Method and device for evaluating supporting quality of fully mechanized coal mining face

Abstract

The invention relates to the technical field of coal mine fully-mechanized coal mining support, and provides a fully-mechanized coal mining face support quality evaluation method and device, and the method comprises the steps: collecting support related data of a fully-mechanized coal mining face through a multi-source sensor arranged on the fully-mechanized coal mining face; fusing the support related data through a Kalman filtering algorithm and a deep belief network fusion algorithm to obtain fused data; analyzing the fused data through an autoregressive integral moving average model, predicting support parameters in future preset time, and obtaining support prediction parameters; and determining an evaluation threshold value of the support quality according to the support prediction parameters. According to the method, accurate support related data can be obtained, efficient and high-precision fusion is carried out on the support related data, the support parameters are accurately predicted by utilizing the autoregression integral moving average model, then the evaluation threshold is dynamically generated, the accuracy of a fully mechanized coal mining face support quality evaluation result is improved, and the support quality evaluation efficiency is improved. And the method can better adapt to complex working conditions and stope environments.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fully mechanized coal mining support, and particularly relates to a fully mechanized coal mining face support quality evaluation method and device. BACKGROUND

[0002] The support quality monitoring of the fully mechanized coal mining face mainly depends on manual inspection and static threshold judgment, and the data acquisition and processing capacity is limited, so that the dynamic changes of the support quality cannot be captured in real time. Meanwhile, the fixed threshold is used for support quality evaluation, which cannot adapt to the changes of the geological conditions and the mining environment, resulting in inaccurate evaluation results. SUMMARY

[0003] The embodiments of the present application aim to provide a fully mechanized coal mining face support quality evaluation method and device, so as to at least solve the technical problems that the data acquisition and processing capacity is limited when the support quality is evaluated, and the fixed threshold is used for evaluation, which cannot adapt to the changes of the geological conditions and the mining environment, resulting in inaccurate evaluation results.

[0004] In order to solve the above technical problems, the embodiments of the present application adopt the following technical solutions:

[0005] In a first aspect, the embodiments of the present application provide a fully mechanized coal mining face support quality evaluation method, comprising:

[0006] acquiring support related data of the fully mechanized coal mining face through a plurality of source sensors arranged at the fully mechanized coal mining face;

[0007] fusing the support related data through a Kalman filtering algorithm and a deep belief network fusion algorithm to obtain fused data;

[0008] analyzing the fused data through an autoregressive integrated moving average model to predict support parameters in a future preset time to obtain support prediction parameters;

[0009] determining an evaluation threshold of the support quality according to the support prediction parameters.

[0010] In some embodiments, the fusing of the support related data through the Kalman filtering algorithm and the deep belief network fusion algorithm to obtain the fused data comprises:

[0011] processing the support related data through the Kalman filtering algorithm to output a first state estimation value of the support related data;

[0012] optimizing the first state estimation value and the support related data through the deep belief network fusion algorithm to obtain a second state estimation value.

[0013] In some embodiments, the first state estimate and the support-related data are optimized using a deep belief network fusion algorithm to obtain a second state estimate, including:

[0014] The first state estimate and the support-related data are combined to form an input feature vector;

[0015] The input feature vector is input into a preset deep belief network to extract target features, wherein the target features include at least one of nonlinear relationships, hidden patterns, and potential noise patterns;

[0016] The second state estimate is generated based on the target features.

[0017] In some embodiments, the support-related data is processed using a Kalman filter algorithm, including:

[0018] Determine the corresponding state equation and observation equation based on the sensor type;

[0019] Adjust the Kalman gain parameters based on the monitored support-related data and historical support-related data;

[0020] The Kalman gain parameter is adjusted based on the support environment data of the fully mechanized mining face, wherein the support environment data includes geological condition data and mining environment data.

[0021] In some embodiments, the fused data is analyzed using an autoregressive integral moving average model to predict support parameters within a preset time period, resulting in predicted support parameters, including:

[0022] Differential processing is performed on the support-related data and historical support-related data to obtain a stationary time series;

[0023] Construct an autoregressive integral moving average model based on the stationary time series;

[0024] The autoregressive integral moving average model was fitted using the historical support data to obtain the fitting results;

[0025] Based on the fitting results, the support parameters are predicted within a preset time period to obtain the predicted support parameters.

[0026] In some embodiments, determining an evaluation threshold for support quality based on the support prediction parameters includes:

[0027] If the predicted support parameter is higher than the preset support parameter range, the upper limit of the evaluation threshold is increased;

[0028] If the predicted support parameter is lower than the preset support parameter range, the lower limit of the evaluation threshold is lowered.

[0029] In some embodiments, the method further comprises:

[0030] According to the evaluation threshold of the support quality and the deviation of the support-related data, adjusting the evaluation criterion of the support quality through control chart analysis.

[0031] In some embodiments, adjusting the evaluation criterion of the support quality through control chart analysis according to the evaluation threshold of the support quality and the deviation of the support-related data comprises:

[0032] Determining a control limit according to the evaluation threshold of the support quality, wherein the control limit comprises an upper control limit and a lower control limit;

[0033] Marking the support-related data as a data point on a control chart;

[0034] Determining whether the data point is beyond the control limit;

[0035] If the data point is beyond the control limit, adjusting the evaluation criterion;

[0036] If the data point is close to the control limit, fine-tuning the evaluation criterion.

[0037] In some embodiments, the method further comprises:

[0038] If the data points are uniformly distributed, keeping the evaluation criterion unchanged.

[0039] In a second aspect, the embodiments of the present application provide a support quality evaluation device for fully mechanized coal mining face, comprising:

[0040] A collection module configured to collect support-related data of the fully mechanized coal mining face through a plurality of source sensors arranged at the fully mechanized coal mining face;

[0041] A fusion module configured to fuse the support-related data through Kalman filtering algorithm and deep belief network fusion algorithm to obtain fused data;

[0042] A prediction module configured to analyze the fused data through an autoregressive integrated moving average model, predict support parameters in a future preset time, and obtain support prediction parameters;

[0043] A determination module configured to determine an evaluation threshold of support quality according to the support prediction parameters.

[0044] In a third aspect, the embodiments of the present application provide an electronic device comprising at least a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned support quality evaluation method for fully mechanized coal mining face when executing the computer program stored in the memory.

[0045] In a fourth aspect, the embodiments of the present application provide a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the support quality evaluation method for a fully mechanized coal mining face.

[0046] The support quality evaluation method and device for a fully mechanized coal mining face provided by the embodiments of the present application can collect support related data of the fully mechanized coal mining face through a multi-source sensor arranged at the fully mechanized coal mining face, fuse the support related data through a Kalman filtering algorithm and a deep belief network fusion algorithm to obtain fused data, analyze the fused data through a self-regressive integral moving average model to predict support parameters in a future preset time to obtain support prediction parameters, and determine an evaluation threshold of support quality according to the support prediction parameters. The embodiments of the present application can realize real-time acquisition of accurate support related data, efficient and high-precision fusion of the support related data through the Kalman filtering algorithm and the deep belief network fusion algorithm, accurate prediction of support parameters through the self-regressive integral moving average model, dynamic generation of an evaluation threshold, and improvement of the accuracy of the support quality evaluation result of the fully mechanized coal mining face, and can better adapt to complex working conditions and stope environments. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0048] Figure 1 The first flowchart of the support quality evaluation method for a fully mechanized coal mining face of the embodiments of the present application;

[0049] Figure 2 The second flowchart of the support quality evaluation method for a fully mechanized coal mining face of the embodiments of the present application;

[0050] Figure 3 The structural schematic diagram of the support quality evaluation device for a fully mechanized coal mining face of the embodiments of the present application. DETAILED DESCRIPTION

[0051] The various schemes and features of the present application are described herein with reference to the accompanying drawings.

[0052] It should be understood that various modifications can be made to the embodiments of the present application. Therefore, the above description should not be regarded as limiting, but only as an example of the embodiments. Those skilled in the art will think of other modifications within the scope and spirit of the present application.

[0053] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the application and, together with the general description of the application given above, and the detailed description of the embodiments given below, serve to explain the principles of the present application.

[0054] These and other characteristics of the present application will become apparent from the following description of the preferred forms given, by way of non-limiting example, with reference to the attached drawings.

[0055] It is also to be understood that even though a number of specific embodiments of the application have been described herein, many other modifications and variations thereof will be apparent to those skilled in the art.

[0056] The above and other aspects, features and advantages of the present application will become apparent from the following detailed description, taken in conjunction with the accompanying drawings, when considered in conjunction with the following detailed description.

[0057] Specific embodiments of the present application are described hereinafter, with reference to the accompanying drawings; however, it will be understood that the application is not limited to the specific embodiments described and that the application can be practiced with modification and alteration, and is capable of being practiced or carried out in various ways. As such, this detailed description is not meant to limit the scope of the application but is merely meant to provide a specific example of the application.

[0058] The specification can use phrases such as "in one embodiment," "in another embodiment," "in yet another embodiment," or "in other embodiments," which can refer to one or more embodiments of the same or different embodiments.

[0059] In the prior art, the support quality monitoring of fully mechanized coal mining faces relies on manual inspection and static threshold judgment, which has the following defects:

[0060] (1) Poor data timeliness: Traditional monitoring methods (such as pressure gauge readings, manual recording) cannot capture dynamic changes such as support initial support force, roof fragmentation in real time.

[0061] (2) Insufficient multi-source data fusion and analysis capability:

[0062] In fully mechanized coal mining faces, the support quality monitoring of hydraulic supports requires real-time acquisition and processing of multi-source data (such as pressure, displacement, inclination, etc.). However, sensor data is often disturbed by noise, and data from different sensors may have heterogeneity (such as sampling frequency, dimensional difference, etc.), which cannot fully utilize multi-sensor data for comprehensive analysis, resulting in insufficient data fusion accuracy.

[0063] (3) Evaluation criteria are single: using a fixed threshold (such as initial support force ≥ 24 MPa) to judge support quality, without considering the dynamic influence of geological condition changes on support effectiveness, and the fixed threshold evaluation criteria cannot adapt to complex and variable geological conditions;

[0064] (4) Lack of adaptive ability: existing systems are difficult to automatically adjust evaluation model parameters according to stope environment (such as periodic pressure, coal seam inclination).

[0065] Therefore, an embodiment of the present application provides a fully mechanized working face support quality evaluation method and device.

[0066] Embodiment one

[0067] Figure 1 The flowchart of the fully mechanized working face support quality evaluation method of the embodiment of the present application is shown, as shown in Figure 1 The fully mechanized working face support quality evaluation method provided by the embodiment of the present application comprises:

[0068] S101: Collecting support related data of the fully mechanized working face through a multi-source sensor arranged at the fully mechanized working face.

[0069] The fully mechanized working face support quality evaluation method is applied to a fully mechanized working face support quality evaluation system, and the multi-source sensor constitutes a dynamic sensing layer of the system. Through the deployment of multiple types of sensors at the fully mechanized working face, data is collected in real time. The support related data includes support parameter data, support environment data, and mining data, etc. The support parameter data includes support pressure, support force, support spacing, etc. of the hydraulic support. The support environment data includes geological condition data (such as coal seam thickness, roof lithology) and stope environment data (such as periodic pressure law, roof fragmentation, etc.). The mining data includes fully mechanized working condition data, such as coal mining machine cutting data.

[0070] In specific implementation, pressure sensors, displacement sensors, inclination sensors, microseismic sensors, and infrared thermal imagers, etc. can be deployed at key positions such as hydraulic support columns, roofs, and coal walls to collect multi-dimensional data such as support pressure, roof subsidence, coal wall spalling depth, and support posture in real time.

[0071] For example, the pressure sensor is installed on the hydraulic support column to monitor the support pressure in real time. The displacement sensor is installed between the roof and the support to monitor the roof subsidence. The inclination sensor is installed on the support to monitor the posture change of the support. The microseismic sensor is installed near the coal wall to monitor the coal wall spalling depth.

[0072] In some embodiments, the method further comprises:

[0073] S201: Preprocess the collected support-related data, which includes noise reduction, feature extraction, and normalization processing.

[0074] The fully mechanized coal mining face support quality evaluation system includes an edge computing layer for preprocessing and real-time transmission of the collected data. The edge computing layer can utilize the underground communication network (5G network, Ethernet, etc.) to transmit the processed collected data to the cloud server (e.g., upper computer) in real time, realizing remote control of the hydraulic support. The edge computing layer is preferably an edge computing gateway installed on the hydraulic support. The edge computing node can periodically aggregate collected data (e.g., aggregate data every 5 seconds) and upload it to the ground equipment and cloud server through the underground communication network.

[0075] In this embodiment, by deploying multiple types of sensors at the fully mechanized coal mining face and combining edge computing technology for data preprocessing, multi-dimensional support-related data can be collected and processed in real time, ensuring data accuracy and real-time performance, improving data collection and processing capabilities, and effectively solving the problem of poor data timeliness in existing technologies.

[0076] S102: Fuse the support-related data through Kalman filtering algorithm and deep belief network fusion algorithm to obtain fused data.

[0077] In this embodiment, the Kalman filtering algorithm and deep belief network (DBN) fusion algorithm are used to eliminate sensor errors, extract key features of support-related data, improve data fusion accuracy, and enhance data fusion and analysis capabilities. Through deep fusion and analysis of multi-source data, actual working conditions can be more comprehensively reflected, providing more reliable support for support decision-making.

[0078] S103: Analyze the fused data through an autoregressive integrated moving average model to predict support parameters in a future preset time to obtain support prediction parameters.

[0079] The autoregressive integrated moving average (ARIMA) model is a model for time series analysis and prediction, combining autoregressive (AR), difference (I), and moving average (MA) three parts, suitable for data with trend and seasonal characteristics. In this step, based on the ARIMA model, the real-time collected support-related data can be used to predict the trend of support parameters in a future period of time.

[0080] S104: Determine the evaluation threshold of support quality according to the support prediction parameters.

[0081] In this step, the evaluation threshold of support quality can be determined according to the support prediction parameters obtained by the ARIMA model prediction, so that the support quality evaluation threshold can be dynamically adjusted according to the complex and changeable geological conditions, the evaluation result is more accurate, and the adaptability is stronger.

[0082] The support quality evaluation method of the fully mechanized working face provided by the embodiment of the application can collect support related data of the fully mechanized working face through the multi-source sensor arranged in the fully mechanized working face, fuse the support related data through a Kalman filtering algorithm and a deep belief network fusion algorithm to obtain fused data, analyze the fused data through an autoregressive integrated moving average model to predict support parameters in a future preset time to obtain support prediction parameters, determine an evaluation threshold of support quality according to the support prediction parameters, can obtain accurate support related data in real time, and can efficiently and accurately fuse the support related data through the Kalman filtering algorithm and the deep belief network fusion algorithm, and can accurately predict the support parameters through the autoregressive integrated moving average model, and then dynamically generate the evaluation threshold, improve the accuracy of the evaluation result of the support quality of the fully mechanized working face, and can better adapt to complex working conditions and stope environments.

[0083] In some embodiments, in step S102, the support related data is fused through a Kalman filtering algorithm and a deep belief network fusion algorithm to obtain fused data, including:

[0084] S1021: The support related data is processed through a Kalman filtering algorithm to output a first state estimation value of the support related data.

[0085] S1022: The first state estimation value and the support related data are optimized through a deep belief network fusion algorithm to obtain a second state estimation value.

[0086] The traditional Kalman filtering depends on an accurate mathematical model, but the working conditions of the fully mechanized working face of the coal mine are complex and changeable, and it is difficult to establish a completely accurate mathematical model. In the embodiment, the Kalman filtering is combined with deep learning, and the state estimation result of the Kalman filtering is further optimized through the deep belief network.

[0087] Specifically, when data fusion is performed, the first state estimation value of the support related data is obtained through the Kalman filtering algorithm, then the first state estimation value and the support related data collected in step S101 are combined, the first state estimation value and the original support related data are fused and optimized through the deep belief network fusion algorithm, and a second state estimation value more accurate for quality evaluation is obtained.

[0088] Since Kalman filter is difficult to capture nonlinear characteristics, the state estimation accuracy is insufficient. In this embodiment, the state estimation result of Kalman filter can be further optimized by using deep belief network fusion algorithm, the nonlinear characteristics that Kalman filter is difficult to capture are learned, the accuracy of state estimation is improved, and Kalman filter can more accurately predict and update the supporting quality related parameters under complex working conditions. In this embodiment, the state estimation of Kalman filter algorithm can be optimized by combining deep learning, and the real-time performance and robustness of Kalman filter algorithm can be improved.

[0089] In some embodiments, in step S1021, the supporting related data is processed by a Kalman filter algorithm, including:

[0090] S301: Determine the corresponding state equation and observation equation according to the sensor type.

[0091] Traditional Kalman filter can only process single sensor data or simply linearly combined multi-source data, and has limited data fusion capability for multi-type sensors (pressure, displacement, inclination, etc.). In this step, different state equations and observation equations are designed for different types of sensor data (such as pressure sensors, displacement sensors, and inclination sensors), which expands Kalman filter into a multi-source data fusion tool, effectively improves the data fusion accuracy of Kalman filter algorithm, and provides more reliable support for supporting quality evaluation.

[0092] The state equations and observation equations of different types of sensors are described below.

[0093] 1) Pressure sensor

[0094] (1) State equation:

[0095] The pressure sensor is used to monitor the pressure change of the hydraulic support column. The change of the column pressure value is a continuous process, which is affected by factors such as hydraulic system characteristics and geological conditions. The mathematical expression of its state equation is:

[0096] ;

[0097] wherein, is the pressure value at the kth moment; A is a scalar fitted according to historical data; The covariance Q of can be adjusted according to the actual working condition.

[0098] (2) Observation equation

[0099] The observation equation relates the pressure sensor measurement value to the true pressure value, and considers the influence of sensor error. Its mathematical expression is:

[0100] ;

[0101] in, The value is the pressure sensor measurement at time k; H represents the observation matrix, set as an identity matrix (1×1); v k For observing noise. k Set as Gaussian white noise, v k The covariance matrix R was determined through calibration experiments.

[0102] 2) Displacement sensor

[0103] (1) Equations of state

[0104] Displacement sensors are used to monitor changes in roof subsidence. Roof subsidence is a slow and non-linear process, influenced by geological conditions and the supporting force of the scaffolding. The mathematical expression of its state equation is:

[0105] ;

[0106] Where, x k The subsidence at time k represents the amount of roof subsidence; the nonlinear function f(.) describes the relationship between the amount of roof subsidence and time, and is obtained by fitting experimental data; the process noise w k w represents the random disturbance experienced during the subsidence of the roof. k The covariance matrix Q is adjusted based on the actual working conditions.

[0107] (2) Observation equation

[0108] The observation equation correlates the displacement measurements from the displacement sensor with the actual roof settlement, taking into account the influence of sensor error. Its mathematical expression is:

[0109] ;

[0110] Among them, z k Let be the displacement sensor measurement value at time k; h(.) is a nonlinear observation function representing the relationship between the sensor measurement value and the true value, which can be determined based on the sensor characteristics and installation location; v k The observation noise can be set to Gaussian white noise, v k The covariance matrix R can be determined through calibration experiments.

[0111] 3) Tilt sensor

[0112] (1) Equation of state:

[0113] Tilt sensors are used to monitor changes in the attitude of hydraulic supports. The change in support attitude is a smooth process, influenced by ground movement and support adjustments. The mathematical expression of its state equation is: ;

[0114] Where, x k Let be the inclination angle of the support at time k; A is the state transition matrix, representing the trend of the inclination angle changing over time, which is obtained by fitting historical data; w k For process noise, w represents the random disturbance caused by the change in tilt angle. k The covariance matrix Q needs to be adjusted according to the actual working conditions.

[0115] (2) Observation equation

[0116] The observation equation correlates the sensor measurements with the true tilt angle, taking into account the influence of sensor error. Its mathematical expression is:

[0117] ;

[0118] Among them, z k Let H be the tilt sensor measurement at time k; H is the observation matrix, set as an identity matrix (1×1); v k To observe the noise, we set it to Gaussian white noise, v k The covariance matrix R was determined through calibration experiments.

[0119] As can be seen from the above, this application sets different state equations and observation equations for different sensor types, which can enhance the noise suppression capability in multi-source data fusion and improve the data fusion accuracy.

[0120] In other embodiments, step S1021 involves processing the support-related data using a Kalman filter algorithm, including:

[0121] S302: Adjust the Kalman gain parameter based on the monitored support-related data and historical support-related data;

[0122] S303: Adjust the Kalman gain parameter based on the support environment data of the fully mechanized mining face, wherein the support environment data includes geological condition data and mining environment data.

[0123] In traditional Kalman filtering, the Kalman gain is calculated based on fixed model parameters and cannot be dynamically adjusted according to changes in real-time operating conditions. In this embodiment, real-time monitored support-related data can be combined with historical support-related data to dynamically adjust the Kalman gain parameters. For example, historical support-related data can be combined with real-time operating conditions to dynamically adjust the Kalman gain parameters. Simultaneously, a feedback mechanism can be introduced to use real-time monitored support-related data such as roof subsidence and support posture to perform online correction of the Kalman gain parameters.

[0124] Further, considering environmental factors such as changes in geological conditions, support environmental data such as geological condition data (such as periodic pressure, coal seam inclination) are taken as input variables of Kalman filtering, and the Kalman gain parameter is adaptively adjusted to adapt to complex and variable geological conditions and stope environment.

[0125] In the embodiments of the application, the Kalman filtering algorithm is improved by adaptively adjusting the Kalman gain parameter and enhancing the noise suppression capability in multi-source data fusion, and the improved Kalman filtering algorithm is used to process multi-source data such as pressure and displacement, thereby improving the data fusion accuracy. The improved Kalman filtering algorithm can better adapt to the complex and variable working conditions of the fully mechanized coal mining face, and provide more reliable technical support for dynamic perception and adaptive evaluation of support quality.

[0126] In some embodiments, as shown in Figure 2 In step S1021, the first state estimate value and the support-related data are optimized by a deep belief network fusion algorithm to obtain a second state estimate value, including:

[0127] S401: Combine the first state estimate value and the support-related data to form an input feature vector;

[0128] S402: Input the input feature vector into a preset deep belief network to extract target features, wherein the target features include at least one of a nonlinear relationship, an implicit pattern, and a potential noise rule;

[0129] S403: Generate the second state estimate value according to the target features.

[0130] In a specific implementation, the first state estimate value (such as pressure, displacement, inclination, etc.) output by the Kalman filter is first normalized, and is combined with the original sensor data (support-related data) to form a complete input feature vector, thereby completing the preprocessing of the deep belief network input data. When normalizing the first state estimate value, it is necessary to ensure that the value range is suitable for neural network training.

[0131] Further, a deep belief network (DBN) is constructed. The DBN is stacked by multiple restricted Boltzmann machines (RBM), and usually contains 3-5 hidden layers, each of which is responsible for learning different levels of features of the input data.

[0132] Then, an unsupervised pre-training method is used to initialize the DBN, train each restricted Boltzmann machine (RBM) layer by layer, and learn the deep distribution of the data. The goal of the pre-training stage is to restore the features of the input data as much as possible.

[0133] After pre-training, the entire DBN is supervised trained using labeled data, and the network weights are adjusted through back propagation algorithm (supervised fine-tuning) to optimize the state estimation accuracy.

[0134] During supervised training, a target function is first determined, which is a loss function used to measure the difference between the DBN output and the actual state value. For example, mean square error (MSE) or cross-entropy loss. The loss function can be expressed as:

[0135]

[0136] After fine-tuning the DBN through supervised training, the DBN is used for feature extraction and state optimization. The DBN can extract deep features from the state estimation value output by the Kalman filter and the original sensor data. Deep features include nonlinear relationships, hidden patterns, and potential noise patterns. After extracting the target features, the DBN generates a more accurate second state estimation value based on the extracted target features. The optimized second state estimation value can better reflect the actual working condition of the support and improve the accuracy of the support quality evaluation.

[0137] In practical applications, the support condition will change over time. Therefore, in this embodiment, an incremental learning method can be used to dynamically update the DBN model parameters, and the second state estimation value output can be corrected online combined with real-time data to ensure the stability of the results. Using the incremental learning method, the model parameters can be gradually adjusted without retraining the entire network, improving the model adjustment efficiency. The correction process can be achieved through weighted averaging or other fusion strategies to ensure the stability of the output results.

[0138] In some embodiments, in step S103, the fusion data is analyzed by an autoregressive integrated moving average model to predict support parameters in a future preset time, obtaining support prediction parameters, including:

[0139] S1031: The support-related data and historical support-related data are subjected to difference processing to obtain a stationary time series;

[0140] S1032: An autoregressive integrated moving average model is constructed according to the stationary time series;

[0141] S1033: The autoregressive integrated moving average model is fitted using the historical support-related data to obtain a fitting result;

[0142] S1034: According to the fitting result, the support parameters in a future preset time are predicted to obtain the support prediction parameters.

[0143] The ARIMA model is used to predict the trend of the future support parameters, and dynamically generates the evaluation threshold to effectively deal with the complex and changeable working conditions of the fully mechanized coal mining face, and improve the accuracy and adaptability of the support quality evaluation.

[0144] In a specific implementation, historical support-related data (such as pressure, displacement, inclination, etc.) and real-time collected support-related data can be used as input data of the ARIMA model. The ARIMA model pre-processes the input data, for example, denoising the input data to remove outliers and noise interference; then, the input data is stabilized by difference operation to convert it into a stationary time series. That is, in step S1031, the historical support-related data and the real-time collected support-related data are denoised and stabilized.

[0145] Further, an ARIMA model is constructed, and the model form is represented as:

[0146] ;

[0147] Wherein, p is the number of autoregressive terms, representing the influence degree of historical data on the current value; d is the difference order, used to convert a non-stationary sequence into a stationary sequence; q is the number of moving average terms, representing the smoothing degree of error terms.

[0148] When selecting parameters for the ARIMA model, the AIC (Akaike Information Criterion) or BIC (Bayesian Information Criterion) criterion is used to determine the best p, d, q parameters. For example, in this embodiment, after multiple experiments, the final ARIMA model is ARIMA(2,1,1).

[0149] After the ARIMA model is constructed, the ARIMA model is used to predict the support parameters, such as predicting the trend of the support parameters in the next 24 hours. The specific prediction process includes: fitting the ARIMA model using historical support-related data; according to the fitting result, predicting the support parameter value (support prediction parameter) in the future preset time, and then dynamically generating the evaluation threshold based on the predicted parameter value.

[0150] In some embodiments, in step S104, the evaluation threshold of the support quality is determined according to the support prediction parameter, including:

[0151] S1041: If the support prediction parameter is higher than the preset support parameter range, the upper limit of the evaluation threshold is increased;

[0152] S1042: If the support prediction parameter is lower than the preset support parameter range, the lower limit of the evaluation threshold is decreased.

[0153] When the evaluation threshold is generated by using the predicted parameter value, if the predicted parameter value is higher than the normal supporting parameter range, the upper limit of the evaluation threshold is increased, so as to avoid that the upper limit of the evaluation threshold is too low, and the higher predicted parameter value in the normal supporting parameter range is mistakenly considered as an abnormal supporting parameter, thereby affecting the accuracy of the evaluation result; and if the predicted parameter value is lower than the normal supporting parameter range, the lower limit of the evaluation threshold is reduced, so as to avoid that the lower limit of the evaluation threshold is too high, and the lower predicted parameter value in the normal supporting parameter range is mistakenly considered as an abnormal supporting parameter, thereby affecting the accuracy of the evaluation result.

[0154] For example, the supporting pressure data collected by the pressure sensor is taken as an example to illustrate the generation of the dynamic evaluation threshold. The input of the ARIMA model is the historical supporting pressure data [x1, x2,..., xn] of the current pressure sensor, and the output is the dynamically generated pressure threshold interval [L t ,U t ], wherein L t represents the dynamic lower limit threshold, and U t represents the dynamic upper limit threshold. The calculation formula of L t and U t is as follows:

[0155] ;

[0156] wherein μ t is the future mean value predicted by the ARIMA model; σ t is the standard deviation of the predicted value; and k is a confidence level coefficient (for example, 2 or 3).

[0157] In some embodiments, the method further comprises:

[0158] S105: adjusting the evaluation standard of the supporting quality by control chart analysis according to the evaluation threshold of the supporting quality and the deviation of the supporting related data.

[0159] The control chart is a statistical process control tool for monitoring whether the process is in a controlled state. In this step, the evaluation standard of the supporting quality is adjusted in real time by control chart analysis.

[0160] In this embodiment, the actual supporting related data is compared with the dynamic evaluation threshold in real time by control chart analysis, and it is judged whether the evaluation standard of the supporting quality needs to be adjusted. The control chart analysis is responsible for monitoring the deviation between the actual data and the dynamic threshold in real time, and the evaluation standard is flexibly adjusted, so that the complex and changeable working conditions of the fully mechanized coal mining face in the coal mine can be effectively coped with, and the accuracy and adaptability of the supporting quality evaluation can be improved.

[0161] In some embodiments, in step S105, the evaluation criterion of the support quality is adjusted by control chart analysis according to the deviation of the support-related data from the evaluation threshold of the support quality, including:

[0162] S1051: determining a control limit according to the evaluation threshold of the support quality, wherein the control limit includes an upper control limit and a lower control limit;

[0163] S1052: marking the support-related data as a data point on the control chart;

[0164] S1053: determining whether the data point is beyond the control limit;

[0165] S1054: adjusting the evaluation criterion if the data point is beyond the control limit;

[0166] S1055: fine-tuning the evaluation criterion if the data point is close to the control limit.

[0167] The input data of the control chart is the real-time collected support parameter data (such as pressure, displacement, inclination, etc.) and the dynamically generated evaluation threshold interval [L t ,U t ].

[0168] First, the real-time collected support parameter data is normalized to ensure that the numerical range is suitable for control chart analysis. Then, the deviation of the actual support parameter value at each time from the threshold interval is calculated.

[0169] When constructing the control chart, the mean control chart (X-bar chart) and the standard deviation control chart (S-chart) are combined to generate the control chart. The key indicators of the control chart analysis include the mean of the real-time support parameter data (X ) and the standard deviation of the real-time support parameter data (s).

[0170] After the control chart is constructed, the control limit is calculated. In this embodiment, the upper control limit (UCL) and the lower control limit (LCL) are calculated according to the dynamic threshold interval [L t ,U t ].

[0171] According to the control chart analysis results, the evaluation criterion is adjusted in real time. When drawing the control chart, the real-time data points are marked on the chart to determine whether the data points are beyond the control limit:

[0172] If the data points are beyond the UCL or LCL, an alarm is triggered and the evaluation criterion is adjusted;

[0173] If the data points are close to the control limit, the evaluation criterion is gradually fine-tuned.

[0174] For example, based on the real-time data from the current pressure sensor [y1, y2, ..., y m ] and dynamic threshold range [L t U t The adjusted evaluation standard range is determined to be [L]. t ′,U t If most data points are close to UCL, then appropriately increase Ut′ so that most data points are within the evaluation criterion range [L]. t ′,U t Similarly, if most data points are close to LCL, Lt′ should be appropriately reduced to achieve dynamic adjustment and updating of the evaluation criteria, ensuring that the evaluation criteria are adapted to the current support conditions.

[0175] In some embodiments, the method further includes:

[0176] S1056: If the data points are evenly distributed, the evaluation criteria remain unchanged.

[0177] If the data points in the control chart are evenly distributed, then the original evaluation criteria can evenly cover the support parameter data under the previous support conditions. The original evaluation criteria are good evaluation criteria, so the evaluation criteria should be kept unchanged.

[0178] In existing technologies, a closed-loop feedback mechanism is lacking, making it difficult to dynamically adjust support strategies based on real-time data. In this embodiment, a closed-loop feedback mechanism can be used to transmit optimized evaluation thresholds and standards to a cloud server. The cloud server evaluates the support parameter data based on these thresholds and standards, optimizes the support parameters according to the evaluation results, and then transmits the optimized support parameters to the electro-hydraulic control system. This system controls the hydraulic support to provide support according to the optimized parameters. Simultaneously, the adjustment effect of the support parameters is verified based on the optimized parameters, and the model parameters of the Kalman filter algorithm, deep belief network fusion algorithm, and autoregressive integral moving average model are dynamically updated based on the adjustment effect, achieving closed-loop control and ensuring continuous optimization of the support quality evaluation system. In other words, this application can form a closed-loop process of "data acquisition → model prediction → optimization suggestion → automatic execution → effect verification → model update," ensuring that the support quality evaluation system can adapt to complex and changing working conditions in real time, continuously and accurately evaluate support parameters, and optimize support strategies.

[0179] In summary, the embodiment of the present application can significantly improve the data acquisition and processing capability by deploying multiple types of sensors, combining edge computing technology for data preprocessing, ensuring the accuracy and real-time performance of the data; using improved Kalman filtering algorithm and dynamic threshold generation algorithm, which can better adapt to complex working conditions and improve the data fusion accuracy; then, through the ARIMA model to predict the future supporting parameter trend, dynamically generate evaluation threshold, and use control chart to analyze the real-time comparison of actual supporting related data and dynamic evaluation threshold, to determine whether to adjust the evaluation standard, so that the evaluation result is more accurate, and the evaluation adaptability is strong, which can adapt to complex geological conditions and stope environment.

[0180] Embodiment two

[0181] Figure 3 The structure schematic diagram of the fully mechanized working face support quality evaluation device is shown in the embodiment of the present application, as shown in Figure 3 The embodiment of the present application provides a fully mechanized working face support quality evaluation device, which comprises:

[0182] The acquisition module 10 is configured to acquire support related data of the fully mechanized working face through the multi-source sensor arranged in the fully mechanized working face;

[0183] The fusion module 20 is configured to fuse the support related data through Kalman filtering algorithm and deep belief network fusion algorithm to obtain fusion data;

[0184] The prediction module 30 is configured to analyze the fusion data through the autoregressive integrated moving average model to predict the support parameters in the future within a preset time to obtain support prediction parameters;

[0185] The determination module 40 is configured to determine the evaluation threshold of the support quality according to the support prediction parameters.

[0186] In some embodiments, the fusion module 20 is further configured to:

[0187] processing the support related data through the Kalman filtering algorithm to output the first state estimation value of the support related data;

[0188] optimizing the first state estimation value and the support related data through the deep belief network fusion algorithm to obtain the second state estimation value.

[0189] In some embodiments, the fusion module 20 is further configured to:

[0190] combining the first state estimation value and the support related data to form an input feature vector;

[0191] inputting the input feature vector into a preset deep belief network to extract a target feature, wherein the target feature comprises at least one of a nonlinear relationship, an implicit pattern, and a potential noise rule;

[0192] generating the second state estimation value according to the target feature.

[0193] In some embodiments, the fusion module 20 is further configured to:

[0194] determining a corresponding state equation and an observation equation according to a sensor type;

[0195] adjusting a Kalman gain parameter according to the monitored support-related data and historical support-related data;

[0196] adjusting the Kalman gain parameter according to support environment data of the fully-mechanized coal mining face, wherein the support environment data comprises geological condition data and stope environment data.

[0197] In some embodiments, the prediction module 30 is further configured to:

[0198] differentially processing the support-related data and historical support-related data to obtain a stationary time series;

[0199] constructing an autoregressive integrated moving average model according to the stationary time series;

[0200] fitting the autoregressive integrated moving average model using the historical support-related data to obtain a fitting result;

[0201] predicting a support parameter in a future preset time according to the fitting result to obtain a support prediction parameter.

[0202] In some embodiments, the determination module 40 is further configured to:

[0203] if the support prediction parameter is higher than a preset support parameter range, increasing an upper limit of the evaluation threshold;

[0204] if the support prediction parameter is lower than the preset support parameter range, decreasing a lower limit of the evaluation threshold.

[0205] In some embodiments, the fully-mechanized coal mining face support quality evaluation device further comprises a control chart analysis module configured to:

[0206] adjusting an evaluation standard of support quality by control chart analysis according to the evaluation threshold of support quality and a deviation of the support-related data.

[0207] In some embodiments, the control chart analysis module is further configured to:

[0208] determine a control limit according to the evaluation threshold of the support quality, wherein the control limit comprises an upper control limit and a lower control limit;

[0209] mark the support-related data as a data point on a control chart;

[0210] determine whether the data point is beyond the control limit;

[0211] if the data point is beyond the control limit, adjust the evaluation criterion;

[0212] if the data point is close to the control limit, fine-tune the evaluation criterion.

[0213] In some embodiments, the control chart analysis module is further configured to:

[0214] if the data points are uniformly distributed, keep the evaluation criterion unchanged.

[0215] The support quality evaluation device for fully-mechanized coal mining face provided by the embodiments of the present application corresponds to the support quality evaluation method for fully-mechanized coal mining face of the above-mentioned embodiments, and any optional item in the support quality evaluation method for fully-mechanized coal mining face is also applicable to the embodiments of the support quality evaluation device for fully-mechanized coal mining face, which will not be described here again.

[0216] Embodiment Three

[0217] The embodiments of the present application also provide an electronic device comprising at least a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned support quality evaluation method for fully-mechanized coal mining face when executing the computer program stored in the memory.

[0218] In some embodiments, the processor executing the computer program can be a processing device comprising one or more general-purpose processing devices, such as a microprocessor, a central processing unit (CPU), a graphics processing unit (GPU), and the like. More specifically, the processor can be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, a processor operating other instruction sets, or a processor operating a combination of instruction sets. The processor can also be one or more special-purpose processing devices, such as an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a digital signal processor (DSP), a system on a chip (SoC), and the like.

[0219] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++, PHP, Python, conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the passenger computer, partly on the passenger computer, as a stand-alone software package, partly on the passenger computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the passenger computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0220] The memory can be read-only memory (ROM), random access memory (RAM), phase change random access memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), electrically erasable programmable read-only memory (EEPROM), other types of random access memory (RAM), flash memory or other forms of flash storage, cache, register, static memory, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes or other magnetic storage devices, or any other possible non-transitory medium that can be used to store information or instructions that can be accessed by a computer device, and the like.

[0221] The electronic device of the embodiments of the present application can include, but is not limited to, fixed terminal devices such as servers, desktop computers, digital TVs, and the like, and mobile terminal devices such as in-vehicle devices, handheld devices (for example, mobile phones, tablet computers, and the like), wearable devices (for example, smart watches, smart bands, and the like), and the like.

[0222] Embodiment Four

[0223] The embodiments of the present application also provide a computer readable storage medium, the computer readable medium stores a computer program, and the computer program is executed by a processor to implement the above-mentioned fully mechanized coal mining face support quality evaluation method.

[0224] The computer readable storage medium of the embodiments of the present application can adopt any combination of one or more computer readable media. The computer readable medium can be a computer readable signal medium or a computer readable storage medium. The computer readable storage medium may, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination of the above. The computer readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, device or apparatus, for example, the memory described above.

[0225] The computer program of the embodiments of the present application can be organized into one or more computer-executable components or modules. Any number and combination of such components or modules can be employed to implement aspects of the present application. For example, aspects of the present application are not limited to the particular computer-executable instructions or particular components or modules described herein and shown in the Figures. Other embodiments can include different computer-executable instructions or components with more or less functionality than that described herein.

[0226] The above description is merely exemplary of the application and the inventive concept described herein. It is understood that the scope of the disclosure is not limited to the particular details described above and that the embodiments of the application described herein are susceptible to modifications and variations. For example, the features described above and other features disclosed herein (but not limited to) can be interchanged among the various embodiments of the application. It is therefore contemplated that the application shall cover any and all modifications and variations of the various features described above and / or illustrated in the drawings. It is understood that the features described above and / or illustrated in the drawings are susceptible to modifications and variations.

[0227] Further, while operations are depicted in a particular order, this should not be understood as requiring performance of such operations in the particular order shown or in sequential order, as some operations can beneficially precede other operations, or be performed in parallel. Also, while each of the above examples is described as a separate embodiment, it will be appreciated that features from different embodiments can be combined, and that the application is intended to encompass each and every combination of the features described herein. Finally, while the above description describes particular examples, it should be understood that the application is not limited to these particular examples, but rather encompasses all features that would be described by the language of the claims.

Claims

1. A method for evaluating the quality of support in fully mechanized mining faces, characterized in that, include: Support-related data of the fully mechanized mining face are collected by multi-source sensors deployed on the fully mechanized mining face; The support-related data are fused using the Kalman filter algorithm and the deep belief network fusion algorithm to obtain fused data; The fused data is analyzed using an autoregressive integral moving average model to predict support parameters within a preset time period, thus obtaining the predicted support parameters. The evaluation threshold for support quality is determined based on the support prediction parameters.

2. The method according to claim 1, characterized in that, The support-related data are fused using the Kalman filter algorithm and the deep belief network fusion algorithm to obtain fused data, including: The support-related data is processed using the Kalman filter algorithm to output the first state estimate of the support-related data; The first state estimate and the support-related data are optimized using a deep belief network fusion algorithm to obtain a second state estimate.

3. The method according to claim 2, characterized in that, The first state estimate and the support-related data are optimized using a deep belief network fusion algorithm to obtain a second state estimate, including: The first state estimate and the support-related data are combined to form an input feature vector; The input feature vector is input into a preset deep belief network to extract target features, wherein the target features include at least one of nonlinear relationships, hidden patterns, and potential noise patterns; The second state estimate is generated based on the target features.

4. The method according to claim 3, characterized in that, The support-related data are processed using the Kalman filter algorithm, including: Determine the corresponding state equation and observation equation based on the sensor type; Adjust the Kalman gain parameters based on the monitored support-related data and historical support-related data; The Kalman gain parameter is adjusted based on the support environment data of the fully mechanized mining face, wherein the support environment data includes geological condition data and mining environment data.

5. The method according to claim 1, characterized in that, The fused data is analyzed using an autoregressive integral moving average model to predict support parameters within a preset time period, resulting in predicted support parameters, including: Differential processing is performed on the support-related data and historical support-related data to obtain a stationary time series; Construct an autoregressive integral moving average model based on the stationary time series; The autoregressive integral moving average model was fitted using the historical support data to obtain the fitting results; Based on the fitting results, the support parameters are predicted within a preset time period to obtain the predicted support parameters.

6. The method according to claim 1, characterized in that, Determining the evaluation threshold for support quality based on the support prediction parameters includes: If the predicted support parameter is higher than the preset support parameter range, the upper limit of the evaluation threshold is increased; If the predicted support parameter is lower than the preset support parameter range, the lower limit of the evaluation threshold is lowered.

7. The method according to claim 1, characterized in that, The method further includes: Based on the evaluation threshold of the support quality and the deviation of the support-related data, the evaluation criteria for support quality are adjusted through control chart analysis.

8. The method according to claim 7, characterized in that, Based on the deviation of the support-related data according to the evaluation threshold of the support quality, the evaluation criteria for support quality are adjusted through control chart analysis, including: The control limits are determined based on the evaluation threshold of the support quality, wherein the control limits include an upper control limit and a lower control limit; The support-related data are marked as data points on the control chart; Determine whether the data point exceeds the control limit; If the data points exceed the control limits, the evaluation criteria will be adjusted. If the data point is close to the control limit, the evaluation criteria are fine-tuned.

9. The method according to claim 8, characterized in that, The method further includes: If the data points are evenly distributed, the evaluation criteria remain unchanged.

10. A device for evaluating the quality of support in fully mechanized mining faces, characterized in that, include: The data acquisition module is configured to acquire support-related data of the fully mechanized mining face through multi-source sensors deployed on the fully mechanized mining face; The fusion module is configured to fuse the support-related data using a Kalman filter algorithm and a deep belief network fusion algorithm to obtain fused data. The prediction module is configured to analyze the fused data using an autoregressive integral moving average model to predict the support parameters within a preset time period, thereby obtaining the predicted support parameters. The determination module is configured to determine the evaluation threshold for support quality based on the support prediction parameters.

Images