Search

CN-121502151-B - Method for recognizing cycle pressure big data of coal face

CN121502151BCN 121502151 BCN121502151 BCN 121502151BCN-121502151-B

Abstract

The invention discloses a method for identifying big data of periodic pressure of a coal face, and belongs to the technical field of coal mine pressure analysis. The method comprises the steps of S1, cleaning support resistance data based on pressure gradient characteristics, eliminating process noise, S2, constructing a nonlinear propulsion model, realizing accurate mapping from time domain data to space thrust progress, S3, partitioning a working surface into areas, dividing the areas into boxes, calculating energy of each area group, constructing a multi-dimensional energy sequence, S4, extracting static load baselines by an anti-noise trend separation method based on quantiles, separating load disturbance characteristics, enhancing by a morphological method, S5, constructing a self-adaptive threshold based on stable statistics, and carrying out consistency verification by combining geological priori period steps to realize automatic and accurate judgment of period pressing. The method solves the problems of incomplete data cleaning, space-time mapping distortion, weak noise resistance, dependence on experience threshold value in judgment and identification and the like in the traditional method, and remarkably improves the accuracy, the robustness and the automation level of the incoming pressure identification.

Inventors

  • WANG KUN
  • LIU XUESHENG
  • BAI SHANGYU
  • FAN DEYUAN
  • HUANG ZHENDI

Assignees

  • 山东科技大学

Dates

Publication Date
20260508
Application Date
20260113

Claims (6)

  1. 1. The method for recognizing the periodical compaction big data of the coal face is characterized by comprising the following steps of: step 1, acquiring and cleaning hydraulic support resistance data; Calculating the empty value missing rate of each measuring point, judging the validity of the measuring point data based on a preset missing rate threshold value, counting the missing rate of the measuring point, eliminating invalid measuring points, and carrying out space-time interpolation restoration on the partial empty value of the valid measuring point to obtain the working resistance data of the hydraulic support after cleaning; Step 2, nonlinear space-time alignment; Acquiring working face daily footage report data, constructing a nonlinear propulsion model, and mapping the cleaned resistance data set from a time dimension to a space propulsion dimension by using the model; step 3, energy aggregation of the subarea space; Dividing the edge trend of the working face into a plurality of independent monitoring areas, carrying out space binning according to a preset propulsion step length, respectively calculating group energy values of each area in each space bin, and constructing a multi-area space energy representation sequence; Step 4, anti-noise trend separation and morphological characteristic enhancement; The method comprises the steps of carrying out filtering pretreatment on the multi-region space energy representing sequence, constructing a static load evolution base line by adopting an anti-noise trend separation method based on quantile estimation, separating periodic load disturbance characteristics from the energy representing sequence, and enhancing the periodic load disturbance characteristics by utilizing a mathematical morphology method to obtain a periodic pressure characteristic enhancement sequence; Step 5, self-adaptive threshold value and cycle consistency are used for judging; And (3) based on the periodic pressing characteristic enhancement sequence obtained in the step (4), constructing a self-adaptive judgment threshold by utilizing a robust statistical method, combining the periodic pressing step range of geological priori, performing space-time logic verification on candidate events exceeding the threshold, and finally identifying and outputting the periodic pressing event and multidimensional state parameters thereof.
  2. 2. The method according to claim 1, wherein in step 1, the low resistance data and the working resistance data generated by the working procedures including moving the frame and lowering the column are separated by using the pressure gradient characteristics, and specifically comprising: Step 1.1, collecting an original data set; Acquiring front and rear column resistance monitoring data of the hydraulic support of the working face by using an electrohydraulic control system of the hydraulic support of the fully mechanized mining face at a set sampling frequency to form an original pressure time matrix; Step 1.2, identifying process characteristics; Calculating a first-order difference absolute value of the hydraulic support at adjacent moments; If the first-order difference absolute value is larger than a preset procedure judging gradient threshold value, judging that the bracket is in a bracket moving or column descending action period in the period, wherein the moment data is procedure interference data and marked as invalid; If the absolute value of the first-order difference is lower than a preset effective pressure lower limit value, judging that the bracket is in an unloading or non-jacking state, wherein the moment data is unloading or non-jacking state data and marked as invalid; Step 1.3, repairing the integrity; In order to prevent information faults caused by data cleaning, a deletion rate threshold R m is introduced; And (3) counting a missing rate threshold R m of the bracket pressure data of the current shift, if R m is more than 0.3, judging that the bracket sensor fails, directly removing the bracket data, and if R m is less than 0.3, filling a partial null value by using an interpolation algorithm, wherein the maximum continuous filling length is limited to 5 sampling points.
  3. 3. The method according to claim 1, characterized in that step 2 comprises the sub-steps of: Step 2.1, obtaining shift footage data; Acquiring a daily report of the coal mining shift of the working face corresponding to the date, reading a daily footage report, and determining the start time and the end time of the shift and the total footage of the shift ; Step 2.2, constructing a non-uniform footage function by adopting a Sigmoid nonlinear space-time mapping model ; (1); Wherein, x is normalized time, x is 0,1, k is slope control factor, and the value range is 5-15; step 2.3, coordinate mapping, namely mapping the working resistance data set cleaned in the step 1 from a time dimension to a space coordinate axis of the accumulated thrust to generate a space-time alignment data set, and defining a space-time coordinate mapping formula as follows: (2); in the formula, The accumulated drag progress coordinate corresponding to the monitoring time t is obtained; the accumulated pushing progress is the accumulated pushing progress at the end of the previous coal mining shift; The total ruler feeding amount is the number of shifts; Is a non-uniform footage function.
  4. 4. The method according to claim 1, characterized in that step 3 comprises the sub-steps of: Step 3.1, space division; Setting the step length of the propulsion degree to be 0.5m, and dividing the accumulated propulsion degree of the whole mine into continuous space boxes; step 3.2, partitioning the working surface; Dividing the working surface into three areas, namely an upper end area Z L , a middle area Z M and a tail area Z R ; step 3.3, energy calculation; For each bin in each region, calculating its population energy value : (3); Wherein, the A group energy value is used for representing the intensity of incoming pressure characteristic of the propulsion position; the total number of valid data points of the area in the current space box is calculated; Is the first in the area An effective work resistance value; Is a summation index; And 3.4, constructing a multi-region spatial energy representation sequence.
  5. 5. The method according to claim 1, characterized in that step 4 comprises the sub-steps of: step 4.1, filtering pretreatment; median filtering is carried out on the multi-region space energy representing sequence; Step 4.2, empirical mode decomposition based on the self-adaptive noise complete set; Based on the adaptive noise complete set empirical mode decomposition, performing multi-scale decomposition on the filtered multi-region spatial energy representation sequence to obtain a plurality of inherent mode components and a residual error item; (4); in the formula, Representing the sequence for energy; Is the first Each natural modal component; is the total number of modal components; is a monotonic trend residual term; Index for sequence number of intrinsic mode component; step 4.3, anti-noise trend separation; construction of static load evolution base line by adopting anti-noise trend separation method based on quantile estimation Specifically, it includes; Step 4.3.1, for the energy representative sequence E (x), calculating an initial estimate of the low quantile background baseline within the sliding window ; (4); In the formula, Q is a low quantile proportion, L is a sliding window length; to be at the propulsion position A population energy value at; step 4.3.2, pair Performing adaptive cascade Gaussian smoothing to obtain continuous static load evolution base line ; Step 4.3.3, the energy representative sequence E (x) is differenced from the base line B (x) to obtain periodic load disturbance characteristics ; (5); Step 4.4, enhancing morphological characteristics; performing feature enhancement by using a mathematical morphology method, specifically performing top hat transformation on the periodic load disturbance feature D (x); (6); in the formula, Is corrosion operation; b is a structural element; The feature enhancement sequence is compressed for a period.
  6. 6. The method according to claim 1, characterized in that step5 comprises the following sub-steps: Step 5.1, monitoring multi-region characteristics and primarily judging a threshold value; periodic based incoming compression feature enhancement sequence Constructing self-adaptive judgment threshold by using robust statistical method ; (7); In the formula, Enhancing sequences for periodic tap features Absolute deviation of bits in the local window; A robust scale estimator constructed based on the median absolute deviation; is a sensitivity adjustment coefficient; is an absolute safety margin term; step 5.2, performing space-time logic verification by combining a geological priori period to press a step distance range, wherein the method specifically comprises the following steps of; setting a theoretical period step interval [ D min , D max ] which accords with the movement rule of the rock stratum of the working face; if the step distance between the candidate incoming call event and the last confirmed incoming call event is smaller than D min , judging that the incoming call event is a false signal and eliminating the false signal; If the continuous pushing distance exceeds D max and no candidate event is generated, carrying out supplementary searching in the overrun interval; step 5.3, calculating and outputting multidimensional characteristic parameters; And 5.2, carrying out parameterization calculation on the pressing event confirmed by the step 5.2, and finally outputting dimension characteristic parameters including a periodical pressing position, a pressing step distance, a pressing intensity level and a pressing region type.

Description

Method for recognizing cycle pressure big data of coal face Technical Field The invention belongs to the technical field of analysis of large data of mine pressure of a coal mine, and particularly relates to a method for identifying large data of periodic pressure of a coal face. Background In the stoping process of the fully mechanized coal face, with the continuous pushing of the coal mining machine, the direct roof collapses, and the basic roof stratum with the hard upper layer can form a cantilever beam or masonry beam structure. When the suspension length reaches the limit, the basic overlying strata is periodically broken, and the process is accompanied by huge elastic energy to be released instantaneously and acts on the hydraulic support in a dynamic load mode, so that the hydraulic support resistance presents a periodic pressure characteristic. In recent years, as resources are gradually depleted, coal mining activities are continuously extended to deep parts, the complexity and stress level of a overlying strata structure are remarkably increased, and the occurrence frequency of roof disasters is in an ascending state. However, the conventional periodical compression judgment method based on manual statistics or an empirical formula is difficult to adapt to the safety production requirement under deep complex geological conditions. Therefore, how to utilize intelligent monitoring means, accurately extract and quantify the periodic step-by-step distance and the step-by-step intensity of the working face from mass monitoring data, and has urgent and practical engineering significance for optimizing supporting parameters and preventing and controlling mining pressure disasters. At present, a multichannel hydraulic support resistance monitoring system is widely applied, and in an actual field, support resistance is heterogeneous, sampling intervals are different, sensors are likely to fall off and are susceptible to extreme value interference, and meanwhile, a propulsion log is asynchronous with shift information and pressure data, so that a stable and reusable cycle pressure automatic judgment method is difficult to form by a traditional method mainly comprising a single sensor, threshold experience or manual inspection. The Chinese patent application with publication number of CN118760916A discloses a fully mechanized mining face pressure analysis early warning method, electronic equipment and medium, the method firstly carries out whole cutter identification and circulation division of a coal mining machine, then extracts the characteristics of the resistance of the circulation end and the like, adopts a threshold method and cluster identification to realize pressure area, finally calculates the step distance and intensity, and uses an index to smoothly predict the next pressure footage and step distance to realize grading early warning. According to the method, the final resistance is defined as the maximum value of the column pressure in a set time period before the hydraulic support column descending action starts, and the accurate recognition cycle number is relied on by the coal mining machine so as to carry out the coming pressure judgment. The method has the following defects that the final resistance value is extremely easy to be distorted due to the influence of manual operations such as frame moving in advance by a bracket worker, the cycle characteristic is completely invalid and cannot reflect the whole process energy of the action of a top plate once the waveform cutting algorithm misjudges the column descending action, the system depends on external hardware, the hardware usually causes data drift due to vibration or accumulated errors, once hardware faults cause the whole cutter identification interruption, the data filtering means mainly focuses on the limitation of sampling frequency and amplitude, a noise depth cleaning mechanism is lacked, and non-working resistance data is easy to mix. The Chinese patent application with publication number of CN117266936A discloses a mining pressure display characteristic monitoring method and device based on stent resistance, the method screens high resistance suspected points by setting a threshold value, screens a set in a defined coordinate system to perform spatial clustering, extracts the centroid abscissa of a pressed cluster, takes the transverse difference value of the centroids of adjacent clusters as the step distance, and thus obtains a step distance sequence and a visual result. The patent application calculates the Euclidean distance between the point pairs by multiplying the opening times of the safety valve by a coefficient as a fixed threshold value and clusters the Euclidean distance to generate a pressure area. The method has the limitation in practical application that effective process information with obvious resistance increasing trend is cleaned by screening data only through a threshold method, and local abnormality is d