A hydraulic support course angle correction method and system for steeply inclined large mining height working face

Abstract

The present application belongs to the technical field of hydraulic support group control, and particularly relates to a hydraulic support drift angle correction method and system for steeply inclined large-mining-height working face. Anchor points are arranged on the end and chute side, and data such as IMU, stroke / pressure, UWB / laser and conveyor attitude are collected, and a virtual strike reference angle is obtained through robust regression and consistency constraint; the overall deviation is determined according to the strike number and sliding window (same number rate / variance / mean), and the staggered strategy of remainder grouping is adopted to perform continuous small-step shaping, so that the adjacent drift angle deviation RMS is less than or equal to the threshold value, and then the residual deviation is zeroed in the single-frame closed loop, thereby solving the problem that the center line of the conveyor or chute is taken as a single reference datum, and finally leading to the problem of more and more deviation of the drift angle of the whole line support.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of hydraulic support group control, and particularly relates to a hydraulic support azimuth correction method and system for an acute-inclined large-mining-height working face. BACKGROUND

[0002] In a fully-mechanized coal mining face, automatic correction of hydraulic supports is a key technology for ensuring normal operation of a coal mining machine and safety of the working face. Existing automatic correction methods of hydraulic supports usually take the center line of a conveyer or a chute as a single reference datum. However, in the working condition of acute inclination and large mining height, the lateral component of gravity will continuously move the conveyer to the downside, and the chute will also be bent due to extrusion of the support, resulting in deviation of the reference datum. Azimuth correction according to the deviated reference datum will eventually cause the azimuth of the whole line of supports to be adjusted more and more deviated. SUMMARY

[0003] To solve the above technical problems, the application realizes the following technical scheme:

[0004] In a first aspect, a hydraulic support azimuth correction method for an acute-inclined large-mining-height working face is provided, including the following steps: collecting pose deviation data between a support cluster and an anchor point, relative position relationship data between adjacent supports, and working condition state data of the support cluster, and establishing a data set; the anchor point is fixed on the side of the chute and the coal wall near the two ends of the working face; performing robust regression analysis on the data set and applying consistency constraint to obtain a reference angle; the reference angle is used to represent the unified target orientation of the support cluster; sequentially numbering the support cluster along the strike of the working face, and sequentially arranging the working condition state data of the support cluster according to the number to obtain a data sequence; using a first sliding window to calculate the sign consistency ratio, sample variance and mean absolute value of the azimuth deviation on the data sequence; the azimuth deviation is the difference between the reference angle and the azimuth of the support; if the following conditions are met simultaneously: the same sign rate is greater than or equal to a first threshold value, the sample variance is less than or equal to a second threshold value, and the mean absolute value is greater than or equal to a third threshold value, S1 to S3 are executed; otherwise, S3 is executed; S1: the support cluster is divided into multiple groups by modulo operation; S2: for each group, the azimuth of the whole group of supports is adjusted continuously multiple times by using a second sliding window until the root mean square of the azimuth deviation between adjacent supports is less than or equal to a fourth threshold value; S3: the azimuth deviation adjustment value of each support is within the residual error threshold range.

[0005] In a second aspect, a hydraulic support azimuth angle correction system for steeply inclined large mining height working face is provided, comprising: a data acquisition module, configured to acquire pose deviation data between a support cluster and an anchor point, relative position relationship data between adjacent supports, and working condition state data of the support cluster, and establish a data set; the anchor point is fixed on the side of the chute and the coal wall near the two ends of the working face; a first data processing module, configured to perform robust regression analysis on the data set and impose consistency constraints, and obtain a reference angle; the reference angle is used to represent the unified target orientation of the support cluster; a second data processing module, configured to sequentially number the support cluster along the strike of the working face, and sequentially arrange the working condition state data of the support cluster according to the number, and obtain a data sequence; a third data processing module, configured to calculate the sign consistency ratio, sample variance and mean absolute value of the azimuth angle deviation on the data sequence by using a first sliding window; the azimuth angle deviation is the difference between the reference angle and the azimuth angle of the support; a first analysis control module, configured to control a fourth data processing module, a first azimuth angle adjustment module and a second azimuth angle adjustment module to work when the following conditions are met simultaneously: same sign rate ≥ first threshold value, sample variance ≤ second threshold value, and mean absolute value ≥ third threshold value; otherwise, the second azimuth angle adjustment module is controlled to work; the fourth data processing module, configured to group the support cluster by modulo operation to divide it into multiple groups; the first azimuth angle adjustment module, configured to adjust the azimuth angle of the entire group of supports continuously multiple times by using a second sliding window for each group until the root mean square of the azimuth angle deviation between adjacent supports ≤ fourth threshold value; the second azimuth angle adjustment module, configured to adjust the azimuth angle deviation of each support within the residual error threshold range.

[0006] Compared with the prior art, the present application has the following advantages and beneficial effects: by arranging anchor points on the chute side and the two end coal walls and collecting multi-source data such as "support-anchor point, adjacent support, working condition state", a reference angle insensitive to local drift and outliers is obtained by using robust regression and imposing consistency constraints, thereby avoiding misjudgment of the chute gradual change bend as an attitude error from the source; then, based on the first sliding window numbered along the strike, the same sign rate, variance and absolute value of the mean are calculated simultaneously, and the overall deviation is identified by "same sign, high mean, low variance", and it is decided to enter "remainder grouping + continuous small step shaping (S1, S2)", otherwise only single support refinement (S3) is performed, thereby reducing unnecessary group action; in the shaping stage, through staggered grouping and the termination condition of "adjacent azimuth deviation RMS < threshold", the space attitude signal is equivalent to low-pass and second-order smoothing, and the chain propagation and end reflection snake oscillation are suppressed; then in the refinement stage, the azimuth deviation of each support is zeroed one by one, and two-stage smooth convergence is realized; since the action is dispersed and multiple small steps are used, the hydraulic instantaneous flow and pressure impact are significantly reduced, and the energy consumption and valve group wear are simultaneously reduced; at the same time, multi-source fusion + window statistics have fault tolerance to sensing abnormalities, so that the system is more matched and reliable in the scene of "steep inclination + large mining height" with gravity change as the main disturbance, and finally the group shape is smooth, the false triggering is significantly reduced, the production rhythm is smoother, and all support azimuths can be stabilized within the engineering allowable range and significantly reduce resonance and energy consumption. BRIEF DESCRIPTION OF DRAWINGS

[0007] The drawings described herein are used to provide further understanding of the embodiments of the present application, form a part of the present application, and do not constitute a limitation on the embodiments of the present application. In the drawings:

[0008] Figure 1 A hydraulic support cluster structure diagram of a steeply inclined and large mining height working face is provided for the embodiment 1 of the present application;

[0009] Figure 2 A local enlarged view of a single support hydraulic support is provided for the embodiment 1 of the present application;

[0010] Figure 3 A hydraulic support azimuth correction method flowchart of a steeply inclined and large mining height working face is provided for the embodiment 1 of the present application.

[0011] Markings in the drawings and corresponding component names:

[0012] 1-support, 2-first anchor point, 3-stand, 4-traveling jack, 5-chute, 6-top beam, 7-second anchor point, 41-push rod, 42-traveling cylinder. DETAILED DESCRIPTION

[0013] In order to make the objects, technical solutions and advantages of the present application clearer, further detailed description will be made to the present application in combination with embodiments, the illustrative embodiments and the description thereof are only used to explain the present application, and do not limit the present application. The embodiments described below are part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0014] In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present application. However, it will be apparent to one skilled in the art that the present application can be practiced without these specific details. In other embodiments, well-known structures, materials or methods have not been specifically described in order to avoid obscuring the present application. The materials, instruments and reagents used in the following embodiments, etc. can be obtained from commercial channels unless otherwise specified. The technical means used in the embodiments are conventional means known to those skilled in the art unless otherwise specified.

[0015] In addition, the terms "first", "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise specifically limited.

[0016] Embodiment 1: A hydraulic support azimuth angle correction method for steeply inclined large mining height working face is provided, which is suitable for Figure 1 The hydraulic support cluster is shown. Figure 1 In the embodiment, the hydraulic support cluster is arranged along the coal wall and is composed of a plurality of supports 1, and a first anchor point 2 is arranged on the coal wall near the two ends of the working face. Figure 2 It is a local enlarged view of a single support. Figure 2 The support comprises a stand column 3, a push-moving jack 4 composed of a push rod 41 and a push-moving cylinder 42, a chute 5, a top beam 6, and a second anchor point 7 arranged on the side wall of the chute 5.

[0017] The hydraulic support azimuth angle correction method comprises Figure 3 The following steps are shown.

[0018] Step 1: Collecting the pose deviation data between the support cluster and the anchor points, the relative position relationship data between adjacent supports, and the working condition state data of the support cluster, and establishing a data set.

[0019] Among them, the anchor points include: a first anchor point (hereinafter referred to as "end anchor point") and a second anchor point (hereinafter referred to as "light anchor point"). As Figure 1As shown, the end anchor points are fixed on the coal wall near the two ends of the working face, and the light anchor points are fixed on the chute side at intervals.

[0020] The pose deviation data includes: lateral relative position offset and relative heading angle deviation;

[0021] The relative position relationship data includes: relative lateral displacement and interval;

[0022] The working condition state data includes: support heading angle, push jack stroke, push cylinder pressure and column pressure.

[0023] I. End anchor points and light anchor points

[0024] This embodiment adopts the anchor point layout method of "setting end anchor points at both ends + arranging light anchor points in the middle section", which is designed based on three targets of "reference stability requirement, data coverage integrity, and balance between cost and practicality" under steeply inclined conditions. The essence is to solve the problem of easy drift of traditional single reference (such as conveyer) and local data loss through the combination of "fixed strong reference + dynamic auxiliary reference". Specifically:

[0025] (1) One end is arranged on the coal wall near the two ends of the working face, which aims to anchor the "fixed boundary" that cannot be moved in the working face, and provides "absolute reference that does not drift with dynamic equipment" for the whole line support heading angle correction:

[0026] 1. The drift characteristics of dynamic equipment under steeply inclined conditions determine that "there must be a fixed reference". The steeply inclined working face (inclination ≥ 35°) is affected by the lateral component of gravity, and the conveyer will drift 15-20 mm downward each cycle, and the chute will slowly bend due to the extrusion of the support (the monthly bending amount can reach 50-80 mm). If only the conveyer / chute is used as the reference, the support will drift with the reference "more and more deviated" (the traditional technology has a heading angle deviation of more than 5° per week). The reason why the end anchor point is fixed on the coal wall is that the coal wall is the only fixed boundary that does not move dynamically with mining, and its position and strike are basically unchanged during the mining period, which can be used as an "absolute reference" to anchor the target strike of the whole line support (i.e. the physical reference of reference angle θ 0), avoiding the mistake of "judging the drift of the conveyer as the attitude error of the support".

[0027] 2. If only one end anchor point is set, it is easy to cause deviation of the whole line strike due to "single-end reference" (such as the anchor point corresponding to the first support at the end is normal, and the whole line strike will deviate towards the tail end due to the drift of the conveyer at the Nth support at the tail end). Setting one end anchor point at the first support and the Nth support can form a "two-way fixed reference": (1) Through the line connecting the two end anchor points, an "ideal reference line" consistent with the strike of the coal wall is directly fitted, and the reference angle θ0 that is, the line is generated along the line to avoid one-way deviation of single-end reference; (2) when the whole conveyor drifts, the two end anchor points can monitor the drift amount of the two ends of the conveyor (for example, the anchor point corresponding to the first support measures the east end drift of 5 mm, and the anchor point corresponding to the Nth support measures the west end drift of 6 mm), which is judged as "whole conveyor drift" rather than "local deviation of support", and further triggers the dynamic fine adjustment of the reference angle (rather than blind correction of the support), thereby reducing invalid actions.

[0028] 3. The end support (the first and the Nth) needs to be connected to the fixed equipment such as the transfer machine and the crusher, and the lateral stress is more complex (affected by the traction force of the transfer machine and the pushing force of the support), which is easy to become a "heading angle deviation amplification source". The end anchor point adopts a "hydraulic fixed structure" (such as a GY-300 type hydraulic fixed displacement sensor, anchor force ≥ 50 kN), which has higher stability than the middle light anchor point, and can perform "strong constraint calibration" on the heading angle deviation of the end support. For example, when the heading angle of the first support deviates from the reference by 0.8°, the end anchor point can directly provide the "absolute deviation value", avoiding the "chain wave transmission" caused by the transmission of end deviation to the middle section.

[0029] (2) In the middle section of the working face, light anchor points are fixed on the side of the chute at intervals (synchronously move with the chute but are independent of the support), which aims to fill the "local data blank between the two end anchor points", correct the local deviation of the reference angle in the middle section, and avoid the distortion of the whole line posture caused by "two ends fixed and middle section out of control":

[0030] 1. The total length of the steeply inclined large-height working face is usually 150-300 m (corresponding to 100-200 supports), and if only two end anchor points are relied on, the distance between the middle section support (such as the 20th-80th support) and the end anchor point is too far (≥ 50 m), and the reference signal of the end anchor point is easy to be disturbed in the transmission process by "support extrusion and chute bending", which leads to distortion of the reference angle in the middle section (for example, the actual local bending of the chute in the middle section is 2°, but the reference angle fitted by the end anchor point still considers that the middle section is straight). Along the strike, M second anchor points are arranged (such as one every 10-15 supports, M = 8-15), which can realize "one local reference every 30-50 m", real-time monitor the lateral deviation and bending angle of the chute in the middle section, and provide data support for "segment fitting" of the reference angle - for example, the light anchor point at the 50th support in the middle section monitors that the local deviation of the chute is 3 mm, which can correct the reference angle of 0.1° in this area, avoiding the posture distortion caused by the adjustment of the middle section support according to the "distorted reference".

[0031] 2. Under steep tilt conditions, the offset of the transport aircraft is divided into "overall drift" (synchronous lateral movement of the entire line) and "local bending" (local bulge / depression in the middle section due to uneven pressure on the top plate). The two have completely different effects on the reference angle: (1) When there is overall drift, the reference angle needs to be adjusted synchronously to adapt; (2) When there is local bending, the local deviation needs to be retained (to avoid damage to the chute caused by the support being forcibly corrected). The light anchor points can be distinguished by the "relative offset of adjacent light anchor points": if the offset of M light anchor points is "linearly increasing / decreasing" (such as shifting 2mm, 3mm, and 4mm from the east end to the west end respectively), it is judged as "overall drift"; if the offset of two adjacent light anchor points changes abruptly (such as the light anchor point at the 40th frame shifting 2mm and the light anchor point at the 50th frame shifting 5mm), it is judged as "local bending".

[0032] 3. The lightweight anchor points adopt a "lightweight design" (such as the UWB-100 wireless positioning module, weighing ≤1kg, and requiring no damage to the chute structure during installation). The number of M points can be flexibly adjusted according to the length of the working face (e.g., 8 M points for a 150m working face, and 15 M points for a 300m working face). Simultaneously, the lightweight anchor points move synchronously with the chute, reflecting the dynamic changes in the middle section of the chute in real time—for example, if the middle section of the chute shifts 1mm laterally per cycle, the lightweight anchor points can synchronously collect this data, providing middle section data input for the "dynamic smooth update" (first-order low-pass filtering) of the reference angle, avoiding abrupt changes in the reference angle due to "relying solely on end data".

[0033] (iii) The end anchor point and the middle light anchor point do not exist independently, but form a "global-local" collaborative benchmark system through "data fusion" to solve the inherent defects of a single anchor point type.

[0034] 1. The logic for generating the reference angle, which constrains the "global orientation" with end anchor points and corrects "local deviations" with mid-section light anchor points, is as follows: First, the "global ideal orientation" is fitted using two end anchor points. Then, the local data from M mid-section light anchor points are used to "segmentally calibrate" the orientation, ultimately generating a dynamic virtual orientation with "fixed ends and smooth transition in the middle". θ 0. For example: through the positioning and data fitting analysis of the end anchor points, the overall trend deviation is found to be within 1.2°; if the 30th light anchor point in the middle section of the working face detects a local attitude deviation from the reference, the target orientation angle of that area is corrected to 1.3°; if the 70th light anchor point in the middle section detects another local offset, the target orientation angle of the corresponding area is simultaneously adjusted to 1.1°; the final reference angle presents a smooth curve of "1.2°→1.3°→1.1°→1.2°", which not only conforms to the fixed boundary of the end but also adapts to the local dynamics of the middle section, avoiding equipment damage caused by the "rigid alignment" of the entire line support.

[0035] 2. The environment of steeply inclined working face is poor (dust, vibration, impact), and single anchor point is prone to data failure (such as end anchor point sensor failure). The multi-anchor point layout of "end + middle" can form redundancy: if one end anchor point fails, the reference angle can be refitted through another end anchor point + middle light anchor point; if one middle light anchor point fails, the data can be completed by interpolation of adjacent light anchor points; compared with single reference, it can ensure that the heading angle correction can still operate normally when local anchor point fails.

[0036] (Four) Adaptation logic of anchor point number and position

[0037] 1. End anchor point: 2 is the "minimum and optimal" number. One end anchor point cannot determine the direction (only single-end position can be provided, and a straight line cannot be fitted); three or more end anchor points will cause reference conflict due to local unevenness of coal wall (such as the directions fitted by anchor points corresponding to the first support, the second support and the Nth support are inconsistent); two end anchor points can accurately fit the "overall direction of coal wall".

[0038] 2. Middle light anchor point: the value of M is determined according to "coverage density and cost balance". If M is too small (such as M = 3), it is easy to cause insufficient coverage of the middle section, and there is still a data blind area; if M is too large (such as M = 30), it is easy to cause redundancy conflict of multi-anchor point data, usually M is determined according to the density of "one every 10-15 supports" (coverage radius 30-50m) to ensure local data accuracy.

[0039] Two, pose deviation data

[0040] Collect end anchor point / light anchor point pose deviation data to solve the "reference drift" problem and provide reliable reference for heading angle correction.

[0041] Under steeply inclined working conditions, the conveyor / chute will continuously drift due to gravity side shift and support extrusion, and if only dynamic equipment is used as reference, the support will "deviate with the deviation". The core of collecting anchor point pose deviation data is to anchor the absolute boundary of the working face through "fixed anchor point" to generate reference angle that does not drift with the equipment θ 0: 1. End anchor point relative pose reflects the relative deviation of support / conveyor and coal wall fixed boundary, judges whether it is "overall drift of conveyor" (not the deviation of support itself), avoids reference misjudgment; 2. Light anchor point relative pose: fills in the middle section data blind area, distinguishes "overall drift of conveyor" and "local bending" (such as middle chute protrusion), corrects local distortion of reference angle, prevents equipment damage caused by hard correction of "straight line reference" of middle support.

[0042] Three, relative position relationship data of adjacent supports

[0043] Collecting solves the "group wave transmission" problem and ensures the stability of support group cooperation.

[0044] The lateral stiffness of the support on the steeply inclined working face is low (the lateral stiffness decreases by 40% when the mining height is greater than or equal to 6 m), and the adjustment of a single support can easily cause "chain transmission" through lateral force transmission (one support offset drives multiple supports to offset). The core of collecting the relative position relationship data of adjacent supports is to build "group space constraints":

[0045] (1) Real-time monitoring of the transverse misalignment and spacing deviation of adjacent supports to avoid excessive adjustment of a single support causing adjacent supports to squeeze the chute and collide with the coal mining machine.

[0046] (2) Providing data input for distributed wave suppression coordination (such as module 3 grouping and cost function "wave suppression term") to ensure consistent support posture along the line and cut off the wave transmission path.

[0047] Four, support cluster working condition state data

[0048] Collecting support cluster working condition state data solves the "action safety" problem and realizes precise and controllable azimuth adjustment.

[0049] Azimuth correction ultimately relies on the action execution of the push jack, and the working condition state data directly reflects "whether the support can be safely and effectively adjusted":

[0050] (1) Azimuth: judging the deviation of a single support from the reference angle is the core basis for adjustment.

[0051] (2) Stroke: controlling the adjustment amplitude (such as less than or equal to 2 mm per time) to avoid excessive action leading to increased energy consumption and hydraulic oil temperature exceeding the limit.

[0052] (3) Pressure: monitoring the load of the push cylinder / stand to prevent damage to the cylinder body due to excessive pressure (> 30 MPa) or ineffective adjustment due to insufficient pressure (< 15 MPa).

[0053] In summary, in the azimuth correction scheme of the hydraulic support on the steeply inclined and high-height working face, collecting "end anchor point / light anchor point pose deviation data, adjacent support relative position relationship data, and support cluster working condition state data" is essentially to provide data support for building a "dynamic reference system with multiple references", "group coordination space constraints", and "safe action execution guarantee". The three types of data correspond to the technical links of "reference generation-group coordination-action control", and are indispensable.

[0054] Five, sensor selection, installation method and position of the three types of data

[0055] (1) Pose deviation data includes "lateral relative position offset" and "relative azimuth deviation", which requires the cooperation of two types of sensors, as shown in Table 1.

[0056] Table 1 shows the details of the pose deviation data collection device:

[0057] Data type Sensor model / type Installation location Installation method Core parameters (accuracy / range) Relative pose of anchor point GY-300 hydraulic fixed displacement sensor Anchor point (coal wall side) + conveyor east / west end chute 1. Anchor point end: fixed to the coal wall by expansion bolts (depth ≥ 200mm, anchoring force ≥ 50kN); 2. Chute end: weld metal target (thickness 10mm), sensor probe directly opposite the target (distance 50-80mm). Lateral offset: ±2mm / 0-500mm; relative azimuth: ±0.05° / ±180° Relative pose of light anchor point UWB-100 wireless positioning module Middle section chute side (1 every 10-15) + corresponding support canopy beam 1. Chute end: fixed to the chute side ear plate by bolts (avoiding the scraper chain motion track); 2. Support end: adsorbed installation (built-in strong magnet) on the side of the canopy beam (height flush with the chute end module) Lateral offset: ±3mm / 0-300mm; relative azimuth: ±0.1° / ±90°

[0058] (II) Relative position relationship data contains "relative lateral displacement" and "distance", mainly relying on UWB positioning technology, and needs to be collected by two types of sensors, see Table 2.

[0059] Table 2 is a detail table of relative position relationship data collection device:

[0060] Data type Sensor model / type Installation location Installation method Core parameters (accuracy / range) Relative lateral displacement of adjacent supports + distance UWB-100 multi-antenna module Adjacent support canopy beam side (1 each) 1. Support end: fixed to the side of the canopy beam by bolts (1.5m from the ground, avoiding the hydraulic cylinder action area); 2. Two modules are installed directly opposite (distance 1.2-1.5m, no obstruction) Lateral displacement: ±2mm / 0-200mm; distance: ±1mm / 1200-1800mm Auxiliary distance monitoring (end) Laser ranging sensor (LR-200) Front end of the support top beam at the end Support end: bolted to the inside of the front end of the top beam, the probe points to the top beam of the adjacent support (horizontal distance 500-800mm) Distance: ±1mm / 0-1000mm

[0061] (III) Working condition state data contains "heading angle, stroke, pressure", needs to be collected by three types of sensors, see Table 3.

[0062] Table 3 is a detail table of working condition state data collection device:

[0063] Data type Sensor model / type Installation location Installation method Core parameters (accuracy / range) Support azimuth JY901 nine-axis IMU sensor Center position of the support top beam 1. Welded mounting seat (size 100x100mm); 2. Sensor fixed to the mounting seat by bolts, horizontal error ≤0.5° (avoiding inclination leading to azimuth misjudgment) Azimuth: ±0.05° / ±180°; sampling frequency: 10Hz Push jacking jack stroke XC-100 magnetostrictive displacement sensor Outside of the push jacking cylinder 1. Cylinder end: weld sensor fixing sleeve (adapt to cylinder diameter 160mm); 2. Piston rod end: install magnetic ring (synchronous with the piston rod), sensor inserted into the fixing sleeve (coaxiality ≤0.2mm) Stroke: ±0.1%FS / 0-1000mm (FS represents full scale) Push jacking cylinder / stand column pressure YZ-100 pressure sensor Push jacking cylinder oil inlet / stand column lower cavity 1. Hydraulic pipeline end: connected through a tee joint (pressure rating 40MPa); 2. Sensor outlet leads into the support electrical control box through an explosion-proof gland, avoiding oil contamination Pressure: ±0.5%FS / 0-60Mpa (FS represents full scale)

[0064] Step 2: Robust regression analysis is performed on the data set and consistency constraints are applied to obtain the reference angle.

[0065] The reference angle described in the embodiment is an angle parameter obtained by robust regression analysis and applying consistency constraints to multi-source data such as the relative pose of the support cluster and the anchor point, the position relationship between adjacent supports, and the working condition state. Its core role is to define a unified target attitude direction for all hydraulic supports (i.e. support cluster) in the working face that meets the safety production and equipment collaboration requirements. Specifically, it refers to the dynamic virtual trend that all supports need to follow along the working face trend (i.e. the overall direction of coal mining operation advancement), which is adapted to the fixed boundary of the coal wall (anchored by the anchor points at both ends of the coal wall) and the dynamic trend of the chute (corrected by the light anchor point in the middle section of the chute), ensuring that the overall attitude of the support cluster is smooth and there is no significant misalignment, avoiding chain wave or equipment collision caused by single support deviation, and adapting to the slow drift of equipment caused by gravity component in steep inclined working conditions, providing a stable and reliable reference benchmark for subsequent heading angle correction.

[0066] To suppress the influence of outliers and local drift on the estimation of the reference angle, the pose deviation data (lateral relative position offset and relative heading angle deviation) between the support cluster and the anchor point, the relative position relationship data (relative lateral displacement and distance) between adjacent supports, and the working condition state data (support heading angle, stroke of push jacking, and column pressure) of the support cluster are used to solve the unified target orientation using iterative weighted least squares with robust loss. At the same time, consistency constraints are introduced, and physical relationships such as "constant angle between anchor points" and "difference between geometric angle and IMU angle not exceeding tolerance" are written into linear inequalities and checked at each update, making the reference angle insensitive to noise and slow drift, providing a stable reference for group shaping and single support refinement.

[0067] I. Robust regression analysis: resist outliers interference, extract "trend reference" of multi-anchor point data.

[0068] Anchor data of steeply inclined working face is prone to interference (such as instantaneous data jump caused by coal block impact on end anchor, single abnormal value of middle light anchor due to chute vibration), and traditional least squares regression deviates from the true benchmark due to "outliers" (data points with large deviation from normal range). Robust regression reduces the weight of abnormal data, prioritizes fitting the overall trend of the data, and avoids local anomalies dominating the benchmark angle generation. The specific steps are as follows:

[0069] (1) Determine the regression variable and data source.

[0070] Independent variable x: support number (from 1 to N along the working face, such as N=100, x=1,2,...,100), representing the spatial position.

[0071] Dependent variable y: relative lateral displacement of anchor (unit: mm), including two types of data - end anchor data: 2 fixed points (located at support positions x=1 and x=100), recording the lateral displacement between the coal wall anchor and the chute; middle light anchor data: corresponding to M dynamic collection points (for example, 1 point per 10 supports, located at support positions x=10, x=20, …, x=90), recording the lateral displacement between the chute light anchor and the corresponding support.

[0072] (2) Select robust regression algorithm (weighted least squares M estimation) Huber-M estimation with strong anti-outlier ability, the core is to reduce the influence of abnormal data through "dynamic weight function", the specific formula and logic are as follows:

[0073] Initial regression: first use ordinary least squares (OLS) to fit the initial straight line x i , y i ) for all anchor data ( ). Among them, a 0 is the slope, reflecting the trend along the trend; b 0 is the intercept, reflecting the overall deviation.

[0074] Calculate residual and weight: residual (actual deviation and fitted deviation, the larger the residual, the more likely the data is an outlier); the weight function is , where σ=1.4826×MAD (MAD is the median of absolute deviation of residual, used to estimate the standard deviation of data, anti-outlier), k =1.345 (Huber default parameter, ensuring that 95% of normal data weight is 1, and the weight of abnormal data decreases with the increase of residual).

[0075] Weighted iterative regression: use weightw i Re-fitting the straight line, objective function is , repeat iteration until weight change is less than 0.01 (usually 3-5 times convergence).

[0076] (Three) Outlier determination and elimination (auxiliary verification) to the residual data points after iteration (such as the end anchor point appears r i = 15mm, far beyond the normal range ± 5mm), is determined as an outlier and is eliminated, and the regression is re-executed to ensure that the reference angle only reflects the "overall drift of the device / normal trend", rather than local anomalies.

[0077] The process of the above weighted least squares iteration is:

[0078] First, use the full anchor point data (including end anchor points and middle section light anchor points) to do initial ordinary least squares regression to get the "initial fitting line"; then calculate the "residual" of each anchor point data and the fitting line (the difference between the actual offset and the fitting offset) - the larger the residual, the more likely it is that the anchor point data is an outlier (such as a light anchor point with 18mm offset due to vibration, far exceeding the normal range of 5mm, with a significantly large residual). Then assign a weight to each anchor point through a weight function (Huber-M function): normal anchor points with small residuals (such as end anchor points with a residual of 2mm) have a weight = 1 and participate fully in subsequent regression; outliers with large residuals (such as the 18mm offset light anchor point mentioned above) have a weight < 1 (such as 0.2), which significantly reduces their impact on the regression result. Finally, do weighted least squares regression again with "anchor point data x corresponding weight", and repeat iteration until the weight is stable. Through weighted least squares iteration, the interference data for fitting the final reference angle is suppressed, thereby not reducing the amount of fitting data and improving the fitting accuracy (suitable for the case where the number of anchor points on an acute inclined working face is limited).

[0079] Two, consistency constraint

[0080] The data of end anchor points (fixed to the coal wall) and middle section light anchor points (moving with the chute) may conflict (such as the end anchor point showing "overall drift 5mm to the west", but the middle section light anchor point showing "offset 3mm to the east"), and the consistency constraint corrects the data through "physical logic rules" to ensure that the reference angle meets both "end fixed boundary" and "middle section device dynamic" without contradictions. The following are the core:

[0081] (One) Constraint 1: End anchor point "absolute boundary constraint" - force the reference angle to anchor the coal wall.

[0082] End anchor points ( x = 1 and x = Nis the only "fixed boundary" of the working face that does not move with the mining, and the data thereof has the highest priority, and the constraint rule is:

[0083] Let the true value of the lateral offset of the end anchor point be y 1,true , y N,true (the position of the coal wall is fixed, y 1,true , y N,true changes by ≤1mm during the mining period); the subscript "true" represents the meaning of "true value", that is, y 1,true represents the true value of the lateral offset of the end anchor point corresponding to the first support, y N,true represents the true value of the lateral offset of the end anchor point corresponding to the Nth support.

[0084] If the regression result of the middle light anchor point leads to the end offset of the benchmark angle , deviates from the true value y 1,true , y N,true by >3mm (for example, the calculated =8mm, and the true y 1,true =3mm, the intercept of the regression straight line is forcibly corrected b , = y 1,true , = y N,true , and the "two end points" of the benchmark angle are strictly aligned with the fixed boundary of the coal wall, so as to avoid the "overall deviation of the benchmark".

[0085] (2) Constraint 2: Dynamic smoothing constraint of the middle light anchor point - forcibly adapt the benchmark angle to the continuity of the chute

[0086] The chute is a continuous rigid structure, and the lateral offsets of adjacent light anchor points should show "smooth changes" (no abrupt changes), and the constraint rule is:

[0087] For the data of M light anchor points in the middle section y m ( m =1, 2,..., M ), the offset difference of adjacent anchor points is calculated ; if a certain adjacent difference (for example, y 5=3mm, yThe difference of 6mm (6=9mm, far exceeding the normal range of 2mm for chute bending) is judged as "data anomaly" (e.g., loose anchor points). "Linear interpolation" is used to correct the anomaly. (Replace with the mean of the normal points before and after); the corrected data is resubmitted into the robust regression to ensure that the reference angle reflects the "continuous bending trend" of the chute, rather than local abnormal jumps, and to avoid jamming caused by adjusting the middle support according to the "abrupt reference".

[0088] (III) Constraint 3: Angle of Flight - Offset "Physical Association Constraint" - Force the reference angle to conform to mechanical logic

[0089] There is a fixed mechanical relationship between the gantry angle and the lateral relative position offset (between the gantry cluster and the anchor point) (a 1° relative gantry angle deviation corresponds to a lateral relative position offset ≈ L × sin(1°), where L is the length of the gantry top beam; for example, when L = 5m, the corresponding lateral relative position offset is ≈ 87mm; relative gantry angle deviation = reference angle - current gantry angle of a single gantry). The constraint rule is: if the reference angle... θ 0 The calculated "theoretical offset" The "physical offset" calculated from the actual flight angle of the support is "If the deviation is greater than 10mm, the reference angle is deemed unreasonable." θ i For the first i The current flight angle of the support frame; if the reference angle is unreasonable, readjust the weights of the robust regression until the deviation is ≤10mm, ensuring the reference angle. θ 0 It not only fits the anchor point data but also conforms to the physical and mechanical laws of the support posture. Specifically:

[0090] Reference angle θ 0 It serves as the "reference standard" for all subsequent support corrections; if the reference angle... θ 0 If the support itself is off-center, the more you try to correct it, the more chaotic it becomes. The "theoretical offset" is based on a reference angle. θ 0 The calculated "ideal value" and "physical offset" are "true values" calculated based on the actual condition of the stent. The reference angle is determined by comparing the deviation between the two. θ 0 Is it reasonable? For example, the difference between the "theoretical offset of 10mm" and the "physical offset of 25mm" for the 5th support is 15mm, which exceeds 10mm. There are two possibilities: one is the reference angle. θ 0 Fitting errors (such as deviations in anchor point data causing issues with the reference angle) θ 0 (Calculation deviation), and secondly, the reference angle. θ 0Without considering the actual stress state of the support (such as the gravity component of the steeply inclined part deviating from the ideal value), in these two cases, the reference angle θ 0 is identified as unreasonable and cannot be used as a correction reference. If the difference between the theoretical deviation and the physical deviation is ≤10 mm (for example, the theoretical deviation is 10 mm, the physical deviation is 12 mm, and the difference between the two is 2 mm), the reference angle θ 0 both meets the "ideal trend" of the anchor point data and conforms to the "physical state" of the support, and the reference angle θ 0 can be used as a correction reference.

[0091] In the case where the difference between the theoretical deviation and the physical deviation is >10 mm, the weight of the robust regression needs to be adjusted again - since the reference angle θ 0 is fitted based on anchor point data, when it is now found that the reference angle θ 0 deviates from the actual value, it means that the weight distribution of the anchor point data in the weighted least squares fitting is unreasonable, and the weight of the anchor point data needs to be distributed according to the weight function described above, and then the fitting function is used to fit the reference angle θ 0 again by robust regression (the reference angle θ 0 is the horizontal angle of the straight line fitted by all anchor point data) until it is ≤10 mm.

[0092] It should be noted that robust regression and consistency constraint are not independent steps, but a collaborative process of "first anti-interference fitting, then logical correction": robust regression is responsible for "extracting the trend from chaotic data" and solving "outlier interference"; consistency constraint is responsible for "verifying whether the trend is reasonable" and solving "data conflict and physical contradiction"; the final reference angle θ 0 neither deviates from the overall trend due to local abnormal data nor causes "reference and actual boundary contradiction" due to data conflict, providing a "stable, reliable, and logical" reference for subsequent azimuth correction.

[0093] For example: a steeply inclined working face (N=100 supports) collected 2 end anchor points and 8 middle light anchor point data, of which 1 light anchor point had an abnormal value (deviation 18 mm) due to vibration. Through "robust regression to reduce its weight + consistency constraint to correct adjacent data", the final reference angle deviation is only 0.08°, which is much better than the traditional method of 0.7°, ensuring accurate and controllable adjustment of the azimuth angle of the entire line support.

[0094] In addition, in order to realize smooth updating of the reference angle, the embodiment considers that field noise and short-time impact will induce actuator following jitter, and a first-order low-pass filter is used to filter the estimated reference angle, the cut-off frequency is preferably 0.2-0.5 Hz, and the updating coefficient is adaptively calculated according to the sampling period; the formula of the first-order low-pass filter is: ; wherein, is the reference angle obtained after the first-order low-pass filtering, and a is the filtering coefficient, a ∈ (0, 1), is the reference angle obtained after the first-order low-pass filtering of the previous time, is the reference angle obtained at present. This processing retains the slow change trend of the working condition and filters out high-frequency disturbances, so that the reference angle changes more continuously and unnecessary actuation is reduced.

[0095] It should be noted that the reference angle is not calculated once and then fixed, but is dynamically updated with the working condition of the working face - in the steeply inclined working face, the chute will continuously change due to gravity lateral displacement and support extrusion (such as 1-2 mm lateral displacement per cycle), and the anchor point data (offset of the end anchor point and the chute, local offset of the light anchor point) will also be collected and updated in real time over time; after each new anchor point data is collected, the reference angle is recalculated through "robust regression + consistency constraint" to form "a plurality of reference angle values at different time points" (such as 1.2° at the first minute, 1.25° at the second minute, 1.18° at the third minute, …), which are arranged in time sequence to form a "reference angle value sequence". By processing the reference angle value sequence which is dynamically updated over time, the reference angle jump caused by short-time fluctuation of the anchor point data is avoided, and the subsequent azimuth angle correction action is ensured to be stable.

[0096] The reference angle sequence is subjected to first-order low-pass filtering.

[0097] Step 3: sequentially number the support clusters along the working face trend, and arrange the working condition state data of the support clusters in sequence according to the number to obtain a data sequence.

[0098] In order to detect the overall deviation along the working face direction, the present application sequentially numbers the supports according to the trend, and arranges the working condition data into a time-space sequence accordingly; the sequenced data not only provides an index for the sliding window statistics, but also lays a structural foundation for the subsequent staggered adjustment of "modular grouping". Therefore, the spatial problem is converted into a sequence problem which is easy to handle in the algorithm.

[0099] Step 4: using the first sliding window to calculate the same sign rate, sample variance and mean absolute value of the azimuth angle deviation on the data sequence.

[0100] A first sliding window with length of 7, 9 or 11 is set on the sequence, and the same sign rate, sample variance and mean absolute value of the drift angle deviation are calculated window by window. The overall deviation usually shows high same sign rate, low variance and significant non-zero mean; the random disturbance is the opposite. With the help of the three-index combination criterion, the situation that needs to be shaped preferentially can be identified online without interrupting production, and the pertinence of control decision is improved.

[0101] The first sliding window described in the embodiment includes three window functions:

[0102] The first window function is used to calculate the same sign rate of the drift angle deviation;

[0103] The second window function is used to calculate the sample variance of the drift angle deviation;

[0104] The third window function is used to calculate the mean absolute value of the drift angle deviation.

[0105] Taking the first sliding window length = 7 (i.e. single calculation covers the drift angle deviation data of 7 consecutive supports) as an example:

[0106] I. Basic data preparation

[0107] Reference angle θ 0: generated by robust regression and consistency constraint, assuming that in this case θ 0 = 1.2° (dynamic virtual trend); drift angle deviation : the actual drift angle of the i th support θ i The difference between the actual drift angle and the reference angle θ 0, i.e. (the deviation is positive if the support drifts to the right, and negative if it drifts to the left);

[0108] Sliding window: fixed length = 7, sliding along the "support number order" (from the 1st to the Nth), sliding 1 support position each time, covering the deviation data of 7 consecutive supports.

[0109] Actual drift angles of 10 consecutive supports on the working face θ i and corresponding drift angle deviations See Table 4.

[0110] Table 4 is a comparison table of actual drift angles and drift angle deviations:

[0111] Support No. i 1 2 3 4 5 6 7 8 9 10 True heading θ i ]] 1.5 1.4 1.3 1.2 1.3 1.4 1.5 1.6 1.7 1.6 Yaw angle deviation Δ θ i ]]> +0.3 +0.2 +0.1 0 +0.1 +0.2 +0.3 +0.4 +0.5 +0.4

[0112] II. Sliding window calculation demonstration

[0113] Taking the "1st-7th support" as an example:

[0114] The heading angle deviation data in the extraction window: Δ θ = [+0.3, +0.2, +0.1, 0, +0.1, +0.2, +0.3] (unit: °).

[0115] (I) Calculate the same sign rate

[0116] The same sign rate is defined as the ratio of the number of brackets with non-zero deviation and the same sign to the total number of non-zero deviations (deviation of 0 is not counted in the sign statistics because there is no clear bias).

[0117] Step 1: Screen non-zero deviation data and sign.

[0118] Non-zero deviation: +0.3 (positive), +0.2 (positive), +0.1 (positive), +0.1 (positive), +0.2 (positive), +0.3 (positive) → total of 6 non-zero values, all positive.

[0119] Step 2: Calculate the same sign rate:

[0120] The same sign rate = (number of non-zero deviations with the same sign / total number of non-zero deviations) x 100% = (6 / 6) x 100% = 100%.

[0121] The deviations of the first 7 brackets are all positive, indicating that the overall bracket in this section is right, with no left-right alternating offset, which is consistent with the characteristics of "overall drift".

[0122] (II) Calculate the sample variance

[0123] The sample variance is defined as the dispersion degree of the deviation data in the window (the smaller the variance, the more concentrated the deviation, and the more uniform the bracket attitude), and the sample variance calculation formula is: ; Where, S 2 is the sample variance of the heading angle deviation, n is the length of the sliding window, n = 7; represents the heading angle deviation of the k th bracket, is the mean value of the deviation in the window.

[0124] Step 1: Calculate the mean value of the deviation in the window .

[0125] .

[0126] Step 2: Calculate the squared difference between each deviation and the mean value.

[0127] (0.3−0.171) 2 ≈0.0166, (0.2−0.171) 2 ≈0.0008, (0.1−0.171) 2≈0.0050, (0−0.171) 2 ≈0.0292, (0.1−0.171) 2 ≈0.0050, (0.2−0.171) 2 ≈0.0008, (0.3−0.171) 2 ≈0.0166.

[0128] Step 3: Sum and calculate sample variance.

[0129] Sum of squared differences ≈0.0166+0.0008+0.0050+0.0292+0.0050+0.0008+0.0166≈0.074;

[0130] Sample variance S 2 =0.074 / (7−1)≈0.0123.

[0131] The variance is small (≈0.0123), indicating that the 1-7th deviation is concentrated, with no large discrete deviation, and is suitable for subsequent group shaping adjustment.

[0132] (Three) Calculate the absolute value of the mean

[0133] Definition of absolute value of mean: the absolute value of the mean of the deviation in the window, reflecting the "strength of the overall deviation direction" of the segment (the larger the absolute value of the mean, the more significant the overall deviation);

[0134] Calculation process: absolute value of mean = |mean| .

[0135] The absolute value of the mean is approximately 0.171°, indicating that the 1-7th segment is overall right and the deviation is moderate, and needs to be adjusted back to the reference direction through shaping.

[0136] Step 5: If the same rate is greater than or equal to the first threshold, the sample variance is less than or equal to the second threshold, and the absolute value of the mean is greater than or equal to the third threshold, execute S1 to S3; otherwise, execute S3.

[0137] Based on the calculation example of the "1-7th segment": the same rate is 100%> 90%, the variance is 0.0123<0.02, and the absolute value of the mean is 0.171>0.1 → meets the "group shaping condition", triggering S1-S2 (grouping + continuous adjustment); if the window calculation result is: the same rate is 70% (part of the deviation is negative), the variance is 0.03 (the dispersion is large), and the absolute value of the mean is 0.08° (the deviation is weak) → does not meet the condition, directly executes S3 (single segment fine-tuning), avoiding invalid group action.

[0138] Taking a window length of 7 as an example, the calculation process essentially uses "local continuous data statistics" to accurately identify the "overall offset trend" (same number rate), "attitude uniformity" (sample variance), and "offset intensity" (absolute value of mean) of the support group, providing a quantitative basis for the subsequent "whether to perform group shaping" - which avoids misjudgment due to abnormal data of a single support and can capture the collaborative offset characteristics of continuous supports, which is the key to the "dynamic decision-making correction strategy" under steep tilt conditions.

[0139] S1: Divide the support cluster into multiple groups according to the modulo operation.

[0140] In the hydraulic support navigation angle correction scheme for steeply inclined and high-extraction working faces, "dividing the support cluster into multiple groups according to the modal operation" is a prerequisite for realizing "distributed wave suppression coordination". Its purpose is to force "adjacent supports to be in different groups" through mathematical grouping logic, providing a basic framework for subsequent "group token scheduling", and avoiding chain wave transmission and group resonance caused by the synchronous action of adjacent supports from the root.

[0141] I. Core objective: To break the causal chain of "synchronous operation of adjacent supports → wave resonance".

[0142] Hydraulic supports in steeply inclined, high-extraction working faces have low lateral stiffness (lateral stiffness decreases by 40% compared to intermediate-extraction working faces when the extraction height is ≥6m), and the supports are indirectly coupled through chutes and pushing jacks—if adjacent supports (such as the first) i Frame support and the first i When one or more supports move simultaneously, the lateral forces will be superimposed and transmitted, triggering a "chain reaction" (the displacement of the first support causes the subsequent 5th to 8th supports to shift in a chain), and may even trigger resonance at the end, leading to chute damage and coal mining machine collision. The core purpose of "grouping by model calculation" is to force adjacent supports to belong to different groups through mathematical rules, ensuring that adjacent supports will not move synchronously during subsequent "token scheduling" (when only one group of supports moves at the same time), physically cutting off the wave transmission path and avoiding the risk of resonance.

[0143] II. Grouping Principle: Based on the mathematical logic of "modulo N remainder", it ensures that adjacent supports are in different groups.

[0144] "Grouping by Modulus" essentially uses the mathematical concept of "modulo operation." It takes the "physical number of the support structure" as input and groups the data based on the remainder when divided by the group number N. The core principle is that "the remainder of consecutive integers divided by N must be unique." The specific logic is as follows:

[0145] (a) Determine the number of groups N: Select N=3 based on the working conditions.

[0146] (II) Grouping Rules: Let the support cluster be numbered sequentially along the working face as 1, 2, 3, ..., N (total number of supports). Perform a "modulo 3 remainder" operation on the number of each support, and divide the support into groups based on the remainder (0, 1, 2):

[0147] Number divided by 3 leaves a remainder of 1 → Group 1;

[0148] Number divided by 3 leaves a remainder of 2 → Group 2;

[0149] Number ÷ 3 leaves a remainder of 0 (i.e., divisible by 3) → Group 3.

[0150] (III) Core characteristics: Any two adjacent support frames (such as the first one) i Frame support and the first i +1 support frame), whose number is a consecutive integer. The remainders of consecutive integers divided by 3 must be different (for example: the remainder is 2 when i=2, and 0 when i+1=3; the remainder is 2 when i=5, and 0 when i+1=6). Therefore, adjacent supports must belong to different groups, which is the mathematical basis for subsequent wave suppression.

[0151] III. Function: To achieve "peak-shifting operation → wave suppression and energy saving".

[0152] The grouping step itself does not directly perform adjustments, but rather provides a "grouping basis" for subsequent scheduling and control. Its role is reflected in three aspects:

[0153] (a) Suppressing chain propagation: Because adjacent supports are in different groups, the "first" wave will not occur. i frame and the first i In the case of "+1 support moving simultaneously", the lateral force cannot be superimposed and transmitted, completely cutting off the transmission path of "1 support shift → multiple supports shifting in a chain".

[0154] (ii) Avoid group resonance: The end supports of steeply inclined working faces are prone to resonance due to "multiple supports moving in a concentrated manner". After grouping, the number of supports in each group should be moderate (e.g., 100 supports divided into 3 groups, with 33-34 supports in each group). Only one group moves at the same time, reducing the number of supports moving in a concentrated manner and dispersing the resonance energy.

[0155] (III) Balancing the correction efficiency and hydraulic load: If the entire line is operated directly without grouping, the instantaneous flow rate of the hydraulic system will exceed the rated value (e.g., the flow rate required for 100 supports to operate at the same time is 400L / min, while the system is rated at only 200L / min), resulting in a sudden drop in pressure and insufficient adjustment. After grouping, the flow rate requirement for each group to operate is reduced to 130-140L / min (100 supports are divided into 3 groups), matching the system load capacity. At the same time, a full working face cycle is completed every 3 seconds, and the correction response speed meets the requirements of dynamic adjustment of the reference angle.

[0156] For example:

[0157] Assuming that there are 100 hydraulic supports in the working face, numbered from the end to the tail end along the strike as No. 1 (end) → No. 100 (tail end), grouped according to "mod 3 remainder", the specific steps and results are as follows:

[0158] Grouping steps:

[0159] Step 1: Determine the number of groups N = 3, and clarify the remainder correspondence (remainder 1 → Group 1, remainder 2 → Group 2, remainder 0 → Group 3);

[0160] Step 2: Perform "mod 3 remainder" operation on each numbered support (e.g. No. 1 support: 1 ÷ 3 = 0 remainder 1, No. 3 support: 3 ÷ 3 = 1 remainder 0, No. 100 support: 100 ÷ 3 = 33 remainder 1);

[0161] Step 3: Classify according to the remainder, count the number of supports and the number range of each group.

[0162] S2: For each group: use the second sliding window to continuously adjust the azimuth of the entire group of supports for multiple times until the root mean square of the azimuth deviation between adjacent supports is ≤ the fourth threshold value.

[0163] In the "group shaping phase" (S2) of the hydraulic support azimuth correction of the steeply inclined large mining height working face, "use the second sliding window to continuously adjust the azimuth of the entire group of supports for multiple times until the root mean square of the azimuth deviation between adjacent supports is ≤ the fourth threshold value", which is the core operation of "smooth attitude of the entire group of supports" through "local continuous fine tuning + dynamic convergence verification". Its essence is to further refine the action coordination of the supports within a single group on the basis of "mod 3 grouping", to avoid excessive adjustment of local supports within the group causing new wave transmission risks, and at the same time to ensure that the attitude of the entire line of supports finally reaches the convergence standard of "group consistency" through the root mean square (RMS) threshold.

[0164] Specifically, the following steps are included:

[0165] S2.1: Construct the undirected graph of the support cluster and the cost function of the single support; the expression of the cost function is: ; wherein, J i represents the cost function value of the i th support, θ 0 is the reference angle, θ i is the current azimuth of the i th support, w e is the weight of the difference between the reference angle and the current azimuth, θ ij is the azimuth difference between the i th support and the adjacent j th support, i andj The number of the support, N i represents a support cluster, w n The weight of the difference in azimuth angle of adjacent supports.

[0166] In the "group shaping stage" (S2.1) of the hydraulic support azimuth angle correction of the steeply inclined large mining height working face, "constructing the undirected graph of the support cluster and the cost function of the single support" is the core step of converting the "coupling relationship of the physical support group" into a "mathematical solvable model", and its essence is to provide a "relationship description framework" and "optimization decision criteria" for the subsequent "distributed wave suppression coordination", and to solve the "balance between single action and group coordination" problem from the mathematical level.

[0167] I. Purpose: Convert "complex physical coupling" into "mathematical quantifiable model" to solve the "group control difficult modeling" problem.

[0168] The hydraulic support group (usually 100-200 supports) of the steeply inclined large mining height working face is a typical "strongly coupled distributed system":

[0169] (1) Each support is connected through the chute and is laterally extruded to form a coupling relationship (the deviation of one support will affect 3-5 adjacent supports through lateral force transmission, i.e. "chain wave transmission");

[0170] (2) If a control strategy is directly designed for a single support (such as traditional PID), the group coupling effect will be ignored, resulting in "single optimal ≠ group optimal" (such as a single support making a large adjustment to the alignment reference, causing adjacent supports to deviate in a chain); if the entire group is centrally controlled, it will also cause response lag due to insufficient computing power and communication delay (especially when the number of supports exceeds 100).

[0171] (3) Use an undirected graph to describe the "coupling relationship": convert the physically dispersed supports and their mutual influence into a clear "vertex-edge" network model, so that "who is coupled and how strong the coupling is" can be directly quantified.

[0172] (4) Use the cost function to define the "optimization target": integrate the three mutually restrictive control requirements of "alignment reference, wave suppression, and energy consumption reduction" into a single solvable mathematical expression for the support, so that there is a unified decision standard for "how the single support should act"; ultimately achieve the leap from "empirical control to mathematical controllability", and lay the foundation for subsequent calculation of "single optimal action quantity considering group coordination".

[0173] II. Principle: The double logic of "relationship modeling" of the undirected graph and "multi-objective optimization" of the cost function.

[0174] (1) The construction principle of the undirected graph of the support cluster: use "vertex-edge-weight" to describe physical coupling.​

[0175] Mathematical definition of undirected graph is G=(V, E, W), where V is the vertex set, E is the edge set, and W is the edge weight set, which constructs the logical complete matching of the physical characteristics of the support group:

[0176] 1. Vertex v i ∈ V : 1 vertex corresponds to 1 hydraulic support, and the vertex number is consistent with the physical number of the support (for example, 1 corresponds to the first support, 100 corresponds to the 100th support), and the vertex attribute contains the current heading angle of the support v v 100 . θ i .

[0177] 2. Undirected edge e ij ∈ E : If the first support and the second support have physical coupling (usually directly adjacent supports, such as i = j ±1), add an undirected edge between j i v i and v j , representing "bidirectional mutual influence" (the first support deviation will affect the second support, and vice versa); non-adjacent supports (such as i = j ±2) are not added due to weak coupling, avoiding model redundancy. j i

[0178] 3. Undirected edge w ij ∈ W : The core parameter of quantifying the coupling strength, the weight value is positively related to the "physical correlation degree of adjacent supports" - for example: end support (1-2 supports, 99-100 supports) due to the connection of the transfer machine, the lateral constraint is strong, the coupling strength is high, and the weight w ij = w 99,100 =1.5; the middle support (10-90 supports) has weak constraint and low coupling strength, and the weight w ij ==1.0; if the distance between certain adjacent supports is too large (such as more than 1.8m, normally 1.5m), the coupling is weakened, and the weight can be adjusted to 0.8.

[0179] ​​​​​Through this model, the originally abstract "scaffold coupling relationship" is transformed into a computable "graph theory parameter" (such as vertex degree, edge weight), providing a mathematical basis for subsequent identification of transmission path and constraint of adjacent action.

[0180] (II) Construction principle of single-scaffold scaffold cost function: fusion of multi-objective requirements by "weighted sum".

[0181] The expression of the cost function is The core principle is to "transform multi-objective requirements into a single-objective optimization problem", and balance the priority of each objective through weight distribution:

[0182] 1. Reference alignment term : Core objective (weight w e =1.0), highest priority), quantifying the "deviation cost of single-scaffold azimuth from reference angle θ 0" - the greater the deviation, the greater the value of this term, forcing single-scaffold action to approach the reference.

[0183] 2. Wave suppression constraint term : Group objective (weight w n =0.6, second highest priority), where N ( i ) is the set of adjacent vertices in the undirected graph to v i (the adjacent scaffold number of the ith scaffold), is the difference in azimuth angle between adjacent scaffolds; this term quantifies the "cost of single-scaffold adjustment to the transmission risk of adjacent scaffolds" - the greater the difference in azimuth angle, the greater the value of this term, forcing single-scaffold action to consider adjacent scaffold attitude and avoid transmission.

[0184] The essence of the cost function is a "total cost" measure of single-scaffold action, and the minimum value of J i (by gradient descent, least squares, etc. optimization algorithm) can obtain the "single-scaffold optimal action quantity" (adjustment direction and travel) that takes into account the core objective and the group.

[0185] S2.2: Based on the undirected graph, obtain the azimuth adjustment quantity of each scaffold with the minimum cost function value.

[0186] Example of 100-scaffold undirected graph and cost function

[0187] (I) Construction of undirected graph: (100 vertices), E ={ e 1,2 , e 2,3 ,…, e99,100} (99 edges); the weight of the edge between the first vertex and the second vertex is 1.5, the weight of the edge between the last vertex and the second last vertex is 1.5, and the weight of the edge between the rest of the vertices is 1.0.

[0188] (II) Cost function calculation.

[0189] Let the heading adjustment amount of the ith support be , then the adjusted target heading is , which is substituted into the cost function to obtain: . At this time J i becomes a quadratic function about , the image of which is an "open upward parabola", and the minimum value corresponds to the vertex of the parabola, which can be solved by derivation.

[0190] Take the 5th support as an example: it is known that θ 0 = 1.2°, θ 5 = 1.8°, θ 4 = 1.7°, θ 6 = 1.6°;

[0191] Reference alignment term: 1.0 x (1.2-1.8) 2 = 0.36;

[0192] Wave suppression constraint term: 0.6 x [(1.8-1.7) 2 +(1.8-1.6) 2 ]= 0.6 x (0.01+0.04) = 0.03;

[0193] Total cost J 5 = 0.36+0.03 = 0.39.

[0194] (III) Calculate the optimal adjustment amount by "derivation to find extreme value".

[0195] Take the derivative of J i about , and let the derivative equal 0 (extreme value condition) to obtain .

[0196] For example, take the derivative of J i , and let the derivative equal 0 to obtain: . This result directly reflects the balance between the "reference alignment weight w e " and the "wave suppression weight w n ". w eThe larger, the more the adjustment is biased towards the alignment reference angle θ 0 ; w n The larger, the more the adjustment is biased towards the alignment reference angle θ i±1 The larger, the more the adjustment is biased towards the alignment reference angle

[0197] If the algorithm adjusts the azimuth angle θ 5=1.6°, at which J 5=1.0×(0.4) 2 +0.6×[(0.1) 2 +(0.0) 2 ]=0.16+0.006=0.166. The cost function value decreases, both aligning the reference and not interfering with the adjacent frame, achieving "balance between single frame and group".

[0198] In summary, the essence of constructing an undirected graph and a cost function is to build a bridge between "physical support group" and "mathematical control model": the undirected graph solves the problem of "how to describe the coupling relationship", providing a constraint framework for group coordination; the cost function solves the problem of "how to optimize the action of a single frame", providing a quantitative criterion for single frame decision-making; the two work together to ensure that subsequent adjustments meet the "single frame's accurate alignment with the reference" and achieve "group resonance-free", which is the core mathematical innovation that distinguishes this invention from traditional "single frame PID control" and "simple linkage control", and is also the key technical support for stabilizing and correcting the support group of an acute inclined large mining height working face.

[0199] S2.3: Adjust the azimuth angle of each support frame according to the corresponding azimuth angle adjustment amount.

[0200] S2.4: Compare the root mean square of the azimuth angle deviation between adjacent frames with the fourth threshold value, if the root mean square of the azimuth angle deviation between adjacent frames ≤ the fourth threshold value, execute S3, otherwise, return to S2.2.

[0201] The purpose of this step is to ensure the core convergence verification step of "finishing after the group posture meets the standard". Its essence is to judge whether the "group shaping stage has completed the target" through quantitative indicators, avoiding the problem of "single frame alignment but group misplacement" caused by directly entering the fine-tuning without smooth group posture, specifically:

[0202] I. Purpose: To clarify the switching standard of "group shaping → single frame fine-tuning" to avoid confusion of stage goals.

[0203] Under acute inclined conditions, azimuth angle correction is divided into two progressive stages: "group shaping (S2)" and "single frame fine-tuning (S3)", with completely different goals:

[0204] (1) Group shaping (S2): The goal is "rough adjustment" - through grouping adjustment and window fine-tuning, let the whole line support change from "overall deviation and high risk of wave transmission" to "smooth posture and small deviation between adjacent supports" (do not pursue absolute precision of single support heading angle, allow residual deviation, such as 0.1-0.3°).

[0205] (2) Single support fine-tuning (S3): The goal is "fine-tuning" - on the basis of smooth group posture, reduce the residual deviation of single support heading angle from 0.1-0.3° to within 0.1°, and achieve absolute precise alignment.

[0206] If the switching condition of "RMS ≤ fourth threshold value" is not set, two problems may occur:

[0207] 1. Enter S3 too early: Group shaping is not completed (e.g., adjacent deviation RMS = 0.5°, far exceeding the safety value), at this time single support fine-tuning only focuses on its own alignment reference, which may lead to further increase of the heading angle difference between adjacent supports, causing new wave transmission.

[0208] 2. Stay in S2 too long: Group shaping has reached the standard (e.g., RMS ≤ 0.3°), but still continues to adjust repeatedly, leading to increased hydraulic energy consumption (e.g., traditional technology over-adjustment increases energy consumption) and accelerated equipment wear and tear (e.g., service life of push jack seal is shortened).

[0209] Therefore, the core purpose of this step is to set "quantitative switching threshold" for the two stages: only when the group posture meets the conditions of "small deviation between adjacent supports and no risk of wave transmission", can the fine-tuning stage be entered, ensuring that the stage goals are not confused and the adjustment process is efficient and controllable.

[0210] S3: Adjust the heading angle deviation of each support within the residual error threshold range.

[0211] This step is the final execution goal of the entire correction process (corresponding to the core task of S3 single support fine-tuning stage), its essence is to eliminate residual deviation through single support precise fine-tuning on the basis of "smooth posture achieved in group shaping stage (S2)", ensuring that each support is absolutely aligned with the dynamic virtual alignment reference angle θ 0, from "group smoothness" to "single support precision".

[0212] Specifically, the following steps are included:

[0213] S3.1: Obtain the heading angle residual error of each support.

[0214] The heading angle residual error is the difference between the current heading angle and the reference angle.

[0215] S3.2: For each support: judge whether the heading angle residual error is less than or equal to the residual error threshold; if yes, end the support heading angle correction, otherwise, execute S3.3.

[0216] The remaining error threshold is set to 0.2° in this embodiment.

[0217] This step is the "termination decision and iteration trigger" core link of single-frame fine-tuning. Its essence is to accurately control the "start-stop timing" of single-frame adjustment through the quantitative error threshold, ensuring that the heading angle finally meets the standard and avoiding energy waste and equipment risks caused by excessive adjustment.

[0218] The core task of the single-frame fine-tuning stage is to eliminate the "residual heading angle error after group shaping" (usually 0.1°-0.3°), but in actual working conditions, affected by "sensor precision limitation", "hydraulic system hysteresis", and "top plate dynamic interference", it is impossible to achieve "absolute zero" of the residual error of the heading angle (for example, the IMU sensor precision is ±0.05°, theoretically the error cannot be completely eliminated). If there is no "error decision link", two extreme problems may occur:

[0219] 1. Early termination: the adjustment is ended before the error is reduced to a safe range (such as residual 0.2°), which may cause the support posture to deviate from the reference, leading to risks such as collision of the coal mining machine and bending of the chute.

[0220] 2. Infinite iteration: pursuing "absolute zero error", repeatedly fine-tuning (such as error is 0.03° but still adjusting), leading to frequent action of the hydraulic system.

[0221] The core purpose of this step is to set a "quantitative termination threshold" for single-frame fine-tuning (the allowed error range in engineering, not the absolute zero in mathematical sense), and terminate the adjustment when the error meets the "precision and safety" requirements; if not, trigger the next round of fine-tuning (S3.3) to ensure the balance between "precision meeting the standard" and "efficiency controllable".

[0222] S3.3: Adjust the push jack speed, push cylinder pressure, and stroke, and return to S3.1.

[0223] This step is the "core execution link" to eliminate the residual error of the heading angle, and its essence is to accurately control the key parameters of the hydraulic actuator to correct the residual deviation of the support heading angle from "0.1°-0.3°" after group shaping to "engineering zero error", and through "return to S3.1" to form a closed loop verification to ensure that the adjustment effect meets the standard.

[0224] The core contradiction of single-frame fine-tuning is "how to accurately eliminate the residual heading angle error with the least hydraulic action" - if only one parameter is adjusted (such as only the stroke is adjusted), it is easy to cause "insufficient adjustment" (error not eliminated) or "excessive adjustment" (error increases in the opposite direction), and even cause equipment risks. The core purpose of this step can be divided into three points:

[0225] I. Precise elimination of residual error: Through the coordination of "stroke matching error, pressure guaranteeing thrust, and speed control rhythm", it is ensured that each adjustment can "quantitatively eliminate error" (such as 1mm stroke corresponding to 0.1° change of heading angle), avoiding the blindness of traditional "empirical adjustment".

[0226] II. Avoiding hydraulic system risks: Under steep inclination conditions, the support is subjected to complex lateral forces (gravity component + roof pressure), and if the push cylinder pressure is too high, it is easy to cause damage to the cylinder seal; if the speed is too fast, it is easy to cause hydraulic impact; the adjustment parameters can forcibly constrain the system to operate within a safe range.

[0227] III. Dynamic adaptation to changes in working conditions: The roof pressure and hydraulic oil temperature will fluctuate in real time (such as an increase of 10% in pressure after roof caving and an increase of 5℃ in oil temperature), and by adjusting the parameters in real time (such as increasing the pressure in synchronization with the roof pressure), it is ensured that the adjustment effect is not affected by the working conditions, avoiding the problem of "static parameters unable to eliminate differences".

[0228] The specific adjustment method is as follows:

[0229] 1. Pushing jack speed adjustment:

[0230] Implementation logic: According to the size of the heading angle deviation, the speed is dynamically adjusted to avoid the support posture jumping due to too fast speed or affecting the mining efficiency due to too slow speed.

[0231] When the deviation is large (such as θᵢ - θ 0>0.5°, lateral deviation>15mm): "low-speed start→medium-speed push" mode is adopted (such as the speed is gradually increased from 5mm / s to 15mm / s), ensuring that the support posture is slowly corrected and avoiding impact on the jacks;

[0232] When the deviation is small (such as θᵢ - θ 0≤0.2°, lateral deviation≤5mm): "low-speed fine-tuning" mode is adopted (such as the speed is kept at 3-5mm / s), precisely controlling the deviation to prevent over-adjustment.

[0233] 2. Pushing cylinder pressure adjustment:

[0234] Implementation logic: According to the working face roof and floor pressure and chute resistance, the pressure is dynamically adjusted to ensure that the thrust can overcome the load to complete the posture correction and not exceed the rated pressure of the jack.

[0235] When the load is large (such as the chute resistance increases due to coal wall extrusion, and the pressure sensor shows that the current load pressure is 25MPa): the pushing cylinder pressure is set to "load pressure + 2-3MPa" (such as 27MPa), providing sufficient thrust to push the chute and then correct the support heading angle;

[0236] Load hours (such as chute has no obvious resistance, load pressure 15 MPa): set the pressure to "load pressure + 1-2 MPa" (such as 16 MPa), avoid high pressure leading to jack seal damage or chute deformation.

[0237] 3. Pushing stroke adjustment: control "lateral offset correction amount"

[0238] Implementation logic: calculate the required pushing stroke according to the "target lateral offset" (combined with the mechanical correlation formula: stroke ΔL = target lateral offset - current lateral offset / sin(θᵢ-θ0), ensure that the denominator is not 0), avoid insufficient stroke leading to deviation not corrected, or excessive stroke causing new deviation. Among them, the target lateral offset refers to the difference between the actual lateral position and the preset target lateral position in the lateral direction perpendicular to the working face trend.

[0239] Calculation example: if the current lateral offset = 12 mm, the target offset target = 5 mm, the azimuth deviation θᵢ-θ0 = 0.5° (sin 0.5° ≈ 0.0087), then the required stroke ΔL = (5-12) / 0.0087 ≈ -804 mm (negative sign indicates "reverse pushing" i.e. pushing to the coal wall side);

[0240] Stroke constraint: pushing stroke should not exceed the maximum stroke of the jack (such as the maximum stroke of the support jack in the file is 800 mm), if the calculated required stroke exceeds the maximum stroke, it needs to be adjusted in 2-3 cycles (after each adjustment, reacquire the azimuth data and correct the next stroke target).

[0241] Termination condition: when the actual pushing stroke reaches the calculated value within ±1 mm, or the azimuth deviation ≤0.1°, stop pushing. Step 3.4: control the adjustable damper adjuster to adjust the damping coefficient of the working face end support.

[0242] This step is a key reinforcement means to specifically solve the "end support resonance amplification" problem. Its essence is to dynamically adjust the damping coefficient of the end support, consume transmission energy and suppress vibration amplitude, avoid resonance risk caused by "energy convergence" in the end area, and provide "local anti-interference protection" for the stability of the whole line support posture.

[0243] The end support of steeply inclined large mining height working face is a "sensitive area of group resonance", and its special working condition determines that the damping adjustment must be strengthened to suppress the wave capacity. The adjustable damping regulator consumes vibration energy by "oil throttling to generate damping force", and dynamically adjusts the damping coefficient based on the real-time working condition (vibration amplitude, stress state) of the end support, realizing "on-demand wave suppression". The adjustable damping regulator consumes vibration energy by "oil throttling to generate damping force", and dynamically adjusts the damping coefficient based on the real-time working condition (vibration amplitude, stress state) of the end support, realizing "on-demand wave suppression", which contains "valve core, spring, electromagnetic proportional valve" inside. The oil flowing out of the push cylinder needs to pass through the throttle hole of the regulator to return to the oil tank; the electromagnetic proportional valve can control the displacement of the valve core through the current signal to change the cross-sectional area of the throttle hole (the smaller the cross-sectional area, the greater the damping coefficient).

[0244] The specific adjustment method is as follows:

[0245] First step: Real-time acquisition of "stress-pose data" of end support (determine the basis for adjustment)

[0246] Through the deployment of multiple types of sensors on the end support, two types of key data are obtained in real time as the input of damping adjustment:

[0247] Stress data: The side pressure of the support top beam collected by the pressure sensor (such as the extrusion force of the roadway side on the support, unit MPa), the load pressure of the push jack (such as the resistance when pushing the chute, unit MPa);

[0248] Pose data: The current heading angle and lateral offset of the end support collected by the IMU sensor.

[0249] For example: When it is monitored that the side pressure of the first end support reaches 28 MPa (exceeding the normal threshold of 25 MPa), and the lateral offset y1 = 8 mm (approaching the 10 mm warning value), it is determined that the damping needs to be increased to suppress the lateral displacement.

[0250] Second step: Match the "damping coefficient adjustment target" based on the collected data, see Table 5.

[0251] Table 5 is a comparison table of monitoring data characteristics and damping adjustment targets:

[0252] Monitoring data features Damping coefficient adjustment target Large lateral pressure (>25 MPa), fast lateral offset increase (>2 mm / min) Increase damping coefficient (e.g. from 0.6 to 0.8) Small lateral pressure (<15 MPa), lateral offset approaching 0 mm, small deviation of adjacent support heading angle (<0.2°) Reduce damping coefficient (e.g. from 0.6 to 0.4) Large lateral pressure fluctuation (20-28 MPa frequent switching), stable lateral offset (3-5 mm) Maintain damping coefficient (0.6) and enable "dynamic cushioning mode"

[0253] Third step: Execute "damping adjustment" through the adjustable damping regulator.

[0254] The adjustable damping regulator is integrated into the hydraulic control system of the end support (such as the return oil line of the push jack), and adjusts the opening of the internal throttle valve through electro-hydraulic control command to realize precise control of the damping coefficient:

[0255] Hardware operation: the damping adjuster receives the PWM (pulse width modulation) signal of the electro-hydraulic control system to control the opening of the throttle valve. The smaller the opening, the greater the resistance of the hydraulic oil flow, and the greater the damping coefficient; otherwise, the smaller the damping coefficient.

[0256] Example: when the damping needs to be increased, the signal adjusts the throttle valve opening from 50% to 30%, and the damping coefficient is increased from 0.6 to 0.8;

[0257] Precision control: the opening adjustment accuracy is 1%, and the damping coefficient adjustment accuracy is 0.01, which ensures smooth adjustment process without impact.

[0258] Feedback check: continuously monitor the lateral offset acceleration of the end head support after adjustment (such as reducing the offset acceleration from 3 mm / min to 1 mm / min), and the stability of the azimuth angle (the fluctuation is reduced from 0.3° to 0.1°). If the data does not meet the stable target, the damping coefficient needs to be adjusted again (such as reducing the opening by 5%), until the requirements are met.

[0259] In summary, the hydraulic support azimuth correction method for steeply inclined large mining height working face provided by the embodiment is to arrange anchor points on the end and the chute side, collect IMU, stroke / pressure, UWB / laser and conveyor attitude data, and obtain a virtual strike reference angle through robust regression and consistency constraint; determine the overall offset according to the strike number and using a sliding window (same number rate / variance / mean), and use the remainder grouping staggered strategy to perform continuous small step shaping, so that the adjacent azimuth deviation RMS is less than or equal to the threshold value, and then enter the single frame closed loop to zero the residual deviation, thereby solving the problem that the center line of the conveyor or the chute is used as a single reference datum, which eventually leads to the problem of more and more deviation of the azimuth angle of the whole line support.

[0260] Embodiment 2: Corresponding to embodiment 1, the embodiment provides a hydraulic support azimuth correction system for steeply inclined large mining height working face, comprising:

[0261] A data acquisition module is configured to collect pose deviation data between the support cluster and the anchor points, relative position relationship data between adjacent supports, and working condition state data of the support cluster, and establish a data set; the anchor points are fixed on the chute side and the coal wall near the two ends of the working face;

[0262] A first data processing module is configured to perform robust regression analysis on the data set and apply consistency constraint to obtain a reference angle; the reference angle is used to represent the unified target orientation of the support cluster;

[0263] A second data processing module is configured to sequentially number the support cluster along the strike of the working face, and sequentially arrange the working condition state data of the support cluster according to the number to obtain a data sequence;

[0264] The third data processing module is configured to calculate, by using the first sliding window, a proportion of consistent signs, a sample variance and an absolute value of a mean of the yaw angle deviations on the data sequence; the yaw angle deviation is a difference between the reference angle and a yaw angle of the support;

[0265] The first analysis control module is configured to control the fourth data processing module, the first yaw angle adjustment module and the second yaw angle adjustment module to work when the same sign rate is greater than or equal to the first threshold value, the sample variance is less than or equal to the second threshold value and the absolute value of the mean is greater than or equal to the third threshold value, and control the second yaw angle adjustment module to work otherwise.

[0266] The fourth data processing module is configured to divide the support cluster into a plurality of groups by modulo operation.

[0267] The first yaw angle adjustment module is configured to, for each group: continuously adjust the yaw angles of all the supports in the group for multiple times by using the second sliding window until a root mean square of the yaw angle deviations between adjacent supports is less than or equal to the fourth threshold value.

[0268] The second yaw angle adjustment module is configured to adjust the yaw angle deviation of each support within a residual error threshold value range.

[0269] Further, the first yaw angle adjustment module comprises:

[0270] The model construction unit is configured to construct an undirected graph of the support cluster and a cost function of a single support; an expression of the cost function is: ; wherein, J i represents a cost function value of the i th support, i θ 0 is a reference angle, and θ i is a current yaw angle of the i th support, w e is a weight of a difference between the reference angle and the current yaw angle, and θ ij is a yaw angle difference between the i th support and the j th adjacent support, i j is a number of the support, i j represents the support cluster, N i is a number of the support cluster, w n is a weight of the yaw angle difference between the adjacent supports;

[0271] The yaw angle adjustment amount acquisition unit is configured to acquire, based on the undirected graph, a yaw angle adjustment amount with which the cost function value of each support is minimum.

[0272] The yaw angle adjustment unit is configured to control each support to adjust the yaw angle according to the corresponding yaw angle adjustment amount.

[0273] ​​​​The first analysis control unit is configured to compare the root mean square of the heading angle deviation between adjacent supports with a fourth threshold value, and if the root mean square of the heading angle deviation between adjacent supports is less than or equal to the fourth threshold value, control the second heading angle adjustment module to work, otherwise, control the heading angle adjustment amount acquisition unit to work.

[0274] Further, the second heading angle adjustment module comprises:

[0275] The heading angle residual error acquisition unit is configured to acquire the heading angle residual error of each support; the heading angle residual error is the difference between the current heading angle and the reference angle;

[0276] The second analysis control unit is configured to, for each support: determine whether the heading angle residual error is less than or equal to the residual error threshold value; if yes, end the support heading angle correction, otherwise, drive the comprehensive control unit to work;

[0277] The comprehensive control unit is configured to adjust the push jack speed, the push cylinder pressure and the stroke, and control the heading angle residual error acquisition unit to work.

[0278] Further, the hydraulic support heading angle correction system for the steeply inclined large-mining-height working face further comprises a damping adjuster control unit configured to control the adjustable damping adjuster to adjust the damping coefficient of the supports at the two ends of the working face.

[0279] Further, the hydraulic support heading angle correction system for the steeply inclined large-mining-height working face further comprises a low-pass filtering module configured to perform first-order low-pass filtering on the reference angle sequence; the formula of the first-order low-pass filtering is: ; wherein, is the reference angle obtained after the first-order low-pass filtering, a is a filtering coefficient, a ∈ (0, 1), is the reference angle obtained after the first-order low-pass filtering in the previous time, is the reference angle obtained at present.

[0280] It should be understood that the "system", "device", "unit" and / or "module" used in the specification is a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

[0281] As shown in the specification and claims, unless the context clearly indicates otherwise, the words "one", "a", "an" and / or "the" do not refer to the singular, but also include the plural. Generally, the terms "include" and "contain" only indicate the inclusion of the steps and elements explicitly identified, and these steps and elements do not constitute an exclusive list, and the method or device can also include other steps or elements.

[0282] The above detailed description of the specific implementation, the purpose, technical solutions and beneficial effects of the present application are further described in detail, it should be understood that the above description is only a specific embodiment of the present application, and is not used to limit the protection scope of the present application, any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application should be included in the protection scope of the present application.

[0283] It should be noted that the structure, proportion, size, etc. shown in the drawings attached to the present specification are only used to cooperate with the content disclosed in the specification, so that those skilled in the art can understand and read, and are not used to limit the implementation conditions of the present application, so they do not have technical significance. Any modification of structure, change of proportion relationship or adjustment of size, without affecting the effect and purpose that can be achieved by the present application, should still fall within the scope of the technical content disclosed by the present application. At the same time, the terms such as "upper", "lower", "left", "right", "middle" and the like in the present specification are only for the convenience of clear description, and are not used to limit the implementation scope of the present application. The change or adjustment of the relative relationship is also considered as the implementation range of the present application without substantial changes in technical content.

Claims

1. A hydraulic support azimuth angle correction method for steeply inclined large mining height working face, characterized in that, The method comprises the following steps: Collecting pose deviation data between the support cluster and the anchor point, relative position relationship data between adjacent supports, and working condition state data of the support cluster, and establishing a data set; Performing robust regression analysis on the data set and applying consistency constraints to obtain a reference angle; The reference angle is used to represent the unified target orientation of the support cluster; The support cluster is sequentially numbered along the working face trend, and the working condition state data of the support cluster is sequentially arranged according to the number to obtain a data sequence; The sign consistency ratio, sample variance and mean absolute value of the drift angle deviation are calculated on the data sequence by using a first sliding window; the drift angle deviation is the difference between the reference angle and the drift angle of the support; If the same sign rate is greater than or equal to a first threshold value, the sample variance is less than or equal to a second threshold value, and the mean absolute value is greater than or equal to a third threshold value, S1 to S3 are executed; Otherwise, S3 is executed; S1: The support cluster is divided into multiple groups by modulo operation; S2: For each group, the drift angle of the entire group of supports is adjusted continuously multiple times by using a second sliding window until the root mean square of the drift angle deviation between adjacent supports is less than or equal to a fourth threshold value; S3: The drift angle deviation adjustment value of each support is within a residual error threshold range; S2 comprises: S2.1: Constructing an undirected graph of the cluster of struts and a cost function for a single strut; the cost function is expressed as: ; wherein, J i represents the cost function value of the i th strut, θ 0 is the reference angle, i is the current yaw angle of the i th strut, w e is the weight of the difference between the reference angle and the current yaw angle, ij is the yaw angle difference between the i th strut and the adjacent j th strut, i and j is the number of the strut, N i represents the cluster of struts, w n is the weight of the yaw angle difference of the adjacent struts;​ S2.2: Obtain the drift angle adjustment amount of each support based on the minimum cost function value of the undirected graph; S2.3: Control each support to adjust the drift angle according to the corresponding drift angle adjustment amount; S2.4: Compare the root mean square of the drift angle deviation between adjacent supports with the fourth threshold value, if the root mean square of the drift angle deviation between adjacent supports is less than or equal to the fourth threshold value, S3 is executed, otherwise, return to S2.

2.

2. The hydraulic support azimuth angle correction method for steeply inclined large mining height working face according to claim 1, characterized in that the anchor point Comprise: A first anchor point and a second anchor point; The first anchor point is fixed on the coal wall near the two ends of the working face; The second anchor point is fixed on the side of the chute at intervals; The pose deviation data includes: lateral relative position offset and relative drift angle deviation; The relative position relationship data includes: relative lateral displacement and interval; The working condition state data includes: drift angle, stroke and pressure.

3. The hydraulic support azimuth angle correction method for steeply inclined large mining height working face according to claim 1, characterized in that S3 Comprise: S3.1: Obtain the drift angle residual error of each support; the drift angle residual error is the difference between the current drift angle and the reference angle; S3.2: For each support: determine whether the drift angle residual error is less than or equal to the residual error threshold value; if yes, end the support drift angle correction, otherwise, execute S3.3: S3.3: Adjust the push jack speed, push cylinder pressure and stroke, and return to S3.

1.

4. The hydraulic support azimuth angle deviation correction method for steeply inclined large mining height working face according to claim 3, characterized in that, Further comprising S3.4: controlling the adjustable damping adjuster to adjust the damping coefficient of the support at the two ends of the working face.

5. The hydraulic support azimuth angle correction method for steeply inclined large mining height working face according to any one of claims 1-4, characterized in that, Further comprising the following steps: The reference angle sequence is first-order low-pass filtered; the formula of the first-order low-pass filtering is: ; wherein, is the reference angle obtained after the first-order low-pass filtering, a is a filtering coefficient, a ∈ (0, 1), is the reference angle obtained after the first-order low-pass filtering of the previous time, is the reference angle obtained at present.

6. A hydraulic support azimuth angle correction system for steeply inclined large mining height working face, characterized in that, Comprise: A data acquisition module for acquiring pose deviation data between a support cluster and an anchor point, relative position relationship data between adjacent supports, and working condition state data of the support cluster, and establishing a data set; the anchor point is fixed on the side of the chute and the coal wall near the two ends of the working face; A first data processing module for performing robust regression analysis on the data set and applying consistency constraints to obtain a reference angle; The reference angle is used to represent the unified target orientation of the support cluster; A second data processing module for sequentially numbering the support cluster along the working face trend, and sequentially arranging the working condition state data of the support cluster according to the number to obtain a data sequence; The third data processing module is configured to calculate, by using the first sliding window, a sign consistency ratio, a sample variance, and a mean absolute value of the azimuth deviation on the data sequence; the azimuth deviation is a difference between the reference angle and the azimuth of the support; The first analysis control module is configured to control the fourth data processing module, the first azimuth adjustment module, and the second azimuth adjustment module to work when the same sign rate is greater than or equal to the first threshold value, the sample variance is less than or equal to the second threshold value, and the mean absolute value is greater than or equal to the third threshold value, and control the second azimuth adjustment module to work otherwise. The fourth data processing module is configured to divide the support cluster into a plurality of groups by modulo operation. The first azimuth adjustment module is configured to, for each group: continuously adjust the azimuth of the entire group of supports for multiple times by using the second sliding window until the root mean square of the azimuth deviation between adjacent supports is less than or equal to the fourth threshold value. The second azimuth adjustment module is configured to adjust the azimuth deviation of each support within a residual error threshold range. The first azimuth adjustment module comprises: The model building unit is used to construct the undirected graph of the support cluster and the cost function of a single support frame; the expression of the cost function is: ;in, J i Indicates the first i The cost function value of the scaffold. θ 0 As the reference angle, θ i For the first i The current flight angle of the support frame, w e The weights are the differences between the reference angle and the current course angle. θ ij For the first i The support frame and the adjacent first j The angle difference between the support frames, i and j For the stent number, N ( i ) indicates a support cluster. w n The weight of the flight angle difference between adjacent supports; The azimuth adjustment amount acquisition unit is configured to acquire, based on the undirected graph, an azimuth adjustment amount with a minimum cost function value for each support; The azimuth adjustment unit is configured to control each support to adjust the azimuth according to the corresponding azimuth adjustment amount; The first analysis control unit is configured to compare the root mean square of the azimuth deviation between adjacent supports with the fourth threshold value, and control the second azimuth adjustment module to work if the root mean square of the azimuth deviation between adjacent supports is less than or equal to the fourth threshold value, and control the azimuth adjustment amount acquisition unit to work otherwise.

7. The hydraulic support azimuth angle correction system for steeply inclined large mining height working face according to claim 6, characterized in that, The second azimuth adjustment module comprises: The azimuth residual error acquisition unit is configured to acquire the azimuth residual error of each support; the azimuth residual error is a difference between the current azimuth and the reference angle; The second analysis control unit is configured to, for each support: determine whether the azimuth residual error is less than or equal to the residual error threshold value; if yes, end the support azimuth correction, and if no, drive the comprehensive control unit to work; The comprehensive control unit is configured to adjust the push jack speed, the push cylinder pressure, and the stroke, and control the azimuth residual error acquisition unit to work.

8. The hydraulic support azimuth angle correction system for steeply inclined large mining height working face according to claim 7, characterized in that, Further comprising: The damping regulator control unit is configured to control the adjustable damping regulator to adjust the damping coefficient of the supports at both ends of the working face.

9. The hydraulic support azimuth angle correction system for steeply inclined large mining height working face according to any one of claims 6-8, characterized in that, Further comprising: a low-pass filtering module, configured to perform first-order low-pass filtering on the reference angle sequence; a formula of the first-order low-pass filtering is: ; wherein, is the reference angle obtained after the first-order low-pass filtering, a is a filtering coefficient, and a ∈ (0, 1), is the reference angle obtained after the first-order low-pass filtering of the previous time, is the reference angle obtained at the current time.

Images