A Mine Safety Hazard Early Warning Method and System Based on Big Data

Abstract

The present application provides a method and system for mine safety hazard early warning based on big data, relating to the technical field of landslide early warning. The method includes: constructing a spatial coordinate system and collecting the position data of a monitoring target, and collecting the depth information sequence before slope instability. Subsequently, a prediction model is trained through the depth information sequence. In the data processing stage, the depth information of the monitoring target is calculated, and this data is input into the trained model to predict the slope instability probability. According to the comparison result between the predicted probability and a preset threshold, it is decided whether to continue data collection or send out an early warning signal, thereby improving the accuracy of mine safety hazard early warning.

Description

Technical Field

[0001] This application relates to the technical field of landslide early warning, and in particular to a method and system for early warning of mine safety hazards based on big data. Background Art

[0002] During the process of mine exploitation, the stability of the slope is one of the key factors for safe production. However, due to complex geological conditions, frequent mining activities, and the influence of natural factors, mine slopes are often at risk of instability. Once the slope becomes unstable, it may lead to serious safety accidents such as landslides and collapses, posing a huge threat to the safety of people's lives and property.

[0003] Traditional methods for monitoring the stability of mine slopes mainly rely on manual inspections and simple instrument measurements. These methods have many limitations, such as low monitoring frequency, limited coverage, and insufficient data accuracy, which will reduce the timeliness and accuracy of early warning of mine safety hazards. Summary of the Invention

[0004] In order to improve the accuracy of early warning of mine safety hazards, this application provides a method and system for early warning of mine safety hazards based on big data.

[0005] In the first aspect, this application provides a method for early warning of mine safety hazards based on big data, adopting the following technical solution:

[0006] A method for early warning of mine safety hazards based on big data includes the following steps:

[0007] First collection: Construct a spatial coordinate system, collect the position data of the monitoring target in the spatial coordinate system, denoted as the first data; collect the depth information sequence corresponding to different slope instabilities before.

[0008] First modeling and training: Establish a prediction model, and use the depth information sequence to train the prediction model to obtain the trained prediction model.

[0009] Data processing: Calculate the depth information of the monitoring target based on the first data, denoted as the second data.

[0010] Prediction: Input the second data into the trained prediction model to obtain the probability of slope instability.

[0011] First judgment: Judge whether the probability is less than the preset threshold. If so, execute the steps of the first collection; if not, execute the steps of early warning.

[0012] Early warning: Send out an early warning signal.

[0013] This application first constructs a spatial coordinate system and collects the position data of the monitoring target. By constructing the spatial coordinate system, the three-dimensional position information of the monitoring target can be accurately obtained, which helps with subsequent data processing and prediction. This application also collects the depth information sequences corresponding to different slope instabilities. The depth information sequences can reflect the characteristic changes before slope instability and provide valuable training data for the prediction model. Then, a prediction model is established, and the depth information sequences are used as training data to train the prediction model, which helps the prediction model learn the characteristic change rules during slope instability. Then, based on the position data, the depth information of the monitoring target is calculated, and the depth information is input into the trained prediction model to obtain the probability of slope instability. The prediction model can quickly give the probability of slope instability according to the input depth information, facilitating timely measures. This application also determines whether the predicted probability is less than a preset threshold. If so, continuous monitoring is carried out; if not, a warning signal is issued. By using the depth information to train the prediction model, this application enables the prediction model to use the currently collected position data to predict the probability of future slope instability, and thus can issue a warning in a timely manner to reduce the potential hazards and losses caused by slope instability.

[0014] Optionally, after performing the step of data processing and before performing the step of prediction, it further includes:

[0015] Second collection: Obtain the n depth information of the monitoring target before the current moment, denoted as the first depth information, where n is equal to the number of samples in the depth information sequence minus one;

[0016] Information integration: Integrate the n first depth information with the second data into a second data sequence;

[0017] First calculation: Standardize both the k-th depth information sequence and the second data sequence, and calculate the Euclidean distance between the standardized k-th depth information sequence and the standardized second data sequence;

[0018] Statistics: Count the number of Euclidean distances less than the preset Euclidean distance threshold, denoted as the third data;

[0019] Second judgment: Determine whether the third data is greater than the preset quantity threshold. If so, perform the step of prediction; if not, perform the iterative step;

[0020] Iteration: Take the (k + 1)-th depth information sequence as the new k-th depth information sequence, and perform the step of the first calculation until the preset stop condition is met.

[0021] This application obtains n depth information of the monitoring target before the current moment, integrates the n first depth information and the second data into a second data sequence, and then standardizes both the k-th depth information sequence and the second data sequence. Calculate the Euclidean distance between the standardized k-th depth information sequence and the standardized second data sequence. By calculating the Euclidean distance, the similarity between different depth information sequences and the second data sequence can be quantified. Subsequently, count the number of Euclidean distances less than the preset Euclidean distance threshold to quantify the similarity degree between the second data sequence and the k-th depth information sequence. Then, count the number of depth information sequences with a high similarity to the second data sequence to obtain the third data. Subsequently, determine whether the third data is greater than the preset quantity threshold to achieve automatic judgment of the similarity of the data sequence, thereby deciding whether to execute the prediction step, thus improving the accuracy of the prediction. This application can determine whether the second data is suitable for prediction by calculating the Euclidean distance and counting the number of similar samples, and will only perform prediction when the data similarity is high, so as to improve the accuracy of the prediction.

[0022] Optionally, after the step of performing the second judgment and before the step of performing the prediction, it further includes:

[0023] Construct a matrix: Construct a distance matrix, and the calculation model of the element in the i-th row and j-th column of the distance matrix is as follows:

[0024] ;

[0025] Wherein, is the element in the i-th row and j-th column of the distance matrix; is the Euclidean distance between the i-th sample in the k-th depth information sequence and the j-th sample in the second data sequence;

[0026] First acquisition: Use the dynamic path planning algorithm to find the shortest path through the distance matrix, and calculate the sum of all elements included in the shortest path, denoted as the fourth data;

[0027] Third judgment: Determine whether the fourth data is greater than the preset threshold. If so, execute the iterative step; if not, execute the prediction step.

[0028] This application constructs a distance matrix, which can intuitively display the cumulative distances between samples in different depth information sequences, providing a basis for subsequent shortest path search. Subsequently, this application uses a dynamic path planning algorithm to search for the shortest path and obtain the shortest path through the distance matrix to evaluate the overall similarity and correlation between the second data sequence and the k-th depth information sequence. Subsequently, the sum of all elements included in the shortest path is calculated to time-align the second data sequence with the k-th depth information sequence. Subsequently, it is determined whether the fourth data is greater than a preset threshold to automatically determine the overall similarity between the second data sequence and the k-th depth information sequence, improving the accuracy of prediction.

[0029] Optionally, after the step of performing the third determination and before the step of performing the prediction, it further includes:

[0030] Second acquisition: Obtain the timestamp of the first data, delete the year information in the timestamp, and denote the remaining part in the timestamp as the m moment;

[0031] Third acquisition: Respectively obtain the historical position data of the monitoring target at the m moment of each year, and the offset corresponding to each historical position data;

[0032] Second calculation: Calculate the average value of the historical position data, denoted as the fifth data; calculate the average value of the offsets, denoted as the sixth data;

[0033] Third calculation: Calculate the difference between the fifth data and the sixth data, denoted as the first difference;

[0034] First adjustment: Input the first difference and the first data into the FOPID controller to obtain the adjustment amount of the first data;

[0035] Fourth calculation: Perform a subtraction operation on the first data and the adjustment amount to obtain an operation result, and use the operation result as the new first data.

[0036] This application obtains the timestamp of the first data, deletes the year information, and only retains the month, date, and time parts (and this part is denoted as m moment), so that the subsequent obtained historical location data can be aligned with the current data in the time dimension and is not affected by the year difference, thereby enabling the analysis of historical behavior patterns at the same time point. Subsequently, the historical location data and its corresponding offset at the m moment of each year for the monitoring target are obtained to reveal the location change law of the target at this time point, as well as possible anomalies or trends. Subsequently, the average value of the historical location data (i.e., the fifth data) and the average value of the offset (i.e., the sixth data) are calculated. By calculating the average value, the abnormal fluctuations of individual data can be smoothed out to obtain a more representative historical location and offset benchmark. Subsequently, the difference between the fifth data and the sixth data (i.e., the first difference) is calculated, and the deviation of the fifth data can be corrected through the sixth data. Subsequently, the first difference and the first data are input into the FOPID controller to obtain the adjustment amount of the first data. The FOPID controller (fractional order proportional-integral-derivative controller) can calculate an appropriate adjustment amount according to the input data to correct the deviation or error in the current data and improve the accuracy of the data. Subsequently, a subtraction operation is performed on the first data and the adjustment amount to obtain a new first data. By adopting the above solution, this application realizes the correction of the original first data, makes the new first data closer to the expected position of the target at the m moment, provides a more accurate data basis for the subsequent prediction step, and helps to improve the accuracy of the prediction, especially when dealing with data with periodic or seasonal changes.

[0037] Optionally, after the step of performing the first adjustment and before the step of performing the fourth calculation, it further includes:

[0038] Second modeling and training: Establish an adjustment model, and use the fifth data and the sixth data to train the adjustment model to obtain a trained adjustment model;

[0039] Second prediction: Input the first data into the trained adjustment model to obtain a predicted adjustment amount;

[0040] Fifth calculation: Calculate the difference between the adjustment amount and the predicted adjustment amount, and denote it as the second difference;

[0041] Fourth judgment: Judge whether the second difference is less than the preset difference threshold. If so, perform the fourth calculation step; if not, perform the feedback step;

[0042] Feedback: Send a feedback signal to the operation and maintenance personnel.

[0043] This application establishes an adjustment model and trains the model using the fifth data (the average of historical position data) and the sixth data (the average of offsets). Through training, the model can learn the relationship between historical positions and offsets. Subsequently, the current first data is input into the trained adjustment model to obtain a predicted adjustment amount. Subsequently, the difference between the actual adjustment amount (the adjustment amount output from the FOPID controller) and the predicted adjustment amount is calculated and denoted as the second difference. By comparing the actual and predicted adjustment amounts, the prediction accuracy of the model can be evaluated. Subsequently, it is determined whether the second difference is less than a preset difference threshold. If so, the steps of the fourth calculation are continued; if not, a feedback signal is sent to the operation and maintenance personnel indicating that the difference between the predicted adjustment amount and the actual adjustment amount exceeds the acceptable range, enabling the operation and maintenance personnel to promptly understand the prediction performance problem of the FOPID controller and take appropriate measures for correction, thereby maintaining the accuracy and reliability of the entire data processing process.

[0044] Optionally, after performing the steps of the fourth determination and before performing the feedback step, it further includes:

[0045] Construct a model: Construct a reinforcement learning model;

[0046] Set the state: Set the state space of the reinforcement learning model, where the state space includes: the difference between the adjustment amount output from the FOPID controller and the predicted adjustment amount output from the adjustment model; set the action space of the reinforcement learning model, where the action space includes: the parameter adjustment strategy of the FOPID controller;

[0047] Set the agent: Set the agent, which is used to select the parameter adjustment strategy of the FOPID controller according to the current state of the FOPID controller;

[0048] First adjustment: Input the second difference into the reinforcement learning model to obtain the parameter adjustment strategy corresponding to the second difference, and adjust the current parameters of the FOPID controller according to the parameter adjustment strategy corresponding to the second difference;

[0049] Model update: Use the adjusted FOPID controller as the new controller and perform the steps of the first adjustment to obtain a new adjustment amount;

[0050] Fifth determination: Determine whether the new adjustment amount meets the expectation. If so, perform the steps of the fourth calculation; if not, perform the feedback step.

[0051] In this application, by constructing a reinforcement learning model and introducing the basis of reinforcement learning, a necessary algorithm framework is provided for subsequent steps. Subsequently, the state space of the reinforcement learning model is defined, including the difference between the adjustment amount output by the FOPID controller and the predicted adjustment amount output by the adjustment model, and the action space of the reinforcement learning model is set, including the parameter adjustment strategy of the FOPID controller. By clarifying the state space and the action space, the agent can clearly know how to select actions according to the current state. The selection of the state space enables the model to pay attention to the difference between the adjustment amount and the predicted adjustment amount, while the action space allows the model to optimize this difference by adjusting the parameters of the FOPID controller. Subsequently, an agent is set to select a parameter adjustment strategy according to the current state of the FOPID controller, that is, to select the optimal action according to the current state. Subsequently, the second difference is input into the reinforcement learning model to obtain the corresponding parameter adjustment strategy, and the current parameters of the FOPID controller are adjusted according to this strategy. Subsequently, the adjusted FOPID controller is used as a new controller, and the first adjustment step is executed to obtain a new adjustment amount. Subsequently, it is judged whether the new adjustment amount meets the expectation. If the new adjustment amount meets the expectation, subsequent steps can be continued; if not, a feedback step is executed, and the parameters of the reinforcement learning model or the FOPID controller need to be further adjusted.

[0052] Optionally, after performing the first adjustment step and before performing the model update step, it further includes:

[0053] Fourth acquisition: Acquire training data, where the training data includes the difference between the adjustment amount output by the FOPID controller and the predicted adjustment amount output by the adjustment model, and the parameters of the FOPID controller;

[0054] Third modeling and training: Establish an MLP model and train the MLP model with the training data to obtain a trained MLP model;

[0055] Third prediction: Input the second difference into the trained MLP model to obtain the predicted parameters of the FOPID controller, denoted as the first parameters;

[0056] Fifth acquisition: Acquire the parameters of the FOPID controller after adjustment in the first adjustment step, denoted as the second parameters;

[0057] Sixth judgment: Judge whether the difference between the first parameters and the second parameters is less than the preset parameter difference threshold. If so, execute the model update step; if not, execute the feedback step.

[0058] This application obtains training data. Subsequently, an MLP model is established, and the MLP model is trained using the training data to obtain the trained MLP model. By training the MLP model, the MLP model can learn the relationship between the difference between the adjustment amount and the predicted adjustment amount and the FOPID controller parameters, so that the MLP model can predict the optimal FOPID controller parameters according to the new difference in the future. Subsequently, the second difference (i.e., the difference between the current adjustment amount and the predicted adjustment amount) is input into the trained MLP model to obtain the predicted parameters of the FOPID controller, denoted as the first parameter. By adopting the above solution, this application realizes using the MLP model to predict the optimal FOPID controller parameters in the current situation. Subsequently, the parameters of the FOPID controller after adjustment in the first adjustment step are obtained, denoted as the second parameter, and it is judged whether the difference between the first parameter (the predicted parameter of the MLP model) and the second parameter (the actually adjusted parameter) is less than the preset parameter difference threshold. If the predicted parameter is close enough to the actually adjusted parameter (i.e., the difference between the two parameters is less than the preset threshold), it can be considered that the performance of the FOPID controller meets the expectations, and the model update step can be continued; if the difference is too large, it indicates that the performance of the FOPID controller does not meet the expectations, and the feedback step needs to be executed to further adjust the model or parameters, so as to improve the stability and reliability of the FOPID controller.

[0059] Optionally, after performing the fourth acquisition step and before performing the third modeling and training step, it further includes:

[0060] The seventh judgment: Judge whether the number of samples in the training data meets the preset number threshold. If so, perform the third modeling and training step; if not, perform the fourth modeling step;

[0061] The fourth modeling: Establish a generative adversarial network;

[0062] Generate data: Use the generative adversarial network to generate new training data;

[0063] Data update: Use the new training data and the training data in the fourth acquisition step as the new training data.

[0064] By adopting the above solution, this application can utilize the generation ability of the generative adversarial network to generate new training samples. These new samples will be used together with the existing training data to train the MLP model, thereby enhancing the generalization ability and prediction accuracy of the model. Subsequently, the new training data and the training data in the fourth acquisition step will be used as the new training data to form a larger and more comprehensive training set. This updated training set will be used for subsequent MLP model training, which helps to improve the performance and stability of the MLP model. By increasing the amount of training data, the problem of insufficient model training can also be reduced, improving the stability and reliability of the entire control system.

[0065] Optionally, after performing the fourth modeling step and before performing the data generation step, it further includes:

[0066] Fourth training: Use the training data in the fourth acquisition step to train the generative adversarial network to obtain the trained generative adversarial network;

[0067] In the data generation step, use the trained generative adversarial network to generate new training data.

[0068] This application uses the training data in the fourth acquisition step to train the generative adversarial network. Through training, the generator gradually learns how to generate samples similar to the real data, while the discriminator learns how to more accurately identify real data and generated data. This process is an iterative process until the generator can generate new samples close enough to the real data and the discriminator has difficulty distinguishing them. Subsequently, in the data generation step (using the trained generative adversarial network), that is, use the trained generative adversarial network to generate new training data. After the GAN is fully trained, the generator has learned how to generate high-quality new samples similar to the real data. These new samples can be used as additional training data to enhance the training effect of other machine learning models.

[0069] In the second aspect, this application provides a mine safety hazard warning system based on big data, adopting the following technical solution:

[0070] A mine safety hazard warning system based on big data, including:

[0071] A processor and a memory,

[0072] Program code is stored in the memory;

[0073] When the processor calls the program code in the memory, it executes the steps of the method.

[0074] In summary, this application includes at least one of the following beneficial technical effects:

[0075] 1. The present application first constructs a spatial coordinate system and collects the position data of the monitoring target. By constructing the spatial coordinate system, the three-dimensional position information of the monitoring target can be accurately obtained, which is helpful for subsequent data processing and prediction. The present application also collects the depth information sequences corresponding to different slope instabilities. The depth information sequences can reflect the characteristic changes before slope instability and provide valuable training data for the prediction model.

[0076] 2. The present application establishes a prediction model and trains the prediction model with the depth information sequences as training data, which helps the prediction model learn the characteristic change rules during slope instability. Then, based on the position data, the depth information of the monitoring target is calculated and input into the trained prediction model to obtain the probability of slope instability. The prediction model can quickly give the probability of slope instability according to the input depth information, facilitating timely measures to be taken.

[0077] 3. The present application also determines whether the predicted probability is less than a preset threshold. If so, continuous monitoring is carried out; if not, a warning signal is issued. By training the prediction model using the depth information, the prediction model can use the currently collected position data to predict the probability of future slope instability, and thus can issue a warning in a timely manner to reduce the potential hazards and losses caused by slope instability. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 is the flowchart of Embodiment 1 of the present application;

[0079] Figure 2 is the flowchart of Embodiment 2 of the present application;

[0080] Figure 3 is the flowchart of Embodiment 3 of the present application;

[0081] Figure 4 is the flowchart from S81 Second Modeling and Training to S86 Feedback of Embodiment 4 of the present application;

[0082] Figure 5 is the flowchart from S8541 Fourth Acquisition to S8545 Sixth Judgment of Embodiment 4 of the present application;

[0083] Figure 6 is the flowchart of Embodiment 5 of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0084] The following is a further detailed description of the present application in conjunction with Figures 1 to 6 to the present application.

[0085] Embodiment 1: This embodiment discloses a method for warning about potential safety hazards in mines based on big data. Referring to Figure 1, the method includes: S1 First Collection, S2 First Modeling and Training, S3 Data Processing, S4 Prediction, S5 First Judgment, and S6 Early Warning. First, a spatial coordinate system is constructed and the position data of the monitoring target is collected, and a depth information sequence before slope instability is collected. Subsequently, a prediction model is trained through the depth information sequence. In the data processing stage, the depth information of the monitoring target is calculated, and this data is input into the trained model to predict the slope instability probability. According to the comparison result between the predicted probability and the preset threshold, it is decided whether to continue data collection or send an early warning signal. This embodiment includes the following steps:

[0086] S1 First Collection, using a camera to periodically or real-time photograph the monitoring target (i.e., the slope) to obtain image data.

[0087] Taking the camera position as the coordinate origin, a three-dimensional spatial coordinate system is established, and the position data of the monitoring target (such as specific points or areas on the slope) is recorded. These position data are marked as the first data.

[0088] For example, a GNSS receiver is deployed at the camera installation position to receive and record the position data of the camera. This GNSS receiver continuously receives signals from multiple satellite systems such as GPS, Beidou, and GLONASS, and calculates the longitude, latitude, and altitude information of the camera through built-in algorithms. These information will be used as the origin data of the three-dimensional spatial coordinate system.

[0089] At the same time, additional GNSS receivers need to be deployed at specific points or key areas on the slope. These GNSS receivers will also receive signals from satellites and calculate their respective three-dimensional position data. These three-dimensional position data will be used as the position data of the monitoring target, recorded in the database, and associated with the position data of the camera (i.e., the origin data) to obtain the coordinate values of the monitoring target in the three-dimensional spatial coordinate system.

[0090] Calculate the difference between the longitude of the monitoring target and the longitude of the camera, calculate the difference between the latitude of the monitoring target and the latitude of the camera, calculate the difference between the altitude of the monitoring target and the altitude of the camera, and use the three differences as the coordinate values of the corresponding coordinate axes.

[0091] Through the above method, a three-dimensional spatial coordinate system with the camera position as the origin can be established, and specific points or areas on the slope can be accurately located in this coordinate system. These position data (the first data) will be used for subsequent slope stability analysis. By comparing the changes in position data at different time points, the morphological changes of the slope can be detected in a timely manner, so as to predict the stability state of the slope.

[0092] For multiple known slope instability cases, collect the depth information sequences before their instability. These depth information reflect the morphological changes of the slope before instability.

[0093] S2 First modeling and training, construct a prediction model capable of processing depth information sequences, such as deep learning models (such as convolutional neural network CNN or recurrent neural network RNN) or traditional machine learning algorithms.

[0094] Use the collected depth information sequences before different slope instabilities as training data to train the prediction model, enabling the prediction model to learn to identify the patterns or features of slope instability from the depth information sequences.

[0095] S3 Data processing, based on the first data (i.e., the position data of the monitoring target in the spatial coordinate system), calculate the depth information of the monitoring target in the image data, which reflects the three-dimensional shape of the slope surface and is denoted as the second data.

[0096] S4 Prediction, input the calculated depth information (second data) into the trained prediction model, and the trained prediction model outputs the probability of slope instability, thereby obtaining the probability of future slope instability under the condition of the second data.

[0097] S5 First judgment, preset a threshold for judging the level of slope instability risk. If the predicted probability is less than this threshold, it is considered that the current slope state is relatively stable, and continue to execute S1 First acquisition to continuously monitor the changes of the slope. If the probability is not less than the threshold, it is considered that the slope has a high risk of instability and S6 warning needs to be executed.

[0098] S6 Warning, send a warning signal.

[0099] This embodiment uses a camera to collect the image data of the monitoring target and constructs a spatial coordinate system to obtain the position data of the monitoring target (i.e., the first data). By collecting the depth information sequences before different slope instabilities, a prediction model is established and trained. Subsequently, the depth information of the monitoring target calculated based on the image data (i.e., the second data) is input into the trained model to predict the probability of slope instability. If the predicted probability exceeds the preset threshold, a warning signal is sent; otherwise, continue to collect data to continuously monitor the slope state.

[0100] Embodiment 2: Refer to Figure 2 , the difference between this embodiment and Embodiment 1 is that after executing S3 Data processing and before executing S4 Prediction, it further includes:

[0101] S31 Second acquisition, in the previous data records, obtain the n depth information of the monitoring target before the current moment, and denote the depth information of the monitoring target before the current moment as the first depth information.

[0102] For the convenience of subsequent data calculation, n is equal to the number of samples in the depth information sequence minus one.

[0103] S32 Information integration: Integrate the n first depth information and the second data into a second data sequence. The integration method can be: Arrange the n first depth information and the second data in chronological order, and denote the arranged sequence as the second data sequence; The integration method can also be: Arrange the n first depth information and the second data in ascending or descending order of magnitude, and denote the arranged sequence as the second data sequence.

[0104] S33 First calculation: Standardize both the k-th depth information sequence and the second data sequence, and calculate the Euclidean distance between the standardized k-th depth information sequence and the standardized second data sequence. The calculation model is as follows:

[0105] ;

[0106] where p = q, is the Euclidean distance between the standardized k-th depth information sequence and the standardized second data sequence; is the q-th sample in the standardized second data sequence. is the -th sample in the standardized k-th depth information sequence.

[0107] S34 Statistics: Set a Euclidean distance threshold, and count the number of all Euclidean distances less than this threshold in the first calculation of S33 as the third data.

[0108] S35 Second judgment: Judge whether the third data is greater than the preset quantity threshold. If so, it indicates that there are enough sample pairs showing similarity, and execute S36 to construct a matrix; If not, it indicates that the second data sequence and the current depth information sequence are not similar enough, and S39 iteration needs to be executed.

[0109] S36 Matrix construction: Construct a distance matrix. The calculation model of the element in the i-th row and j-th column of the distance matrix is as follows:

[0110] ;

[0111] where, is the element in the i-th row and j-th column of the distance matrix; is the Euclidean distance between the i-th sample in the k-th depth information sequence and the j-th sample in the second data sequence.

[0112] S37 First acquisition: Use the dynamic path planning algorithm to find the shortest path from the starting point (the first row and the first column) to the end point (the (n + 1)-th row and the (n + 1)-th column) in the distance matrix, and calculate the sum of all elements on this path as the fourth data.

[0113] S38 Third judgment: Determine whether the fourth data is greater than a preset threshold. If so, execute S39 iteration; if not, execute S4 prediction.

[0114] S39 iteration: Take the (k + 1)-th depth information sequence as the new k-th depth information sequence, and execute S33 first calculation until a preset stop condition is met. The preset stop condition can be: the number of depth information sequences that are sufficiently similar to the second data sequence is greater than a preset quantity threshold, or all depth information sequences have been calculated for similarity with the second data sequence.

[0115] The following elaborates on this embodiment in combination with specific cases.

[0116] Set the current time as T, the depth information at time T is 9.8m, and it is necessary to obtain the depth information of the previous 5 time points, which are respectively:

[0117] At time T - 1: 10m, at time T - 2: 9.9m, at time T - 3: 9.9m, at time T - 4: 9.9m, at time T - 5: 9.9m.

[0118] Then the second data sequence is: {10, 9.9, 9.9, 9.9, 9.9, 9.8}. After normalization, it becomes {1.73, 0, 0, 0, 0, -1.73}.

[0119] There are 4 depth information sequences, which are respectively: {18, 18, 18, 18, 18, 17.8}, {5.3, 5.2, 5.1, 5, 4.9, 4.7}, {40, 40, 40, 39, 38, 37}, {17, 17, 17, 16.9, 16.8, 16.7}.

[0120] The 4 depth information sequences after normalization are respectively: {0.7023, 0.7023, 0.7023, 0.7023, 0.7023, -0.1395}, {1.0913, 0.6546, 0.2179, -0.2179, -0.6546, -1.0913}, {0.6198, 0.6198, 0.6198, -0.1240, -0.9300, -1.7400}, {1.0, 1.0, 1.0, 0.0, -1.0, -2.0}.

[0121] Using the Euclidean distance calculation formula in S33 first calculation, the Euclidean distance between {1.73, 0, 0, 0, 0, -1.73} and {0.7023, 0.7023, 0.7023, 0.7023, 0.7023, -0.1395} is: .

[0122] The Euclidean distance between {1.73, 0, 0, 0, 0, -1.73} and {1.0913, 0.6546, 0.2179, −0.2179, −0.6546, −1.0913} is: .

[0123] The Euclidean distance between {1.73, 0, 0, 0, 0, -1.73} and {0.6198, 0.6198, 0.6198, −0.1240, −0.9300, −1.7400} is: .

[0124] The Euclidean distance between {1.73, 0, 0, 0, 0, -1.73} and {1.0, 1.0, 1.0, 0.0, −1.0, −2.0} is: .

[0125] Set the Euclidean distance threshold to 1.9, then the third data is 3. For example, set the preset quantity threshold to k / 2.

[0126] It is easy to know that in this embodiment, 3 is greater than 4 / 2, so execute S36 to construct a matrix.

[0127] This embodiment is described by taking the first depth information sequence as an example, and the calculation methods of other depth information sequences are the same as those of the first depth information sequence.

[0128] ;

[0129] ;

[0130] Among them, The distance matrix corresponding to the first depth information sequence.

[0131] The shortest path found by using the dynamic path planning algorithm is:

[0132] , then the fourth data is 13.8804.

[0133] Assume the preset threshold is 15, then 13.8804 < 15, that is, S4 prediction can be executed. Otherwise, execute S39 iteration.

[0134] In this embodiment, the Euclidean distance between the depth information sequence and the second data sequence is calculated to determine whether there is a sufficient depth information sequence similar to the second data sequence. If so, by constructing a distance matrix, a method is used to find whether the depth information sequence can be aligned with the second data sequence. Especially when the acquisition frequencies of the depth information sequence and the second data sequence are inconsistent, this embodiment can utilize the distance matrix and the dynamic programming algorithm to explore an optimal alignment path, so that the depth information sequence and the second data sequence can achieve the best match in the time or space dimension. Furthermore, each data point in the second data sequence is mapped to the most similar position in the depth information sequence according to its position information in the distance matrix. This mapping not only considers the similarity of individual data points but also comprehensively takes into account the coherence and consistency of the overall sequence, thereby improving the accuracy and reliability of the mapping result. By mapping the second data sequence into the depth information sequence, the purpose is to improve the accuracy of the slope instability probability in step S4 prediction by using the depth information change situation in the depth information sequence.

[0135] Embodiment 3: When collecting the data of the monitored target position, due to periodic changes such as seasons, the collected position data will change periodically. Therefore, in this embodiment, an FOPID controller is set to adaptively calculate the offset of the position data to correct this periodic offset. Refer to Figure 3 , the difference between this embodiment and Embodiment 2 is that after performing the third judgment in S38 and before performing the S4 prediction, it further includes:

[0136] S71 Second acquisition, obtain the timestamp when the first data is collected, delete the year information in the timestamp, and record the remaining part in the timestamp as the m moment.

[0137] For example, if the first data is collected at 10:00:00 on January 1, 2025, then the m moment is 10:00:00 on January 1.

[0138] S72 Third acquisition, for each past year, at the corresponding m moment (that is, 10:00:00 on January 1 of each year), search for and record the historical position data of the monitored target. The position data includes longitude, latitude, and altitude, and the offset corresponding to each historical position data. The offset represents the deviation of the actual position data from the expected position data, and this deviation comes from the periodic offset of the position data caused by natural factors and other factors. The natural factors may be: the periodic freezing and thawing of the soil.

[0139] S73 Second calculation, perform an average calculation on the position data at the m moment of all years to obtain an average historical position data as the fifth data. Calculate the average value of the offsets at the m moment of all years as the sixth data.

[0140] S74 Third calculation: Calculate the difference between the fifth data and the sixth data, denoted as the first difference.

[0141] S75 First adjustment: Input the first difference and the first data together into a FOPID (fractional order proportional-integral-derivative) controller.

[0142] The FOPID controller can calculate the adjustment amount based on the input signal to optimize or adjust the system performance.

[0143] S76 Fourth calculation: Perform a subtraction operation on the first data and the adjustment amount to obtain the operation result, and use the operation result as the new first data.

[0144] In this embodiment, historical position data and offsets at the same moment in previous years are collected, and the average value of the two is calculated to provide reference data for the position data and offsets at the current moment. The first data and the first difference are input into the FOPID controller to output the adjustment amount corresponding to the first data, and the adjustment amount is used to correct the first data to obtain the corrected first data (i.e., the new first data). This embodiment takes into account the periodic offset of the position data caused by external factors, realizes the correction of the current position data, improves the accuracy of the current position data, and further improves the accuracy of early warning.

[0145] Example 4: Refer to Figure 4 , the difference between this embodiment and Embodiment 3 is that after performing S75 First adjustment and before performing S76 Fourth calculation, it further includes:

[0146] S81 Second modeling and training: Establish an adjustment model, and use the fifth data and the sixth data to train the adjustment model to obtain the trained adjustment model. During the training process, the sixth data is used as the true label of the fifth data to train the adjustment model, so that the trained adjustment model can predict the corresponding offset according to the input position data.

[0147] The adjustment model can adopt various neural network models, such as: CNN, RNN or LSTM, etc., as long as the adjustment model can predict the corresponding offset according to the input position data after being trained by the fifth data and the sixth data.

[0148] S82 Second prediction: Input the first data into the trained adjustment model to obtain the predicted adjustment amount, and the predicted adjustment amount is obtained by the adjustment model's understanding of historical data and analysis of current data.

[0149] S83 Fifth calculation: Calculate the difference between the adjustment amount of the first data in S75 First adjustment and the predicted adjustment amount, denoted as the second difference, and the second difference reflects the magnitude of the difference between the adjustment amount output by the FOPID controller and the expected adjustment amount.

[0150] S84 Fourth judgment: Judge whether the second difference is less than the preset difference threshold. If so, it indicates that the performance of the FOPID controller meets the expectation, and then execute S76 Fourth calculation; if not, it indicates that the performance of the FOPID controller does not meet the expectation, and S85 adaptive learning needs to be executed.

[0151] S85 Adaptive learning includes S851 Building a model, S852 Setting the state, S853 Setting the agent, S854 First adjustment, S855 Model update, and S856 Fifth judgment.

[0152] S851 Building a model: Build a reinforcement learning model.

[0153] S852 Setting the state: Define the state space of the reinforcement learning model. The state space includes: the difference between the adjustment amount output by the FOPID controller and the predicted adjustment amount output by the adjustment model; Set the action space of the reinforcement learning model. The action space includes: the parameter adjustment strategy of the FOPID controller, such as the adjustment strategies of gain, integral time, derivative time, etc.

[0154] S853 Setting the agent: Set the agent, which is used to select the parameter adjustment strategy of the FOPID controller according to the current state of the FOPID controller.

[0155] S854 First adjustment: Input the second difference into the reinforcement learning model to obtain the parameter adjustment strategy corresponding to the second difference, and adjust the current parameters of the FOPID controller according to the parameter adjustment strategy corresponding to the second difference.

[0156] S855 Model update: Take the adjusted FOPID controller as the new controller, and execute S75 First adjustment to obtain a new adjustment amount.

[0157] S856 Fifth judgment: Judge whether the new adjustment amount meets the expectation. If so, execute S76 Fourth calculation; if not, execute S86 Feedback.

[0158] S86 Feedback: Send a feedback signal to the operation and maintenance personnel, and provide the operation and maintenance personnel with information about the accuracy of the adjustment model and the parameter adjustment of the FOPID controller.

[0159] In this embodiment, by setting an adjustment model, the adjustment amount corresponding to the first data is predicted to obtain the predicted adjustment amount. Through the trained adjustment model, an accurate adjustment amount can be output. The subtraction operation is performed on the predicted adjustment amount and the adjustment amount output by the FOPID controller to obtain the difference between the two, that is, the second difference. According to the second difference, it can be determined whether the adjustment amount output by the FOPID controller meets the expectation. If so, the steps of the fourth calculation are executed; otherwise, the steps of adaptive learning are executed to adjust the parameters of the FOPID controller to improve the calculation accuracy of the subsequent adjustment amount. If the adjustment amount output by the FOPID controller still does not meet the expectation after adjusting the parameters, information about the model accuracy and the adjustment of the FOPID controller parameters is provided to the operation and maintenance personnel, enabling the operation and maintenance personnel to adjust the FOPID controller in a timely manner. This embodiment further verifies the output of the FOPID controller to determine whether the performance of the FOPID controller meets the expectation. If not, it is adjusted in a timely manner to improve the accuracy of the subsequent output results, thereby improving the accuracy of the early warning.

[0160] In other embodiments, referring to Figure 5 , after performing the first adjustment in S854 and before performing the model update in S855, it further includes:

[0161] S8541 Fourth acquisition: Acquire training data, where the training data includes the difference between the adjustment amount output by the FOPID controller and the predicted adjustment amount output by the adjustment model, and the parameters of the FOPID controller.

[0162] S8542 Third modeling and training: Establish an MLP model and train the MLP model using the training data to obtain the trained MLP model. During the training process, the parameters of the FOPID controller are used as the true label of the difference between the adjustment amount output by the FOPID controller and the predicted adjustment amount output by the adjustment model to train the MLP model, so that the trained MLP model can predict the corresponding parameters of the FOPID controller based on the difference between the adjustment amount output by the FOPID controller and the predicted adjustment amount output by the adjustment model.

[0163] S8543 Third prediction: Input the second difference into the trained MLP model to obtain the predicted parameters of the FOPID controller, denoted as the first parameters.

[0164] S8544 Fifth acquisition: Acquire the adjusted parameters of the FOPID controller in the first adjustment in S854, denoted as the second parameters.

[0165] Sixth judgment: Determine whether the difference between the first parameter and the second parameter is less than the preset parameter difference threshold. If so, it indicates that the prediction of the MLP model is accurate and the parameter adjustment strategy is effective, and then execute S855 model update; if not, execute S86 feedback.

[0166] In this embodiment, the training data including the difference between the adjustment amount output by the FOPID controller and the adjustment amount predicted by the adjustment model, as well as the corresponding FOPID controller parameters (as the true labels) is first obtained. Then, an MLP model is established and trained using this training data so that the model can predict the parameters of the FOPID controller based on the adjustment amount difference. Then, the new adjustment amount difference is input into the trained MLP model to predict the parameters of the FOPID controller (the first parameter). After that, the adjusted FOPID controller parameters (the second parameter) are obtained. Finally, the difference between the two parameters is compared. If the difference is less than the preset parameter threshold, it is considered that the MLP model prediction is accurate and the parameter adjustment strategy is effective, and the model update step is executed; otherwise, the feedback step is executed to further optimize the model or adjust the strategy.

[0167] Example 5: Refer to Figure 6 , after the execution of S8541 Fourth Acquisition and before the execution of S8542 Third Modeling and Training, it further includes:

[0168] S91 Seventh judgment: Determine whether the number of samples in the training data in S8541 Fourth Acquisition meets the preset number threshold. If so, execute S8542 Third Modeling and Training; if not, execute S92 Fourth Modeling.

[0169] S92 Fourth Modeling: Establish a generative adversarial network.

[0170] S93 Fourth Training: Use the training data in S8541 Fourth Acquisition to train the generative adversarial network to obtain the trained generative adversarial network.

[0171] S94 Generate Data: Use the trained generative adversarial network to generate new training data.

[0172] S95 Data Update: Use the new training data and the training data in S8541 Fourth Acquisition as the new training data.

[0173] In this embodiment, the number of samples in the training data is determined. When the number of samples is not sufficient, a generative adversarial network is used to supplement it to enrich the number of samples and improve the accuracy of the prediction results of the MLP model.

[0174] Example 6: This embodiment discloses a mine safety hazard warning system based on big data, and the system includes:

[0175] A processor and a memory,

[0176] wherein program code is stored in the memory;

[0177] When the processor calls the program code in the memory, it executes the steps of the method.

[0178] The above are all preferred embodiments of the present application. The protection scope of the present application is not limited thereby. Therefore, all equivalent changes made according to the structure, shape, and principle of the present application shall be covered by the protection scope of the present application.

Claims

1. A mine safety hazard warning method based on big data, characterized in that, Including: First acquisition: Construct a spatial coordinate system, and acquire the position data of the monitoring target in the spatial coordinate system, denoted as the first data; Acquire the depth information sequences corresponding to different slope instabilities; First modeling and training: Establish a prediction model, and use the depth information sequences to train the prediction model to obtain the trained prediction model; Data processing: Calculate the depth information of the monitoring target based on the first data, denoted as the second data; Prediction: Input the second data into the trained prediction model to obtain the probability of slope instability; First judgment: Judge whether the probability is less than a preset threshold. If so, execute the steps of the first acquisition; If not, execute the warning steps; Warning: Send a warning signal; After executing the steps of data processing and before executing the steps of prediction, it further includes: Second acquisition: Obtain n depth information of the monitoring target before the current moment, denoted as the first depth information, where n is equal to the number of samples in the depth information sequence minus one; Information integration: Integrate the n first depth information and the second data into a second data sequence; First calculation: Standardize both the k-th depth information sequence and the second data sequence, and calculate the Euclidean distance between the standardized k-th depth information sequence and the standardized second data sequence; Statistics: Count the number of Euclidean distances less than a preset Euclidean distance threshold, denoted as the third data; Second judgment: Judge whether the third data is greater than a preset quantity threshold. If so, execute the steps of prediction; if not, execute the iteration steps; Iteration: Use the (k + 1)-th depth information sequence as the new k-th depth information sequence, and execute the steps of the first calculation until a preset stop condition is met; After executing the steps of the second judgment and before executing the steps of prediction, it further includes: Construct a matrix: Construct a distance matrix, and the calculation model of the element in the i-th row and j-th column of the distance matrix is as follows: ； Among them, is the element in the i-th row and j-th column of the distance matrix; is the Euclidean distance between the i-th sample in the k-th depth information sequence and the j-th sample in the second data sequence; First acquisition: Use the dynamic path planning algorithm to find the shortest path through the distance matrix, and calculate the sum of all elements included in the shortest path, denoted as the fourth data; Third judgment: Judge whether the fourth data is greater than a preset threshold. If so, execute the iteration steps; if not, execute the steps of prediction; After executing the steps of the third judgment and before executing the steps of prediction, it further includes: Second acquisition: Obtain the timestamp of the first data, delete the year information in the timestamp, and denote the remaining part of the timestamp as the m moment; Third acquisition: Respectively obtain the historical position data of the monitoring target at the m moment of each year, and the offset corresponding to each historical position data; Second calculation: Calculate the average value of the historical position data, denoted as the fifth data; calculate the average value of the offsets, denoted as the sixth data; Third calculation: Calculate the difference between the fifth data and the sixth data, denoted as the first difference; First adjustment: Input the first difference and the first data into the FOPID controller to obtain the adjustment amount of the first data; Fourth calculation: Perform a subtraction operation on the first data and the adjustment amount to obtain the operation result, and use the operation result as the new first data.

2. The method for early warning of mine safety hazards based on big data according to claim 1, wherein, After executing the steps of the first adjustment and before executing the steps of the fourth calculation, it further includes: Second Modeling and Training: Establish a regulation model, and use the fifth data and the sixth data to train the regulation model to obtain the trained regulation model; Second Prediction: Input the first data into the trained regulation model to obtain the predicted regulation amount; Fifth Calculation: Calculate the difference between the regulation amount and the predicted regulation amount, denoted as the second difference; Fourth Judgment: Judge whether the second difference is less than the preset difference threshold. If so, execute the steps of the fourth calculation; if not, execute the feedback step; Feedback: Send a feedback signal to the operation and maintenance personnel.

3. The mine safety hazard early warning method based on big data according to claim 2, wherein After executing the steps of the fourth judgment and before executing the feedback step, it also includes: Model Construction: Construct a reinforcement learning model; State Setting: Set the state space of the reinforcement learning model, and the state space includes: the difference between the regulation amount output by the FOPID controller and the predicted regulation amount output by the regulation model; set the action space of the reinforcement learning model, and the action space includes: the parameter adjustment strategy of the FOPID controller; Agent Setting: Set an agent, and the agent is used to select the parameter adjustment strategy of the FOPID controller according to the current state of the FOPID controller; First Adjustment: Input the second difference into the reinforcement learning model to obtain the parameter adjustment strategy corresponding to the second difference, and adjust the current parameters of the FOPID controller according to the parameter adjustment strategy corresponding to the second difference; Model Update: Take the adjusted FOPID controller as the new controller, and execute the steps of the first regulation to obtain the new regulation amount; Fifth Judgment: Judge whether the new regulation amount meets the expectation. If so, execute the steps of the fourth calculation; if not, execute the feedback step.

4. The method for early warning of mine safety hazards based on big data according to claim 3, wherein, After executing the steps of the first adjustment and before executing the model update step, it also includes: Fourth Acquisition: Acquire training data, and the training data includes the difference between the regulation amount output by the FOPID controller and the predicted regulation amount output by the regulation model, and the parameters of the FOPID controller; Third Modeling and Training: Establish an MLP model, and use the training data to train the MLP model to obtain the trained MLP model; Third Prediction: Input the second difference into the trained MLP model to obtain the predicted parameters of the FOPID controller, denoted as the first parameter; Fifth Acquisition: Acquire the parameters of the FOPID controller after adjustment in the steps of the first adjustment, denoted as the second parameter; Sixth Judgment: Judge whether the difference between the first parameter and the second parameter is less than the preset parameter difference threshold. If so, execute the model update step; if not, execute the feedback step.

5. The mine safety hazard early warning method based on big data according to claim 4, characterized in that After executing the steps of the fourth acquisition and before executing the steps of the third modeling and training, it also includes: Seventh Judgment: Judge whether the number of samples in the training data meets the preset number threshold. If so, execute the steps of the third modeling and training; if not, execute the steps of the fourth modeling; Fourth Modeling: Establish a generative adversarial network; Data Generation: Use the generative adversarial network to generate new training data; Data Update: Take the new training data and the training data in the steps of the fourth acquisition as the new training data.

6. The method for early warning of mine safety hazards based on big data according to claim 5, wherein After executing the steps of the fourth modeling and before executing the data generation step, it also includes: Fourth training: Use the training data in the fourth acquisition step to train the generative adversarial network to obtain the trained generative adversarial network; In the step of generating data, use the trained generative adversarial network to generate new training data.

7. A mine safety hazard early warning system based on big data, characterized in that, Including: A processor and a memory, Program code is stored in the memory; When the processor calls the program code in the memory, it executes the steps of the method described in any one of claims 1-6.

Images