state representation; Anomaly detection unit, used to: Based on the results of machine learning, anomaly detection is performed on multi-dimensional monitoring data to obtain the abnormal data corresponding to each dimensional monitoring data. At the same time, the topological structure between the multi-dimensional monitoring data and the data features corresponding to the multi-dimensional monitoring data are determined based on the preset knowledge system, and the interaction parameters of the abnormal data in different dimensions are determined based on the topological structure and data features. Correct the abnormal data based on the interaction parameters and recheck the corrected abnormal data; Data isolation unit for: Determine the final abnormal data points based on the re-inspection results, and determine the distribution status of the abnormal data points in the corresponding dimension monitoring data; Construct a target hyperplane, and divide the abnormal data points and normal data points in each dimension of the monitoring data according to the distribution state based on the target hyperplane, and output the abnormal data points based on the division result.

8. The coal mine environment abnormality monitoring system based on machine learning according to claim 1 is characterized in that: Abnormal monitoring module, including: A real-time data connection unit is used to construct a real-time data flow channel between the monitoring center and the data source of the multi-dimensional monitoring data based on the configuration results, and input the multi-dimensional monitoring data acquired in real time into the monitoring center for anomaly detection based on the real-time data flow channel; The abnormal data analysis unit analyzes the abnormal data points obtained based on the abnormal detection results to obtain the abnormal causes corresponding to the abnormal data points. At the same time, the target weight of each dimension of monitoring data is determined based on the monitoring requirements, and the obtained abnormal data points are prioritized based on the target weight; The report generation unit is used to create a report template, and based on the priority sorting result, the dimensions corresponding to the abnormal data points and the abnormal causes are recorded in the report template in turn to obtain an abnormal analysis report.

9. A method for monitoring coal mine environmental anomalies based on machine learning, characterized in that: include: Step 1: Access the multi-dimensional monitoring data of the coal mine environment in the monitoring center; Step 2: Establish and configure exception rules in the monitoring center; Step 3: Perform anomaly detection on multi-dimensional monitoring data based on the configuration results and machine learning, and output abnormal data points and anomaly analysis reports; Step 1 includes: Obtain the sensors corresponding to each monitoring dimension and the sensor model. At the same time, establish a data storage node in the monitoring center according to the sensor model. Obtain the data access format of the monitoring center and read the monitoring data of each sensor in real time. At the same time, convert the monitoring data format according to the data access format to obtain standard monitoring data; Sort the standard monitoring data corresponding to each sensor and generate an information queue corresponding to each sensor; Generate a verification code according to the type of sensor, and encapsulate the verification code with the information queue to generate an access queue; The access queue is uploaded to the monitoring center, and the access queue is verified by the monitoring center, and when the verification passes, it is transmitted to the corresponding data storage node.

Images