Pressure Prediction Method and Device, Storage Medium, and Terminal

Abstract

The present invention discloses a pressure prediction method, device, storage medium, and terminal, which relate to the technical field of mine pressure, and mainly aim to solve the problem of low accuracy of pressure prediction. The method includes: collecting initial stroke data and initial pressure data during the advancement process of a hydraulic support; determining the cumulative moving distance of at least one support moving process according to the initial stroke data, classifying the initial pressure data corresponding to the support moving process, and determining the pressure value of the support moving process; using the trained prediction model to perform prediction processing on the cumulative moving distance, the pressure value, the initial stroke data, and the initial pressure data to obtain at least one target cumulative moving distance and a predicted pressure value corresponding to the target cumulative moving distance. It is mainly used for pressure prediction during the mine mining process.

Description

Technical Field

[0001] The present invention relates to the technical field of mine pressure, and particularly to a pressure prediction method, device, storage medium, and terminal. Background Art

[0002] Before the ore body is mined, the rock mass is in a balanced state. After the ore body is mined, an underground space is formed, which destroys the original stress of the rock mass, causes the stress of the rock mass to redistribute, and seeks a new stress balance. A temporarily balanced rock loosening circle is formed above the roof. At this time, the working face support mainly supports the weight of the rock in the loosening circle. The force caused in the rock mass around the roadway and the mining face during the mining and excavation activities, as well as on the support, is called roof weighting. Roof weighting is closely related to mine safety. If the roof weighting is too large and no pressure relief treatment is carried out, it will easily cause safety accidents such as roof fall and support pressing. By predicting the roof weighting, abnormal early warning and pre-disposal of the roof weighting can be carried out, so as to meet the requirements of mine safety production of prevention first and comprehensive treatment.

[0003] The existing roof weighting prediction methods mainly target the mining equipment and mining environment of medium and high coal seams, and are based on the time cycle law of roof weighting for prediction. However, the error of the roof weighting prediction value will continuously increase over time, and there is no direct connection with the driving depth of comprehensive mining, and secondary calculation is required, resulting in low accuracy, high complexity, and low applicability to the mining environment of roof weighting prediction. Summary of the Invention

[0004] In view of this, the present invention provides a pressure prediction method and device, mainly aiming at the problems of low accuracy, high complexity, and low applicability to the mining environment of the existing roof weighting prediction.

[0005] According to one aspect of the present invention, a pressure prediction method is provided, including:

[0006] Collect the initial stroke data and initial pressure data during the advancement of the hydraulic support;

[0007] Determine the cumulative moving distance of at least one support moving process according to the initial stroke data, and classify the initial pressure data corresponding to the support moving process to determine the pressure value of the support moving process;

[0008] Use the trained prediction model to perform prediction processing on the cumulative moving distance, the pressure value, the initial stroke data, and the initial pressure data to obtain at least one target cumulative moving distance and the predicted pressure value corresponding to the target cumulative moving distance.

[0009] Further, the prediction model includes a first prediction sub-model and a second prediction sub-model. The process of using the trained prediction model to predict the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data to obtain at least one target cumulative support moving distance and the predicted pressure value corresponding to the target cumulative support moving distance includes:

[0010] Using the first prediction sub-model to perform prediction processing on the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data to obtain first prediction data;

[0011] Using the second prediction sub-model to perform cycle, trend, and residual decomposition on the first prediction data to obtain cycle data, trend data, and residual data;

[0012] Using the second prediction sub-model to perform fitting processing on the cycle data, the trend data, and the residual data to obtain at least one target cumulative support moving distance and the predicted pressure value corresponding to the target cumulative support moving distance.

[0013] Further, the second prediction sub-model includes a cycle prediction model, a trend prediction model, and a residual prediction model. The cycle prediction model, the trend prediction model, and the residual prediction model are in a parallel relationship. The process of using the second prediction sub-model to perform fitting processing on the cycle data, the trend data, and the residual data to obtain at least one target cumulative support moving distance and the predicted pressure value corresponding to the target cumulative support moving distance includes:

[0014] Using the cycle prediction model to fit the cycle data to obtain first fitting data;

[0015] Using the trend prediction model to fit the trend data to obtain second fitting data;

[0016] Using the residual prediction model to fit the residual data to obtain third fitting data;

[0017] Superimposing the first fitting data, the second fitting data, and the third fitting data to obtain at least one target cumulative support moving distance and the predicted pressure value corresponding to the target cumulative support moving distance.

[0018] Further, the process of determining the cumulative support moving distance of at least one support moving process according to the initial stroke data includes:

[0019] Performing preprocessing on the initial stroke data, and determining at least one stroke maximum value and one stroke minimum value from the results of the preprocessing. The preprocessing includes filtering processing and duplicate removal processing;

[0020] Calculate the vector distance between the maximum stroke value and the minimum stroke value, and determine at least one stroke combination according to the vector distance, where the stroke combination includes one maximum stroke value and one minimum stroke value;

[0021] Perform cumulative calculation according to the stroke combination to obtain the cumulative support moving distance of at least one support moving process.

[0022] Further, the classifying the initial pressure data corresponding to the support moving process to determine the pressure value of the support moving process includes:

[0023] Classify the initial pressure data according to a preset label to obtain multiple pressure data categories. The preset label corresponds to multiple pressure stages of the support moving process. The multiple stages of the support moving process include the initial support stage, the resistance increasing stage, the constant resistance stage, the resistance decreasing stage, and the abnormal stage;

[0024] Calculate the central values of the maximum data volume category and the minimum data volume category among the multiple pressure data categories, and use the larger value of the central values as the pressure value of the support moving process.

[0025] Further, before collecting the initial stroke data and the initial pressure data of the hydraulic support during the advancing process, the method further includes:

[0026] Construct an initial prediction model including a first prediction sub-model and a second prediction sub-model. The second prediction sub-model includes a periodic prediction model, a trend prediction model, and a residual prediction model. The periodic prediction model, the trend prediction model, and the residual prediction model are in a parallel relationship;

[0027] Train the initial prediction model using the obtained training sample data to obtain a trained prediction model. The training sample data includes stroke sample data and pressure sample data.

[0028] Further, the method further includes:

[0029] Construct an initial prediction model, where the initial prediction model is the first prediction sub-model;

[0030] Train the first prediction sub-model using the obtained training sample data to obtain a trained first prediction sub-model;

[0031] Use the trained first prediction sub-model to perform prediction processing on the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance.

[0032] According to another aspect of the present invention, there is provided a pressure prediction device, comprising:

[0033] A collection module for collecting initial stroke data and initial pressure data during the advancement of a hydraulic support;

[0034] A determination module for determining the cumulative support moving distance during at least one support moving process according to the initial stroke data, classifying the initial pressure data corresponding to the support moving process, and determining the pressure value of the support moving process;

[0035] A prediction module for performing prediction processing on the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data by using a trained prediction model to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance.

[0036] Further, the prediction module includes:

[0037] A first prediction unit for performing prediction processing on the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data by using the first prediction sub-model to obtain first prediction data;

[0038] A decomposition unit for performing cycle, trend, and residual decomposition on the first prediction data by using the second prediction sub-model to obtain cycle data, trend data, and residual data;

[0039] A second prediction unit for performing fitting processing on the cycle data, the trend data, and the residual data by using the second prediction sub-model to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance.

[0040] Further, the second prediction unit is specifically configured to use the cycle prediction model to fit the cycle data to obtain first fitting data; use the trend prediction model to fit the trend data to obtain second fitting data; use the residual prediction model to fit the residual data to obtain third fitting data; and superimpose the first fitting data, the second fitting data, and the third fitting data to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance.

[0041] Further, the determination module includes:

[0042] A preprocessing unit for preprocessing the initial stroke data and determining at least one stroke maximum value and one stroke minimum value from the result of the preprocessing, where the preprocessing includes filtering processing and duplicate removal processing;

[0043] A first calculation unit for calculating a vector distance between the maximum stroke and the minimum stroke, and determining at least one stroke combination according to the vector distance, where the stroke combination includes one of the maximum stroke and one of the minimum stroke;

[0044] A second calculation unit for performing cumulative calculation according to the stroke combination to obtain a cumulative support moving distance of at least one support moving process

[0045] Further, the determination module includes:

[0046] A classification unit for classifying the initial pressure data according to a preset label to obtain a plurality of pressure data categories, where the preset label corresponds to a plurality of pressure stages of the support moving process, and the plurality of stages of the support moving process include an initial support stage, a resistance increasing stage, a constant resistance stage, a resistance decreasing stage, and an abnormal stage;

[0047] A third calculation unit for calculating central values of the maximum data volume category and the minimum data volume category among the plurality of pressure data categories, and taking the larger value of the central values as the pressure value of the support moving process.

[0048] Further, the device further includes:

[0049] A construction module for constructing an initial prediction model including a first prediction sub-model and a second prediction sub-model, where the second prediction sub-model includes a periodic prediction model, a trend prediction model, and a residual prediction model, and the periodic prediction model, the trend prediction model, and the residual prediction model are in a parallel relationship;

[0050] A training module for training the initial prediction model using the obtained training sample data to obtain a trained prediction model, where the training sample data includes stroke sample data and pressure sample data.

[0051] Further, the device further includes:

[0052] The construction module is further configured to construct an initial prediction model, and the initial prediction model is the first prediction sub-model;

[0053] The training module is further configured to train the first prediction sub-model using the obtained training sample data to obtain a trained first prediction sub-model;

[0054] The prediction module is further configured to perform prediction processing on the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data using the trained first prediction sub-model to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance.

[0055] According to another aspect of the present invention, there is provided a storage medium in which at least one executable instruction is stored, and the executable instruction causes a processor to perform operations corresponding to the above pressure prediction method.

[0056] According to still another aspect of the present invention, there is provided a terminal, including: a processor, a memory, a communication interface, and a communication bus, and the processor, the memory, and the communication interface complete communication with each other through the communication bus;

[0057] The memory is used to store at least one executable instruction, and the executable instruction causes the processor to perform operations corresponding to the above pressure prediction method.

[0058] By means of the above technical solutions, the technical solutions provided by the embodiments of the present invention have at least the following advantages:

[0059] The present invention provides a pressure prediction method, device, storage medium, and terminal. In the embodiments of the present invention, initial stroke data and initial pressure data during the advancement of a hydraulic support are collected; the cumulative support moving distance during at least one support moving process is determined according to the initial stroke data, and the initial pressure data corresponding to the support moving process is classified to determine the pressure value of the support moving process; the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data are subjected to prediction processing by using a trained prediction model to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance, establishing a connection between the cumulative support moving distance and the pressure value, fully considering the internal growth trend of the pressure changing with the cumulative support moving distance. At the same time, the secondary calculation of converting time-corresponding pressure to distance-corresponding pressure is avoided, thereby effectively reducing the complexity of pressure prediction and improving the accuracy of pressure prediction.

[0060] The above description is only an overview of the technical solutions of the present invention. In order to be able to understand the technical means of the present invention more clearly, it can be implemented according to the content of the description. And in order to make the above and other purposes, features, and advantages of the present invention more obvious and understandable, the following specifically describes the embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] By reading the detailed description of the preferred embodiments below, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0062] Figure 1 A flowchart of a pressure prediction method provided by an embodiment of the present invention is shown;

[0063] Figure 2 Shows a flowchart of another pressure prediction method provided by an embodiment of the present invention;

[0064] Figure 3 Shows a relationship diagram between the cumulative support moving distance and the predicted pressure value provided by an embodiment of the present invention;

[0065] Figure 4 Shows a flowchart of yet another pressure prediction method provided by an embodiment of the present invention;

[0066] Figure 5 Shows a relationship diagram between the pushing distance and the pressure value provided by an embodiment of the present invention;

[0067] Figure 6 Shows a schematic diagram of the training process of a prediction model provided by an embodiment of the present invention;

[0068] Figure 7 Shows a block diagram of the composition of a pressure prediction device provided by an embodiment of the present invention;

[0069] Figure 8 Shows a schematic diagram of the structure of a terminal provided by an embodiment of the present invention. Detailed implementation manners

[0070] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art.

[0071] Regarding the existing roof weighting prediction method, it mainly aims at the mining equipment and mining environment of medium and high coal seams and makes predictions based on the time cycle law of roof weighting. However, the error of the roof weighting prediction value will continuously increase over time, and it is impossible to establish a direct connection with the driving depth of fully mechanized mining, requiring secondary calculations, resulting in low accuracy, high complexity, and low applicability to the mining environment of roof weighting prediction. An embodiment of the present invention provides a pressure prediction method, as Figure 1 shown, the method includes:

[0072] 101. Collect the initial stroke data and initial pressure data during the advancement of the hydraulic support.

[0073] In the embodiments of the present invention, a hydraulic support is a hydraulic power device that uses liquid pressure to generate a supporting force and realizes automatic relocation for roof support and management. During the process of fully mechanized coal mining, the pressure generated by the surrounding rock of the mining face acts on the hydraulic support in the form of an external load, and the resultant force of each supporting member of the hydraulic support is equal to the resultant force of the external load acting on the hydraulic support by the roof. Among them, the resultant force of the external load acting on the hydraulic support by the roof is also called the roof weighting. The hydraulic support relies on the reciprocating telescopic movement of the internal cylinders to generate a forward thrust, thereby realizing the forward movement of the entire hydraulic support, which is also called moving the support. The time for a cylinder to complete a full telescopic movement is one cylinder telescopic cycle, and the process of the hydraulic support moving forward within one cylinder telescopic cycle is one support moving process. During the process of the hydraulic support moving forward, the telescopic stroke data of the cylinder, that is, the initial stroke data, is collected by the stroke sensor inside the hydraulic support, and the roof weighting data, that is, the initial pressure data, is collected by the pressure sensor.

[0074] It should be noted that both the stroke sensor and the pressure sensor are components of the hydraulic support equipment, which can accurately collect the accurate stroke data and pressure data inside the hydraulic support, and do not require additional data collection equipment. While ensuring the accuracy of the data, it greatly reduces the cost of data collection and pressure prediction. In addition, based on the real-time collected and transmitted telescopic stroke data of the internal cylinders of the hydraulic support and the roof weighting data of the hydraulic support, prediction can be carried out to realize real-time prediction and feedback of the pressure, thereby improving the timeliness of the prediction.

[0075] 102. Determine the cumulative support moving distance of at least one support moving process according to the initial stroke data, and classify the initial pressure data corresponding to the support moving process to determine the pressure value of the support moving process.

[0076] In the embodiments of the present invention, in order to establish the connection between the mining depth and the roof weighting, it is necessary to obtain the cumulative support moving distance of at least one support moving process of the hydraulic support. Among them, the cumulative support moving distance is the cumulative moving distance generated by the current support moving process and the previous support moving processes. For example, if the support moving distance of the first support moving process is 15 and the support moving distance of the second support moving process is 20, then the cumulative support moving distance of the first support moving process is 15, and the cumulative support moving distance of the second support moving process is 35. Since the forward movement of the hydraulic support is completed by the thrust generated during the retraction process of the cylinders, based on the cylinder telescopic stroke data (initial stroke data), the cumulative support moving distance of each support moving process can be determined, and the cumulative support moving distance can be determined by cumulative calculation of the support moving distances.

[0077] In the embodiments of the present invention, after obtaining the cumulative support moving distance of at least one support moving process, it is necessary to determine the pressure value corresponding to each support moving process to establish the relationship between the cumulative support moving distance and the pressure value. Since the pressure value borne by the hydraulic support during a support moving process is constantly changing, and the change frequency of the stroke data is much lower than that of the pressure data, a large amount of initial pressure data will be generated during a support moving process. Therefore, it is necessary to classify the large amount of initial pressure data to extract the most representative pressure value from the large amount of pressure data as the pressure value of each support moving process.

[0078] It should be noted that taking a support moving process as the minimum unit for determining the cumulative support moving distance and the pressure value can make full use of the cylinder telescopic stroke data to obtain an accurate cumulative support moving distance, and can effectively divide the initial stroke data and the initial pressure data, so that the cumulative support moving distance and the pressure value obtained based on the initial stroke data and the initial pressure data form a corresponding relationship, thereby establishing the relationship between the cumulative support moving distance and the pressure value, that is, the relationship between the mining depth and the roof weighting.

[0079] 103. Use the trained prediction model to perform prediction processing on the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data to obtain at least one target cumulative support moving distance and the predicted pressure value corresponding to the target cumulative support moving distance.

[0080] In the embodiments of the present invention, in order to achieve accurate prediction of the pressure, the cumulative support moving distance and the pressure value corresponding to each support moving process are used as representative labeled data, and the initial stroke data and the initial pressure data are used as the process data between the two labeled data and filled between the labeled data. (If there is only the cumulative support moving distance and the pressure value of one support moving process, the initial stroke data and the initial pressure data are filled before the support moving distance and the pressure value) to form a complete [distance, pressure] sequence for input into the prediction model, and the [distance, pressure] sequence is input into the pre-trained pressure prediction model, thereby obtaining the target cumulative support moving distance and the predicted pressure value corresponding to the target cumulative support moving distance. Among them, the predicted value output by the prediction model can be the prediction of the pressure value of the next support moving process, or the prediction of the pressure values of multiple subsequent support moving processes. The embodiments of the present invention do not make specific limitations here. In addition, in order to further reduce the data processing volume, the initial stroke data after filtering and the initial pressure data corresponding to the initial stroke data after filtering can also be used to replace the initial stroke data and the initial pressure data as the input of the prediction model.

[0081] It should be noted that, due to the different hardness of the coal quality in coal seams with different thicknesses, for example, the hardness of extremely thin coal seams is relatively large, resulting in a longer mining process time. If prediction processing is based on the time sequence cycle, the amount of data will be very large, and there will be a large amount of redundant data. Therefore, prediction processing based on distance and travel data can greatly reduce the computational amount of the model, thereby improving the model calculation efficiency. At the same time, the applicability of the pressure prediction method to coal seams with different thicknesses is improved. In addition, by using the cumulative support moving distance and pressure value of each support moving process as the annotation values, and forming a complete [distance, pressure] data sequence with the initial travel data and initial pressure data, it is possible to improve the data accuracy while ensuring the amount of data input to the model, thereby effectively improving the accuracy of pressure prediction.

[0082] In an embodiment of the present invention, for further explanation and limitation, as Figure 2 shown, step 103 of using the trained prediction model to perform prediction processing on the cumulative support moving distance, the pressure value, the initial travel data, and the initial pressure data to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance includes:

[0083] 201. Use the first prediction sub-model to perform prediction processing on the cumulative support moving distance, the pressure value, the initial travel data, and the initial pressure data to obtain first prediction data.

[0084] 202. Use the second prediction sub-model to perform cycle, trend, and residual decomposition on the first prediction data to obtain cycle data, trend data, and residual data.

[0085] 203. Use the second prediction sub-model to perform fitting processing on the cycle data, the trend data, and the residual data to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance.

[0086] In an embodiment of the present invention, the basic model of the first prediction sub-model is an autoregressive integrated moving average model (ARIMA: Autoregressive Integrated Moving Average Model). ARIMA is a statistical model used for time series prediction. The prediction principle is based on the time series difference algorithm on the time axis, and the new data value is predicted according to the difference value of the time series [time, data]. In order to enable the model to perform prediction processing on the [distance, pressure] series that is not a time series, the time axis, time period, and time index attributes of the model are improved. Specifically, the time axis of the time series is changed to a numerical travel axis in the source code, the default time period interval value of 12 for the time period is changed to the training period value of 27 of the current [distance, pressure] series, and the time index attribute in the prediction method is modified to a travel numerical type index. The improved ARIMA is used to perform prediction processing on the [distance, pressure] series, and the predicted results of the target cumulative support moving distance and the pressure value corresponding to the target cumulative support moving distance are obtained, that is, the first prediction data. Moreover, the pressure values corresponding to the cumulative support moving distances in the first test data exhibit periodic characteristics, trend characteristics, and deviation characteristics. Furthermore, the trained second prediction sub-model is used to perform cycle, trend, and residual decomposition on the first prediction data to obtain the weights of each characteristic data, and each characteristic data is respectively fitted to obtain the predicted pressure value corresponding to the target cumulative support moving distance. Among them, the basic model of the first prediction sub-model can also select a Markov algorithm model or a long short-term memory model, and the embodiments of the present invention do not make specific limitations on this.

[0087] It should be noted that using the improved ARIMA model as the first prediction sub-model and performing further prediction processing on the first prediction data obtained based on the first prediction sub-model can effectively avoid overfitting of the prediction model, thereby improving the accuracy of model prediction.

[0088] In an embodiment of the present invention, for further illustration and limitation, step 203 of using the second prediction sub-model to fit the cycle data, the trend data, and the residual data to obtain at least one target cumulative support moving distance and the predicted pressure value corresponding to the target cumulative support moving distance includes:

[0089] Using the cycle prediction model to fit the cycle data to obtain the first fitting data; using the trend prediction model to fit the trend data to obtain the second fitting data; using the residual prediction model to fit the residual data to obtain the third fitting data; superimposing the first fitting data, the second fitting data, and the third fitting data to obtain at least one target cumulative support moving distance and the predicted pressure value corresponding to the target cumulative support moving distance.

[0090] In an embodiment of the present invention, after obtaining the characteristic data representing the period, trend, and residual, in order to process the data in a targeted manner, three parallel predictions are respectively used for fitting processing for different characteristic data. Specifically, the period prediction model selects the improved ARIMA model, and the period processing sub-module in the improved ARIMA model is used to perform period fitting on the period data to obtain more accurate period data, that is, the first fitting data; the trend prediction model selects the polynomial nonlinear regression model, and this model is used to fit the trend data to obtain more accurate trend data, that is, the second fitting data; the residual prediction model selects the improved ARIMA model, and the residual processing sub-module in the improved ARIMA model is used to obtain more accurate residual data, that is, the third fitting data. Further, the first fitting data, the second fitting data, and the third fitting data are added together to obtain the target cumulative support moving distance and the predicted pressure value.

[0091] It should be noted that through practical verification, by decomposing the first prediction data into period, trend, and residual, and respectively performing targeted fitting processing on the characteristic data representing the period characteristics, trend characteristics, and residual characteristics obtained by the decomposition, the accuracy of each characteristic data can be effectively improved, thereby improving the accuracy of the pressure value prediction. In addition, for the convenience of viewing the prediction data, a relationship graph of the target cumulative support moving distance and the predicted pressure value can be generated based on the predicted target cumulative support moving distance and the predicted pressure value, as Figure 3 shown. By visualizing the predicted pressure data, users can more intuitively obtain the change of the predicted pressure value with the movement of the hydraulic support, thereby improving the user experience.

[0092] In an embodiment of the present invention, for further illustration and limitation, as Figure 4 shown, step 102 of determining the cumulative support moving distance of at least one support moving process according to the initial stroke data includes:

[0093] 301. Preprocess the initial stroke data, and determine at least one stroke maximum value and one stroke minimum value from the results of the preprocessing.

[0094] 302. Calculate the vector distance between the stroke maximum value and the stroke minimum value, and determine at least one stroke combination according to the vector distance.

[0095] 303. Perform cumulative calculation according to the stroke combination to obtain the cumulative support moving distance of at least one support moving process.

[0096] In the embodiments of the present invention, since the mining environments corresponding to different coal seam heights are different, different hydraulic support equipment is used. For example, in the mining of medium-high coal seams, the mining space is large and the geological structure is relatively stable, so large four-leg hydraulic supports can be used; while in the mining of extremely thin coal seams, the mining space is small and the geological structure changes greatly, requiring equipment miniaturization, and two-leg hydraulic supports are mostly used. In order to improve the applicability of pressure prediction to different mining environments, the initial stroke data is processed according to the data characteristics during the support moving process in extremely thin coal seams. Since during the support moving process in extremely thin coal seams, due to the limited space of the working face, there are few rib protection plates, and at the same time the geology is hard, the telescopic rigidity of the hydraulic support cylinder is affected, and the phenomenon of springback often occurs, resulting in large fluctuations in the data collected by the sensor. The working face environment of extremely thin coal seams is harsh, affecting the sensor operation, resulting in abnormal data being collected, and the initial stroke data collected by the hydraulic support sensor is discrete stroke data with white noise and has periodic fluctuations. Therefore, it is necessary to filter the initial stroke data to remove the jitter data, periodic fluctuation data and abnormal data. For the jitter data in the initial stroke data, the amplitude-limiting filtering method is adopted to filter out the jitter data lower than the preset threshold. Among them, the preset threshold can be set according to the actual springback experience value of the hydraulic support cylinder, and in order to reduce the amount of data processed each time, the time range corresponding to the filtered target data can be set. The embodiments of the present invention do not make specific limitations on the above preset threshold and preset time range. For the periodic fluctuation data in the initial stroke data, the smooth arithmetic mean filtering method is adopted. First, a data queue with a preset length is set. Each time the old stroke data at the previous moment is removed and the new stroke data at the current moment is added, and then the mean value of the entire current queue is calculated to represent the latest stroke data, which is used to remove the periodic fluctuations of the data. Among them, the preset length of the data queue can be customized according to actual needs, and the embodiments of the present invention do not make specific limitations here. For the abnormal data in the initial stroke data, program filtering is adopted to filter out the abnormal data exceeding the range of the cylinder. The above filtering method can also be replaced by other filtering methods with the same function, and the embodiments of the present invention do not make specific limitations here.

[0097] In the embodiments of the present invention, since the moving distance of the hydraulic support in one moving process is equal to the maximum contraction stroke value in each telescopic stroke cycle of the hydraulic support cylinder, and the maximum contraction stroke value is equal to the difference between the maximum value and the minimum value of the initial stroke data in one telescopic stroke cycle, therefore, the moving distance of the hydraulic support can be calculated based on the maximum value and the minimum value of the initial stroke data in one telescopic stroke cycle, and the cumulative moving distance of this moving process can be obtained by cumulative calculation based on the moving distance in the previous moving process. Here, the difference method is used to calculate the maximum value and the minimum value in the initial stroke data. To avoid the vanishing gradient problem during the calculation of the maximum and minimum values using the difference method, first, the filtered initial stroke data is de-duplicated, and then, for the de-duplicated initial stroke data, a maximum value set and a minimum value set are obtained using the difference method. The vector distance algorithm is used to traverse and calculate the values in the maximum value set and the minimum value set to obtain the vector distance between each maximum value and each minimum value, and a pair of the maximum value and the minimum value with the smallest vector distance is determined as a stroke combination. The maximum value and the minimum value of a stroke combination correspond to the maximum value and the minimum value of the stroke data in one telescopic stroke cycle. Further, the moving distance of one moving process is calculated based on the maximum value and the minimum value of each stroke combination, and the cumulative moving distance of this moving process is obtained by cumulative calculation based on the moving distance in the previous moving process. Among them, the vector distance algorithm can be the Euclidean distance algorithm or the Manhattan distance algorithm, and the embodiments of the present invention do not make specific limitations.

[0098] It should be noted that, based on the characteristics of the pressure data and the stroke data collected during the moving process of the extremely thin coal seam, the initial stroke data is preprocessed, fully considering the limitations and influences of the extreme mining environment on the mining face and the hydraulic support equipment. When applied to the mining process of medium-high coal seams with a larger mining space and more stable geological conditions, it can better ensure the accuracy of pressure prediction, thereby greatly expanding the applicable range of the pressure prediction method while ensuring the accuracy of pressure prediction.

[0099] In an embodiment of the present invention, for further illustration and limitation, as Figure 4 shown, step 102 of classifying the initial pressure data corresponding to the moving process to determine the pressure value of the moving process includes:

[0100] 304. Classify the initial pressure data according to a preset label to obtain multiple pressure data categories.

[0101] 305. Calculate the central values of the maximum data volume category and the minimum data volume category among the multiple pressure data categories, and use the larger value of the central values as the pressure value of the moving process.

[0102] In the embodiments of the present invention, during one support moving process, each stroke data corresponds to multiple pressure data, and the amount of pressure data is relatively large. In order to determine the pressure value during one support moving process, it is necessary to determine a representative pressure data from a large amount of data as the pressure value for the entire support moving process. Since the pressure data generated during the support moving process changes with the pushing distance of the hydraulic support and shows a concentrated change trend, as Figure 5 shown, therefore, a large amount of pressure data is classified and divided, and the pressure value of this support moving process is determined according to multiple classified categories. According to the working principle of the hydraulic support, one support moving process is divided into an initial support stage, a resistance increasing stage, and a constant resistance stage. Considering that there is also a situation of pressure drop and an abnormal situation of sudden roof pressure during this process, the entire support moving process is refined into five stages: an initial support stage, a resistance increasing stage, a constant resistance stage, a resistance decreasing stage, and an abnormal stage. The Gaussian mixture clustering algorithm is used to cluster the initial pressure data according to the above five stages, and the probability of each pressure value belonging to each cluster category is calculated. When new pressure data appears, according to the probability of each pressure value belonging to each cluster category obtained, combined with the Bayesian posterior verification probability formula: P = max{Pinitial support = (initial support cluster | current pressure data), Presistance increasing = (resistance increasing cluster | current pressure data), Pconstant resistance = (constant resistance cluster | current pressure data), Presistance decreasing = (resistance decreasing cluster | current pressure data), Ppressure coming = (pressure coming cluster | current pressure data), Pabnormal = (abnormal cluster | current pressure data)}(1); determine the maximum probability that the current pressure data belongs to a certain cluster, and assign the data to the cluster category corresponding to the maximum probability. After completing the clustering, determine the two cluster categories with the largest and smallest data amounts, and calculate the central values of the cluster category with the largest data amount and the cluster category with the smallest data amount. Since the pressure value prediction is to prevent accidents caused by excessive pressure, the largest pressure data among the central values is determined as the pressure value for one support moving process.

[0103] Among them, in order to further reduce the data processing volume, while filtering and de-duplicating the initial stroke data, the initial pressure data corresponding to the initial stroke data filtered and de-duplicated and removed can be removed together, and the initial stroke data and initial pressure data removed during the de-duplication process are recorded. After calculating the cumulative support moving distance, the initial stroke data and initial pressure data removed by de-duplication are supplemented back into the data for calculating the pressure value of the support moving process. Regarding this, the embodiments of the present invention do not make specific limitations.

[0104] It should be noted that clustering the pressure data can obtain the overall distribution of the pressure quantity, further determine the pressure value from the central values of the largest quantity cluster category and the smallest data quantity category, pay attention to the pressure data with the most concentrated data quantity while also paying attention to the pressure outliers with the smallest data quantity, which can more comprehensively reflect the pressure value during the support moving process, thereby improving the accuracy of pressure value prediction.

[0105] In an embodiment of the present invention, for further illustration and limitation, the method further includes:

[0106] Construct an initial prediction model including a first prediction sub-model and a second prediction sub-model. The second prediction sub-model includes a periodic prediction model, a trend prediction model, and a residual prediction model, and the periodic prediction model, the trend prediction model, and the residual prediction model are in a parallel relationship.

[0107] Use the obtained training sample data to train the initial prediction model to obtain a trained prediction model. The training sample data includes stroke sample data and pressure sample data.

[0108] In the embodiment of the present invention, the stroke sample data and the pressure sample data can be the stroke data and pressure data after establishing a mapping relationship and normalizing and removing the dimension of the cumulative support moving distance and pressure value during multiple support moving processes. Among them, the cumulative support moving distance and the pressure value are determined based on the initial stroke data and the initial pressure data, and the determination process is the same as steps 301-305, which will not be elaborated here in the embodiment of the present invention. Use the stroke sample data and the pressure sample data to train the first prediction sub-model, and use the X-11-ARIMA decomposition method to decompose the prediction result of the first prediction sub-model into periodic data, trend data, and residual data, and use the obtained periodic data, trend data, and residual data to train the periodic prediction model, the trend prediction model, and the residual prediction model respectively, and integrate the trained periodic prediction model, trend prediction model, and residual prediction model to obtain an integrated sub-model, that is, the second prediction sub-model, and then obtain a trained prediction model. Among them, the periodic prediction model, the trend prediction model, and the residual prediction model in the integrated sub-model are in a parallel relationship. And during the actual application process of the prediction model, the second prediction sub-model can perform periodic, trend, and residual decompositions on the first prediction data based on the trained parameters. In addition, based on the cumulative support moving distance and the pressure value, the initial stroke data after filtering and the initial pressure data corresponding to the initial stroke data after filtering can be added as training sample data to train the initial prediction model to increase the amount of training sample data, thereby improving the accuracy of model training.

[0109] It should be noted that when the amount of training sample data is small, the pressure prediction result can be obtained based on the first prediction sub-model. When the amount of training sample data is large, adding a second prediction sub-model on the basis of the first prediction sub-model can effectively improve the generalization ability of the model, thereby improving the accuracy of pressure prediction.

[0110] It should be noted that after the model training and integration are completed, the prediction model can be published through a Web site built by an application programming interface (WebAPI) or packaged as an executable file for external access and retrieval of the pressure prediction result. In addition, after the model is published, the predicted pressure value can also be detected based on the actual pressure value. When the deviation between the predicted pressure value and the actual pressure value exceeds the preset deviation threshold, the prediction model can be updated and trained to improve the accuracy of pressure value prediction. At the same time, the application programming interface provides an interface for manually updating the model parameters, and the parameters of the prediction model can be modified. Among them, the preset threshold can be customized according to actual experience, and the embodiments of the present invention do not make specific limitations.

[0111] In a specific application scenario, the process of training sample data extraction, training the initial prediction model, and service publishing is as Figure 6 shown, including: 1. Perform jitter removal processing and data filtering processing on the stroke data and the corresponding mine pressure data in the support moving production data to obtain the original data. 2. Perform duplicate removal processing on the original data. 3. Calculate the maximum value and the minimum value in the stroke data according to the data after duplicate removal processing, and perform cumulative calculation according to the maximum value and the minimum value to obtain the cumulative support moving distance. 4. Extract the production mine pressure data associated and matched with the support moving distance, and perform outlier processing on the mine pressure data. 5. Backfill the duplicate-removed mine pressure data into the data set to update the data set to obtain the initial training set. 6. Use Gaussian mixture clustering to cluster the pressure data in the initial training set according to the data labels. 7. Use Bayesian screening to obtain the category with the largest data volume and the category with the smallest data volume after clustering. 8. Obtain the single-point mine pressure data (the pressure value corresponding to one support moving process) through numerical screening. 9. Data set update: Map the mine pressure data and the cumulative support moving distance associatively, and perform normalization to remove the dimension to obtain the final training set. 10. Use the original data and the final training set to train the improved ARIMA and perform stability detection. 11. Perform period, trend, and residual difference on the output result of ARIMA. 12. Use the decomposed period data to train the period model, use the decomposed trend data to train the trend model, and use the decomposed residual data to train the residual model. 13. Perform model integration on the trained period model, trend model, and residual model. 14. Perform service publishing on the integrated prediction model. 15. Listen for service requests and process them.

[0112] In an embodiment of the present invention, for further illustration and limitation, the method further includes:

[0113] Construct an initial prediction model, where the initial prediction model is the first prediction sub-model;

[0114] Use the obtained training sample data to train the first prediction sub-model to obtain a trained first prediction sub-model;

[0115] Use the trained first prediction sub-model to perform prediction processing on the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance.

[0116] In the embodiment of the present invention, the first prediction sub-model can be used as an independent prediction model to predict the cumulative support moving distance and the corresponding pressure value. Compared with the prediction model including the first prediction sub-model and the second prediction sub-model, although the prediction accuracy is lower, the model training difficulty and the operation processing workload are both reduced, and it has a higher prediction efficiency, and can be applied to application scenarios where the prediction efficiency requirement is higher than the prediction accuracy requirement.

[0117] The present invention provides a pressure prediction method. In the embodiment of the present invention, the initial stroke data and the initial pressure data during the advancement of the hydraulic support are collected; the cumulative support moving distance of at least one support moving process is determined according to the initial stroke data, and the initial pressure data corresponding to the support moving process is classified to determine the pressure value of the support moving process; the trained prediction model is used to perform prediction processing on the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance, establishing a connection between the cumulative support moving distance and the pressure value, fully considering the internal growth trend of the pressure changing with the cumulative support moving distance. At the same time, the secondary calculation of converting the time-corresponding pressure to the distance-corresponding pressure is avoided, thereby effectively reducing the complexity of pressure prediction and improving the accuracy of pressure prediction.

[0118] Further, as an implementation of the above Figure 1 shown method, the embodiment of the present invention provides a pressure prediction device, as Figure 7 shown, the device includes:

[0119] A collection module 41, configured to collect the initial stroke data and the initial pressure data during the advancement of the hydraulic support;

[0120] A determination module 42, configured to determine the cumulative support moving distance of at least one support moving process according to the initial stroke data, and classify the initial pressure data corresponding to the support moving process to determine the pressure value of the support moving process;

[0121] A prediction module 43, configured to perform prediction processing on the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data by using a trained prediction model to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance.

[0122] Further, the prediction module includes:

[0123] A first prediction unit, configured to perform prediction processing on the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data by using the first prediction sub-model to obtain first prediction data;

[0124] A decomposition unit, configured to perform period, trend, and residual decomposition on the first prediction data by using the second prediction sub-model to obtain period data, trend data, and residual data;

[0125] A second prediction unit, configured to perform fitting processing on the period data, the trend data, and the residual data by using the second prediction sub-model to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance.

[0126] Further, the second prediction unit is specifically configured to perform fitting on the period data by using the period prediction model to obtain first fitting data; perform fitting on the trend data by using the trend prediction model to obtain second fitting data; perform fitting on the residual data by using the residual prediction model to obtain third fitting data; and superimpose the first fitting data, the second fitting data, and the third fitting data to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance.

[0127] Further, the determination module includes:

[0128] A preprocessing unit, configured to preprocess the initial stroke data, and determine at least one stroke maximum value and one stroke minimum value from the result of the preprocessing, where the preprocessing includes filtering processing and duplicate removal processing;

[0129] A first calculation unit, configured to calculate the vector distance between the stroke maximum value and the stroke minimum value, and determine at least one stroke combination according to the vector distance, where the stroke combination includes one stroke maximum value and one stroke minimum value;

[0130] A second calculation unit, configured to perform cumulative calculation according to the stroke combination to obtain the cumulative stroke distance of at least one support moving process.

[0131] Further, the determination module includes:

[0132] A classification unit, configured to classify the initial pressure data according to a preset label to obtain multiple pressure data categories, where the preset label corresponds to multiple pressure stages of the support moving process, and the multiple stages of the support moving process include an initial support stage, a resistance increase stage, a constant resistance stage, a resistance decrease stage, and an abnormal stage;

[0133] A third calculation unit, configured to calculate the central values of the category with the largest data volume and the category with the smallest data volume among the multiple pressure data categories, and use the larger value among the central values as the pressure value of the support moving process.

[0134] Further, the device further includes:

[0135] A construction module, configured to construct an initial prediction model including a first prediction sub-model and a second prediction sub-model, where the second prediction sub-model includes a periodic prediction model, a trend prediction model, and a residual prediction model, and the periodic prediction model, the trend prediction model, and the residual prediction model are in a parallel relationship;

[0136] A training module, configured to train the initial prediction model using the obtained training sample data to obtain a trained prediction model, where the training sample data includes stroke sample data and pressure sample data.

[0137] Further, the device further includes:

[0138] The construction module is further configured to construct an initial prediction model, and the initial prediction model is the first prediction sub-model;

[0139] The training module is further configured to train the first prediction sub-model using the obtained training sample data to obtain a trained first prediction sub-model;

[0140] The prediction module is further configured to perform prediction processing on the cumulative stroke distance, the pressure value, the initial stroke data, and the initial pressure data using the trained first prediction sub-model to obtain at least one target cumulative stroke distance and a predicted pressure value corresponding to the target cumulative stroke distance.

[0141] The present invention provides a pressure prediction device. In an embodiment of the present invention, initial stroke data and initial pressure data during the advancement process of a hydraulic support are collected; the cumulative support moving distance during at least one support moving process is determined according to the initial stroke data, and the initial pressure data corresponding to the support moving process is classified and processed to determine the pressure value during the support moving process; the trained prediction model is used to perform prediction processing on the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance, establishing a connection between the cumulative support moving distance and the pressure value, fully considering the internal growth trend of the pressure changing with the cumulative support moving distance. At the same time, the secondary calculation of converting time-corresponding pressure to distance-corresponding pressure is avoided, thereby effectively reducing the complexity of pressure prediction and improving the accuracy of pressure prediction.

[0142] According to an embodiment of the present invention, a storage medium is provided. The storage medium stores at least one executable instruction, and the computer executable instruction can execute the pressure prediction method in any of the above method embodiments.

[0143] Figure 8 The structural schematic diagram of a terminal provided according to an embodiment of the present invention is shown. The specific implementation of the terminal is not limited in the specific embodiments of the present invention.

[0144] As Figure 8 shown, the terminal may include: a processor 502, a communication interface 504, a memory 506, and a communication bus 508.

[0145] Among them: the processor 502, the communication interface 504, and the memory 506 communicate with each other through the communication bus 508.

[0146] The communication interface 504 is used to communicate with network elements of other devices such as clients or other servers.

[0147] The processor 502 is used to execute the program 510, and specifically can execute the relevant steps in the above pressure prediction method embodiment.

[0148] Specifically, the program 510 may include program codes, and the program codes include computer operation instructions.

[0149] The processor 502 may be a central processing unit (CPU), or a specific application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention. One or more processors included in the terminal may be of the same type of processor, such as one or more CPUs; or they may be of different types of processors, such as one or more CPUs and one or more ASICs.

[0150] A memory 506 for storing a program 510. The memory 506 may include high-speed RAM memory and may also include non-volatile memory, such as at least one disk memory.

[0151] The program 510 may specifically be used to cause the processor 502 to perform the following operations:

[0152] Collect the initial stroke data and the initial pressure data during the advancement process of the hydraulic support;

[0153] Determine the cumulative advancing distance of at least one support moving process according to the initial stroke data, classify the initial pressure data corresponding to the support moving process, and determine the pressure value of the support moving process;

[0154] Use the trained prediction model to perform prediction processing on the cumulative advancing distance, the pressure value, the initial stroke data, and the initial pressure data to obtain at least one target cumulative advancing distance and a predicted pressure value corresponding to the target cumulative advancing distance.

[0155] Obviously, those skilled in the art should understand that the above-mentioned modules or steps of the present invention can be implemented by a general-purpose computing device. They can be concentrated on a single computing device or distributed on a plurality of computing devices. Optionally, they can be implemented by program codes executable by the computing device. Thus, they can be stored in a storage device and executed by the computing device. And in some cases, the steps shown or described can be executed in a different order than here, or they can be separately made into individual integrated circuit modules, or multiple modules or steps among them can be made into a single integrated circuit module to implement. In this way, the present invention is not limited to any specific combination of hardware and software.

[0156] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for pressure prediction, characterized in that, it includes: Collect the initial stroke data and initial pressure data during the advancement of the hydraulic support; Determine the cumulative moving distance of at least one support moving process according to the initial stroke data, classify the initial pressure data corresponding to the support moving process, and determine the pressure value of the support moving process; Use the trained prediction model to perform prediction processing on the cumulative moving distance, the pressure value, the initial stroke data, and the initial pressure data to obtain at least one target cumulative moving distance and the predicted pressure value corresponding to the target cumulative moving distance; Wherein, the prediction model includes a first prediction sub-model and a second prediction sub-model. The step of using the trained prediction model to perform prediction processing on the cumulative moving distance, the pressure value, the initial stroke data, and the initial pressure data to obtain at least one target cumulative moving distance and the predicted pressure value corresponding to the target cumulative moving distance includes: Use the first prediction sub-model to perform prediction processing on the cumulative moving distance, the pressure value, the initial stroke data, and the initial pressure data to obtain first prediction data; Use the second prediction sub-model to perform cycle, trend, and residual decomposition on the first prediction data to obtain cycle data, trend data, and residual data; Use the second prediction sub-model to perform fitting processing on the cycle data, the trend data, and the residual data to obtain at least one target cumulative moving distance and the predicted pressure value corresponding to the target cumulative moving distance output by the second prediction sub-model.

2. The method according to claim 1, characterized in that, The second prediction sub-model includes a cycle prediction model, a trend prediction model, and a residual prediction model. The cycle prediction model, the trend prediction model, and the residual prediction model are in a parallel relationship. The step of using the second prediction sub-model to perform fitting processing on the cycle data, the trend data, and the residual data to obtain at least one target cumulative moving distance and the predicted pressure value corresponding to the target cumulative moving distance includes: Use the cycle prediction model to fit the cycle data to obtain first fitting data; Use the trend prediction model to fit the trend data to obtain second fitting data; Use the residual prediction model to fit the residual data to obtain third fitting data; Superimpose the first fitting data, the second fitting data, and the third fitting data to obtain at least one target cumulative moving distance and the predicted pressure value corresponding to the target cumulative moving distance.

3. The method according to claim 1, characterized in that, The step of determining the cumulative moving distance of at least one support moving process according to the initial stroke data includes: Preprocess the initial stroke data, and determine at least one stroke maximum value and one stroke minimum value from the result of the preprocessing. The preprocessing includes filtering processing and duplicate removal processing; Calculate the vector distance between the maximum stroke value and the minimum stroke value, and determine at least one stroke combination according to the vector distance, where the stroke combination includes one maximum stroke value and one minimum stroke value; Perform cumulative calculation according to the stroke combination to obtain the cumulative support moving distance of at least one support moving process.

4. The method according to claim 1, wherein, the classifying the initial pressure data corresponding to the support moving process to determine the pressure value of the support moving process includes: classifying and processing the initial pressure data according to a preset label to obtain a plurality of pressure data categories, where the preset label corresponds to a plurality of pressure stages of the support moving process, and the plurality of stages of the support moving process include an initial support stage, a resistance increasing stage, a constant resistance stage, a resistance decreasing stage, and an abnormal stage; calculate the central values of the maximum data volume category and the minimum data volume category among the plurality of pressure data categories, and use the larger value of the central values as the pressure value of the support moving process.

5. The method according to claim 1, wherein, before collecting the initial stroke data and the initial pressure data during the advancement of the hydraulic support, the method further includes: construct an initial prediction model including a first prediction sub-model and a second prediction sub-model, where the second prediction sub-model includes a periodic prediction model, a trend prediction model, and a residual prediction model, and the periodic prediction model, the trend prediction model, and the residual prediction model are in a parallel relationship; train the initial prediction model using the obtained training sample data to obtain a trained prediction model, where the training sample data includes stroke sample data and pressure sample data.

6. The method according to claim 1, wherein, the method further includes: construct an initial prediction model, where the initial prediction model is the first prediction sub-model; train the first prediction sub-model using the obtained training sample data to obtain a trained first prediction sub-model; perform prediction processing on the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data using the trained first prediction sub-model to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance.

7. A pressure prediction device, wherein, comprises: a collection module for collecting the initial stroke data and the initial pressure data during the advancement of the hydraulic support; a determination module for determining the cumulative support moving distance of at least one support moving process according to the initial stroke data, and classifying and processing the initial pressure data corresponding to the support moving process to determine the pressure value of the support moving process; a prediction module for performing prediction processing on the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data using a trained prediction model to obtain at least one target cumulative support moving distance and a predicted pressure value corresponding to the target cumulative support moving distance, where the prediction model includes a first prediction sub-model and a second prediction sub-model; the prediction module includes: A first prediction unit, configured to perform prediction processing on the cumulative support moving distance, the pressure value, the initial stroke data, and the initial pressure data by using the first prediction sub-model, so as to obtain first prediction data; A decomposition unit, configured to perform periodic, trend, and residual decomposition on the first prediction data by using the second prediction sub-model, so as to obtain periodic data, trend data, and residual data; A second prediction unit, configured to perform fitting processing on the periodic data, the trend data, and the residual data by using the second prediction sub-model, so as to obtain at least one target cumulative support moving distance output by the second prediction sub-model and a predicted pressure value corresponding to the target cumulative support moving distance; 8. A storage medium, in which at least one executable instruction is stored, and the executable instruction enables a processor to perform operations corresponding to the pressure prediction method according to any one of claims 1-6; 9. A terminal, comprising: a processor, a memory, a communication interface, and a communication bus, where the processor, the memory, and the communication interface complete communication with each other through the communication bus; the memory is configured to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the pressure prediction method according to any one of claims 1-6.

Images