Coal mine rock burst disaster early warning method and system based on feature drift

By dynamically adjusting the sampling points and observation window lengths, combined with the time series algorithm model, the limitations of traditional coal mine impact ground pressure disaster monitoring methods are solved, more accurate risk assessment and early warning are achieved, and monitoring efficiency and economic benefits are improved.

CN118966426APending Publication Date: 2024-11-15UNIV OF SCI & TECH BEIJING
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411019983.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-29
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

Traditional coal mine impact ground pressure disaster monitoring methods are difficult to effectively adapt to risk changes in different regions under complex and changing geological conditions, resulting in insufficient monitoring efficiency and accuracy, and lack of dynamic analysis and prediction of real-time data, which affects the timeliness and accuracy of early warnings. sex.

Method used

Using a method based on feature drift, the number of sampling points and the length of the observation window is dynamically adjusted, and real-time data analysis and prediction are combined with the time series algorithm model. By preprocessing and correcting stress data and temperature data, the monitoring density is adaptively adjusted, and resource allocation is optimized. And generate a risk level signal.

Benefits of technology

It improves monitoring efficiency and accuracy, can more accurately judge the frequency and intensity of geological activities, enhances the timeliness and accuracy of early warnings, optimizes risk management strategies, and reduces operating costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118966426A_ABST
    Figure CN118966426A_ABST
Patent Text Reader

Abstract

The invention provides a coal mine rock burst disaster early warning method and system based on feature drifting, and relates to the technical field of coal mine rock burst disaster early warning. Real-time data analysis and prediction are carried out by dynamically adjusting the number of sampling points and the length of an observation window and combining a time sequence algorithm model; and the limitation of a traditional monitoring method is effectively solved. Specifically, according to the scheme, the monitoring density can be adaptively adjusted according to the actual risk level, the resource allocation is optimized, and the monitoring efficiency and precision are improved. Meanwhile, by predicting and correcting stress data and temperature data and combining multi-dimensional risk assessment, the method can more accurately judge the frequency degree and the intensity degree of the geological activity, so that the timeliness and the accuracy of early warning are improved. The beneficial effects not only enhance the performance of the coal mine rock burst disaster early warning system, but also optimize the risk management strategy, reduce the operation cost and improve the economic benefit.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of coal mine rock burst disaster early warning, and in particular to a coal mine rock burst disaster early warning method and system based on feature drift. Background Art

[0002] Feature drift refers to the phenomenon that the statistical characteristics of data in the data stream (including the relationship between input features and output) gradually change over time. In the process of coal mining, rock burst disasters are a serious safety hazard that may cause significant casualties and property losses. Traditional monitoring methods usually use stress, vibration, temperature and other factors in the monitoring area as geological characteristics, and study their relationship with time for monitoring. This method relies on a fixed number of sampling points and a static monitoring window. This method has limitations when dealing with complex and changeable geological conditions and cannot effectively adapt to changes in risks in different regions, resulting in insufficient monitoring efficiency and accuracy. In addition, existing risk assessment methods often lack dynamic analysis and prediction of real-time data, making it difficult to accurately judge the frequency and intensity of geological activities, thereby affecting the timeliness and accuracy of early warnings.

[0003] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0004] The purpose of the present invention is to provide a coal mine rock burst disaster early warning method and system based on characteristic drift to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A coal mine rock burst disaster early warning method based on characteristic drift, the specific steps include:

[0007] S1: Based on user settings, at least two groups of monitoring areas are divided, and at least two groups of sampling points are set in each monitoring area for sampling to obtain stress data and temperature data corresponding to each monitoring area;

[0008] S2: preprocessing the collected stress data and temperature data, correcting the stress data according to the preprocessed temperature data, and then adjusting the number of sampling points in the corresponding monitoring area according to the corrected stress data;

[0009] S3: After the number of sampling points is adjusted, continue to collect stress data and temperature data in each monitoring area and preprocess them, then build a time series algorithm model, and use the preprocessed stress data and temperature data to train and optimize the time series algorithm model;

[0010] S4: After the time series algorithm model training is completed, the stress data and temperature data collected in real time in each monitoring area are preprocessed and input into the time series algorithm model for calculation to generate the predicted stress data and temperature data of each monitoring area, and the risk level of each monitoring area is judged based on the predicted stress data and temperature data.

[0011] Preferably, the preprocessing includes normalization processing and data cleaning, and the specific method is:

[0012] The stress data and temperature data are calibrated as and The superscript i represents the area number of the monitoring area, i = 1, 2, 3, ..., M, and the subscript j represents the sampling number of the sampling point, j = 1, 2, 3, ..., N;

[0013] The stress data and temperature data are respectively used as input data for normalization, and the calculation method is:

[0014]

[0015] Where: represents the input data after normalization. Respectively represent the maximum and minimum values ​​of the input data;

[0016] Then calculate the corresponding mean and standard deviation as follows:

[0017]

[0018] Where: σ i Respectively represent the mean and standard deviation calculated based on the normalized input data;

[0019] When the input data after normalization processing exists, there is J-th data that satisfies:

[0020]

[0021] The data is considered abnormal, and the corresponding input data is updated in the following way:

[0022]

[0023] Where: α represents the preset volatility coefficient, 3≤α≤6.

[0024] Preferably, the calculation method for correcting the stress data according to the preprocessed temperature data is:

[0025]

[0026] Where: They represent the mean values ​​of stress data before and after correction in a certain monitoring area after preprocessing, It represents the mean value of the pre-processed temperature data in a certain monitoring area, μ represents the temperature calibration coefficient, and the calculation method of the temperature calibration coefficient μ is:

[0027]

[0028] Where: T0 represents the preset reference temperature.

[0029] Preferably, the corrected stress data mean value of each monitoring area is and standard deviation Calculate the overall mean stress data and standard deviation of the mean The calculation method is:

[0030]

[0031]

[0032] Then adjust the number of sampling points N, and the calculation method is:

[0033]

[0034] Where: Express The internal data is rounded up, and N0 represents the initial number of sampling points.

[0035] Preferably, the method for training and optimizing the time series algorithm model using the preprocessed stress data and temperature data is:

[0036] The initial length of the observation window is set according to the sampling sequence, and then the length of the observation window is adjusted according to the number of sampling points in each monitoring area;

[0037] The length of the adjusted observation window is calibrated as L i , when the sampling points in the corresponding monitoring area collect L i When grouping data, i The group data is analyzed, the first 80% of the data in the observation window is divided into a training set, and the last 20% of the data is divided into a validation set. The training set is used to train the time series algorithm model, and the validation set is used to validate the time series algorithm model.

[0038] Preferably, the length L of the adjusted observation window is i The calculation method is:

[0039]

[0040] Where: L0 represents the initial length of the observation window.

[0041] Preferably, generate the pre-processed stress data predicted for each monitoring area and temperature data The predicted temperature data is then used to correct the predicted stress data to obtain the mean value of the corrected predicted stress data for each monitoring area after preprocessing. Prediction standard deviation σ c i , and the overall pre-processed corrected predicted stress data mean Prediction standard deviation mean

[0042] When the monitoring area meets the following conditions at the same time:

[0043]

[0044] The risk level of the monitoring area is considered high;

[0045] When the monitoring area meets:

[0046]

[0047] or

[0048]

[0049] The risk level of this monitoring area is considered to be medium;

[0050] When the monitoring area meets the following conditions at the same time:

[0051]

[0052] The risk level of the monitoring area is considered to be low;

[0053] Where: δ represents the preset monitoring threshold, 0.3≤δ≤1.

[0054] A coal mine rock burst disaster early warning system based on characteristic drift, the early warning system adopts the above-mentioned early warning method, including:

[0055] A data acquisition module, which is electrically connected to the data analysis module and is used to collect stress data and temperature data in different monitoring areas;

[0056] A data analysis module, which is electrically connected to the data prediction module and is used to preprocess the stress data and the temperature data, and adjust the number of acquisition points according to the preprocessed stress data and the temperature data;

[0057] A data prediction module, which is electrically connected to the risk judgment module and is used to construct a time series algorithm model to generate predicted stress data and temperature data;

[0058] A risk judgment module, which is electrically connected to the alarm module and is used to generate different risk level signals according to the predicted stress data and temperature data;

[0059] An alarm module is used to send out different types of alarms according to different risk level signals.

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

[0061] The present invention effectively solves the limitations of traditional monitoring methods by dynamically adjusting the number of sampling points and the length of the observation window, and combining the time series algorithm model for real-time data analysis and prediction. Specifically, this solution can adaptively adjust the monitoring density according to the actual risk level, optimize resource allocation, and improve monitoring efficiency and accuracy. At the same time, by predicting and correcting stress data and temperature data, combined with multi-dimensional risk assessment, this solution can more accurately judge the frequency and intensity of geological activities, thereby improving the timeliness and accuracy of early warning. These beneficial effects not only enhance the performance of the coal mine rock burst disaster early warning system, but also optimize the risk management strategy, reduce operating costs, and improve economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a schematic diagram of the overall method flow of the present invention;

[0063] Figure 2 It is a schematic diagram of the module structure of the present invention. DETAILED DESCRIPTION

[0064] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0065] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0066] Example:

[0067] See also Figure 1-2 , the present invention provides a technical solution:

[0068] A coal mine rock burst disaster early warning method based on characteristic drift, the specific steps include:

[0069] S1: Based on user settings, at least two groups of monitoring areas are divided, and at least two groups of sampling points are set in each monitoring area for sampling to obtain stress data and temperature data corresponding to each monitoring area.

[0070] S2: pre-processing the collected stress data and temperature data, correcting the stress data according to the pre-processed temperature data, and then adjusting the number of sampling points in the corresponding monitoring area according to the corrected stress data.

[0071] The preprocessing includes normalization and data cleaning, and the specific method is as follows:

[0072] The stress data and temperature data are calibrated as and The superscript i represents the area number of the monitoring area, i = 1, 2, 3, ..., M, and the subscript j represents the sampling number of the sampling point, j = 1, 2, 3, ..., N;

[0073] The stress data and temperature data are respectively used as input data for normalization, and the calculation method is:

[0074]

[0075] Where: represents the input data after normalization. Respectively represent the maximum and minimum values ​​of the input data;

[0076] Then calculate the corresponding mean and standard deviation as follows:

[0077]

[0078] Where: σ i Respectively represent the mean and standard deviation calculated based on the normalized input data;

[0079] When the input data after normalization processing exists, there is J-th data that satisfies:

[0080]

[0081] The data is considered abnormal, and the corresponding input data is updated in the following way:

[0082]

[0083] Where: α represents the preset fluctuation coefficient. In Gaussian distribution, the fluctuation coefficient is generally defined as 3. However, in order to improve the system's adaptability to different environments, the fluctuation coefficient is allowed to be adjusted manually and the value is between [3,6].

[0084] In this step, all data are scaled to the same scale through normalization, which helps to eliminate the impact between data of different magnitudes and makes data analysis more accurate and effective. At the same time, by identifying and correcting abnormal data, the noise and errors in the data can be reduced, and the overall quality of the data can be improved, so that the system can more accurately identify and predict coal mine rock burst disasters, thereby improving the timeliness and accuracy of early warning.

[0085] The calculation method for correcting the stress data based on the preprocessed temperature data is:

[0086]

[0087] Where: They represent the mean values ​​of stress data before and after correction in a certain monitoring area after preprocessing, It represents the mean value of the pre-processed temperature data in a certain monitoring area, μ represents the temperature calibration coefficient, and the calculation method of the temperature calibration coefficient μ is:

[0088]

[0089] Where: T0 represents the preset reference temperature. Since the stress sensor is usually calculated by detecting the strain of the strain gauge and its own elastic modulus, the strain of the strain gauge is detected by detecting the resistance of the strain gauge, and the temperature will have a certain impact on the resistance of the strain gauge, so temperature compensation is required to improve the detection accuracy. The stress sensor used in the coal mine usually needs to use a metal strain gauge with stronger resistance to external interference. The resistance of the metal foil on its surface will increase with a small curve as the temperature increases. Therefore, the temperature calibration coefficient is used here for fitting to obtain more accurate stress data.

[0090] In this step, by considering the impact of temperature on stress data, the corrected stress data can more accurately reflect the actual stress state and reduce the error caused by temperature changes. Using the dynamically calculated temperature calibration coefficient, the correction strength can be adjusted according to the actual temperature changes, making the correction process more flexible and adaptable to different environmental conditions, which can reduce false alarms caused by temperature changes, reduce the false alarm rate of the system, and improve the credibility of the early warning.

[0091] According to the corrected stress data mean of each monitoring area and standard deviation Calculate the overall mean stress data and standard deviation of the mean The calculation method is:

[0092]

[0093] Then adjust the number of sampling points N, and the calculation method is:

[0094]

[0095] Where: Express The internal data is rounded up, and N0 represents the initial number of sampling points.

[0096] According to the above formula, the corrected mean stress data of each monitoring area is The higher the value, the more intense the geological activity in the area. The standard deviation σ in each monitoring area o The higher the value, the more frequent the geological activities in the area, so more sampling points need to be added for more comprehensive monitoring. In other words, the number of sampling points is related to the corrected mean stress data of each monitoring area. and standard deviation σ o Directly proportional relationship.

[0097] In this step, by dynamically adjusting the number of sampling points, the early warning system can more accurately monitor and predict coal mine rock burst disasters, thereby improving the accuracy and timeliness of the early warning. It can also adapt to monitoring needs under different risk conditions, enhance the system's environmental adaptability, and at the same time optimize resource allocation, improve monitoring efficiency, reduce the system's operating costs, and improve economic benefits.

[0098] S3: After the number of sampling points is adjusted, continue to collect stress data and temperature data in each monitoring area and preprocess them, then build a time series algorithm model, and use the preprocessed stress data and temperature data to train and optimize the time series algorithm model.

[0099] The method for training and optimizing the time series algorithm model using preprocessed stress data and temperature data is as follows:

[0100] The initial length of the observation window is set according to the sampling sequence, and then the length of the observation window is adjusted according to the number of sampling points in each monitoring area;

[0101] The length of the adjusted observation window is calibrated as L o , when the sampling points in the corresponding monitoring area collect L i When grouping data, i The group data is analyzed, the first 80% of the data in the observation window is divided into a training set, and the last 20% of the data is divided into a validation set. The training set is used to train the time series algorithm model, and the validation set is used to validate the time series algorithm model.

[0102] The length of the adjusted observation window is L i The calculation method is:

[0103]

[0104] Where: L0 represents the initial length of the observation window.

[0105] It can be seen from the above formula that when the sampling points in the monitoring area increase, the length of the observation window will also increase accordingly to obtain more data for analysis and improve the accuracy of the system. Conversely, when the sampling points decrease, the length of the observation window will be reduced to improve the response speed of the system.

[0106] In this step, the length of the observation window is dynamically adjusted according to the number of sampling points in the monitoring area, so that the model training is more adaptable to the changes in the actual monitoring data, the adaptability and flexibility of the model are improved, the system can adapt to the changes in different monitoring data, and the environmental adaptability of the system is enhanced. At the same time, by automatically calculating and adjusting the length of the observation window, the operation process is simplified, human intervention is reduced, and the automation level of the system is improved.

[0107] S4: After the time series algorithm model training is completed, the stress data and temperature data collected in real time in each monitoring area are preprocessed and input into the time series algorithm model for calculation to generate the predicted stress data and temperature data of each monitoring area, and the risk level of each monitoring area is judged based on the predicted stress data and temperature data.

[0108] Generates pre-processed stress data predicted for each monitoring area and temperature data The predicted temperature data is then used to correct the predicted stress data to obtain the mean value of the corrected predicted stress data for each monitoring area after preprocessing. Prediction standard deviation σ c i , and the overall pre-processed corrected predicted stress data mean Prediction standard deviation mean

[0109] When the monitoring area meets the following conditions at the same time:

[0110]

[0111] It is believed that the geological activities in the monitoring area are both frequent and intense, and the risk level is high;

[0112] When the monitoring area meets:

[0113]

[0114] or

[0115]

[0116] It is considered that the geological activities in the monitoring area are frequent or intense, and the risk level is medium;

[0117] When the monitoring area meets the following conditions at the same time:

[0118]

[0119] It is believed that the geological activities in the monitoring area are relatively small and the risk level is low;

[0120] Where: δ represents the preset monitoring threshold, 0.3≤δ≤1. Generally speaking, if the mean and standard deviation of the predicted stress data in a certain monitoring area exceed 30% of the overall corresponding data, it can be considered to exceed the threshold. The actual threshold value can be slightly adjusted according to user needs or environmental factors.

[0121] A coal mine rock burst disaster early warning system based on characteristic drift, the early warning system adopts the above-mentioned early warning method, including:

[0122] A data acquisition module, which is electrically connected to the data analysis module and is used to collect stress data and temperature data in different monitoring areas;

[0123] A data analysis module, which is electrically connected to the data prediction module and is used to preprocess the stress data and the temperature data, and adjust the number of acquisition points according to the preprocessed stress data and the temperature data;

[0124] A data prediction module, which is electrically connected to the risk judgment module and is used to construct a time series algorithm model to generate predicted stress data and temperature data;

[0125] A risk judgment module, which is electrically connected to the alarm module and is used to generate different risk level signals according to the predicted stress data and temperature data;

[0126] An alarm module is used to send out different types of alarms according to different risk level signals.

[0127] In summary, the present invention effectively solves the limitations of traditional monitoring methods by dynamically adjusting the number of sampling points and the length of the observation window, and combining the time series algorithm model for real-time data analysis and prediction. Specifically, this solution can adaptively adjust the monitoring density according to the actual risk level, optimize resource allocation, and improve monitoring efficiency and accuracy. At the same time, by predicting and correcting stress data and temperature data, combined with multi-dimensional risk assessment, this solution can more accurately judge the frequency and intensity of geological activities, thereby improving the timeliness and accuracy of early warning. These beneficial effects not only enhance the performance of the coal mine rock burst disaster early warning system, but also optimize the risk management strategy, reduce operating costs, and improve economic benefits.

[0128] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0129] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0130] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0131] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A coal mine rock burst disaster early warning method based on feature drift, characterized in that: The specific steps include: S1: Based on user settings, at least two groups of monitoring areas are divided, and at least two groups of sampling points are set in each monitoring area for sampling to obtain stress data and temperature data corresponding to each monitoring area; S2: preprocessing the collected stress data and temperature data, correcting the stress data according to the preprocessed temperature data, and then adjusting the number of sampling points in the corresponding monitoring area according to the corrected stress data; S3: After the number of sampling points is adjusted, continue to collect stress data and temperature data in each monitoring area and preprocess them, then build a time series algorithm model, and use the preprocessed stress data and temperature data to train and optimize the time series algorithm model; S4: After the time series algorithm model training is completed, the stress data and temperature data collected in real time in each monitoring area are preprocessed and input into the time series algorithm model for calculation to generate the predicted stress data and temperature data of each monitoring area, and the risk level of each monitoring area is judged based on the predicted stress data and temperature data.

2. The coal mine rock burst disaster early warning method based on feature drift according to claim 1, characterized in that: The preprocessing includes normalization and data cleaning, and the specific method is as follows: The stress data and temperature data are calibrated as and The superscript i represents the area number of the monitoring area, i = 1, 2, 3, ..., M, and the subscript j represents the sampling number of the sampling point, j = 1, 2, 3, ..., N; The stress data and temperature data are respectively used as input data for normalization, and the calculation method is: Where: represents the normalized input data, Respectively represent the maximum and minimum values ​​of the input data; Then calculate the corresponding mean and standard deviation as follows: Where: σ i They represent the mean and standard deviation calculated based on the normalized input data; When the input data after normalization processing exists, there is J-th data that satisfies: The data is considered abnormal, and the corresponding input data is updated in the following way: Where: α represents the preset volatility coefficient, 3≤α≤6.

3. The coal mine rock burst disaster early warning method based on feature drift according to claim 2, characterized in that: The calculation method for correcting the stress data based on the preprocessed temperature data is: Where: They represent the mean values ​​of stress data before and after correction after preprocessing in a monitoring area, It represents the mean value of the pre-processed temperature data in a certain monitoring area, μ represents the temperature calibration coefficient, and the calculation method of the temperature calibration coefficient μ is: Where: T0 represents the preset reference temperature.

4. The coal mine rock burst disaster early warning method based on feature drift according to claim 3 is characterized by: According to the corrected stress data mean of each monitoring area and standard deviation Calculate the overall mean stress data and standard deviation of the mean The calculation method is: Then adjust the number of sampling points N, and the calculation method is: Where: Express The internal data is rounded up, and N0 represents the initial number of sampling points.

5. The coal mine rock burst disaster early warning method based on feature drift according to claim 4, characterized in that: The method for training and optimizing the time series algorithm model using preprocessed stress data and temperature data is as follows: The initial length of the observation window is set according to the sampling sequence, and then the length of the observation window is adjusted according to the number of sampling points in each monitoring area; The length of the adjusted observation window is calibrated as L i , when the sampling points in the corresponding monitoring area collect L i When grouping data, i The group data is analyzed, the first 80% of the data in the observation window is divided into a training set, and the last 20% of the data is divided into a validation set. The training set is used to train the time series algorithm model, and the validation set is used to validate the time series algorithm model.

6. The method for early warning of coal mine rock burst disaster based on feature drift according to claim 5, characterized in that: The length of the adjusted observation window is L i The calculation method is: Where: L0 represents the initial length of the observation window.

7. The method for early warning of coal mine rock burst disaster based on characteristic drift according to claim 6, characterized in that: Generates pre-processed stress data predicted for each monitoring area and temperature data The predicted temperature data is then used to correct the predicted stress data to obtain the mean value of the corrected predicted stress data for each monitoring area after preprocessing. Prediction standard deviation σ c i , and the overall pre-processed corrected predicted stress data mean Prediction standard deviation mean When the monitoring area meets the following conditions at the same time: The risk level of the monitoring area is considered high; When the monitoring area meets: or The risk level of this monitoring area is considered to be medium; When the monitoring area meets the following conditions at the same time: The risk level of the monitoring area is considered to be low; Where: δ represents the preset monitoring threshold, 0.3≤δ≤1.

8. A coal mine rock burst disaster early warning system based on characteristic drift, characterized in that: The early warning system adopts the early warning method according to any one of claims 1 to 7, including: A data acquisition module, which is electrically connected to the data analysis module and is used to collect stress data and temperature data in different monitoring areas; A data analysis module, which is electrically connected to the data prediction module and is used to preprocess the stress data and the temperature data, and adjust the number of acquisition points according to the preprocessed stress data and the temperature data; A data prediction module, which is electrically connected to the risk judgment module and is used to construct a time series algorithm model to generate predicted stress data and temperature data; A risk judgment module, which is electrically connected to the alarm module and is used to generate different risk level signals according to the predicted stress data and temperature data; An alarm module is used to send out different types of alarms according to different risk level signals.